Primary ocular melanoma from melanocytes within the eye constitutes uveal melanoma (UM), whereas ocular melanoma from melanocytes on the globe surface is considered conjunctival melanoma (CM). Although both UM and CM develop from similar neural crest lineages and are classified as ocular melanomas, they have distinct clinical behaviours, management, prognosis, cancer staging features, and molecular characteristics. Overall, 85% of melanomas in ocular regions arise from the uvea (iris, choroidal, and ciliary body), 5% arise from the conjunctiva, and 10% occur in other sites.1 The most common site of UM is the choroid, which is involved in 90% of cases.2

For most cases of UM, the primary treatment is enucleation. Alternative options include plaque brachytherapy and proton beam radiation; these options are currently unavailable in Hong Kong and similar Asian countries (eg, Singapore). Systemic metastases of UM most commonly affect the liver, followed by lung and bone. Conjunctival melanoma is usually managed by complete excisional biopsy with wide surgical margins, using a ‘no-touch’ technique. Adjuvant therapies such as cryotherapy or topical mitomycin C are offered, followed by reconstruction with an amniotic membrane graft. Orbital exenteration may be necessary for advanced tumours where local resection is not feasible. Metastatic disease, often to regional lymph nodes and the brain, occurs in 20% to 30% of patients with CM.3

Melanomas are generally rare tumours and their incidences are particularly low in Asian populations. The annual incidence of UM in Caucasian populations is 5.1 cases per million,2 whereas the respective incidences in Japanese4 and Korean5 populations are 0.2 and 0.6 cases per million. The incidences of CM are similar: 0.2 to 0.5 cases per million in Caucasian populations3 and 0.15 cases per million in Asian populations.6 Because of this rarity among Asian populations, very few studies have focused on Chinese patients. Epidemiological studies have largely involved Caucasian populations. Clinical characteristics may differ in Asian populations; for example, Asian patients are initially diagnosed with UM5 and CM6 at younger ages.

Known risk factors for UM include fair skin, light-coloured eyes, congenital ocular melanocytosis, ocular melanocytoma, and BAP1 tumour predisposition syndrome.7 In contrast, risk factors for CM include increased conjunctival pigmentation and a history of primary acquired melanosis. Predictors of recurrence or new tumour formation after CM treatment are older age, a history of prior conjunctival surgery, and advanced T subclassification in the American Joint Committee on Cancer (AJCC) staging system.8

This study explored the disease course and outcomes of UM and CM in Chinese patients without nodal metastasis on presentation.

This retrospective study included patients who received treatment for UM or CM in a single tertiary ophthalmic centre in Hong Kong between January 1994 and December 2019. Inclusion criteria included Chinese ethnicity and imaging-confirmed lack of tumour dissemination at diagnosis. Exclusion criteria were <6 months of follow-up, insufficient available information, loss to follow-up, or presence of any precursor lesions (eg, atypical primary acquired melanosis). The following data were recorded: patient demographics, tumour laterality, tumour characteristics (eg, presentation and staging according to the AJCC Cancer Staging Manual [8th Edition]9 10), investigations performed, treatment regimen, and final outcomes. The study adhered to the principles outlined in the Declaration of Helsinki.

Provisional clinical diagnoses of UM and CM were made based on each patient’s medical history, primary acquired melanosis status, and clinical presentation. Final diagnoses of UM and CM were confirmed by excisional biopsy or analysis of a specimen collected during definitive surgical treatment. Unless refused by the patient, positron emission tomography–computed tomography (PET-CT) scans of UM and CM were performed after 2001 (when such scans became commercially available). Magnetic resonance imaging scans of the brain and orbit, as well as fundus fluorescein angiography and indocyanine green angiography, were conducted to investigate suspected UM. All patients with pathologically confirmed UM or CM underwent definitive surgical treatment. The specific surgical treatment was selected according to melanoma location and size, depth of invasion, systemic metastasis status, and the patient’s physical condition.

SPSS software (Windows version 20; IBM Corp, Armonk [NY], US) was utilised for statistical analysis. Differences between patient groups were calculated using the Chi squared and Mann-Whitney U tests. P values <0.05 were considered statistically significant. Kaplan-Meier survival analysis was performed to estimate disease-free survival and overall survival in patients who had received definitive treatment. Continuous data were reported as mean ± standard deviation.

In this 25-year retrospective study, there were 13 and 11 patients with pathologically confirmed UM and CM, respectively. Two other patients were excluded: both displayed ocular symptoms (upper eyelid mass and bloody discharge) but were subsequently diagnosed with ocular metastases of primary nasal melanoma. The mean follow-up durations for UM and CM were 67 ± 53.2 and 74 ± 83.3 months, respectively. There were more men among patients with UM than among those with CM (P=0.042). The incidence of pre-existing conjunctival nevus was higher among patients with CM than among those with UM (P=0.049). There were no other significant differences in terms of age at diagnosis, tumour laterality, duration of symptoms, history of ocular nevi, or pathologically determined lesion diameter (Table 1).

Twenty-three of the 24 included patients had visual symptoms on presentation. Among patients with UM, 84.6% (n=11) reported blurring of vision and 7.7% (n=1) presented with photopsia. In one male patient (7.7%), the tumour was discovered during a routine ophthalmological examination. Patients with CM had more diverse visual symptoms: 63.6% (n=7) exhibited conjunctival pigmentation, 18.2% (n=2) had either an upper or lower eyelid mass, 9.1% (n=1) displayed bloody ocular discharge, and 9.1% (n=1) had experienced bleeding from the mass.

Retrospective melanoma staging of UM9 and CM10 was conducted in accordance with the AJCC Cancer Staging Manual (8th Edition). Uveal melanoma stages varied from T1a to T4a and from stage I to IIIa. Most UM cases were clinical stage II: 38.5% (n=5), 38.5% (n=5), 7.7% (n=1), and 15.3% (n=2) of patients with UM had clinical stage I, IIA, IIB, and IIIA tumours, respectively. Pathological staging of the tumours revealed that 46.2% (n=6) were spindle cell type, 38.5% (n=5) were epithelioid cell type, and the remaining 15.3% (n=2) were mixed cell type (ie, >10% epithelioid cells and <90% spindle cells). Pathological staging tended to be equivalent to or higher than the initial clinical staging: two patients with pathological stage IIB disease were initially diagnosed with clinical stage I disease, and three patients with pathological stage IIB disease were initially diagnosed with clinical stage IIA disease. For the remaining eight patients, the clinical and pathological stages were identical. Overall, 23.1% (n=3), 15.4% (n=2), 46.2% (n=6), and 15.3% (n=2) of patients with UM had pathological stage I, IIA, IIB and III tumours, respectively.

At diagnosis of CM, the clinical T (cT) stage was cT2 in 72.7% (n=8) of patients and cT3 in 27.3% (n=3) of patients. There were no cT1 or cT4 tumours in our cohort. The pathological T (pT) stage was pT2 in 54.5% (n=6) of patients and pT3 in 45.5% (n=5) of patients. No patients were diagnosed with pT1 or pT4 disease. Unlike the approach or UM, the AJCC staging system for CM does not include guidance regarding overall stage; there is only clinical and pathological staging for the T (tumour) component. One patient had a lower clinical stage (T2b) than pathological stage (T3b); all other patients had identical clinical and pathological stages.

In terms of disease management, biopsies were more frequently performed before definitive treatment in patients with CM than in those with UM (P=0.003). Enucleation was performed in 92.3% (n=12) of patients with UM, whereas orbital exenteration was performed in 81.8% (n=9) of patients with CM. The interventions and treatments performed are shown in Table 2.

After definitive treatment, all patients initially attended weekly follow-up visits; they gradually transitioned to follow-up at 6-month intervals. Clinical examinations were performed for any surgical complications or disease recurrence. In cases of suspected disease recurrence or metastasis, contrast CT scans of the brain and orbit were performed. No patients with UM or CM had short-term or long-term wound complications. There were no instances of local recurrence during follow-up.

One patient with UM (7.7%) had metastasis to bone, lung, and breast tissue, as determined by high-resolution CT at 105 months of follow-up. The patient died 117 months after initial diagnosis. Two patients (15.3%) died of causes unrelated to melanoma, such as hepatocellular carcinoma or squamous cell carcinoma of the lung.

Patients with CM had worse outcomes than those with had UM, in terms of systemic metastasis (P=0.031) and tumour-related mortality rates (P=0.006). Overall, 45.5% (n=4) of patients with CM developed systemic metastasis during follow-up; the mean time from definitive treatment to systemic metastasis was 55.8 ± 66 months (range, 14-168). Cases were detected when patients were symptomatic and admitted to an acute care hospital for whole-body PET-CT or brain magnetic resonance imaging. Lymph node metastasis in two patients and liver metastasis in one patient were confirmed by fine needle aspiration cytology. Brain metastasis was identified by brain CT in two patients, one of whom had simultaneous bone and lung metastases confirmed by PET-CT. Overall, 18.2% (n=2) of patients died of causes unrelated to melanoma. The mean time from definitive treatment to death was 39 ± 26 months (range, 1-64). Patients with CM had shorter median disease-free survival (P=0.004) and overall survival (P=0.006). Detailed results are presented in Table 3.

At 2 years after definitive treatment, the probabilities of disease-free survival were approximately 0.55 for CM and 1.0 for UM. At 10 years, these probabilities decreased to 0.4 for CM and 0.7 for UM. Overall survival was 100% among patients with UM at 10 years after definitive treatment. In contrast, patients with CM displayed a progressive decrease in overall survival, reaching 35% at 5 years after definitive treatment. Survival curves are depicted in Figures 1 and 2.

To our knowledge, this is the first retrospective study of disease course and outcomes among patients with UM and CM in southern China.

In the present study, the mean ages at diagnosis of UM and CM were 59 and 57 years, respectively. These findings are similar to the ages at diagnosis of UM in Taiwan (55 years)11 and the US (58 years),12 but slightly older than the ages in Singapore (52 years)13 and South Korea (53 years).14 The findings are also similar to the ages at diagnosis of CM in the US and mainland China (61 years15 and 54 years,16 respectively). It is clear that UM and CM both affect patients in their late 50s to early 60s, regardless of ethnicity.

Uveal melanoma primarily affected men in the present study, consistent with findings in Australian17 and European18 Caucasian populations. No previous studies have identified oestrogen receptors in normal uveal tissue or UM tumours,19 and there is no evidence that oral contraceptives or postmenopausal oestrogens participate in UM aetiology.20 Because female hormones do not have protective or exacerbating roles in UM, there is speculation that testosterone receptors are present on UM tumours, leading to a higher incidence in men.21 The present study revealed that CM primarily affected women, but previous studies have not found a relationship between sex and CM incidence.22 23 In contrast to our results, a study in mainland China showed that CM primarily affected men.15 Further studies are needed regarding the relationship between sex and CM.

Studies in Caucasian populations have demonstrated that the incidence of UM is much higher than that of CM, with a ratio of approximately 3 to 1.21 24 In the present study, CM constituted a larger proportion of ocular melanoma cases than previously identified in Caucasian populations, with results similar to epidemiological findings from Korea.5 Asian populations may have a lower risk of UM, but further research is warranted to confirm this speculation.

Zloto et al21 described an intriguing phenomenon whereby men were less likely than women to report symptoms of UM. Among our patients with UM, nearly all reported symptoms; one male patient did not report symptoms and had a tumour identified during a routine examination. Although this finding was not statistically significant, it is consistent with the previous results in a Caucasian population.2

In the present study, more patients with CM had pre-existing conjunctival pigmented lesions, compared with those who had UM. Somatic mutations in the BRAF gene frequently occur in human melanomas, including CM; such mutations are strongly associated with ultraviolent light exposure.25 In subtropical regions such as Hong Kong where there is abundant sunlight, analyses of underlying BRAF gene mutations could be insightful.

Among 10 ophthalmology centres on four continents (North and South America, Europe, and Asia), the proportions of UM stages I, IIA, IIB, and IIIA were 32%, 34%, 22.1%, and 8.8%, respectively10; the present study demonstrated a comparatively greater proportion of cases with stage IIB or higher.

Comparative analysis of CM data revealed similar results: patients in the present study had higher clinical stages relative to patients in the US, where three-quarters and more than half of the patients had clinical and pathological stage I disease. This discrepancy could be explained by the rare nature of CM, particularly in Asian populations, leading to lower disease awareness and delayed referral to an ophthalmologist.

Primary enucleation was an effective treatment for patients with UM in the present study; only one patient (7.7%) had systemic metastasis. In that patient, although no systemic metastasis was detected on PET-CT at diagnosis, the maximum standardised uptake value of the UM tumour was particularly high (9.4); UM tumours in other patients had maximum standardised uptake values of 0 to 3.9. Considering the exceptionally high metabolic rate in the UM tumour of the patient with systemic metastasis, microscopic metastasis may have been present before enucleation.

Despite radical treatment with wide excisional biopsy or primary orbital exenteration, systemic metastasis occurred in nearly half of the patients with CM. Regular imaging surveillance by PET-CT may be beneficial for patients with CM after orbital exenteration.

Among patients with UM, we observed a much lower rate of systemic metastasis than reported in Singapore (7.7% vs 45.5%26). This difference could be attributed to the performance of early radical treatment (ie, enucleation) before detection of systemic metastasis. Notably, the disease-free survival rate was better in our cohort of patients with UM than in another study of Chinese patients with UM (5- and 10-year disease-free survival rates of 80% and 70%, respectively)27 and better than in a study of Singaporean patients with UM (5-year disease-free survival rate of 56.8%).13 For comparison, in Caucasian populations, the 5-year and 10-year disease-free survival rates were 81.6%17 and 50%,28 respectively.

The rates of systemic metastasis and tumour-related mortality are consistently higher in patients with CM than in those with UM. In the present study, tumour-related mortality at 5 years was similar to the rate in Chinese patients (30.5%)16 but higher than that among patients in the US (7%).29 In five large studies (n=734 cases overall) of CM after surgical resection with tumour-free margins, the 5-year overall survival rates ranged from 74% to 86%.30 31 32 33 34 A recent Singaporean study revealed a slightly lower 5-year overall survival rate (68.8%).13 In the present study, the 5-year overall survival rate was substantially lower than the rates in other countries. Shields et al29 reported that a pathologically confirmed positive tumour margin and the absence of limbal involvement were risk factors for CM metastasis. Although all patients with CM in our cohort had pathologically confirmed clear margins, most patients with metastasis (n=6) had palpebral CM (83.3%, n=5) in which the melanoma did not reach the limbus. Thus, tumour location and ethnicity may explain the poor overall survival among patients with CM in the present study.

The higher risk of metastasis and lower rates of 5- and 10-year overall survival in CM, compared with UM, could be attributed to multiple factors. We did not find significant differences between UM and CM in terms of tumour stage or delays in diagnosis/treatment. We suspect that the differences between tumours are related to the nature of the disease, the primary mode of metastasis (UM spreads through the vasculature, whereas CM spreads through the lymphatic system), and genetic alterations (UM is associated with chromosomal abnormalities and CM is associated with mutations in specific genes). In terms of monitoring UM recurrence, the use of single-cell technologies to identify circulating tumour cells has implications for clinical stratification, particularly in cases of UM where specific genetic mutations have been identified.35 Because circulating tumour cell tests have received US Food and Drug Administration’s approval for clinical use in the management of various tumours (eg, metastatic breast and prostate cancers), they may be utilised in future efforts to detect circulating UM tumour cells.

In our centre, sentinel lymph node biopsy (SLNB) is not performed as a component of CM management. Mor et al36 recommend SLNB for patients with CM because false-negative findings are rare and 5-year survival can reach 79%. Therefore, early diagnosis of CM, including SLNB in cases with poor prognosticating factors (lack of limbal involvement and positive biopsy margin), and radical neck dissection as appropriate (with support from head and neck surgeons) should be considered before systemic treatment is offered.

