Laparoscopy has replaced laparotomy in most gynaecological procedures, and laparoscopic cystectomy is currently the mainstay of treatment for apparently benign ovarian cysts. However, the absence of depth perception and limited instrument dexterity are major drawbacks of laparoscopy. Advances in three-dimensional (3D) video imaging technology allow 3D laparoscopy to provide better precision than conventional two-dimensional (2D) laparoscopy, especially in depth perception and spatial orientation. This increased precision may help improve surgeon’s performance during laparoscopic surgery.

Studies have shown that 3D laparoscopy objectively1 2 and subjectively3 4 improves surgical performance, especially during complex tasks.5 In addition, 3D laparoscopy lessens the learning curve for beginners.6 The durations of laparoscopic cholecystectomy and pelvic lymphadenectomy have also been shortened when performed using 3D technologies.7 8 However, the clinical value of 3D laparoscopy in benign gynaecological surgery remains uncertain. This study aimed to evaluate any advantages of using 3D laparoscopy over 2D laparoscopy for ovarian cystectomy.

This prospective randomised study was conducted from May 2014 to May 2016 at the Queen Mary Hospital, Hong Kong, a teaching hospital affiliated with The University of Hong Kong. Women with apparently benign ovarian cysts who were scheduled for elective laparoscopic ovarian cystectomy and who were eligible for the study were invited at the pre-admission clinic to enrol in the study. Inclusion criteria were being older than 18 years; ability to understand Cantonese, Putonghua, or English; and ability to understand the study information during the consent process. Patients who were intra-operatively found to have no ovarian cyst were excluded from further analysis.

Patients were allocated by block randomisation to undergo surgery with 2D laparoscopy (2D group) or 3D laparoscopy (3D group) according to a computer-generated random sequence, in blocks of five. The group allocation for each patient was disclosed to the surgeon on the day before the surgery using a consecutively numbered, opaque, sealed envelope. Demographic data of patients and duration of surgeries were collected by a research nurse.

A pneumoperitoneum was created using a Veress needle to provide visually guided closed access. For 3D laparoscopy, a 10-mm 3D telescopic videoscope was used (Endoeye Flex 3D; Olympus, Center Valley [PA], US). All surgeons were trained for 3D laparoscopy using a pelvic trainer with standardised tasks including peg transfer, precision cutting, duct cannulation, and suturing with knot tying. The 3D laparoscopy training was continued until the surgeons could confidently operate using 3D images. All non-specialist surgeons were supervised by a laparoscopist accredited at the advanced level in gynaecological laparoscopic surgery, according to the Hong Kong College of Obstetricians and Gynaecologists.9 At their discretion, surgeons were allowed to switch from 3D laparoscopy to traditional 2D laparoscopy if difficulty was encountered during surgery. All 2D laparoscopies were performed using a 10-mm laparoscope (26033AP; Karl Storz Endoscopy-America Inc, Culver City [CA], US). The same 32-inch high-definition monitor (LMD-3215MT; Sony Corporation, Tokyo, Japan) was used for all operations. In the 2D and 3D groups, cystectomy was performed in the usual manner, using two or three 5-mm accessory ports inserted in the lower abdomen under direct vision. The start time of the operation (first skin incision), insertion of primary trocar, completion of cystectomy, and end of operation (final skin closure) were recorded by the research nurse.

After the operation, all surgeons were required to self-evaluate their performance by using the Global Operative Assessment of Laparoscopic Skills (GOALS) assessment tool.10 The five-item GOALS score includes assessment of depth perception, bimanual dexterity, efficiency, tissue handling, and autonomy. Any operator discomfort encountered during the surgery, any need to convert to 2D laparoscopy, and the surgeon’s preference for the type of laparoscopy based on experience were also recorded. Demographic data and operative findings, such as size and laterality of cysts, operative duration, and presence of adhesions were analysed. Duration of cystectomy was defined as the time from completion of primary port insertion to separation of the cyst from the ovary and completion of haemostasis. The time spent on specimen retrieval was not included, owing to variations in the specimen retrieval method with or without use of a specimen bag.

The primary outcome of the present study was the difference between the GOALS score of 2D and 3D groups. The secondary outcomes were the duration of cystectomy and surgeon’s preferences and reported adverse effects. Subgroup analysis was performed to compare the outcomes for different experience levels among the surgeons. The surgeons were categorised according to their experience in performing laparoscopic surgery (≤5 years or >5 years). Surgeons with more than 5 years of experience had achieved competency in gynaecological laparoscopic surgery to at least an intermediate level, according to the Hong Kong College of Obstetricians and Gynaecologists, and had completed a required number of laparoscopic operations as requested by the College.9

A sample size of 36 patients was required in each group, as calculated using an alpha of 0.05 and a beta of 0.2 for detection of a difference in the sum of four items of the GOALS score (excluding tissue handling) of 13 (interquartile range [IQR], 11-16) in the 2D group and 16 (IQR, 12-18) in the 3D group, as based on a previous study,11 using a two-sided test. To allow for a 10% dropout rate, 40 patients were recruited into each group. For randomised patients whose operations were subsequently rescheduled outside the study period, treatment assignment numbers were reallocated to subsequent eligible patients who provided consent. Statistical analysis was performed using SPSS Windows version 21.0 (IBM Corp, Armonk [NY], US). Data were presented as proportions or mean and standard deviation. Student’s t test and Chi squared test were used for statistical analyses. A P value of <0.05 was considered statistically significant.

Of the 83 patients recruited into the study from May 2014 to May 2016, operations were rescheduled for three patients who were therefore withdrawn from the study; 80 patients completed the trial (Fig). Of these 80 patients, two from the 2D group and three from the 3D group were excluded from analysis because no cysts were identified. Finally, 38 patients in the 2D group and 37 patients in the 3D group were included for analysis. Patient characteristics and surgical outcomes are presented in Table 1. There were no significant differences between the 2D and 3D groups in terms of patient age, laterality of the ovarian cyst, histological diagnosis of the cyst, presence of severe adhesions, volume of blood loss, and experience level of the surgeon. Three accessory ports were used in four patients in the 2D group and in five patients in the 3D group. In all other patients, two accessory ports were used. The mean (standard deviation) diameter of the ovarian cyst was smaller in the 3D group than that in the 2D group (5.1 [2.1] cm vs 6.1 cm [2.1] cm; P=0.031). Body mass index in the 2D group was significantly higher than that in the 3D group (23.4 [4.4] kg/m² vs 21.3 [2.6] kg/m²; P=0.011). Severe adhesion was defined as a score of >20 for adnexal adhesion unilaterally12 or a score of >40 for endometriosis,13 according to the American Society for Reproductive Medicine classifications.

The differences between 2D and 3D groups in terms of GOALS score and duration of cystectomy are presented in Table 2. A total of 15 surgeons participated in the study and there were 13 in each group: 11 in both, while two for each were involved in 2D and 3D groups, respectively. Participating surgeons in the 2D group reported more efficient tissue handling than did those in the 3D group. Adverse effects, including nausea, dizziness, ocular fatigue, and blurring of vision were reported less frequently by participating surgeons in 2D group than those in 3D group (Table 3). However, none of the participating surgeons requested intra-operative conversion from 3D to 2D laparoscopy. At the end of surgery, more participating surgeons in the 3D group expressed a preference for 2D laparoscopy (43.3%) than for 3D laparoscopy (18.9%), whereas 37.8% had no preference. A subgroup analysis of participating surgeons in the two groups did not show statistically significant differences in terms of GOALS score (2D vs 3D; 28.9 [5.1] vs 28.2 [46.0]; P=0.585), tissue handling (4.2 [0.8] vs 3.9 [0.8]; P=0.060), and duration of cystectomy (93.7 [46.1] min vs 97.7 [52.2] min; P=0.737).

Subgroup analyses according to the experience level of the surgeon and the presence of dense adhesions are shown in Tables 4 and 5, respectively. Two of the surgeons in the 3D group and three of the surgeons in the 2D laparoscopy are accredited at the advanced level in gynaecological laparoscopic surgery by the Hong Kong College of Obstetricians and Gynaecologists. Surgeons with more than 5 years of laparoscopic experience reported lower scores in tissue handling and efficiency when using 3D laparoscopy. There were no differences in terms of GOALS score and duration of cystectomy in the subgroup with dense adhesions.

Three-dimensional laparoscopy is gaining popularity in modern gynaecological surgery owing to improved depth perception and spatial orientation compared with 2D laparoscopy. Improved effectiveness using 3D laparoscopy has been shown extensively in training models, especially when performing complex tasks5 and in beginners.6 8 14 However, our study was unable to show an improvement in terms of GOALS score and duration of operation (Table 2) despite the 3D laparoscopy group having a smaller mean ovarian cyst diameter (Table 1). This finding contradicts a recent meta-analysis that 3D laparoscopy was associated with shortened surgical time and hospital study, less blood loss, and fewer perioperative complications.15

The addition of binocular vision and depth perception in 3D laparoscopy is associated with more frequent adverse effects such as ocular fatigue, nausea, and dizziness.16 In the present study, participating surgeons in the 3D group more frequently reported nausea, dizziness, ocular fatigue, and blurring of vision than did those in the 2D group. However, this result may be because the participating surgeons were unfamiliar with 3D images; with experience, this discomfort may be lessened. Maintaining stability of the telescope is of utmost importance during 3D laparoscopy; therefore, familiarity with 3D images is important for assistants to mitigate adverse effects. Furthermore, maintaining an appropriate distance between the screen and the surgeon also alleviates nausea and ocular fatigue.16

Previous studies have shown that 3D laparoscopy is beneficial for less experienced surgeons6 8 14 and for any surgeon performing complex tasks.5 However, in our subgroup analysis, we were unable to confirm any benefits of 3D laparoscopy in relation to the experience level of the surgeons. All participating surgeons were much more familiar with 2D laparoscopy and, thus, the difference between groups might simply reflect the surgeon’s assessment of what they are used to. This familiarity effect may explain the lower scores in tissue handling and efficiency with 3D laparoscopy attained by the more experienced surgeons.

The surgeon’s preference for 2D laparoscopy and the heterogeneity of the participating surgeons and patients make the subgroup analyses underpowered and represents a constitute limitation of the present study. The differences in mean diameter of the ovarian cysts and body mass index between the two groups also suggest ineffective randomisation. Other limitations include ineffective randomisation, withdrawal of patients after randomisation, and surgeon’s lack of experience with 3D laparoscopy. During data analysis, there were also no controls for possible confounding factors, such as experience of each surgeon with 3D laparoscopy or significant differences in patient characteristics between the groups.

In conclusion, the results show that there is no significant benefit to using 3D laparoscopy for ovarian cystectomy compared with conventional 2D laparoscopy. Moreover, 3D laparoscopy is associated with more frequent adverse effects for surgeons. However, it is possible that more complex procedures, such as those involving laparoscopic suturing and knot tying, might be easier to perform with 3D laparoscopy than with 2D laparoscopy. Therefore, further evaluation of the clinical performance of 3D laparoscopy in operations of different complexities and of surgeons with different experience levels are warranted.

All authors have made substantial contributions to the concept of this study; acquisition of data; analysis or interpretation of data; drafting of the article; and critical revision for important intellectual content.

We wish to thank Ms Wai-ki Choi for helping in patient recruitment and data collection.

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

The authors have no conflicts of interest to disclose. 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 study was presented as oral presentation in the 25th Asian and Oceanic Congress of Obstetrics and Gynaecology, 16 June 2017, Hong Kong.

Ethical approval was obtained from the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster. Written informed consent was obtained from all participating patients and surgeons. The study was registered with ClinicalTrials.gov (NCT02775344).

