Inequities in postnatal care in low- and middle-income countries: a systematic review and meta-analysis

1Each year an estimated 289 000 women die worldwide from complications related to pregnancy, childbirth or the postnatal period and up to two thirds of such maternal deaths occur after delivery., Poor outcomes of maternal and neonatal care also include 2.9 million neonatal deaths per year. Of the maternal and neonatal deaths that occur globally, 99% occur in low- and middle-income countries.,

2According to the World Health Organization (WHO), the postnatal period begins immediately after childbirth and lasts six weeks. In low-income countries, almost 40% of women experience complications after delivery and an estimated 15% develop potentially life-threatening problems. Postnatal care services are a fundamental element of the continuum of essential obstetric care – which also includes antenatal care and skilled birth attendance – that decreases maternal and neonatal morbidity and mortality in low- and middle-income countries., Compared with other maternal and infant health services, coverage for postnatal care tends to be relatively poor. Increasing such coverage has been highlighted as a priority. In the Democratic Republic of the Congo, for example, at least 93% of pregnant women receive antenatal care and skilled birth attendance but only 35% of birthing women receive postnatal care. In Kenya, fewer than 20% of women use postnatal care services. In 2014, WHO recommended that a mother and her newborn child should receive postnatal care within 24 hours of the birth and then at least three more times – i.e. at least on day three after the birth, in the second week after the birth and six weeks after the birth. Postnatal care services can be defined as preventive care practices and assessments that are designed to identify and manage or refer complications for both the mother and the neonate. Typically, such services include an integrated package of routine maternal and neonatal care as well as extra care for neonates that are considered particularly vulnerable because, for example, they are preterm, have a low birth weight, are small for gestational age or have mothers infected with human immunodeficiency virus (HIV). Possible postnatal interventions for the mother include: (i) iron and folic acid supplementation for at least three months; (ii) screening for – and treatment of – infection, haemorrhage, thromboembolism, postnatal depression and other conditions; (iii) prophylactic antibiotics given to women who have a third- or fourth-degree perineal tear; and (iv) counselling on early and exclusive breastfeeding, nutrition, birth spacing and family planning options – including any available contraception.,, Possible interventions for the neonate include: (i) care of the umbilical cord (ii) special care for preterm, low-birth-weight and HIV-infected neonates;,, (iii) screening and treatment of infections and postnatal growth restriction; (iv) assessment of factors predisposing to infant anaemia; and (v) teaching the mother to seek additional care for her neonate if she notices danger signs such as convulsions or problems with feeding.

3Low use of postnatal care services is associated with lack of education, poverty and limited access to health-care facilities. However, these associations have not been assessed systematically. We therefore conducted a systematic review of the relevant evidence from low- and middle-income countries, to inform policy-making, help strengthen health systems and increase access to – and use of – postnatal care services.

4We followed guidelines for systematic reviews from the Cochrane Collaboration and a standardized methodology described in an explicit protocol. The review was registered with the Prospero database (registration number: CRD42013004661) and results were reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.

5To identify the studies of interest, we searched the Medline, Embase and Cochrane Central databases and grey literature for relevant medical subject headings and keywords. We focused on articles published between 1 January 1960 and 31 May 2013 in English, French, Spanish, Portuguese and Chinese and were assisted by an expert librarian. Our search strategy combined terms related to postnatal or postpartum care, use or accessibility, determinants or inequities and low- or middle-income countries. Our full search strategy is detailed in Appendix A (available at: https://dl.dropboxusercontent.com/u/28446882/Appendix%20A.pdf). To identify further data that might be useful, we also checked the reference lists of the articles found to be of potential interest, visited institutional web sites and contacted the authors of some of the articles of interest and other experts in the field.