This retrospective study included a low number of patients. Additionally, genetic testing was not performed on tissue samples from patients with pathologically confirmed tumours. Chromosomal abnormalities (monosomy 3, gain of chromosome 8q, and monosomy 3 combined with loss of 1p36) have been associated with decreased survival in UM,37 whereas mutations in the BRAF, RAS, cKIT, and NF1 genes have been associated with CM38; thus, survival could be more closely related to specific genetic features, and it may be inappropriate to consider these tumours as single entities.

Because UM and CM are rare conditions, they represent challenges for primary physicians (ie, timely referral) and ophthalmologists (ie, appropriate treatment and adequate long-term follow-up). Currently, there is limited information regarding the roles of newer targeted therapies for UM and CM, compared with the application of such therapies to cutaneous melanoma. Among patients with CM, long-term mortality remains high despite definitive radical treatment. This study explored the disease course and outcomes in Hong Kong Chinese patients, then compared the findings with data from patients in other countries. For patients with UM and CM, we recommend long-term follow-up with close monitoring, a detailed medical history, holistic assessment involving cervical and head lymph node palpation, and ophthalmological examination. Collaborations with oncologists to provide regular systemic evaluation during long-term followup, with the goal of early detection for distant metastases, are also important. Chest X-ray, brain magnetic resonance imaging, and cytology with SLNB should be performed regularly (annually if possible) to improve survival, particularly in patients with CM.

Concept or design: HKL Yuen.
Acquisition of data: JYY Chan.
Analysis or interpretation of data: SC Lam, JYY Chan.
Drafting of the manuscript: JYY Chan.
Critical revision of the manuscript for important intellectual content: SC Lam, HKL Yuen.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

All authors have disclosed no conflicts of interest.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

The study protocol was approved by the Research Ethics Committee (Kowloon Central/Kowloon East) of Hospital Authority, Hong Kong (Ref No.: KC/KE-21-0129/ER-1). Patients were treated in accordance with the tenets of the Declaration of Helsinki, provided written informed consent for all treatments and procedures, and consented to publication of this report.

1. Chang AE, Karnell LH, Menck HR. The National Cancer Data Base report on cutaneous and noncutaneous melanoma: a summary of 84,836 cases from the past decade. The American College of Surgeons Commission on Cancer and the American Cancer Society. Cancer 1998;83:1664-78. Crossref

2. Kaliki S, Shields CL. Uveal melanoma: relatively rare but deadly cancer. Eye (Lond) 2017;31:241-57. Crossref

3. Vora GK, Demirci H, Marr B, Mruthyunjaya P. Advances in the management of conjunctival melanoma. Surv Ophthalmol 2017;62:26-42. Crossref

4. Stang A, Parkin DM, Ferlay J, Jöckel KH. International uveal melanoma incidence trends in view of a decreasing proportion of morphological verification. Int J Cancer 2005;114:114-23. Crossref

5. Park SJ, Oh CM, Kim BW, Woo SJ, Cho H, Park KH. Nationwide incidence of ocular melanoma in South Korea by using the national cancer registry database (1999-2011). Invest Ophthalmol Vis Sci 2015;56:4719-24. Crossref

6. Hu DN, Yu G, McCormick SA, Finger PT. Population-based incidence of conjunctival melanoma in various races and ethnic groups and comparison with other melanomas. Am J Ophthalmol 2008;145:418-23. Crossref

7. Jager MJ, Shields CL, Cebulla CM, et al. Uveal melanoma. Nat Rev Dis Primers 2020;6:24. Crossref

8. Vaidya S, Dalvin LA, Yaghy A, et al. Conjunctival melanoma: risk factors for recurrent or new tumor in 540 patients at a single ocular oncology center. Eur J Ophthalmol 2021;31:2675-85. Crossref

9. Baron ED, Di Nicola M, Shields CL. Updated AJCC classification for posterior uveal melanoma. Available from: https://retinatoday.com/pdfs/0518RT_Oncology.pdf. Accessed 1 Mar 2021.

10. Jain P, Finger PT, Damato B, et al. Multicenter, international assessment of the Eighth Edition of the American Joint Committee on Cancer Cancer Staging Manual for conjunctival melanoma. JAMA Ophthalmol 2019;137:905-11. Crossref

11. Ma ST, Hsieh YT, Wei YH, Liao SL. A 45-year experience of uveal melanoma in Taiwan: verification of American Joint Committee on Cancer staging system and prognostic factors. J Formos Med Assoc 2021;120:1361-8. Crossref

12. Shields CL, Furuta M, Thangappan A, et al. Metastasis of uveal melanoma millimeter-by-millimeter in 8033 consecutive eyes. Arch Ophthalmol 2009;127:989-98. Crossref

13. Tan LL, Hong J, Goh WL, et al. Clinical features and survival outcomes of ocular melanoma in a multi-ethnic Asian cohort. Sci Rep 2020;10:16367. Crossref

14. Lee CS, Lee J, Choi JJ, et al. Cytogenetics and prognosis for uveal melanoma in Korean patients. Acta Ophthalmol 2011;89:e310-4. Crossref

15. Yousef YA, Finger PT. Predictive value of the seventh edition American Joint Committee on Cancer staging system for conjunctival melanoma. Arch Ophthalmol 2012;130:599-606. Crossref

16. Zhou C, Wang Y, Jia R, Fan X. Conjunctival melanoma in Chinese patients: local recurrence, metastasis, mortality, and comparisons with Caucasian patients. Invest Ophthalmol Vis Sci 2017;58:5452-9. Crossref

17. Vajdic CM, Kricker A, Giblin M, et al. Incidence of ocular melanoma in Australia from 1990 to 1998. Int J Cancer 2003;105:117-22. Crossref

18. Singh AD, Turell ME, Topham AK. Uveal melanoma: trends in incidence, treatment, and survival. Ophthalmology 2011;118:1881-5. Crossref

19. Mäkitie T, Tarkkanen A, Kivelä T. Comparative immunohistochemical oestrogen receptor analysis in primary and metastatic uveal melanoma. Graefes Arch Clin Exp Ophthalmol 1998;236:415-9. Crossref

20. Holly EA, Aston DA, Ahn DK, Kristiansen JJ, Char DH. Uveal melanoma, hormonal and reproductive factors in women. Cancer Res 1991;51:1370-2.

21. Zloto O, Pe’er J, Frenkel S. Gender differences in clinical presentation and prognosis of uveal melanoma. Invest Ophthalmol Vis Sci 2013;54:652-6. Crossref

22. McLaughlin CC, Wu XC, Jemal A, Martin HJ, Roche LM, Chen VW. Incidence of noncutaneous melanomas in the U.S. Cancer 2005;103:1000-7. Crossref

23. Wong JR, Nanji AA, Galor A, Karp CL. Management of conjunctival malignant melanoma: a review and update. Expert Rev Ophthalmol 2014;9:185-204. Crossref

24. Hu DN, Yu GP, McCormick SA, Schneider S, Finger PT. Population-based incidence of uveal melanoma in various races and ethnic groups. Am J Ophthalmol 2005;140:612-7. Crossref

25. Besaratinia A, Pfeifer GP. Sunlight ultraviolet irradiation and BRAF V600 mutagenesis in human melanoma. Hum Mutat 2008;29:983-91. Crossref

26. Wong W, Sundar G, Chee C, Zhao PS, Rajagopalan R, Gopal L. Clinical spectrum, treatment and outcomes of uveal melanoma in a tertiary centre. Singapore Med J 2019;60:474-8. Crossref

27. Yue H, Qian J, Yuan Y, et al. Clinicopathological characteristics and prognosis for survival after enucleation of uveal melanoma in Chinese patients: long-term follow-up. Curr Eye Res 2017;42:759-65. Crossref

28. Virgili G, Gatta G, Ciccolallo L, et al. Survival in patients with uveal melanoma in Europe. Arch Ophthalmol 2008;126:1413-8. Crossref

29. Shields CL, Shields JA, Gündüz K, et al. Conjunctival melanoma: risk factors for recurrence, exenteration, metastasis, and death in 150 consecutive patients. Arch Ophthalmol 2000;118:1497-507. Crossref

30. Tuomaala S, Eskelin S, Tarkkanen A, Kivelä T. Population-based assessment of clinical characteristics predicting outcome of conjunctival melanoma in whites. Invest Ophthalmol Vis Sci 2002;43:3399-408.

31. Missotten GS, Keijser S, De Keizer RJ, De Wolff-Rouendaal D. Conjunctival melanoma in the Netherlands: a nationwide study. Invest Ophthalmol Vis Sci 2005;46:75-82. Crossref

32. Shields CL, Kaliki S, Al-Dahmash SA, Lally SE, Shields JA. American Joint Committee on Cancer (AJCC) clinical classification predicts conjunctival melanoma outcomes. Ophthal Plast Reconstr Surg 2012;28:313-23. Crossref

33. Esmaeli B, Wang X, Youssef A, Gershenwald JE. Patterns of regional and distant metastasis in patients with conjunctival melanoma: experience at a cancer center over four decades. Ophthalmology 2001;108:2101-5. Crossref

34. Werschnik C, Lommatzsch PK. Long-term follow-up of patients with conjunctival melanoma. Am J Clin Oncol 2002;25:248-55. Crossref

35. Wang MM, Chen C, Lynn MN, et al. Applying single-cell technology in uveal melanomas: current trends and perspectives for improving uveal melanoma metastasis surveillance and tumor profiling. Front Mol Biosci 2021;7:611584. Crossref

36. Mor JM, Rokohl AC, Koch KR, Heindl LM. Sentinel lymph node biopsy in the management of conjunctival melanoma: current insights. Clin Ophthalmol 2019;13:1297-302. Crossref

37. van den Bosch T, Kilic E, Paridaens D, de Klein A. Genetics of uveal melanoma and cutaneous melanoma: two of a kind? Dermatol Res Pract 2010;2010:360136. Crossref

38. Gkiala A, Palioura S. Conjunctival melanoma: update on genetics, epigenetics and targeted molecular and immune-based therapies. Clin Ophthalmol 2020;14:3137-52. Crossref

Paediatric intensive care units (PICUs) are highly specialised workplaces that support children with critical illnesses and their caregivers. Advances in paediatric critical care have significantly improved survival among critically ill children, although this improvement has also led to higher rates of morbidity, more disabilities, and longer hospital stays.1 2 3 4 5 These changes have resulted in potentially conflicting views regarding expectations and treatment goals among healthcare workers and patients’ families, increasing the incidence of moral distress among healthcare workers.6

Moral distress is a term that refers to experiences of frustration and failure arising from healthcare workers’ attempts to fulfil their moral obligations to patients, families, and the public.7 8 In an intensive care setting, healthcare workers frequently encounter ethical issues. Moral distress arises when a healthcare worker has determined the right course of action but cannot follow it because of internal or external constraints (eg, limited resources, institutional policies, or family preferences).9 Moral distress has been identified among healthcare workers in both adult ICUs and PICUs.10 11 It is associated with greater experience and lower staff retention.12

Depression and stress-related symptoms are common in healthcare workers, particularly among ICU staff.13 14 Studies have shown that these symptoms can ultimately impair patient care quality.15 16 Thus far, most literature regarding moral distress has been published in Western countries; the concept of moral distress is not well-known outside of the Western world.17 To our knowledge, there have been few analyses of moral distress and psychological status among healthcare workers in non-Western PICUs. Factors that can influence the level and type of moral distress include cultural backgrounds; beliefs of the patient, their family, and the clinical team; and differences among healthcare systems. Hong Kong is a multicultural city influenced by both Eastern and Western cultures; challenges in this setting may be unique. This study assessed moral distress prevalence and psychological status among PICU healthcare workers in Hong Kong.

This prospective single-centre cross-sectional study was conducted from June to July 2020 in the six-bed tertiary PICU of the Hong Kong Children’s Hospital (HKCH), which began operation at the end of March 2019. The HKCH is the only dedicated paediatric oncology centre in the region, and most PICU admissions (54%) during the study period involved patients with cancer.

Study participants were healthcare workers involved in direct clinical care within the HKCH PICU, including doctors, nurses, and allied health professionals (ie, physiotherapists, occupational therapists, speech therapists, pharmacists, and dietitians). Healthcare workers were excluded if they had <3 months of critical care experience in the PICU or were temporarily on leave from the PICU during the study period. The survey was distributed to all eligible healthcare workers in the HKCH PICU during working hours within the study period.

The survey included two validated instruments (Revised Moral Distress Scale [MDS-R] Paediatric Version and Depression Anxiety and Stress Scale–21 items [DASS-21]) to measure levels of moral stress, depression, anxiety, and stress in all participants.18 19 The participants’ demographic details were also collected. The survey explored job-quitting intentions related to moral distress or other reasons. It was piloted with two HKCH PICU staff members; questions were refined based on feedback from them. The final survey was paper-based. An email was sent to all participants before study commencement with information regarding the aim and details of the study. The survey was distributed by hand, and all copies were collected in a sealed box after completion. To ensure anonymity, the survey did not contain any identifiers.

Moral distress, the main outcome of the study, was measured using the validated paediatric version of the MDS-R (online supplementary Appendix 1).18 It consists of 21 items describing predetermined potentially morally distressing situations. There are five predetermined categories of situations: end-of-life care and quality of life, poor communication, staffing and material resources, hierarchies of decision making, and witnessing unethical behaviour. Each item on the MDS-R is scored according to the frequency and intensity that a healthcare worker experienced, using a Likert scale that ranges from 0 to 4. If a specific situation has never been experienced, participants are asked to indicate how disturbing the situation would be if they encountered it in their workplace. The frequency and intensity scores are then multiplied to produce an overall score for each item. The total moral distress score is the sum of the 21 overall scores for each item, ranging from 0 to 336. The English version of this instrument was used.

Psychological status was assessed using the DASS-21 (online supplementary Appendix 2).19 It is a set of three self-reporting subscales that measure participants’ emotional states: depression, anxiety, and stress. Each scale contains seven items for each emotional state. Each item is scored on a four-point Likert scale ranging from 0 (‘Did not apply to me at all’) to 3 (‘Applied to me very much or most of the time’). The total score for each emotional state is the sum of the subscale scores multiplied by 2. Depression, anxiety, or stress was considered present if the relevant scores exceeded the normal cut-off. The emotional state was categorised as mild, moderate, severe, or extremely severe, based on published cut-offs. The English and Chinese versions of this instrument were used; both language versions have been validated.19 20

Outcome measures were demographic data and the levels of moral distress, depression, anxiety, and stress. Data were expressed using median (interquartile range [IQR]) for continuous variables and count (percentage) for categorical variables. Results of the MDS-R and DASS-21 were compared among doctors, nurses, and allied health professionals using the Chi squared test, Kruskal-Wallis test, or Cohen’s d. Correlations between participant variables and outcome measures were evaluated using Spearman’s rank correlation coefficient. P values <0.05 were considered statistically significant. Statistical analysis was performed using SPSS (Windows version 26.0; IBM Corp, Armonk [NY], United States).

In total, 46 of 56 healthcare workers in the PICU completed the survey; the response rate was 82%. On one survey, the moral distress section was incomplete; that survey was excluded from the analysis of moral distress.

Most participants were women (n=36, 78%) and were aged ≥30 years (n=35, 76%). More than half of the participants were nurses (n=26, 57%). Approximately half of the participants (n=24, 52%) had >5 years of PICU experience. Detailed participant characteristics are presented in Table 1.