1. Storz P, Buess GF, Kunert W, Kirschniak A. 3D HD versus 2D HD: surgical task efficiency in standardised phantom tasks. Surg Endosc 2012;26:1454-60. Crossref

2. Lusch A, Bucur PL, Menhadji AD, et al. Evaluation of the impact of three-dimensional vision on laparoscopic performance. J Endourol 2014;28:261-6. Crossref

3. Tanagho YS, Andriole GL, Paradis AG, et al. 2D versus 3D visualization: impact on laparoscopic proficiency using the fundamentals of laparoscopic surgery skill set. J Laparoendosc Adv Surg Tech A 2012;22:865-70. Crossref

4. Sørensen SM, Savran MM, Konge L, Bjerrum F. Three-dimensional versus two-dimensional vision in laparoscopy: a systematic review. Surg Endosc 2016;30:11-23. Crossref

5. Wagner OJ, Hagen M, Kurmann A, Horgan S, Candinas D, Vorburger SA. Three-dimensional vision enhances task performance independently of the surgical method. Surg Endosc 2012;26:2961-8. Crossref

6. Cicione A, Autorino R, Breda A, et al. Three-dimensional vs standard laparoscopy: comparative assessment using a validated program for laparoscopic urologic skills. Urology 2013;82:1444-50. Crossref

7. Bilgen K, Ustun M, Karakahya M, et al. Comparison of 3D imaging and 2D imaging for performance time of laparoscopic cholecystectomy. Surg Laparosc Endosc Percutan Tech 2013;23:180-3. Crossref

8. Fanfani F, Rossitto C, Restaino S, et al. How technology can impact surgeon performance: a randomized trial comparing 3-dimensional versus 2-dimensional laparoscopy in gynecology oncology. J Minim Invasive Gynecol 2016;23:810-7. Crossref

9. Hong Kong College of Obstetricians and Gynaecologists. Endoscopic surgery: accreditation of gynaecological laparoscopic surgery. Available from: http://www.hkcog.org.hk/hkcog/pages_2_64.html. Accessed 4 Jun 2017.

10. Vassiliou MC, Feldman LS, Andrew CG, et al. A global assessment tool for evaluation of intraoperative laparoscopic skills. Am J Surg 2005;190:107-13. Crossref

11. Ko JK, Li RH, Cheung VY. Two-dimensional versus three-dimensional laparoscopy: evaluation of physicians’ performance and preference using a pelvic trainer. J Minim Invasive Gynecol 2015;22:421-7. Crossref

12. Hulka JF, Omran K, Berger GS. Classification of adnexal adhesions: a proposal and evaluation of its prognostic value. Fertil Steril 1978;30:661-5. Crossref

13. Revised American Fertility Society classification of endometriosis: 1985. Fertil Steril 1985;43:351-2. Crossref

14. Alaraimi B, El Bakbak W, Sarker S, et al. A randomized prospective study comparing acquisition of laparoscopic skills in three-dimensional (3D) vs. two-dimensional (2D) laparoscopy. World J Surg 2014;38:2746-52. Crossref

15. Cheng J, Gao J, Shuai X, Wang G, Tao K. Two-dimensional versus three-dimensional laparoscopy in surgical efficacy: a systematic review and meta-analysis. Oncotarget 2016;7:70979-90. Crossref

16. Kunert W, Storz P, Kirschniak A. For 3D laparoscopy: a step toward advanced surgical navigation: how to get maximum benefit from 3D vision. Surg Endosc 2013;27:696-9. Crossref

All newborns have elevated unconjugated bilirubin concentrations relative to normal adult values. Two thirds or more of breastfed infants have unconjugated hyperbilirubinaemia that extends into the second and third weeks of life, and often up to age 8 to 12 weeks.1 Guidelines on the evaluation of cholestasis in infants recommend establishing cholestasis before proceeding to investigate the underlying pathology.2 3 However, of the hospitals administered by the Hong Kong Hospital Authority that care for newborns, most check full liver function at the first Neonatal Jaundice Clinic (NNJC) visit.

The Food and Health Bureau established the Committee on Promotion of Breastfeeding in Hong Kong in 2014. Since then, there has been a substantial improvement in exclusive breastfeeding rates and a continuous upward trend in ever-breastfeeding rates.4

The Tuen Mun Hospital is a regional hospital in the New Territories West region of Hong Kong. An outpatient Neonatal Jaundice Clinic is available every Monday to Friday from 14:00 to 17:00 to treat infants with jaundice. The increasing breastfeeding rates have led to increasing numbers of infants with prolonged jaundice being referred to the NNJC.

The present study aimed to learn about the clinical and biochemical characteristics of prolonged neonatal jaundice, and to study the diagnostic yield of a full liver function check at first NNJC visit. We reviewed the clinical and laboratory records of infants visiting the NNJC between 8 July 2015 and 8 March 2016. On the basis of the findings, we propose a more effective evaluation procedure for prolonged neonatal jaundice.

Attendance records maintained by the NNJC included the registration number, sex, and attendance date of the patients. Using the recorded patient registration numbers, gestational age, birth weight, glucose-6-phosphate dehydrogenase (G6PD) deficiency status, mode of feeding, phototherapy history, liver function test results, and clinical progress were retrieved from the clinical management system and electronic patient record. The study period was chosen for convenience. All infants who visited the NNJC during the study period were included to prevent possible selection bias. All suspicious data were verified by revisiting the electronic patient record to ensure data accuracy.

For gestational age, completed weeks of gestation were recorded. Preterm births were categorised as very preterm (before 34 weeks of gestation) or late preterm (between 34 weeks and 36 weeks 6 days of gestation). Full-term births were those at 37 weeks of gestation or later. Birth weights were categorised as very low (<1500 g), low (1500-2499 g), normal (2500-3999 g), or high (>3999 g). Modes of feeding were categorised as exclusive breastfeeding, mixed feeding, or exclusive formula feeding.

The SPSS for Windows version 15.0 (SPSS Inc, Chicago [IL], United States) was used for all statistical analyses. Student’s t test and univariate analysis of variance (ANOVA) were used in comparing the outcomes between groups. Tukey’s test was used for pairwise comparisons in ANOVA. Cross-tabulation was used to measure associations between binary outcome variables and binary predictor variables.

In total, 1164 infants (663 males, 501 females; male-to-female ratio=1.3:1) with prolonged jaundice were referred to the NNJC during the 8-month study period. The gestational ages of the infants ranged from 29 to 41 weeks, and there were eight (0.69%) very preterm infants and 94 (8.08%) late preterm infants. The birth weights of the infants ranged from 1425 g to 4670 g. The sample included only one (0.09%) very low birth weight infant, in addition to 80 (6.87%) low birth weight infants and 13 (1.12%) high birth weight infants. In total, 34 (5.13%) male infants and two (0.40%) female infants had G6PD deficiency. The mode of feeding was exclusive breastfeeding in 648 (55.70%) infants, mixed feeding in 400 (34.36%) infants, and exclusive formula feeding in 114 (9.79%) infants; the mode of feeding was not recorded in two infants. Among the 114 exclusively formula fed infants, 24 (21.10%) had a history of breastfeeding. At the first NNJC visit, 70 (6.01%) infants were 2 weeks old, 156 (13.40%) 3 weeks old, 758 (65.12%) 4 weeks old, 165 (14.18%) 5 weeks old, 10 (0.86%) 6 weeks old, and three (0.26%) 7 weeks old. One (0.09%) infant first visited at 8.7 weeks old and another (0.09%) first visited at 11.4 weeks old.

At the first NNJC visit, 1139 (97.90%) infants received a full liver function test, which included taking measurements of alkaline phosphatase (ALP), calcium, and phosphate. In seven infants, low transcutaneous bilirubinometer readings were observed (peak reading, 42-86 μmol/L). Therefore, blood tests were not performed in these infants. At the first NNJC visit, 17 infants were seen by a doctor who adopted the stepwise approach and who screened only for cholestasis; one infant visiting on day 14 had only total bilirubin (TB) checked.

There was a significant effect of mode of feeding on TB (F_2,1155=18.058; P<0.01). The mean (M) TB levels in the exclusive breastfeeding group (M=146.73) were significantly higher than those in the mixed feeding group (M=131.05) and in the exclusive formula feeding group (M=121.21). Tukey’s test showed significant differences in TB levels between the exclusive breastfeeding and the mixed feeding groups (P<0.01) and between the exclusive breastfeeding and the exclusive formula feeding groups (P<0.01). There was no significant difference in TB level between the mixed feeding and exclusive formula feeding groups.

There was a significant effect of feeding on alanine transaminase (ALT) levels (F_2,1133=15.015; P<0.01). The ALT levels in the exclusive breastfeeding group (M=21.79) were significantly higher than those in the mixed feeding group (M=18.97) and in the exclusive formula feeding group (M=16.55). Tukey’s test showed significant differences in ALT levels between the exclusive breastfeeding and the mixed feeding groups (P<0.01) and between the exclusive breastfeeding and the exclusive formula feeding groups (P<0.01). There was no significant difference in ALT levels between the mixed feeding and exclusive formula feeding groups.

There was a significant effect of feeding on ALP (F_2,1135=6.276; P<0.01). The ALP levels in the exclusive breastfeeding group (M=348.83) were significantly higher than those in the mixed feeding group (M=329.08) and in the exclusive formula feeding group (M=327.76). Tukey’s test showed a significant difference in ALP level between the exclusive breastfeeding and the mixed feeding groups (P<0.01). There was no significant difference in ALP levels between the exclusive breastfeeding and the exclusive formula feeding groups (P=0.07) or between the mixed feeding and the exclusive formula feeding groups (P=0.99).

There was a significant effect for infants who had received phototherapy (t(1141)=3.57; P<0.01). Infants who had received phototherapy had higher TB levels (M=147.12) than those who had not (M=135.06).

There was no significant difference in TB levels between male and female infants. There was also no statistically significant difference in TB levels between G6PD-deficient and G6PD-sufficient infants.

At the first NNJC visit, 16 (1.4%) infants had conjugated hyperbilirubinaemia. Diagnoses included biliary atresia in one, cytomegalovirus (CMV) infection in three, neonatal hepatitis syndrome in two, and transient cholestasis in 10 infants (Table 1). All infants were thriving well and did not have dark urine or pale stool at their first NNJC visit.

At the first NNJC visit, 98 (8.4%) infants had elevated ALT levels. The reference range used for ALT level is 5 U/L to 33 U/L. The proportion of infants that were followed up increased with increasing ALT level, as shown in Figure 1. Specific causes for elevated ALT level included biliary atresia in one infant, hepatic congestion related to congestive heart failure in one infant, and CMV in five infants (Table 2). Of the remaining 91 infants with elevated ALT levels, 16 had neonatal hepatitis syndrome and 75 had non-specific elevated ALT levels.

Urine tests revealed CMV infection in five infants with elevated ALT levels; CMV infection was believed to be acquired postnatally. These infants were all full-term births of normal birth weight and were asymptomatic. Of these five infants, four were exclusively breastfed and one was fed a mix of breastmilk and formula. After their elevated ALT levels resolved, three infants were discharged from the NNJC and the other two were followed up for coincidental findings (developmental concern in one and familial small head in the other) and not for concern over CMV infection. These five infants and the 75 infants with non-specific elevated ALT were all healthy and asymptomatic and had good weight gain; their elevated ALT levels resolved without treatment.

In this study, inflammation of the liver occurring in early infancy that could not be attributed to a specific cause of liver disease was termed neonatal hepatitis syndrome. The peak ALT levels and the duration of elevated ALT levels in 16 infants with neonatal hepatitis syndrome are shown in Figure 2. The duration of elevated ALT level ranged from 13 to 69 weeks and was shorter than 28 weeks in only four infants. Infants with neonatal hepatitis syndrome were followed up until ALT levels returned to normal and then for an average of 1 month longer.

In 75 infants who were otherwise healthy, non-specific elevated ALT levels were within double the usual upper limit. These infants were either not followed up or their liver function was monitored periodically (time intervals in months determined on case-by-case basis) with limited diagnostic testing.

The mode of feeding, gestational age, and birth weight were all found to affect ALP levels. Among 132 preterm or low birth weight infants, 21 (15.9%) had elevated ALP, compared with 31 (3.0%) among 1032 full-term and normal birth weight infants. For preterm or low birth weight infants versus term infants with birth weight >2499 g, the odds ratio for elevated ALP was 6.109 (95% confidence interval=3.394-10.994) [Table 3]. Among 1139 infants who underwent full liver function tests, 10 had low phosphate levels: seven had concomitant high ALP and low phosphate levels, and the remaining two late preterm infants and one full-term infant had isolated low phosphate levels.

We found that TB levels were significantly higher in the exclusive breastfeeding and mixed feeding groups than in the exclusive formula feeding group. Eight infants that were exposed to breastmilk and two exclusively formula-fed infants had transient cholestasis. Our findings also revealed that ALT levels were significantly higher in the exclusive breastfeeding and mixed feeding groups than in the exclusive formula feeding group. Besides, preterm or low birth weight infants had increased odds of high ALP levels compared with term infants with birth weight >2499 g.