6We retrieved data from experimental, quasi-experimental and observational studies of women aged 15–49 years, that had been implemented in low- or middle-income countries as defined by the World Bank. The primary outcome of interest was the use of postnatal care services – i.e. at least one follow-up visit in the 42 days post-childbirth. We included studies in which the potential socioeconomic, geographical and/or demographic determinants of the use of postnatal care had been assessed. The potential socioeconomic determinants that we investigated were socioeconomic status, occupation and education. We investigated distance and travel time to a health centre and place of residence – i.e. urban or rural – as potential geographical determinants and ethnicity, marital status, religion and immigration status as potential demographic determinants. We analysed data from studies that included at least one association measure – such as a frequency ratio or difference – or the result of at least one statistical test in which use of postnatal care had been compared across two or more categories. We included relative comparisons to a reference group (e.g. concentration indexes) and absolute comparisons (e.g. slope indexes of inequality). In some relevant studies, a concentration index was used to measure the relationship between accumulated proportions of mothers ranked by their socioeconomic status against the cumulative proportion of postnatal care use. In these studies, a positive value for the index indicates that rich households have greater coverage than poor households, a negative index indicates that poor households have greater coverage than rich households and zero values for the index that coverage is independent of socioeconomic status. Other studies used a slope index of inequality to estimate the absolute difference in percentage postnatal care coverage between individuals at the top and bottom of the socioeconomic status scale. In such studies, a high slope index of inequality would have indicated great inequity in coverage.

7The eligibility of each study identified in the initial screening was assessed by two reviewers using a standardized form with explicit inclusion and exclusion criteria. There was a high level of agreement between the reviewers (Cohen’s kappa, κ: 0.92). Data were extracted with a standardized data collection form that had been pilot tested on a random sample of studies. We collected data on country, setting, year of publication, study design, sample size, population attributes, outcome definition, comparison groups, point estimates and precision measures.

8Two individuals, working independently, assessed the scientific quality of each selected study using the Effective Public Health Practice Project’s quality assessment tool for quantitative studies – after extending the criteria for selection bias assessment. Scientific quality was categorized as high, moderate or low if, respectively, the risk of bias in the study results was considered to be very low, low or high. The level of agreement between the two assessors of quality was good (κ: 0.75). Discrepancies in the assessment of eligibility or scientific quality were resolved in discussions with an experienced researcher.

9Evidence tables were generated to summarize the selected studies and results descriptively. We conducted a qualitative synthesis of the findings. We also conducted a meta-analysis of selected studies that provided a comparable classification of the outcome and determinants of interest. For this purpose, we also required either estimates of the standard errors for the association measure or confidence intervals that allowed us to derive such estimates. Many of the studies included in the systematic review had to be excluded from the meta-analysis because of differences in the classification or definition of determinants. We pooled the association measures for socioeconomic status and geography, as represented by socioeconomic status quintile and an indicator of urban/rural place of residence, respectively. We assessed heterogeneity of these results using Cochran’s Q testand the I2 statistic. We used random-effects meta-analysis models when heterogeneity was statistically significant (P > 0.1) and I2 was moderate or high according to the criteria of Higgins et al. We conducted sensitivity analyses by removing studies deemed to be of low quality or potential outliers., We assessed publication bias in the meta-analyses with funnel plots. Data analysis was performed using Stata version 12.0 (StataCorp LP, College Station, United States of America).

10Our initial search produced 3152 articles of potential interest and articles describing 36 studies,,,,,– contributed to our qualitative synthesis of evidence (Fig. 1). Data from 10 of the studies were included in the meta-analysis.,–,,,,,, The 36 studies included in our qualitative synthesis of evidence comprised two randomized controlled trials, three quasi-experimental studies, two cohort and 28 cross-sectional studies, and one investigation of 31 demographic and health surveys. Of these 36 studies, 11 were conducted in low-income countries, 24 in middle-income countries and one in both low- and middle-income countries. Three, 26 and six of the 36 studies were deemed to be of high, moderate and low scientific quality, respectively. A lack of information on methodology prevented the assessment of the scientific quality of one study included in the qualitative synthesis (Table 1, available at: http://www.who.int/bulletin/volumes/93/4/14-140996).