The median MDS-R score was 71 (IQR=34-115). There was a significant difference in MDS-R score among the three professions (P<0.001). Doctors and nurses had significantly higher MDS-R scores, compared with allied health professionals (P<0.05). Nurses had the highest median MDS-R score (102, IQR=71-126), whereas allied health professionals had the lowest (20, IQR=6-39). There were no significant differences in MDS-R score according to sex, age, or duration of PICU experience (Table 1).

Among the 21 items on the MDS-R, the most morally distressing item was related to end-of-life care and quality of life: ‘Honour the family’s wishes to continue life support even though I believe it is not in the child’s best interest’. This item also scored highest in frequency and intensity among the 21 items. All three groups of health professionals ranked this item as the most morally distressing situation in the clinical setting. The second most morally distressing item was also related to end-of-life care and quality of life: ‘Initiate extensive life-saving actions when I think they only prolong death’. This item also consistently scored high in frequency and intensity (Table 2). Situations involving poor communication constituted the remaining three most morally distressing items in this study. The top five most morally distressing items, as well as the top five items with the highest frequency and intensity, are presented in Table 2.

A higher MDS-R moral distress score was associated with the intention to quit. Healthcare workers who intended to quit their jobs had significantly higher moral distress scores (P<0.05). A higher moral distress score was also associated with higher DASS-21 depression factor (r=0.445; P<0.05) and anxiety factor scores (r=0.417; P<0.05). Nurses who had worked for a greater number of years in the PICU also experienced higher moral distress (r=0.512; P<0.05). Twenty-eight percent of all participants and 35% of nurses reported they intended to quit their jobs because of moral distress.

The median depression, anxiety, and stress scores were 11 (IQR=0.5-18), 8 (IQR=3-145), and 30 (IQR=21-38), respectively; these scores corresponded to mild depression, mild anxiety, and severe stress. Among the three groups, nurses had the highest median anxiety (11, IQR=6-16) and stress scores (20, IQR=12-26) [Fig]; these scores corresponded to mild depression, moderate anxiety, and moderate stress. Participants with significantly higher depression and anxiety (both P<0.05) scores also intended to quit their jobs. There was no significant difference in stress score between participants who did and did not intend to quit their jobs (P=0.434).

In this study, various levels of moral distress were present in all three groups of PICU healthcare workers. There was a significant difference in MDS-R scores among the three professions, and nurses had the highest median MDS-R score. This finding is contrary to the results of previous PICU studies, which showed that moral distress did not differ among various healthcare workers.21 22 The literature suggests that nurses exhibit higher moral distress scores because they often have less autonomy concerning options in situations that involve moral dilemmas, and they are required to implement care plans with which they do not agree.23 24 25 26 Studies of PICU healthcare workers’ behaviours in ethical and morally distressing dilemmas have shown that 48% of PICU nurses reported needing to perform actions that violated their conscience. These results reflect the culture and hierarchies of power in the PICU.23 26 27 Moreover, nurses are the frontline workers who directly experience the impacts of clinical decisions on patients and their families.26 28 In newly established PICUs, decreased self-confidence or increased fear in a new working environment, combined with an uncertain ethical climate, unclear team dynamics, and less decision-making autonomy regarding care plans, can cause nurses to perceive less moral agency (ie, ability to act morally and change a situation).22 24 25 26 29 30 31 A reduced sense of moral agency can result in moral distress, which may be more apparent in newly established PICUs.29 31

We note that moral distress scores among nurses in the present study are among the highest in published studies of PICU healthcare workers (Table 3). In addition to the aforementioned lack of clarity in working environment and team dynamics, the diverse levels of experience among nurses might have also contributed to their high moral distress scores. In the present study, 54% of nurses had <3 years of PICU experience, whereas 39% of nurses had >10 years of PICU experience. These proportions of nurses with extensive and minimal experience were both larger than the proportions reported in previous PICU studies.32 33 The presence of such a large number of inexperienced junior nurses in the PICU may place additional stress on more experienced nurses. Indeed, survey items related to staffing (item 17 ‘Work with nurses or other care providers who are less competent than the child’s care requires’ and item 21 ‘Work with levels of care provider staffing that I consider unsafe’ in the MDS-R) were ranked by nurses as the seventh and eighth most morally distressing items; these rankings were higher than in other professions.

The PICU case mix might also contribute to moral distress. The majority of PICU admissions during the study period involved patients with cancer, who had considerably higher mortality rates; care for such patients frequently involved end-of-life and palliative care issues.34 35 In a study of nurses’ experiences while caring for dying children, Davies et al36 found that when nurses recognise a child’s death is inevitable, they often have to manage conflicting obligations: follow the doctor’s treatment orders and allow the child to die without unnecessary pain. These disparate treatment goals for critically ill children with terminal cancer can exacerbate moral distress.36 37 In a comparison of moral distress scores among various paediatric disciplines (eg, general care and surgical service), Trotochaud et al21 found that healthcare workers in haematology/oncology areas experienced the second highest amount of moral distress on the list, second to healthcare workers in PICUs. Moreover, the proportion of patients with cancer in our PICU is much higher than the proportions in previous PICU studies.38 39 Therefore, it is entirely understandable that moral distress in our PICU was particularly high among nurses.

The present study revealed a positive correlation between years of PICU experience and moral distress scores among nurses, consistent with previous results concerning healthcare workers in PICUs and adult ICUs.12 26 This correlation may be related to effective utilisation of clinical knowledge and experience, along with greater awareness concerning the impacts of potentially inappropriate treatment plans on patients.40 Conversely, a study by Larson et al26 revealed a negative correlation between moral distress scores and years of experience among doctors in the PICU. However, the present study showed no correlation between moral distress scores and years of experience among doctors. This finding might be attributed to the small number of doctors involved, which was insufficient to demonstrate an association.

Moral distress is often associated with the intention to quit a job.41 42 43 44 The results of the study were consistent with previous findings. Studies by Sannino et al11 and Trotochaud et al21 showed that 10.3% to 25% of PICU nurses intended to quit their jobs because of moral distress. The proportion of nurses in our study who intended to quit their job because of moral distress (34.6%) was higher than the proportions in previous PICU studies,11 21 which could be explained by their high moral distress scores. However, further studies are needed to determine the impact of moral distress alone on a healthcare worker’s intention to quit their job, compared with other possible distressing factors (eg, working hours and promotional opportunities) that can have a synergistic effect on the decision to quit.

To our knowledge, this is the first study of moral distress among healthcare workers in an East Asian PICU. The results of this study provide insights concerning the broader understanding of moral distress in newly established PICUs. The high response rate also suggests strong participation and indicates that the study sample is representative of healthcare workers in our PICU.

However, the results of this study should be interpreted with the following caveats. First, this was a single-centre study with a relatively small sample size, which limits the generalisability of the findings. The small sample size also hindered further evaluation of identifiable demographic factors, such as education level and whether participants had any children; another study indicated that such factors may be associated with moral distress.11 Moreover, the small sample size precluded subgroup analysis. Second, this study was susceptible to ‘survivorship’ bias because the sample did not include PICU staff who already quit their jobs, including some who quit because of moral distress. Third, considering the cross-sectional nature of this study, causal relationships among various factors could not be established. For example, although participants with higher depression and anxiety scores reported a stronger intention to quit their jobs, we could not determine whether these participants reported more psychological symptoms because of their intention to quit, or if their intention to quit led to more psychological symptoms. Larger multicentre studies are needed to further explore moral distress among healthcare workers in Hong Kong PICUs. As our unit expands to a 16-bed PICU and a five-bed high-dependency unit, a longitudinal study will also enhance the broader understanding of moral distress dynamics in a developing PICU, as well as the efficacies of various strategies to address moral distress.

Considering the results of this study, moral distress should be regarded as a key area for service improvement. The high levels of moral distress experienced by nurses, as well as the substantial moral distress in relation to end-of-life care, suggest that coping strategies should target nurses and focus on end-of-life education. These coping strategies are urgently needed to improve staff retention and quality of care; they can be implemented at the individual, organisational, and administrative levels.20

At the individual level, ethics education is essential for improvements in coping capacity and sense of moral agency, which can reduce the levels of moral distress.22 45 Education can be provided through interactive workshops or self-guided programmes.41 Prentice et al42 suggested that education should focus on improving knowledge regarding patient outcomes, the degree of uncertainty in specific situations, and appropriate pain control. Instead of emphasising ethical dilemmas and underlying principles, education should highlight communication skills, clarify values, and enhance the overall understanding of the healthcare system to address potential environmental conflicts.31 This approach can ultimately increase staff confidence (ie, moral courage) in constructively communicating their concerns.42 Screening tools for various emotional states, such as the DASS-21, should also be included to help individuals gain better awareness of their own psychological well-being and seek professional help if necessary. Additionally, these tools can be used to monitor emotions that might cause moral distress.

At the organisational level, efforts should be made to promote intra- and interdisciplinary communication. Poor communication, one of the five most morally distressing items, can lead to diminished quality of care, reduced job satisfaction, and poor patient outcomes.46 Ethics rounds, formal and informal discussions, and debriefing sessions regarding morally distressing cases could improve interdisciplinary communication.22 These initiatives can help promote better mutual understanding of viewpoints across disciplines and individuals.22 Participation in these events may also allow nurses to feel more empowered and experience a greater sense of decision-making autonomy.43 Finally, the establishment of formal ethical consultation services may provide support and clarification with respect to ethical dilemmas.44

At the administrative level, administrators should recognise that it is acceptable for staff to perceive moral distress; this perception is a sign of humanity and an affirmation of moral values.44 Improvements in clinical environments (eg, reduction of staff shortages, promotion of intra- and interdisciplinary collaboration, and encouragement of a safe and supported ethical climate) can help decrease moral distress.47 These measures include providing respectful feedback to staff, empowering staff to voice perceptions and emotions, and making difficult decisions in a timely manner after open discussion.48

This study revealed significant differences in moral distress among doctors, nurses, and allied health professionals in a newly established PICU in Hong Kong. Nurses had the highest moral distress scores among the three groups of PICU healthcare workers in this study and among published studies involving PICU nurses. Most areas of moral distress were related to end-of-life care and poor communication. Higher moral distress was also associated with greater depression, anxiety, and intention to quit. There is an urgent need for interventions to help healthcare workers cope with moral distress and create a more sustainable working environment.

Concept or design: WL Cheung, KL Hon, KKY Leung, WF Hui.
Acquisition of data: WL Cheung, KL Hon, KKY Leung, WF Hui.
Analysis or interpretation of data: All authors.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

As an editor of the journal, KL Hon was not involved in the review process. Other authors have disclosed no conflicts of interest.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

This research was approved by the Hong Kong Children’s Hospital Research Ethics Committee (Ref No.: HKCH-REC-2020-008) and was conducted in accordance with the Declaration of Helsinki and International Conference on Harmonisation Good Clinical Practice Guideline. All participants provided informed consent to take part in the research.

1. Jones S, Rantell K, Stevens K, et al. Outcome at 6 months after admission for pediatric intensive care: a report of a national study of pediatric intensive care units in the United Kingdom. Pediatrics 2006;118:2101-8. Crossref

2. Cremer R, Leclerc F, Lacroix J, Ploin D; GFRUP/RMEF Chronic Diseases in PICU Study Group. Children with chronic conditions in pediatric intensive care units located in predominantly French-speaking regions: prevalence and implications on rehabilitation care need and utilization. Crit Care Med 2009;37:1456-62. Crossref

3. Namachivayam P, Shann F, Shekerdemian L, et al. Three decades of pediatric intensive care: who was admitted, what happened in intensive care, and what happened afterward. Pediatr Crit Care Med 2010;11:549-55. Crossref

4. Edwards JD, Houtrow AJ, Vasilevskis EE, et al. Chronic conditions among children admitted to U.S. pediatric intensive care units: their prevalence and impact on risk for mortality and prolonged length of stay. Crit Care Med 2012;40:2196-203. Crossref

5. Ebrahim S, Singh S, Hutchison JS, et al. Adaptive behavior, functional outcomes, and quality of life outcomes of children requiring urgent ICU admission. Pediatr Crit Care Med 2013;14:10-8. Crossref

6. Bruce CR, Miller SM, Zimmerman JL. A qualitative study exploring moral distress in the ICU team: the importance of unit functionality and intrateam dynamics. Crit Care Med 2015;43:823-31. Crossref

7. Austin W. Moral distress and the contemporary plight of health professionals. HEC Forum 2012;24:27-38. Crossref

8. Jameton A. Nursing practice: the ethical issues. Englewood Cliffs (NJ): Prentice Hall Inc; 1984.

9. Garros D, Austin W, Carnevale FA. Moral distress in pediatric intensive care. JAMA Pediatr 2015;169:885-6. Crossref

10. Lusignani M, Giannì ML, Re LG, Buffon ML. Moral distress among nurses in medical, surgical and intensive-care units. J Nurs Manag 2017;25:477-85. Crossref

11. Sannino P, Giannì ML, Carini M, et al. Moral distress in the pediatric intensive care unit: an Italian study. Front Pediatr 2019;7:338. Crossref

12. Dodek PM, Wong H, Norena M, et al. Moral distress in intensive care unit professionals is associated with profession, age, and years of experience. J Crit Care 2016;31:178-82. Crossref

13. Ohayon MM, Priest RG, Guilleminault C, Caulet M. The prevalence of depressive disorders in the United Kingdom. Biol Psychiatry 1999;45:300-7.Crossref

14. de Vargas D, Dias AP. Depression prevalence in intensive care unit nursing workers: a study at hospitals in a northwestern city of São Paulo State. Rev Lat Am Enfermagem 2011;19:1114-21. Crossref

15. Poghosyan L, Clarke SP, Finlayson M, Aiken LH. Nurse burnout and quality of care: cross-national investigation in six countries. Res Nurs Health 2010;33:288-98. Crossref

16. Humphries N, Morgan K, Conry MC, McGowan Y, Montgomery A, McGee H. Quality of care and health professional burnout: narrative literature review. Int J Health Care Qual Assur 2014;27:293-307. Crossref

17. Prompahakul C, Epstein EG. Moral distress experienced by non-Western nurses: an integrative review. Nurs Ethics 2020;27:778-95. Crossref

18. Hamric AB, Borchers CT, Epstein EG. Development and testing of an instrument to measure moral distress in healthcare professionals. AJOB Prim Res 2012;3:1-9. Crossref

19. Tran TD, Tran T, Fisher J. Validation of the Depression Anxiety Stress Scales (DASS) 21 as a screening instrument for depression and anxiety in a rural community-based cohort of northern Vietnamese women. BMC Psychiatry 2013;13:24. Crossref

20. Wang K, Shi HS, Geng FL, et al. Cross-cultural validation of the Depression Anxiety Stress Scale–21 in China. Psychol Assess 2016;28:e88-100. Crossref

21. Trotochaud K, Coleman JR, Krawiecki N, McCracken C. Moral distress in pediatric healthcare providers. J Pediatr Nurs 2015;30:908-14. Crossref

22. Sauerland J, Marotta K, Peinemann MA, Berndt A, Robichaux C. Assessing and addressing moral distress and ethical climate part II: neonatal and pediatric perspectives. Dimens Crit Care Nurs 2015;34:33-46. Crossref

23. Papathanassoglou ED, Karanikola MN, Kalafati M, Giannakopoulou M, Lemonidou C, Albarran JW. Professional autonomy, collaboration with physicians, and moral distress among European intensive care nurses. Am J Crit Care 2012;21:e41-52. Crossref

24. Corley MC. Nurse moral distress: a proposed theory and research agenda. Nurs Ethics 2002;9:636-50. Crossref

25. Wilkinson JM. Moral distress in nursing practice: experience and effect. Nurs Forum 1988;23:16-29. Crossref

26. Larson CP, Dryden-Palmer KD, Gibbons C, Parshuram CS. Moral distress in PICU and neonatal ICU practitioners: a cross-sectional evaluation. Pediatr Crit Care Med 2017;18:e318-26. Crossref

27. Solomon MZ, Sellers DE, Heller KS, et al. New and lingering controversies in pediatric end-of-life care. Pediatrics 2005;116:872-83. Crossref

28. Molloy J, Evans M, Coughlin K. Moral distress in the resuscitation of extremely premature infants. Nurs Ethics 2015;22:52-63. Crossref

29. McCarthy J, Gastmans C. Moral distress: a review of the argument-based nursing ethics literature. Nurs Ethics 2015;22:131-52. Crossref

30. LaSala CA, Bjarnson D. Creating workplace environments that support moral courage. Online J Issues Nurs 2010;15:Manuscript 4. Crossref

31. Epstein EG, Delgado S. Understanding and addressing moral distress. Online J Issues Nurs 2010;15:Manuscript 1. Crossref

32. Ahmed GE, Abosamra OM. Knowledge of pediatric critical care nurses regarding evidence based guidelines for prevention of ventilator associated pneumonia (VAP). J Educ Pract 2015;6:94-101.