Breast milk jaundice (BMJ) was first described by Newman and Gross in 1963.5 Subsequently, BMJ has been reported to be associated with increased conjugated and unconjugated bilirubin levels.6 In 1991, investigators in Japan studied 58 breastfed infants with indirect hyperbilirubinaemia and found that 18 (31%) with BMJ had elevated ALP, gamma-glutamyltranspeptidase or serum bile acid, reflecting alterations in the hepatobiliary system.7 The serum bile acid levels in patients with BMJ are similar to those with cholestatic jaundice caused by diseases such as extrahepatic biliary atresia.7 This finding suggests that BMJ may be caused by hepatic dysfunction related to cholestasis. Moreover, for infants with BMJ and increased fasting serum bile acid levels, discontinuation of breastfeeding did not cause a rapid normalisation of the serum TB levels.7 This observation suggests that hyperbilirubinaemia in infants with increased serum bile acid levels is not directly related to breastfeeding. Our finding of transient cholestasis in 1.75% (2/114) of exclusive formula feeding infants versus 0.76% (8/1048) exclusive breast milk feeding and mixed feeding infants supported the hypothesis that cholestasis is not related to breastfeeding.

In this study, we used the term ‘transient cholestasis’ to describe the delay in the decline of conjugated bilirubin levels, as observed in 10 infants. Because TB levels declined first without a concomitant decline or even with an increase in conjugated bilirubin levels, there seemed to be a transitional stage of cholestasis. When TB levels decreased further, the conjugated bilirubin levels then decreased. Moreover, the conjugated bilirubin levels in these 10 infants were lower than those with biliary atresia or hepatitis. This finding supports the use of direct bilirubin as a surrogate marker in assessing the severity of cholestasis to ensure optimal timing of hepatobiliary scanning.8

In 1981, Landaas et al9 first reported a significant difference in ALT levels between breastfed and formula-fed infants. They proposed a normal range for ALT level of 14 to 84 IU/L until 4.5 months of age. In 1984, Gómez et al10 examined 2099 out-patient children and found that ALT levels (40-97 IU/L, 3rd to 97th percentile) were higher in children <1 year than in older children. In 2003, investigators in Denmark found higher mean serum bilirubin, albumin, and aspartate transaminase (AST) levels in healthy exclusively breastfed full-term infants; they also found a strong positive association between AST and insulin-like growth factor-1 levels at 2 months (r=0.47; P=0.004).11 Protein levels in breast milk are lower than those in infant formulas. Serum albumin levels have been used to evaluate the adequacy of protein levels in infant formula. Thus, the finding of higher serum albumin levels in breastfed infants than formula feeding infants suggests that there were no protein deficiencies or abnormalities affecting albumin production in breastfed infants.11 Insulin-like growth factor-1 is an anabolic hormone in infants; thus, those authors believed that the most likely explanation for the elevated AST values among breastfed infants is a stimulation of liver metabolism through one of several growth factors in human milk.11 Therefore, the higher AST levels in breastfed infants were believed to be a reflection of a higher liver metabolism, rather than a reflection of liver cell damage. Alanine transaminase is present primarily in the liver and thus is a more specific marker of hepatocellular cell injury. Aspartate transaminase is present in the liver and other organs, a less specific marker of hepatocellular function. The aforementioned study used AST in a restrictive sense to reflect liver biochemistry. Therefore, the higher ALT levels among breastfed infants should logically be interpreted by the same token as reflection of higher liver metabolism rather than liver cell injury.

In the present study, other than those in which CMV was identified, findings were negative. Elevated ALT concentrations resolved in all infants other than the infant with biliary atresia. These findings support the hypothesis that elevated ALT in breastfed infants is a reflection of higher metabolism rather than of any pathology.

Serum ALP is derived predominately from the liver and bones. In the present study, elevated ALP in seven infants was of hepatic origin. Tests of ALP’s heat stability index revealed that 52 infants with isolated ALP elevation were of bone origin.

In this study, 21 (15.9%) preterm or low birth weight infants had isolated elevated ALP (Table 3). Among the eight very preterm infants, only one had isolated elevated ALP and none had low phosphate. Among 94 late preterm infants, 20 (21.3%) had isolated elevated ALP and among these, six (30%) also had low phosphate.

The biochemical characteristics of high ALP and low phosphate are compatible with osteopenia of prematurity. During pregnancy, calcium and phosphorus are actively transferred from the mother to the fetus, reaching a peak accretion rate at 32 to 36 weeks of gestation. The third trimester is the period of most active growth and the increased accretion rate is in response to the higher fetal needs for the developing skeleton. As a result, in preterm infants, calcium and phosphate requirements increase with decreasing gestational age, to compensate for the loss of accretion of these minerals. In the present study, very preterm infants were managed in neonatal wards for long durations; therefore, this need for increased mineral supplements was recognised and addressed. However, late preterm infants may not be clearly distinguished from full-term infants, and their need for additional minerals may not be apparent or addressed during their short hospital stay.

Breastfed infants have been reported to acquire CMV via breastmilk.12 13 Cytomegalovirus excreted in breastmilk is likely caused by reactivated infection in the presence of maternal antibody transferred transplacentally.12 This type of milk-borne CMV infection apparently protects children from CMV diseases and the seropositivity for CMV may protect the next generation from CMV inclusion disease.

Poddighe et al14 reported a full-term breastfed infant with prolonged jaundice who had undergone extensive tests but with negative findings. Liver function test results returned to normal by age 7 months when breastmilk intake was significantly reduced. The authors14 then proposed that, in otherwise healthy infants and in the absence of risk factors, elevated ALT levels should be monitored for 7 months before performing further sophisticated tests. If this proposal were applied to the present study, only four infants had elevated ALT for less than 7 months and would have avoided further tests.

The above discussion suggests that measuring ALT levels at the first NNJC visit is of limited benefit. For jaundiced infants, repeated follow-up examinations for elevated ALT levels may increase the risk of premature cessation of breastfeeding and of development of vulnerable child syndrome.15 Therefore, we propose measuring ALT levels only after noting cholestasis.

In the present study, late preterm infants were identified to be at high risk of having ALP elevation. Preterm or low birth weight infants accounted for <10% of those visiting the NNJC. In addition to cholestasis screening at the first NNJC visit, ALP, albumin, calcium, and phosphate should be checked for late preterm or low birth weight infants.

Vitamin D deficiency has been reported in 18% of Hong Kong women.16 In this study, the prevalence of elevated ALP was 4.6%. The recently published global consensus on prevention and management of nutritional rickets recommends vitamin D supplementation at 400 IU daily in all infants, independent of their mode of feeding in the first year of life.17 Therefore, the best way to protect infants may be to educate pregnant mothers to take vitamin D supplements during pregnancy and to give 400 IU daily vitamin D supplements to their infants.17

Information bias and selection bias are two potential limitations of our study. First, for simplicity, all infants taking just one mouthful of breastmilk and those taking just one mouthful of formula were classified in the mixed feeding group. In the mixed feeding group, this information bias may mask the effect of breastfeeding on liver biochemistry. Second, the study period was chosen for convenience and not at random. This may create selection bias because outbreaks of diseases in infants tend to create clusters of clinic visits within a certain timeframe.

For full-term and normal or high birth weight infants, the most effective way to manage prolonged neonatal jaundice is to screen for cholestasis before full liver function examination. For late preterm or low birth weight infants, the most effective way to manage prolonged neonatal jaundice is to screen for cholestasis and to check bone profile (ALP, albumin, calcium, and phosphate) at the first NNJC visit.

All authors have made substantial contributions to the concept or design of this study; acquisition of data; analysis or interpretation of data; drafting of the article; and critical revision for important intellectual content.

The authors thank Ms CK Ho and Mr WF Wu for maintaining the neonatal jaundice clinic attendance records that made this retrospective study possible.

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

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. This paper was presented orally in Joint Annual Research & Scientific Meeting 2017, 19 August 2017, Hong Kong.

Ethical approval for the study was obtained and patient/parental consent was waived by the New Territories West Cluster Clinical and Research Ethics Committee.

1. Academy of Breastfeeding Medicine Protocol Committee. ABM clinical protocol #22: guidelines for management of jaundice in the breastfeeding infant equal to or greater than 35 weeks’ gestation. Breastfeed Med 2010;5:87-93. Crossref

2. Fawaz R, Baumann U, Ekong U, et al. Guideline for the evaluation of cholestatic jaundice in infants: joint recommendations of the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition and the European Society for Pediatric Gastroenterology, Hepatology, and Nutrition. J Pediatr Gastroenterol Nutr 2017;64:154-68. Crossref

3. Moyer V, Freese DK, Whitington PF, et al. Guideline for the evaluation of cholestatic jaundice in infants: recommendations of the North American Society for Pediatric Gastroenterology, Hepatology and Nutrition. J Pediatr Gastroenterol Nutr 2004;39:115-28. Crossref

4. Family Health Service, Department of Health, HKSAR Government. Breastfeeding Survey 2017.

5. Newman AJ, Gross S. Hyperbilirubinemia in breast-fed infants. Pediatrics 1963;32:995-1001.

6. Winfield CR, Macfaul R. Clinical study of prolonged jaundice in breast- and bottle-fed babies. Arch Dis Child 1978;53:506-7. Crossref

7. Tazawa Y, Abukawa D, Watabe M, et al. Abnormal results of biochemical liver function tests in breast-fed infants with prolonged indirect hyperbilirubinaemia. Eur J Pediatr 1991;150:310-3. Crossref

8. Siu LY, Wong KN, Li KW, et al. Outcome of hepatobiliary scanning: preterm versus full-term cholestatic infants. J Paediatr Child Health 2013;49:E46-51. Crossref

9. Landaas S, Skrede S, Steen JA. The levels of serum enzymes, plasma proteins and lipids in normal infants and small children. J Clin Chem Clin Biochem 1981;19:1075-80. Crossref

10. Gómez P, Coca C, Vargas C, et al. Normal reference-intervals for 20 biochemical variables in healthy infants, children and adolescents. Clin Chem 1984;30:407-12.

11. Jørgensen MH, Ott P, Juul A, et al. Does breast feeding influence liver biochemistry? J Pediatr Gastroenterol Nutr 2003;37:559-65. Crossref

12. Minamishima I, Ueda K, Minematsu T, et al. Role of breast milk in acquisition of cytomegalovirus infection. Microbiol Immunol 1994;38:549-52. Crossref

13. Hamprecht K, Maschmann J, Vochem M, et al. Epidemiology of transmission of cytomegalovirus from mother to preterm infants by breastfeeding. Lancet 2001;357:513-8. Crossref

14. Poddighe D, Castelli L, Marseglia GL, et al. Prolonged, but transient, elevation of liver and biliary functions tests in a healthy infant affected with breast milk jaundice. BMJ Case Rep 2014;pii:bcr2014204124. Crossref

15. Kemper K, Forsyth B, McCarthy P. Jaundice, terminating breast-feeding and the vulnerable child. Pediatrics 1989;84:773-8.

16. Woo J, Lam CW, Leung J, et al. Very high rates of vitamin D insufficiency in women of child-bearing age living in Beijing and Hong Kong. Br J Nutr 2008;99:1330-4. Crossref

17. Munns CF, Shaw N, Kiely M, et al. Global consensus recommendations on prevention and management of nutritional rickets. J Clin Endocrinol Metab 2016;10:394-415. Crossref

Prostate cancer (PCa) is the most common non-cutaneous malignancy among men in western countries, and is the third most common cancer among men in Hong Kong.1 Increasing public awareness in the Chinese community, as well as the common use of prostate-specific antigen (PSA) tests by primary health physicians, have led to detection of PCa at an earlier stage, when it is amenable to either radical surgery or radiotherapy (RT).2 Because of recent advancements in operative management, such as robotic-assisted laparoscopic prostatectomy,3 many patients have found radical prostatectomy (RP) the preferred treatment option. Nevertheless, adjuvant radiotherapy (ART) to the prostatic fossa is indicated postoperatively in cases with positive surgical margin (SM), or residual disease from extracapsular extension (ECE). Alternatively, patients may receive salvage radiotherapy (SRT) when there is PSA failure, defined as any detectable and rising PSA level after RP.

