Objective To provide an in-depth systematic assessment of the global epidemiology of gonorrhoea infection in infertile populations.
Methods A systematic literature review was conducted up to 29 April 2019 on international databases and WHO regional databases, and reported following Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. All prevalence measures of gonorrhoea infection among infertile populations, based on primary data, qualified for inclusion. Infertile populations were broadly defined to encompass women/men undergoing infertility evaluation or treatment (infertility clinic attendees and partners). Pooled mean prevalence by relevant strata was estimated using random-effects meta-analysis. Associations with prevalence and sources of heterogeneity were explored using metaregression. Risk of bias was assessed using four quality domains.
Findings A total of 147 gonorrhoea prevalence studies were identified from 56 countries. The pooled mean prevalence of current gonorrhoea infection was estimated globally at 2.2% (95% CI 1.3% to 3.2%), with the highest prevalence in Africa at 5.0% (95% CI 1.9% to 9.3%). The mean prevalence was higher for populations with tubal factor infertility (3.6%, 95% CI 0.9%–7.7%) and mixed cause and unexplained infertility (3.6%, 95% CI 0.0% to 11.6%) compared with other diagnoses, such as ovarian and non-tubal infertility (0.1%, 95% CI 0.0% to 0.8%), and for secondary (2.5%, 95% CI 0.2% to 6.5%) compared with primary (0.5%, 95% CI 0.0% to 1.7%) infertility. Metaregression identified evidence of variations in prevalence by region and by infertility diagnosis, higher prevalence in women than men and a small-study effect. There was a trend of declining prevalence by about 3% per year over the last four decades (OR=0.97, 95% CI 0.95 to 0.99).
Conclusions Gonorrhoea prevalence in infertile populations is several folds higher than that in the general population, with even higher prevalence in women with tubal factor infertility and in individuals with secondary infertility. These findings support the potential role of gonorrhoea in infertility and suggest that some infertility is possibly preventable by controlling gonorrhoea transmission.
PROSPERO registration number CRD42018102934.
- Neisseria gonorrhoeae
- epidemiology (general)
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Infertility, ‘a disease characterised by failure to establish clinical pregnancy after 12 months of regular, unprotected sexual intercourse’,1 2 affects ~2% of reproductive-age women with no prior live birth and >10% of those with an earlier successful delivery.3 While infertility in men remains poorly quantified,4 available estimates by world region suggest a range of 2.5%–12.0%.5
A potential contributor to infertility, for both women and men, is a common STI caused by the bacterium Neisseria gonorrhoeae,6 in addition to Chlamydia trachomatis (CT).7–10 In 2016, the WHO estimated that nearly 87 million individuals acquired this infection globally, with incidence rates estimated at 20 per 1000 women and 26 per 1000 men.11 In women, gonorrhoea is often asymptomatic, complicating early detection and treatment and increasing their risk of cervicitis and pelvic inflammatory disease,12 13 while in men, it has been associated with epididymitis, epididymo-orchitis and chronic prostatitis.14–17 Untreated, these conditions may lead to subfertility/infertility.12 14 15 18
Despite their health, social and economic implications,19 20 STIs and infertility have long been a low priority on national policy agendas. Recently, the WHO formulated the ‘Global Health Sector Strategy on STIs, 2016–2021’, with the goal of ending STI epidemics as a public health concern by 2030.21 A key target is achieving by 2030 a 90% reduction in N. gonorrhoeae incidence.21 The urgency in addressing gonorrhoea is compounded by its recent classification as a ‘superbug’,22 given the widespread antimicrobial resistance, even to infection’s last-line treatment.23–26 Consequently, the WHO launched a global action plan to control gonorrhoea transmission and sequelae,27 28 including building a business case for the global public health value of gonococcal vaccines.29 30 Achieving WHO set targets entails fulfilment of five strategic directions/actions; the first is to understand the STI epidemic and burden, including subfertility/infertility, as a basis for advocacy, political commitment, national planning, resource mobilisation and allocation, implementation and programme improvement.21
This study was motivated by our recent work assessing CT prevalence levels in different at-risk populations in the Middle East and North Africa, where we identified an association between CT prevalence and infertility, with prevalence among infertile populations being three-fold higher than that among the general population.10 The present study aimed to characterise the global epidemiology of gonorrhoea infection in infertile populations by (1) systematically reviewing and synthesising evidence of infection prevalence, (2) estimating the pooled mean prevalence, stratified by WHO region among other key factors, and (3) exploring population-level associations with prevalence and sources of between-study heterogeneity.
