Impact of social determinants of health on progression from potentially life-threatening complications to near miss events and death during pregnancy and post partum in a middle-income setting: an observational study

Introduction

Ill health during pregnancy represents a continuum between normal health and death extremes.1 Potentially life-threatening complications (PLTC) lie on this spectrum and could either progress to recovery or clinical deterioration, depending on various factors related to the primary condition and access to and appropriateness of care. Pregnant patients with PLTC who experience clinical deterioration either die (maternal mortality) or nearly die (maternal near-miss event). These two outcomes—maternal mortality and near-miss, which represent the most severe pregnancy complications—affect approximately 2% of pregnancies and are referred to as severe maternal outcomes (SMO).1–3

Maternal mortality is a key performance indicator of maternal health and health systems functioning, tracked and monitored globally.4 5 With maternal mortality rates falling, following the measures taken under the WHO’s millennium developmental goals, PLTC and near-miss rates are increasingly considered the more appropriate indicators of maternity care.1 6

It is recognised that along with the clinical factors, social determinants of health (SDH), which represent the non-medical factors that influence health outcomes,7 may contribute to the development of maternal morbidity and mortality, and information on these determinants may help better understand the social structural framework and the contributory, non-medical mechanisms associated with SMO.7–10 However, information on the association of SDH with SMO is limited in the literature.8 10 11

SDH are broadly categorised as structural and intermediary determinants.7 8 Structural determinants are those factors referring to the interplay between the socioeconomic–political context that produces social stratification and differential social positions,12 13 whereas the intermediary determinants include individual-level influences and health system and community contextual factors that influence health outcomes. This study aimed to determine the association of various SDHs with the development of SMO among those with PLTC during pregnancy and post partum.

Methods

Study design and setting

This cohort study was conducted using prospectively collected data for a study on the incidence and determinants of maternal near-miss and their impact on long-term maternal health between May 2018 and September 2021. The setting was the Women and Children’s Hospital attached to the Jawaharlal Institute of Postgraduate Medical Education and Research, Pondicherry, India, which has an annual birth rate of 17 000–18 000 and provides tertiary care to those with high-risk pregnancies, mainly from rural (based on the census 2011 definition as as an area not included as cities, and those fulfilling the following criteria of having a population of less than 5000, density of population less than 400 per sq km, and more than ‘25 per cent of the male working population’ is engaged in agricultural pursuits14) regions of Pondicherry and the neighbouring districts of Tamil Nadu. Informed consent was obtained from all women and/or their relatives in the study.

Study population

Women with PLTC identified from the intensive care and high-dependency unit, based on the criteria by the WHO, were included in the study.1 3 These conditions include hypertensive and other medical disorders, labour-related complications, obstetric haemorrhage and infections developing during any stage of pregnancy or post partum, as shown in table 1

Table 1

List of potentially life-threatening complications

Data collection

As part of the original study, research staff collected details on SDH after obtaining informed written consent from participating mothers. These data were categorised into structural and intermediary determinants. Structural determinants included the level of education, religion and occupation. All belonged to the same ethnicity, that is, South Indian descent.

Intermediary determinants included information on

  1. Community context: this included socioeconomic status as determined by the BG Prasad classification scale, grouped into five categories based on per capita monthly income (calculated as total monthly family income/total number of family members),15 and whether they belong to a socially, economically and educationally challenged classes compared with other social groups, as defined by the National Commission for Backward Class under the government of India.16 17

  2. Individual and family context: maternal age, marital status and family size (classified into three categories based on the number of members: nuclear with two, medium sized with three to four, and large family with five or more members), number of live children, category of PLTC.

  3. Health services context: distance to the health facility (a cut-off of 10 km chose based on the previous study on referral patterns, unpublished work), self-referral for admission versus admission following transfer from a health facility, and the number of centres visited before the admission to the tertiary centre (in addition to the booking centre), number of antenatal clinic visits, place of delivery and the information provided regarding the complications during pregnancy.

The delays at the various levels include (1) delay in deciding to seek appropriate medical help (personal or family reasons), (2) delay in reaching the appropriate health facility (lack of logistics or communication), and (3) delay in receiving appropriate treatment at the facility (lack of equipment or health personnel in the facility) were also collected.18 We also collected the details of previous miscarriages (any pregnancy loss at less than 28 weeks of gestation)19 and previous perinatal loss (stillbirth and early neonatal loss).

Outcomes and analysis

The primary outcome was the development of SMO, defined as the occurrence of either (1) maternal near miss based on the WHO criteria1 or (2) maternal death, any time during pregnancy till 6 weeks post partum.

