Reference intervals (RIs) are an indispensable tool for clinical decision making in the interpretation of numerical pathology results. Simple yet elegant comparisons with reference subjects empower the interpreter with objective judgements and aid clinicians in diagnosis, treatment, monitoring, prognostication, and screening.1

Reference intervals are commonly defined as limiting values, usually upper and lower limits, between which a prespecified percentage (usually 95%) of results would fall.2 3 In daily practice, for most tests, there exists some degree of laboratory-specific bias related to differences in pre-analytical and analytical factors, such as the choices of analytical platform, methodology, and reagent. Therefore, it is desirable for laboratories to provide sets of laboratory-specific RIs following Clinical and Laboratory Standards Institute guideline C28-A3c. A laboratory may establish a new set of RIs by conducting an RI study with at least 120 reference individuals from each subgroup stratified by sex, age, and other parameters as appropriate.2 Conducting an RI study is challenging, as enormous efforts of human and financial resources are needed. As the list of analytes is long, it is almost impossible for every laboratory to repeat an RI study to accommodate future changes in methodology or analytical platforms.2 4 Alternatively, a laboratory may adopt the RIs established by other sources such as manufacturers or the literature and validate them with at least 20 reference individuals’ results. An additional option is for the laboratory to transfer previously established RIs according to mathematical formulas to account for differences in analytical factors.2 These methods ensure that each laboratory provides a set of clinically meaningful intervals to clinicians, aiding their management.

Therefore, for the same analyte, it is not uncommon to see different RIs across laboratories. This inter-laboratory coefficient of variation was reported by Ceriotti et al3 to be as high as 15% to 20% for the RIs of urea and creatinine. This could be reasonable for hormonal tests that are not optimally standardised, as demonstrated by the marked variations in RIs for thyroid hormones between four analytical platforms shown by a recent study in the UK.5 For analytes that are generally well standardised across platforms, such as plasma electrolytes, one would expect results generated by different laboratories to be comparable. Logically, with insignificant methodological bias, the RIs should be same for the specified homogenous population.

In 2007, the UK Pathology Harmony Group showed that laboratories were using different sets of Rls with no sound scientific basis despite using the same analytical platform and reagents.6 7 The same problem was later also revealed by a survey on RIs in Australasia.8 The differences in RIs were concluded to be unnecessary and would have created unneeded confusion during interpretation, which might lead to inappropriate investigations or treatments.9 10 Common RIs were offered as a solution to unite laboratory practices.4

In Hong Kong, we have observed a general trend of variation in RIs that resembles those in the UK and Australasia, with various RIs adopted for most tests, including general chemistry laboratory tests. Hence, we conducted the first local study to explore the situation with a territory-wide survey on RIs. The aim was to scientifically review the analytical variation of general chemistry laboratory tests between local laboratories and to examine the evidence for such variations.

Fourteen blood general chemistry analytes were included in this study, namely albumin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total bilirubin, calcium, chloride, creatinine, gamma-glutamyl transferase (GGT), phosphate, potassium, sodium, total protein, and urea. We conducted a territory-wide survey involving 10 chemical pathology laboratories. All laboratories provided routine services to assess the 14 analytes, except for AST, chloride, and GGT, which were not evaluated in three laboratories. The instruments were Abbott Architect (labs 1-3), Beckman Coulter AU (labs 4-5), Roche Cobas (labs 6-9), and Siemens Dimension EXL (lab 10). Table 1 summarises the analytical platforms and methodologies.

The laboratories participated in the Condensed General Chemistry Programme provided by the Royal College of Pathologists of Australasia Quality Assurance Programs. In each cycle of the external quality assurance programme (EQAP), identical sets of QAP materials were analysed by each individual laboratory for the aforementioned blood general chemistry analytes using their own analytical platform. The use of QAP materials, which were commutable samples with the same properties as routinely analysed clinical samples, minimises the matrix effect. In routine clinical practice, EQAP safeguards laboratory performance by comparison with peers and reference methods. In the present study, we retrospectively review these readily available EQAP data from local laboratories for bias assessments. The participants provided their responses by email to the following items: (1) historical EQAP results of six cycles (105-11, 105-12, 105-15, 105-16, 106-03, and 106-04); (2) adult RIs in use for clinical service, and (3) analytical specification of assays.

Laboratory performance bias was assessed by percentage differences of EQAP results. Percentage difference was defined as the laboratory result minus the target value divided by the target value. The feasibility of applying common RIs among the laboratories was determined by the degree of agreement between the percentage differences and the corresponding allowable limits of performance.11 Data analyses were performed using Microsoft Excel 2016.

Figure 1 shows that half of the 14 analytes showed agreement across all laboratories. The inter-laboratory differences are within the corresponding target allowable limit of error (ALE) for AST (-3% to +9%; target ALE ±12%), chloride (-1% to +2%; ±3%), phosphate (-1% to 4%; ±8%), potassium (-2% to 3%; ±5%), sodium (-1% to 2%; ±2%). Three other analytes (albumin, ALT, and GGT) also showed agreement across nine laboratories with the Abbott, Beckman, and Roche platforms, except Siemens which was only used by one laboratory.

Figure 2 shows the inter-laboratory comparison of RIs for the 14 general chemistry analytes. Laboratories using the same platform generally adopted the same RIs, except for one laboratory using the Roche platform.

Notably, for the seven analytes mentioned above that showed agreement within the target ALE, the RIs differed substantially across the 10 laboratories. Particularly, the upper RI limit ranged from 34 to 50 U/L (coefficient of variation [CV]: 11%): in male samples and 30 to 40 U/L (9%) in female samples in AST; 107 to 109 mmol/L (0.9%) in chloride; 1.39 to 1.52 mmol/L (2.7%) in phosphate; 4.4 to 5.2 mmol/L (6.7%) in potassium; 144 to 148 mmol/L (0.7%) in sodium; 79 to 87 g/L (2.2%) in total protein; and 6.3 to 8.1 mmol/L (8.1%) in urea. The lower RIs ranged from 98 to 102 mmol/L (1.7%) in chloride; 0.72 to 0.88 mmol/L (6.2%) in phosphate; 3.4 to 3.6 mmol/L (2.6%) in potassium; 136 to 137 mmol/L (0.2%) in sodium; 63 to 68 g/L (2.2%) in total protein; and 2.5 to 3.1 mmol/L (7.4%) in urea.

The remaining analytes (albumin, ALT, ALP, calcium, creatinine, GGT, and total bilirubin) demonstrated substantial platform-specific bias exceeding the target ALE. High bias exceeding the ALE was observed for ALT (+12% to +20%; target ALE ±12%) and GGT (+11% to +14%; ±12%), with negative bias exceeding the ALE in albumin (-5.3% to -7.1%; ±6%), ALP (-11.4% to -15.3%; ±12%), and calcium (-5.6% to -7.1%; ±4%) present on the Siemens platform. Negative bias exceeding the ALE in ALP (-12.2% to -14.8%; ±12%) was also detected on the Roche platform. For calcium, negative bias exceeding the ALE (-4% to -6%; ±4%) was also detected at one laboratory using the Beckman platform. For creatinine, all laboratories were in agreement about concentrations ranging from 152 to 349 μmol/L. However, significant negative bias (-13% to -22%; ±12%) was observed for creatinine levels at the target value of 67 μmol/L on the Abbott, Siemens, and Roche instruments. For total bilirubin, half of the laboratories showed agreement within the ALE, while the remaining laboratories had significant negative bias (-14% to 17%; ±12%).

The investigated laboratories used different RIs despite employing the same analytical platforms, methods and reagents, for 11 out of the 14 analytes among those using the Abbott platform (labs 1-3), 11 out of 14 of analytes among those using Roche platforms (labs 6-9), and three out of the 14 of analytes among those using the Beckman platforms (labs 4-5).

Sex-specific RIs were not consistently provided for eight analytes (ALP, ALT, AST, phosphate, potassium, total bilirubin, total protein, and urea). For instance, sex-specific RIs were not provided by two laboratories for ALP, two for ALT, two for AST, five for potassium, five for urea, seven for total protein, eight for phosphate, and nine for total bilirubin.

Reference intervals are provided by laboratories as interpretative tools to aid clinical decision making. Theoretically, RIs could be affected by patient factors (eg, sex, age, ethnicity, biological variability), pre-analytical and analytical factors (eg, choice of method, reagents, platform, calibration), and statistical methodology.12 Therefore, for the same population, the RIs used for a test are inevitably influenced by the bias of the laboratory assays. In other words, RIs should theoretically be the same if the above-listed factors do not introduce significant bias.

In local practice, 10 analytes surveyed demonstrated sufficient agreement within the ALE between different analytical platforms across laboratories (Fig 1: AST, chloride, phosphate, potassium, sodium, total protein, and urea for all four platforms; albumin, ALT and GGT for Abbott, Beckman, and Roche platforms) [Fig 1]. These results confirmed the previous findings of bias assessment by the Australasian Association of Clinical Biochemists, which concluded that chloride, phosphate, potassium, sodium, total protein, and urea measurements showed sufficient similarity across analytical platforms and laboratories and that common RIs could be adopted.10 The same study found method-specific bias in AST levels averaging +22% for assays using pyridoxal-5-phosphate as an activator compared with those not using pyridoxal-5-phosphate.10 Our results showed a lesser degree of pyridoxal-5-phosphate–related bias (+7%), so this issue would not prevent the use of common RIs in the local scenario.

For analytes with demonstrated agreement across platforms and laboratories, the RIs are theoretically expected to be the same if obtained from the same group of reference (ie, ‘healthy’) individuals. In actual practice, for the seven analytes mentioned above, all of the adult RIs varied across laboratories, with the CV of the upper and lower limits of the RIs up to 11% and 7.4%, respectively. The inter-laboratory percentage differences of upper RI limits were up to 47% for AST (absolute difference: 16 U/L), 29% for urea (1.8 mmol/L), and 18% for potassium (0.8 mmol/L), and those of the lower RI limits were up to 24% for urea (0.6 mmol/L), 22% for phosphate (0.16 mmol/L), and 8% for total protein (5 g/L). We can compare the CVs of these analytes’ RIs against the corresponding between-subject biological variation (CV-G) values published by Ricos et al.13 The CV of the upper RI limits of potassium and sodium were 1.2 and 1.0 times those of CV-G, respectively while that of the lower RI limits of sodium and phosphate were 1.1 and 0.6 times those of CV-G, respectively. These RI variations generate significant additional bias during interpretation, which is often overlooked and unchecked. Furthermore, laboratories were using different RIs despite using the same analytical platforms and methodologies for these analytes. For example, among users of the Abbott platform, the potassium RIs of labs 1 and 2 were 3.6 to 5.2 mmol/L for samples of both sexes, while that of lab 3 was 3.5 to 4.5 mmol/L for male and 3.4 to 4.4 mmol/L for female samples. These variations were unnecessary, as supported by the sufficient agreement across analytical platforms and laboratories. Application of different RIs in various circumstances could lead to confusion among interpreters and hinder data management in the era of electronic health records.4 14 Similar trends of unexplained RI variations were previously observed in the UK for sodium, potassium, and other analytes, and this eventually lead to the Pathology Harmony group’s recommendation of harmonised RIs.7 At present, local laboratories often spend substantial human resources on decisions and maintenance regarding the appropriate RIs for large numbers of analytes. The use of common RIs for these seven analytes would unite local laboratory practices, facilitate electronic communications between laboratory information and electronic patient record systems, and streamline the maintenance of RIs.

For creatinine, low bias was noted for seven laboratories using the Jaffe methods, but this tendency spared the laboratories that used the enzymatic method on the Beckman platform. This bias was likely related to the high variability of the Jaffe method at low creatinine concentrations, which has been reported to be up to 30% on some platforms.15 While the remarkably good analytical agreement shown for the remaining five higher concentrations of creatinine support the use of common RIs, this should be cautiously reviewed, as the lowest concentration of creatinine (67 μmol/L) is very close to the lower RI limit. Further study of bias may be warranted for creatinine.

Substantial bias exceeding the ALE was demonstrated for the remaining six analytes, with high bias for ALT and GGT and low bias for albumin, ALP, and calcium on the Siemens platform; low bias for ALP on the Roche platform; low bias for calcium at one laboratory using the Beckman platform; and low bias for total bilirubin in labs 1 to 3, 7, and 8. Positive bias averaging 8% for albumin was observed for laboratories using the bromocresol green method compared with the bromocresol purple method, a pattern similar to the findings of Koerbin et al.10 16 Bias in ALT measurement could be attributed to differences in assay design,10 with an average of +7% bias shown for the assay using pyridoxal-5-phosphate over the assay that does not use it. Bias for calcium and total bilirubin could be related to methodological differences between platforms, while bias for ALP and GGT were likely to be specific to the analytical platform. While the feasibility of local common RIs for these six analytes was not confirmed by this study, our findings indicate that common RIs could still be considered for albumin, ALT, and GGT in laboratories using the Abbott, Beckman, and Roche platforms, which all laboratories except one use.