Longitudinal studies examining gonorrhoea’s adverse health outcomes (a curable infection) are difficult/unethical to conduct. A recent study attempted to overcome this challenge through linking national testing databases to hospital records, but identified too few cases to reach conclusive evidence about gonorrhoea’s role in infertility.31 In the absence of direct evidence, our study aimed to provide indirect evidence for a link between gonorrhoea and infertility but strictly did not aim to nor can it establish causality. The underlying hypothesis is that current infection is of unknown duration and persistence to establish a causal link with infertility, but is often predictive of past exposure.32–34 This assertion is supported by several lines of evidence. It is established through tens of studies of different designs that gonorrhoea as well as chlamydia, being curable infections, carry a high risk of reinfection because of re-exposure to the same sexual partner or to other high-risk partners.33 35–37 As such, it can be assumed that a current gonorrhoea infection is strongly indicative of a previous gonorrhoea infection33 38 39; indeed studies have shown that the strongest predictor of current gonorrhoea infection is a history of gonorrhoea infection.32 40 For example, in the UK, a history of gonorrhoea infection was found to be the strongest predictor of current gonorrhoea infection even after controlling for other demographic and behavioural factors (adjusted OR 4.36, 95% CI 1.78 to 10.71).32
It is also established that there are strong correlations between exposure and the prevalence of different STIs, such as gonorrhoea and chlamydia,41 42 herpes simplex virus type 2 (HSV-2) and HIV43–45 (beyond the debated biological synergy46), even though these STIs could be acquired at different time frames. As such, exposure to an STI is a predictor of exposure to another STI. For instance, HSV-2 is often used as a proxy biomarker for HIV exposure and epidemic potential.43–45 Just as STI exposures acquired at different time points are correlated with each other, it is reasonable to expect that measures of gonorrhoea prevalence assessed at different times in the same population are also correlated.32 40 This is because, fundamentally, the driving factor of STI exposure is sexual risk behaviour47; current gonorrhoea infection in a population/person can be seen as a proxy of the past and present sexual risk behaviour of that population/person or person’s sexual partners.48 49 Studies also show that people tend to be consistent in their sexual risk behaviour over at least a few years’ duration.50–52
Detailed methodology has been previously published as a study protocol.53 A brief description is provided as follows.
Search strategy and selection criteria
A systematic review of gonorrhoea prevalence in infertile populations was conducted following Cochrane Collaboration guidelines,54 and reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines55 (checklist in online supplementary table 1).
Literature was searched, up to 29 April 2019 on PubMed and Embase, and up to 5 February 2019 on the WHO Index Medicus regional databases, using broad terms with no language or year restrictions (online supplementary box 1). Duplicate citations were excluded using a reference manager, EndNote (Thomson Reuters, USA). Title and abstract screening and full-text screening of relevant/potentially relevant citations were performed by HC and AM. Reference lists of reviews and relevant articles were further hand-searched.
Any article reporting prevalence of current urogenital infection or serological markers of gonorrhoea in infertile populations, based on primary data, qualified for inclusion. Infertile populations were broadly defined to include women/men undergoing infertility evaluation or treatment (infertility clinic attendees and partners). Studies in voluntarily sterile populations, based on the infection’s self-report, including <10 participants, or assessing gonorrhoea in tissue samples from the upper genital tract, were excluded.
Data extraction and synthesis
Data were extracted by HC and AM and double extracted by FA (extraction list in online supplementary box 2). In addition to the overall gonorrhoea measure, stratified measures were extracted whenever a stratum included ≥10 participants.
Studies assessing gonorrhoea using different assay types (nucleic acid amplification test (NAAT), culture, Gram stain and Ig among others) were extracted separately for different analyses. Studies applying the same assay to different biological specimens were included once based on a predefined order prioritising, for women, gonorrhoea detection in endocervical swabs, followed by vaginal and urine samples; and for men, detection in urethral swabs, followed by urine and semen samples.