Currently, ART is still being compared with SRT in three randomised controlled trials (RADICALS, RAVES, GETUG-AFU 17).4 5 6 While the results of these European and Australasian studies are still pending, the American Society for Radiation Oncology/American Urological Association guidelines recommend that physicians offer SRT to patients with PSA or local recurrence after RP in whom there is no evidence of distant metastasis (DM).7 Patients should be advised that SRT should be administered at the earliest sign of PSA recurrence. Approximately 60% of patients who are treated with SRT before the PSA level rises to >0.5 ng/mL will achieve an undetectable PSA level, providing long-term PSA control in nearly half of them.8

However, after SRT, some patients may still experience further clinical progression, including DM and cancer-related death. The effect of SRT on the long-term outcomes including metastasis-free survival (MFS) and overall survival—especially in Chinese patients—is not well understood. Herein we report the long-term survival data of patients at a single institution in Hong Kong who received SRT to the prostatic fossa using modern RT techniques.

Using the MOSAIQ system (version 2.62, IMPAC Medical Systems, Inc.; Sunnyvale [CA], US), we identified 91 Chinese patients treated with postoperative RT to the prostatic fossa at Tuen Mun Hospital, Hong Kong, between 2006 and 2017. The treatment records and clinical data of these patients were reviewed. Two patients who received ART with undetectable PSA were excluded. Patients who had received androgen deprivation therapy (ADT) prior to SRT were also excluded. These selection criteria yielded 84 evaluable individuals who received SRT to the prostatic fossa alone for PSA failure (defined as detection of PSA concentration at 0.2 ng/mL, with a second confirmatory level detected at 0.2 ng/mL) more than 3 months after RP.

A planning computed tomographic scan was performed for each patient with 3-mm slice thickness, and the clinical target volume was determined with reference to one of the published consensus guidelines.9 10 11 The usual boundaries of the clinical target volume are: inferiorly, 5 mm below the urethral anastomosis; anteriorly, the posterior aspect of the symphysis pubis or the posterior third of the bladder; laterally, the medial border of the obturator internus and levator ani muscles; posteriorly, the anterior mesorectal fascia; and superiorly, 5 mm above the surgical bed. The planning target volume was defined as clinical target volume with a margin of 4 to 5 mm posteriorly and 0.7 to 1 cm in all other directions. Organs at risk, including the rectum, bladder, and bilateral femoral heads were contoured. Conformal radiotherapy or inverse planning techniques with intensity-modulated radiotherapy (IMRT) using seven to nine static beams were used before October 2010. After that, volumetric modulated arc radiotherapy (VMAT) was employed using the Pinnacle treatment planning system (Philips Medical Systems, Fitchburg [WI], US) with treatment delivered through one to two dynamic cone arcs.

Clinical data included age at SRT, time from surgery to RT (≤24 months vs >24 months), SRT dose, pre-SRT PSA level, and post-SRT nadir PSA. Pathological data consisted of pathological T stages (T2a vs T2b vs T2c vs T3a or T3b), ECE, seminal vesicle invasion, SM, and pathological Gleason scores (≤7 or ≥8).

After SRT, patients were followed up with PSA level checks every 3 months in the first 2 years, every 6 months from year 3 to year 5, then annually. A complete response was defined as an undetectable nadir PSA (<0.1 ng/mL). Biochemical progression (PSA failure) was defined as a rise of PSA level by 0.2 ng/mL above the nadir with a second confirmation at least 1 week apart.12 Biochemical progression-free survival (bPFS) was defined as the date from SRT completion to the first date of biochemical progression. Patients who showed biochemical progression or symptoms suggestive of metastasis received imaging studies at the discretion of the oncologist. Metastasis-free survival was defined as the date from SRT completion to the date of occurrence of metastasis on imaging. Patients who showed biochemical progression with or without metastasis were counselled on the use of ADT; ADT-free survival was defined as the date of SRT completion to the first date of ADT administration.

The Kaplan-Meier method was used to estimate bPFS, MFS, and ADT-free survival. Log-rank tests and Cox regression analysis were used to test the association between groups and oncologic outcomes. Covariates consisted of continuous variables, including patient age at SRT, SRT dose, and pre-SRT PSA, and discrete variables including post-SRT nadir PSA (detectable vs undetectable), pathological T stages (T2a vs T2b vs T2c vs T3a vs T3b), pathological Gleason score (≤7 vs ≥8), SM (negative vs positive), ECE (negative vs positive), seminal vesicle invasion (negative vs positive), and time of SRT (≤24 months after RP or >24 months after RP). Only variables that were significantly associated with outcomes on univariate analyses were further tested for association in multivariate analyses.

Statistical analyses were performed using IBM SPSS Statistics for Windows, version 24.0 (IBM Corp, Armonk [NY], US), and numerical data were presented according to Cole.13

The median age of the 84 patients was 68 years (range, 52-79 years) when they received SRT. The patients’ median pre-SRT PSA level was 0.4 ng/mL (range, 0.2-7.4 ng/mL). Of the patients, 63 (75%) had positive SM in their prostatectomy specimens. Extracapsular extension was detected in 25 (29.8%) patients. Pelvic lymph nodes of 41 patients were sampled during RP and were all found to be negative for malignancy. These and other pathological characteristics are summarised in Table 1. The median time from surgery to start of SRT was 18.4 months (range, 3.8-121 months).

Before October 2010, one patient was treated with conformal RT and 10 patients were treated with IMRT. Subsequently the other 73 patients were treated with VMAT. The median dose given to the prostatic fossa was 70 Gy (range, 64-76 Gy), with 66 (79%) patients receiving a dose of ≥70 Gy. The mean dose delivered using VMAT (69.5 Gy) was slightly higher than that delivered using IMRT/conformal RT (68.1 Gy) [independent-samples t test, t=2.1; P=0.028].

Of 84 patients, 47 (56%) had undetectable PSA levels (complete response; <0.1 ng/mL) after SRT. After a median follow-up of 48 months (range, 2-120 months), 25 (30%) patients had biochemical progression with an estimated 5-year bPFS of 62.7% (95% confidence interval [CI], 50.1-75.3%) [Fig 1a]. Among the 25 patients who developed biochemical progression after SRT, seven were found to have DM and subsequently received ADT, and five started ADT in the absence of DM, two of whom later developed DM and had their disease became castration-resistant. Overall, 12 patients received ADT and nine (11%) patients developed DM. The 5-year ADT-free survival and MFS were 83.5% (95% CI, 73.7-93.3%) and 86.7% (95% CI, 77.7-95.7%), respectively (Fig 1b, c). Notably, only six patients died, all from causes other than PCa.

On univariate analysis, a post-SRT nadir PSA ≥0.1 ng/mL, positive ECE, and bPFS ≤12 months were significantly associated with a shorter MFS (all P<0.001; Fig 2). Similarly, a post-SRT nadir PSA ≥0.1 ng/mL (P<0.001), positive ECE (P<0.001), negative SM (P=0.045), pathological Gleason score ≥8 (P=0.002), and time from surgery to SRT ≤24 months (P=0.008) were significant predictors of a shorter bPFS (Fig 3). The pre-SRT PSA level, age, and SRT dose were not associated with either MFS or bPFS in this cohort on univariate analysis. On multivariate analysis using the Cox regression method, negative SM (hazard ratio [HR]=2.7; 95% confidence interval [CI], 1.1-6.6), positive ECE (HR=4.6; 95% CI, 1.8-11.7), post-SRT nadir PSA ≥0.1 ng/mL (HR=17.3; 95% CI, 5.3-57.0), and time from surgery to SRT ≤24 months (HR=7.4; 95% CI, 2.2-24.0) retained significant association with a shorter bPFS (Table 2). There was no variable significantly associated with MFS after multivariate analysis.

Most patients who develop biochemical recurrence after RP for localised PCa remain asymptomatic for many years.14 However, patients with increasing PSA level are at high risk of developing DM. Salvage radiotherapy is effective in terms of biochemical control when it is administered at low PSA level. Stephenson et al12 reported a 6-year progression-free probability of 32% after SRT. In their multi-institutional retrospective cohort of 1603 consecutive patients from 17 North American tertiary referral centres who received SRT after RP for PSA recurrence between 1987 and 2005, the median dose was only 64.8 Gy (interquartile range, 63-66 Gy) delivered using older techniques. The 5-year bPFS of 62.7% in the present study is similar or better than those reported in western countries.12 15 16 This might be due to better selection of patients (most patients started SRT when their PSA level was ≤0.5 ng/mL), or the higher dose of SRT to the prostatic fossa (median 70 Gy). In our cohort, all patients but one were treated using IMRT/VMAT. Intensity-modulated radiotherapy was introduced in the 1990s and it has since enabled radiation oncologists to deliver higher doses of radiation to treat patients with PCa—including patients with residual disease at the prostatic fossa—without causing excessive radiation damage to healthy tissue.17 18 19 Volumetric modulated arc radiotherapy has recently attracted much interest because it can dynamically deliver a radiation dose during rotation of the gantry; this is also superior to IMRT in terms of its plan qualities and efficiency in the treatment of PCa.20 21

Pisansky et al22 reported that SRT doses of ≥66.0 Gy were associated with reduced cumulative incidence of biochemical progression. A systemic review by King23 provides level 2a evidence for escalated SRT dose of at least 70 Gy. A 2% improvement in relapse-free survival can be achieved for each additional Gy from 60 Gy to 70 Gy.23 However, higher SRT dose was not shown to be associated with better bPFS/MFS in our 84 patients by univariate analysis, because most (79%) had been treated with an SRT dose of ≥70 Gy, and the follow-up time may still be too short to demonstrate any further dose-response relationship. We postulated that such high-dose SRT can be delivered safely with modern techniques using VMAT, therefore our current usual prescribed dose is 70 Gy to the prostatic fossa, unless limited by dose constraints of the organ at risk. We have previously shown the efficiency and low toxicities using VMAT for SRT to the prostatic fossa.24 Longer follow-up is necessary to ensure that late complications are within safety limits.

Despite the success of SRT in biochemical control, some patients may develop further biochemical progression. In our present study, patients whose surgical pathology revealed negative margin and positive ECE had a shorter bPFS (HRs of 2.7 and 4.6, respectively). Patients who start SRT within 2 years of RP may also have a shorter PSA doubling time, leading to earlier detection of recurrence. These patients have a greater than 7-fold higher risk of biochemical regression after SRT than those with later recurrence. Salvage radiotherapy to the prostatic fossa alone cannot eradicate cancer that has spread outside the surgical bed after RP. In fact, negative SM, positive ECE, and shorter PSA doubling time are three of the many adverse factors which predict a shorter bPFS after SRT, using the nomogram proposed by Stephenson et al.25 However, we cannot demonstrate the importance of pre-SRT PSA level in our patient cohort because more than 65% of the patients had started SRT when their PSA level was ≤0.5 ng/mL.

Overall, the role of SRT in improving MFS and overall survival is less certain, because the disease can be indolent and mortality due to causes other than PCa is more likely in older patients. Patients also have other complications related to disease progression, such as painful bone metastasis. Efforts have been made to identify surrogate endpoints that can predict further disease progression, metastasis, and even cancer-related death after SRT. In a single institution review, Johnson et al26 reported approximately 50% of men experience further biochemical progression after SRT. Those who have a short interval to biochemical progression of ≤18 months after SRT are most likely to experience DM, PCa-specific mortality, and overall mortality. Bartkowiak et al27 reported on the long-term outcomes of patients with a median follow-up of 7 years (maximum, 14 years) after SRT. They found that a post-SRT nadir PSA <0.1 ng/mL was associated with improved bPFS and overall survival. The results of our univariate analysis support the abovementioned findings27 (Fig 2a, b). On multivariate analysis, we found that undetectable nadir PSA (<0.1 ng/mL) is the most important factor for predicting longer bPFS (Table 2). In the present study, of the 47 patients who achieved biochemical complete response after SRT, none developed DM. In contrast, among the 25 patients who had biochemical progression, nine whose disease progressed within 1 year after SRT eventually developed DM. Although our result of a 5-year MFS of nearly 90% is encouraging, with the median follow-up of only 4 years, we can hypothesise only that better biochemical control is correlated with improvements in other clinical outcomes. For patients whose PSA level does not become undetectable and rapidly rises within 1 year after SRT (bPFS ≤12 months), close monitoring for DM may be needed.