Variable adoptions of sex-specific RIs were another key finding of the survey. Heterogeneous and inconsistent practices of sex partitioning for RIs were noted in eight analytes (ALP, ALT, AST, phosphate, potassium, total bilirubin, total protein, and urea). Moreover, sex-specific RIs were sometimes different even within the same platform. For example, the upper RI limit of GGT in male samples differed by 35 U/L among users of the Roche platform, and the upper RI limit of ALP differed by 40 U/L and 7 U/L in male and female samples, respectively, among users of the Abbott platform. Common RIs with united practice of sex partitioning could be the solution to converge these practices.

Historically, heterogeneous and sometimes incomparable results of the same measurands could be obtained with different assays because of suboptimal standardisations in pre-analytical and analytical factors. Laboratory-specific RIs were advocated to compensate and allow for sound interpretations of laboratory results in clinical settings.17 Realising the need for assay standardisation, an enormous global effort has been taken in the past 60 years to study biological variability, standardise pre-analytical conditions and analytical methods, improve quality control, establish traceability of reference materials and methods, and implement EQAPs for various kinds of assays, led by the International Federation of Clinical Chemistry (IFCC) and other international/national organisations.18 Major successes have been realised for a large number of measurands, as listed on the website of the International Consortium for Harmonization of Clinical Laboratory Results.19

The concept of common RIs emerged in the early 2000s and has gained huge popularity over the past decade.4 The theory is simple: if the measured results of different assays are comparable, ie with adequate assay standardisation, the same RIs should be adopted given that the tests are performed on the same reference population.17 Redundant variations of RIs merely impair interpretation.

Presently, there are two types of common RIs: ‘objective’ and ‘subjective’ ones.20 Subjective common RIs were generally defined by scientific surveys and expert guidance with the harmonisation approach. Examples include the “agreed Pathology Harmony clinical biochemistry reference intervals for adults” for 15 general chemistry analytes recommended by the UK Pathology Harmony Group in 201121 and the “adult harmonised reference intervals” for 18 general chemistry analytes recommended by the Australasian Association of Clinical Biochemists and endorsed by the Royal College of Pathologists of Australasia in 2016.16 22 23 The two groups have since continued their work on harmonisation of various aspects of pathology in the past decade, with the UK Pathology Harmony Group working on the Pathology Harmony bookmark for tumour markers and requesting guidance for non-specialists, and the Australasian Association of Clinical Biochemists working on harmonisation of paediatric common RIs, serum protein electrophoresis reporting, lipid reporting, management and communication of high-risk lab results, arterial and venous blood gas RIs, and reporting of dynamic endocrine testing for adults and paediatric patients.6 7 8 16 24 25 26

Objective common RIs refer to those defined by well-conducted, multicentre reference studies, such as the Nordic Trueness Project, which was conducted with well-standardised pre-analytical and analytical handlings and the use of five control materials. The project involved 102 Nordic routine clinical biochemistry laboratories and more than 2500 carefully selected healthy reference individuals.27 The Nordic Reference Interval Project RIs for 25 general chemistry analytes were established and published in 2002 and implemented throughout Nordic countries in 2004 with the help of the Scandinavian Society of Clinical Chemistry.27 28 29 30 Among Asian countries, the Japan Society of Clinical Chemistry has recently published their nationwide common RIs for 40 laboratory tests determined by three multicentre RI studies.31 Table 2 8 21 23 28 31 summarises the common RIs published in different parts of world for the general chemistry analytes surveyed and the common RIs proposed by our study.

In 2017, the IFCC Committee on Reference Intervals and Decision Limits (C-RIDL) published two landmark papers on the results of their global multicentre study on reference values of 25 chemistry analytes in 13 386 healthy adults recruited from 12 countries, including China,32 with the use of a specially designed serum panel.33 34 The study explored the regionality and ethnicity of these reference values globally and provided invaluable information for the possibility of future derivation and transference of the established RIs through use of the C-RIDL serum panel.34

The relatively small number and choice of QAP specimens for retrospective methodological comparisons represent a major limitation of our survey. Artificial materials used in QAP specimens generally gave rise to more variable and method-dependent results due to matrix effects.9 Despite this, our survey demonstrated that methodological bias would not prevent the use of common RIs for seven general chemistry analytes. For the remaining analytes, we speculate that the degree of methodological bias may be exaggerated by the matrix effect of the QAP, ie, the actual analytical difference is likely to be smaller when tested with a patient sample. Our findings should be verified with a formal prospective bias study with a standardised protocol and the use of another set of blood specimens, preferably unadulterated human samples, with pre-assigned reference values to ensure commutability.

This survey compared the adult RIs of 14 general chemistry analytes among 10 chemical pathology laboratories using four different analytical platforms. Bias assessments and comparisons of RIs revealed that different and variable RIs were provided by the laboratories despite sufficient inter-laboratory and inter-platform agreement regarding the RIs of 10 general chemistry analytes. The use of common RIs was found to be feasible and is recommended for these 10 analytes. Such use would unify and improve local standards of clinical laboratory practice. A well-designed implementation plan for common RIs with support from stakeholders including clinicians, pathologists, and scientists would be vital for the success of such a substantial project. Figure 3 shows our proposed implementation plan for the introduction of common RIs in Hong Kong, modified from the plan suggested by Tate et al8 for the harmonisation of adult and paediatric RIs in Australasia. Furthermore, the concept of common RIs could be expanded to cover more general chemistry analytes, eg, creatine kinase and magnesium; special chemical tests, eg, therapeutic drug monitoring and hormones; other clinical laboratory specialties, such as haematology and immunology; and paediatric RIs.6 7 8

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

All authors have disclosed no conflicts of interest.

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

The present survey is a retrospective observational review of local laboratory practice and external quality assurance program data with no patient participation, no access to private or sensitive patient data, no collection or analysis of human body fluid or tissue. The quality assurance program materials used for the data collection in the survey are processed samples designed by the external quality assurance program organiser to mimic the properties of clinical sample. Therefore, ethics approval was not applicable for this study.

1. Plebani M, Lippi G. Reference values and the journal: why the past is now present. Clin Chem Lab Med 2012;50:761-3. Crossref

2. Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline—Third Edition. Clinical and Laboratory Standards Institute; 2010.

3. Ceriotti F, Hinzmann R, Panteghini M. Reference intervals: the way forward. Ann Clin Biochem 2009;46:8-17. Crossref

4. Jones GR, Barker A, Tate J, Lim CF, Robertson K. The case for common reference intervals. Clin Biochem Rev 2004;25:99-104.

5. Barth JH, Luvai A, Jassam N, et al. Comparison of method-related reference intervals for thyroid hormones: studies from a prospective reference population and a literature review. Ann Clin Biochem 2018;55:107-12. Crossref

6. Berg J. The approach to pathology harmony in the UK. Clin Biochem Rev 2012;33:89-93.

7. Berg J. The UK Pathology Harmony initiative; The foundation of a global model. Clin Chim Acta 2014;432:22-6. Crossref

8. Tate JR, Sikaris KA, Jones GR, et al. Harmonising adult and paediatric reference intervals in Australia and New Zealand: an evidence-based approach for establishing a first panel of chemistry analytes. Clin Biochem Rev 2014;35:213-35.

9. Jones G, Barker A. Standardisation of reference intervals: an Australasian view. Clin Biochem Rev 2007;28:169-73.

10. Koerbin G, Tate JR, Ryan J, et al. Bias assessment of general chemistry analytes using commutable samples. Clin Biochem Rev 2014;35:203-11.

11. Jones GR, Sikaris K, Gill J. ‘Allowable limits of performance’ for external quality assurance programs—an approach to application of the Stockholm Criteria by the RCPA Quality Assurance Programs. Clin Biochem Rev 2012;33:133-9.

12. Ozarda Y. Reference intervals: current status, recent developments and future considerations. Biochem Med (Zagreb) 2016;26:5-16. Crossref

13. Ricós C, Alvarez V, Cava F, et al. Current databases on biological variation: pros, cons and progress. Scand J Clin Lab Invest 1999;59:491-500. Crossref

14. Koerbin G, Sikaris KA, Jones GR, et al. Evidence-based approach to harmonised reference intervals. Clin Chim Acta 2014;432:99-107. Crossref

15. Hoste L, Deiteren K, Pottel H, Callewaert N, Martens F. Routine serum creatinine measurements: how well do we perform? BMC Nephrol 2015;16:21. Crossref

16. Koerbin G, Sikaris K, Jones GR, et al. An update report on the harmonization of adult reference intervals in Australasia. Clin Chem Lab Med 2018;57:38-41. Crossref

17. Ceriotti F. Prerequisites for use of common reference intervals. Clin Biochem Rev 2007;28:115-21.

18. Siest G, Henny J, Gräsbeck R, et al. The theory of reference values: an unfinished symphony. Clin Chem Lab Med 2013;51:47-64. Crossref

19. International Consortium for Harmonization of Clinical Laboratory Results. Summary of measurand harmonization activities. Available from: www.harmonization.net/measurands/. Accessed 27 Mar 2019.

20. Ozarda Y, Sikaris K, Streichert T, Macri J; IFCC Committee on Reference Intervals and Decision Limits (C-RIDL). Distinguishing reference intervals and clinical decision limits—a review by the IFCC Committee on Reference Intervals and Decision Limits. Crit Rev Clin Lab Sci 2018;55:420-31. Crossref

21. Berg J, Lane V. Pathology Harmony; a pragmatic and scientific approach to unfounded variation in the clinical laboratory. Ann Clin Biochem 2011;48(Pt 3):195-7. Crossref

22. Jones GR, Koetsier S. Uptake of recommended common reference intervals for chemical pathology in Australia. Ann Clin Biochem 2017;54:395-7. Crossref

23. Koerbin G, Tate JR. Harmonising adult reference intervals in Australia and New Zealand—the continuing story. Clin Biochem Rev 2016;37:121-9.

24. Campbell C, Caldwell G, Coates P, et al. Consensus statement for the management and communication of high risk laboratory results. Clin Biochem Rev 2015;36:97-105.

25. Tate J, Caldwell G, Daly J, et al. Recommendations for standardized reporting of protein electrophoresis in Australia and New Zealand. Ann Clin Biochem 2012;49:242-56. Crossref

26. Australasian Association of Clinical Biochemists. Harmonisation: guideline, publications, history, committee. Available from: www.aacb.asn.au/aboutus/harmonisation. Accessed 27 Mar 2019.

27. Pedersen MM, Ornemark U, Rustad P, et al. The Nordic Trueness Project 2002: use of reference measurement procedure values in a general clinical chemistry survey. Scand J Clin Lab Invest 2004;64:309-20. Crossref

28. Rustad P, Felding P, Franzson L, et al. The Nordic Reference Interval Project 2000: recommended reference intervals for 25 common biochemical properties. Scand J Clin Lab Invest 2004;64:271-84. Crossref

29. Strømme JH, Rustad P, Steensland H, Theodorsen L, Urdal P. Reference intervals for eight enzymes in blood of adult females and males measured in accordance with the International Federation of Clinical Chemistry reference system at 37 degrees C: part of the Nordic Reference Interval Project. Scand J Clin Lab Invest 2004;64:371-84. Crossref

30. Rustad P, Felding P, Lahti A, Hyltoft Petersen P. Descriptive analytical data and consequences for calculation of common reference intervals in the Nordic Reference Interval Project 2000. Scand J Clin Lab Invest 2004;64:343-70. Crossref

31. Ichihara K, Yomamoto Y, Hotta T, et al. Collaborative derivation of reference intervals for major clinical laboratory tests in Japan. Ann Clin Biochem 2016;53:347-56. Crossref

32. Xia L, Chen M, Liu M, et al. Nationwide multicenter reference interval study for 28 common biochemical analytes in China. Medicine (Baltimore) 2016;95:e2915. Crossref

33. Ichihara K, Ozarda Y, Barth JH, et al. A global multicenter study on reference values: 1. Assessment of methods for derivation and comparison of reference intervals. Clin Chim Acta 2017;467:70-82. Crossref

34. Ichihara K, Ozarda Y, Barth JH, et al. A global multicenter study on reference values: 2. Exploration of sources of variation across the countries. Clin Chim Acta 2017;467:83-97. Crossref

Obstetric anal sphincter injuries (OASIS) is a serious complication of vaginal delivery that is associated with an increased risk of anal incontinence (complaint of involuntary loss of faeces or flatus).1 The incidence of OASIS is reportedly much lower in Hong Kong (0.32%) than in other countries, such as the United Kingdom, Norway, and Sweden (2.9%-4.2%).2 3 4 5 This could be affected by a number of factors. First, delivery practices in Hong Kong are quite different from elsewhere in the world, such that they include the use of a hands-on approach to protect the perineum and liberal use of episiotomy.6 The episiotomy rates are reportedly high in Hong Kong: 83.7% for primiparous women and 54.8% for multiparous women.5 Moreover, in Hong Kong, a left mediolateral episiotomy is used, whereas midline episiotomy or right mediolateral episiotomy are used in many other parts of the world.7 Second, there may be ethnic differences in pelvic floor biometry. In particular, Chinese women have a smaller hiatal dimension and reduced pelvic organ mobility.8 It is unclear how these differences in practice and pelvic floor biometry influence the incidence of OASIS.