Categorical variables such as level of education, religion, parity and category of PLTC were summarised as frequencies and percentages; continuous variables, such as maternal age, were expressed as means and SD. Univariate and multivariable logistic regression was performed to assess the association of various structural and intermediary determinants with the development of SMO. In multivariable analysis, those factors with p values <0.05 in univariate analysis were assessed for independent association with SMO. Unadjusted and adjusted ORs (OR and aOR) and their 95% CIs were reported. A p value <0.05 was considered to be statistically significant. Analysis was performed using STATA V.17.0 (Stata Corp).

Patient and public involvement

Patients or the public were not involved in the design, conduct, reporting or dissemination plans of our research.

Results

Of the 38 292 births at the hospital between May 2018 and April 2021, 37 590 were live births, and 1833 (4.9%) had PLTC. SMO occurred in 380 women (10 per 1000 live births), of which 57 (15.0%) succumbed to various complications, and 323 (85.0%) were classified as maternal near misses (figure 1).

Figure 1
Figure 1

Participant flow in the study.

As seen in table 2, the distribution of women belonging to lower socioeconomic status, larger family size (five or more adults), a greater number of live children, older maternal age, multiparity, previous miscarriage, haemorrhagic disorders, hypertensive and other medical diseases, and occurrence of sepsis/infections were significantly different between those who developed SMO with those who did not develop them. The incidence of SMO was found to be higher when the delivery was in a level one centre (34/69, 49%) and home births (3/10, 30%) compared with those delivering at tertiary centres (343/1754, 19.6%, p <0.001). Analysis of factors relating to the three-delay model showed that delays in receiving adequate care in the facility and seeking care differed significantly. The occurrence of SMO was lower among those living more than 10 km from the health facility.

Table 2

Distribution of various determinants of health among those with and without the severe maternal outcome in the study

On multivariable analysis, after adjusting for the risk factors shown in table 3, an association between SMO and some clinical and SDH-related factors persisted (figure 2). Clinical conditions that increased the odds of SMO included medical disorders other than hypertension (aOR=1.50), obstetric haemorrhage (aOR=4.63) and infections (aOR=2.93), whereas a history of one previous miscarriage (aOR=0.51) lowered the odds. Determinants such as those belonging to a socially backward community (aOR=0.45), middle socioeconomic class (aOR=0.61), living 10 or more kilometres from a health facility (aOR=0.56), who had 4 or more antenatal visits before admission (aOR=0.53), and those referred from resource-limited facilities (aOR=0.624) had significantly lower odds of SMO. Advancing maternal age (aOR=1.04, 4% increase in odds for SMO with each year increase in age), multiparity (aOR=1.44, compared with nulliparity), those admitted following a referral from a lower centre to the tertiary centre (compared with self-referred, aOR=2.95) and delays in seeking care (aOR=3.30) increased the odds of development of SMO.

Table 3

Results of multivariate logistic regression analysis assessing the association of various determinants of health with the severe maternal outcome

Figure 2
Figure 2

Association of social determinants of health with severe maternal outcomes among patients with potentially life-threatening complications after multivariate analysis.

Discussion

At our tertiary referral centre in a middle-income setting, the incidence of SMO in the study period was 20.7% among those admitted with PLTC. There was a strong association between clinical conditions such as obstetric haemorrhage, infections and medical disorders other than hypertension and SMO. Those with one previous miscarriage, compared with those without any miscarriage, had lower odds for the development of SMO. The study was also able to identify independent associations between SDH and SMO. Determinants such as belonging to a socially backward community and middle socioeconomic class were associated with 50% lower odds of an SMO, whereas with each year of increasing age, the odds increased by 4%, after adjusting for other determinants. Intermediary determinants noted to be protective included four or more antenatal clinic visits before admission, living farther than 10 km from a health facility, and facilities with resource limitations. Determinants such as increasing maternal age at childbirth, multiparity, admission following referrals and delays in seeking care increased the risk of SMO.

The prospectively gathered data, ensuring the completeness of the information and parameters, with a large sample size of women with PLTC, can be considered the study’s strength. Despite its strengths, the study has some limitations. The first pertains to self-identification as belonging to lower social classes due to the stigma they face to avoid them, resulting in misclassification bias and an underestimation of the magnitude of the association. Second, being limited to a tertiary referral centre for high-risk pregnancies in the south-eastern region of India, with inclusion of all consecutive women admitted with PLTC, the data do not include those with PLTC in the community who were not referred or had either outcome before the referral. It is possible, therefore, that some individuals experienced SMO outside the tertiary hospital, resulting in underestimating or overestimating the strength of the association of the determinants with the progression of PLTC to SMO. However, given the regional challenges with data acquisition from community hospitals, a study of this nature would be difficult. Since we were assessing the association of the determinants with the development of SMO, we excluded those referred or transferred, which would have made it impossible to study the associations using regression analysis. Third, data on sociopolitical and community-level determinants and factors such as women’s autonomy were not collected, which would have helped understand the policy, macroeconomic and governance-related factors associated with the SMO. And finally, the data on other determinants, structural like the government policies and political context, and the intermediate determinants like the psychosocial and the social and environmental factors were not available to be analysed in the study.