33. Hickey PA, Pasquali SK, Gaynor JW, et al. Critical care nursing’s impact on pediatric patient outcomes. Ann Thorac Surg 2016;102:1375-80. Crossref

34. Earle CC, Park ER, Lai B, Weeks JC, Ayanian JZ, Block S. Identifying potential indicators of the quality of end-of-life cancer care from administrative data. J Clin Oncol 2003;21:1133-8. Crossref

35. Wösten-van Asperen RM, van Gestel JP, van Grotel M, et al. PICU mortality of children with cancer admitted to pediatric intensive care unit a systematic review and meta-analysis. Crit Rev Oncol Hematol 2019;142:153-63. Crossref

36. Davies B, Clarke D, Connaughty S, et al. Caring for dying children: nurses’ experiences. Pediatr Nurs 1996;22:500-7.

37. Johnston EE, Alvarez E, Saynina O, Sanders L, Bhatia S, Chamberlain LJ. Disparities in the intensity of end-of-life care for children with cancer. Pediatrics 2017;140:e20170671. Crossref

38. Demaret P, Pettersen G, Hubert P, Teira P, Emeriaud G. The critically-ill pediatric hemato-oncology patient: epidemiology, management, and strategy of transfer to the pediatric intensive care unit. Ann Intensive Care 2012;2:14. Crossref

39. Dalton HJ, Slonim AD, Pollack MM. Multicenter outcome of pediatric oncology patients requiring intensive care. Pediatr Hematol Oncol 2003;20:643-9. Crossref

40. Zhang W, Wu X, Zhan Y, Ci L, Sun C. Moral distress and its influencing factors: a cross-sectional study in China. Nurs Ethics 2018;25:470-80. Crossref

41. Beumer CM. Innovative solutions: the effect of a workshop on reducing the experience of moral distress in an intensive care unit setting. Dimens Crit Care Nurs 2008;27:263-7. Crossref

42. Prentice T, Janvier A, Gillam L, Davis PG. Moral distress within neonatal and paediatric intensive care units: a systematic review. Arch Dis Child 2016;101:701-8. Crossref

43. Browning AM. CNE article: moral distress and psychological empowerment in critical care nurses caring for adults at end of life. Am J Crit Care 2013;22:143-51. Crossref

44. Santos RP, Garros D, Carnevale F. Difficult decisions in pediatric practice and moral distress in the intensive care unit. Rev Bras Ter Intensiva 2018;30:226-32. Crossref

45. Burston AS, Tuckett AG. Moral distress in nursing: contributing factors, outcomes and interventions. Nurs Ethics 2013;20:312-24. Crossref

46. Whitehead PB, Herbertson RK, Hamric AB, Epstein EG, Fisher JM. Moral distress among healthcare professionals: report of an institution-wide survey. J Nurs Scholarsh 2015;47:117-25. Crossref

47. Atabay G, Çangarli BG, Penbek Ş. Impact of ethical climate on moral distress revisited: multidimensional view. Nurs Ethics 2015;22:103-16. Crossref

48. Benoit DD, Jensen HI, Malmgren J, et al. Outcome in patients perceived as receiving excessive care across different ethical climates: a prospective study in 68 intensive care units in Europe and the USA. Intensive Care Med 2018;44:1039-49. Crossref

Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), was first identified in December 2019 in Wuhan, Hubei, China.1 2 According to World Health Organization Coronavirus (COVID-19) data, as of 7 May 2022, 188 countries and territories had reported more than 510.2 million cumulative confirmed cases and more than 6.23 million deaths3; in Hong Kong, there were 330 670 confirmed cases and 9308 (2.81%) deaths.4 Common symptoms of COVID-19 include fever, sore throat, cough, malaise, dyspnoea, and anosmia or aguesia.1 Although most people have mild symptoms, some develop acute respiratory distress syndrome, which may lead to cytokine storm, multiorgan failure, septic shock, and even death.5

There is evidence that rash is an early symptom or the only symptom in patients who are ‘asymptomatic’ or paucisymptomatic.6 7 8 9 Early detection of this ‘silent’ sign and corresponding diagnosis are important for epidemiologic management because asymptomatic or paucisymptomatic cases may function as sources of community spread. Various dermatologic manifestations of COVID-19 have been reported including maculopapular eruption, urticarial eruption, livedo reticularis, pernio/chilblain, vasculitis, vesicular eruption, and papulo-necrotic eruption.10 11 12 13 14 15 The incidences of cutaneous manifestations in patients with COVID-19 have varied among case series (from 0.2% to 20.4%10 11 12 13), possibly because of the under-recognition of asymptomatic or paucisymptomatic cases.

The spread of SARS-CoV-2 mainly involves droplets; it can also occur via direct contact and is speculated to occur through faecal excretion.2 The primary target of SARS-CoV-2 is the upper respiratory mucosa, where angiotensin-converting enzyme 2 (ACE2) serves as a functional receptor for viral spikes and eventual viral entry into host cells. Gene expression of the SARS-CoV-2 cellular receptor ACE2 has been demonstrated in multiple human tissues, including skin and adipose tissue.16 17 18 Therefore, the proposed mechanisms by which SARS-CoV-2 affect cutaneous tissues include direct attacks on epidermal basal cells and vascular endothelial cells (possibly targeting ACE2 expressed on skin keratinocytes) and indirect impacts through the antiviral inflammatory response.16 17 18

There is speculation that patients with rash occurrence may have a better prognosis because they display better antiviral immunity.19 Early in the COVID-19 pandemic, little was known about relationships among cutaneous manifestations, viral load, co-morbidities, and clinical outcomes. A recent systematic review showed inconclusive results about the relationship between COVID-19 severity and viral load; however, it suggested that older age and higher SARS-CoV-2 viral load were directly related.20 Likewise, some rashes such as maculopapular rash and chilblain-like lesions were found to be strongly associated with paucisymptomatic disease course and lower severity of COVID-19 while skin changes such as acro-ischaemia, livedo reticularis and purpura may be useful indicators of higher severity of COVID-19.21 22 In 2020, according to the Public Health Ordinance of Hong Kong, all patients with SARS-CoV-2–positive test results were hospitalised for quarantine, regardless of symptoms.4 Here, we explored the incidences and patterns of clinical and cutaneous manifestations among hospitalised patients with confirmed COVID-19, then investigated associations with viral load, co-morbidities, and prognosis.

This retrospective cohort study was conducted from 1 July to 30 September 2020 in an acute tertiary hospital, Queen Mary Hospital (ie, a major public hospital within one of seven hospital clusters) serving one-fifth of the population of 7.5 million in Hong Kong. Electronic hospital records were used to identify adult patients aged ≥18 years who were admitted during the study period for suspected COVID-19.

The flow of patient recruitment is illustrated in Figure 1. Patients included in this study were adults with laboratory confirmation of COVID-19 by real-time reverse transcription polymerase chain reaction (rRT-PCR) assay from a nasopharyngeal swab. Clinical information was collected from electronic clinical photographs of patients who had provided informed consent to receive treatment. A physical examination was performed by a dermatologist within 48 hours of rash onset to confirm clinical signs; follow-up was conducted monthly until 3 months after discharge. Rashes were considered COVID-19–related if they were new, could not be explained by the patient’s previous or pre-existing skin conditions or an alternative diagnosis (eg, drug eruption or other viral exanthem of varicella, parvovirus, enterovirus, influenza, parainfluenza, adenovirus, or respiratory syncytial virus detected in nasopharyngeal swab [performed as clinically indicated and excluded]), occurred along with the SARS-CoV-2–positive rRT-PCR test results, and resolved when other symptoms improved.

Clinical and laboratory data, including patient demographics, initial COVID-19 viral load according to cycle threshold (Ct) value, treatment received, co-morbidities (diabetes mellitus, hypertension, and chronic kidney disease [CKD]), and pre-existing skin diseases, were retrieved from electronic medical records for analysis. For the detection of viral nucleic acids, rRT-PCR is considered a gold standard diagnostic assay. The Ct value refers to the number of rRT-PCR cycles needed to amplify viral RNA to a detectable level; it is inversely related to viral load.23 Thus, the Ct value can indicate the relative quantity of viral RNA in a specimen (lower Ct values reflect greater quantities of viral RNA). In this study, Ct values of <26, 26-30, and ≥31 were regarded as high, intermediate, and low viral load, respectively.24 25

Continuous variables were expressed as medians (interquartile ranges) or means (± standard deviations), as appropriate. The Mann-Whitney U test and Kruskal-Wallis test were used to compare median values between two groups and among ≥3 groups, respectively. Categorical variables, expressed as proportions, were compared using the Chi squared test or Fisher’s exact test, as appropriate.

To identify factors independently associated with outcomes, variables with P values <0.1 in univariate analyses were subsequently entered into binary logistic regression multivariate analyses; odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. All statistical analyses were performed using SPSS (Windows version 26.0; IBM Corp, Armonk [NY], United States). Two-tailed P values <0.05 were considered statistically significant.

From 1 July to 30 September 2020, 414 patients with suspected COVID-19 were admitted to our hospital. This study included 219 patients who had SARS-CoV-2–positive rRT-PCR results in analyses of nasopharyngeal swab samples (from 213 recovered patients and six patients who had died). One hundred and ninety-five patients were excluded because of non–COVID-19 diagnosis, unconfirmed status, non-Asian ethnicity, or refusal to consent (Fig 1).

The mean patient age was 54.7 ± 17.5 years (range, 18-99), the male-to-female ratio was approximately 1:1, and 90.4% of the patients were Chinese (Table 1). The mean duration of hospitalisation was 9.87 ± 6.99 days and the overall mortality rate was 2.7%. The mean SARS-CoV-2 rRT-PCR Ct values for nasopharyngeal swab on admission was 24.2 ± 7.1. The median time to the first post-discharge visit was 38 days (range, 28-42) and the median duration of follow-up was 14 weeks (range, 13.1-15.5).

The three most frequent symptoms were upper respiratory symptoms: cough (51.5%), fever (42.5%), and sputum production (27.8%). Among the 219 patients with positive SARS-CoV-2 test results, 58 (26.5%) were asymptomatic and had undergone compulsory SARS-CoV-2 testing in accordance with the Public Health Ordinance. Of the 58 patients, 75.9% reported contact with identifiable index cases, such as household members, domestic helpers, or work colleagues.

Twenty patients presented with new rash. The incidence of new rash was 9.1% in this 3-month study period (95% CI=6.25%-14.4%). At the time of this study, there were no biomarkers or diagnostic tests for COVID-19–related cutaneous manifestations. Any new cutaneous manifestation not attributable to a previous/pre-existing skin disease or alternative diagnosis was considered COVID-19–related. Upon review by a dermatologist, six patients were diagnosed with localised urticarial eruptions after interferon injection treatment; 6.4% of patients (14/219) displayed various forms of COVID-19–related rash (Fig 2 and Table 2).

The most common manifestations were maculopapular exanthem (n=6, 42.9%, median Ct value: 24.8), followed by livedo reticularis (n=4, 28.6%, median Ct value: 21.3), varicella-like lesions (n=2, 14.3%, median Ct value: 19.3), urticaria (n=1, 7.1%, median Ct value: 14.4), and acral chilblain and petechiae (n=1, 7.1%, median Ct value: 33.1) [Fig 2]. The median Ct values for patients with and without rash were 22.9 and 24.1, respectively (P=0.58). The timing of symptom onset ranged from day 1 to day 9 (median, 4; mean, 4.28 ± 2.26). Skin symptoms were the sole symptoms in two patients with COVID-19 (0.91%), highlighting the importance of carefully evaluating patients who only display initial cutaneous symptoms or signs.

Compared with patients without rash, patients with rash were more likely to exhibit fever (OR=5.73; P=0.008) and display pulmonary infiltrates on chest X-ray (OR=5.06; P=0.013). Among patients with pulmonary infiltrates (n=19), four of them had rash. The episodes of desaturation requiring supplemental oxygen were less common in patients with rash (25%, 1/4) than in those without (93.3%, 14/15; OR=0.02, 95% CI=0.001-0.49; P=0.02). Furthermore, among these 19 patients with pulmonary infiltrates, systemic corticosteroids were less frequently required by patients with rash (25%, 1/4) than by those without (73.3%, 11/15; OR=0.12, 95% CI=0.01-1.53; P=0.10), but it was not statistically significant. There were no significant differences in age, sex, co-morbidities or Ct values between patients with and without rash. The estimated glomerular filtration rate (eGFR) was slightly lower in older patients without rash (P=0.024); 10.2% of these patients had CKD stage ≥3. In terms of outcomes, patients with and without rash had mortalities of 0.0% and 2.9%, respectively (P=0.97). The length of hospitalisation was similar in both groups (Table 3).

Patients aged ≥70 years had a significantly higher viral load (as reflected by a lower Ct value), compared with those aged <70 years (mean Ct value: 21.97 vs 24.65, P=0.03). Regardless of age, patients with hypertension and CKD stage ≥3 had a significantly higher viral load and lower initial Ct value on admission (OR=2.65, 95% CI=1.08-6.45 and OR=3.65, 95% CI=1.18-11.3, respectively; both P<0.05).

All six patients who died were men; their mean age was 87.0 ± 7.3 years. The rates of hypertension, diabetes mellitus, a glycated haemoglobin level of ≥6.5%, CKD stage ≥3, and higher viral load (ie, lower Ct value on admission) were significantly greater among patients who died than among those who survived (Table 4). Older age, hypertension, and low eGFR were associated with a higher risk of mortality (all P<0.05) [Table 5].

Treatment varied in this cohort because there was no standard of care in the early days of the COVID-19 pandemic. Symptomatic treatment was administered to 53 patients (24.2%); 166 patients (75.8%) received early treatment within the first week of symptom onset, including interferon beta-1b and ribavirin, which were administered based on the results of a triple therapy clinical study.21 Emollients and topical corticosteroids of mild to moderate potency (1% hydrocortisone cream and 0.1% mometasone furoate cream) were prescribed for symptomatic relief.

During follow-up, we observed that urticarial eruption after interferon injection resolved within 10 to 14 days upon completion of treatment. With respect to COVID-19–related skin eruptions, most lesions (maculopapular exanthem, livedo reticularis, and urticaria) were self-limiting and spontaneously resolved without specific treatment; there were no severe sequelae. Two patients with varicella-like lesions had mild post-inflammatory hyperpigmentation without scarring.