Importantly, it is also possible that the reduced incidence of OASIS in Hong Kong is a result of underdetection. In a recent local prospective observational study, women were assessed by a single experienced clinician via rectal examination after either normal or instrumental vaginal delivery; the results of that study showed that the incidence of OASIS in primiparous Asian women in Hong Kong was 10%,6 which suggests that the OASIS rate might be higher than previously published. Obstetric anal sphincter injuries that are identified after an extended interval (such as during postnatal follow-up) is regarded as occult OASIS. There is limited information in the literature regarding occult OASIS; thus far, studies have been conducted in the United Kingdom and Australia.9 10

The use of endoanal ultrasound (US) may facilitate identification of OASIS.11 Endoanal US comprises a non-invasive assessment modality and is regarded as the gold standard in studies of anal sphincter injury.9 11 Moreover, all cases of clinically identified OASIS can also be identified on endoanal US.9 The aim of this study was to determine the prevalence of OASIS in primiparous women after normal vaginal delivery or instrumental delivery using endoanal US during postnatal follow-up. Understanding the prevalence and detection rates of OASIS can help inform training policies for midwives and doctors on the awareness and detection of OASIS.

This was a retrospective analysis of archived US volumes from two previously published studies that were performed at a tertiary university hospital in Hong Kong. The initial study recruited 442 nulliparous women in the first trimester, during the period from August 2009 to September 2010.12 13 The second study recruited 292 primiparous women at 1 to 3 days after instrumental delivery, during the period from September 2011 to May 2012. None of the women in either study reported symptoms of pelvic floor disorders, including faecal incontinence to solid or loose stool, before pregnancy.14 Details of deliveries, including any occurrence of perineal tearing, were recorded after each delivery. Ethics approval was obtained from The Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (Ref CRE-2013.332). The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines were followed in the preparation of this report.15

Generally, each woman underwent perineal examination by the attending midwife or doctor who conducted the delivery, immediately after vaginal delivery. This information was immediately recorded in the medical record. Third- or fourth-degree tears were assessed and repaired by a trained obstetrician. The anorectal mucosa was repaired by continuous or interrupted sutures with 3-O Vicryl. Internal anal sphincter tears were repaired separately by interrupted end-to-end sutures with 2-O Vicryl. External anal sphincter (EAS) tears were repaired by overlapping or end-to-end sutures with 2-O Vicryl. Perineal muscles and the vagina were repaired with 2-O Vicryl. The diagnosis and operative record of each woman were immediately entered into the electronic medical record. The degree of perineal tear was defined using Sultan’s classification of perineal trauma.16

During postnatal follow-up (6-12 months after delivery), the urinary, bowel, and prolapse symptoms of each woman, as well as their quality of life, were assessed using the Chinese Pelvic Floor Distress Inventory (PFDI) and Pelvic Floor Impact Questionnaire (PFIQ).17 Assessment of the anal sphincter was performed with endoanal US using a 10-MHz 360-degree rotating probe (Focus 400, BK Medical; Gentofte, Denmark) with the woman in the lithotomy position. Automatic image acquisition was performed with two volumes stored for each woman.

Offline analysis of the endoanal US volumes was performed in 2018 by two experienced obstetricians (OYKW, SSCC) who were blinded to the clinical diagnosis and questionnaire information. An anal sphincter defect was defined as a discontinuity of >30 degrees in endosonographic images of the internal (hypoechoic ring) and/or external (mixed echogenic ring) sphincters.18 A partial-thickness EAS injury was defined as a defect of <50% thickness of the EAS, whereas a defect of >50% of the EAS was regarded as a full-thickness injury. We considered any EAS and/or internal anal sphincter injury to be OASIS. This follows the clinical classification of OASIS by Sultan.16 Each researcher reviewed all endoanal US volumes independently. Any discrepancies were resolved by consensus review of the relevant US volumes.

The PFDI and PFIQ are comprehensive validated instruments which assess the symptoms and impact of pelvic floor disorders.17 In this study, faecal incontinence was defined as an affirmative response to either item 38 (“Do you usually lose stool beyond your control if your stool is well formed?”) or item 39 (“Do you lose stool beyond your control if your stool is loose or liquid?”) of the PFDI. Flatal incontinence was defined as an affirmative response to item 40 (“Do you usually lose gas from the rectum beyond your control?”) of the PFDI.

Data were analysed by SPSS (Window version 22.0; IBM Corp, Armonk [NY], United States). Descriptive analyses were used to study the prevalence of OASIS on endoanal US. Means were compared between groups using the independent-samples t test. Comparisons of frequencies were made using the Chi squared test or Fisher’s exact test, where appropriate. Univariate analysis was performed to evaluate the influence of potential risk factors on OASIS. Differences with P<0.05 were considered to be statistically significant. Power calculations were not performed with regard to this specific research question, as this study comprised a subanalysis of two prior projects, as described earlier in this paper.

A total of 544 women who had vaginal delivery were enrolled in this study; 207 had normal vaginal delivery and 337 had instrumental delivery (285 vacuum extraction, 52 forceps). Ultrasound images were suboptimal for two women who had normal vaginal delivery; these women were excluded from the analysis.

The demographic data and delivery information are shown in Table 1. Left mediolateral episiotomy was performed in 187 (91.2%) women in the normal vaginal delivery and 336 (99.7%) women in the instrumental delivery group. The duration of active second stage was longer in the instrumental delivery group than in the normal vaginal delivery group (62.7 ± 40.9 min vs 27.9 ± 22.4 min, P<0.005), as a prolonged second stage was the most common indication for instrumental delivery in this cohort (48.4%). More women had epidural analgesia in the instrumental delivery group than in the normal vaginal delivery group (15.7% vs 8.8%, P=0.028). There was no significant difference between the normal vaginal delivery and instrumental delivery groups regarding the timing of endoanal US assessment (P=0.22).

The Figure shows endoanal US images of intact anal sphincters, as well as sphincters with different degrees of OASIS. There were discrepancies or uncertainties in the endoanal US analysis of 16 women with respect to the diagnosis of OASIS. The two researchers determined the diagnoses of these women by consensus review; six were diagnosed with OASIS and 10 were regarded as normal.

The prevalence of clinically detected OASIS was 0% in the normal vaginal delivery group and 1.8% (n=6) in the instrumental delivery group. Table 2 shows that the prevalence of OASIS detected by endoanal US was 7.8% (n=16; 95% confidence interval [CI]=4.1%-11.5%) in the normal vaginal delivery group and 5.6% (n=19; 95% CI=3.1%-8.1%) in the instrumental delivery group (P=0.415). Twenty-nine (82.9%) women had OASIS, as detected by endoanal US, that was not diagnosed during clinical assessment immediately after delivery. Therefore, the occult OASIS rate was 7.8% (95% CI=4.1%-11.5%) in the normal vaginal delivery group and 3.8% (95% CI=1.8%-5.8%) in the instrumental delivery group. In addition, 63.6% (n=21) of occult EAS injuries comprised partial-thickness EAS injuries, whereas 36.4% (n=12) comprised full-thickness EAS injuries. When women with OASIS were compared to those without OASIS, increased birth weight was the only delivery factor associated with an increased risk of OASIS (odds ratio [OR]=3.1, 95% CI=1.3%-7.6%, P=0.012) [Table 3].

Overall, nine (1.7%) and 29 (5.4%) women reported faecal incontinence to solid and loose stool, whereas 97 (17.9%) women reported flatal incontinence (Table 4). All affected women reported mild symptoms. Among the women with OASIS, only one (2.9%) with a repaired third degree (3a) tear reported symptoms of both (faecal incontinence to loose stool and flatal incontinence). Three women (10.3%) who had occult injury reported flatal incontinence. There were no associations between the presence of OASIS and faecal incontinence (P=0.71) or between the presence of OASIS and flatal incontinence (P=0.37).

Primiparity has been associated with increased risks of OASIS (ORs of 2.39 and 8.34) in large retrospective studies.19 20 In the present study, which included large number of primiparous women, the findings on endoanal US were compared with women’s reported symptoms of faecal and flatal incontinence. Importantly, there were no associations between faecal or flatal incontinence and the presence of OASIS.

After assessment by endoanal US, the prevalence of OASIS in the normal vaginal delivery group increased from 0% to 7.8% and that in the instrumental delivery group increased from 1.8% to 5.6%. Overall, 82.9% of women with OASIS detected by endoanal US had not been diagnosed with OASIS during clinical assessment immediately after delivery. This finding is consistent with the results of the study by Andrews et al.9 In that study, the prevalence of OASIS markedly increased from 11% to 24.5% when women were re-examined by an experienced research fellow; 87% of OASIS diagnoses were missed by midwives and 28% were missed by junior doctors.9 In our study, normal vaginal deliveries were primarily attended by midwives, whereas instrumental deliveries were performed by residents. The higher rate of occult OASIS in the normal vaginal delivery group suggests that midwives currently receive inadequate training for clinical identification of OASIS. Thus, to improve the detection of OASIS, midwives and doctors should be trained to recognise OASIS by performing a standardised vaginal and rectal examination after delivery.

Compared with previous studies, the rate of OASIS determined by endoanal US in our study (6.5%) was lower than the rate of 10% determined by a single examiner in a prospective observational study conducted in the same unit.6 This could be a result of the small sample size (70 subjects) in the prior study. Furthermore, most patients with OASIS (5/7) in that study were reported to have small 3a tears. There were no 3c or fourth-degree tears in that study. Following the same delivery practices, clinically detected small 3a tears may therefore appear normal in endoanal US. Furthermore, these tears might not result in long-term consequences.6 21

The finding of an overall lower OASIS rate in Hong Kong, compared with that in Asian women who deliver in Caucasian countries, is not new.6 Asian women who deliver in locations with more restrictive policies regarding episiotomy have shown higher rates of OASIS.22 23 24 In a study conducted in the United States, OASIS was found significantly more frequently in Asian women than in women of other ethnicities.23 In Australia, nulliparous women born in South Asia and South-East Asia were 2.6-fold and 2.1-fold more likely to exhibit OASIS than women born in Australia or New Zealand women.24 It is uncertain whether the increased rate of episiotomy might protect against OASIS in Asian women and contribute to the relative reduction in the rate of OASIS in Hong Kong. Thus, our unit is currently conducting a randomised controlled trial to compare restrictive and routine episiotomy. In addition to episiotomy, the delivery technique and hands-on approach might contribute to the relative reduction in the rate of OASIS. All deliveries in our study were conducted with women in a lithotomy position, with their feet on footplates or in stirrups. All midwives and doctors conducting the deliveries used hands-on techniques to protect the perineum in each woman. Either firm pressure or pressure with squeezing of the perineum, also known as the modified Ritgen manoeuvre, was used.6 Warm compresses were not commonly used by midwives and doctors in our study.

The OASIS rate in the normal vaginal delivery group was higher than that in the the instrumental delivery group, but this difference was not statistically significant. The majority of deliveries by women in the instrumental delivery group were performed using vacuum extraction. The rate of OASIS in these women could be similar to that of women in the normal vaginal delivery group. The OASIS rates were similar in women who delivered with the aid of vacuum extraction or with forceps, whereas previous studies showed that forceps delivery was associated with an increased risk of OASIS.19 20 25 The small number of forceps deliveries in this study might have led to insufficient statistical power to detect a difference between the two types of instrumental deliveries. Furthermore, the use of forceps was primarily restricted to patients who were low risk, and mostly comprised outlet/low-cavity forceps deliveries. Previous studies reported that macrosomia, higher birth weight (OR=1.14, 95% CI=1.0-1.3, P=0.039), and shorter perineal length were risk factors for OASIS.6 19 20 The present study had similar findings, in that higher birth weight was a risk factor for OASIS (OR=3.1, 95% CI=1.3-7.6, P=0.012). However, perineal length was not assessed, which is an important limitation of this study.