The improvement in overall survival and MFS of adjuvant ADT with SRT has been demonstrated by Shipley et al28 in a phase III study. However, ADT is not routinely recommended to our patients because of the known metabolic and cardiovascular toxicities and the negative impact on patients’ quality of life. In addition, most of our patients have fewer adverse features than those reported by Shipley et al.28 For patients with biochemical regression alone after SRT, we suggest monitoring for any site of disease recurrence such that further SRT could still be feasible. Nonetheless, we applied positron emission tomography with 68 Ga-labelled prostate-specific membrane antigen (PET-PSMA) to identify the site of recurrence in four of our patients when their PSA levels increased to ≥2.2 ng/mL (Table 3). All four patients were found to have DM which was not amenable to further local treatment and ADT had become their only option. It remains unclear whether PET-PSMA or other imaging studies at lower PSA levels are sensitive or useful enough to alter the management decision.29 Further research to study the use of novel radiological examinations in this situation is needed.

This is the first report to demonstrate the therapeutic effects in terms of bPFS and MFS of SRT in Chinese patients in a Hong Kong centre. Salvage radiotherapy is an effective local treatment that can prevent DM and avoid the need for ADT in most patients who have PSA failure after RP in Chinese patients. Our results appear to be better than those of some studies in western countries, in which older radiotherapy techniques and lower radiation doses were used. The limitations of our study include the retrospective design with lack of evaluation of patients’ reported outcome, small sample size, and short duration of follow-up. A multi-institutional study is recommended to collect more local data and experiences.

Concept or design: EKC Lee, Y Tung.
Acquisition of data: EKC Lee, AW Chan.
Analysis or interpretation of data: EKC Lee.
Drafting of the article: EKC Lee, WH Mui, FCS Wong.
Critical revision for important intellectual content: EKC Lee, WH Mui, FCS Wong.

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

All authors have no conflicts of interest to disclose. 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. An earlier version of this paper was presented as poster presentation at the 9th European Multidisciplinary Meeting on Urological Cancers, 16-19 November 2017, Barcelona, Spain.

The study was conducted with approval from the New Territories West Cluster Clinical and Research Ethics Committee.

1. Hospital Authority, HKSAR Government. Leading cancer sites in Hong Kong in 2014. Available from: http://www3.ha.org.hk/cancereg/pdf/top10/rank_2014.pdf. Accessed Jul 2017.

2. Poon DM, Chan SL, Leung CM, et al. Efficacy and toxicity of intensity-modulated radiation therapy for prostate cancer in Chinese patients. Hong Kong Med J 2013;19:407-15. Crossref

3. Ng AT, Tam PC. Current status of robot-assisted surgery. Hong Kong Med J 2014;20:241-50. Crossref

4. Parker C, Clarke N, Logue J, et al. RADICALS (Radiotherapy and Androgen Deprivation In Combination After Local Surgery). Clin Oncol (R Coll Radiol) 2007;19:167-71. Crossref

5. Richaud P, Sargos P, Henriques de Figueiredo B, et al. Postoperative radiotherapy of prostate cancer [in French]. Cancer Radiother 2010;14:500-3. Crossref

6. Trans Tasman Radiation Oncology Group. RAVES trial: radiotherapy—adjuvant versus early salvage. Available from: http://www.clinicaltrial.gov/ct2/show/NCT00860652. Accessed Jul 2017.

7. Valicenti RK, Thompson I Jr, Albertsen P, et al. Adjuvant and salvage radiation therapy after prostatectomy: American Society for Radiation Oncology/American Urological Association guidelines. Int J Radiat Oncol Biol Phys 2013;86:822-8. Crossref

8. Wiegel T, Lohm G, Bottke D, et al. Achieving an undetectable PSA after radiotherapy for biochemical progression after radical prostatectomy is an independent predictor of biochemical outcome—results of a retrospective study. Int J Radiat Oncol Biol Phys 2009;73:1009-16. Crossref

9. Michalski JM, Lawton C, El Naqa I, et al. Development of RTOG consensus guidelines for the definition of the clinical target volume for postoperative conformal radiation therapy for prostate cancer. Int J Radiat Oncol Biol Phys 2010;76:361-8. Crossref

10. Poortmans P, Bossi A, Vandeputte K, et al. Guidelines for target volume definition in post-operative radiotherapy for prostate cancer, on behalf of the EORTC Radiation Oncology Group. Radiother Oncol 2007;84:121-7. Crossref

11. Sidhom MA, Kneebone AB, Lehman M, et al. Post-prostatectomy radiation therapy: consensus guidelines of the Australian and New Zealand Radiation Oncology Genito-Urinary Group. Radiother Oncol 2008;88:10-9. Crossref

12. Stephenson AJ, Scardion PT, Kattan MW, et al. Predicting the outcome of salvage radiation therapy for recurrent prostate cancer after radical prostatectomy. J Clin Oncol 2007;25:2035-41. Crossref

13. Cole TJ. Too many digits: the presentation of numerical data. Arch Dis Child 2015;100:608-9. Crossref

14. Pound CR, Partin AW, Eisenberger MA, Chan DW, Pearson JD, Walsh PC. Natural history of progression after PSA elevation following radical prostatectomy. JAMA 1999;281:1591-7. Crossref

15. Geinitz H, Riegel MG, Thamm R, et al. Outcome after conformal salvage radiotherapy in patients with rising prostate-specific antigen levels after radical prostatectomy. Int J Radiat Biol Oncol Phys 2012;82:1930-7. Crossref

16. Fossati N, Karnes RJ, Boorjian SA, et al. Long-term impact of adjuvant versus early salvage radiation therapy in pT3N0 prostate cancer patients treated with radical prostatectomy: results from a multi-institutional series. Eur Urol 2017;71:886-93. Crossref

17. Goldin GH, Sheets NC, Meyer A, et al. Patterns of intensity modulated radiation therapy (IMRT) use for the definitive and postoperative treatments of prostate cancer: a SEER-medicare analysis. Int J Radiat Oncol Biol Phys 2011;81(2 Suppl):S408. Crossref

18. Nath SK, Sandhu AP, Rose BS, et al. Toxicity analysis of postoperative image-guided intensity-modulated radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys 2010;78:435-41. Crossref

19. Ost P, De Troyer B, Fonteyne V, Oosterlinck W, De Meerleer G. A matched control analysis of adjuvant and salvage high-dose postoperative intensity-modulated radiotherapy for prostate cancer. Int J Radiat Oncol Biol Phys 2011;80:1316-22. Crossref

20. Kopp RW, Duff M, Catalfamo F, Shah D, Rajecki M, Ahmad K. VMAT vs. 7-field-IMRT: assessing the dosimetric parameters of prostate cancer treatment with a 292-patient sample. Med Dosim 2011;36:365-72. Crossref

21. Palma D, Vollans E, James K, et al. Volumetric modulated arc therapy for delivery of prostate radiotherapy: reduction in treatment time and monitor unit requirements compared to intensity modulated radiotherapy. Int J Radiat Oncol Biol Phys 2008;72(1 Suppl):S312. Crossref

22. Pisansky TM, Agrawal S, Hamstra DA, et al. Salvage radiation therapy dose response for biochemical failure of prostate cancer after prostatectomy—A multi-institutional observational study. Int J Radiat Oncol Biol Phys 2016;96:1046-53. Crossref

23. King CR. The dose-response of salvage radiotherapy following radical prostatectomy: a systemic review and meta-analysis. Radiother Oncol 2016;121:199-203. Crossref

24. Lee EK, Yuen KK, Mui WH, et al. Salvage radiotherapy to the prostatic fossa using volumetric-modulated arc therapy: early results. Hong Kong J Radiol 2013;16:191-7. Crossref

25. Stephenson AJ, Shariat SF, Zelefsky MJ, et al. Salvage radiotherapy for recurrent prostate cancer after radical prostatectomy. JAMA 2004;291:1325-32. Crossref

26. Johnson S, Jackson W, Li D, et al. The interval to biochemical failure is prognostic for metastatic, prostate cancer-specific mortality, and overall mortality after salvage radiation therapy for prostate cancer. Int J Radiat Oncol Biol Phys 2013;86:554-61. Crossref

27. Bartkowiak D, Bottke D, Thamm R, Siegmann A, Hinkelbein W, Wiegel T. The PSA-response to salvage radiotherapy after radical prostatectomy correlates with freedom from progression and overall survival. Radiother Oncol 2016;118:131-5. Crossref

28. Shipley WU, Seiferheld W, Lukka HR, et al. Radiation with or without antiandrogen therapy in recurrent prostate cancer. N Engl J Med 2017;376:417-28. Crossref

29. Perera M, Papa N, Christidis D, et al. Sensitivity, specificity, and predictors of positive 68Ga-prostate-specific membrane antigen positron emission tomography in advanced prostate cancer: a systemic review and meta-analysis. Eur Urol 2016;70:926-37. Crossref

Elderly patients have multiple co-morbidities and they are consequently prone to multiple medication use. Inappropriate medication use is common among hospitalised older adults. The number of drugs taken is one of the important determinants of risk for receiving an inappropriate medication.1 There is a high prevalence of unnecessary drug use in frail older people. In one hospital study, 44% of patients were prescribed at least one unnecessary drug, with the most common reason being lack of indication.2 The most commonly prescribed unnecessary drug classes were gastrointestinal, central nervous system, and therapeutic nutrients/minerals.2 Appropriate use of drugs is particularly important in the frail older people who are especially at risk of adverse drug reactions.3 It has been shown that implementation of a clinical pharmacist service has a positive effect on medication use and health care service utilisation among hospitalised patients.4 5 A local study in a geriatric hospital demonstrated the effectiveness of a drug rationalisation programme with involvement of a clinical pharmacist in reducing the incidence of polypharmacy and inappropriate medications.6 Interacting with the health care team on patient rounds, interviewing patients, reconciling medications, and providing patient discharge counselling and follow-up all resulted in improved outcomes.7 It is for this reason that patient safety strategies encourage the use of medication reconciliation and clinical pharmacists in health care systems to reduce adverse drug events.8 9 10

There is not much information about the effectiveness of a medical review programme among hospitalised elderly patients in Hong Kong. Two recent local reports that examined the effects of a clinical pharmacist–led medication review on hospital readmissions showed conflicting results and did not specifically address elderly patients.11 12 We therefore conducted a prospective controlled study to investigate the effectiveness of a comprehensive pharmacist intervention on medication use and hospital readmission among a group of geriatric in-patients in Hong Kong.

This prospective controlled study was conducted in the geriatric unit of a regional hospital in Hong Kong. The unit has 38 in-patient beds and admits older people aged 65 years or above who are transferred from an acute hospital after initial stabilisation of medical and/or geriatric problems. The unit admits more than 1000 patients per year and provides medical treatment, rehabilitation, and discharge planning services by a multidisciplinary team composed of a geriatrician, residents, nurses, physiotherapists, occupational therapists, and medical social workers. All patients admitted to the unit during December 2013 to September 2014 were included. Patients were excluded if they refused to participate, were terminally ill with a life expectancy of less than 3 months, or if they had already received pharmacist intervention in another hospital prior to this admission. Eligible subjects were assigned to an intervention or control group according to the admission day of the week. Those who were admitted on Monday through Thursday were assigned to the intervention group, and those admitted on Friday through Sunday to the control group. This arrangement was to ensure that pharmacist intervention could be initiated promptly within 48 hours of patient admission. Demographic data, functional status, co-morbidities, and number of drugs on admission were collected at admission.

The intervention was conducted by a pharmacist who was present in the unit from Monday to Saturday. The pharmacist provided pharmaceutical care from admission to discharge. Interventions performed by the pharmacist consisted of the following:

(1) Medication reconciliation on admission to identify unintended discrepancies between medications prescribed on admission and the usual medications prior to admission—sources to assist medication reconciliation included: electronic patient record; patient’s ward case notes; interview with patient and/or patient carer. The number and type of unidentified discrepancies were recorded.

(2) Medication review to check for medication appropriateness on admission and also at discharge—medication appropriateness was assessed by the Medication Appropriateness Index (MAI).13 There are 10 criteria to assess for appropriateness, namely indication, effectiveness, dosage, correct direction, practical direction, drug-drug interaction, drug-disease interaction, duplication, duration, and expense. For a drug item coded as ‘inappropriate’, relative weights for each criterion would apply. A sum of MAI scores could then be calculated to give a score ranging from 0 to 18. The higher the score, the more inappropriate the drug. Recommendations from the pharmacist after the reconciliation and medication review in the intervention group were then communicated to the in-charge doctor via a written note in the medical records. Recommendations were reinforced verbally if deemed appropriate by the pharmacist.