Cutaneous manifestations of the COVID-19 pandemic have been gaining increasing attention because they may be useful in the early diagnosis of COVID-19, triage of patients with SARS-CoV-2–positive test results, and risk stratification. There is speculation that the mechanism involves the direct action of SARS-CoV-2 on tissues, the complement/interferon-driven immune response, and the coagulation system; alternatively, it involves nonspecific skin symptoms of systemic viral infection.16 17 22 26 27 28 Although more investigations are needed, it is possible that some symptoms are clinical signs of milder COVID-19, whereas others are indicators of more severe clinical illness.

Our study showed that patients with confirmed COVID-19 could display various cutaneous manifestations. The most common manifestation attributed to COVID-19 was maculopapular exanthem, followed by livedo reticularis. Because most skin lesions were transient and self-limiting, skin biopsy was only performed in one patient. In that 40-year-old female patient, skin biopsy of the left trunk revealed low to moderate numbers of perivascular lymphocytes and histiocytes, as well as sparse eosinophils, in the superficial dermis; focal parakeratosis was present in the epidermis. There was no evidence of vasculitis or interfacial changes. These findings were compatible with maculopapular exanthem.

In previous reports, erythema multiforme–like lesions, chilblain-like acral eruptions, and livedo erythema were identified in children and young adult patients with asymptomatic or mild disease.26 28 29 In contrast, maculopapular rash and acro-ischaemic lesions were often observed among adult patients with more severe disease. Among our patients who presented with rash, there were no instances of mortality; the duration of hospitalisation was similar regardless of rash status. The results of a previous study has suggested that the cutaneous manifestation is the only manifestation of COVID-19 in some patients30; thus, careful documentation of any cutaneous symptoms during the COVID-19 pandemic may be necessary for early recognition and diagnosis.30 Additionally, urticaria with fever has diagnostic implications because this combination may be an early symptom of subsequently confirmed SARS-CoV-2 infection.19 In our cohort, patients with cutaneous manifestations were more likely to present with fever. Although most of our patients had symptoms other than rash alone, two patients (0.9%) presented with rash only (one with urticaria and one with maculopapular exanthem); the clinical significance of these symptoms should not be ignored. Informal extrapolation of these results to the general population in Hong Kong suggested that 2477 cases (2/219; ie, 0.91% × 272 235 confirmed cases)4 solely involve rash presentation; these patients would remain undiagnosed if they did not undergo SARS-CoV-2 testing. This lack of diagnosis is a potential health threat and could facilitate viral spread.

In our cohort, the incidence of new rash was 13.6%. In the study by Guan et al31 in China, the prevalence of rash was much lower in patients with COVID-19 (0.2%; 2/1099). In that study, patients with rash may have been underdiagnosed because patients with suspected COVID-19 were managed by general practitioners or hospitalists who had less familiarity with cutaneous manifestations.31 In contrast, our patients underwent prompt assessment by in-hospital dermatologists to detect cutaneous manifestations. In an Italian study, the prevalence of rash presentation was much higher (20.4%),13 presumably because asymptomatic patients were excluded through a lack of testing. However, if we exclude the 58 asymptomatic patients in our cohort (all of whom underwent compulsory testing in accordance with the Public Health Ordinance), the incidence of new rash in our study was 16.7% (95% CI=14.5-18.8), which remains lower than the incidence in the Italian study. We speculate that this difference is related to the early initiation of combined treatment (ribavirin and interferon beta-1b) in our cohort, which may modify or halt the SARS-CoV-2–induced inflammatory process.21 Importantly, the genomic characteristics of SARS-CoV-2 spread are under investigation worldwide; this approach helps identify transmission routes in various regions. In a case series in the United States, SARS-CoV-2 genomes in one region were predominantly associated with isolates that originated in Europe (>80%), similar to the distributions of viral strains in other regions in the United States32; a smaller subgroup of SARS-CoV-2 genomes displayed similarity to strains that originated in Asia (15%), indicating multiple sources of viral spread within the community.32 Differences in the prevalences of cutaneous manifestations may represent variations in SARS-CoV-2 genomic characteristics among regions; in Hong Kong, a cosmopolitan city with many travellers from mainland China and other countries, the prevalences of cutaneous manifestations may be the result of viral strains from all provinces of China as well as Europe and other regions. Further studies are needed concerning genomic variations and clinical manifestations.

In terms of viral load and prognosis, higher viral load on admission was significantly associated with greater mortality in patients with older age, history of hypertension, and CKD stage ≥3. Univariate analysis showed that the risk of mortality was the greatest among patients with older age, hypertension, higher glycated haemoglobin level, and renal impairment. Multivariate Cox regression analysis confirmed that older age, hypertension, and low eGFR were significantly associated with greater mortality risk.

Conversely, patients with renal impairment were less likely to present with rash, suggesting that the immune response is weaker in patients with renal impairment. However, the length of hospitalisation was similar regardless of cutaneous manifestations; the presence of cutaneous manifestations was not associated with other co-morbidities. There was no clear association between Ct values and rash occurrence. Additional studies with larger sample sizes may be necessary to explore the relationship between rash subtype and viral load.

The results of a previous study suggested that cutaneous manifestations of COVID-19 were related to immunological responses rather than the direct results of viral invasion17; cutaneous manifestations may be an early sign of immunological responses elsewhere in the body, similar to pulmonary infiltrates secondary to cytokine storm. The present study showed that the incidence of pulmonary infiltrates was considerably higher among patients with rash (28.6%) than among those without (7.3%) [Table 3]; conversely, patients with rash were less likely to display further deterioration, such as oxygen desaturation and a requirement for oxygen supplementation (P=0.016). Only one patient with rash (25%) received dexamethasone, whereas multiple patients without rash required such treatment (73.3%) [P=0.11]. Another explanation is that, overall, patients with rash tended to seek medical attention earlier than those without, which would increase the likelihood of prompt treatment. A previous study has indicated that patients with cutaneous manifestations may have a better prognosis because those patients develop a more robust immune response.17

In patients with new pulmonary infiltrates as well as evidence of respiratory decompensation/failure (eg, desaturation and/or tachypnoea), systemic corticosteroids have been used to prevent tissue destruction from cytokine storm after other causes had been ruled out. In this context, patients receiving systemic corticosteroids had more severe disease that involved evidence or features of respiratory decompensation and carried a greater risk of mortality.

In the present study, after the exclusion of patients with nosocomial/secondary bacterial pneumonia, heart failure, or pulmonary changes related to prior disease, 19 patients (8.7%) had new pulmonary infiltrates on admission. All 19 patients received interferon beta-1b and ribavirin treatment; 12 patients received dexamethasone (daily dosage range, 6-8 mg; mean duration, 8.63 ± 2.53 days) [Table 4]. Among the 12 patients receiving dexamethasone, five patients (41.7%) died despite the use of systemic corticosteroids, together with empirical antibiotics, interferon beta-1b, and ribavirin; in contrast, only one death (14.3%) occurred among seven patients receiving interferon beta-1b and ribavirin without corticosteroids (OR=4.28, 95% CI=0.38-47.6; P=0.23). The mean interval from symptom onset to systemic corticosteroid initiation was shorter among patients who recovered than among those who died (5.14 ± 2.14 days vs 8.61 ± 2.30 days; P=0.0026). These results suggest that the early use of systemic corticosteroids may lead to a better survival outcome.

Although no deaths occurred among patients with cutaneous manifestations, the mortality rate did not significantly differ from the rate of 2.9% among patients without rash. Most patients received treatment within the first week after diagnosis of COVID-19 (according to detection of SARS-CoV-2–specific immunoglobulin G within 14 days after symptom onset; mean, 7.71 ± 3.05 days; range, 4-13), which may have improved disease outcomes and shortened hospitalisation. These findings highlighted the importance of early treatment beginning at symptom onset (ie, in the first week) and supported the use of interferon therapy described in a previous report.21

First, this study had a small number of patients. Second, there was potential selection bias because only hospitalised patients with SARS-CoV-2–positive test results were included in the analysis; patients with COVID-19 who did not undergo screening or seek medical consultation were not diagnosed, and thus they were excluded from the study. Third, Ct value analysis was not conducted according to rash subtype and severity because of the limited number of patients. Fourth, some viral laboratory tests (eg, test for human herpesvirus 6) were not routinely available in our hospital, which may have hindered the interpretation of possible causes of rash or the identification of coexisting infections. Nevertheless, most other possible viral infections were excluded from this study. Additional studies with larger sample sizes and comparisons with treatment outcomes are needed.

This study did not demonstrate direct relationships among rash, viral load, and mortality. Furthermore, cutaneous manifestations may be early signs of immunological responses (similar to pulmonary infiltrates). Patients with older age, hypertension, and renal impairment have greater mortality risk and higher viral load. These high-risk groups should be prioritised in early screening and vaccination efforts to avoid poor clinical outcomes.

Concept or design: CSM Wong, IFN Hung.
Acquisition of data: CSM Wong, MMH Chung, MWM Chan, AKC Cheng, YM Lau.
Analysis or interpretation of data: CSM Wong, IFN Hung. Drafting of the manuscript: CSM Wong, IFN Hung, CK Yeung, HHL Chan.
Critical revision of the manuscript for important intellectual content: CSM Wong, IFN Hung, MYW Kwan, CK Yeung, HHL Chan, CS Lau.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

All authors declare no conflicts of interest.

The authors thank all patients for their participation.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

This research was approved by the Institutional Review Board of The University of Hong Kong/Hospital Authority Hong Kong West Cluster (Ref No.: UW20-725) and was conducted in full compliance with the ICH E6 guideline for Good Clinical Practice and the principles of the Declaration of Helsinki. Appropriate patient consent was obtained for clinical information and images to be publicly reported. All participants’ clinical data and reports were deidentified to maintain anonymity.

1. Chen N, Zhou M, Dong X, et al. Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study. Lancet 2020;395:507-13. Crossref

2. Zhu N, Zhang D, Wang W, et al. A novel coronavirus from patients with pneumonia in China, 2019. N Engl J Med 2020;382:727-33. Crossref

3. World Health Organization. WHO coronavirus (COVID-19) dashboard. Situation by region, country, territory & area. Available from: https://covid19.who.int/table. Accessed 7 May 2022.

4. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Latest situation of coronavirus disease (COVID-19) dashboard in Hong Kong. Available from: https://chp-dashboard.geodata.gov.hk/covid-19/en.html. Accessed 7 May 2022.

5. Ye Q, Wang B, Mao J. The pathogenesis and treatment of the ‘cytokine storm’ in COVID-19. J Infect 2020;80:607-13. Crossref

6. Joob B, Wiwanitkit V. COVID-19 can present with a rash and be mistaken for dengue. J Am Acad Dermatol 2020;82:e177. Crossref

7. Hassan K. Urticaria and angioedema as a prodromal cutaneous manifestation of SARS-CoV-2 (COVID-19) infection. BMJ Case Rep 2020;13:e236981. Crossref

8. Andina D, Noguera-Morel L, Bascuas-Arribas M, et al. Chilblains in children in the setting of COVID-19 pandemic. Pediatr Dermatol 2020;37:406-11. Crossref

9. Marzano AV, Genovese G, Fabbrocini G, et al. Varicella-like exanthem as a specific COVID-19-associated skin manifestation: multicenter case series of 22 patients. J Am Acad Dermatol 2020;83:280-5. Crossref

10. De Giorgi V, Recalcati S, Jia Z, et al. Cutaneous manifestations related to coronavirus disease 2019 (COVID-19): a prospective study from China and Italy. J Am Acad Dermatol 2020;83:674-5. Crossref

11. Tammaro A, Adebanjo GA, Parisella FR, Pezzuto A, Rello J. Cutaneous manifestations in COVID-19: the experiences of Barcelona and Rome. J Eur Acad Dermatol Venereol 2020;34:e306-7. Crossref

12. Marraha F, Al Faker I, Gallouj S. A review of the dermatological manifestations of coronavirus disease 2019 (COVID-19). Dermatol Res Pract 2020;2020:9360476. Crossref

13. Recalcati S. Cutaneous manifestations in COVID-19: a first perspective. J Eur Acad Dermatol Venereol 2020;34:e212-3. Crossref

14. Galván Casas C, Català A, Carretero Hernández G, et al. Classification of the cutaneous manifestations of COVID-19: a rapid prospective nationwide consensus study in Spain with 375 cases. Br J Dermatol 2020;183:71-7. Crossref

15. Zhao Q, Fang X, Pang Z, Zhang B, Liu H, Zhang F. COVID-19 and cutaneous manifestations: a systematic review. J Eur Acad Dermatol Venereol 2020;34:2505-10. Crossref

16. Hamming I, Timens W, Bulthuis ML, Lely AT, Navis G, van Goor H. Tissue distribution of ACE2 protein, the functional receptor for SARS coronavirus. A first step in understanding SARS pathogenesis. J Pathol 2004;203:631-7. Crossref

17. Novak N, Peng W, Naegeli MC, et al. SARS-CoV-2, COVID-19, skin and immunology—what do we know so far? Allergy 2021;76:698-713. Crossref

18. Li MY, Li L, Zhang Y, Wang XS. Expression of the SARS-CoV-2 cell receptor gene ACE2 in a wide variety of human tissues. Infect Dis Poverty 2020;9:45. Crossref

19. Rahimi H, Tehranchinia Z. A comprehensive review of cutaneous manifestations associated with COVID-19. Biomed Res Int 2020;2020:1236520. Crossref

20. Dadras O, Afsahi AM, Pashaei Z, et al. The relationship between COVID-19 viral load and disease severity: a systematic review. Immun Inflamm Dis 2022;10:e580. Crossref

21. Hung IF, Lung KC, Tso EY, et al. Triple combination of interferon beta-1b, lopinavir-ritonavir, and ribavirin in the treatment of patients admitted to hospital with COVID-19: an open-label, randomised, phase 2 trial. Lancet 2020;395:1695-704. Crossref

22. Wollina U, Karadağ AS, Rowland-Payne C, Chiriac A, Lotti T. Cutaneous signs in COVID-19 patients: a review. Dermatol Ther 2020;33:e13549. Crossref

23. Infectious Diseases Society of America; Association for Molecular Pathology. IDSA and AMP joint statement on the use of SARS-CoV-2 PCR cycle threshold (Ct) values for clinical decision-making. Available from: https://www.idsociety.org/globalassets/idsa/public-health/covid-19/idsa-amp-statement.pdf. Accessed 12 Apr 2022.

