Objectives Over 2.4 million people have been displaced within the Thailand-Myanmar border region since 1988. The efficacy of community-driven health models within displaced populations is largely unstudied. Here, we examined the relationship between maternal healthcare access and delivery outcomes to evaluate the impact of community-provided health services for marginalised populations.
Setting Study setting was the Thailand-Myanmar border region’s single largest provider of reproductive health services to displaced mothers.
Participants All women who had a delivery (n=34 240) between 2008 and 2019 at the study clinic were included in the performed retrospective analyses.
Primary and secondary outcome measures Low birth weight was measured as the study outcome to understand the relationship between antenatal care access, family planning service utilisation, demographics and healthy deliveries.
Results First trimester (OR=0.86; 95% CI=0.81 to 0.91) and second trimester (OR=0.86; 95% CI=0.83 to 0.90) antenatal care visits emerged as independent protective factors against low birthweight delivery, as did prior utilisation of family planning services (OR=0.82; 95% CI=0.73 to 0.92). Additionally, advanced maternal age (OR=1.36; 95% CI=1.21 to 1.52) and teenage pregnancy (OR=1.27, 95% CI=1.13 to 1.42) were notable risk factors, while maternal gravidity (OR=0.914; 95% CI=0.89 to 0.94) displayed a protective effect against low birth weight.
Conclusion Access to community-delivered maternal health services is strongly associated with positive delivery outcomes among displaced mothers. This study calls for further inquiry into how to best engage migrant and refugee populations in their own reproductive healthcare, in order to develop resilient models of care for a growing displaced population globally.
- maternal medicine
- reproductive medicine
- preventive medicine
Data availability statement
No data are available. No additional data available.
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.
Strengths and limitations of this study
The impact of community-delivered reproductive health services on maternal health outcomes is largely unstudied in populations impacted by civil unrest, such as the 2.4 million people who have been displaced within the Thailand-Myanmar border region since 1988.
This is the first study to directly examine the impact of community-based reproductive health services on delivery outcomes within a displaced and refugee population, demonstrating a clear relationship between community-based care and healthy deliveries.
These results corroborate the clear association between antenatal care access, family planning services access and infant and maternal health outcomes established by population modelling studies globally.
This study invites further exploration of how to best engage displaced and refugee populations in their own reproductive healthcare, which has the potential to improve care quality and access for those unsupported by formal healthcare systems globally.
Limitations of this study include the inability to quantify the impact of geography on maternal health outcomes in this highly mobile population, and ongoing lack of consensus on the value of birth weight as a prognostic indicator of health and point of intervention.
Refugee and displaced mothers experiencing pregnancy are among the world’s most vulnerable populations. Inadequate shelter, increased prevalence of gender-based violence and limited access to education and employment are among the challenges faced by displaced mothers during critical childbearing months, putting them distinctly at risk compared with refugee men and the non-displaced at large.1 2 The issue of refugee maternal health is underscored by the rising population of displaced persons worldwide. The United Nations High Commissioner for Refugees (UNHCR) estimates that over 80 million people globally were forcibly displaced as of 2020, up from <20 million in 2010.3 Among them is a growing population of women of reproductive age, whose experiences of pregnancy will have implications for their own health and the health of future generations numbering in the millions.4
The pressing need for reproductive healthcare among displaced populations is clearly established5 6; however, challenges to implementing effective systems of care delivery persist.7 8 There is an emerging consensus on the value of community-delivered maternal healthcare in low-resource healthcare settings.9–12 The role of midwifery has also seen renewed interest, with a number of studies establishing the potential impact to improving access and quality of maternal healthcare.13 However, few examples support the feasibility of such community-centred care models within displaced populations,14 and none to our knowledge directly evaluate the impact on maternal health outcomes.
Directly studying maternal healthcare delivery within actively displaced populations is challenging. The transient and mobile nature of migrant groups fundamentally limits opportunities for investigation to rare cases of sizeable and persistently displaced populations. One such setting is the Thailand-Myanmar border region, home to one of the longest-standing refugee and migrant populations globally. Over 2 million people have lived displaced between eastern Myanmar and northwestern Thailand for the last 30 years. A network of community-based ethnic health organisations serves the large population with significant barriers to accessing health services through formal healthcare systems.15 This ethnic health system presents the opportunity to examine the efficacy of a community-established system of reproductive healthcare that has continuously served its displaced population through ongoing political unrest for nearly a generation.