Flatal incontinence was present in 17.9% of women after delivery, which is comparable to the rate reported in previous studies.26 27 In addition to OASIS, irritable bowel syndrome, high body mass index, and mode of delivery constitute factors associated with flatal incontinence.20 21 Overall, 5.5% of women reported faecal incontinence; most of these women reported faecal incontinence to loose stool and mild symptoms only. Most obstetric anal sphincter injuries were not detected during clinical examination. Shortly after delivery, the presence of OASIS was not associated with symptoms of faecal or flatal incontinence, but a longer-term study is needed to confirm these findings. However, we previously found that only antenatal faecal incontinence symptoms increased the likelihood of faecal incontinence at 12 months after delivery (OR=6.1, 95% CI=1.8-21.5, P=0.005), whereas maternal characteristics, mode of delivery, and the presence of OASIS did not.28 In longer-term follow-up (3-5 years after delivery), 2.1% and 5.9% of women who had one vaginal delivery reported faecal incontinence to solid and loose stool, respectively.29

To the best of our knowledge, there have been no randomised controlled trials regarding the optimal timing for the use of endoanal US to assess OASIS after vaginal delivery. One randomised controlled trial has been conducted to compare clinical examination alone (control group) and clinical examination with additional endoanal US immediately after delivery (intervention group).30 31 The results of that study showed that US performed immediately after delivery—before repair—might detect more cases of OASIS: 5.6% of women were found to have full-thickness OASIS that was not recognised during clinical examination alone.31 However, the study also showed that five of 21 women underwent unnecessary intervention, as the sonographic defect could not be clinically located, despite surgical exploration.31 Therefore, the use of endoanal US immediately after delivery and before repair was not recommended.

Women with OASIS should undergo follow-up after delivery to assess symptoms of faecal incontinence. Currently, there is no consensus regarding the optimal mode of delivery for these women in subsequent pregnancies. Scheer et al32 and Karmarkar et al33 assessed women who had OASIS in subsequent pregnancies using a questionnaire, endoanal US, and manometry. Vaginal delivery was recommended for asymptomatic women with normal findings. Women were reassessed after subsequent deliveries. There were no statistically significant differences in anal manometry findings, anal symptoms, or quality of life following subsequent vaginal delivery or caesarean section.32 33 In the study by Scheer et al,32 new OASIS occurred in only one woman after a vaginal delivery. Therefore, decisions regarding the mode of delivery for subsequent pregnancies after OASIS should be based on clinical symptoms, anal manometry, and endoanal US. This would help to preserve anal sphincter function and avoid unnecessary caesarean sections. Currently, the value of the above assessments is limited in Hong Kong. The significance of an incidental finding of occult anal sphincter defect remains uncertain.

The prevalence of OASIS determined by endoanal US was higher than the rate determined by clinical practice. This may indicate that additional training for midwives and doctors may be required to improve the detection of OASIS. At 6 to 12 months after delivery, OASIS was not associated with symptoms of faecal or flatal incontinence, but a longer-term study is needed to confirm these findings.

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.

Concept or design of the study: RYK Cheung, SSC Chan.
Acquisition of data: OYK Wan, RYK Cheung, LL Lee, SSC Chan.
Analysis or interpretation of data: SPK Kwok, SSC Chan.
Drafting of the article: All authors.
Critical revision for important intellectual content: SPK Kwok, OYK Wan, RYK Cheung, SSC Chan.

The results from this research have been presented, in part, at the following conferences:
1. Wan OYK, Cheung RYK, Chan SSC. 6th Annual Meeting of the Asia-Pacific Urogynecology Association and 13th Japanese Society of Pelvic Organ Prolapse Surgery Joint Conference–Young Doctors Session. Okinawa, Japan, 22-24 March 2019 (oral abstract presentation).
2. Wan OYK, Kwok SPK, Cheung RYK, Chan SSC. Hospital Authority Convention 2019, Hong Kong, 14-15 May 2019 (e-poster presentation).
3. Kwok SPK, Wan OYK, Cheung RYK, Lee LL, Chung JPW, Chan SSC. Obstetrical and Gynaecological Society of Hong Kong Annual Scientific Meeting 2019, Hong Kong, 1-2 June 2019 (oral presentation).

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

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

Ethics approval was obtained from local institute, The Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (Ref CRE-2013.332). Written informed consent was obtained from all participants.

1. Fenner DE, Genberg B, Brahma P, Marek L, DeLancey JO. Fecal and urinary incontinence after vaginal delivery with anal sphincter disruption in an obstetrics unit in the United States. Am J Obstet Gynecol 2003;189:1543-50. Crossref

2. Thiagamoorthy G, Johnson A, Thakar R, Sultan AH. National survey of perineal trauma and its subsequent management in the United Kingdom. Int Urogynecol J 2014;25:1621-7. Crossref

3. Baghestan E, Irgens LM, Børdahl PE, Rasmussen S. Trends in risk factors for obstetric anal sphincter injuries in Norway. Obstet Gynecol 2010;116:25-34. Crossref

4. Ekéus C, Nilsson E, Gottvall K. Increasing incidence of anal sphincter tears among primiparas in Sweden: a population-based register study. Acta Obstet Gynecol Scand 2008;87:564-73. Crossref

5. Hong Kong College of Obstetricians and Gynaecologists. Territory-wide audit in Obstetrics & Gynaecology; 2009.

6. Bates LJ, Melon J, Turner R, Chan SS, Karantanis E. Prospective comparison of obstetric anal sphincter injury incidence between an Asian and Western hospital. Int Urogynecol J 2019;30:429-37. Crossref

7. Carroli G, Mignini L. Episiotomy for vaginal birth. Cochrane Database Syst Rev 2009;(1):CD000081. Crossref

8. Cheung RY, Shek KL, Chan SS, Chung TK, Dietz HP. Pelvic floor muscle biometry and pelvic organ mobility in East Asian and Caucasian nulliparae. Ultrasound Obstet Gynecol 2015;45:599-604. Crossref

9. Andrews V, Sultan AH, Thakar R, Jones PW. Occult anal sphincter injuries—myth or reality? BJOG 2006;113:195-200. Crossref

10. Guzmán Rojas RA, Shek KL, Langer SM, Dietz HP. Prevalence of anal sphincter injury in primiparous women. Ultrasound Obstet Gynecol 2013;42:461-6. Crossref

11. Sultan AH, Kamm MA, Hudson CN, Thomas JM, Bartram CI. Anal-sphincter disruption during vaginal delivery. N Engl J Med 1993;329:1905-11. Crossref

12. Chan SS, Cheung RY, Yiu KW, Lee LL, Leung TY, Chung TK. Pelvic floor biometry during first singleton pregnancy and the relationship with symptoms of pelvic floor disorders: a prospective observational study. BJOG 2014;121:121-9. Crossref

13. Chan SS, Cheung RY, Yiu KW, Lee LL, Chung TK. Pelvic floor biometry in Chinese primiparous women 1 year after delivery: a prospective observational study. Ultrasound Obstet Gynecol 2014;43:466-74. Crossref

14. Chung MY, Wan OY, Cheung RY, Chung TK, Chan SS. Prevalence of levator ani muscle injury and health-related quality of life in primiparous Chinese women after instrumental delivery. Ultrasound Obstet Gynecol 2015;45:728-33. Crossref

15. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol 2008;61:344-9. Crossref

16. Sultan AH. Obstetric perineal injury and anal incontinence. Clinical Risk 1999;5:193-6. Crossref

17. Chan SS, Cheung RY, Yiu AK, et al. Chinese validation of Pelvic Floor Distress Inventory and Pelvic Floor Impact Questionnaire. Int Urogynecol J 2011;22:1305-12. Crossref

18. Roos AM, Thakar R, Sultan A. Outcome of primary repair of obstetric anal sphincter injuries (OASIS): does the grade of tear matter? Ultrasound Obstet Gynecol 2010;36:368-74. Crossref

19. de Leeuw JW, Struijk PC, Vierhout ME, Wallenburg HC. Risk factors for third degree perineal ruptures during delivery. BJOG 2001;108:383-7. Crossref

20. Aukee P, Sundström H, Kairaluoma MV. The role of mediolateral episiotomy during labour: analysis of risk factors for obstetric anal sphincter tears. Acta Obstet Gynecol Scand 2006;85:856-60. Crossref

21. Ramalingam K, Monga AK. Outcomes and follow-up after obstetric anal sphincter injuries. Int Urogynecol J 2013;24:1495-500. Crossref

22. Hauck YL, Lewis L, Nathan EA, White C, Doherty DA. Risk factors for severe perineal trauma during vaginal childbirth: a Western Australian retrospective cohort study. Women Birth 2015;28:16-20. Crossref

23. Grobman WA, Bailit JL, Rice MM, et al. Racial and ethnic disparities in maternal morbidity and obstetric care. Obstet Gynecol 2015;125:1460-7. Crossref

24. Davies-Tuck M, Biro MA, Mockler J, Stewart L, Wallace EM, East C. Maternal Asian ethnicity and the risk of anal sphincter injury. Acta Obstet Gynecol Scand 2015;94:308-15.Crossref

25. Stedenfeldt M, Øian P, Gissler M, Blix E, Pirhonen J. Risk factors for obstetric anal sphincter injury after a successful multicentre interventional programme. BJOG 2014;121:83-91. Crossref

26. Boreham MK, Richter HE, Kenton KS, et al. Anal incontinence in women presenting for gynecologic care: prevalence, risk factors, and impact upon quality of life. Am J Obstet Gynecol 2005;192:1637-42. Crossref

27. Melville JL, Fan MY, Newton K, Fenner D. Fecal incontinence in US women: a population-based study. Am J Obstet Gynecol 2005;193:2071-6. Crossref

28. Chan SS, Cheung RY, Yiu KW, Lee LL, Chung TK. Prevalence of urinary and fecal incontinence in Chinese women during and after first pregnancy. Int Urogynecol J 2013;24:1473-9. Crossref

29. Ng K, Cheung RY, Lee LL, Chung TK, Chan SS. An observational follow-up study on pelvic floor disorders to 3-5 years after delivery. Int Urogynecol J 2017;28:1393-9. Crossref

30. Walsh KA, Grivell RM. Use of endoanal ultrasound for reducing the risk of complications related to anal sphincter injury after vaginal birth. Cochrane Database Syst Rev 2015;(10):CD010826. Crossref

31. Faltin DL, Boulvain M, Floris LA, Irion O. Diagnosis of anal sphincter tears to prevent fecal incontinence: a randomized controlled trial. Obstet Gynecol 2005;106:6-13. Crossref

32. Scheer I, Thakar R, Sultan AH. Mode of delivery after previous obstetric anal sphincter injuries (OASIS)—a reappraisal? Int Urogynecol J Pelvic Floor Dysfunct 2009;20:1095-101. Crossref

33. Karmarkar R, Bhide A, Digesu A, Khullar V, Fernando R. Mode of delivery after obstetric anal sphincter injury. Eur J Obstet Gynecol Reprod Biol 2015;194:7-10. Crossref

According to the Hong Kong Cancer Registry,1 prostate cancer is the third most common cancer in men. As in other cancers, biopsy is needed for histological confirmation of the diagnosis of prostate cancer before treatment is initiated. With the increasing age of the population, the incidence rate of this cancer is expected to increase; the frequency of prostate biopsy will therefore also increase. Hodge et al2 introduced the systematic sextant biopsy protocol under transrectal ultrasound guidance. Transrectal ultrasound-guided prostate biopsy (TRUSPB) has since become a widely accepted and routinely performed technique to detect prostate cancer.3 In Hong Kong, most urologists use TRUSPB to confirm the diagnosis of prostate cancer, particularly in patients with elevated prostate-specific antigen (PSA) or abnormal digital rectal examination; TRUSPB is also used in patients undergoing active surveillance of prostate cancer. However, there are complications associated with the use of TRUSPB. Most notably, because the procedure is performed via the rectum, there is a risk of postprocedural sepsis; the incidence of sepsis ranged was 2% to 4% in contemporary series,4 5 and sepsis-related mortality has also been reported.6

An increasing number of studies have demonstrated success in cancer diagnosis with extended biopsy using transperineal ultrasound-guided prostate biopsy (TPUSPB). In an early report, Kojima et al7 retrospectively assessed the usefulness of TPUSPB, which differs from TRUSPB in terms of patient position, puncture route, puncture site, and ultrasound probe.8 Most importantly, the TPUSPB enables urologists to thoroughly prepare the perineum with a disinfectant solution to eliminate the possibility of skin flora contamination of the puncture site.9 In addition, this procedure involves puncture of perineal skin under the guidance of a side-fire ultrasound probe without penetration of rectal mucosa, thereby avoiding the possibility that intestinal flora are transferred to the blood stream. An Australian study group showed that TPUSPB, in combination with antibiotic prophylaxis, could almost entirely prevent sepsis complications.10 Some authors have suggested that TPUSPB may be performed without antibiotic prophylaxis, thus reducing the risk of generating antibiotic resistance.4 Based on these potential benefits, our centre introduced TPUSPB beginning in January 2018. Subsequently, we have completely replaced TRUSPB with TPUSPB. In this study, we aimed to compare the outcomes of our initial series of patients who underwent TPUSPB with those of our previous cohort of patients who underwent TRUSPB.