11Our qualitative assessment of relevant studies indicates that there was a gradient in the use of postnatal care according to socioeconomic status – as measured on different scales (Table 2, available at: http://www.who.int/bulletin/volumes/93/4/14-140996).,,,,,–, Results of our meta-analysis that included data on socioeconomic status from studies of moderate quality also indicated that the higher the socioeconomic status of the mother, the more likely she was to access postnatal care (Fig. 2, Fig. 3, Fig. 4 and Fig. 5).

12Meta-analysis was used to derive pooled adjusted odds ratios (OR) from 10 studies and a total of 136 431 women. For each quintile of socioeconomic status, the Q test gave a significant result and the I2 statistic fell between 50% and 75% – indicating moderate heterogeneity. When the lowest quintile (Q1) was used as the reference, the pooled OR for the highest quintile (Q5) was 2.27 (95% confidence interval, CI: 1.75–2.93). The corresponding OR for Q4, Q3 and Q2 were lower, at 1.60 (95% CI: 1.30–1.98; I2: 70%), 1.32 (95% CI: 1.12–1.55; I2: 50%) and 1.14 (95% CI: 0.96–1.34; I2: 52%), respectively.

13In a sensitivity analysis, we removed the potentially atypical data reported by Amin et al. The pooled OR for Q5, Q4, Q3 and Q2 – with Q1 used as the reference – were reduced to 2.09 (95% CI: 1.70–2.56), 1.55 (95% CI: 1.27–1.90), 1.30 (95% CI: 1.10–1.54) and 1.08 (95% CI: 0.95–1.24), respectively.

14The data in a report included in the systematic review showed concentration indexes and slope indexes of inequality for use of postnatal care in 31 countries (Table 3). For the low-income countries, the mean concentration index was 0.23 and the mean slope index of inequality was 53%. The corresponding values for the middle-income countries were 0.18 and 61%, respectively. In Pakistan, exposure to a voucher scheme led to significant increase in the use of postnatal care (OR: 4.98; P < 0.001).

15Our qualitative assessment of studies indicated marked variations in the use of postnatal care according to the level of education of the women investigated – or their partners (Table 2). Compared to women who had received no formal education, women who had attended primary education were more likely to use postnatal care,,,, and women who had completed secondary school were the most likely to access postnatal care.,,,,,,, In three studies, the duration of maternal schooling was found to be positively correlated with postnatal care use.,, Compared with other women, those with husbands who had completed secondary school also appeared more likely to use postnatal care.,, In Lebanon, an educational intervention to emphasize the importance of postnatal care led to a marked increase in the use of such care (relative risk: 2.8; 95% CI: 2.2–3.4). Inconsistent classification of education status prevented us from performing a meta-analysis of these apparent determinants of the use of postnatal care.

16The income-earning occupations of women and their husbands appear to influence the women’s use of postnatal care (Table 2). For example, women married to men with professional, technical or managerial occupations were more likely to use postnatal care than women married to manual labourers (OR: 2.22; 95% CI: 1.62–2.81). Similarly, women married to men with well paid jobs were more likely to use postnatal care than women married to farmers (OR: 1.45; P < 0.05). In China, women with so-called white-collar occupations were more likely to use postnatal care than other women (OR: 2.17; P < 0.001). Inconsistent classification of occupation impeded any corresponding meta-analysis.

17A qualitative assessment of the evidence indicated that postnatal care was more commonly used by women living in urban areas than by their rural counterparts (Table 4).,,,,,,–, Our meta-analysis of this trend was based on five studies and a total of 46 913 women.,,,, As a Q test gave a significant result (P < 0.001) and I2 was 83.7%, heterogeneity was considered high. With women in rural areas used as the reference, our initial estimate of the pooled OR for use of postnatal care by women residing in urban areas was 1.36 (95% CI: 1.01–1.81; Fig. 6). After removing the study deemed to be of low quality, the estimated pooled OR became 1.21 (95% CI: 0.95–1.53). In several studies included in our systematic review, distance to the nearest health facility was also found to be associated with use of postnatal care services. In India, for example, the relevant OR for distances of 2–5 and at least 6 km – with a distance of less than 2 km used as the reference – were 0.80 (95% CI: 0.67–0.95) and 0.64 (95% CI: 0.50–0.83), respectively. In rural areas of India, the presence of a bus service has been found to increase the use of postnatal care services (OR: 1.18; P < 0.01).