In this study, approximately 20% of those admitted with PLTC experienced SMO, higher than the WHO estimates of 15% of pregnancies.1 As the index hospital is the preferred regional centre for high-risk pregnancies and those with various complications in pregnancy, the rates may reflect the condition/ severity at admission.

The strong association between medical conditions and SMO is not surprising. In particular, there was a strong positive association between obstetric haemorrhage, infections and medical disorders other than hypertension and the development of SMO.20 21 The presence of these individual health conditions, which either (1) developed during pregnancy and postpartum, such as haemorrhage or infection, or (2) aggravated during the pregnancy, such as pre-existing medical disorders, were independently associated with SMO, similar to other studies.21

Surprisingly, hypertensive disorders were not associated with an increased risk of developing SMO as seen in earlier studies, probably due to the increased awareness among pregnant women and healthcare personnel about its impact, leading to early recognition, protocol-based management using prophylactic magnesium sulphate or antihypertensive medications, more aggressive follow-up, and prompt delivery.22

The association between SDH and SMO and some of the structural and intermediary determinants have been described in the literature. For example, increased risk with increasing maternal age, economic status/household income, ethnicity, maternal education, women’s autonomy, place of residence and women’s awareness of complications through exposure to mass media have been shown to modify the risk of SMO factors.10 11 23

Increasing maternal age was a significant risk factor.11 A higher prevalence of pre-existing medical conditions and age-related physiological changes may have increased the risk associated with increasing maternal age. Similarly, higher birth order, or multiparity, increased the risk of SMO following lower utilisation of maternal health services.11 24

The occurrence of adverse pregnancy outcomes in the previous pregnancy is observed to increase the utilisation of the care and institutionalised deliveries compared with those who did not have it.11 25 In particular, women with previous miscarriages often fear the consequences to themselves and the baby in the subsequent pregnancy.26 This could have resulted in greater vigilance, possibly early admission, and a reduction in the odds of SMO among those with a prior miscarriage.

The literature demonstrates that, as in most families, especially those from the rural regions, pregnant individuals were primarily not the decision-makers and depended on their husbands (who mostly worked away from their hometowns to earn livelihood) and families for travel, including hospital visits. This could contribute to delays in seeking care, which, in our study, considerably increased the risk of developing SMO.11 27

Some of the associations between SDH and SMO seemed paradoxical. For example, the structural determinant, belonging to socially backward communities, was found to be protective. While many communities in low to middle-income countries remain underprivileged, with restricted access and affordability to healthcare, making them vulnerable to increased risk of complications,28 a possible explanation for our finding could be that women belonging to socially backward communities tend to use the support provided by the various maternity welfare schemes and the transport to the hospital, which could explain the lower rates of SMO.29 These maternity benefits schemes include Dr. Muthulakshmi Reddy Maternity Benefit Scheme for financial assistance as well as nutritional kits, with delivery till postnatal care provided free of cost along with the medicines, availability of 24-hour emergency services in the district (to facilitate accessibility within an hour of travel, upgraded after scheme after geo mapping) and referral transport availability through 108 ambulance service on a 24×7 basis which hands over to a higher level centre.30 Other studies have demonstrated that although most women from rural areas do not have formal schooling, the strong community ties and advice from the elders and community health workers help them become more knowledgeable about availing of health services, possibly resulting in early recognition and contact with the health system.31 32 A third possible explanation for this finding could be misclassification bias, wherein a considerable proportion of people belonging to these communities may have indicated belonging to a higher class.

Healthcare services, accessibility and affordability are more significant determinants of health in most low-income and middle-income countries.11 33 In our study, those living farther away from healthcare facilities were observed to have a lower risk of SMO. This finding can be explained by the possibility that through the motivation of village health workers and provision of free transport and healthcare under the governmental schemes, healthcare providers and women with complications plan for early referral and admission to higher care to facilitate treatment therefore are less likely to have PLTCs which then progress to SMO.32 34 Finally, those who self-referred or were referred from resource-limited facilities had a 38% reduction of odds for SMO compared with those referred from higher level centres. While a possible explanation for this could be that those presenting with complications to resource-limited centres get referred early to a higher centre due to the non-availability of specialists or equipment, it is equally possible that the more severe cases progress to near miss or death and, therefore, don’t get referred.35

This post was originally published on https://bmjopen.bmj.com