In 2016, there were an estimated 10.4 million new tuberculosis (TB) cases worldwide, of which 1.03 million (10%) were among people living with human immunodeficiency virus (HIV).1 Of the estimated 1.7 million deaths resulting from TB disease, 0.4 million were among people living with HIV.1 The success rate of TB treatment is generally worse for patients with HIV than for those without. Globally, the proportion of patients with TB who died during treatment has been reported as approximately three-fold higher among patients with HIV than those without (11% vs 4%); this difference is greatest in the Western Pacific region (13% vs 2%).1

Identification of risk factors associated with mortality and innovative strategies to manage disease in high-risk groups may contribute to reduction in mortality among patients coinfected with TB and HIV. Lack of drug susceptibility testing, suboptimal initial anti-TB treatment, disseminated TB, and a low CD4 cell count have been reported as prognostic factors for increased TB-related mortality in Eastern Europe.2 3 In some areas of Africa with high HIV prevalence, increased odds of mortality are associated with the following factors: advanced age, pretreatment sputum smear status, history of TB, and lack of antiretroviral therapy (ART).4 5 6 However, data regarding risk factors for mortality in the Western Pacific region are scarce.7 8 In particular, there have been few assessments in areas with intermediate burden of TB in the Western Pacific region.

Hong Kong has been classified by the World Health Organization as a place with intermediate TB burden and good infrastructure. The TB notification rate in Hong Kong was 60.5 per 100 000 population in 2015, of which 40.2% were patients aged >65 years.9 The overall TB mortality rate was 2.3 per 100 000 population in the same year. The prevalence of HIV infection in the general population in Hong Kong is low (<0.1%).10 Notably, patients with HIV-associated TB constituted approximately 1% of all TB notifications in Hong Kong.9 Treatment outcomes of a cohort of patients reported to the territory-wide TB-HIV Registry of the Hong Kong Department of Health, as of 31 December 2009, have been reported previously.11 The aim of the present study was to assess the 1-year mortality rate among patients with HIV-associated TB and to identify risk factors associated with mortality. Therefore, we retrospectively reviewed the data of patients reported to the TB-HIV Registry between 1 January 2006 and 31 December 2015. The results of this study are expected to guide targeted measures to improve survival and to inform TB control policies in areas with intermediate TB burden, such in Hong Kong and throughout the Western Pacific region.

Data regarding sex, age, ethnicity, case category, history of drug addiction, site of TB infection, sputum smear and culture results, CD4 cell count at TB diagnosis, and ART usage of consecutive patients reported to the TB-HIV Registry between 1 January 2006 and 31 December 2015 were retrieved and retrospectively reviewed. Tuberculosis treatment outcomes at 12 months after initiation of treatment, and mortality data such as the date and cause(s) of death, were obtained from the database. Further mortality data were obtained by cross-matching with the statutory death registry by using patients’ identity card numbers/passport numbers as unique identifiers. For patients who died within 12 months during anti-TB treatment, relevant clinical records from chest clinics and hospitals were traced and reviewed.

All the data collected were imported into Epi-Info software, and exported into SPSS (Windows version 16.0; SPSS Inc, Chicago [IL], US) for analysis. For the identification of variables potentially associated with mortality, comparisons of survival status were made by using Kaplan-Meier estimates. A two-tailed P value <0.05 was considered statistically significant. An evaluation of the effects of covariates on mortality was performed by using Cox proportional hazards regression modelling. Variables that were significant in the univariate analysis or were of clinical significance were included in the model. Patients who had been transferred out of Hong Kong before completion of treatment were censored at the date they were last known to be alive, based on the date of their most recent contact. For patients who were lost to follow-up, but were not deceased according to the statutory death registry, the censor date was 12 months. Kaplan-Meier survival curves were generated to demonstrate differences in mortality, stratified for a variety of covariates.

Of 337 patients reported to the TB-HIV Registry between 2006 and 2015, 17 were excluded either because of duplicate reporting or revision of diagnosis. Twenty-one patients were tourists or illegal immigrants with a very short stay in Hong Kong; these were also excluded, resulting in 299 patients in the analysis. Baseline demographic and clinical characteristics of these 299 patients, stratified by survival status at 12 months from the date of start of anti-TB treatment, are shown in Table 1. Overall, extrapulmonary TB was common, present in approximately 65% of patients. Sputum smear positivity was present in more than one-third of patients. Of note, the multidrug resistance rate among the bacteriologically proven cases was low (2.2% overall). The overall median CD4 cell count at the time of TB diagnosis was 100/μL (interquartile range [IQR], 36-204/μL). Of 284 patients who were alive and had not defaulted or transferred out at 8 weeks from the initiation of TB treatment, 73 (25.7%) were undergoing ART. The proportion undergoing ART at 8 weeks was higher among patients reported to the TB-HIV Registry between 2011 and 2015 than among those reported between 2006 and 2010 (37.3% vs 18.4%; P=0.0005).

