Background
The maternal mortality ratio (MMR) in Haiti is estimated to be 359/100,000 live births [
1], and while this represents a 43% decrease from the 1990 estimate of 625/100,000 live births, it remains the highest MMR in the Latin American and Caribbean region and on par with several countries in sub-Saharan Africa. Improvements in access to and quality of intrapartum care, specifically basic and comprehensive emergency obstetrical care, will lead to the largest reductions in maternal death. Both high quality antenatal care and postpartum care with an emphasis on identifying complications and offering family planning methods to achieve healthy birth spacing, can contribute to reductions in maternal deaths [
2].
The majority of pregnant women in Haiti receive at least one antenatal visit (90%), however far fewer access the recommended 4 visits (67%) [
3]. Anecdotal evidence suggests that care is variable in terms of service components and quality, however there are no systematic data addressing quality of antenatal care in Haiti. Standardized, high quality antenatal care could potentially influence a pregnant woman’s satisfaction with care, her knowledge of danger signs in pregnancy, her decision to seek a facility based delivery and her knowledge of available postpartum family planning methods [
4]. Much less is known about the quality and uptake of postnatal care. Only 30% of women and only 19% of infants received any postnatal care in the first 2 days after delivery [
3]. Little is known about the services provided at these visits and there are no current estimates of postnatal care uptake at the recommended 1-week or 6-weeks visits.
In order to ensure comprehensive, high quality antenatal and postnatal care, it is imperative to characterize the current quality of service delivery in the realms of antenatal and postpartum care and to better understand the gaps in completeness of care in accordance with the World Health Organization (WHO) recommended guidelines [
5]. To address this, a study assessing quality and uptake of antenatal and postnatal care in large health facilities in three departments within Haiti was conducted.
Methods
Study sites
All public and private clinics who provide antenatal services to approximately 100 women per week within three target areas of Haiti (Nord department, Nord-Est department, and the communes of St Marc, Verrettes, and Petite Rivière within the Artibonite department– referred to as the Artibonite sub-region for the remainder of this manuscript– were included in the study. These regions were conveniently selected and represent the regions where most CDC provided technical assistance related to maternity care has occurred in Haiti. The approximate number of women seen weekly was estimated using data from the publicly available Monitoring Evaluation and Surveillance Intégrée (MESI). Ten health facilities met the above criteria and were included in the study (2 in Nord department, 3 in Nord-Est department, and 5 in Artibonite sub-region). These ten studies represent an exhaustive sample of antenatal care clinics servicing 100 women per week or more in our three selected regions.
Data collection
Three data collection methods were used in this study 1 exit interviews with women seeking antenatal or postnatal care, 2) antenatal chart abstraction and 3) provider knowledge assessments. This study uses data from these three methods to triangulate quality of antenatal and postnatal care. Each method is described below.
Exit interviews
Exit interviews were conducted in Haitian Creole with pregnant or postpartum women as they exited the facility after their antenatal or postnatal care visit. Four days were spent at each of the Nord and Nord-Est department health facilities (except one health facility in Nord-Est where the clinic is open only 2 days per week) and 3 days were spent at each of the Artibonite health facilities between March-April, 2015. This duration was deemed sufficient to capture a sample of approximately 400 women seeking antenatal care from each department (Nord, Nord-Est, and Artibonite sub-region) for a total of 1200 women. The sample size of 400 women per department was calculated based on an anticipated 10% change in services provided from baseline to endline, assuming a baseline frequency of 50%, a power of 80%, a 95% confidence interval and an anticipated non-response rate of 3%. The sample size of 400 women within each department was distributed population proportional to estimated size among all included health facilities, resulting in a representative sample of women seeking antenatal care from large health facilities by department. All women who were seeking antenatal or postnatal care services on study days were approached for inclusion and nearly the same number of women recorded in the antenatal or prenatal registrar were approached on any given day. Interviews took approximately 10 min to complete and questions included standardized information on demographics, previous pregnancies, and current pregnancy, with a focused section on services and satisfaction regarding the visit they had just completed. Data was entered directly onto Samsung Galaxy 10® tablets (Seoul, South Korea) using Open Data Kit (ODK; open-source). Data were transferred off tablets onto a password protected centralized computer daily. Data were converted to a comma separated value (csv) file using ODK briefcase 10. No personally identifiable information was collected and only women who gave verbal consent were interviewed.