There are many obstacles to providing effective care to refugee populations. Programmatic challenges to designing effective systems of care delivery in displaced populations are widely documented,7 8 and range from limitations of international funding to operationalisation concerns. Non-governmental and community-based organisations, especially those disconnected from formal healthcare systems, must often rely on external grants and inconsistent funding streams to accomplish critical humanitarian work. Even with adequate resources available, displaced and refugee mothers face numerous barriers to accessing their care, ranging from volatile security situations and statelessness to cultural obstacles and transportation issues (table 1).
This study adds to other published studies in suggesting that direct support of community-coordinated reproductive health programmes represent a potential avenue for addressing these concerns. As the international community makes efforts to support those threatened by reinvigorated political instability in the region, the evidence presented here suggests that community-based organisations will be critical for sustaining essential reproductive health services to those already marginalised by government-based systems.
In this study, we examined 12 years of delivery outcomes and reproductive healthcare service utilisation among a cohort of refugee and internally displaced mothers who delivered at the Mae Tao Clinic (MTC). MTC is a community-established clinic in Mae Sot, Thailand that serves as the region’s single largest provider of maternal health services for the displaced. This study focused on the outcome of low birth weight, a critical prognostic indicator of infant health,16 to better understanding the relationship between antenatal care access, family planning service utilisation, demographics and healthy deliveries among this underserved population. Through assessing trends in birth weight outcomes in this context, we sought to understand the impact of community-delivered care in this marginalised population of mothers in the Thailand-Myanmar border region.
MTC is located in Mae Sot, a city in Thailand’s northwestern Tak Province that shares its western border with Karen State, Myanmar. The border region has been the site of conflict between the Myanmar military and ethnic minority groups since 1988.17 18 Ongoing violence in eastern Myanmar has displaced 2.4 million migrants into Thailand during that time, with reasons for migration ranging from violence and persecution to economic motivations.19 20 Approximately 110 000 reside as officially registered refugees in camps along the Thailand-Myanmar border.21 The remaining majority are categorised as ‘irregular migrants’, lacking official political status in Thailand.21 Although Myanmar’s 2015 democratic elections suggested a transition towards peace in the region, the 1 February 2021 military coup d’état was a reminder of the ongoing instability in the region.
Since 1989, a system of Ethnic Health Organizations (EHOs) has developed to provide care to the largely unserved group of migrants, refugees and displaced persons in the region. MTC has played a central role in coordinating the health system development, while simultaneously providing free health services across an estimated catchment area of over 1 million (figure 1). Founded by Dr Cynthia Maung, a refugee herself, the clinic is rooted entirely in the migrant community, operating independently from government health systems. MTC’s health and administrative teams are staffed by local community health workers (CHWs) and managers, and supported largely by international grants (including United States Agency for International Development (US AID), United Kingdom Aid (UK AID) and the International Rescue Committee). Having provided over 135 000 antenatal care consultations, over 75 000 family planning consultations and supported over 34 000 deliveries since 2007, MTC is the region’s single largest provider of reproductive health services for migrants, refugees and the displaced in the region.
The overall landscape of healthcare options for the region’s displaced population is diverse. Health centres exist within the border region’s numerous refugee camps and are supported by international organisation such as UNHCR. Independent institutionally affiliated organisations, such as Shoklo Malaria Research Unit, also provide maternal healthcare services alongside a variety of missionary and religious groups in the region. Finally, numerous other ethnic health organisations, such as The Karen Department of Health and Welfare, Backpack Health Workers and The Burma Medical Association, work closely with MTC to support the needs of the region’s most underserved. In light of this complex system, care access patterns are difficult to characterise. While mothers with risk factors or complicated antenatal courses may be more likely to present to MTC for health services, these motivations are difficult to untangle from cultural, political and financial reasons for using the clinic over alternative sources of care.
