Factors associated with screening for preeclampsia in the Democratic Republic of Congo: evidence from a nationwide cross-sectional study

Delays in identication of women with preeclampsia is a contributing factor to maternal deaths in developing countries. We conducted this study to assess the prevalence of screening for preeclampsia in the Democratic Republic of Congo (DRC), and to identify factors associated with the screening. We used data from the 2017-2018 DRC Service Provision Assessment survey, a nationwide survey that samples services, including antenatal care (ANC), in representative health facilities. A pregnant woman was deemed screened for preeclampsia if she had received the two following services: blood pressure measurement and assessment for warning signs of preeclampsia. We used multiple logistic regression, with generalized estimating equations, to identify factors associated with screening for preeclampsia.


Abstract
Background Delays in identi cation of women with preeclampsia is a contributing factor to maternal deaths in developing countries. We conducted this study to assess the prevalence of screening for preeclampsia in the Democratic Republic of Congo (DRC), and to identify factors associated with the screening.

Methods
We used data from the 2017-2018 DRC Service Provision Assessment survey, a nationwide survey that samples services, including antenatal care (ANC), in representative health facilities. A pregnant woman was deemed screened for preeclampsia if she had received the two following services: blood pressure measurement and assessment for warning signs of preeclampsia. We used multiple logistic regression, with generalized estimating equations, to identify factors associated with screening for preeclampsia.
There was a regional variation in the screening for preeclampsia, with the lowest prevalence observed in the Central (10.7%) and the highest in the Eastern provinces (22.5%).

Conclusion
Our study shows that screening for preeclampsia during ANC visits in DRC is not systematic, with disparities between provinces. We found that a low schooling level, a low number of ANC visits, rural areas, province of residence, and a high volume of ANC visits at the facility were associated with a low screening for preeclampsia. These factors should be considered for setting up strategies in order to increase the screening for preeclampsia during ANC visits in provinces of DRC with too low screening. Background Page 3/21 Hypertensive Disorders in Pregnancy (HDP) including preeclampsia are the leading cause of direct maternal deaths worldwide (1). Globally, preeclampsia affects 2 to 8% of pregnancies annually, and is responsible for over 50 000 maternal deaths (2). Most of deaths occur in low-and middle-income countries (LMICs) and are however preventable. A contributing factor in these deaths is a delay in identifying preeclamptic women (3)(4)(5)(6)(7).
However, a study from six sub-Saharan Africa countries indicated that only 31% of ANC attendees are screened for preeclampsia (21).
According to the World Health Organization (WHO), maternal mortality in the DRC is estimated to be 473 deaths per 100 000 live births (22), despite a high rate of both skilled attendance at delivery (80%) and antenatal care attendance (89%) (23). The contrast between the high maternal mortality rate and the high antenatal service coverage may indicates that the majority of antenatal attendees miss screening for clinical conditions, such as preeclampsia, known to result in poor birth outcomes (24, 25).
According to the DRC's National Health Information System (NHIS), 3379 (62%) of the 5459 cases of preeclampsia reported in DRC in 2019 developed eclampsia (28), the major cause of deaths due to HDPs.
Studies focusing on the screening for preeclampsia in DRC are scarce. A previous study assessed the screening for HDP in Kinshasa, the capital city of DRC, by combining proteinuria testing, blood pressure measurement, and a comprehensive maternal history as a screening package. The study indicated that only 26.7% of pregnant women were screened for HDP (25).
A services provision assessment survey was conducted in DRC (DRC-SPA) from 2017 to 2018 to evaluate the quality of healthcare services nationwide (13).
Data from the DRC-SPA survey have been used to assess the provision of individual screening tasks for preeclampsia including blood pressure measurement and inquiring about warning signs of preeclampsia.
However, no study to date has assessed the provision of these individual tasks during ANC visit as a screening package nationwide in DRC. Our study therefore aims to ll this gap by using data from the DRC-SPA survey to assess the frequency of screening for preeclampsia nationwide in DRC and to identify factors associated with the screening.