In this retrospective cohort study, we compared 100 patients who underwent TPUSPB with 100 patients who underwent TRUSPB in our centre. This study was approved by our institutional ethics committee. All 100 patients who underwent TPUSPB from January 2018 to May 2018 (TPUSPB group) were included; 100 patients who underwent TRUSPB from January 2016 to April 2016 were also included. The indications for biopsy for both groups were serum PSA >4 ng/dL, abnormal digital rectal examination, and surveillance biopsy for patients under active surveillance.

The following data were retrieved from hospital records and compared between the two groups: age, serum PSA level, prostate size, PSA density, prostate cancer detection rate, and complications (eg, admission due to acute retention of urine, rectal bleeding, haematuria, fever, and sepsis). We used the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) as an acute change in total sequential organ failure assessment score ≥2 points due to the infection11: (1) respiratory rate ≥22/min, (2) altered mental activity, and (3) systolic blood pressure ≤100 mm Hg.

For both groups, the indications for prostatic biopsies were serum PSA level >4 ng/dL, abnormal digital rectal examination, or follow-up biopsy for patients under active surveillance. All patients underwent pre-procedure blood tests and urine tests to ensure there was no bleeding tendency or positive urine culture. Patients using antiplatelet or anticoagulant treatment were required to discontinue drugs prior to undergoing biopsy. All patients used a sodium phosphate rectal enema in the morning of the procedure and took oral prophylactic antibiotics (1 g amoxicillin-clavulanate and 500 mg ciprofloxacin) 2 hours before the procedure. For TPUSPB, numbing cream (2.5% lidocaine and 2.5% prilocaine) was applied over the perineal region 1 hour before the procedure and 1% lidocaine (20 mL) was injected into the perineum as local anaesthesia (LA) immediately prior to prostate biopsy, at a 45-degree angle from the midline and approximately 15 mm above the anus on either side. Details of the two procedures are described below. After either procedure, all patients were given an additional 1-day course of oral antibiotics (1 g amoxicillin-clavulanate and 500 mg ciprofloxacin).

When undergoing TRUSPB, patients assumed the left lateral position, as shown in Figure 1. Prostate size was measured using a transrectal biplanar ultrasound probe. Subsequently, 10 core biopsies were taken: five cores were taken from each side of the prostate at the base, mid, apex, upper lateral, and lower lateral regions. Each procedure was 5 to 10 minutes in duration.

When undergoing TPUSPB, patients assumed the Lloyd-Davies position prior to injection of lidocaine for LA (described above). After lidocaine injection, a biplanar ultrasound probe was inserted through the anus. The prostate size was measured, and 14-gauge angiocatheters were then inserted at the sites previously used for LA injection, as shown in Figure 2. Ten core biopsies were obtained in a manner similar to that of TRUSPB. Because of the different orientation of the biopsy needle, the apical biopsy was targeted towards the anterior fibromuscular layer. The biopsy needle was maintained parallel to the probe to ensure clear visualisation of the targeted area, which was possible when the whole needle was completely visualised on ultrasound (Fig 3). Each procedure was 10 to 15 minutes in duration. We also assessed the pain experienced during the TPUSPB at three time points, namely during probe insertion into the anus, LA injection, and biopsy procedures, by verbal analogue scale (0-10) during TPUSPB.

Statistical analyses were performed using SPSS (Windows version 24.0; IBM Corp, Chicago [IL], United States). For continuous variables, age was compared by independent t test, while PSA, prostate size, and PSA density were compared by the Mann-Whitney U test as they did not exhibit normal distributions. Normality was assessed by normal QQ plots and the Shapiro-Wilk test. When comparing categorical variables, including cancer detection rates and complication rates, the Chi squared test was used if the expected count in each cell was >5; otherwise, Fisher’s exact test was used. In addition, the Cochran-Mantel-Haenszel test was performed to assess whether there was an association between the biopsy method and cancer detection rate according to clinical stage. Differences with a two-sided P value of <0.05 were considered to be statistically significant.

The patient characteristics, prostate cancer detection rates, and complications in patients who underwent TPUSPB, compared with those who underwent TRUSPB, are summarised in Table 1. Age did not significantly differ between the two groups. The median serum PSA in the TPUSPB group was higher than that in the TRUSPB group (12.0 ng/dL vs 9.5 ng/dL, P=0.047). Moreover, the median prostate size in the TPUSPB was smaller than that in the TRUSPB group (46.2 mL vs 56.8 mL, P=0.003). Therefore, the PSA density of TPUSPB group was higher than that in the TRUSPB group (0.27 vs 0.16, P=0.001).

Stratified prostate cancer detection rates in patients who underwent TPUSPB, compared with those who underwent TRUSPB, are listed in Table 2. There was no statistically significant difference in overall prostate cancer detection rate between the two groups. In subgroup analysis stratified by serum PSA level, the TPUSPB group had a higher prostate cancer detection rate than the TRUSPB group among patients with 20 to 100 ng/mL PSA (50% vs 15%, P=0.036). However, there were no statistically significant differences in prostate cancer detection rates among other subgroups according to PSA levels. There were also no statistically significant differences in prostate cancer detection rates between the two groups upon stratification according to PSA density or clinical staging.

In analysis of 35 patients with positive cores in the TPUSPB group, 16 (45.7%) patients had at least one positive core in the anterior fibromuscular stroma. Among these 16 patients, 14 were diagnosed with high-risk prostate cancers, with multiple positive cores in each patient. The relatively high proportion of high-risk prostate cancer in the TPUSPB cohort might explain the relatively high number of positive cores in the anterior fibromuscular stroma.

In the TPUSPB group, 19 patients had previous negative findings in TRUSPB; three of these 19 (15.7%) were diagnosed with prostate cancer based on the findings of TPUSPB. Two of the three tumours were detected in the anterior fibromuscular layer, and the remaining tumour was found in the apical zone. In the TRUSPB group, 36 patients had previous negative findings in TRUSPB; six (16.7%) of these were diagnosed with prostate cancer based on the findings of the current TRUSPB.

Concerning about the pain experienced during TPUSPB, the pain scores reported by patients during probe insertion into the anus, LA injection, and biopsy procedures were 1-2, 1-2, and 2-4, respectively.

With respect to complications, we initially planned to use the Sepsis-3 described above, but no patients in this study developed clinical signs of infection that met the criteria for sepsis. However, four (4%) patients in the TRUSPB group developed fever requiring hospital admission compared with none in the TPUSPB group (P=0.121) [Table 3]. At least 1 week of intravenous antibiotic treatment was prescribed for all four of those patients with fever. Three patients in the TPUSPB group and one in the TRUSPB group developed acute retention of urine (P=0.621). No patients in the TPUSPB group and one in the TRUSPB group had rectal bleeding (P=1.000). There were no admissions due to haematuria in either group.

Since Hodge et al2 introduced the systematic sextant biopsy protocol, TRUSPB has become the main approach to detect prostate cancer worldwide. In recent years, many urologists have described increased risks of infection and sepsis associated with TRUSPB.12 Fluoroquinolone was previously thought to provide effective antibiotic prophylaxis, thus preventing infections associated with TRUSPB; however, fluoroquinolone-resistant bacteria and extended-spectrum beta-lactamase–producing bacteria are present within the intestinal flora of 40.4% and 41.0% of Chinese patients, respectively.13 Accordingly, the rate of post-TRUSPB sepsis is rising both in Hong Kong6 and worldwide.14 To reduce the rates of infection, many strategies have been attempted, including augmented prophylactic antibiotic protocols. However, no strategies have prevented the development of sepsis due to the transfer of faecal bacteria into the blood stream through TRUSPB puncture sites.10 Transperineal ultrasound-guided prostate biopsy has been suggested as a potentially safer alternative. Notably, the indications, workups, medications, and numbers of cores are identical between TRUSPB and TPUSPB. However, the techniques differ with respect to multiple aspects, including patients’ position and puncture route,15 as described in the Methods section of this paper.

From the pain scores recorded during TPUSPB, most patients tolerated the procedure well. Because the entire procedure was performed under LA, all patients could be discharged on the same day without the need for general anaesthesia or monitored anaesthesia care. In Asian nations, many patients must travel a considerable distance to reach the hospital; therefore, 1-day out-patient procedures are preferable for patients and their relatives. Moreover, some patients are high-risk or unfit for general anaesthesia or monitored anaesthesia care; procedures performed under LA are therefore much safer and more practical for them.

Transrectal ultrasound-guided prostate biopsy neglects prostate cancer located in the anterior fibromuscular stroma, whereas the TPUSPB does not.16 In our study, a significant proportion of positive prostatic cores were found in the anterior fibromuscular stroma among patients in the TPUSPB group. Moreover, some patients with prior negative findings in TRUSPB were diagnosed with prostate cancer in the anterior fibromuscular stroma based on the results of TPUSPB. Therefore, TPUSPB may have an advantage over TRUSPB in patients with previous negative biopsy findings.

This study had some limitations. First, a consistent number of cores was biopsied in all patients in the TPUSPB group, irrespective of prostate size. To improve the rate of prostate cancer detection, some experts have advocated for the use of different numbers of prostate biopsies, based on prostate size16—more biopsies should be taken in patients with larger prostates. Second, TPUSPB required more time than TRUSPB. However, as each step is standardised, the duration of the procedure may decrease. Third, there was no documentation of pain scores in the TRUSPB group; thus, a comparison could not be performed. Future studies should address these limitations. In particular, a larger sample size is needed to confirm whether TPUSPB is superior with respect to the rate of prostate cancer detection.

In summary, TPUSPB avoids penetration of the rectal mucosa and possible transfer of intestinal flora to the blood stream during the procedure. This contributed to the lack of infections in the present study. With respect to prostate cancer detection rate, TPUSPB is at least comparable to TRUSPB. Therefore, an increasing number of urologists may adopt this technique in the future.

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.

Concept or design of study: KL Lo, KL Chui, CF Ng.
Acquisition of data: KL Lo, K Lim, JKM Li, J Wong, SK Mak.
Analysis or interpretation of data: KL Lo, CF Ng, SCH Leung, SF Ma.
Drafting of the manuscript: KL Lo, CF Ng.
Critical revision for important intellectual content: All authors.

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

We would like to thank for the support of the nursing staff in the Integrated Ambulatory Care Centre, North District Hospital for their support to the procedures.

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

This study was approved by the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (Ref CREC 2018.323).

1. Hong Kong Cancer Registry, Hospital Authority, Hong Kong SAR Government. Prostate cancer in 2016. Available from: http://www3.ha.org.hk/cancereg/pdf/top10/rank_2016.pdf. Accessed 17 May 2019.

2. Hodge KK, McNeal JE, Terris MK, Stamey TA. Random systematic versus directed ultrasound guided transrectal core biopsies of the prostate. J Urol 1989;142:71-4. Crossref

3. Heidenreich A, Bastian PJ, Bellmunt J, et al. EAU guidelines on prostate cancer. Part 1: screening, diagnosis, and local treatment with curative intent-update 2013. Eur Urol 2014;65:124-37. Crossref

4. Toner L, Bolton DM, Lawrentschuk N. Prevention of sepsis prior to prostate biopsy. Investig Clin Urol 2016;57:94-9. Crossref

5. Chan ES, Lo KL, Ng CF, Hou SM, Yip SK. Randomized controlled trial of antibiotic prophylaxis regimens for transrectal ultrasound-guided prostate biopsy. Chin Med J (Engl) 2012;125:2432-5.