Risk of bias and precision assessments
Informed by the Cochrane approach54 and existing literature,56–59 each study was rated as having ‘low’ versus ‘high’ risk of bias on four quality domains: (1) validity of infertility definition (follows WHO definition vs otherwise), (2) lack of exposure to antimicrobials for ≥1 week prior to collection of biological samples (ascertained vs otherwise), (3) consistency in assay used for infection ascertainment (same assay used to test all participants vs otherwise) and (4) response rate (≥80% vs <80%). A study with missing information for a specific domain was considered as having ‘unclear’ risk of bias for that domain. A study was deemed of ‘higher’ precision if its original sample tested ≥100 participants.
Pooled mean gonorrhoea prevalence and 95% CIs were estimated using random-effects meta-analysis. Here, overall prevalence was replaced by strata, whenever possible. For each study, only one final stratification was considered, based on a predefined priority order: country, sex, infertility diagnosis, infertility type, age and year of data collection. Stratified meta-analyses by relevant factors were further performed, and heterogeneity assessment was conducted.60 61
Metaregression analyses were conducted to explore sources of between-study heterogeneity and to examine associations with prevalence for the following predefined factors: WHO region (African region (AFRO), Americas (AMRO), Eastern Mediterranean (EMRO), European (EURO), Southeast Asia (SEARO), Western Pacific (WPRO)), sex, infertility type, infertility diagnosis, presence of urogenital signs and symptoms, assay type, median year of data collection, sample size/precision (to assess small-study effect) and risk of bias domains. Variables’ details/subgroupings are in online supplementary box 2 and online supplementary table 2.
Strength of evidence for an association with prevalence was deemed ‘good’ at 0.05<p value≤0.10 and ‘strong’ at p value≤0.05. Sensitivity analysis focusing on studies assessing current infection was performed.
Search results and scope of evidence
Figure 1 shows the study selection process. Search identified 9937 citations: 3603 through PubMed, 5141 through Embase and 1193 through the WHO Index Medicus databases. After excluding duplicates and screening titles and abstracts, 1410 unique reports underwent full-text screening. Of these, 89 were eligible for inclusion. The rest were excluded for reasons outlined in figure 1. Twenty-six additional reports were identified through reference list hand-searching. In sum, 115 reports contributing 147 gonorrhoea prevalence studies were included in the review. These yielded 184 stratified measures for meta-analyses.
There were 27 264 gonorrhoea test results from 56 countries. EURO contributed 44.2% of studies (n=65), AMRO 16.3% (n=24), AFRO 13.6% (n=20), SEARO 8.2% (n=12), WPRO and EMRO 7.5% (n=11) each, and multicentre/multiregional studies 2.7% (n=4). Most studies (n=107, 72.8%) assessed current infection, of which 26.2% were NAAT-based; 67.3% were culture-based; and 6.5% were Gram stain/gonozyme/fluorescent antibody-based. The rest either reported ever infection using IgG (n=20, 13.6%) or IgA (n=3, 2.0%), or were based on unclear assays (n=17, 11.6%). Studies are detailed in online supplementary tables 3-8.
Reported current infection prevalence across regions ranged from 0% to 53.0% (online supplementary tables 3-8). The median was 0%, as 57 out of 107 studies reported zero prevalence; it is difficult to identify a positive case for a low-prevalence infection in a study of a small sample size. The highest median current infection prevalence was for AFRO at 3.3%. Ever infection prevalence (IgG) ranged from 1.3% to 65.0%, with a median of 25.0%; the median per region ranged from 2.5% in EMRO to 39.1% in AFRO (online supplementary tables 3-8).
Risk of bias and precision assessments
Online supplementary tables 9 and 10 show the summarised and study-specific precision and risk of bias assessments. Briefly, 50.3% of studies were of higher precision (≥100 participants). Over a third (34.7%) followed WHO infertility definition; 1.3% included infertile participants for <12 months, while the rest (64.0%) did not report an infertility definition. Only 14.3% of studies excluded infertile participants exposed to antimicrobials in the week prior to sample collection; 6.1% may have included such participants; and information was missing for the rest of studies (79.6%). Almost all studies (96.6%) demonstrated consistency in gonorrhoea testing across infertile participants. Response rate was mostly unavailable (97.3%); studies were almost entirely facility/clinic-based or retrospective charts were review-based.
