Experiences of women seeking care for abortion complications in health facilities: Secondary analysis of the WHO Multi‐Country Survey on Abortion in 11 African countries

Despite evidence of acute and long‐term consequences of suboptimal experiences of care, standardized measurements across countries remain limited, particularly for postabortion care. We aimed to determine the proportion of women reporting negative experiences of care for abortion complications, identify risk factors, and assess the potential association with complication severity.

tion, societal stigmatization, and harassment were found to be barriers to PAC uptake. 6 Health system constraints and patient socioeconomic disadvantages have also been identified as drivers of negative experiences of care. Research suggests adolescents are less likely to be treated respectfully in maternity care and PAC and may delay care to avoid provider bias and discrimination. 12,13,17 Women of low socioeconomic status (SES) were found to have increased risk of unnecessary obstetric procedures, discrimination, and stigmatization from providers. 8 This effect was heightened when intersections with religious, ethnic, or racial minority status were considered. Facility environments can also obstruct practices; analyses of mistreatment during childbirth in Kenya found an association between facility infrastructure and observed verbal abuse, unhygienic practices, and lack of consent. 12 This secondary analysis of the WHO Multi-Country Survey on Abortion (WHO MCS-A) had three objectives: (1) to estimate proportions of women reporting positive and negative experience of care for abortion complications in 11 African countries; (2) to identify risk factors associated with negative experience of care; and (3) to examine the association between negative experience of care and abortion complication severity.

| MATERIAL S AND ME THODS
Data were sourced from phase one of the WHO MCS-A in Benin, Burkina Faso, Chad, the Democratic Republic of the Congo, Ghana, Kenya, Malawi, Mozambique, Niger, Nigeria, and Uganda. The study protocol has been published, detailing recruitment and collection methodologies. 18,19 Participating countries, provinces, and facilities were identified with multistage sampling. Facilities had the following characteristics: more than 1000 deliveries per year, a gynecology ward, and surgical capability (defined as providing the signal functions for comprehensive emergency obstetric care, which includes removal of retained products of conceptus and surgical capability and, if available, abortion provision and/or postabortion care). Data collection occurred between February 2017 and April 2018 via medical record extraction and a facility assessment, typically overseen severity on experience of care appeared significant, such that women with moderate and severe complications had 12% and 40% higher odds of reporting negative experiences, respectively.
Conclusion: There were widespread reports of negative experiences of care among women receiving treatment for abortion complications in health facilities. Our findings contribute to the scant understanding of the risk factors for negative experiences of postabortion care and highlight the need to address harmful provider biases and behaviors, alleviate health system constraints, and empower women in demanding better care.

K E Y W O R D S
abortion complications, experience of care, quality of care, respectful care, Sub-Saharan Africa by coordinating obstetrician/gynecologists or midwives, and an exit survey in the form of an audio computer-assisted self-interview (ACASI). Medical records of patients with signs and symptoms consistent with complications related to spontaneous and induced abortions who presented at selected facilities were eligible for inclusion.
A sample of eligible women admitted for at least 24 h were invited to participate in the ACASI at discharge after informed consent.
Experience of care was measured as a composite variable of eight questions from the ACASI. 7,14 Dimensions were comparable to themes identified by WHO and established measures of disrespectful maternity care. 7,11,13,14 Six of the questions were yes/no responses and two were a five-point Likert scale ranging from very satisfied to very dissatisfied, regrouped into satisfied/dissatisfied with neutral included in dissatisfied. Questions were coded such that a negative answer (i.e. no, dissatisfied) constituted a negative experience.
We categorized risk factors as sociodemographic, clinical, facility, and geopolitical. Sociodemographic factors included age group, marital status, and highest education level, obtained from medical records, and household SES and economic independence from the ACASI. Household SES was a composite of the presence of running water in the participant's home and sufficiency of household income the previous month was to cover food, health needs, and savings.
Economic independence was determined if the participants self- F I G U R E 1 Flow diagram for inclusion and exclusion criteria for analytical sample. a WHO maternal near-miss criteria (organ dysfunction of either one or more of the following: cardiovascular, respiratory, renal, coagulation, hepatic, neurologic, or uterine dysfunction). b WHO potentially life-threatening conditions (severe hemorrhage, severe systemic infection, or suspected uterine perforation). c Moderate complications (heavy bleeding, suspected intra-abdominal injury, or infection). d Mild complications based on abnormal physical examination findings on initial assessment (vital signs, appearance, mental status, abdominal examination, gynecological examination) Bivariate analyses were performed for individual-, facility-, and country-level factors, and cross-tabulated with women's negative experience of care, construed as a binary variable (i.e. yes reported for at least one negative experience versus no reports). We conducted crude analyses to assess the association of each factor with reported experiences of care, using generalized estimating equations to account for clustering by facility. Multivariate regression analyses evaluated the strength of association between each factor and reported experience of care, after adjusting for age group, household SES, and country. One-way ANOVAs were used to identify statistically significant differences in the mean number of negative care report counts within risk factors. A Tukey post-hoc test was used to determine the directionality of between-group differences.
Multiple regression determined the strength of association between complication severity and negative experience of care, of particular interest after initial analyses of the WHO MCS-A found gaps in care for the subset of women with severe complications. 19 A generalized estimating equations model was fitted, adjusting for country and any risk factor that led to a change of 10% or more in the association between complication severity and negative experience of care.  Table 1. Overall, 62% (n = 1821) reported at least one incident of poor experience of care, of whom 88% (n = 1598) reported this for 1-4 questions and 12% (n = 223) for 5-8 ( Figure 2). The question with the highest percentage of responses for negative experience of care was: "Were you able to ask questions during the examination and treatment?" (n = 1009, 34%).
Descriptive statistics for the study population are given in Table 2. Participant age ranged from 12-50 years, with a mean age of 27 years. Most participants reported that they were married or cohabitating (75%, n = 2116), with secondary or higher-level education allowed abortion "to preserve health," and 48% (n = 1412) in countries that permitted abortion "in certain cases" (i.e. less restrictive).
Estimates of the association between risk factors and negative experiences of care are given in Table 2 As determined by one-way ANOVAs, we found significant dif-

| DISCUSS ION
We found that three in five women reported at least one negative experience of care for facility-based treatment of abortion complications, suggesting that quality of PAC is a major public health concern in participating African countries. Consistent with previous studies, women who were adolescents, single, of lower household SES, and who were economically dependent were more likely to report negative experiences of care. 5  reports consultancy fees to her institute received from WHO. SG reports institutional funding from WHO for data collection in this study.
VF reports WHO funding to support data analysis in this study.