6. Ng CF, Chan SY. Re: the incidence of fluoroquinolone resistant infections after prostate biopsy—are fluoroquinolones still effective prophylaxis? J Urol 2008;180:1570-1. Crossref

7. Kojima M, Hayakawa T, Saito T, Mitsuya H, Hayase Y. Transperineal 12-core systematic biopsy in the detection of prostate cancer. Int J Urol 2001;8:301-7. Crossref

8. Xue J, Qin Z, Cai H, et al. Comparison between transrectal and transperineal prostate biopsy for detection of prostate cancer: a meta-analysis and trial sequential analysis. Oncotarget 2017;8:23322-36. Crossref

9. Chang DT, Challacombe B, Lawrentschuk N. Transperineal biopsy of the prostate—is this the future? Nat Rev Urol 2013;10:690-702. Crossref

10. Grummet JP, Weerakoon M, Huang S, et al. Sepsis and ‘superbugs’: should we favour the transperineal over the transrectal approach for prostate biopsy? BJU Int 2014;114:384-8. Crossref

11. Singer M, Deutschman CS, Seymour CW, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016;315:801-10. Crossref

12. Steensels D, Slabbaert K, De Wever L, Vermeersch P, Van Poppel H, Verhaegen J. Fluoroquinolone-resistant E. coli in intestinal flora of patients undergoing transrectal ultrasound-guided prostate biopsy—should we reassess our practices for antibiotic prophylaxis? Clin Microbiol Infect 2012;18:575-81. Crossref

13. Tsu JH, Ma WK, Chan WK, et al. Prevalence and predictive factors of harboring fluoroquinolone-resistant and extended-spectrum β-lactamase-producing rectal flora in Hong Kong Chinese men undergoing transrectal ultrasound-guided prostate biopsy. Urology 2015;85:15-21. Crossref

14. Williamson DA, Barrett LK, Rogers BA, Freeman JT, Hadway P, Paterson DL. Infectious complications following transrectal ultrasound-guided prostate biopsy: new challenges in the era of multidrug-resistant Escherichia coli. Clin Infect Dis. 2013;57:267-74. Crossref

15. Emiliozzi P, Corsetti A, Tassi B, Federico G, Martini M, Pansadoro V. Best approach for prostate cancer detection: a prospective study on transperineal versus transrectal six-core prostate biopsy. Urology 2003;61:961-6. Crossref

16. Ong WL, Weerakoon M, Huang S, et al. Transperineal biopsy prostate cancer detection in first biopsy and repeat biopsy after negative transrectal ultrasound-guided biopsy: the Victorian Transperineal Biopsy Collaboration experience. BJU Int 2015;116:568-76. Crossref

‘Consanguinity’ is a term derived from the Latin word ‘consanguineus’, meaning ‘of the same blood’. In medical genetics, consanguineous union is generally referred as a union between couples related as second cousins or closer.1 The prevalence of consanguinity varies significantly worldwide, depending on cultural background, religious belief, and geography. The highest rates are estimated in the Near and Middle East and in Northern Africa, where 20% to 50% of marriages are consanguineous.1 2 The prevalence in Southern Europe, South America, and Japan is about 1% to 5%, whereas Western European countries, North America, and Oceania have the lowest prevalence of <1%.1 2

Consanguineous union increases the risk of genetic disorders in offspring, especially for autosomal recessive diseases. However, recent studies suggest that parental consanguinity is also a risk factor for other adverse outcomes, even in developed multi-ethnic countries where the prevalence of consanguineous marriages is perceived as lower. For example, in Vienna where the background consanguinity rate was <1%, Posch et al3 reported that 39.7% of consanguineous couples had obstetric history of congenital malformations or genetic disorders. Becker et al4 reported that 6.1% of consanguineous couples were referred to a specialist centre in Germany for a history of major fetal anomalies. A 10-year retrospective analysis conducted in Australia, where the consanguinity rate is 5.5%, concluded that parental consanguinity was associated with higher rates of threatened premature labour, fetal congenital abnormality, stillbirth, and perinatal mortality.5 In that study, consanguinity was also found to be an independent risk factor of nearly 3-fold for stillbirth.

In Hong Kong, parental consanguinity is more frequent among non-Chinese ethnic minorities, which account for 8% of the total population.6 Internationally, healthcare workers lack knowledge on the risks of consanguinity.7 8 9 Inconsistencies in information provided during genetic counselling and screening has been observed.10 Consanguineous couples are often unaware of the potential health hazards in their offspring.11 12 13 The level of concern and awareness of the adverse effects of parental consanguinity among patients and physicians is low, and available data on consanguinity in Hong Kong are limited. Therefore, in the present study, we aimed to clarify the prevalence and characteristics of pregnancies from consanguineous unions in Hong Kong, and to assess the related effects on maternal, perinatal, and child health outcomes.

The Prenatal Diagnosis Clinic in Tuen Mun Hospital is responsible for counselling consanguineous couples. Dating ultrasound and counselling sessions for Down syndrome screening are arranged for all pregnant women who have their booking appointment in our locality. At the booking appointment, patients are also asked about consanguinity. Hospital-accredited interpreters are arranged for couples who are not fluent in Cantonese or English. Identification of consanguineous cases depends on self-reporting by couples. A pedigree chart is constructed for each case. Couples are counselled about the possible effects of parental consanguinity on pregnancy outcomes, and advised to attend antenatal care regularly.

A retrospective cohort study of all parental consanguinity cases over a 10-year period from 1 January 2007 to 31 December 2016 was conducted. The antenatal records of these cases were reviewed. Details were gathered about pregnancy loss, fetal congenital abnormalities, pregnancy and perinatal outcomes, and neonatal and childhood development in the preceding pregnancy. The family history of each case was also collected from patient records, including known genetic or congenital anomalies, or intellectual or developmental disabilities. A morphology scan was arranged for consanguineous cases. Each family pedigree was studied to determine the degree of parental consanguinity (Fig). Only couples fulfilling the definition of consanguineous unions (second cousins or closer) were included for analysis in the present study.

Socio-demographic characteristics were collected, including ethnicity, maternal and paternal age, religious beliefs, working status, education level, and occupation. Maternal antepartum and peripartum characteristics, and fetal and perinatal information were available. Information about the neonatal, infancy, and childhood outcomes of the offspring were retrieved from the public sector electronic record system.

The relationship between consanguinity and fetal, neonatal, infant, or childhood diseases that required long-term paediatric management was evaluated and categorised into one of three categories:

Category A—Improbable association with consanguinity: cases known to be caused by numerical or structural chromosomal abnormalities, or not to have an autosomal recessive mode of inheritance;

Category B—Probable association with consanguinity: cases known to have an autosomal recessive mode of inheritance, particularly when both parents were found to be the carriers of genetic disorders; and

Category C—Possible/unclear association with consanguinity: cases where the mode of inheritance was unclear, or when genetic testing was unremarkable.

The characteristics and outcomes of consanguineous cases were compared with a control group of non-consanguineous unions. The next record of a non-consanguineous case of the same ethnicity after that of a case of consanguineous union was selected as the control. This ensured the similar composition of ethnicity which might have socio-economic effects on the maternal and fetal outcomes within the study and control groups.14 As some consanguineous couples might have contributed more than one pregnancies in our database, only adverse past obstetric outcome in the immediately preceding pregnancy was counted in the analysis, and any positive family history reported by such couples was counted as one case only, in order to prevent duplicated entries for multigravida women. Most previous studies have not evaluated the effects of closer consanguinity that might increase risks of hereditary disorders.5 15 16 To evaluate the effect of degree of inbreeding, comparisons were made among ‘first cousin or closer’ (including first cousin and double first cousin), ‘beyond first cousin’ (including first cousin once removed and second cousin), and non-consanguineous relationships.

Approval of this study was granted by the research and ethics committee of the study hospital. Guidelines for reporting observational studies according to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement were followed.

Statistical analysis was performed using SPSS (Windows version 22.0; IBM Corp, Armonk [NY], US). Cross-tabulation between degrees of consanguinity and the different variables was performed in order to evaluate the characteristics of the study population. Differences in continuous variables were compared using t test or one-way analysis of variance. Differences in categorical variables were analysed with Chi squared test or Fisher’s exact test. Linear regression was carried out to adjust the collinearity among variables. Multivariate logistic regression analysis was used to determine the risk of consanguinity for adverse pregnancy and perinatal outcomes, with adjustment of significant confounders. Adjusted odds ratio (OR) with 95% confidence interval (CI) were calculated. Statistical significance was established for P<0.05.

Of 56 657 fetuses, 334 (0.6%) were conceived by consanguineous parents; of these, the majority (85.0%, 284 of 334) were ethnic Pakistani (among whom the prevalence of consanguineous union is highest, at 30.5%), followed by Indian (6.2%), Nepalese (2.7%), Filipino (0.4%), and Chinese (0.04%) [Table 1]. Of all consanguineous unions, the majority were first cousin consanguineous unions (76.6%) and double first cousin unions (1.8%); together, these were categorised as first cousin or closer (≤1C) unions. The remainder were categorised as beyond first cousin (>1C) unions, and included first cousin once removed unions (5.1%), and second cousin unions (16.5%). Comparison of background variables including maternal and paternal age, education level, religion, length of stay in Hong Kong, marital status, working status, occupation, parity, and body mass index showed no significant differences between the consanguineous group and the non-consanguineous control group (Table 2).

Women in consanguineous unions were significantly more likely to have experienced congenital abnormality (10.5% vs 0.4%; P<0.001), unexplained intrauterine fetal demise (4.2% vs 0.4%; P=0.005) and neonatal death (4.6% vs 1.2%; P=0.024) in the preceding pregnancy, and family history of congenital abnormality (4.6% vs 0%; P<0.001) than were non-consanguineous controls (Table 3). Down syndrome screening was offered to all women, but the attendance was only about one-fifth for all groups.

In terms of major maternal and perinatal complications, there were no significant differences between the non-consanguineous control group and the overall consanguineous group or the subgroups, except that pregnancies of ≤1C unions were more often complicated with pre-eclampsia (4.2% vs 1.2%; P=0.02) than were those of the non-consanguineous control group (Table 4).

Altogether there were 58 fetuses and 14 fetuses having different abnormalities, from 55 consanguineous and 14 control couples respectively (Table 5). Offspring of consanguineous couples had a higher risk of having category C disorders (OR=4.60; 95% CI=2.35-9.00) or category B disorders (OR=8.70; 95% CI=1.06-71.36), compared with those of non-consanguineous couples. The overall prevalence of category C disorders (14.7%) was higher than that of category B disorders (2.4%). Compared with the non-consanguineous control group, the prevalence of category C disorders was significantly higher in the ≤1C subgroup (OR=5.59; 95% CI=2.83-11.06); it was lower in the >1C subgroup, but the difference was not significant.

The prevalence of structural malformations was higher in the consanguineous group than that in the non-consanguineous control group, especially for those abnormalities involving cardiovascular, musculoskeletal, and urological systems (Table 5). Parental consanguinity also significantly increased the risk of developmental delay in offspring of consanguineous couples (OR=6.72, 95% CI=1.48-30.63) and in those of ≤1C couples (OR=7.64, 95% CI=1.64-35.58). Autism spectrum disorder was more prevalent in offspring of consanguineous couples (2.1%) than in those of non-consanguineous couples (0%) [P=0.008]. The diseases recorded in the consanguineous group and in the control group are detailed in online supplementary Appendices 1 and 2, respectively.

To the best of our knowledge, this is the first comprehensive study in Hong Kong describing the prevalence of parental consanguinity. Our results support those of previous studies that revealed a higher prevalence of parental consanguinity in certain ethnic groups, and the higher prevalence of known genetic disorders (category B) among their offspring. In addition, our study has revealed that the prevalence of fetal structural abnormalities, developmental delay, and autism spectrum disorders (category C) are also high. This has implications for prenatal counselling and diagnosis, and related healthcare services.

Our comparison of maternal age and parity showed no significant difference between the consanguineous group and control group. This is in contrast to findings by Islam et al16 and Hosseini-Chavoshi et al,17 who found that women in consanguineous unions were younger and of higher parity in Iran and Oman, where the consanguinity rate was more than 30%. Studies in India and Pakistan populations also showed that mothers in consanguineous relationships were more likely to be socially and economically disadvantaged.11 18 The similarity in the socio-economic characteristics between the consanguineous and non-consanguineous unions of our study indicates that socio-economic factors are unlikely to be causes of the poorer fetal outcomes, both in the index pregnancy and the preceding pregnancy, found in our consanguineous group.