Chart abstraction
A sample of individual patient charts was obtained by acquiring the antenatal register used 9 months prior to the data collection period and compiling a list of women seen for antenatal care services starting on the first day of the month and continuing sequentially for 1–3 days of chart abstraction per facility. A minimum of 100 antenatal charts were attempted from all health facilities, however the difficulty and time required to abstract information from a patient chart varied by health facility as a result of differences in chart recording practices and chart storage practice. This number was selected to ensure at least 400 charts from each region were abstracted to approximately match the antenatal care exit interview sample size. From the most recent pregnancy identified in the patient chart, the total number of antenatal care visits was recorded on a paper form which was then entered daily onto a Samsung Galaxy 10 tablet using ODK. Data were transferred off tablets onto a password protected centralized computer daily. Data were converted to a csv file using ODK briefcase 10. Lab-only visits and result-return visits were not included in the total count.
Provider knowledge assessment
Knowledge questionnaires were administered to all providers who delivered antenatal and/or postnatal care to women during any of the site visit days. The number of providers varied by facility (n = 2–6). Verbal consent was obtained and a questionnaire written in Haitian Creole was completed by the provider. No personally identifiable information was collected. Provider assessments were scored by a single study staff member and data was entered directly into an excel database.
Data analysis
All analyses were performed using survey procedures in SAS Version 9.3 for Windows (SAS Institute Inc.; Cary, North Carolina; USA). Answers of “I don’t know” were recoded to “no”, except where specified. Weighted data are presented as unweighted frequencies and weighted percentages, except where specified.
Antenatal care (ANC)
Weights for overall antenatal exit interviews were calculated based on the number of days of data collection at a facility and the response rate within each department. For family planning questions, spontaneous answers were recorded. Patients missing observations for any outcome were excluded from all analyses; to determine how those patients who were excluded differed from the included sample, chi-square tests and Fisher’s exact test for categorical outcomes and t-test for continuous outcomes were conducted to assess for significant differences between the two groups.
A total of 10 counseling messages were expected during each visit. A point system was applied to counseling messages such that receipt of each message provided a single point. Univariable and multivariable logistic regression were used to examine determinants of reporting having received at least five of ten standard counseling services during the antenatal care visit. Multivariable models were adjusted for demographics with a known impact on quality of antenatal care services including age, marital/co-habiting status, and education. Prenatal visit number and gestational age were included to assess how repeat visit or gestational age may impact the delivery of recommended counseling messages. Parity was included to assess whether women attending care for a later pregnancy may receive different messaging, and department where services were received (Nord, Nord-Est, Artibonite) was included to account for potential differences in antenatal care across the country.
Postnatal care (PNC)
Weights for postnatal exit interviews were calculated based on the number of days of data collection at a facility and the response rate within each department. Patients missing observations for any outcome were excluded from all analyses; to determine how those patients who were excluded differed from the included sample, chi-square tests and Fisher’s exact test for categorical outcomes and t-test for continuous outcomes were conducted to assess for significant differences between the two groups.
Antenatal care chart abstraction
A total of 1,411 charts were abstracted from the ten study facilities. Weights for the charts were calculated based on the number of women seeking antenatal care in each facility.
Provider knowledge questionnaire
An exhaustive survey of providers was conducted in each of the facilities, however only those providers who reported delivering antenatal care were analyzed for the questions on antenatal care knowledge; similarly, only those providers who reported delivering postnatal care were analyzed for the questions on postnatal care knowledge. One provider who did not specify which services he or she provided was excluded from all analyses. All questions were scored with correct answers receiving one point and missing answers were marked as incorrect. All presented data are unweighted due to the small sample size, and thus the inability to make inference on these findings.
Discussion
Results from a sample of pregnant and postpartum women seeking care at large health facilities in three departments in Haiti suggest underutilization of both antenatal and postpartum services. Additionally, while routine antenatal clinical services, such as blood pressure checks, were systematically provided, education messages were reported to occur far less frequently and knowledge of key life-saving measures, such as danger signs of pregnancy and postpartum contraceptive methods was low among pregnant and postpartum women. Uptake of antenatal care has been linked to a reduction in maternal morbidity and mortality and improvement in neonatal survival [
6,
7]. Antenatal care represents a point of contact between providers and pregnant women at which key antenatal services and educational messages may be given [
8]. Antenatal care components can identify potential complications in pregnancy and counseling can help women recognize danger signs of pregnancy, understand the benefits of facility deliveries, and the importance of healthy spacing of pregnancies through the use of postpartum contraception [
9]. Likewise, postpartum care is an opportunity to identify complications after delivery, and to provide health promotion messages and preventive services, such as family planning methods to prevent a rapid repeat pregnancy and improve both maternal and newborn survival [
10]. The low level of delivery and or retention of educational messages among our sample participants is concerning and emphasizes the importance of standardizing and delivering quality antenatal care and measuring quality indicators alongside quantity of antenatal care visits [see ref in comment] [
11].