Overall, 135 (45.2%) were cured or had completed treatment at 12 months, whereas 96 (32.1%) remained on an anti-TB treatment regimen at 12 months due to extensive disease, use of a non-standard treatment regimen, and/or drug resistance. Twenty-three (7.7%) patients defaulted treatment for more than 2 months, while 24 (8.0%) patients were transferred out; 21 (7.0%) patients died within 12 months of the initiation of anti-TB treatment. The proportions of patients who died within 12 months among patients reported to the TB-HIV Registry between 2011 and 2015 and those reported between 2006 and 2010 were similar (7.1% vs 6.9%). The median time interval between the initiation of anti-TB treatment and death was 7.5 months (IQR, 3.8-10 months); the median age of death was 54 (IQR, 40.5-75.0) years. The cause of death was due to TB in five patients and unrelated to TB in the remaining 16 (Table 2). A similar proportion of patients with early death within 2 months died from TB (25.0%), compared with those who died after 2 months (23.5%; P=0.72).

Table 3 summarises the results of univariate (using Kaplan-Meier estimates) and Cox proportional hazards regression analysis with respect to 1-year mortality of the 299 cases. In the univariate analysis, older age, case categories other than new cases, a positive sputum smear, a history of drug addiction, and a low CD4 cell count (<50/μL) at the time of TB diagnosis were associated with mortality. Further analysis with Cox proportional hazards regression showed that older age (adjusted hazard ratio=4.5; 95% confidence interval [CI]=1.4-14.9; P=0.012), a history of drug addiction (adjusted hazard ratio=4.6; 95% CI=1.6-13.0; P=0.005) and a low CD4 cell count (<50/μL) at the time of TB diagnosis (adjusted hazard ratio=2.9; 95% CI=1.1-7.7; P=0.03) remained the sole independent risk factors for death within 12 months. Similar findings were observed when ART at 8 weeks was included in the Cox proportional hazards regression model (results not shown). Unadjusted Kaplan-Meier survival curves stratified by covariates are shown in the Figure.

This retrospective cohort study was designed to examine the mortality pattern and factors associated with 1-year mortality among patients with HIV-associated TB reported to the Hong Kong TB-HIV Registry. In our cohort, 7.0% died within 12 months of anti-TB treatment (median [IQR], 7.5 months [3.8-10 months]). The median age of death was 54 (IQR, 40.5-75.0) years. The cause of death was unrelated to TB in >75% of the cohort. Older age, history of drug addiction, and a low CD4 cell count (<50/μL) at the time of TB diagnosis were independent risk factors for death within 12 months.

The 1-year mortality rate of 7.0% among patients with HIV-associated TB in our cohort was lower than the rates in Eastern Europe (29%) and Latin America (11%) reported in a study by Podlekareva et al.2 The mortality rate in our study was also lower than the rates reported in other countries in the Western Pacific region, such as Thailand (17%)7 and Cambodia (22.5%)8 among enrolled patients with HIV-associated TB. Nonetheless, the rate was higher than the 4% reported for Western Europe.2 Although these rates might not be directly comparable because of differences in the study periods and methodologies used, the results from our study add important data to the current body of literature regarding mortality risk among patients with HIV-associated TB in the Western Pacific region.

In our cohort, 76.2% of deaths were unrelated to TB. This differed from the proportion reported among patients with HIV-associated TB in Eastern Europe; TB was reported as the cause of death in >60% of those who died.2 Differences in patient profile, degree of immunosuppression, proportion of patients receiving ART, and proportion of patients with other opportunistic infections might have contributed to the differences observed with respect to the cause(s) of death.

The proportion of patients who died within the first 2 months of anti-TB treatment due to TB was similar to that among patients who died after 2 months (25.0% vs 23.5%). This finding differed from the outcome of a prospective observational study that examined the causes of death among patients with HIV-associated TB in Thailand: 18 of 33 deaths (55%) within 60 days after initiation of TB treatment were caused by TB, compared with 11 of 41 deaths (27%) beyond 60 days after initiation of TB treatment (P=0.02).7 However, the patients in that study exhibited a greater degree of immunosuppression: the median CD4 count for patients in that study was 55/μL (IQR, 18-142/μL), compared with a median CD4 count of 100/μL (IQR, 36-204/μL) in our cohort.

