Article Text
PDF
Abstract
Objectives To identify the risk factors for neonatal sepsis in Sub-Saharan Africa.
Design Systematic review and meta-analysis.
Data sources PubMed, Embase, Web of Science, African Index Medicus and ClinicalTrials.gov were searched for observational studies from January 2010 to August 2020.
Setting Sub-Saharan Africa, at all levels of healthcare facilities.
Participants ‘Neonates’ (<28 days of age) at risk of developing either clinical and/or laboratory-dependent diagnosis of sepsis.
Outcome measures Identification of any risk factors for neonatal sepsis.
Results A total of 36 studies with 23 605 patients from secondary or tertiary level of care facilities in 10 countries were included. Six studies were rated as good quality, 8 as fair and 22 as poor. Four studies were omitted in the meta-analysis due to insufficient data. The significant risk factors were resuscitation (OR 2.70, 95% CI 1.36 to 5.35), low birth weight <1.5 kg (OR 3.37, 95% CI 1.59 to 7.13) and 1.5–2.5 kg (OR 1.36, 95% CI 1.01 to 1.83), low Apgar score at the first minute (OR 3.69, 95% CI 2.34 to 5.81) and fifth minute (OR 2.55, 95% CI 1.46 to 4.45), prematurity <37 weeks (OR 1.91, 95% CI 1.27 to 2.86), no crying at birth (OR 3.49, 95% CI 1.42 to 8.55), male sex (OR 1.30, 95% CI 1.01 to 1.67), prolonged labour (OR 1.57, 95% CI 1.08 to 2.27), premature rupture of membranes (OR 2.15, 95% CI 1.34 to 3.47), multiple digital vaginal examinations (OR 2.22, 95% CI 1.27 to 3.89), meconium-stained amniotic fluid (OR 2.72, 95% CI 1.58 to 4.69), intrapartum maternal fever (OR 2.28, 95% CI 1.18 to 4.39), foul-smelling vaginal discharge (OR 3.31, 95% CI 2.16 to 5.09) and low socioeconomic status (OR 1.93, 95% CI 1.11 to 3.35). We found considerable heterogeneity in the meta-analysis of 11 out of 15 identified risk factors.
Conclusion Multiple risk factors for neonatal sepsis in Sub-Saharan Africa were identified. We revealed risk factors not listed by the WHO guidelines. The included studies overall had high risk of bias and high heterogeneity and thus, additional research of high quality is needed.
PROSPERO registration number CRD42020191067.
- OBSTETRICS
- PAEDIATRICS
- Paediatric infectious disease & immunisation
- NEONATOLOGY
Data availability statement
No data are available. The only original data published were the protocol of the study, which can be found on the PROSPERO web page (ID: CRD42020191067). The remaining data are published in the articles included in this systematic review and meta-analysis.
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
STRENGTHS AND LIMITATIONS OF THIS STUDY
This systematic review and meta-analysis has a high number of included studies (36) as well as a large sample size (23 605 neonates).
This systematic review has a broad search strategy, with a meta-analysis performed on 33 risk factors.
Heterogeneity in the study design of the included studies is a limitation.
The overall high risk of bias in the included studies is a limitation.