(3) Pharmacist counselling on admission and also at discharge was provided to improve patients’ drug knowledge to ensure proper use of drugs and compliance after discharge. A discharge counselling service was provided for all patients who returned home. The counselling included any changes to drug regimen; an explanation of each drug’s indication; any untoward effects that might occur and when to seek medical advice; and drug storage and administration instructions. To ensure patient understanding, written information such as patient information leaflets were given to patients and their carers to remind them of the correct drug regimen. If the patient was illiterate, a simple diagram was drawn on drug labels to demonstrate the time of day and number of tablets to be taken. If necessary, individualised pictorial schedules with drug images and administration instructions could be produced for patients and their carers. The assistance of a family member or external care services such as a community nurse was enlisted if the patient was found to have compliance issues.

The control group received routine clinical services. Records of the control group were retrospectively reviewed by the pharmacist after patient discharge to check for medication appropriateness on admission and also at discharge. The primary outcome measure was the appropriateness of prescription as measured by the MAI. Secondary outcomes included the acceptance rate by physicians, number of subjects with unintended discrepancies, patient satisfaction with the programme (for those home-living only), and unplanned hospitalisations 1 and 3 months after discharge.

A sample size of 98 patients per group was required to have 85% power to detect an effect size of 0.9 on the MAI. Our sample size was finally set at 210 patients to account for loss of participants due to dropout or death. This sample size was comparable with a study by Spinewine et al14 in which MAI was used as one of the tools to assess appropriateness of prescribing in an acute geriatric care unit and 203 patients were recruited. Our study included 212 patients and was expected to have adequate statistical power to detect differences between groups. Descriptive analyses were performed and included the number and types of unintended discrepancies, MAI score upon admission and at discharge, types of drug-related problems, number of interventions made by pharmacists, and number of recommendations accepted by doctors and implemented. Outcomes for the two groups before and after the programme were compared using the t test and Chi squared test. The Statistical Package for the Social Sciences (Windows version 17.0; SPSS Inc, Chicago [IL], United States) was used and a P value of <0.05 was regarded as statistically significant. The study was approved by the Cluster Research Ethics Committee of the Hospital Authority Hong Kong West Cluster. Written consent was obtained from the patient or their caregiver. The absence of pharmacist intervention in the control group was considered acceptable because a pharmacy service was not a part of routine care at the institution.

Figure 1 summarises the patient flow from recruitment to hospital discharge, the components of the medication review programme, and the planned outcome measures. A total of 212 patients were recruited. There were 108 subjects in the intervention group and 104 in the control group (Fig 2). There were no statistical differences in the baseline characteristics of patients (Table 1).

On admission, 51.9% (56/108) of the intervention group and 58.7% (61/104) of the control group had at least one drug classified as inappropriate (P=0.319). Overall, 1996 drug items were reviewed by a pharmacist on admission of which 1020 were from the intervention group and 976 from the control group. Among them, 9.3% and 11.1% of the drugs, respectively, were classified as inappropriate (P=0.282). In the intervention group, 93 recommendations were made by the pharmacist of which 60 (64.5%) were accepted by the physicians and implemented. The mean (standard deviation) MAI score per patient was 2.19 (3.03) in the intervention group and 2.28 (3.09) in the control group (P=0.841). The mean MAI score per drug was 0.23 (0.30) in the intervention group and 0.25 (0.31) in the control group (P=0.628) [Table 2].

After the program and at discharge, the proportion of subjects with inappropriate medications in the intervention group was significantly lower than that in the control group (28.0% vs 56.4%; P<0.001). Among the 1999 drug items reviewed by the pharmacist on patient discharge, 3.5% (37 of 1048) of the intervention group and 9.7% (92 of 951) of the control group were classified as inappropriate (P<0.001). The intervention group also had a significantly lower MAI score per patient (0.95 (2.02) vs 2.02 (2.53); P<0.001) and MAI score per drug (0.09 (0.17) vs 0.24 (0.30); P<0.001) implying a significant reduction in medication inappropriateness after the pharmacist medication review (Table 2).

Types of inappropriateness according to the MAI in the two groups are illustrated in Figure 3. In both the intervention and control groups, the common causes were indication, effectiveness, dosage, practical direction, duration, and expense. After the programme, there was a significant reduction in the number of these drug-related problems in the intervention group.

Among the 108 subjects in the intervention group, 19.4% had at least one unintended discrepancy in medications, involving a total of 43 drug factors. The majority (90.7%) of these factors were omission of drugs, and 4.6% were due to inappropriate dosages. Of all the drug factors involved, 69.8% involved prescribed drugs from hospitals, 25.6% were from a private clinic, and 4.6% were over-the-counter drugs. Overall, 41 recommendations were made, of which 32 (78.0%) were accepted by physicians and implemented.

Contact was made with 50 of the 90 non-institutionalised subjects 1 month after discharge to assess satisfaction with the programme. Of those contacted, 98.0% were satisfied with the programme and only one (2.0%) patient expressed a neutral opinion.

There were no statistical differences in the length of hospital stay, in-patient mortality, or mortality at 1 month or 3 months after discharge. There was also no statistical difference in the number of attendances at the accident and emergency department 1 month or 3 months after discharge or in the unplanned hospital readmission rate at 3 months after discharge. The unplanned hospital readmission rate 1 month after discharge, however, was significantly lower in the intervention group than that in the control group (13.2% vs 29.1%; P=0.005) [Table 3].

To the best of our knowledge, this is the first local prospective controlled study to investigate the effectiveness of a pharmacist-led medication review programme on medication appropriateness and clinical outcomes among geriatric in-patients in Hong Kong. This study has demonstrated superior outcomes that favour a pharmacist-led intervention. There was a substantial reduction in the use of inappropriate medications and all-cause unscheduled readmissions 1 month after hospital discharge. Nonetheless, analysis of length of hospital stay, number of all-cause emergency department visits, and mortality rate favoured neither the intervention nor the usual pharmacist care.

This study showed that one in five geriatric in-patients had an unintended medication discrepancy on admission. This figure was slightly higher than that found in a group of 3317 hospitalised medical patients (13%) over 1 year in an acute hospital in Hong Kong by Kwok et al.15 Subjects in our study were all elderly patients, whereas those in Kwok et al’s study were adults of all ages. Elderly subjects tend to have polypharmacy and thus are more vulnerable to unintended medication discrepancy when they move in and out of hospital or are transferred to another health care unit for further care. Unjustifiable medication discrepancies account for more than half of the medication errors that occur during transition of care and up to one third have the potential to cause harm.16 17 This does not bode well for our elderly patients with multiple co-morbidities.

Up to 30% of the discrepancies in our study involved medications that had been prescribed by private practitioners or purchased over-the-counter. Unlike medications prescribed from the Hospital Authority, these medications might be overlooked unless the admitting doctor specifically asks for a detailed drug history from the patient. Knowing the medication history and hence resuming these medications are important if new health problems are to be prevented. Pharmacist-led medication reconciliation is therefore a critical process that can enhance patient medication safety by compiling a complete and accurate medication list for patients in hospital.

This study revealed that more than half of the subjects (55.2%) received inappropriate medications. The majority of reasons for inappropriateness related to effectiveness, dosage, practical directions, and expense as reflected by the MAI. The inappropriate dosage and the questionable effectiveness might lead to not only failed pharmacological effects, but also potentially an untoward adverse drug reaction, especially in elderly individuals with pre-existing organ dysfunction.18 When a medication is not used according to the practical directions, it may lead to patient non-compliance. Optimising outcomes while reducing costs are the keys for medication management in today’s health care environment.19 Often there are several choices of drugs available to treat a disease or health condition and some are more expensive than others. The involvement of a ward-based pharmacist to review medication can enhance the use of appropriate medications in hospitalised patients and potentially reduce medication costs.

Following the medication review (60/93) and medication reconciliation (32/41), there were 92 recommendations that were accepted by the physician. The overall acceptance rate by physicians and the implementation of pharmacist recommendations in our study was 68.7% (92/134). This figure ranged from 39% to 100% in a previous systematic review of 32 studies.4 Our study did not specifically record recommendations accepted by physicians but not implemented. Hence the acceptance rate in our study may be underestimated. Nevertheless, the clinical pharmacist is encouraged to discuss the medication-related problems in person with the physician as well as contacting the patient in order to enhance the implementation rate.4

Pharmacy departments within the public hospital system in Hong Kong have strived to implement the aforementioned patient safety strategies in different specialties. Nonetheless, this is not a standard practice simply because of insufficient pharmacist staffing resources. In some hospitals, a pharmacist service is provided in wards, but this does not apply in all cases and is not standardised. Experience of a pharmacist-led medication reconciliation service from an acute teaching hospital in Hong Kong showed promising results over a 1-year trial run with high acceptance and recognition by other health care professionals.16 It is hoped that the clinical role of clinical pharmacists in patient medication management in hospitals can be encouraged. Another local study has demonstrated a positive impact on medication safety in patients with diabetes by pharmacists’ intervention in collaboration with a multidisciplinary team.20 The feasibility of incorporating a pharmacist as part of a multidisciplinary team of health care professionals must be explored in geriatric wards in Hong Kong. With increasing life expectancy, the expanding elderly population will equate to an increase in morbidity and mortality owing to drug-related problems where the need for trained health care professionals to perform medication reviews will be in even greater demand. To enhance safe drug use with limited resources, a systematic approach must be adopted to cover all aspects that affect drug therapy.

In terms of the impact on health care services utilisation, a recent systematic review and meta-analysis of the effectiveness of a pharmacist-led medication reconciliation programme revealed a substantial reduction in the rate of all-cause readmissions (19%), all-cause emergency department visits (28%), and adverse drug event–related hospital visits (67%).21 Our study revealed a significant reduction in unplanned hospital admissions (all-cause admission) at 1 month but not at 3 months. This implication might be due to an inadequate sample size to show the difference at 3 months. Alternatively, it might also imply that pharmacist intervention needs to be continued after patient discharge in order to have a sustained effect. This is supported by a study by Schnipper et al22 in which pharmacist intervention after patient discharge was associated with a lower rate of preventable adverse drug events 30 days after hospital discharge.

During the pharmacist review, cost-effectiveness of drug use was assessed through MAI. Alternative options such as less-expensive formulations or drugs but of the same quality would be recommended to the doctor in-charge. This was a means of encouraging cost-effective use of drugs in a hospital. Furthermore, the reduction in unscheduled hospital readmissions in the intervention group implies a potential saving in hospital costs. Although a detailed analysis was not performed in this study, a rough estimation is that nearly HK$2 million may be saved annually as a result of lower drug costs and reduced hospital admissions, even after considering the cost of employing a pharmacist. The estimation was based on the following calculations. The estimated drug saving as a result of a switch to a more cost-effective alternative was HK$7500 among the 108 patients in the intervention group. This can be projected to a saving of about HK$69 500 in a unit that admits 1000 patients annually. In the current study, there were 16 fewer readmissions in the intervention group compared with the control group. Assuming a daily cost of an acute hospital bed is HK$4680 and the mean length of hospital stay is 3 days, this equates to a potential saving of about HK$2 080 000 per year in a unit that admits 1000 patients annually. If it is assumed that a pharmacist spends 30 minutes for each patient at an hourly salary of HK$433, the projected cost of an additional pharmacist to run the intervention would be HK$216 500 per year. The net annual saving of this programme to serve 1000 patients in this unit would thus still be close to HK$2 million.