The finding that older age constitutes a risk factor for mortality in our cohort is consistent with findings in some other studies.4 5 12 In a recently reported retrospective 10-year electronic record review of patients with TB in a South African province, although the large majority of patients with TB that died were aged 18 to 49 years, the odds of dying were incrementally higher in older age-groups: 8 to 17 years (adjusted odds ratio [AOR]=2.0; 95% CI=1.5-2.7), 18 to 49 years (AOR=5.8; 95% CI=4.0-8.4), 50 to 64 years (AOR=7.7; 95% CI=4.6-12.7), and ≥65 years (AOR=14.4; 95% CI=10.3-20.2).4 A similar finding was observed in another South African study.5 In our cohort, the median age of death was 54 years. Patients aged ≥65 years had an adjusted hazard ratio of 4.5 for dying, compared with younger patients. Atypical presentation, increased comorbidity, and a higher proportion of drug-related adverse events might have contributed to the increased odds of death in the older age-group.

In our cohort, a history of drug addiction was significantly associated with death at 12 months (adjusted hazard ratio=4.6; 95% CI=1.6-13.0). In the study by The TB:HIV Study Writing Group, injection drug use was reported as a significant predictor of death (hazard ratio=2.11; 95% CI=1.04-4.26); notably, this became non-significant after adjustment due to high correlation with the region of residence.3 The association between a history of drug abuse and mortality in our cohort—also reported in some other studies—is not surprising. Poor access to and uptake of health services, poor adherence, and treatment default are well-known to pose unique challenges for treatment of drug users.13 In a study of 291 patients with smear- and culture-positive pulmonary TB presenting for retreatment in Morocco, substance users were 2.7 times more likely to default treatment, compared with non-substance users (AOR=2.73; 95% CI=1.04-7.15).14 Because injection drug users exhibit a high prevalence of chronic viral hepatitis and alcohol abuse, hepatotoxic reactions to anti-TB drugs may be more common.15 In a study that examined the role of hepatitis C virus and HIV in anti-TB drug-induced hepatotoxicity, Ungo et al16 found that patients with TB exhibited a four-to-five-fold increased risk of drug-induced hepatitis if coinfected with viral hepatitis or HIV; moreover, they exhibited a 14-fold increased risk if coinfected with both. In addition, efavirenz, nevirapine, and rifampicin may increase methadone clearance, thus causing methadone withdrawal; this may dissuade drug users from continuing treatment.15

In a study that examined TB-related mortality in people living with HIV in Eastern Europe and Latin America, a CD4 cell count of <50/μL was associated with increased TB-related mortality (hazard ratio=3.46; 95% CI=2.02-5.95; P<0.0001) after adjustment for multidrug-resistant TB status.2 In another study conducted in Ethiopia, a CD4 cell count of <75/μL was also associated with increased mortality (adjusted hazard ratio=4.83; 95% CI=1.98-11.77).6 In our cohort, consistent with the findings from prior studies, patients with HIV-associated TB with a low CD4 cell count (<50/μL) at the time of TB diagnosis had an adjusted hazard ratio of 2.9 (95% CI=1.1-7.7; P=0.03) for mortality at 1 year. The findings from our study and other studies support the recommendation that ART be initiated early during TB treatment in patients with HIV.

Case category other than new cases was associated with increased mortality at 12 months in univariate analysis in our study. However, the association disappeared in the adjusted model. Of note, prior history of TB or retreatment has been associated with increased mortality in some other studies.5 17 Because our study was a retrospective case review, it is uncertain whether misclassification or inadequate control for other unknown confounding factors might have obscured an association between case category and mortality.

Studies regarding the association between sputum smear status and death have reported conflicting results. In a retrospective cohort study of predictors of TB mortality in Khayelitsha, South Africa, an area with high HIV prevalence, no association was found between smear status and death.5 Similarly, in a study of patients coinfected with TB and HIV at a teaching hospital in Ghana, no differences were reported in clinical outcome between patients with smear-negative TB status and those with smear-positive TB status.18 In a prospective study of 827 adult TB in-patients registered at Zomba Hospital, Malawi, of whom 77% were HIV-seropositive, patients with smear-negative pulmonary TB exhibited the highest death rates during 32 months of follow-up (hazard ratio=2.7; 95% CI=2.1-3.5; P<0.001, compared to smear-positive patients).19 Similar findings were reported in studies of patients with HIV-associated TB in South Africa and in Cameroon.17 20 Conversely, in a study that examined patients with HIV with culture-confirmed pulmonary TB in the US from 1993 to 2006, patients with smear-negative TB exhibited better survival than patients with smear-positive disease, both before (hazard ratio=0.82; 95% CI=0.75-0.90) and after (hazard ratio=0.81; 95% CI=0.71-0.92) the introduction of combination ART.21 In our study, a positive sputum smear was associated with mortality in univariate analysis, but not in multivariate analysis. The results from our study regarding the effect of AFB smear status on survival were similar to those from the study in the US. The reasons for observed differences in survival among patients with HIV with sputum smear-negative TB between resource-rich and resource-limited settings are likely complex; these might have been a result of differences in study populations, the availability of diagnostics, misdiagnosis, or differences in treatment regimens.21