24. Aranha C, Patel V, Bhor V, Gogoi D. Cycle threshold values in RT-PCR to determine dynamics of SARS-CoV-2 viral load: an approach to reduce the isolation period for COVID-19 patients. J Med Virol 2021;93:6794-7. Crossref

25. Cevik M, Tate M, Lloyd O, Maraolo AE, Schafers J, Ho A. SARS-CoV-2, SARS-CoV, and MERS-CoV viral load dynamics, duration of viral shedding, and infectiousness: a systematic review and meta-analysis. Lancet Microbe 2021;2:e13-22. Crossref

26. Genovese G, Moltrasio C, Berti E, Marzano AV. Skin manifestations associated with COVID-19: current knowledge and future perspectives. Dermatology 2021;237:1-12. Crossref

27. Bastard P, Rosen LB, Zhang Q, et al. Autoantibodies against type I IFNs in patients with life-threatening COVID-19. Science 2020;370:eabd4585. Crossref

28. Chua GT, Wong JS, Lam I, et al. Clinical characteristics and transmission of COVID-19 in children and youths during 3 waves of outbreaks in Hong Kong. JAMA Netw Open 2021;4:e218824. Crossref

29. Chua GT, Wong JS, Chung J, et al. Paediatric multisystem inflammatory syndrome temporally associated with SARS-CoV-2: a case report. Hong Kong Med J 2022;28:76-8. Crossref

30. Leung TY, Chan AY, Chan EW, et al. Short- and potential long-term adverse health outcomes of COVID-19: a rapid review. Emerg Microbes Infect 2020;9:2190-9. Crossref

31. Guan WJ, Ni ZY, Hu Y, et al. Clinical characteristics of coronavirus disease 2019 in China. N Engl J Med 2020;382:1708-20. Crossref

32. Zhang W, Govindavari JP, Davis BD, et al. Analysis of genomic characteristics and transmission routes of patients with confirmed SARS-CoV-2 in southern California during the early stage of the US COVID-19 pandemic. JAMA Netw Open 2020;3:e2024191. Crossref

In the early days of the coronavirus disease 2019 (COVID-19) pandemic, telemedicine was recommended as a solution to provide safe access to healthcare.1 In 2020, the World Health Organization reported that most global health authorities regarded telemedicine as a potential method to provide services for patients with non-communicable diseases.2 For example, in a national survey of healthcare providers in Germany, approximately 60% of participants reported routine or partial use of telemedicine during the COVID-19 pandemic.3 In the US in 2020, telemedicine was used for COVID-19 screening, monitoring of patients with positive COVID-19 test results, management of chronic diseases, and virtual monitoring and follow-up.4 Telemedicine reduces patient travel costs and improves access, while reducing the use of personal protective equipment and the risk of COVID-19 transmission. In China, ‘internet hospitals’ offered essential medical support to the public during the early days of the COVID-19 pandemic.5

However, from a patient perspective, there are limitations and barriers to the use of telemedicine. A qualitative study conducted in Hong Kong in 20166 revealed that patient concerns included technical and logistical issues (eg, difficulty in accessing and using computers), limited personal interactions (eg, lack of in-person physical examinations and risk of incorrect diagnosis related to poor communication), concerns regarding cybersecurity and safety, and problems with prescriptions (eg, distrust of local community pharmacies). In 2020, clinical guidelines were published concerning the performance of remote primary care assessments and treatments for patients with suspected COVID-19 in the United Kingdom.7 During the COVID-19 pandemic, a digital health ecosystem may benefit the healthcare system, as well as the broader population (eg, through tracking and communication strategies) and research and health technology sectors (eg, online activity monitoring and digital support for isolation and quarantine situations).8

Telemedicine consists of remote healthcare service delivery by healthcare professionals for diagnosis, treatment, and prevention efforts, as well as research and continuing education.9 The fundamental goal of telemedicine is to increase access to care, thereby serving populations that otherwise would not receive timely medical evaluation and treatment. Research concerning telemedicine effectiveness has been controversial.10 11 12 13 14 A previous systematic review found that many studies showed no difference between telemedicine and usual care, and there remains limited evidence concerning the cost-effectiveness of telemedicine.10 With respect to chronic disease management via telemedicine, another review found publication bias and a tendency to report only short-term outcomes.11 Moreover, a Cochrane review revealed improved control of blood glucose among patients with diabetes using telemedicine, compared with patients receiving in-person care; conversely, the care modality did not affect health outcomes among patients with heart failure.12 Regarding the impacts and costs of telemedicine services, economic analyses have been problematic because of complex and unpredictable collaborative achievement processes13; generalisability has been limited by poor quality and low reporting standards.14 There is a need to focus on patient perspectives and telemedicine innovations.

In 2020, the health system in Hong Kong was considerably impacted by COVID-19 transmission risk and public health responses.15 During the early days of the COVID-19 pandemic, the number of in-person medical consultations decreased worldwide.16 17 In the Netherlands, the use of telemedicine offset this decrease.17 Before the COVID-19 pandemic, telemedicine was not widely used in Hong Kong6 18; user perspectives concerning the role of telemedicine during the COVID-19 pandemic and future pandemics require investigation. This study of telemedicine users in Hong Kong explored their perceptions of telemedicine use during the COVID-19 pandemic, their perceptions of disease risk, and their COVID-19 preparedness measures. We hypothesised that telemedicine users have distinct perceptions of risk, compared with the general public; we sought to determine the effects of such differences on the delivery of health services.

We conducted a cross-sectional online survey from 6 April to 11 May 2020. Participants in this study were users of COVID-19 testing services provided by The Chinese University of Hong Kong (CUHK) Medical Clinic,19 a multispecialty clinic offering specialist consultations, health screenings, vaccinations, and COVID-19 testing services.

All service users completed a telemedicine consultation, followed by COVID-19 testing. The consultation, delivered using a standard telephone, focused on assessment of the user’s COVID-19 risk prior to the time of testing. Users were offered a deep throat saliva test for COVID-19 detection. After the telemedicine consultation and COVID-19 test, all users were invited to complete an anonymous online survey within 1 month for service evaluation and data collection.

The online survey consisted of 24 questions regarding telemedicine services, COVID-19 risk perception, and preparedness measures in Hong Kong. The survey specifically focused on reasons for using CUHK telemedicine and testing services, user perceptions and attitudes regarding telemedicine consultations, user perceptions and attitudes regarding the COVID-19 pandemic and prevention, and user characteristics (eg, age and sex). Questions about user perceptions and attitudes regarding the current COVID-19 pandemic and prevention were phrased in a manner identical to a previous study,20 allowing direct comparison with survey results from a study focused on the Hong Kong general public. The survey was pilot tested and subsequently modified to ensure content validity. Most questions included ‘yes’ or ‘no’ answers and a 5-point Likert scale. Written consent to take part in the study was obtained from all participants before they began the survey.

Descriptive statistics were reported for participant characteristics, perceptions, and attitudes regarding telemedicine consultations. Perceptions and attitudes regarding the COVID-19 pandemic and prevention were recorded in a manner that allowed comparison with the results of a prior COVID-19–focused telephone survey of the Hong Kong general public.20 The prior telephone survey—a cross-sectional, population-based landline telephone survey using a computerised random-digit dialling method—included 765 adult Hong Kong residents during the period from 22 March to 1 April 2020. The participants in that study were representative of Hong Kong Census population data with respect to age, sex, marital status, and residential district, although they had higher levels of education and household income.20 Chi squared tests were used for comparisons between the prior survey and the present survey. The threshold of statistical significance was set at 0.05. All statistical analyses were conducted using SPSS (Windows version 21.0; IBM Corp, Armonk [NY], US).

Among 187 telemedicine users of COVID-19 testing services at the CUHK Medical Clinic, we identified 150 responses to the online survey during the period from 6 April to 11 May 2020. In total, 141 telemedicine users (response rate of 75.4%) were willing to participate in this study. Table 1 shows the participants’ characteristics. Notably, 56.1% were men, over half (50.7%) were aged 18 to 24 years (all participants were aged <65 years), most (59.4%) lived in private housing, and most (65.4%) resided in the New Territories. The most common reason for COVID-19 testing was a work-related requirement (56.7%, 80/141), followed by recent international travel (39.7%, 56/141); 14.9% (21/141) of the participants sought testing because of concerns about the spread of COVID-19. Overall, 14.2% (20/141) of the participants had either been in close contact with a confirmed case or suspected that they had symptoms of COVID-19.

In total, 95.1% (116/122) participants believed that COVID-19 telemedicine consultations were useful. Most participants believed that telemedicine consultations could perform the functions of ‘health protection, promotion and disease prevention’ (73.6%) and ‘diagnosis’ (64.0%) [Fig 1]. Concerning the choice of telemedicine provider, almost all participants (99.2%) would accept medical doctors; more than half of the participants (54.1%) would accept trained nurses, but only 13.1% would accept non-clinical staff who had been trained to provide telemedicine services.

Nearly half of the participants (49.0%) agreed or strongly agreed that telemedicine consultations were appropriate during the COVID-19 pandemic (Fig 2). More than half (61.0%) of the participants reported satisfaction with telemedicine consultations (‘agree’ or ‘strongly agree’) and services provided by clinic staff (73.8% responded ‘agree’ or ‘strongly agree’). However, only 36.2% agreed or strongly agreed that service quality was identical between telemedicine consultations and in-person consultations.

Household capacities for potential COVID-19—related quarantine were compared between telemedicine users in this study and respondents in the prior Hong Kong general public telephone survey20 (Table 2). Telemedicine users reported having less space at home, fewer masks, fewer gloves, and less detergent for potential quarantine situations, compared with respondents in the Hong Kong general public telephone survey (all P<0.05).

In response to the question ‘How many designated caregivers are appropriate for each isolated/quarantined person?’, telemedicine users were less likely to answer correctly (limit to one main carer, 8.1%), compared with respondents in the Hong Kong general public telephone survey (51.3%)20 [P<0.001, Chi squared with continuity correction].

Most telemedicine users (76.4%) believed that household prevention could prevent COVID-19; approximately 67% of them believed that they had sufficient knowledge to manage COVID-19—related health risks. These percentages were higher than the percentages in the Hong Kong general public telephone survey20 (Table 3). Additionally, >80% of telemedicine users believed that Hong Kong had achieved better control of COVID-19, compared with other major cities.

Regarding their main channel for infectious disease information, approximately 64% of participants were using the internet or mobile applications, whereas 24% were using television; these percentages differed from the Hong Kong general public telephone survey, in which 56.5% of respondents used the internet or mobile applications and 36.2% used television20 (P<0.001). Overall, telemedicine users preferred their main channel for infectious disease information to be the internet or mobile applications and personal sources (eg, family, friends, or healthcare professionals).

This study demonstrated high satisfaction with telemedicine consultations among users and revealed that users considered telemedicine to be appropriate during the COVID-19 pandemic. Despite the perception that telemedicine users have sufficient knowledge to manage health risks from COVID-19, when responses were compared between telemedicine users and the Hong Kong general public, we found that household preventative measures were inadequate among telemedicine users.

Most of our telemedicine users did not agree that quality was identical between telemedicine consultations and in-person consultations. This perspective possibly resulted from the provision of telemedicine consultations via telephone without a video component; moreover, the users might not have been familiar with the concept of telemedicine consultation. Indeed, a 2014 nationwide survey in the US revealed that only 15% of family physicians reported using telemedicine in the previous year; barriers included a lack of training, a lack of reimbursement, the cost of equipment, and potential liability issues.21 Furthermore, Schwamm22 has described telemedicine as a disruptive technology that may threaten traditional healthcare delivery. Obstacles to the expansion of telemedicine in the US include state-level statutes that require the clinician to be located in the same state as the patient and to have previously completed an in-person consultation with that patient.23 Similar regulatory requirements exist in Hong Kong.24

Various studies in the US have detected increasing uptake of telemedicine, particularly in primary care.25 26 27 28 29 A study of telemedicine users in a large commercially insured population in the US from 2005 to 2017 showed that the mean age was 38.3 years; on average, users of primary care telemedicine were younger than users of telemental healthcare and more likely to reside in urban areas.25 A study regarding Teladoc, one of the largest telemedicine providers in the US, revealed that Teladoc users were younger and less likely to have used healthcare before the introduction of Teladoc; they were also less likely to have a follow-up visit in any setting, compared with patients who visited a clinic for a similar condition.26 These findings are consistent with our observations that most telemedicine users were young; they also suggest that the use of telemedicine services can provide opportunities for healthcare access that are not otherwise available.

Intriguingly, although more telemedicine users agreed that household preparation could prevent COVID-19, they were less likely to believe that their household preventative measures were adequate, compared with respondents in the Hong Kong general public telephone survey.20 This disparity may be attributed to differences in participant characteristics: telemedicine users in this study were younger (51% aged 18-24 years, vs 9% in the Hong Kong general public20), were male (56% vs 47%20), and were living in the New Territories (65% vs 51%20). Telemedicine users were also more likely to use the internet (64% vs 57%20) as the main channel for infectious disease information and to prefer using the internet for such information (62% vs 54%20). These findings have important implications for the use of telemedicine to fill gaps in health promotion and disease prevention. Wu et al30 found that secondary cases from household transmission of COVID-19 were common in Zhuhai, China, and one-third of these secondary cases were asymptomatic. Sufficient household preparation measures are needed to limit the spread of COVID-19.31

Telemedicine in Hong Kong had a ‘late start’ (in 1994); in 1998, Hjelm32 predicted that the telemedicine would be rapidly implemented in Hong Kong. However, in the December 2019 version of the Ethical Guidelines on Practice of Telemedicine by the Medical Council of Hong Kong, it was noted that telemedicine in Hong Kong has not fully developed.24 Despite the obvious convenience benefit and reduced risk of COVID-19 transmission, the guidelines remind practitioners that they remain fully responsible for legal and ethical obligations during telemedicine consultations.24 Furthermore, the guidelines mention that standards of care to protect patients are applicable during in-person and telemedicine consultations; practitioners should familiarise themselves with the World Medical Association Statement on the Ethics of Telemedicine.33

Training and patient assessment guidelines for healthcare practitioners are urgently needed, considering the unique circumstances surrounding the use of telemedicine, such as technology (eg, technical limitations of virtual consultations, including assessments), patient education and informed consent, cybersecurity, and other concerns.6 These guidelines would ensure that the same standards of telemedicine consultations can be implemented, as described by the Medical Council of Hong Kong.

The present study focused on 141 telemedicine users of a single private clinic providing deep throat saliva polymerase chain reaction tests for detection of severe acute respiratory syndrome coronavirus 2; thus, the study participants may not be representative of all telemedicine users in Hong Kong. Responder bias may have affected the results (the response rate was 75.4%), but it was not possible to compare participants with individuals who refused to participate. Furthermore, the recent experience of a telemedicine consultation may have biased participants’ responses in favour of telemedicine.

Concerning the comparison of COVID-19 risk perception and preparedness measures, methodological discrepancies between the online survey of telemedicine users and the telephone survey of the Hong Kong general public may have led to differences in responses, particularly with respect to including younger and more computer-literate individuals in the online survey. Although the present study was conducted from 6 April to 11 May 2020 (after a surge of COVID-19 cases in Hong Kong), the Hong Kong general public telephone survey used for comparison was conducted from 22 March to 1 April 2020 (during a surge in COVID-19 cases).20 The difference in data collection periods may have contributed to different perceptions of COVID-19 preparedness. Various physical distancing measures were implemented during the COVID-19 surge (from late March to early April), which may also have contributed to differences between the telephone survey and online survey cohorts.

Finally, because of sample size limitations and problems with representativeness, the findings of the study may be restricted to understanding views regarding telemedicine consultations among participants in the present study. However, factors such as young age, residence in private housing, and residence in the New Territories may have contributed to response bias; because no information was collected concerning occupation, education level, income, or ethnicity, we could not control for bias related to these factors.

In this study, telemedicine users in Hong Kong agreed that telemedicine consultations were appropriate during the COVID-19 pandemic. Participants agreed that telemedicine consultations could perform the functions of health protection, promotion, disease prevention, and diagnosis. The use of telemedicine for screening and patient education can be encouraged during the COVID-19 pandemic in Hong Kong.

Concept or design: KKC Hung, EYY Chan, JCY Wu.
Acquisition of data: KKC Hung, JCY Wu, CA Graham.
Analysis or interpretation of data: KKC Hung, EYY Chan, ESK Lo, Z Huang.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: KKC Hung.

All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

All authors have disclosed no conflicts of interest.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

This research was approved by the Survey and Behavioural Research Ethics Committee of The Chinese University of Hong Kong (Ref No.: SBRE-19-730). Patients were treated in accordance with the tenets of the Declaration of Helsinki and provided written informed consent for all treatments and procedures, as well as publication of their anonymised data.

1. Hollander JE, Carr BG. Virtually perfect? Telemedicine for COVID-19. N Engl J Med 2020;382:1679-81. Crossref

2. World Health Organization. Rapid assessment of service delivery for NCDs during the COVID-19 pandemic. Available from: https://www.who.int/publications/m/item/rapid-assessment-of-service-delivery-for-ncds-during-the-covid-19-pandemic. Accessed 10 Aug 2020.