Introduction
The Millennium Development Goals from 1990 identified newborn health as a key priority for global development.1 The global neonatal mortality rate has decreased by 37%, from 33 to 21 deaths per 1000 live births since then.2 In 2016, the Sustainable Development Goals (SDGs) were announced.3 SDG goal 3 aims to ensure healthy lives and promote well-being for all at all ages, and includes subtarget 3.2: by 2030, to end preventable deaths of newborns and children under 5 years of age, with all countries aiming to reduce neonatal mortality to at least 12 per 1000 live births and under-5 mortality to at least as low as 25 per 1000 live births.3 However, today a child born in Sub-Saharan Africa is still 10 times more likely to die in the first month compared with a child born in a high-income country.4 In 2018, 2.5 million children died within the first 28 days of life globally.4 In the same year, countries in Sub-Saharan Africa had the highest mortality, with 28 neonatal deaths per 1000 live births.2 4
The majority of the 2.5 million neonatal deaths in 2018 worldwide can be divided into three main causes, each contributing approximately one-third to neonatal deaths: infections, intrapartum asphyxia and preterm birth complications.2 5 However, the causes of neonatal death vary among countries and regions.5 In countries with high neonatal mortality, almost 50% of deaths are due to severe infection with sepsis, making sepsis a leading cause of admissions and deaths in neonatal units in low-income and middle-income countries (LMICs).5 6 The Sub-Saharan African region includes some of the highest rates of neonatal mortality due to neonatal sepsis, yet prevention strategies are and remain unsatisfactory.7 Improved understanding of the underlying causes of neonatal sepsis is necessary to optimise prevention and management guidelines. Evidence from reviews of risk factors has been used globally to guide the development of management guidelines and prevention strategies for neonatal sepsis.8 The WHO recommends prophylactic antibiotics to newborns within 48 hours after delivery if membranes ruptured >18 hours before delivery, the mother had fever >38°C before delivery or during labour, or the amniotic fluid was foul-smelling or purulent.9 However, there might be discrepancies in the risk factors in different parts of the world. In a paper from 2020 on neonatal mortality, the authors conclude that there is a need to develop clinical guidelines for prevention and management of neonatal sepsis that are specific to the Sub-Saharan African context.10
Multiple studies aiming to identify the risk factors for neonatal sepsis have been performed in Sub-Saharan Africa during the last 10 years. With this systematic review and meta-analysis, we aim to provide quality evidence to identify the risk factors for neonatal sepsis in Sub-Saharan Africa. To the best of our knowledge, this is the first systematic review and meta-analysis to address neonatal risk factors for sepsis in the Sub-Saharan African context.
Methods and materials
This systematic review with meta-analysis has been reported in accordance with the ‘Preferred Reporting Items for Systematic reviews and Meta-analysis’ guidelines (online supplemental appendix 1).11 A protocol (online supplemental appendix 2) was developed for our review in accordance with the ‘Preferred Reporting Items for Systematic reviews and Meta-analysis protocols’ guidelines.12 It was registered on 12 July 2020 with the ‘International prospective register of systematic reviews PROSPERO’ (ID: CRD42020191067), which can be accessed on its website (https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020191067).
Supplemental material
Supplemental material
Search strategy and selection criteria
A comprehensive search strategy including all possible risk factors for neonatal sepsis in Sub-Saharan Africa was developed in cooperation with subject experts and an information scientist. Free text and database-specific subject headings were included. Publication date was restricted to 1 January 2010–7 August 2020 and language was restricted to English. A search strategy was first developed for PubMed (online supplemental appendix 3) and subsequently adapted in other databases.
Supplemental material
One author (CMB) searched PubMed, Embase, Web of Science (Clarivate Analytics) and African Index Medicus (accessed through the WHO) for published materials. ClinicalTrials.gov was searched for ongoing trials (grey literature). Additionally, the reference lists of the included studies were screened for potentially relevant studies. Systematic reviews and literature reviews were excluded from this systematic review, but the reference lists of these were screened as well. The authors of published abstracts were furthermore contacted to identify the full studies.
The following were the inclusion criteria:
Neonates (<28 days of age) with sepsis,4 that is, septicaemia/sepsis, pneumonia, meningitis, osteomyelitis, arthritis, urinary tract infections, malaria and candidiasis. Sepsis could be either clinical or laboratory-dependent diagnosis.
Reported on one or more risk factor for neonatal sepsis.
Observational prospective and/or retrospective analytical design, reporting on two outcome groups: one with sepsis and one without sepsis.
For inclusion in the meta-analysis, studies had to present quantitative data on the two above-mentioned outcome groups and the risk factors had to be reported on in at least three studies or found to be significant factors in at least two studies.
Data extraction
One author (CMB) screened the studies in Covidence (www.covidence.com) in the title stage. Two authors independently performed abstract screening and full-text study selection, where both authors had to approve the inclusion of the study in the systematic review. Disagreements during full-text study selection were resolved by discussion and consensus was reached in the presence of senior authors (AP and SL). If needed data were missing (eg, full article or raw data for meta-analysis), the authors were contacted in order to obtain the data. A predesigned extraction tool, specific to this review, was developed in Excel. This tool included study identification, location, study period, setting, definition of a neonate, definition of early-onset and late-onset neonatal sepsis (EONS and LONS), study design, sample size associated with risk factors, risk factors examined (neonatal and/or maternal), and limitations in relation to our review’s objective (eg, studies only examining risk factors for EONS). Only unadjusted/‘raw’ data were pooled in the meta-analysis.