Prior research within the region’s EHO network have helped to characterise the successes and ongoing challenges of task-shifting maternal health services for displaced mothers to a community-based approach.15 In 2008, Mullaney et al completed a comprehensive survey of maternal health access in the border region, highlighting the significant impact that the overarching political, developmental and human rights context has had on reproductive and maternal health among displaced populations.22 23 More recent works have used focus group interviews with migrant mothers and lay maternal CHWs to illuminate ongoing barriers to care, and understand local perceptions of reproductive health.24–27 To date, there has been no investigation of the relationship between access to community-delivered health services and individual delivery outcomes among the mothers in the region, leaving the direct impact of these community care models on obstetric outcomes largely unquantified at the population level.
Despite growing consensus around the value of task-shifting to improve maternal health in low-income and middle-income country,28 29 little work has demonstrated the feasibility and impact of community care models within refugee and migrant populations, usually among the most vulnerable and underserved in any low-resource setting. In addition to establishing the impact of community-based reproductive healthcare, the results of this study call for further exploration of the ongoing barriers to care facing displaced mothers in the region. Notably, 33% of mothers in the cohort met the current WHO recommendation of four antenatal care (ANC) visits during pregnancy. This figure lags behind national averages for Myanmar (59%) and Thailand (91%),30 underscoring the underserved nature of this population. Since March 2020, periodic closures of the Thailand-Myanmar border, due to the COVID-19 pandemic, have presented yet another barrier to care. The critical need to improve reproductive care access among the region’s displaced population is further heightened by the Myanmar military’s seizure of government power on 1 February 2021. Although the ramifications of this political development are still unfolding, it has already shown signs of exacerbating the issues of statelessness, political autonomy and discrimination that reinforce inequalities in care access and health outcomes in the region.
This study was a retrospective review of all deliveries that occurred at MTC between June 2007 and December 2019 (n=34 240). Data were extracted from MTC’s Health Information System (HIS; details given in online supplemental material) located in Mae Sot, Thailand. Data aggregation and processing were performed in Python (V.3.8.3) using the Pandas (V.1.0.1) analytical package. Visualisation and logistic regression were performed in the R Studio software environment (V.3.6.2). Cartography was conducted using ArcGIS Pro (V.2.6).
Patient and public involvement
This study did not involve direct interaction with patients. All data were de-identified prior to their analysis. The Community Ethics Advisory Board reviewed the proposed data contained herein, and allowed their analysis and publication.
Independent and dependent variables
Low birthweight delivery was the primary outcome in this study. Low birth weight was defined as any weight <2.5 kg. Reproductive health service utilisation was quantified by number of ANC visits attended during each trimester of pregnancy, as well as each mother’s history of accessing family planning services at MTC at any point prior to conception. Two demographic variables were included as covariates: (1) ethnicity and (2) maternal age. Ethnicity was aggregated into the three predominant ethnicities among MTC’s patient population (Burmese, Karen and Mon) and one additional category to encompass the remaining heterogenous mix of other ethnic identities. Maternal age was stratified into three groups to capture high-risk cohorts defined by teenage pregnancy and advanced maternal age: (1) <20 years old; (2) 20–35 years old and (3) <35 years old. Two variables related to pregnancy history were included as potential maternal health risk factors: (1) maternal gravidity and (2) any history of prior miscarriage. Cases for which any variable was missing were excluded, leaving a total of 30 209 deliveries for final analysis. Validation was performed to confirm that the excluded mothers were similar to the final study cohort in terms of outcomes, demographics, care utilisation and risk factors.
Bivariate analyses and logistic regression
Deliveries were segregated based on birthweight outcome (ie, low birth weight or normal birth weight) and bivariate analysis was performed across all variables. The χ2 calculations were conducted for categorical variables (ethnicity, age, history of miscarriage and family planning utilisation). Following bivariate analysis, a multiple logistic regression was performed to assess for independent predictors of low birth weight across all independent variables. An alpha of 0.05 was set for determination of statistical significance in all analyses.