We identified eight offspring with autosomal recessive diseases in the consanguineous group, including three cases of beta-thalassaemia major and five cases of other rarer diseases (online supplementary Appendix 1). Although the carrier status of thalassaemia can be screened by low mean corpuscular volume of red blood cells, the carrier status of other recessive disorders can be more complex. For some disorders, comprehensive genetic carrier screening using exome sequencing is required.4 19 20 21 Our data provide useful information for preconception counselling for consanguineous couples. However, exome sequencing is expensive, and this screening test is not yet available in public hospitals. Health education and information about the availability of carrier screening should be provided to all pregnant women, regardless of cultural, religious, or socio-economic background. Once a consanguineous couple is diagnosed to be the carrier of a genetic disease, they should be encouraged to discuss carrier screening with their siblings, who may also carry the same recessive gene and be in consanguineous union. Access to obstetric care and genetic counselling services in prenatal diagnosis clinics allows couples to make informed choices. Knowledge on various cultural, religious, or socio-economic issues allows healthcare workers to provide appropriate support and to best advise patients.

Our results revealed that category C disorders are more prevalent among offspring of consanguineous couples, especially in the ≤1C subgroup. Fetal structural ultrasonographic examination should be offered to ≤1C couples, especially for the cardiovascular, urological, and skeletal systems.22 23 24 25 26 Detailed genetic counselling and investigation services must be offered to ≤1C couples if fetal abnormalities are detected.3 4

Our results revealed increased risk of developmental and behavioural disorders for offspring of consanguineous couples. However, disorders such as developmental delay and autism spectrum disorder are not diagnosable before birth. Preconception and prenatal counselling should be offered to consanguineous couples, who should also be reminded about regular postnatal follow-up examinations, in order to avoid any delay in diagnosing any developmental or behavioural disorders.27

Pakistani ethnicity accounted for only 1.6% of all fetuses but 85% of consanguineous couples in our study. According to the Hong Kong 2016 population by-census, 0.25% of the total Hong Kong population was of Pakistani ethnicity.6 However, the majority of this local Pakistani population is within potentially reproductive age-groups (15-24 years, 19.2%; 25-34 years, 14.9%; 35-44 years, 21.3%), and they tend to have more children per couple than do ethnic Chinese couples.6 It is essential to include information about the increased risks of parental consanguinity during the antenatal care and provide appropriate genetic counselling once a consanguineous couple is identified.

In addition to poor fetal outcomes, we also found a 3-fold increased risk of pre-eclampsia among women in ≤1C unions. Familial aggregation and possible genetic correlation of pre-eclampsia have been observed, but the exact effect of consanguinity remains controversial.28 29 Mumtaz et al15 suggested that parental consanguinity is a risk factor of 1.6-fold for preterm birth at less than 33 weeks of gestation. Low birth weight has also been associated with first-cousin relationships, but the risk increase was found to be marginal (OR=1.36)30. Our study did not confirm higher incidences of antepartum, peripartum, neonatal and perinatal complications in overall consanguinity. Findings on the effect of consanguinity on various complications are inconsistent, especially when these complications are multifactorial in pathogenesis.5 15 27 29 30

One limitation of our study is the retrospective nature that might have led to incompleteness of information for analysis, especially when previous pregnancies were not in Hong Kong. Another limitation is that some of the fetal abnormalities classified under category C may in fact be category B disorders, as some of them recurred in the same couples (online supplementary Appendix 1); the majority of category C disorders did not receive genetic investigations. However, there is a high dependence on public health service in our locality, and this facilitated data retrieval of postnatal, infancy, and childhood outcomes of the offspring. Different types of parental consanguinity were also included in our analysis to provide the stratified risks according to the degree of inbreeding. Collection of socio-economic characteristics was also comprehensive. The same composition of ethnicity in both the consanguineous and control groups further minimised the socio-economic confounding effects in the analysis. Another limitation is that the genetic data were often incomplete or not up-to-date for the studied cases, which were recorded from 2007 to 2016.

It is recommended that a territory-wide prospective study is conducted on consanguineous couples to further delineate their healthcare needs in Hong Kong.

Identification of consanguineous couples is essential to ensure appropriate referral for preconception or prenatal counselling and diagnosis. Our study showed the majority of consanguineous unions in Hong Kong are of Pakistani ethnicity. International studies have reported that in addition to recessive genetic diseases, offspring of consanguineous unions have higher incidences of non–genetically confirmed structural abnormalities, developmental delay, and autism spectrum disorders. The present study confirms this in the Hong Kong population. Information on the increased risks associated with parental consanguinity should be included in genetic counselling for consanguineous couples, so that they can make informed decisions.

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.

Concept or design of the study: All authors.
Acquisition of data: KH Siong.
Analysis or interpretation of data: KH Siong, TY Leung.
Drafting of the manuscript: KH Siong, TY Leung.
Critical revision for important intellectual content: All authors.

The authors have no conflicts of interest to disclose.

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

Ethics approval was obtained from New Territories West Cluster Clinical Research Ethics Committee (Ref NTWC/CREC/18012).

1. Bittles A. Consanguinity and its relevance to clinical genetics. Clin Genet 2001;60:89-98. Crossref

2. Consang.net. Global prevalence of consanguinity. Available from: http://www.consang.net/index.php/Global_prevalence. Accessed 1 Apr 2018.

3. Posch A, Springer S, Langer M, Blaicher W, Streubel B, Schmid M. Prenatal genetic counseling and consanguinity. Prenat Diagn 2012;32:1133-8. Crossref

4. Becker R, Keller T, Wegner RD, et al. Consanguinity and pregnancy outcomes in a multi-ethnic, metropolitan European population. Prenat Diagn 2015;35:81-9. Crossref

5. Kapurubandara S, Melov S, Shalou E, Alahakoon I. Consanguinity and associated perinatal outcomes, including stillbirth. Aust N Z J Obstet Gynaecol 2016;56:599-604. Crossref

6. Hong Kong SAR Government. Hong Kong 2016 Population By-census. Available from: https://www.bycensus2016.gov.hk/en/bc-articles.html. Accessed 1 Apr 2018.

7. Barrett P. A review of consanguinity in Ireland—estimation of frequency and approaches to mitigate risks. Ir J Med Sci 2016;185:17-28. Crossref

8. Teeuw ME, Hagelaar A, ten Kate LP, Cornel MC, Henneman L. Challenges in the care for consanguineous couples: an exploratory interview study among general practitioners and midwives. BMC Fam Pract 2012;13:105. Crossref

9. Aalfs CM, Smets EM, de Haes HC, Leschot NJ. Referral for genetic counselling during pregnancy: limited alertness and awareness about genetic risk factors among GPs. Fam Pract 2003;20:135-41. Crossref

10. Bennett RL, Hudgins L, Smith CO, Motulsky AG. Inconsistencies in genetic counseling and screening for consanguineous couples and their offspring: the need for practice guidelines. Genet Med 1999;1:286-92. Crossref

11. Joseph N, Pavan KK, Ganapathi K, Apoorva P, Sharma P, Jhamb JA. Health awareness and consequences of consanguineous marriages: a community-based study. J Prim Care Community Health 2015;6:121-7. Crossref

12. Jaber L, Romano O, Halpern GJ, Livne I, Green M, Shohat T. Awareness about problems associated with consanguineous marriages: survey among Israeli Arab adolescents. J Adolesc Health 2005;36:530. Crossref

13. Teeuw ME, Loukili G, Bartels EA, ten Kate LP, Cornel MC, Henneman L. Consanguineous marriage and reproductive risk: attitudes and understanding of ethnic groups practising consanguinity in Western society. Eur J Hum Genet 2014;22:452-7. Crossref

14. Khalil A, Rezende J, Akolekar R, Syngelaki A, Nicolaides KH. Maternal racial origin and adverse pregnancy outcome: a cohort study. Ultrasound Obstet Gynecol 2013;41:278-85. Crossref

15. Mumtaz G, Nassar AH, Mahfoud Z, et al. Consanguinity: a risk factor for preterm birth at less than 33 weeks’ gestation. Am J Epidemiol 2010;172:1424-30. Crossref

16. Islam MM. Effects of consanguineous marriage on reproductive behaviour, adverse pregnancy outcomes and offspring mortality in Oman. Ann Hum Biol 2013;40:243-55. Crossref

17. Hosseini-Chavoshi M, Abbasi-Shavazi MJ, Bittles AH. Consanguineous marriage, reproductive behaviour and postnatal mortality in contemporary Iran. Hum Hered 2014;77:16-25. Crossref

18. Bhopal RS, Petherick ES, Wright J, Small N. Potential social, economic and general health benefits of consanguineous marriage: results from the Born in Bradford cohort study. Eur J Public Health 2014;24:862-9. Crossref

19. Teebi AS, El-Shanti HI. Consanguinity: implications for practice, research, and policy. Lancet 2006;367:970-1. Crossref

20. Fareed M, Afzal M. Genetics of consanguinity and inbreeding in health and disease. Ann Hum Biol 2017;44:99-107. Crossref

21. Committee on Genetics. Committee Opinion No. 690: Carrier screening in the age of genomic medicine. Obstet Gynecol 2017;129:e35-40. Crossref

22. Stoll C, Alembik Y, Roth MP, Dott B. Parental consanguinity as a cause for increased incidence of births defects in a study of 238,942 consecutive births. Ann Genet 1999;42:133-9.

23. Sheridan E, Wright J, Small N, et al. Risk factors for congenital anomaly in a multiethnic birth cohort: an analysis of the born in Bradford study. Lancet 2013;382:1350-9. Crossref

24. Nabulsi MM, Tamim H, Sabbagh M, Obeid MY, Yunis KA, Bitar FF. Parental consanguinity and congenital heart malformations in a developing country. Am J Med Genet A 2003;116:342-7. Crossref

25. Becker SM, Al Halees Z, Molina C, Paterson RM. Consanguinity and congenital heart disease in Saudi Arabia. Am J Med Genet 2001;99:8-13. Crossref

26. Yunis K, Mumtaz G, Bitar F, et al. Consanguineous marriage and congenital heart defects: a case-control study in the neonatal period. Am J Med Genet A 2006;140:1524-30.

27. Abbas HA, Yunis K. The effect of consanguinity on neonatal outcomes and health. Hum Hered 2014;77:87-92. Crossref

28. Berends AL, Steegers EA, Isaacs A, et al. Familial aggregation of preeclampsia and intrauterine growth restriction in a genetically isolated population in The Netherlands. Eur J Hum Genet 2008;16:1437-42. Crossref

29. Sezik M, Ozkaya O, Sezik HT, Yapar EG, Kaya H. Does marriage between first cousins have any predictive value for maternal and perinatal outcomes in pre-eclampsia? J Obstet Gynaecol Res 2006;32:475-81.

30. Poorolajal J, Ameri P, Soltanian A, Bahrami M. Effect of consanguinity on low birth weight: a meta-analysis. Arch Iran Med 2017;20:178-84.

Clostridioides difficile infection (CDI) is a leading cause of healthcare-associated diarrhoea and is associated with significant morbidity and mortality.1 In the United States, C difficile has become the most lethal acute enteric pathogen, with the Centers for Disease Control designating it as an urgent antibiotic resistance threat in 2015, one of only three pathogens to earn this distinction (http://www.cdc.gov/drugresistance/biggest_threats.html).2 In Hong Kong, an annual increase of 26% in CDI was noted from 2006 to 2014.3 This is even more alarming for older adults aged ≥65 years, in whom the estimated crude incidence rate is 133 to 207 cases per 100 000 population.4 It is believed that CDI arises when the normal gut microbiota is disrupted. This allows for the colonisation of C difficile, whose production of cytotoxins leads to disease.5 A recent study showed that apart from microbiota dysbiosis, enteric virome dysbiosis may also play an important role.6 Risk factors include nursing home care, recent hospitalisation, antibiotics exposure, and proton-pump inhibitor use.4 First-line treatment for non-fulminant C difficile–associated diarrhoea includes the cessation of unnecessary medications and treatment with vancomycin or fidaxomicin. For more severe cases, a combination of vancomycin and intravenous metronidazole with early consideration of surgery may be required.7 However, it is well known that a significant proportion of cases relapse or recur despite first-line therapy.8 For recurrent cases, taper and pulse regimens of vancomycin can be offered, but response rates decline further with multiple recurrences.9 Fidaxomicin is another promising treatment option, but its usage is limited by its price and availability.10 These difficult-to-treat cases are associated with extended hospital stays and may result in widespread nosocomial outbreaks. In their seminal paper, van Nood et al11 demonstrated that faecal microbiota transplantation (FMT), in which infusion of faeces from healthy donors is given to individuals with disease, was efficacious for recurrent CDI. Since then, this finding has been confirmed by a systematic review and meta-analysis of >160 clinical studies.12 In view of the limited options available for treatment of recurrent or refractory cases, we decided to establish a pilot FMT service to address this treatment gap. Initially, screening of relatives meeting strict donor criteria and willing to donate fresh stools was performed; later on, with the establishment of the Healthy Donor Stool Biobank by the Faculty of Medicine, The Chinese University of Hong Kong in 2017, pre-screened frozen stools were made available. The objectives of this study, the first of its kind in Hong Kong, was to assess the safety, efficacy, and technical and logistical feasibility of performing FMT.