Quality assessment
Two authors (CMB and CNS) independently performed quality assessment of the included studies using the National Heart, Lung, and Blood Institute’s (NHLBI) ‘Quality Assessment of Case-Control Studies’ and ‘Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies’.13 14 If the study design was unclear/poorly reported but the study reported data with a comparison group, we classified the study design as either ‘prospective’ (data collected when the neonate was in the neonatal unit) or ‘retrospective’ (data collected after the neonate had been discharged from the neonatal unit). We assessed each study on its own based on the details reported and considered the concepts for minimising the risk of bias. Discrepancies were resolved by discussion and consensus in the presence of senior authors (CHH, AP and SL) for all the above procedures. Covidence identified duplicate data and the duplicates were manually checked by CNS.
Statistical analysis
For the meta-analysis, a forest plot was created according to a random effects model. We chose the random effects model over the fixed effects model because it accounts for variations between studies, which we expected due to significant differences in the methodology, design of the studies as well as the different healthcare resources.15 ORs with 95% CIs were presented in the meta-analysis for dichotomous data (eg, sepsis vs no sepsis). The degree of heterogeneity across studies was determined using the I-test, with I2 values of 25% or less, 25%–75% and 75% or greater representing low, moderate and high inconsistency, respectively. P<0.05 was considered statistically significant. All statistical calculations were performed with the assistance of a statistician using Review Manager (V.5.4.1; The Cochrane Collaboration).
Patient and public involvement
To our experience from different settings in Sub-Saharan Africa, it is an important issue for the quality of patient treatment to follow guidelines and therefore to have relevant, updated guidelines for health workers to follow. This is what the research question of this study is based on. As it is a systematic review, there are no direct study participants, but we will disseminate the results on international conferences and to WHO and other stakeholders.
Results
A total of 6168 titles were screened after excluding 2674 duplicate records. Of these, 6083 were excluded based on screening of abstracts. The remaining 85 studies underwent full-text assessment for eligibility. Five of these were only available as an abstract online and we requested full text from the authors but only one author replied. Thirty-six full texts met the inclusion criteria of our review after discussion with senior authors and reaching consensus. Reasons for exclusion of 49 full-text records were other focus of study design (eg, not examining risk factors for sepsis) (n=8), wrong patient population/not neonates/no subgroup analysis (n=15), other outcomes/no risk factors studied (n=16), location not according to the protocol setting (eg, not in Sub-Saharan Africa or not in a hospital) (n=4), no full text (n=5) and duplicate (n=1). All included studies were published in peer-reviewed journals. The study selection process is illustrated in figure 1.
PRISMA 2009 flow diagram. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analysis.
All the 36 included studies were of observational study design. Twenty-eight studies were prospective (five cohort, six case–control, eleven cross-sectional studies and six studies of unclear/mixed unspecified design), seven were retrospective (three case–control studies, three cross-sectional studies and one study of unspecified design) and one was combined prospective and retrospective. The total sample size was 23 605 neonates (range: 10016–812917 18), and of these 4014 were diagnosed with sepsis. Ten studies reported the use of clinical guidelines for defining/diagnosing neonatal sepsis, while 26 studies required laboratory testing (eg, positive blood culture or haematological criteria) to establish the diagnosis of neonatal sepsis. All studies were conducted in secondary or tertiary level of care hospitals. The included studies were conducted in 10 different Sub-Saharan African countries, with majority of the studies conducted in Nigeria (n=10) and Ethiopia (n=10) (figure 2). The minimum study duration was 32 days19 and the maximum was 7 years and 6 months.20
Location of the studies.
Some of the included studies had a narrowed approach; for example, some studies only examined one or a few risk factors, and some studies only examined a narrowed population (ie, babies born before arrival). There were variations in defining EONS and LONS, with EONS ranging from 48 hours to 7 days. The characteristics of the included studies are provided in table 1.