Data source
We used data from the 2017-18 DRC-SPA survey, a nationwide facility-based cross-sectional study. The survey was implemented by the Kinshasa School of Public Health (KSPH) with technical assistance from Inner City Fund International through funding from the United States Agency for International Development (USAID).

Sample size and sampling procedure
The sampling process for the DRC-SPA survey has been described in details elsewhere (26). Brie y, the survey sampled 1412 health facilities of which 32 could not be reached due to their location in areas where there were still armed con icts. As a result, 1380 health facilities were surveyed, of which 1286 offered an ANC visit during the study period. In each health facility, ve women per ANC provider were included in the survey with a maximum of 15 women per facility. The nal sample size was 4512 women. As preeclampsia rarely presents prior to 20 weeks' gestation (27), we excluded women with a gestational age less than 20 weeks (n=749). As a result, 3763 pregnant women were included in the analytic sample. These women received care from 1191 providers in 864 health facilities ( Figure 1). Data The survey used four data collection tools: 1) facility inventory questionnaire; 2) healthcare provider interview questionnaire; 3) checklist for direct observation of services for sick children, family planning, and ANC; and 4) exit interview questionnaires for caretakers of sick children, for women receiving ANC and for women receiving family planning whose consultations were observed.
For each pregnant woman, a data collector observed the provider conducting ANC consultation and recorded observations using the ANC observation checklist. Speci cally, the data collector observed whether the provider measured blood pressure, tested or referred the woman for proteinuria testing, or asked about any preeclampsia-related warning signs, including severe headaches or blurred vision, and swollen face or hands. The data collector also observed whether the woman herself mentioned any of these preeclampsia-related warning signs to the provider.
At the completion of ANC consultation, data collectors interviewed the women to collect sociodemographic and reproductive data. Data collectors also collected clinic-level data such as the type of clinics (private or public), location (urban or rural), as well as data regarding the quality of health providers who attended pregnant women (speci cally, whether they had received in-service training in the management of HDP or not).

Operational de nitions
We used the following de nitions:

Preeclampsia screening package
The entire preeclampsia screening package consisted of 1) inquiring about preeclampsia-related warning signs, and 2) blood pressure measurement (21,28).
Inquiring about preeclampsia-related warning signs ANC provider asked about or the woman mentioned any of the following conditions for current pregnancy: headaches or blurred vision, swollen face or hands.
Woman screened for preeclampsia A pregnant woman who had received the entire preeclampsia screening package during the ANC visit.

Data processing and management
For the purpose of analyses, we merged four datasets (facility inventory, provider interview, direct observation of ANC consultation, and exit interview datasets) into one dataset by using the provider and facility identi ers of each dataset. The unit of analysis of the merged dataset was individual women.
Secondary and tertiary health facilities were grouped into one category named "hospitals", as they are all used as referral by primary health centres (PHCs).

Data analysis
Data were analyzed using Stata 14.1 software (StataCorp, College Station, Texas). All analyses were weighted using sampling weights provided in the DRC-SPA datasets. Continuous variables were summarized using weighted mean and standard deviation (±SD) if normally distributed, and median and interquartile range (IQR) otherwise. Categorical variables were summarized using weighted proportions.
Screening for preeclampsia was our major outcome. We used generalized estimating equations (GEE) method while tting a logistic regression model to adjust for correlation among pregnant women of the same clinic. All variables with a P-value less than 0.25 in univariate analysis were included in the multivariable analysis.
Explanatory variables included characteristics of women (schooling level, gestational age, number of previous ANC visits for the current pregnancy, gravidity, province of residence), provider characteristics (quali cation, whether he/she received in-service training on ANC screening within the previous 24 months), and facility characteristics (location, facility type, facility ownership, mean number of pregnant women attended per day). Quantitative independent variables, such as the mean number of ANC visit per health facility per day, were standardized to avoid scale dependence. If a strong correlation was noticed between two independent variables, one amongst the two was eliminated to avoid multicollinearity. We checked multicollinearity among independent variables by estimating the variance in ation factor (29). All variance in ation factor values were less than 10. Therefore, no multicollinearity was observed. The statistical signi cance level was set to 0.05. There were provincial disparities in the screening for preeclampsia, with the lowest percentage of women screened in provinces such as Maniema (8.0%); , and Bandundu (8.7 %), and the highest percentage in Nord Kivu (38.3%); (Figure 2).