Review of 1,141 antenatal charts of recently delivered women demonstrated that 27% of our sample attended only a single antenatal care visit and 47% attended two visits or less. Furthermore, only 36% of women attended the recommended 4 antenatal care visits. This is far less than the national estimate of 67% of women who report attending at least 4 antenatal visits [
3]. While it is possible that some women may have attended a portion of their antenatal care visits at another facility, therefore not recorded in the charts reviewed, this number is likely to be low. Alternatively, regional differences in antenatal care uptake or desire to report higher uptake in the national Demographic Health Survey may explain some differences. It is both a global and national recommendation that pregnant women deliver in a health facility with skilled care, yet the overall facility delivery rate in Haiti is low at 36% [
3]. Attendance at ANC is a determinant of a facility delivery [
4], and under-utilization of antenatal care, even among women who do have a facility delivery, has been shown to increase the risk of poor pregnancy and neonatal outcomes [
6]. Further efforts to increase uptake of antenatal care and examine barriers to seeking care among should be considered.
The majority of pregnant women seeking antenatal care received the recommended clinical services such as HIV testing, blood pressure, weight checks and fetal viability assessments. In contrast relatively few women received adequate counseling messages. For example, 44% reported receiving education on danger signs of pregnancy and 18% on the importance of birth spacing for maternal and newborn health. Antenatal care is an opportunity to provide services, screen for complications, and counseling at multiple time points during pregnancy [
2,
4]. Components of counseling should include education on danger signs of pregnancy, birth preparedness, breastfeeding, and the importance of postpartum family planning for birth spacing A total of 10 counseling messages was expected at each visit. Among our sample of women seeking antenatal care, only 5% reported receiving all 10 messages and 7% reported they received none of the messages, indicating that a critical opportunity to provide key health messages to pregnant women was missed. An alternative explanation to the low report of counseling messages may be a lack of understanding or generalized low health literacy among the study population. This hypothesis is supported by our findings which show women of higher educational attainment and women who were at a repeat visit (compared to women attending their first antenatal care visit) were more likely to report receipt of 5 or more out of 10 counseling messages. It is possible that higher education and/or repeat exposure to the counseling messages may aids in understanding and suggesting that implementation of simple scripted counseling messages that could be repeated at every antenatal visit may improve comprehension and retention. Interestingly 28% spent ≤ 5 min with their provider which may be another contributing factor to the low report of counseling received. While time with the providers was low, clinic wait times were very high (26% waited >4 h) therefore providing scripted messages with a lay counselor in the waiting area may also improve retention and comprehension.
A woman’s ability to achieve recommended birth spacing can improve both maternal and neonatal outcomes [
10]. This is best achieved by adopting a family planning method in the immediate postpartum period. Use of family planning postpartum among many Haitian women is low where only 23% of women reporting use of any family planning method in the first months, yet only 2% of postpartum Haitian women desire subsequent conception in the next 2 years after birth [
12] and desire among recently delivered women living in Northern Haiti to adopt a family planning method postpartum is high [
13]. Our study demonstrated low attendance at postpartum care visits highlighting the need to discuss postpartum family planning during the antenatal care period. Limited knowledge of postpartum family planning methods by both women and providers were also observed. Provider knowledge and comfort with family planning methods can influence uptake of family planning methods by women [
14]. A concerted effort towards improved provider knowledge and comfort with family planning methods postpartum could be explored as a strategy to meet the known demand for delaying next birth among Haitian postpartum women.
This study is the first to characterize antenatal and postpartum care in Haiti through a quality of care lens. Our methodology allowed us to triangulate findings from several sources, including providers, patients and hospital records, in order to strengthen the validity of our findings, however, our study has several limitations. First we limited our study to large health facilities, limiting the generalizability of our findings. Second, we sampled women seeking antenatal and postnatal care at facilities located in 3 departments within Haiti and therefore findings may not be nationally generalizable as these departments were not chosen at random. Third, we surveyed patients immediately after receiving care, which minimizes recall bias, however may have increased the likelihood of desirability bias among patients. Fourth, antenatal care visit numbers abstracted from charts may be lower than recommended as a result of preterm delivery or through seeking antenatal care at multiple facilities, neither explanation this study was designed to address. Finally, although our provider sample was small, it was an exhaustive sample and therefore represents the current knowledge held by antenatal care providers at large hospitals in the three departments, but is not generalizable to the larger national population of providers.
Acknowledgments
We would like to thank the following field supervisors for their invaluable oversight of the study team: Lesly Andrecy, MD; Gregeneve Beauzille, MD; Adam Gedeon, MD; Sherlie Jean-Louis, Ellie Joseph, MD; Samson Marseille, MD. We would also like to thank all the enumerators for contribution during data collection. Finally, we would like to thank Edy Joseph for his translation services and project coordination. This research was supported in part by an appointment to the Research Participation program at the CDC administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the US Department of Energy and the CDC.