Characteristics of the included studies
Supplemental material
According to the the NHLBI quality assessment, 6 studies were rated as good, 8 were rated as fair and 22 were rated as poor (online supplemental appendix 4, table 1). No studies were excluded after quality assessment.
Risk factors were classified as neonatal, maternal or sociodemographic factors in our review. A total of 60 risk factors were reported. Twenty-seven studies examined both neonatal and maternal risk factors.
Meta-analysis
Thirty-two studies were included in the meta-analysis (n=22 731 neonates). For each risk factor, a meta-analysis with adjacent forest plot was performed (not shown). The number of studies and patients in the meta-analysis ranged from 3 studies and 832 patients to 21 studies with 14245 patients. The 33 examined risk factors are provided in table 2.
Risk factors for neonatal sepsis reported in included studies and included in the meta-analysis
Four studies6 20–22 did not provide sufficient data needed to conduct meta-analysis and we did not obtain these data after contacting the authors. These studies were therefore not included in the meta-analysis. Furthermore, some studies did not provide sufficient data for all of the examined risk factors in the studies.
The following neonatal risk factors were found significant1:
Resuscitation at birth (12 studies and 3363 patients) increased the risk of sepsis (OR 2.70, 95% CI 1.36 to 5.35), but with a considerable I2 (92%).
Birth weight <1.5 kg (7 studies, 10 482 patients) increased the risk of sepsis (OR 3.37, 95% CI 1.59 to 7.13), but with a considerable I2 (83%).
Birth weight 1.5–2.5 kg (16 studies and 5151 patients) increased the risk of sepsis (OR 1.36, 95% CI 1.01 to 1.83), but with a considerable I2 (76%).
Low Apgar score at the first minute (7 studies and 2647 patients) increased the risk of sepsis (OR 3.69, 95% CI 2.34 to 5.81), but with a considerable I2 (77%).
Low Apgar score at the fifth minute (12 studies and 4185 patients) increased the risk of sepsis (OR 2.55, 95% CI 1.46 to 4.45), but with a considerable I2 (90%).
Prematurity <37 weeks (21 studies and 14 245 patients) increased the risk of sepsis (OR 1.91, 95% CI 1.27 to 2.86), but with a considerable I2 (90%).
No crying after birth (7 studies and 2772 patients) increased the risk of sepsis (OR 3.49, 95% CI 1.42 to 8.55), but with a considerable I2 (92%).
Male sex (18 studies and 4984 patients) increased the risk of sepsis (OR 1.30, 95% CI 1.01 to 1.67), but with a moderate I2 (73%).
The following maternal risk factors were significant:
Prolonged labour (11 studies and 11 190 patients) increased the risk of sepsis (OR 1.57, 95% CI 1.08 to 2.27), but with a moderate I2 (73%).
Premature rupture of membranes (PROM) (18 studies and 5620 patients) increased the risk of sepsis (OR 2.15, 95% CI 1.34 to 3.47), but with a considerable I2 (88%).
Multiple digital vaginal examinations (3 studies and 8684 patients) increased the risk of sepsis (OR 2.22, 95% CI 1.27 to 3.89), but with a considerable I2 (79%).
Meconium-stained amniotic fluid (8 studies and 10 108 patients) increased the risk of sepsis (OR 2.72, 95% CI 1.58 to 4.69), but with a considerable I2 (84%).
Intrapartum fever (10 studies and 2966 patients) increased the risk of sepsis (OR 2.28, 95% CI 1.18 to 4.39), but with a considerable I2 (84%).
Foul-smelling vaginal discharge (4 studies and 1318 patients) increased the risk of sepsis (OR 3.31, 95% CI 2.16 to 5.09), with no I2 heterogeneity.
The following sociodemographic risk factor was significant:
Low socioeconomic status (3 studies and 832 patients) increased the risk of sepsis (OR 1.93, 95% CI 1.11 to 3.35), but with a moderate I2 (62%).