Characteristics
Overall there was no difference in screening for preeclampsia in public compared to private facilities, and no difference between PHCs and hospitals. However, in Katanga and Bandundu, screening for preeclampsia was higher in private compared to public facilities. A higher percentage of women were screened in hospitals compared to PHCs in Katanga Province (Table 3).

Factors associated with screening for preeclampsia
In multivariable analyses, the percentage of women screened for preeclampsia was signi cantly higher among women with secondary level of schooling or above (adjusted odds ratio [AOR] = 1.21; 95% CI, 1.04-1.42; P=0.016) and among women having a subsequent ANC visit (AOR = 1.39; 95% CI, 1.20-1.62; P<0.001). It was also higher among those receiving care in facilities with a low number of ANC visits (AOR = 1.28; 95% CI, 1.09-1.51; P=0.003) and in urban settings (AOR = 1.36; 95% CI, 1.02-1.82; P=0.036 There was a regional variation in the screening for preeclampsia, with the lowest prevalence observed in the Central (10.7%) and the highest in the Eastern provinces (22.5%). (Table 4).

Discussion
We conducted the current study to assess the prevalence of screening for preeclampsia and to identify factors associated with the screening nationwide in DRC. Our study revealed low provision of screening for preeclampsia in DRC, with only 19.1% of women fully screened. Screening for preeclampsia was signi cantly higher in women with at least secondary level of schooling, in women attending a rst ANC visit, or in those receiving care in a facility with a low number of ANC visits, or among women living in an urban area.
Among women in whom screening was not completed, about one-quarter did not have blood pressure measured. However, failure to inquire about preeclampsia-related warning signs was a more common gap in quality; it was not performed in nearly three-quarters of women. This appears to be a common concern in other Sub-Saharan African countries such as Rwanda (70%), Ethiopia (62 %), and Madagascar (53%) (21).
Preeclampsia-related warning signs can be ascertained by a simple anamnesis of a pregnant woman during ANC visit (17). The failure to assess women for preeclampsia-related warning signs in DRC may be a consequence of providers' knowledge gaps and/or lack of motivation or supervision, but the point needs further investigations.
Although heavy ANC workload could be another explanation for this failure, this is unlikely given that health facilities included in our study had relatively low volumes of ANC attendees on the day of the survey, as half of them received less than 8 ANC visits per day.
Our study found a low provision of urine testing for proteinuria (19.8%). Unlike the WHO recommendations, the current national ANC guidelines in DRC do not recommend systematic urine testing for proteinuria during ANC visits. The guidelines recommend proteinuria screening in women with warning signs of preeclampsia (30). As preeclampsia can present with isolated proteinuria (31,32), it is essential that DRC move towards universal urine testing for proteinuria at every ANC visit.
The proportion of women screened for preeclampsia in DRC is similar to ndings from Rwanda (22%), but lower than the range of 24% to 55% reported in other developing countries such as Madagascar, Tanzania and Ethiopia (21). We found disparities in screening for preeclampsia between provinces, with the highest prevalence in Nord Kivu (38.4%), and the lowest in Maniema (7.9%). We speculate that inequity in the distribution of international agencies involved in maternal health in DRC, with Nord Kivu having the highest number of donors (33), may result in disparities in the provision of health services including screening for preeclampsia. However, additional information is needed to further understand the disparities in screening for preeclampsia between provinces. Information on ANC providers' motivation and satisfaction or the quality of their supervision would be helpful next steps to contextualize results found in this analysis.
According to a previously published study, only 9.1% of health facilities in DRC provide quality emergency obstetric care services, including parenteral administration of magnesium sulfate (34). In the setting of scarce emergency care, universal screening for preeclampsia may be of paramount public health importance because early identi cation of women with preeclampsia allows timely referral of the woman to a facility with appropriate care. However, our ndings indicate that women with ANC visits in rural settings were less likely to be screened than those in urban areas. This nding is in line with other studies reporting a low provision of ANC services in rural compared to urban settings (35). Furthermore, our ndings indicate that women attending a subsequent ANC visit are more likely to be screened than those attending a rst visit. As a result, women with preeclampsia may not be identi ed at an early stage of pregnancy. In DRC, a number of pregnant women have only one ANC visit (36,37). For these women, the rst ANC visit may also be the last one. Therefore, ANC providers should provide all recommended screenings, including screening for preeclampsia, without waiting for the next visit which may not take place.