3. Peine A, Paffenholz P, Martin L, Dohmen S, Marx G, Loosen SH. Telemedicine in Germany during the COVID-19 pandemic: multi-professional national survey. J Med Internet Res 2020;22:e19745. Crossref

4. Ford D, Harvey JB, McElligott J, et al. Leveraging health system telehealth and informatics infrastructure to create a continuum of services for COVID-19 screening, testing, and treatment. J Am Med Inform Assoc 2020;27:1871-7. Crossref

5. Gong K, Xu Z, Cai Z, Chen Y, Wang Z. Internet hospitals help prevent and control the epidemic of COVID-19 in China: multicenter user profiling study. J Med Internet Res 2020;22:e18908. Crossref

6. Kung K, Wong HF, Chen JY. An exploratory qualitative study on patients’ views of medical e-consultation in a public primary care setting. Hong Kong Pract 2016;38:120-7.

7. Greenhalgh T, Koh GC, Car J. COVID-19: a remote assessment in primary care. BMJ 2020;368:m1182. Crossref

8. Fagherazzi G, Goetzinger C, Rashid MA, Aguayo GA, Huiart L. Digital health strategies to fight COVID-19 worldwide: challenges, recommendations, and a call for papers. J Med Internet Res 2020;22:e19284. Crossref

9. World Health Organization. A health telematics policy in support of WHO’s Health-for-all strategy for global health development: report of the WHO Group Consultation on Health Telematics, 11-16 December, Geneva, 1997. 1998. Available from: https://apps.who.int/iris/handle/10665/63857. Accessed 7 Sep 2023.

10. McLean S, Sheikh A, Cresswell K, et al. The impact of telehealthcare on the quality and safety of care: a systematic overview. PLoS One 2013;8:e71238. Crossref

11. Wootton R. Twenty years of telemedicine in chronic disease management—an evidence synthesis. J Telemed Telecare 2012;18:211-20. Crossref

12. Flodgren G, Rachas A, Farmer AJ, Inzitari M, Shepperd S. Interactive telemedicine: effects on professional practice and health care outcomes. Cochrane Database Syst Rev 2015;2015:CD002098. Crossref

13. Ekeland AG, Bowes A, Flottorp S. Effectiveness of telemedicine: a systematic review of reviews. Int J Med Inform 2010;79:736-71. Crossref

14. Eze ND, Mateus C, Cravo Oliveira Hashiguchi T. Telemedicine in the OECD: an umbrella review of clinical and cost-effectiveness, patient experience and implementation. PLoS One 2020;15:e0237585. Crossref

15. Hung KK, Walline JH, Graham CA. COVID-19: emergency medicine perspectives from Hong Kong. Eur J Emerg Med 2020;27:163-4. Crossref

16. Hung KK, Walline JH, Chan EY, et al. Health service utilization in Hong Kong during the COVID-19 pandemic—a cross-sectional public survey. Int J Health Policy Manag 2022;11:508-13. Crossref

17. Auener S, Kroon D, Wackers E, Dulmen SV, Jeurissen P. COVID-19: a window of opportunity for positive healthcare reforms. Int J Health Policy Manag 2020;9:419-22. Crossref

18. Sheng OR, Hu PJ, Chau PY, et al. A survey of physicians’ acceptance of telemedicine. J Telemed Telecare 1998;4 Suppl 1:100-2. Crossref

19. CUHK Medical Clinic. COVID-19 Testing Service (Company/Groups). Available from: https://cuclinic.hk/programmes/covid-19-testing-service/covid-19-testing-testing-service-company-group. Accessed 30 Jul 2020.

20. Chan EY, Huang Z, Lo ES, Hung KK, Wong EL, Wong SY. Sociodemographic predictors of health risk perception, attitude and behavior practices associated with Health-Emergency Disaster Risk Management for biological hazards: the case of COVID-19 pandemic in Hong Kong, SAR China. Int J Environ Res Public Health 2020;17:3869. Crossref

21. Moore MA, Coffman M, Jetty A, Petterson S, Bazemore A. Only 15% of FPs report using telehealth; training and lack of reimbursement are top barriers. Am Fam Physician 2016;93:101.

22. Schwamm LH. Telehealth: seven strategies to successfully implement disruptive technology and transform health care. Health Aff (Millwood) 2014;33:200-6. Crossref

23. Pearl R. Kaiser Permanente Northern California: current experiences with internet, mobile, and video technologies. Health Aff (Millwood) 2014;33:251-7. Crossref

24. The Medical Council of Hong Kong. Guidelines for all registered medical practitioners. Available from: https://www.mchk.org.hk/files/newsletter-26th.pdf. Accessed 30 Jul 2020.

25. Barnett ML, Ray KN, Souza J, Mehrotra A. Trends in telemedicine use in a large commercially insured population, 2005-2017. JAMA 2018;320:2147-9. Crossref

26. Uscher-Pines L, Mehrotra A. Analysis of Teladoc use seems to indicate expanded access to care for patients without prior connection to a provider. Health Aff (Millwood) 2014;33:258-64. Crossref

27. Neufeld JD, Doarn CR. Telemedicine spending by Medicare: a snapshot from 2012. Telemed J E Health 2015;21:686-93. Crossref

28. Mehrotra, A. The convenience revolution for treatment of low-acuity conditions. JAMA 2013;310:35-6. Crossref

29. Courneya PT, Palattao KJ, Gallagher JM. HealthPartners’ online clinic for simple conditions delivers savings of $88 per episode and high patient approval. Health Aff (Millwood) 2013;32:385-92. Crossref

30. Wu J, Huang Y, Tu C, et al. Household transmission of SARS-CoV-2, Zhuhai, China, 2020. Clin Infect Dis 2020;71:2099-108. Crossref

31. Chan EY, Gobat N, Kim JH, et al. Informal home care providers: the forgotten health-care workers during the COVID-19 pandemic. Lancet 2020;395:1957-9. Crossref

32. Hjelm NM. Telemedicine: academic and professional aspects. Hong Kong Med J 1998;4:289-92.

33. World Medical Association. WMA Statement on the Ethics of Telemedicine. Available from: https://www.wma.net/policies-post/wma-statement-on-the-ethics-of-telemedicine/. Accessed 15 Apr 2021.

Because of its close geographical proximity to mainland China, Hong Kong was one of the first regions outside of the Mainland to be affected by the severe acute respiratory syndrome (SARS) and coronavirus disease 2019 (COVID-19) epidemics.1 Since 2020, Hong Kong experienced an initial epidemic wave of imported COVID-19 cases and spillover effects, as well as subsequent epidemic waves of imported COVID-19 cases and associated local transmission.

The COVID-19 epidemic has resulted in millions of deaths worldwide. By late 2020, there had been multiple country-level analyses in other regions; however, there were limited data concerning outcomes among patients with severe or critical COVID-19 in Hong Kong.2 This study analysed 28-day mortality in these patients and explored risk factors for mortality among them during the first several months of the COVID-19 pandemic.

This retrospective multi-centre cohort study included all adult patients aged ≥18 years with severe or critical COVID-19 who were admitted to the medical wards or intensive care units (ICUs) of three public acute hospitals in Hong Kong from 22 January to 30 September 2020. The three hospitals were Pamela Youde Nethersole Eastern Hospital, Princess Margaret Hospital, and Ruttonjee and Tang Shiu Kin Hospitals.

Patients were diagnosed with COVID-19 if their respiratory samples contained severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA, according to reverse transcription polymerase chain reaction (RT-PCR) analysis. Respiratory samples included nasopharyngeal aspirate, nasopharyngeal swab, nasopharyngeal aspirate or swab paired with throat swab, or deep throat saliva. Coronavirus disease 2019 grade was considered mild, severe, or critical, in accordance with the guidelines of the Chinese Center for Disease Control and Prevention.3 Severe COVID-19 was characterised by dyspnoea, respiratory rate of ≥30 breaths/minute, blood oxygen saturation <94%, ratio of arterial oxygen partial pressure to fractional inspired oxygen (P/F ratio) <300, and/or a 50% increase in lung infiltrates within 2 days.3 Critical COVID-19 was characterised by respiratory failure, septic shock, and/or multiple organ dysfunction or failure.3

Hong Kong experienced three epidemic waves during the study period.4 The first epidemic wave, from mid-January to early March 2020, occurred after travellers from mainland China arrived in Hong Kong during the Chinese New Year. The second epidemic wave occurred when Hong Kong residents returned from overseas during the Easter Holiday, from mid-March to May 2020. The third epidemic wave extended from early July until late September 2020; disease transmission may have originated among commercial airline crews.4 Cases were classified as imported or local by the Centre for Health Protection within the Department of Health. Medical comorbidities were defined according to the International Classification of Diseases, Tenth Revision. Comorbidities were selected based on the STOP-COVID (Study of the Treatment and Outcomes in Critically Ill Patients With COVID-19) in the US, which analysed critical COVID-19 cases at 65 ICUs.5

All medical records and data from the Clinical Management System of Hospital Authority and Clinical Information System (IntelliVue Clinical Information Portfolio; Philips Medical, Amsterdam, Netherlands) used by the ICUs were retrospectively reviewed. Upper respiratory tract infection (URI)–related symptoms included cough, rhinorrhoea, and sore throat. For patients admitted to the ICU, disease grade on admission was determined using the APACHE IV (Acute Physiology and Chronic Health Evaluation IV) score and the SOFA (Sequential Organ Failure Assessment) score. For patients requiring mechanical ventilatory support, disease grade was determined by the P/F ratio on the day of intubation. Blood test parameters included minimum lymphocyte count, maximum C-reactive protein level, maximum lactate dehydrogenase (LDH) level, and maximum alanine aminotransferase level. Clinical outcome data included the use of oxygen supplementation, mechanical ventilation, vasopressor or inotrope, renal replacement therapy (RRT), extracorporeal membrane oxygenation, and cardiopulmonary resuscitation, as well as mortality and length of stay in the ICU and hospital. Patients were followed up until death or 31 March 2021, whichever occurred earlier.

The primary outcome was 28-day mortality. Secondary outcomes included length of stay and mortality in the ICU and hospital, duration of oxygen supplementation and mechanical ventilatory support, and time to viral clearance or time to development of antibodies against SARS-CoV-2. Viral clearance was defined as the collection of two consecutive respiratory samples at least 24 hours apart that were both SARS-CoV-2–negative, according to RT-PCR analysis. Cycle threshold (Ct) value indicated the number of RT-PCR cycles required to amplify the viral RNA to a detectable level; this value was inversely related to viral load.6 Minimum Ct values were recorded to determine maximum viral load. Each patient’s blood was collected and qualitatively tested for antibodies against SARS-CoV-2 (ie, antibodies to SARS-CoV-2 nucleoprotein) using the Abbott SARS-CoV-2 Immunoglobulin G assay. Patients were generally released from isolation when they met the requirement for viral clearance or when they displayed serum antibodies, in accordance with recommendations from the Scientific Committee on Emerging and Zoonotic Diseases under the Centre for Health Protection within the Department of Health.7

The clinical characteristics and outcomes of patients with severe or critical COVID-19 were compared between 28-day survivors and non-survivors. Subgroup analysis was performed among patients who received ICU care between 28-day survivors and non-survivors; it was also performed among patients whose laboratory reports provided information regarding viral load (ie, Ct values) in respiratory samples. The frequencies of these characteristics and outcomes were expressed as medians ± interquartile ranges (IQRs) or as numbers of patients and corresponding percentages.

Based on the population of patients with COVID-19 (n=5080) in Hong Kong on 29 September 2020, where 351 patients had ever displayed serious or critical disease, the prevalence of outcome factors in this patient population was 6.9%. Using a confidence limit of 5% and a design effect of 2, the sample sizes required to achieve statistical powers of 80% and 90% were 84 patients and 138 patients, respectively.

For univariate analyses, the Pearson Chi squared test or Fisher’s exact test was used to compare categorical variables; the Mann-Whitney U test was used to compare continuous variables. Variables with a P value of <0.1 in univariate analyses were included in subsequent multivariable analyses. Independent predictors of 28-day mortality were assessed by logistic regression analysis using a forward stepwise approach. Considering the potential for unstable or extreme estimates because of the small sample sizes in subgroup analyses, logistic regression was not performed. All statistical analyses were performed using SPSS (Windows version 23.0; IBM Corp, Armonk [NY], US).

From 22 January to 30 September 2020, 1312 adult patients with COVID-19 were admitted to Princess Margaret Hospital, Pamela Youde Nethersole Eastern Hospital, and Ruttonjee and Tang Shiu Kin Hospitals. In total, 125 patients had severe or critical COVID-19.

The median age was 64 years (IQR=57-75); 69.6% of the patients were men and 93.6% were Chinese (Table 1). In total, 68.8% of the patients were admitted in the third epidemic wave and 86.4% acquired COVID-19 in Hong Kong. Almost half of the patients had hypertension or obesity (48.0% and 54.4%, respectively); 27.2% of patients had diabetes mellitus, and 21.6% had ever smoked. The median duration of symptoms before respiratory samples were confirmed SARS-CoV-2–positive was 3 days. The most common presenting symptom was fever (80.0%), followed by URI-related symptoms (64.8%). Overall, 5.6% of the patients were asymptomatic before their positive test result.

As shown in Table 1, 60% to 80% of patients received lopinavir-ritonavir, ribavirin, interferon beta-1b, or corticosteroids. Nearly half of the patients (48.8%) received anticoagulation treatment; <20% received remdesivir, tocilizumab, or convalescent plasma. More than 80% of the patients received antibiotics during hospitalisation. In total, 45.6% of the patients received ICU care. One patient received non-invasive ventilation, 41 patients (32.8%) received mechanical ventilation, and 11 patients (8.8%) received prone ventilation. Only one patient required extracorporeal membrane oxygenation. Approximately one-fourth of the patients required vasopressor support; 14 patients (11.2%) received acute RRT.

Fifteen patients (12%) died within 28 days (Table 1). Four additional patients died during hospitalisation; thus, the overall hospital mortality rate among patients with severe or critical COVID-19 was 15.2%. The median hospital length of stay was 21.8 days (IQR=15-31.8) and the median duration of oxygen supplementation was 7 days (IQR=4-12.5).

Twenty-eight–day non-survivors were older than 28-day survivors (84 years [IQR=71-86] vs 62 years [IQR=56-71.3]; P<0.001) and more frequently had a history of ischaemic heart disease (26.7% vs 5.5%; P=0.019) or stroke (26.7% vs 4.5%; P=0.012). Moreover, non-survivors had a shorter duration of symptoms before RT-PCR confirmation of SARS-CoV-2 positivity in respiratory samples (1 day [IQR=1-2] vs 3 days [IQR=1-6]; P=0.014); fewer non-survivors displayed URI-related symptoms (33.3% vs 69.1%; P=0.010). Non-survivors had a higher maximum C-reactive protein level (209 mg/L [IQR=82-248] vs 103 mg/L [IQR=63.6-153.5]; P=0.031); they more frequently received mechanical ventilation (66.7% vs 28.2%; P=0.006), acute RRT (40.0% vs 7.3%; P=0.002), and vasopressor support (66.7% vs 19.1%; P<0.001). Finally, non-survivors received a shorter duration of lopinavir-ritonavir treatment (1 day [IQR=0-2] vs 6 days [IQR=0.75-11]; P=0.007) [Table 1].

Logistic regression analysis revealed that age (odds ratio [OR] per 1-year increase in age=1.12, 95% confidence interval [CI]=1.04-1.21; P=0.002), history of stroke (OR=15.96, 95% CI=1.65-154.66; P=0.017), use of acute RRT (OR=15.32, 95% CI=2.67-87.83; P=0.002), and lopinavir-ritonavir duration (OR per 1-day increase=0.82, 95% CI=0.68-0.98; P=0.034) were independent predictors of 28-day mortality (Table 2).