The meta-analysis and forest plots of the four risk factors with the highest OR for neonatal sepsis are provided in figure 3. The Mantel-Haenszel (M-H) formula was used for the analysis. We explored post-hoc for potential causes of heterogeneity via subgroup analyses in the meta-analysis with substantial heterogeneity (I2 >75%), but country, design (retrospective vs prospective design), quality of study and publication year did not indicate a significant difference.
Meta-analysis and forest plots of the four risk factors with the highest OR for neonatal sepsis.
Discussion
It is of importance to prevent neonatal sepsis in order to reduce neonatal mortality to at least as low as 12 per 1000 live births in 2030, as specified by the SDG. One step is to identify the risk factors for neonatal sepsis. In this systematic review and meta-analysis, we found that the significant risk factors for neonatal sepsis in Sub-Saharan Africa were resuscitation at birth, low birth weight (<1.5 kg and 1.5–2.5 kg), low Apgar score at the first and fifth minute, prematurity <37 weeks, no crying right after birth, male sex, prolonged labour, PROM, multiple digital vaginal examinations, meconium-stained amniotic fluid, intrapartum maternal fever, foul-smelling vaginal discharge and low socioeconomic status. Male sex was found to be a significant risk factor in the meta-analysis, even though only 1 of the 23 studies which examined the association found male sex to be a risk factor (table 2).
Our findings are to some extent in line with a literature review from 2009 on the risk factors for maternal sepsis and EONS in Sub-Saharan Africa, where the most common risk factors for EONS were identified as prematurity, PROM, maternal fever, low birth weight and difficulties at delivery (obstructed labour or birth asphyxia).7 Our review and meta-analysis furthermore identified resuscitation at birth, low Apgar score at the first and fifth minute, no crying right after birth, male sex, prolonged labour, multiple digital vaginal examinations, meconium-stained amniotic fluid, foul-smelling vaginal discharge and low socioeconomic status as risk factors. However, we did not find birth asphyxia to be a risk factor. The review from 2009 examined the risk factors for EONS only, whereas our review and meta-analysis examined the risk factors for both EONS and LONS. EONS is more likely to reflect vertically acquired infections from the maternal genital tract and consequently has a different aetiology than LONS, different risk factors and potentially different means of prevention.7 Not all the included studies in our review and meta-analysis differentiate between EONS and LONS and there is no universal consensus on the definitions.
When comparing our findings with a systematic review and meta-analysis of risk factors for neonatal sepsis in India from 2019, we also find that these are to some extent in line. The review from India found that male gender, outborn admission, need for artificial ventilation, birth weight, delivery <37 weeks of gestation and PROM were risk factors for neonatal sepsis.8 Our review and meta-analysis furthermore identified low Apgar score at the first and fifth minute, no crying right after birth, prolonged labour, multiple digital vaginal examinations, meconium-stained amniotic fluid, intrapartum maternal fever, foul-smelling vaginal discharge and low socioeconomic status as risk factors. In our meta-analysis we did not find outborn admission to be a risk factor. The differences between our findings and the findings from India could indicate different risk factors in the two settings, but it could also partly be due to structural differences in the studies included. The Indian review included 13 studies with the diagnosis of neonatal sepsis based on laboratory testing, whereas our review included 36 studies, with 26 studies based on a laboratory-dependent diagnosis of neonatal sepsis and the remaining 10 studies based on clinical diagnosis. Data from studies that used clinical criteria exclusively to diagnose neonatal sepsis were included in our review and meta-analysis due to the fact that not all hospitals in Sub-Saharan Africa have access to validate the sepsis diagnosis with laboratory testing. Furthermore, the studies from the Indian review were solely from hospitals in urban settings, whereas the studies included in this review were conducted at both rural and urban hospitals. Risk factors for neonatal sepsis might be different in urban and rural settings.