Screening for preeclampsia was signi cantly higher among women with secondary or higher level of schooling than those with primary or no level of schooling. This nding is consistent with others studies that reported a high provision of ANC services among more educated women (38,39). It is possible that health providers discriminate against women with low level of schooling by not providing all the required antenatal services (39,40). Moreover, women with high level of schooling may be more knowledgeable about obstetric danger signs including warning signs of preeclampsia and are thus more likely to mention them during ANC visits than women with lower level of schooling (41).
The prevalence of screening for preeclampsia was signi cantly lower in facilities with higher volumes of ANC visits in a given day. As a result, researchers have suggested task shifting as an alternative to improve screening for maternal conditions, including preeclampsia (42).
The low provision of screening for preeclampsia in DRC may also be due to lack of consistent national guidelines on the prevention of preeclampsia, which can subsequently induce knowledge gaps among ANC providers. As in other African countries, the current national guidelines in DRC offer little guidance on preventing preeclampsia in ANC services (30,43). This low provision of screening for preeclampsia in DRC may result in missed opportunities to identify women with preeclampsia as early as possible to prevent adverse maternal and perinatal outcomes. With such low provision of screening, it is not unexpected that preeclampsia is still a public health problem in DRC (29,44). This study has provided information on the factors that should be addressed in strategies aimed at improving screening for preeclampsia in DRC.
The strength of the present study was the use of a comprehensive national dataset, suggesting that our ndings accurately re ect the current situation regarding the quality of screening for preeclampsia in DRC.
The study also has some limitations. Because screening for preeclampsia was assessed by direct observation, health providers might have made an extra effort to give their best quality service at the time when the research team visited the clinic. This Hawthorne effect was mitigated by having up to ve observations per provider, which may have helped to reduce providers' awareness of the presence of the data collector.

Conclusion
Our study shows that screening for preeclampsia during ANC visits in DRC is not systematic. The gap in the quality of screening was in both blood pressure measurement and assessing women for warning signs of preeclampsia. We found that a low schooling level, a low number of ANC visits, rural areas, province of residence, and a high volume of ANC visits at the facility were associated with a low screening for preeclampsia. These factors should be considered for setting up strategies in order to increase the screening for preeclampsia during ANC visits in provinces of DRC with too low

Declarations Ethics approval and consent to participate
The DRC-SPA survey was approved by the institutional review board of the Kinshasa School of Public Health. We obtained permission to use the survey data from the Measure Demographic Health Survey Program. Further ethical approval was not required, as our research is a secondary data analysis of publicly available datasets that do not include any identi able information that links to the survey participants.

Not applicable
Availability of data and material The DRC-SPA data is publically available from the DHS website: http://dhsprogram.com/data/availabledatasets.cfm

Competing interests
The authors declare that they have no competing interests.
Funding None Authors' contributions DNM and AR designed the research; DNM analyzed data, and drafted the manuscript. AR, GW, and PB critically revised the drafted manuscript. All authors read and approved the nal manuscript.

19.
Khooshideh      Weighted proportion of pregnant women screened for preeclampsia during antenatal care visit in DRC, by province A pregnant woman was considered as screened for preeclampsia if she received a blood pressure measurement, and an assessment for warning signs of preeclampsia.