Information on geographic district of origin was available for all mothers included in the final study cohort, providing the opportunity to observe trends in risk factors, care utilisation and outcomes across MTC’s geographically wide catchment area. ArcGIS was used to produce a cartographic representation of low birthweight deliveries based on mothers’ district of origin (figure 2). Study variables were then aggregated by district, and z-scores for each district-variable pair were computed. These z-scores, which reflect geographic variation within the dataset, were then visualised in a heatmap to assess for trends in care utilisation, demographics, risk factors and birthweight outcomes (figure 3).
Temporal analysis was also performed to assess for changes in demographics, risk factors and outcomes in the cohort during the study period. In particular, political developments in 2012 loosened restrictions on cross-border travel and improved access to care in eastern Myanmar. Disparities between the pre-2012 and post-2012 periods lend insight into the health impacts of regional politics, and allow us to evaluate trends in outcomes that co-occurred with community-led efforts to strengthen the region’s ethnic health system. All independent variables, as well as prevalence of low birthweight delivery, were aggregated by year to assess for any overarching temporal trends. The Mann-Kendell test (a non-parametric statistical test for monotonic temporal trend) was performed on each variable’s yearly time series to determine existence and statistical significance of any such trends (alpha=0.05). Additionally, deliveries were segregated based on admission date before and after the start of 2012, to assess for any broad-level changes following the political developments that year which expanded cross-border health service access for migrants. A bivariate analysis of all variables pre-2012 and post-2012 was conducted, following the same methodology noted above.
Yearly low birthweight prevalence displayed a clear downward trend over the study period, with a high of 15.8% in 2008 and low of 9.4% in 2018 (figure 4). Mann-Kendall analysis confirmed this trend in low birthweight delivery (tau=−0.727, alpha=0.001), and also revealed trends towards increased utilisation of first trimester ANC visits (tau=0.795, p<0.001), second trimester ANC visits (tau=0.487, p=0.024), and family planning services (tau=0.939, p<0.001). Rates of prior miscarriage fell during the study period (tau=−0.606, p<0.001) while the proportion of Karen (tau=0.909, p<0.001) and Burmese (tau=0.455, p=0.047) mothers as well as those over the age of 35 (tau=0.545, p=0.016) increased. No significant changes in age were noted during the study period. Bivariate analysis of deliveries pre-2012 and post-2012 corroborated these trends, revealing numerous significant results across our independent variables (online supplemental table S1). Notably, the average number of first trimester ANC visits and second trimester ANC visits were both higher in the post-2012 period, while the proportion of mothers accessing family planning services more than doubled (18% vs 8.8%). Further discussion and context around temporal trends in demographics, care utilisation and outcomes can be found in the online supplemental material.
Bivariate analysis of deliveries segregated by birthweight outcome revealed numerous statistically significant differences between the two cohorts (table 2). Considering reproductive healthcare utilisation, low birthweight deliveries were associated with fewer first trimester ANC visits and second trimester ANC visits compared with normal birthweight deliveries. Mothers who delivered low birthweight infants also used family planning services at a lower rate (11% vs 14%, p<0.001). From the demographic variables, maternal age category was significantly associated with birthweight outcome (p<0.001), with mothers who delivered low birthweight infants more likely to be under the age of 20 (11% vs 8.1%) or over the age of 35 (15% vs 14%) years. Considering pregnancy history, low birthweight deliveries were associated with mothers with lower gravidity, and also concomitant with a lower incidence of prior miscarriage (16% vs 18%, p=0.003). Ethnicity showed no significant correlation with delivery outcome. In separate bivariate analysis assessing for satisfaction of WHO ANC guidelines, mothers who delivered low birthweight infants were less likely to have met the prior recommendation of four total ANC visits during pregnancy (26% vs 34%, p<0.001) than those who were able to meet the guideline.