Our findings add multiple risk factors to the risk factors identified in the WHO’s universal guidelines. In our meta-analysis we identify resuscitation at birth, low birth weight (<1.5 kg and 1.5–2.5 kg), low Apgar score at the first and fifth minute, prematurity <37 weeks, no crying right after birth, male sex, prolonged labour, multiple digital vaginal examinations, meconium-stained amniotic fluid and low socioeconomic status as significant risk factors for neonatal sepsis, none of which are mentioned in the WHO guidelines. However, further research is needed to confirm our findings and they do not necessarily imply expansion of the WHO criteria for prophylactic antibiotics. That is, in our meta-analysis, male sex is a risk factor, but we do not suggest treating all male children with prophylactic antibiotics. If more risk factors were to be treated with prophylactic antibiotics, the risk of overtreatment should be kept in mind since it could lead to high medical cost and use of resources and increased antibiotic resistance.23 Alternative preventive strategies, such as in-hospital observation of the newborn and measurement of C-reactive protein (CRP), are used in high-income countries and could be feasible in some LMICs but also challenging, for example, due to lack of resources. Future research should focus on identifying the risk factors qualifying for preventive measures.
This systematic review and meta-analysis has several strengths and limitations. The broad search strategy and the combination of global and regional databases reduced the risk of missing relevant regional studies and ensured that the evidence in this review was derived from different countries and different hospital settings. The relatively high number of included studies is a strength. However, the geographics of the included studies make our findings not necessarily generalisable; Ethiopia and Nigeria together accounted for more than 50% of the included studies and many Sub-Saharan countries are not represented in this review. Furthermore, the countries in Sub-Saharan Africa differ in the level of hospital expertise, hygiene and medical tools, as well as in climate, diseases and bacteria, limiting the generalisability of the review findings. Another limitation is that the studies were heterogeneous; some were based on a clinical diagnosis of sepsis, some laboratory-dependent, some only examined limited populations, some were retrospective and some were prospective. The studies were also heterogeneous in regard to which risk factors to investigate (table 2). This heterogeneity makes them not perfectly comparable and is thus a limitation. The English language restriction is also a possible risk of bias and is a limitation. Africa has 29 francophone countries and it could be presumed that we could have missed relevant studies written in French. However, a quick search in PubMed with language restricted to French showed 105 studies, of which none was relevant to this review based on their English abstracts. The greatest limitation of this systematic review and meta-analysis is the overall poor quality of the included studies. The study designs used for risk factor analysis (eg, cross-sectional studies) differ from experimental designs and are more prone to bias.24 Furthermore, multiple studies found some factors to be significant risk factors for neonatal sepsis, but when looking at the data, we found that the factors were protecting factors. Despite email correspondence with the authors, agreement was not obtained.
This systematic review and meta-analysis found multiple risk factors for neonatal sepsis in Sub-Saharan Africa, many of which are not on the WHO’s recommendations for prophylactic antibiotics. It has previously been emphasised that there is a need to develop clinical guidelines for prevention and treatment of neonatal sepsis that are specific to the Sub-Saharan African context10 and our review supports this notion. However, even though there are already multiple studies on risk factors for neonatal sepsis in Sub-Saharan Africa, there is a need for research of higher quality in the future as well as research in different settings in order to make presumptions, generalise on the topic or make multinational recommendations for clinical practice. National guidelines for Sub-Saharan African countries might also be beneficial due to differences in risk factors and bacterial agents among the countries. If new guidelines are to be developed, the challenges to implementation and resources should be kept in mind. There are still too many preventable neonatal deaths in LMICs, but with new preventive guidelines it might be possible to save thousands of lives.
Data availability statement
No data are available. The only original data published were the protocol of the study, which can be found on the PROSPERO web page (ID: CRD42020191067). The remaining data are published in the articles included in this systematic review and meta-analysis.
Ethics statements
Patient consent for publication
References
Supplementary materials
Supplementary Data
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.
Footnotes
Contributors CMB: Development of the protocol and search strategy, screening of studies, performance of quality assessment, development of all sections of the manuscript, development of the tables, performance of meta-analysis. Guarantor. CNS: Development of protocol, screening of studies, performance of quality assessment. SL, AP: Development of protocol, screening of studies, development of the tables, development of all sections of the manuscript. CHH: Development of protocol, development of the Introduction section. UN, JSB: Development of the results and discussion sections, performance of meta-analysis.
Funding This work was supported by the University Hospital of Copenhagen, Rigshospitalet. The funder of the study had no role in study design, data collection, data analysis, data interpretation or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Map disclaimer The depiction of boundaries on the map(s) in this article does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. The map(s) are provided without any warranty of any kind, either express or implied.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
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.