Menu
Menu
Menu
Menu
Researchers in Nigeria examined the epidemiological characteristics and factors associated with maternal outcomes using a mixed method approach: a prospective case control study design involving 375 pregnant women who received maternal care from a tertiary facility and in-depth interviews reporting the experience of near-miss survivors. A generalized ordered logit model was used to generate the estimates of partial proportional odds ratios (and 95% confidence intervals) across categories of the outcome variable. Factors strongly associated with maternal morbidity were late referral of women, presence of complications at booking antenatal visits, low birth weight, and severe birth asphyxia. The nearmiss women were further characterized, and a low proportion (25%) had organ dysfunction or failure. The challenge of such diagnoses in resource-constrained settings raises questions about the appropriateness of using organ dysfunction criteria in developing countries.
This study provides a further analysis of a prospective case control study that was carried out at the Obafemi Awolowo University Teaching Hospitals Complex (OAUTHC), Ile-Ife, southwest Nigeria from July 2006 to June 2007. The study was carried out simultaneously at two maternity hospitals under OAUTHC, which are situated in two separate local government areas (LGAs) of Osun State (Wesley Guild Hospital, Ilesa in Ilesa East LGA, and Ife Hospital Unit in Ife Central LGA). The study participants were pregnant women who sought maternal care at the hospitals during the antenatal (third trimester) or intrapartum period or within 42 days after delivery. Four unmatched controls were selected for every near-miss event. The details of the methodology (namely, study setting, population, sample size, and selection) have been described in an earlier publication (Adeoye, Onayade, & Fatusi, 2013). Here, we present fresh perspectives from the quantitative data and also include findings from qualitative aspects of the study based on a collection of narratives from near-miss survivors and subsequent narrative analysis (Hancock, Windridge, & Ockleford, 2007). The qualitative findings were used to put the quantitative data into a social context. Trained research assistants carried out a narrative interview with each of the 75 women who had experienced a near miss. The interview started with a “generative narrative question” whereby each woman was requested to relate her experience of the near-miss event and the associated events/factors. This was then followed by relevant questions from the data collector, drawing from a study guide, to gain better perspectives of the associated social and community factors. Where necessary, the information from the interview of the affected woman was supplemented with information from another adult, usually a woman and a close relative who was caring for the woman at the time of the near-miss event. The initial theme was based on the content of the interview guide. The interviews, on the whole, provided an opportunity to obtain a verbatim account of the pregnancy and the delivery experience as well as information on related maternal health issues including the use of birth preparedness plans, knowledge of warning signs, types of delays encountered, male involvement, access to funds, and perception of the quality of care. The interviews were performed by the bedside when patients became clinically stable enough to respond to the questions. The responses were documented in writing and not recorded because some of the study participants were not comfortable with the use of a tape recorder and did not give consent to that effect. The narrative analysis was carried out manually. The study protocol was approved by the Ethics and Research Committee of the hospital. Dependent variables included maternal outcome categorized into three groups: normal pregnancy, acute maternal morbidities, and near misses: Independent variables included sociodemographics (maternal age, maternal education, marital status), obstetric (parity, gestational age at delivery, antenatal care attendance, complications noted during booking visit, referral status, fetal presentation during labor), and perinatal characteristics (low birth weight, birth asphyxia, stillbirth). Statistical analysis for the quantitative aspect was performed using STATA version 12. The differences in the proportion of women with normal pregnancy, acute maternal morbidities, and near misses with specific characteristics were compared using a chi-square test at a 5% level of statistical significance. Multivariate analysis was done using a generalized ordered logit model with maternal event—normal pregnancy, acute maternal morbidity, and near miss—as the outcome. This was used to generate the estimates of partial proportional odds ratios across the categories of the outcome variable: (a) any maternal morbidity (acute maternal morbidity and near misses) versus normal pregnancy and (b) near misses versus other pregnancy outcomes (acute maternal morbidity and normal pregnancy). The generalized ordered logit model works well in situations where the proportionality or parallel slopes assumption of ordinal logistic regression is violated (Williams, 2005). This assumption was assessed during the preliminary analysis using the omodel test. The test showed a violation of the proportionality assumption of odds across response categories (χ2 = 80.42, p <.001): hence, ordinal logistic regression could not be used for the analysis. The “gologit2” command was used in STATA to fit the generalized ordered logit model. The partial proportional odds ratios and 95% confidence intervals are reported.
N/A