Logistical regression confirmed the relationships established by bivariate analysis (table 3). The number of first trimester ANC visits (OR=0.86; 95% CI=0.81 to 0.91) and number of second trimester ANC visits (OR=0.86; 95% CI=0.83 to 0.90) emerged as independent protective factors against low birthweight delivery across the cohort, as was prior utilisation of family planning services (OR=0.82; 95% CI=0.73 to 0.92). The number of third trimester ANC visits did not show a significant correlation with outcomes. From among demographic factors, maternal age also showed a statistically significant correlation with birthweight outcome, with mothers under the age of 20 years (OR=1.27, 95% CI=1.13 to 1.42) and those over the age of 35 years (OR=1.36; 95% CI=1.21 to 1.52) emerging as independent risk factors. In terms of pregnancy history, maternal gravidity displayed a protective effect against low birthweight delivery (OR=0.92; 95% CI=0.89 to 0.94).
Our cartographic representation of low birthweight deliveries by mothers’ district of origin yielded no obvious trend in outcomes based on geographic distance from MTC (figure 2). However, heatmap visualisation revealed significant variation in the cohort based on district of origin (figure 3). Mothers who self-reported Mae Sot as their district of origin displayed the highest utilisation of MTC’s reproductive health services, while remote districts such as Kyainseikgyi displayed some of the lowest utilisation figures. Among variables tracking demographics and pregnancy history, a number of districts stood out as outliers. For example, mothers from Hlaingbwe displayed meaningfully higher maternal gravidity and history of prior miscarriage compared with those from other districts. Examining age, Mae Sot’s nearby Thai district of Mae Ramat produced the highest proportion of teenage mothers, exceeded only by Umphang. The proportion of mothers over the age of 35 years was far more evenly distributed across all districts.
To our knowledge, this is the first study to directly examine the impact of community-based reproductive health services on delivery outcomes within a displaced and refugee population, demonstrating a clear relationship between community-based care and healthy birthweight delivery. Our results indicate that access to CHW-delivered antenatal care services at MTC during the first two trimesters of pregnancy is independently correlated with healthy birthweight delivery across the cohort. Utilisation of family planning services prior to conception was also found to be independently associated with positive outcomes, with both associations remaining significant even when adjusted for ethnicity, age, gravidity and history of prior miscarriage. Encompassing over 30 000 deliveries spanning 12 years of care, these results represent strong evidence in support of the positive impact of community-based reproductive healthcare on maternal health outcomes in this underserved cohort of mothers.
Our findings corroborate existing consensus in the maternal health literature. Associations between ANC service utilisation and improved infant and maternal health outcomes have been extensively validated globally.31 Population modelling studies assessing the impact of family planning services have also consistently revealed an association between such reproductive services and positive maternal health outcomes.32 Although questions of standardisation of healthcare services across disparate settings remain an issue, MTC’s care protocols follow widely accepted guidelines for ANC services and family planning recommendations established by WHO.33 More information on MTC’s services can be found in the online supplemental material.
Our study also quantifies the significant diversity in outcomes, care utilisation, demographics and pregnancy history based on mothers’ district of origin. Geography is critical for contextualising the complexities of care delivery in this widely dispersed cohort. A patient’s proximity to care, as well as relative mobility and permanence of residence, is an unavoidable confounding factor when examining trends among any mobile population. Beyond contextualising the dataset used in this study, the geographic diversity also represents a tractable starting point for exploring interventions to address disproportionate pregnancy risk faced by certain subcohorts. For example, the heatmap (figure 4) revealed that mothers arriving from the Thai district of Umphang accounted for a disproportionate number of teenage pregnancies during the study period. Teenage pregnancy is independently associated with greater risk of low birthweight delivery in our final model (table 3), and furthermore mothers from Umphang also displayed lower utilisation of first and second trimester visits, both indicated to be protective factors across the population. Umphang mothers did, however, access family planning resources prior to conception more often than their peers. These findings suggest the possibility of providing targeted counselling around pregnancy risk factors during family planning visits, and calls for further investigation into the barriers to seeking care earlier in pregnancy among mothers from this district.