This study had several limitations. First, a substantial proportion (35%) of all the admitted patients were not screened by a pharmacist on admission. This was due to a temporary pause in subject recruitment when patients were admitted on public holidays, when the pharmacist was on holiday or when she had to relieve another pharmacist in the hospital. Moreover, a substantial proportion (44%) of eligible subjects were not included owing to no consent or refusal. These factors might have resulted in selection or self-selection bias. Second, subject recruitment was not randomised, but done according to the day of admission. This might be a source of bias. Nevertheless, this would have minimal influence on the outcomes, as the baseline characteristics of the intervention and control groups were comparable. Third, the pharmacist who carried out the review and data extraction was not blinded to the study hypothesis and the group status of the subjects. This could potentially lead to information bias, although this might be partially offset by the fact that the majority of the information or data on the outcome measures were taken with reference to a well-established and validated tool. Fourth, this study was performed in a single unit, so generalisation to other settings is not possible. Fifth, MAI is an implicit tool that is subjective. A single pharmacist as the rater might limit the reliability of the assessment results. Nevertheless, the more explicit tools of STOPP/START criteria23 had also been referred to in addition to the MAI during the review process. Sixth, this study only addressed appropriateness of drug use, whereas underuse of drugs was not investigated. Finally, this study could not conclude a causal relationship between the reduction in inappropriate medications and the reduction in unscheduled hospital readmissions because there were several components in the intervention that included a medication review, medication reconciliation, and discharge counselling. It is difficult to be certain which of these components alone or in combination gave rise to the positive outcome of this study.

On the other hand, there were several strengths in this study. This was the first prospective controlled study of the effect of a pharmacist-led medication review programme on medication use and health services utilisation involving over 200 Chinese elderly patients in Hong Kong. Second, a well-validated tool was used to assess medication appropriateness. The use of the MAI tool focused on the patient and the entire medication regimen. Third, there was a comprehensive review of outcomes including quality of prescribing, health services utilisation, mortality, length of hospital stay, and patient satisfaction.

This study supported the role of a hospital-based clinical pharmacist to enhance appropriate medication use among elderly Chinese in-patients. A systematic medication review programme in a geriatric unit resulted in a reduced number of drug omissions and fewer inappropriate medications. The service provided by the clinical pharmacist and supported by geriatricians was welcomed by patients and their carers. Together with the potential to reduce hospital readmissions and their associated cost, it is hoped that an in-hospital pharmacist-led medication review programme can be recognised as one of the important strategies to enhance the safety and quality of prescription among elderly patients in hospitals. It is strongly recommended that these programmes be standardised and implemented in all medical and geriatric wards in Hong Kong. Future studies should recruit a larger sample size in a randomised controlled design in other geriatric hospital settings to reiterate our findings. Furthermore, these studies might consider including adverse drug event–related hospital visits as one of the outcome measures.

All authors have disclosed no conflicts of interest.

1. Onder G, Landi F, Cesari M, Gambassi G, Carbonin P, Bernabei R; Investigators of the GIFA Study. Inappropriate medication use among hospitalized older adults in Italy: results from the Italian Group of Pharmacoepidemiology in the Elderly. Eur J Clin Pharmacol 2003;59:157-62. Crossref

2. Hajjar ER, Hanlon JT, Sloane RJ, et al. Unnecessary drug use in frail older people at hospital discharge. J Am Geriatr Soc 2005;53:1518-23. Crossref

3. Routledge PA, O’Mahony MS, Woodhouse KW. Adverse drug reactions in elderly patients. Br J Clin Pharmacol 2004;57:121-6. Crossref

4. Graabaek T, Kjeldsen LJ. Medication reviews by clinical pharmacists at hospitals lead to improved patient outcomes: a systematic review. Basic Clin Pharmacol Toxicol 2013;112:359-73. Crossref

5. Gillespie U, Alassaad A, Henrohn D, et al. A comprehensive pharmacist intervention to reduce morbidity in patients 80 years or older: a randomized controlled trial. Arch Intern Med 2009;169:894-900. Crossref

6. Chan FH, Pei CK, Chiu KC, Tsang EK. Strategies against polypharmacy and inappropriate medication—are they effective? Australas J Ageing 2001;20:85-9. Crossref

7. Kaboli PJ, Hoth AB, McClimon BJ, Schnipper JL. Clinical pharmacists and inpatient medical care: a systematic review. Arch Intern Med 2006;166:955-64. Crossref

8. The ACHS EQuIP5 (5th edition of the ACHS Evaluation and Quality Improvement Program) clinical standard 1.5, criterion 1.5.1.: Medication are managed to ensure safe and effective consumer/patient outcomes. Sydney: The Australian Council on Health Care Standards; 2010.

9. Medication reconciliation. Patient Safety Network. Agency for Healthcare Research and Quality. US Department of Health & Human Services. Available from: http://psnet.ahrq.gov/primer.aspx?primerID=1. Accessed 11 Jun 2017.

10. Gurwitz J, Monane M, Monane S, Avorn J. Polypharmacy. In: Morris JN, Lipsitz LA, Murphy K, Bellville-Taylor P, editors. Quality care in the nursing home. St. Louis, MO: Mosby-Year Book; 1997: 13-25.

11. Wong CY, Lee TO, Lam MP. Pharmacist discharge intervention programme to reduce unplanned hospital use in patients with polypharmacy. Hong Kong Pharm J 2017;24(Suppl 1):S18.

12. Chan HH. To reduce avoidable readmission of patients who are categorized as high risk by 30-day hospital readmission model at medical ward of United Christian Hospital through medication reconciliation and discharge counseling. Hong Kong Pharm J 2017;24(Suppl 1):S21.

13. Hanlon JT, Schmader KE, Samsa GP, et al. A method for assessing drug therapy appropriateness. J Clin Epidemiol 1992;45:1045-51. Crossref

14. Spinewine A, Swine C, Dhillon S, et al. Effect of a collaborative approach on the quality of prescribing for geriatric inpatients: a randomized, controlled trial. J Am Geriatr Soc 2007;55:658-65. Crossref

15. Kwok CC, Mak WM, Chui CM. Implementational experience of medical reconciliation in Queen Mary Hospital. Hong Kong Pharm J 2009;16:50-2.

16. Rozich JD, Howard RJ, Justeson JM, Macken PD, Lindsay ME, Resar RK. Standardization as a mechanism to improve safety in health care. Jt Comm J Qual Saf 2004;30:5-14. Crossref

17. Cornish PL, Knowles SR, Marchesano R, et al. Unintended medication discrepancies at the time of hospital admission. Arch Intern Med 2005;165:424-9. Crossref

18. Cameron KA. Caregivers’ guide to medications and aging. National Center on Caregiving, Family Caregiver Alliance 2004. Available from: https://www.caregiver.org/caregiver%CA%BCs-guide-medications-and-aging. Accessed 11 Jun 2017.

19. Splawski J, Minger H. Value of the pharmacist in the medication reconciliation process. P T 2016;41:176-8.

20. Chung AY, Anand S, Wong IC, et al. Improving medication safety and diabetes management in Hong Kong: a multidisciplinary approach. Hong Kong Med J 2017;23:158-67. Crossref

21. Mekonnen AB, McLachlan AJ, Brien JA. Effectiveness of pharmacist-led medication reconciliation programmes on clinical outcomes at hospital transitions: a systematic review and meta-analysis. BMJ Open 2016;6:e010003. Crossref

22. Schnipper JL, Kirwin JL, Cotugno MC, et al. Role of pharmacist counseling in preventing adverse drug events after hospitalization. Arch Intern Med 2006;166:565-71. Crossref

23. Lam MP, Cheung BM. The use of STOPP/START criteria as a screening tool for assessing the appropriateness of medications in the elderly population. Expert Rev Clin Pharmacol 2012;5:187-97. Crossref

Periprosthetic joint infection (PJI) is an uncommon but serious complication after total knee arthroplasty (TKA). Treatment is often challenging and has a major impact on the patient. Multiple operations are often required and patients may suffer from a long period of disability. Moreover, PJI incurs considerable health care costs.1 2 3 Therefore, multiple strategies including antibiotic prophylaxis, body exhaust systems, and laminar airflow systems have been developed to reduce the incidence of PJI. Studies have also identified modifiable risk factors for PJI after elective total joint replacement,4 5 6 7 8 9 10 11 12 13 14 with the aim of further reducing the incidence of PJI. However, local data on the risk factors and bacteriological features associated with PJI are still lacking.

This study had several aims. First, it aimed to provide the most up-to-date local data on incidence of and risk factors for PJI, including age, sex, presence of diabetes, presence of rheumatoid arthritis, and one-stage bilateral TKA. Second, this study aimed to provide an update on the bacteriology of PJI after elective primary TKA and to examine the association between potential risk factors and bacteriology. Third, we attempted to determine which risk factors, including bacteriology, were more likely to be associated with early-onset infection after elective primary TKA.

It is hoped that risk factors can be optimised or modified to prevent infection after TKA. Furthermore, an improved understanding of local bacteriological patterns and their relationship with various risk factors can help guide antimicrobial therapy.

We reviewed 2543 elective primary TKAs performed at the Queen Mary Hospital, Hong Kong, from 1993 to 2013. Data were collected by an infection control nurse of the Department of Microbiology who was blinded to the study objectives. The cohort data were collected from the Hong Kong Hospital Authority’s Clinical Data Analysis and Reporting System, the Infection Control Team, and the hospital’s joint replacement division registry. The keywords used in the data search were ‘periprosthetic joint infection’, ‘total knee arthroplasty’, and ‘surgical site infection’. Revision arthroplasties and knee arthroplasties for malignant conditions were excluded from the study. In patients with a history of native joint infection, elective primary TKA was performed only after eradication of the infection. Patients with active bacteraemia were also precluded from elective primary TKA until they were infection-free. There were no cases of severe immunosuppression. In relation to infection control, the majority of perioperative protocols for primary TKA were the same throughout the study period. Preoperatively, intravenous antibiotic prophylaxis (1 g of cefazolin) was given within 1 h before skin incision. In patients with penicillin allergy, other antibiotics were prescribed as appropriate. Intra-operatively, laminar airflow and body exhaust systems were used. There was no routine use of antibiotic-loaded cement or postoperative antibiotics. Postoperative wound management was the same throughout the study period.

Cohort characteristics, occurrence of PJI, and bacteriological data were retrieved. Bacterial type was defined as infection with skin flora or non-skin flora. Skin flora included methicillin-susceptible Staphylococcus aureus (MSSA), methicillin-resistant S aureus (MRSA), methicillin-susceptible coagulase-negative staphylococci (MSCNS), and methicillin-resistant coagulase-negative staphylococci (MRCNS). Other organisms were considered non-skin flora.

The following potential risk factors for PJI were analysed: age, sex, presence of diabetes, presence of rheumatoid arthritis, and one-stage bilateral TKA. They were examined by univariable analyses and then multivariable logistic regression to identify potential predictors of PJI, while controlling for confounders. We also studied the association of those potential risk factors with bacteriology and with the timing of infection onset; culture-negative PJI was excluded from these analyses. According to a working party convened by the Musculoskeletal Infection Society in 2014,15 PJI that occurs within 90 days of the index operation is considered early-onset infection, whereas PJI that occurs later is considered late-onset infection.

Both univariable and multivariable logistic regression in this study used the simultaneous entry method, with covariates of age (as a continuous variable) and sex, diabetes, rheumatoid arthritis, and one-stage bilateral TKA (as dichotomous variables). Outcomes are presented as odds ratios (ORs) with 95% confidence intervals (CIs). The regression model and data fitting were assessed using the Hosmer–Lemeshow goodness-of-fit test, and diabetes and one-stage bilateral TKA were excluded from the final model because of poor goodness-of-fit. For associations between potential risk factors and bacteriology and between potential risk factors and early onset of infection, only univariable analyses were used owing to small numbers of events. Categorical variables were compared with the chi-square test, whereas age was compared with the independent t test (two-tailed). Significance was assumed if P<0.05. All statistical analyses were conducted using SPSS version 22.0 (IBM Corporation, Armonk [NY], United States). The study was conducted in accordance with the principles outlined in the Declaration of Helsinki.

The incidence of PJI in our series was 1.34% (n=34). The incidence of early-onset infection was 0.39% (n=10) and that of late-onset infection was 0.94% (n=24). Among the cases PJI, 29.4% were early-onset infection. Early-onset infection occurred within a median of 17 days after arthroplasty (interquartile range, 9-32 days). Late-onset infection occurred within a median of 1 year and 8 months after arthroplasty (interquartile range, 7 months to 2 years and 11 months). Fifty-nine percent of infections occurred in the first year of surgery, whereas 74% occurred in the first 2 years.