Among the 57 patients admitted to the ICU, nine died; the ICU mortality rate was 15.8% (Table 3). The median ICU length of stay was 9.6 days (IQR=4.6-14.9). Univariate analysis demonstrated that 28-day non-survivors were older; more frequently had a history of ischaemic heart disease or stroke; had a shorter duration of symptoms; less frequently presented with URI-related symptoms; more frequently received mechanical ventilation, acute RRT, and vasopressor support; and received a shorter course of lopinavir-ritonavir treatment. Other significant differences between non-survivors and survivors were the minimum lymphocyte count (0.32 × 10⁹/L [IQR=0.17-0.4] vs 0.4 × 10⁹/L [IQR=0.3-0.6]; P=0.042), maximum LDH level (706 U/L [IQR=492.2-5255.5] vs 474.5 U/L [IQR=406.3-616]; P=0.043), anticoagulation duration (8 days [IQR=3- 10] vs 12.5 days [IQR=7.5-26.8]; P=0.045), APACHE IV score upon ICU admission (83 [IQR=64.5-98] vs 54.5 [IQR=46-61.8]; P<0.001), and SOFA score upon ICU admission (10 [IQR=5.5-13] vs 4 [IQR=3-5.8]; P=0.001).

The median P/F ratio on the day of intubation was 94.1 (IQR=81.9-117.7) for patients with COVID-19 requiring mechanical ventilation, and the median duration of ventilation for all ICU patients with COVID-19 was 8.5 days (IQR=5-18); these values were similar between survivors and non-survivors (Table 3).

As shown in Table 4, 28-day non-survivors were older and more frequently had a history of ischaemic heart disease or stroke; fewer of them had URI-related symptoms (36.4% vs 68.8%; P=0.010). Non-survivors more frequently received mechanical ventilation (81.8% vs 29.9%; P=0.001), acute RRT (45.5% vs 7.8%; P=0.004), and vasopressor support (81.8% vs 23.4%; P<0.001); they received a shorter course of lopinavir-ritonavir treatment (1 day [IQR=0-2] vs 6 days [IQR=0.75-11]; P=0.010). Additionally, non-survivors had a lower minimum lymphocyte count (0.37 × 10⁹/L [IQR=0.16-0.4] vs 0.49 × 10⁹/L [IQR=0.3-0.7], P=0.041) and lower minimum Ct value (15.1 [IQR=12.6-18.5] vs 19 [IQR=16.4-22.9]; P=0.004).

Among the 125 patients with severe or critical COVID-19, 38 underwent regular monitoring of viral load via RT-PCR analysis of respiratory samples for SARS-CoV-2 RNA; viral clearance was achieved within a median of 26 days (IQR=19-35). Among the 79 patients with access to antibody testing, a median of 14 days (IQR=10-18) was elapsed between the first positive RT-PCR result and the emergence of antibodies against SARS-CoV-2.

In this cohort of Hong Kong patients with severe or critical COVID-19, the 28-day mortality rate was 12.0%; it was 15.8% among such patients who were admitted to the ICU. In 2020, higher mortality rates were observed among cohorts in the US (35.4% in the STOP-COVID cohort5), Italy (ICU and hospital mortality rates of 48.8% and 53.4%, respectively8), and China (28-day mortality of 38.7% among severely and critically ill patients9). Patient baseline characteristics were similar across the cohorts—the median age was 60 to 70 years and the most common comorbidity was hypertension (40%-50%).5 8 9 The proportion of patients requiring mechanical ventilation varied across cohorts, ranging from 67% to 87% in the US5 and Italian8 cohorts, whereas it was 30% in the Chinese cohort.9 Overall, 70.2% of patients in our ICU subgroup received mechanical ventilation, which is comparable with the percentages in the US5 and Italy.8 The P/F ratio in our cohort was similar to the ratio in the STOP-COVID (median, 94.1 vs 124),5 reflecting moderate to severe hypoxaemia. Extracorporeal membrane oxygenation was required by <3% of patients in our cohort and the three comparison cohorts. In addition to patient factors, non-patient factors (eg, ICU bed availability and patient-to–hospital staff ratio) may affect quality of care and patient outcomes. Among hospitals of the Department of Veterans Affairs in the US, COVID-19 mortality was significantly greater when ICU demand exceeded 75%.10 Hospitals in the US with fewer ICU beds and nurses per COVID-19 case also had greater mortality.11 The relatively low prevalence, limited local transmission, and better ICU availability in Hong Kong may have contributed to the lower mortality rate observed in this study.

In our cohort, older age, history of stroke, use of acute RRT, and shorter duration of lopinavir-ritonavir treatment were independent predictors of 28-day mortality. Across studies in 2020, older age was commonly identified as a risk factor for COVID-19 mortality.5 8 9 A history of stroke was also identified as a significant risk factor for COVID-19 mortality in a large cohort study in China; higher neutrophil and interleukin-6 levels were observed in patients with a history of stroke, possibly because of a stronger inflammatory response to COVID-19.12 In an American cohort, the hospital mortality rate was 50% among patients who developed acute kidney injury, and the highest mortality risk was present in patients requiring dialysis.13 In 2020, the mechanism of acute kidney injury was speculated to involve direct viral invasion of renal tissue, considering post-mortem findings in China.14 Such injury was closely related to respiratory failure; in another American cohort, the median time from mechanical ventilation to the initiation of acute RRT was 0.3 hour.15

A shorter duration of lopinavir-ritonavir was identified as a risk factor in our cohort. Lopinavir-ritonavir was originally a protease inhibitor for the treatment of human immunodeficiency virus infection.16 In 2020, this repurposed drug was included in treatment recommendations in Hong Kong, where it effectively reduced viral load when used in combination with ribavirin and/or interferon beta-1b16; however, there was no difference in mortality between treatment and control groups as no patient died during the study.16 Two Chinese cohorts did not show significant outcome differences when using lopinavir-ritonavir,17 consistent with interim results from the World Health Organization Solidarity trial.18 Lopinavir-ritonavir treatment has been associated with gastrointestinal upset and liver dysfunction. Our findings may differ from the results of other cohort studies because, in the present study, lopinavir-ritonavir treatment was not administered to patients with severe liver dysfunction or to those who were presumably unable to tolerate side-effects because of their illness. In contrast, up to 80% of the ICU patients in an American cohort received hydroxychloroquine and/or azithromycin,5 despite doubtful efficacy and significant arrhythmia risk with QT prolongation.19 The selection of antiviral treatment may partly explain the differences in mortality among cohorts.

Regarding viral load, univariate analysis in our cohort showed that lower minimum Ct values were associated with greater 28-day mortality. In a study at Cornell in the US,6 the SARS-CoV-2 viral load on admission independently predicted the risks of intubation and mortality. In a Brazilian cohort, Ct values of <25 were associated with greater mortality.21 The implications of Ct values may become clearer if patients with mild disease are analysed together.

Lymphopenia on admission has been associated with worse outcomes in terms of ICU care requirement and mortality,22 presumably because of the cytokine storm phenomenon and the infection of T cells by SARS-CoV-2,19 which infection of T cells by SARS-CoV-2 was confirmed both in vitro and by flow cytometry and immunofluorescence studies.20 In the present study, the minimum lymphocyte count was significantly lower among 28-day non-survivors in both subgroups. Multi-centre COVID-19 studies have shown that an elevated LDH level is associated with a six-fold increase in the likelihood of severe disease and a 16-fold increase in the likelihood of mortality.24 In the present study, although univariate analysis showed that the maximum LDH level was associated with greater 28-day mortality in the overall cohort, it was not an independent predictor in logistic regression; this finding may have been influenced by the sample size.

To our knowledge, this large study was the first investigation in Hong Kong concerning the clinical characteristics and outcomes of patients with severe or critical COVID-19; it included three public acute hospitals and covered three waves of the COVID-19 pandemic in Hong Kong. This study also explored the impacts of viral parameters and treatment modalities on 28-day mortality.

First, this retrospective study in Hong Kong may have been subject to confounding factors and selection bias. The results may not be generalisable to patients with COVID-19 worldwide. Second, this study lacked information concerning viral parameters (eg, specific SARS-CoV-2 strains or mutations). Third, although some studies showed that inborn errors in type 1 interferon immunity and autoantibodies to type 1 interferons were associated with critical COVID-19,25 26 this study did not explore such factors because patient immunity data were unavailable. Fourth, the use of remdesivir was limited; most treatment courses were solely available to patients enrolled in studies by the pharmaceutical company concerned. Finally, other novel treatment options, including antivirals such as nirmatrelvir/ritonavir and molnupiravir, anti-inflammatory agents such as baricitinib and tocilizumab, and neutralising monoclonal antibodies against SARS-CoV-2, were not available or introduced during the study period. Fifth, the sample size of this study did not reach the statistical powers of 90% and may then not be of high enough power.

In this Hong Kong cohort, the 28-day mortality among patients with severe or critical COVID-19 was 12.0%. Age, history of stroke, use of RRT, and shorter course of lopinavir-ritonavir treatment were associated with greater 28-day mortality. In the future, larger studies with a focus on viral and host factors (eg, mutations in SARS-CoV-2 spike genes and interferon-1 immunity status) could improve prognosis prediction.

All authors contributed to the concept or design of the study, acquisition of data, analysis or interpretation of data, drafting of the manuscript, and critical revision of the manuscript for important intellectual content. All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

All authors have disclosed no conflicts of interest.

The authors thank the guidance and support of the following seniors and colleagues: Dr KK Chan from the Department of Medicine of Pamela Youde Nethersole Eastern Hospital, Dr Jenny YY Leung and Dr Alwin WT Yeung from the Department of Medicine and Geriatrics of Ruttonjee and Tang Shiu Kin Hospitals, and Dr Dominic HK So from Intensive Care of Princess Margaret Hospital.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

The study protocol was approved by the Hong Kong East Cluster Research Ethics Committee (Ref No.: HKECREC—2020-115) and the Kowloon West Cluster Research Ethics Committee (Ref No.: KW/EX-21-005 [155-05]). The requirement for written informed patient consent was waived by both Committees due to the retrospective nature of the research.

1. Wong SY, Kwok KO, Chan FK. What can countries learn from Hong Kong’s response to the COVID-19 pandemic? CMAJ 2020;192:E511-5. Crossref

2. Ling L, So C, Shum HP, et al. Critically ill patients with COVID-19 in Hong Kong: a multicentre retrospective observational cohort study. Crit Care Resusc 2020;22:119-25. Crossref

3. Wu Z, McGoogan JM. Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention. JAMA 2020;323:1239-42. Crossref

4. To KK, Chan WM, Ip JD, et al. Unique clusters of severe acute respiratory syndrome coronavirus 2 causing a large coronavirus disease 2019 outbreak in Hong Kong. Clin Infect Dis 2021;73:137-42. Crossref

5. Gupta S, Hayek SS, Wang W, et al. Factors associated with death in critically ill patients with coronavirus disease 2019 in the US. JAMA Intern Med 2020;180:1436-47. Crossref

6. Magleby R, Westblade LF, Trzebucki A, et al. Impact of severe acute respiratory syndrome coronavirus 2 viral load on risk of intubation and mortality among hospitalized patients with coronavirus disease 2019. Clin Infect Dis 2021;73:e4197-205. Crossref

7. Scientific Committee on Emerging and Zoonotic Diseases, Centre for Health Protection, Hong Kong SAR Government. Updated consensus recommendations on criteria for releasing confirmed COVID-19 patients from isolation (July 29, 2020). Available from: https://www.chp.gov.hk/files/pdf/updated_consensus_recommendations_on_criteria_for_releasing_confirmed_covid19_patients_from_isolation29july2020.pdf. Accessed 1 Oct 2020.

8. Grasselli G, Greco M, Zanella A, et al. Risk factors associated with mortality among patients with COVID-19 in intensive care units in Lombardy, Italy. JAMA Intern Med 2020;180:1345-55. Crossref

9. Yang X, Yu Y, Xu J, et al. Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study. Lancet Respir Med 2020;8:475-81. Crossref

10. Bravata DM, Perkins AJ, Myers LJ, et al. Association of intensive care unit patient load and demand with mortality rates in US Department of Veterans Affairs hospitals during the COVID-19 pandemic. JAMA Netw Open 2021;4:e2034266. Crossref

11. Janke AT, Mei H, Rothenberg C, Becher RD, Lin Z, Venkatesh AK. Analysis of hospital resource availability and COVID-19 mortality across the United States. J Hosp Med 2021;16:211-4. Crossref

12. Qin C, Zhou L, Hu Z, et al. Clinical characteristics and outcomes of COVID-19 patients with a history of stroke in Wuhan, China. Stroke 2020;51:2219-23. Crossref

13. Chan L, Chaudhary K, Saha A, et al. AKI in hospitalized patients with COVID-19. J Am Soc Nephrol 2021;32:151-60. Crossref

14. Su H, Yang M, Wan C, et al. Renal histopathological analysis of 26 postmortem findings of patients with COVID-19 in China. Kidney Int 2020;98:219-27. Crossref

15. Hirsch JS, Ng JH, Ross DW, et al. Acute kidney injury in patients hospitalized with COVID-19. Kidney Int 2020;98:209-18. Crossref

16. Hung IF, Lung KC, Tso EY, et al. Triple combination of interferon beta-1b, lopinavir-ritonavir, and ribavirin in the treatment of patients admitted to hospital with COVID-19: an open-label, randomised, phase 2 trial. Lancet 2020;395:1695-704. Crossref

17. Wong CK, Wan EY, Luo S, et al. Clinical outcomes of different therapeutic options for COVID-19 in two Chinese case cohorts: a propensity-score analysis. EClinicalMedicine 2021;32:100743. Crossref

18. WHO Solidarity Trial Consortium; Pan H, Peto R, et al. Repurposed antiviral drugs for COVID-19—interim WHO solidarity trial results. N Engl J Med 2021;384:497-511. Crossref

19. Sanders JM, Monogue ML, Jodlowski TZ, Cutrell JB. Pharmacologic treatments for coronavirus disease 2019 (COVID-19): a review. JAMA 2020;323:1824-36. Crossref

20. Pontelli MC, Castro ÍA, Martins RB, et al. SARS-CoV-2 productively infects primary human immune system cells in vitro and in COVID-19 patients. J Mol Cell Biol 2022;14:mjac021. Crossref

21. Faíco-Filho KS, Passarelli VC, Bellei N. Is higher viral load in SARS-CoV-2 associated with death? Am J Trop Med Hyg 2020;103:2019-21. Crossref

22. Huang I, Pranata R. Lymphopenia in severe coronavirus disease-2019 (COVID-19): systematic review and meta-analysis. J Intensive Care 2020;8:36. Crossref

23. Tavakolpour S, Rakhshandehroo T, Wei EX, Rashidian M. Lymphopenia during the COVID-19 infection: what it shows and what can be learned. Immunol Lett 2020;225:31-2. Crossref

24. Henry BM, Aggarwal G, Wong J, et al. Lactate dehydrogenase levels predict coronavirus disease 2019 (COVID-19) severity and mortality: a pooled analysis. Am J Emerg Med 2020;38:1722-6. Crossref

25. Zhang Q, Bastard P, Liu Z, et al. Inborn errors of type I IFN immunity in patients with life-threatening COVID-19. Science 2020;370:eabd4570. Crossref

26. Bastard P, Rosen LB, Zhang Q, et al. Autoantibodies against type I IFNs in patients with life-threatening COVID-19. Science 2020;370:eabd4585. Crossref