Studies were overall of reasonable quality (online supplementary table 9). Nearly all (98.6%) had low risk of bias in ≥1 quality domain and 41.5% had low risk of bias in ≥2 domains. Meanwhile, only 8.8% had high risk of bias in ≥1 quality domain and <1% had high risk of bias in ≥2 domains. Over 90% of studies had unclear risk of bias in ≥2 domains.
Summary estimates of pooled mean gonorrhea prevalence
Pooled mean prevalence of current gonorrhoea infection was globally at 2.2% (95% CI 1.3% to 3.2%), and regionally at 5.0% (95% CI 1.9% to 9.3%) in AFRO, 2.7% (95% CI 0.6% to 5.8%) in EMRO, 2.5% (95% CI 0.4% to 5.7%) in WPRO, 2.4% (95% CI 0.8% to 4.5%) in EURO, 1.0% (95% CI 0.0% to 3.4%) in AMRO and 0.0% (95% CI 0.0% to 0.06%) in SEARO (table 1). Meanwhile, mean ever infection prevalence was globally at 21.0% (95% CI 13.2% to 30.0%) and varied regionally from 5.4% (95% CI 1.2% to 12.0%) in AMRO to 46.6% (95% CI 28.4% to 65.3%) in AFRO (table 1).
Estimates varied by infertility diagnosis (table 1). Mean current infection prevalence was 3.6% (95% CI 0.9% to 7.7%) for tubal factor infertility (TFI), 3.6% (95% CI 0.0% to 11.6%) for mixed (samples combining different diagnoses) and unexplained infertility, 2.6% (95% CI 1.1% to 4.5%) for general/unspecified infertility, 1.4% (95% CI 0.2% to 3.3%) for male factor infertility and 0.06% (95% CI 0.0% to 0.8%) for ovarian and non-TFI infertility. This measure was also 2.5% for secondary infertility (95% CI 0.2% to 6.5%) and 0.5% for primary infertility (95% CI 0.0% to 1.7%).
Mean current infection prevalence varied by assay type: 0.7% (95% CI 0.08% to 1.6%) using NAAT, 2.7% (95% CI 1.4% to 4.3%) using culture, 3.8% (95% CI 0.0% to 24.4%) using other assays assessing current infection and 8.7% (95% CI 0.0% to 31.3%) using Gram stain (table 1).
Mean current infection prevalence was 2.5% in women (95% CI 1.2% to 4.1%) vs 1.5% in men (95% CI 0.5% to 3.0%), 3.0% in studies before 2005 (95% CI 1.5% to 4.8%) vs 1.1% in those after 2005 (95% CI 0.4% to 2.0%), 4.1% in samples including <100 participants (95% CI 1.8% to 6.9%) vs 1.0% in those including ≥100 participants (95% CI 0.3% to 1.9%) and 16.2% in symptomatic individuals (95% CI 7.1% to 27.7%) vs 1.0% in asymptomatic ones (95% CI 0.3% to 2.0%) (table 1).
Mean ever infection prevalence showed similar results, although at much higher prevalence levels (table 1).
There was evidence for heterogeneity in prevalence across studies. Most meta-analyses showed a p value of <0.001 for Cochran’s Q statistic, wide prediction intervals indicating high heterogeneity and I2≥70%, affirming most variability as due to true differences in prevalence across studies rather than chance (table 1).
Associations with prevalence and sources of between-study heterogeneity
Univariable metaregression results are in table 2. There was ‘strong’ evidence for an association with prevalence (p value of ≤0.05) for WHO region, sex, infertility diagnosis, presence of urogenital signs and symptoms, assay type, year of data collection, sample size and exposure to antimicrobials prior to sample collection; ‘good’ evidence for infertility type (0.05<p value≤0.10), but no evidence for validity of infertility definition, consistency in assay used for infection ascertainment and response rate.
Compared with AMRO, AFRO showed four-fold higher odds of gonorrhoea infection (OR=4.0, 95% CI 1.5 to 10.1), while no significant differences were found for the other regions. Women had twice higher odds of infection than men (OR=2.0, 95% CI 1.1 to 3.7). Individuals with secondary infertility also had twice higher odds of infection (OR=2.1, 95% CI 0.9 to 5.2) than those with primary infertility. Odds were 2.4-fold (95% CI 1.2 to 4.6) and 2.0-fold (95% CI 0.8 to 5.0) higher for women with TFI and for individuals with mixed cause and unexplained infertility, respectively, compared with those with general/unspecified infertility. Symptomatic individuals had six-fold higher odds of infection compared with asymptomatic ones (OR=5.9, 95% CI 2.6 to 13.5).