The mean (standard deviation) age was 69 (9) years, with a range from 21 to 91 years; age followed a normal distribution. Overall, PJI developed in 10 males (1.9%) and 24 females (1.2%). In the one-stage bilateral TKA group, PJI occurred in 13 knees (1.2%). For the single-side TKA group, 21 knees (1.4%) developed PJI. Nine patients with diabetes (1.9%) and 25 patients without diabetes (1.2%) developed PJI. The highest rate of PJI, at 3.1%, was found in patients with rheumatoid arthritis, compared with 1.2% in patients without rheumatoid arthritis. The descriptive data are summarised in Table 1.

The most frequent causative organism was MSSA (26.5%, n=9), followed by MRSA (17.6%, n=6), Streptococcus spp (8.8%, n=3), MSCNS (5.9%, n=2), Escherichia coli (5.9%, n=2), Salmonella (5.9%, n=2), MRCNS (2.9%, n=1) and Mycobacterium tuberculosis (2.9%, n=1), The three cases of streptococcal infection comprised two Streptococcus dysgalactiae infections and one Streptococcus agalactiae infection. Culture-negative PJI comprised 23.5% of cases (n=8). Methicillin-resistant strains constituted 39% of all staphylococcal organisms. There was no significant association between the potential risk factors and skin flora infection (Table 2).

Rheumatoid arthritis was a significant risk factor of PJI in the univariable analysis, with an OR of 2.67 (95% CI, 1.15-6.20; P=0.02), as well as in the multivariable analysis, with an OR of 3.12 (CI, 1.29-7.56; P=0.01) [Table 3]. Being male (OR=1.9; P=0.11 in the multivariable analysis) and having diabetes (OR=1.54; P=0.27 in the univariable analysis) were not significantly associated with PJI.

Age (P=0.655), sex (P=0.961), diabetes (P=0.462), and rheumatoid arthritis (P=0.315) were not associated with early-onset infection (Table 4). Infection caused by skin flora was associated with early-onset infection (P=0.099), but the association was not statistically significant.

In this study, the incidence of PJI after primary TKA was 1.34% and the incidence of early-onset infection was 0.39%. The majority of PJIs (70%) were late-onset infections. The reported incidence of PJI after primary TKA ranges from 1.1% to 2.18%.16 17 18 Pulido et al16 reported the incidence of PJI after TKA to be 1.1%, of which 27% were diagnosed during the first 30 days after arthroplasty, and a majority of 65% were diagnosed in the first year after surgery. In our study, the average time to diagnosis was 431 days after the index surgery (range, 11-1699 days).

Rheumatoid arthritis was a significant risk factor for PJI after primary TKA. This finding is in keeping with the current literature.6 8 11 Although various authors have found male sex to be a risk factor for PJI,4 19 20 the association was not significant in this study. The OR of 1.9 may be of clinical importance but not significant as a result of the small number of PJIs and inadequate statistical power. The correlation between age and PJI has been a matter of controversy, with some reports mentioning young age as a risk factor for PJI4 21 and some otherwise.22 In our study, age was not associated with PJI occurrence. For one-stage bilateral TKA, age has been a controversial risk factor for PJI. Some studies16 23 have suggested that one-stage bilateral TKA is associated with an increased risk of superficial and deep infection. Hussain et al24 nonetheless reported a similar infection rate between one- and two-stage bilateral TKA. Our study did not find an association between one-stage bilateral TKA and PJI occurrence.

The local bacteriological pattern for PJI was comparable to that reported in the literature.4 16 In our study, skin flora and gram-positive bacteria were the most commonly isolated organisms, followed by gram-negative bacteria such as Escherichia coli and Salmonella. Coagulase-negative staphylococci were the most common causative organism in one study.4 In contrast, in our series, S aureus was the most common causative organism, particularly methicillin-sensitive strains. Methicillin-resistant strains were less common in our series, constituting 39% of all staphylococcal organisms.

Other authors have reported that male sex is a risk factor for PJI, which may be related to a sex difference in immune response to pathogenic bacteria. Studies6 have shown that males (compared with females) have a significantly higher likelihood of being a persistent S aureus carrier. However, our study did not support male sex as a risk factor for infection with skin flora. With regard to onset of infection, PJI caused by skin flora was positively associated with early-onset infection, although the association did not reach statistical significance (P=0.099). Direct inoculation and spread from contiguous foci of infection are more common in early-onset infection caused by wound complications and local soft-tissue conditions. In contrast, distant foci of infection, such as in bacteraemia, play a more important role in late-onset infection. Therefore, in early-onset periprosthetic joint infection with negative cultures, an empirical antibiotic regimen may provide adequate coverage against skin flora organisms.

Fan et al20 reported 479 TKAs and rates of 1.9% for superficial wound infection, 0.2% for early deep infection (n=1), and 0.6% for late deep infection (n=2). Methicillin-sensitive S aureus and coagulase-negative staphylococci were causative organisms. Lee et al25 reviewed 1133 primary TKAs and found a 0.71% incidence of PJI. The most common causative organisms in descending order were methicillin-sensitive S aureus, coagulase-negative staphylococci, methicillin-resistant S aureus, and Pseudomonas aeruginosa. This finding is in keeping with our data. Among risk factors identified by Lee et al25 were young age, diabetes, anaemia, thyroid disease, heart disease, lung disease, and long operating time. However, the researchers identified limitations of having only a small number of patients with infection (n=8) and insufficient power for analysis. In addition, multivariable analysis should have been performed to account for the effect of confounders among the multiple risk factors. They also reported the limitation that the mean follow-up duration was only 2 years. A short follow-up period may underestimate the occurrence of late-onset infection.

Our study has several limitations. The number of PJI-positive cases was small and thus subgroup analysis was limited. This study included subjects treated at a single centre in Hong Kong; multicentre studies may improve the representativeness of local data. In addition, perioperative management for elective TKA has evolved over the past 20 years, including the introduction of an MRSA-screening programme in 2011. In the screening programme, a nasal swab is taken from all elective joint-replacement patients. Patients with a positive result are prescribed 5 days of decolonisation therapy including a daily chlorhexidine bath. Furthermore, intravenous vancomycin is now administered for prophylaxis instead of cefazolin.26

There are many potential risk factors for PJI documented in the literature. Nonetheless, only a limited number were included in this study, most of which are not be modifiable. Thus, it may not provide the necessary guidance for preoperative optimisation. Furthermore, the exclusion of some potential risk factors may have led to inadequate control for potential confounding factors. Inclusion of more risk factors with better characterisation is needed to provide a more comprehensive understanding and to better account for the confounding effect of other variables.

The incidence of PJI after elective primary TKA in our institution over two decades from 1993 to 2013 was 1.34%. Rheumatoid arthritis was a significant risk factor for PJI in this series. In the early-onset infection group, PJI was caused by skin flora, but this was not statistically significant. It is hoped that this study has updated the local data for PJI after primary TKA and serves as a model for future related studies.

We thank colleagues from the Department of Orthopaedics and Traumatology and the Infection Control Team at the Queen Mary Hospital for their assistance in data collection, and those who advised on this project to make its publication possible.

The authors have no conflicts of interest to disclose.

1. Kurtz SM, Lau E, Watson H, Schmier JK, Parvizi J. Economic burden of periprosthetic joint infection in the United States. J Arthroplasty 2012;27(8 Suppl):61-5.e1. Crossref

2. Lamarsalle L, Hunt B, Schauf M, Szwarcensztein K, Valentine WJ. Evaluating the clinical and economic burden of healthcare-associated infections during hospitalization for surgery in France. Epidemiol Infect 2013;141:2473-82. Crossref

3. Nero DC, Lipp MJ, Callahan MA. The financial impact of hospital-acquired conditions. J Health Care Finance 2012;38:40-9.

4. Crowe B, Payne A, Evangelista PJ, et al. Risk factors for infection following total knee arthroplasty: a series of 3836 cases from one institution. J Arthroplasty 2015;30:2275-8. Crossref

5. Meller MM, Toossi N, Johanson NA, Gonzalez MH, Son MS, Lau EC. Risk and cost of 90-day complications in morbidly and superobese patients after total knee arthroplasty. J Arthroplasty 2016;31:2091-8. Crossref

6. Zmistowski B, Alijanipour P. Risk factors for periprosthetic joint infection. In: Springer BD, Parvizi J, editors. Periprosthetic Joint Infection of the Hip and Knee. New York: Springer; 2014: 15-40. Crossref

7. Jämsen E, Huhtala H, Puolakka T, Moilanen T. Risk factors for infection after knee arthroplasty. A register-based analysis of 43,149 cases. J Bone Joint Surg Am 2009;91:38-47. Crossref

8. Wilson MG, Kelley K, Thornhill TS. Infection as a complication of total knee-replacement arthroplasty. Risk factors and treatment in sixty-seven cases. J Bone Joint Surg Am 1990;72:878-83. Crossref

9. Namba RS, Inacio MC, Paxton EW. Risk factors associated with deep surgical site infections after primary total knee arthroplasty: an analysis of 56,216 knees. J Bone Joint Surg Am 2013;95:775-82. Crossref

10. Pruzansky JS, Bronson MJ, Grelsamer RP, Strauss E, Moucha CS. Prevalence of modifiable surgical site infection risk factors in hip and knee joint arthroplasty patients at an urban academic hospital. J Arthroplasty 2014;29:272-6. Crossref

11. Chesney D, Sales J, Elton R, Brenkel IJ. Infection after knee arthroplasty: a prospective study of 1509 cases. J Arthroplasty 2008;23:355-9. Crossref

12. Moucha CS, Clyburn T, Evans RP, Prokuski L. Modifiable risk factors for surgical site infection. J Bone Joint Surg Am 2011;93:398-404.

13. Peersman G, Laskin R, Davis J, Peterson M. Infection in total knee replacement: a retrospective review of 6489 total knee replacements. Clin Orthop Relat Res 2001;392:15-23. Crossref

14. Rasouli MR, Restrepo C, Maltenfort MG, Purtill JJ, Parvizi J. Risk factors for surgical site infection following total joint arthroplasty. J Bone Joint Surg Am 2014;96:e158. Crossref

15. Parvizi J, Gehrke T; International Consensus Group on Periprosthetic Joint Infection. Definition of periprosthetic joint infection. J Arthroplasty 2014;29:1331. Crossref

16. Pulido L, Ghanem E, Joshi A, Purtill JJ, Parvizi J. Periprosthetic joint infection: the incidence, timing, and predisposing factors. Clin Orthop Relat Res 2008;466:1710-5. Crossref

17. Tsaras G, Osmon DR, Mabry T, et al. Incidence, secular trends, and outcomes of prosthetic joint infection: a population based study, Olmsted County, Minnesota, 1969-2007. Infect Control Hosp Epidemiol 2012;33:1207-12. Crossref

18. Tande AJ, Patel R. Prosthetic joint infection. Clin Microbiol Rev 2014;27:302-45. Crossref

19. Herwaldt LA, Cullen JJ, French P, et al. Preoperative risk factors for nasal carriage of Staphylococcus aureus. Infect Control Hosp Epidemiol 2004;25:481-4. Crossref

20. Fan JC, Hung HH, Fung KY. Infection in primary total knee replacement. Hong Kong Med J 2008;14:40-5.

21. Meehan JP, Danielsen B, Kim SH, Jamali AA, White RH. Younger age is associated with a higher risk of early periprosthetic joint infection and aseptic mechanical failure after total knee arthroplasty. J Bone Joint Surg Am 2014;96:529-35. Crossref

22. Berbari EF, Osmon DR, Lahr B, et al. The Mayo prosthetic joint infection risk score: implication for surgical site infection reporting and risk stratification. Infect Control Hosp Epidemiol 2012;33:774-81. Crossref

23. Luscombe JC, Theivendran K, Abudu A, Carter SR. The relative safety of one-stage bilateral total knee arthroplasty. Int Orthop 2009;33:101-4. Crossref

24. Hussain N, Chien T, Hussain F, et al. Simultaneous versus staged bilateral total knee arthroplasty: a meta-analysis evaluating mortality, peri-operative complications and infection rates. HSS J 2013;9:50-9. Crossref

25. Lee QJ, Mak WP, Wong YC. Risk factors for periprosthetic joint infection in total knee arthroplasty. J Orthop Surg (Hong Kong) 2015;23:282-6. Crossref

26. Cheng VC, Tai JW, Wong ZS, et al. Transmission of methicillin-resistant Staphylococcus aureus in the long term care facilities in Hong Kong. BMC Infect Dis 2013;13:205. Crossref