This study has several limitations. First, the reproductive care utilisation data presented here only reflects care provided by MTC, and as such does not reflect ANC, family planning services and other forms of reproductive care accessed outside of the clinic. To account for this, bivariate analysis and logistical regression modelling were repeated on the subset of mothers reporting no external visits (n=16 769, approximately 56% of the total cohort) and results were highly comparable within this subgroup (online supplemental table S2 and S3). The geographic analysis presented here also comes with a caveat. Given the highly mobile nature of MTC’s patient population, a mother’s self-reported district of origin at the time of delivery may not necessarily reflect her place of residence throughout a pregnancy. It is also common for mothers travelling from afar to temporarily reside close to MTC to ensure access to care. Such tendencies make it difficult to assess the true value of patient-reported geographic information, and more broadly highlight the complexities in understanding how changes in location, access to transportation and presence of local social supports confound trends in outcomes within migrant populations. Finally, the use of birth weight as a prognostic indicator of health and point of intervention is still actively debated, with recent meta-analyses suggesting that the applications of birth weight alone may be more limited than previously thought.34 However, in settings of episodic care such as this, more actionable indicators such as prematurity are sometimes difficult to assess. Birth weight thus remains a reliably measured and widely studied outcome to compare marginalised populations with those globally.35
External factors not accounted for in analysis may also act as confounders in the relationship between healthcare provision and positive delivery outcomes in this setting. For example, increased access to contraceptives in the region has led to fewer unintended pregnancies over time. Alongside decreases in miscarriages and abortions, this is thought to be the single largest factor behind improving maternal outcomes in the region. Mothers who receive care at MTC may well represent a self-selected cohort of those more willing to engage with maternal healthcare and therefore less likely to suffer the consequent risks associated with unintended pregnancy. Furthermore, malaria eradication efforts have reduced infection rates among pregnant migrants, alongside improvements in access to nutrition supplementation including iodine, folate and iron. Both represent other potential confounders of trends in maternal care outcomes in the region over the 12-year study period.
Based on our findings, we propose further study in the Thailand-Myanmar border region to address critical gaps in knowledge regarding the access and utilisation of maternal healthcare. As COVID-19 establishes itself as an ongoing public health concern in the border region, more work must be done to understand the impact of epidemics on vulnerable populations. Emerging evidence from MTC’s HIS points to large declines in migrant patient volumes due to recent cross-border quarantine mandates (figure 5), as well as concomitant staff shortages as community-based groups were forced to dismiss workers to avoid outbreaks. Unfortunately, insufficient data exist to date to properly analyse prepandemic and intrapandemic trends, as the region continues to be greatly affected. However, prior work in other resource-limited regions provide a template for further investigation. For example, observational studies in Sierra Leone have elucidated the impact of the Ebola outbreak on rural reproductive health services.36 Other studies have examined issues of infection after delivery and availability of HIV services in the context of pregnancy, two other pivotal areas for future exploration in the Thailand-Myanmar border region.37 38
Based on these findings we propose that, as international aid enters to region in the wake of increasing political unrest and ongoing COVID-19 concerns, resources be targeted towards ethnic health organisations and community-based groups who provide essential services to displaced individuals unsupported by government health systems. More generally, our findings suggest that community-based reproductive health systems may function as effective care delivery models in long-standing refugee and migrant populations. Furthermore, specific attention must be given to developing effective health information systems in low-resource settings, which are essential to supporting future population health investigation like this study. Such health data systems will be essential as we seek to understand and address the needs of growing displaced populations worldwide.
Data availability statement
No data are available. No additional data available.
Patient consent for publication
Approval for this study was provided by the Community Ethics Advisory Board hosted at Mae Tao Clinic on 30 January 2020. This study does not involve human participants.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Contributors SP, CM, SH, HMM, STL, CB, TS, MB, CA and INS designed the study. SP collected the data with assistance from CM, SH, HMM, STL, CB, TS, MB and CA. Data curation was done by SP. SP led the writing, with CM, SH, HMM, STL, CB, TS, MB, CA and INS contributing to the review and editing. INS provided supervision to SP and had the final responsibility for the decision to submit for publication. INS is the gaurantor of this study, and accepts full responsibility for the work or the conduct of the study, had access to the data, and controlled the decision to publish.
Funding This work was funded by the Brown University Data Science Initiative (no grant identifier), the Biomedical Informatics Global Health Fund at the Brown Centre for Biomedical Informatics (no grant identifier) and National Institutes of Health (grants U54GM11577 and R25MH11440).
Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.