Culture and other assays detecting current infection showed 2.5-fold (95% CI 1.3 to 4.8) and 4.1-fold (95% CI 1.1 to 15.3) higher odds, respectively, compared with NAAT, while assays detecting IgG showed 22.1-fold higher odds (95% CI 9.5 to 51.2). There was evidence for declining prevalence at ~3% per year over the last four decades (OR=0.97, 95% CI 0.95 to 0.99) and for small-study effect, with studies including ≥100 participants showing lower prevalence (OR=0.5, 95% CI 0.3 to 0.9).
Sensitivity analysis using only studies assessing current gonorrhoea infection affirmed the aforementioned results, although some associations failed to reach statistical significance because of the smaller number of studies (online supplementary table 11).
Full multivariable metaregression analysis could not be performed due to lack of statistical power.62 However, backward variable selection yielded a final multivariable model including four predictors: region, presence of urogenital signs and symptoms, sample size and assay type (p value of ≤0.1, online supplementary table 12). The analysis confirmed the univariable metaregression results.
We provided, to our knowledge, the first systematic review of gonorrhoea infection in infertile populations. Current infection prevalence was several folds higher than that in the general population; the global estimate in infertile populations was 2.2%, compared with only 0.8% in the general population (per WHO 2016 estimates).11 Regional estimates followed a similar pattern. Current infection prevalence rates in infertile versus the general population were 5.0% vs 1.8%11 in AFRO, 2.7% vs 0.7%11 in EMRO, 2.5% vs 0.8%11 in WPRO, and 2.4% vs 0.3%11 in EURO, respectively. These findings should be seen against the expectation that infertile populations should be prone to a lower prevalence than the general population; there is higher frequency of STI testing among them, and therefore earlier detection and higher treatment coverage relative to the general population. Infertile populations may also undergo prophylactic antibiotic administration, not necessarily with testing, prior to procedures such as in vitro fertilisation/embryo transfer.63 64
Higher prevalence was also associated with conditions conventionally considered as sequelae of gonorrhoea infection,12 such as TFI and secondary infertility. TFI was associated with two-fold higher odds of gonorrhoea infection. The biological plausibility behind this association has been repeatedly described, with evidence showing that untreated gonococcal infection can lead to pathogen ascension to the upper genital tract, causing pelvic inflammatory disease, tubal scarring, oviduct occlusion and internal tissue adhesion.12 18 65 Higher prevalence was also found in individuals with mixed and unexplained infertility diagnoses. However, samples comprising mixed infertility diagnoses often included individuals with TFI, while more studies are needed to elucidate the association with unexplained infertility. Prevalence was further higher in individuals with secondary infertility, possibly because secondary infertility is more likely to be caused by ‘preventable/acquired factors’, such as recurrent exposure to STIs, as opposed to primary infertility, which is more likely to be caused by non-preventable genetic/congenital abnormalities.3 66 67
These findings attest to the potential role of gonorrhoea, and/or possibly other STIs associated with gonorrhoea, such as chlamydia, in infertility. Since early detection and treatment of gonococcal infections have been challenged by infection’s asymptomatic nature,68 69 and growing antimicrobial resistance,22–26 these findings support the global public health value of developing gonococcal vaccines29 30 as a fundamental solution to gonorrhoea’s adverse implications.70 These findings also support the timeliness of a comprehensive prevention approach promoting sexual health to control N. gonorrhoeae and other STIs, mitigate antimicrobial resistance and achieve WHO global health sector strategy targets.21 Such an approach would focus on the simultaneous implementation of biomedical (rolling-out testing and vaccination), behavioural (promoting healthier sexual lives) and structural prevention interventions (improving access to testing, treatment and care services). Indeed, successful and sustainable implementation of biomedical interventions cannot be achieved without adequate levels of public awareness, access to/uptake of services, and adherence/retention in prevention and treatment cascades.