Of note, ART was not a prognostic factor for mortality in our study, either for the entire cohort or when the analysis was restricted to the subgroup with a CD4 cell count of <50/μL (results for the latter not shown). In addition, the mortality rate among patients in the later cohort (2011 to 2015) remained similar compared to the earlier cohort (2006 to 2010), although the proportion of patients receiving ART at 8 weeks had doubled. This conflicts with the finding that ART reduces mortality, reported in some other studies.3 5 The failure of ART to affect mortality in our study might have been because ART was selectively initiated for patients with more advanced HIV-associated TB (ie, by attending physicians) during the study period. High age-related mortality in the cohort, as discussed above, and inadequate control for other confounding factors might have masked the protective effect of ART.

The strength of this study is that it included a relatively complete set of data regarding clinical features of all patients, as well as the cause(s) and timing of death among patients who died. Such data have been less frequently reported in previous studies. The primary limitation is that it was a retrospective study conducted under programme conditions. Our risk factor evaluation was limited by the availability of data present in the database of the TB-HIV Registry and in clinical records. Information regarding case categories, drug addiction, and CD4 cell count was unavailable for a small subset of patients. Classifying patients with missing case category information as new cases and those with missing substance use data as non-users might have biased our analyses of these risk factors such that no significant differences were found. The time lag between HIV diagnosis and initiation of ART could not be determined because information regarding the date of HIV diagnosis was missing for some patients in the earlier portion of the study period, as well as for patients who were diagnosed outside Hong Kong. Furthermore, information regarding the time lag between presentation of patients to the health care system and the initiation of anti-TB treatment was not captured in the TB-HIV Registry. Therefore, the potential effect of such delay in the initiation of anti-TB treatment on mortality could not be examined. Another limitation is that the accuracy of the data in the programme record forms (on which the database was constructed) and death certificates for coding TB as the underlying cause of death was unknown. Nonetheless, for those patients who died, relevant clinical records from chest clinics and hospitals were traced for review. Misclassification regarding the cause(s) of death should be minimal. Finally, the small number of patients who died, limited statistical power, and inadequate controls for other unknown confounding factors may have concealed the effects of significant variables, as discussed above.

Notwithstanding the aforementioned limitations, we demonstrated that older age, history of drug addiction, and low CD4 cell count (<50/μL) at TB diagnosis were independent risk factors for death within 12 months. This study complements previous studies by providing information regarding risk factors associated with mortality among patients with HIV-associated TB in an area in the Western Pacific region with intermediate TB burden and low HIV prevalence. The results from our study may have important implications for health care programmes; they may aid in formulating innovative strategies and new guidelines to guide targeted measures for the management of disease among subgroups of patients at risk of death from TB or HIV during the course of anti-TB treatment. As such, they may help optimise TB and HIV management in the Western Pacific region.

Concept or design of study: CK Chan, ECC Leung, CC Leung, KH Wong.
Acquisition of data: CK Chan, IKY Mak, WK Chan, KCW Chan, MP Lee.
Analysis or interpretation of data: ECC Leung, CC Leung, CK Chan, MP Lee, KCW Chan, IKY Mak.
Drafting of the article: CK Chan, ECC Leung.
Critical revision for important intellectual content: All authors.

The authors would like to thank Dr CM Tam, previous Consultant Chest Physician in-charge of Tuberculosis and Chest Service, Centre for Health Protection, Department of Health of Hong Kong, for his contributions to the conception and design of this work, as well as his valuable comments on the early draft of this article.

All authors have disclosed no conflicts of interest. 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. The findings of this study were presented in part at the 13th Hong Kong, Macau, Taiwan, Shanghai and Guangdong Tuberculosis Conference, Hong Kong, 2 November 2017.

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

This study was approved by the Ethics Committee of the Department of Health of Hong Kong.