18In one study, use of postnatal care services was higher among Muslim women than among Christian women (OR: 2.01; 95% CI: 1.24–3.25). In contrast, in another study, Muslim women seemed less likely to use such services than their non-Muslim counterparts (OR: 0.77; 95% CI: 0.61–1.34). In Nepal, compared with Hindu women, Buddhist women were less likely to use postnatal care services (OR: 0.25; P < 0.001). Overall, our systematic review of relevant studies revealed no clear trend in the use of such services according to religion (Table 5).

19In India, women belonging to the lower social groups – i.e. those belonging to scheduled castes (OR: 0.69; 95% CI: 0.55–0.86), scheduled tribes (OR: 0.71; 95% CI: 0.54–0.91) or other so-called backward classes (OR: 0.58; 95% CI: 0.48–0.71) – were found to be less likely to use postnatal care services than those belonging to upper castes (Table 5). Although we found statistically significant differences in the use of postnatal care services according to the ethnicity of the women investigated, our systematic review revealed no clear trend in the use of such services according to whether the woman involved belonged to a minority or majority group.,,,,,,

20We have systematically reviewed studies assessing inequities in the use of postnatal care services in low- and middle-income countries. We found strong and consistent evidence indicating that the use of such services was relatively high among women with high socioeconomic status and among more educated women. In general, women with high socioeconomic status belong to those households that can afford the medical, non-medical and opportunity costs of postnatal care. In addition, such women may be relatively empowered and have more autonomy than their poorer counterparts.Educated women are considered to have relatively good access to – and management of – health service information, and relatively accurate and detailed perceptions of diseases and their complications and treatments., There also seems to be an independent association between a woman’s use of maternal services and her partner’s education.

21In addition to increasing household income, employment can increase awareness and modify a person’s behaviour, through social and community interactions. However, in low- and middle-income countries, there seems to be no clear and consistent association between a woman’s income-generating employment and her use of postnatal care services. A woman in gainful employment may still have no control over any of her household’s finances. In addition, a woman’s economic activity may also be poverty-induced, only seasonal and/or relatively poorly remunerated.,

22Compared with women living in rural areas, urban women have generally better access to postnatal care services as well as other advantages of urban life, such as greater exposure to health-promotion programmes., In many rural areas, improvements in the numbers of primary health care facilities, the provision of postnatal care services of high quality and public transportation are required. Although the relationship between ethnicity and use of postnatal care services appears complex, there are some ethnicities, such as India’s lower castes, that often seem to be disadvantaged.

23We found insufficient homogeneous classification of data to conduct meta-analyses for occupation or level of education. Our meta-analysis for place of residence may have been weakened by the suboptimal precision of a between-studies variance estimate. Despite these limitations, our study indicates that the use of postnatal care remains highly inequitable according to socioeconomic status, education and geographical access to health facilities. There are several research and knowledge gaps that need to be filled. For example, we need research to further understand health-seeking behaviours and to inform policy-makers. As most maternal deaths occur during the postnatal period, primary research on postnatal care services should be prioritized. Further research on the contextual and systems-level determinants of the use of such services and the effectiveness of strategies to improve the coverage and quality of postnatal care is also needed. It remains unclear if the number and timing of postnatal consultations recommended by WHO are optimal and achievable in every setting. It also remains to be determined if postnatal care at home can be made as effective and cost-effective as similar care provided by health facilities. We need both community-level interventions to promote the use of postnatal care services and health systems interventions to improve the supply of affordable and quality services – including, but not limited to, alleviation of user-fees and the promotion of postnatal care by health professionals. Strengthening the effectiveness and responsiveness of systems for health-care delivery will also catalyse access to – and use of – postnatal and other obstetric care services. In the current and future elaboration of universal health coverage and equity schemes in low- and middle-income countries, due consideration should be provided to postnatal care services.