Interestingly, there was evidence of declining prevalence by ~3% per year over the last four decades, possibly mirroring declines in prevalence in the population at large,71 72 or growing STI testing and treatment coverage, and use of improved diagnostics in infertility workup.73 There was also evidence of regional variability in prevalence, with AFRO being most affected. This may reflect variability in background prevalence: AFRO has the highest gonorrhoea prevalence in the general population.11
The higher infection levels in infertile women compared with men, possibly reflect larger contribution of gonorrhoea to infertility in women (online supplementary figure 14),74 higher susceptibility to gonorrhoea acquisition in women75 or persistence for longer durations, as this infection is largely asymptomatic in women.68 69 As signs and symptoms are indicative of infection sequelae,76 gonorrhoea prevalence was higher in symptomatic compared with asymptomatic individuals.
There were differences in prevalence by assay type, a result difficult to interpret given differences in sensitivity and specificity,73 and recent and differential use of NAAT in resource-rich versus resource-limited settings.73 Of note, the variation in the use of assays across settings and time is not likely to differentially affect one population, such as infertile populations, as opposed to another, such as the general population. While ever infection prevalence was much higher than current infection prevalence, this finding has probably limited epidemiological relevance, given cross-reaction with other pathogens, such as Staphylococcus aureus.73 77 78
Our study has unavoidable limitations. Data quantity and quality varied across regions and sometimes limited our ability to produce representative summary estimates; there were only six studies assessing current infection in SEARO, all from India. It was not possible to conduct full multivariable metaregression to adjust for potential confounders, with the large number of predictors relative to that of studies. Prevalence estimates by infertility diagnosis may have been affected by unavoidable overlap across categories; samples with mixed infertility often included TFI, and those with non-TFI may have included other infertility diagnoses. An analysis by age could not be performed, given the low number of studies reporting patients’ age. There was evidence for small-study effect in metaregression (table 2 and online supplementary table 12), suggesting publication bias; studies with small sample size reported higher prevalence. Conversely, differential access to quality STI testing and treatment in infertility clinics, in settings with better versus limited access to STI services, may have biassed such studies towards lower prevalence.10 79–81 Risk of bias assessment was limited by studies with missing information. Gonorrhoea prevalence was often reported as a secondary outcome, with no ‘gonorrhoea’ term listed in title/abstract, thereby complicating study identification.
In conclusion, gonorrhoea prevalence in infertile populations is several folds higher than that in the general population. This finding, along with even higher prevalence in women with TFI and individuals with secondary infertility, attests to the potential role of N. gonorrhoeae in infertility and suggests that a fraction of infertility is possibly preventable by controlling N. gonorrhoeae transmission. Expansion of N. gonorrhoeae surveillance and monitoring in infertile populations is warranted as gaps in evidence persist. A multifaceted response should be considered to ensure progress towards WHO global health sector strategy targets.21
Current gonorrhoea infection prevalence in infertile populations varied across regions but was several folds higher than that for the general population across world regions.
Twice higher odds of gonorrhoea infection were found in women with tubal factor infertility and secondary infertility.
A fraction of observed infertility is possibly preventable by controlling Neisseria gonorrhoeae transmission.
The authors thank Ms Adona Canlas for her assistance with locating full-text articles and Professor Nico Nagelkerke for statistical analysis support.
Contributors HC contributed to the study design, conducted the systematic searches of the literature, selected studies for inclusion, performed data extraction and data analyses, and wrote the first draft of the paper. AM contributed to title and abstract screening, full-text screening and data extraction. FAH double extracted the data. IT and LJA conceived and led the design of the study. LJA led the data extraction, analyses and drafting of the article. KB, JK, and TCM contributed to the study design. All authors contributed to the discussion and interpretation of the results and to the writing of the article, and have read and approved the final manuscript.
Funding This work was supported by funding from the WHO. This work was also funded by the UNDP-UNFPA-UNICEF-WHO-World Bank Special Programme of Research, Development and Research Training in Human Reproduction, a cosponsored programme executed by the WHO. JK and IT are staff members of the WHO. The authors alone are responsible for the views expressed in this publication, and they do not necessarily represent the views, decisions or policies of the WHO.
Disclaimer The authors alone are responsible for the views expressed in this publication and they do not necessarily represent the views, decisions or policies of the WHO.
Competing interests The authors have no conflicts of interests to declare.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; internally peer reviewed.
Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. All data are available as part of the manuscript and its supplementary material.