1. World Health Organization. Global Tuberculosis Report 2017. Geneva: World Health Organization; 2017.

2. Podlekareva DN, Efsen AM, Schultze A, et al. Tuberculosis-related mortality in people living with HIV in Europe and Latin America: an international cohort study. Lancet HIV 2016;3:e120-31. Crossref

3. TB:HIV Study Writing Group. One-year mortality of HIV-positive patients treated for rifampicin- and isoniazid-susceptible tuberculosis in Eastern Europe, Western Europe, and Latin America. AIDS 2017;31:375-84. Crossref

4. Heunis JC, Kigozi NG, Chikobvu P, Botha S, van Rensburg HD. Risk factors for mortality in TB patients: a 10-year electronic record review in a South African province. BMC Public Health 2017;17:38. Crossref

5. Pepper DJ, Schomaker M, Wilkinson RJ, de Azevedo V, Maartens G. Independent predictors of tuberculosis mortality in a high HIV prevalence setting: a retrospective cohort study. AIDS Res Ther 2015;12:35. Crossref

6. Sileshi B, Deyessa N, Girma B, Melese M, Suarez P. Predictors of mortality among TB-HIV co-infected patients being treated for tuberculosis in Northwest Ethiopia: a retrospective cohort study. BMC Infect Dis 2013;13:297. Crossref

7. Cain KP, Anekthananon T, Burapat C, et al. Causes of death in HIV-infected persons who have tuberculosis, Thailand. Emerg Infect Dis 2009;15:258-64. Crossref

8. Marcy O, Laureillard D, Madec Y, et al. Causes and determinants of mortality in HIV-infected adults with tuberculosis: an analysis from the CAMELIA ANRS 1295-CIPRA KH001 randomized trial. Clin Infect Dis 2014;59:435-45. Crossref

9. Tuberculosis and Chest Service, Centre for Health Protection, Department of Health, Hong Kong SAR Government. Annual Report 2015. Available from: http://www.info.gov.hk/tb_chest/doc/Annual_Report_2015.pdf. Accessed 18 Sep 2018.

10. Special Preventive Programme, Centre for Health Protection, Department of Health, Hong Kong SAR Government. HIV surveillance report-2015 update. Available from: http://www.info.gov.hk/aids/english/surveillance/sur_report/hiv15.pdf. Accessed 18 Sep 2018.

11. Chan CK, Wong KH, Leung CC, et al. Treatment outcomes after early initiation of antiretroviral therapy for human immunodeficiency virus-associated tuberculosis. Hong Kong Med J 2013;19:474-83. Crossref

12. Waitt CJ, Squire SB. A systematic review of risk factors for death in adults during and after tuberculosis treatment. Int J Tuberc Lung Dis 2011;15:871-85. Crossref

13. Deiss RG, Rodwell TC, Garfein RS. Tuberculosis and illicit drug use: review and update. Clin Infect Dis 2009;48:72-82. Crossref

14. Dooley KE, Lahlou O, Ghali I, et al. Risk factors for tuberculosis treatment failure, default, or relapse and outcomes of retreatment in Morocco. BMC Public Health 2011;11:140. Crossref

15. World Health Organization. Integrating collaborative TB and HIV services within a comprehensive package of care for people who inject drugs. Geneva: World Health Organization; 2016.

16. Ungo JR, Jones D, Ashkin D, et al. Antituberculosis drug-induced hepatotoxicity. The role of hepatitis C virus and the human immunodeficiency virus. Am J Respir Crit Care Med 1998;157:1871-6. Crossref

17. Mabunda TE, Ramalivhana NJ, Dambisya YM. Mortality associated with tuberculosis/HIV co-infection among patients on TB treatment in the Limpopo province, South Africa. Afr Health Sci 2014;14:849-54. Crossref

18. Mudd J, Jaramillo L, Kwara E, et al. Impact of smear-negative results on tuberculosis outcome in HIV co-infected patients at a teaching hospital in Ghana. Ann Glob Health 2016;82:516.

19. Kang’ombe C, Harries AD, Banda H, et al. High mortality rates in tuberculosis patients in Zomba Hospital, Malawi, during 32 months of follow-up. Trans R Soc Trop Med Hyg 2000;94:305-9. Crossref

20. Djouma FN, Noubom M, Ngomba AV, Donfack H, Kouomboua PS, Saah MA. Determinants of death among tuberculosis patients in a semi urban diagnostic and treatment centre of Bafoussam, West Cameroon: a retrospective case-control study. Pan Afr Med J 2015;22:253. Crossref

21. Cavanaugh JS, Shah NS, Cain KP, Winston CA. Survival among patients with HIV infection and smear-negative pulmonary tuberculosis–United States, 1993-2006. PLoS One 2012;7:e47855. Crossref