We conducted a single-centre, retrospective review of all consecutive cases of recurrent or refractory CDI with FMT performed at an academic hospital from 2013 to 2018. Data on patient demographics, co-morbidities, route of administration of FMT, treatment efficacy, adverse events, and other relevant clinical data were extracted from the Clinical Management System of the Hospital Authority, Hong Kong or from review of case notes under the auspices of the FMT registry. Stringent donor criteria based on guidance and protocols were applied to fresh and frozen donor stool to ensure patient safety.13 Screening for viral hepatitis, blood-borne viruses, pathogenic bacteria, enteric viruses, syphilis, multidrug resistant organisms, Helicobacter pylori, C difficile, and parasites was performed (online supplementary Appendix). The FMT product was delivered either via a feeding tube or through scope channels during oesophagogastroduodenoscopy (OGD) or colonoscopy. Feeding tube position was confirmed prior to FMT delivery. For the OGD route, the endoscopist performed a routine OGD to ensure there were no ulcerations or other contra-indications for FMT. Afterwards, the endoscopist tried to pass the scope as distally as possible (at least to the second or third parts of the duodenum) to minimise reflux of the FMT back into the stomach. During the procedure carbon dioxide for luminal insufflation was used to minimise gaseous distention, discomfort, and the urge to retch or vomit. Before FMT delivery, the patient is sat upright to minimise the risk of aspiration. For the colonoscopy route, the endoscopist performed a routine colonoscopy to ensure there are no large ulcers or other contra-indications for FMT. Afterwards, the endoscopist tried to pass the scope as proximally as possible, but if the patient’s tolerance was an issue, administering FMT into the left side of the colon or even the rectum was considered reasonable, depending on the premorbid state. The majority of FMT procedures were performed in the Endoscopy Centre with close monitoring, followed by further observation in the wards for any potential adverse events. The present report was compiled in accordance with the PROCESS statement.14

A total of 24 patients with recurrent or refractory CDI who received FMT were identified. The patients’ baseline characteristics are shown in the Table. Their median age was 70 years (interquartile range=45.0-78.3 years). More than two-thirds of the patients were male. More than half of the patients were in either a bedridden or chair-bound state, a surrogate of poor functional status. The majority (>80%) of patients had been hospitalised within the most recent 3 months or were long-term care facility residents. All patients had at least one co-morbidity. The most common co-morbidity was hypertension (n=10, 41.7%), followed by inflammatory bowel disease (IBD) [n=6, 25.0%] and stroke (n=5, 20.8%). The FMT was performed by experienced gastroenterologists in accordance with published protocols11 and the latest guidelines.15 All patients were given 50 g of FMT product unless otherwise stated. The FMT was delivered via feeding tube in 11 (45.8%), OGD in eight (33.3%), and colonoscopy in six (25.0%) of the patients (one patient had FMT performed via both OGD and colonoscopy during the same session). Resolution of diarrhoea without relapse within 8 weeks was achieved in 21 out of 24 patients (87.5%). Three patients (12.5%) did not show significant improvement in their symptoms after the therapy; therefore, a clinical decision was made to perform repeat FMT, after which all three patients showed resolution of diarrhoea. Two of those were confirmed C difficile–negative on repeat stool testing; the remaining patient was frail and did not save repeat stools, but no documented recurrence within 8 weeks was noted. No deaths occurred at 30 days. The procedure was generally well tolerated, with no serious adverse events attributable to FMT. The most common complication was abdominal pain (n=3, 12.5%) after FMT. Bloating was reported in one (4.2%) patient.

Because it replenishes colonic microbial diversity, FMT removes ecological niches that would otherwise be occupied by C difficile.16 A recent study has shown that the prevalence of CDI in Asia is similar to the high prevalence in North America and Europe.17 Furthermore, it is well known that the treatment efficacy of antibiotics declines with multiple recurrences,9 rendering them less effective in difficult-to-treat cases. We established Hong Kong’s first FMT service to address the treatment gap that currently exists. Our results suggest that the efficacy of FMT for recurrent or refractory CDI is comparable to that reported in the literature, with an excellent safety profile.11 12

According to the Census and Statistics Department, the population of Hong Kong is projected to increase from 6.8 million in mid-2003 to 8.38 million in mid-2033 with a continuous ageing trend.18 The proportion of those aged ≥65 years is projected to rise markedly, from 11.7% in 2003 to 27% in 2033. Life expectancy in Hong Kong has already increased to age 79.5 years, behind only Japan and Switzerland.18 Increasing numbers of elderly people living in residential care homes, together with the widespread use of antibiotics and proton-pump inhibitors4, mean that the incidence of CDI in elderly people will likely increase in Hong Kong, as it has in other developed countries.18

The incidence of IBD in Hong Kong has risen by almost 3-fold in the past two decades.19 Recently in the West, as the incidence and severity of CDI has increased in the general population, even greater increases have been described in patients with IBD.2 In the present study, a significant proportion of patients with CDI also had IBD, and it is likely that the incidence and severity of CDI in IBD patients in Hong Kong will increase further in future. Establishment of FMT services has been advocated for healthcare systems to effectively manage expected increases in refractory or recurrent CDI.20

A concerted effort by policymakers will be necessary to establish a dedicated centre to provide territory-wide FMT services to tackle the growing disease burden. Such a centre would require a multidisciplinary team involving gastroenterologists, microbiologists, infectious disease specialists, specialty nurses, and research personnel together with the cooperation of wards, the healthy donor stool biobank, and endoscopy centres. Such an FMT centre may receive territory-wide consultations and referrals. It could also provide training, knowledge transfer, and accreditation to healthcare professionals. Currently, the status of FMT as a therapeutic agent is still evolving. The United States Food and Drug Administration classifies human stool as a biological agent and asserts that its use in FMT should be regulated, although their draft guidance mentions that it intends to exercise enforcement discretion for its use in recurrent CDI that does not respond to standard therapies.21 This highlights the importance of proper oversight and governance to mitigate any potential risks that may arise. Data on long-term safety outcomes are also scarce, highlighting the importance and role of registries to provide more insight, another function a centralised and specialised FMT centre would be able to undertake.

Our study has several limitations. As a case series, the data presented are mainly descriptive and uncontrolled with the possibility of bias. Further, the sample size is relatively small. Controlled trials with a larger sample size are required to confirm these findings and to optimise the timing of FMT administration. Despite these limitations, resolution of diarrhoea was at a similar rate in the present study to that reported in the literature.11 Furthermore, as the only centre in Hong Kong that has provided FMT, the results reported here would be highly relevant for both clinicians and policymakers concerning its safety, efficacy, and feasibility.

Our results show that FMT is safe, efficacious, and feasible for treating patients with recurrent or refractory CDI in Hong Kong. Given the lack of effective alternatives for difficult-to-treat cases of CDI, and demographic trends that will likely lead to increased incidence of CDI, demand for FMT is expected to rise. A territory-wide FMT service should be established to address this expected increase in demand.

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.

Concept or design of the study: All authors.
Acquisition of data: RN Lui, LHS Lau, KCY Cheung.
Analysis or interpretation of data: RN Lui, LHS Lau, KCY Cheung.
Drafting of the manuscript: RN Lui, SH Wong, PKS Chan, SC Ng.
Critical revision for important intellectual content: All authors.

All authors have disclosed no conflicts of interest.

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

The present study was reviewed and approved by the Joint Chinese University of Hong Kong/New Territories East Cluster Clinical Research Ethics Committee (Ref 2017.260).

1. Lessa FC, Mu Y, Bamberg WM, et al. Burden of Clostridium difficile infection in the United States. N Engl J Med 2015;372:825-34. Crossref

2. Khanna S, Shin A, Kelly CP. Management of Clostridium difficile infection in inflammatory bowel disease: expert review from the clinical practice updates committee of the AGA institute. Clin Gastroenterol Hepatol 2017;15:166-74. Crossref

3. Ho J, Dai RZ, Kwong TN, et al. Disease burden of Clostridium difficile infections in adults, Hong Kong, China, 2006-2014. Emerg Infect Dis 2017;23:1671-9. Crossref

4. Wong SH, Ip M, Hawkey PM, et al. High morbidity and mortality of Clostridium difficile infection and its associations with ribotype 002 in Hong Kong. J Infect 2016;73:115-22. Crossref

5. Mylonakis E, Ryan ET, Calderwood SB. Clostridium difficile–associated diarrhea: a review. Arch Intern Med 2001;161:525-33. Crossref

6. Zuo T, Wong SH, Lam K, et al. Bacteriophage transfer during faecal microbiota transplantation in Clostridium difficile infection is associated with treatment outcome. Gut 2018;67:634-43.

7. McDonald LC, Gerding DN, Johnson S, et al. Clinical Practice Guidelines for Clostridium difficile infection in adults and children: 2017 update by the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA). Clin Infect Dis 2018;66:987-94. Crossref

8. Pépin J, Routhier S, Gagnon S, Brazeau I. Management and outcomes of a first recurrence of Clostridium difficile–associated disease in Quebec, Canada. Clin Infect Dis 2006;42:758-64. Crossref

9. Johnson S. Recurrent Clostridium difficile infection: a review of risk factors, treatments, and outcomes. J Infect 2009;58:403-10. Crossref

10. Cornely OA, Miller MA, Louie TJ, Crook DW, Gorbach SL. Treatment of first recurrence of Clostridium difficile infection: fidaxomicin versus vancomycin. Clin Infect Dis 2012;55 Suppl 2:154-61. Crossref

11. van Nood E, Vrieze A, Nieuwdorp M, et al. Duodenal infusion of donor feces for recurrent Clostridium difficile. N Engl J Med 2013;368:407-15. Crossref

12. Lai CY, Sung J, Cheng F, et al. Systematic review with meta-analysis: review of donor features, procedures and outcomes in 168 clinical studies of faecal microbiota transplantation. Aliment Pharmacol Ther 2019;49:354-63. Crossref

13. Woodworth MH, Neish EM, Miller NS, et al. Laboratory testing of donors and stool samples for fecal microbiota transplantation for recurrent Clostridium difficile infection. J Clin Microbiol 2017;55:1002-10. Crossref

14. Agha RA, Borrelli MR, Farwana R, et al. The PROCESS 2018 statement: updating consensus Preferred Reporting Of CasE Series in Surgery (PROCESS) guidelines. Int J Surg 2018;60:279-82. Crossref

15. Cammarota G, Ianiro G, Tilg H, et al. European consensus conference on faecal microbiota transplantation in clinical practice. Gut 2017;66:569-80. Crossref

16. Seekatz AM, Aas J, Gessert CE, et al. Recovery of the gut microbiome following fecal microbiota transplantation. MBio 2014;5:e00893-14. Crossref

17. Borren NZ, Ghadermarzi S, Hutfless S, Ananthakrishnan AN. The emergence of Clostridium difficile infection in Asia: a systematic review and meta-analysis of incidence and impact. PLoS One 2017;12:e0176797. Crossref

18. Census and Statistics Department, Hong Kong SAR Government. Hong Kong population projection 2004-2033, report of the task force on population policy. Available from: https://www.censtatd.gov.hk/media_workers_corner/pc_rm/hong_kong_population_projections_2004_2033__/index.jsp. Accessed 8 Mar 2019.

19. Lui RN, Ng SC. The same intestinal inflammatory disease despite different genetic risk factors in the East and West? Inflamm Intest Dis 2016;1:78-84. Crossref

20. Costello SP, Tucker EC, La Brooy J, Schoeman MN, Andrews JM. Establishing a fecal microbiota transplant service for the treatment of Clostridium difficile infection. Clin Infect Dis 2016;62:908-14. Crossref

21. Guidance for industry: enforcement policy regarding investigational new drug requirements for use of fecal microbiota for transplantation to treat Clostridium difficile infection not responsive to standard therapies. Food and Drug Administration, US Department of Health and Human Services, US Government; 2016.