Pregnant women follow‐up service, Shewa, Ethiopia

Abstract Background and Aims The goal of this study was to demonstrate the effects of factors related with time to developing pre‐eclampsia (PE) among pregnant women follow‐up service at Arerti Primary Hospital. Methods A survival analysis was employed on a pregnant women's follow‐up service from September 2018 to June 2019 at the Arerti Primary Hospital. A closed‐form sample size formula for estimating the effect of the time‐to‐event data was used. Both the descriptive method and Cox proportional hazards model were applied to compute the research survival data. Results Using the Kaplan–Meier estimation technique, the univariable analysis shows that the survival time median is 7 months and 3 weeks. The graph of Kaplan–Meier estimate of total survival functions indicates a decreasing pattern of survivorship function. We used the Kaplan–Meier estimates to investigate the effects of observed differences among different categories of the factors, we applied the Log‐rank test. The final survival model outcomes weight, marital status, age, history of PE, and multiplicity were related to a substantial hazard of evolving PE. Conclusion On the basis of our final survival model results, we recommended that all pregnant women having such risk factors should see a health care professional and control their medical condition before and during pregnancy. Advising women about proper body weight in each follow‐up period is supported. Finally, health experts should advise pregnant women about potential risk factors related to PE.

In 2017, two-thirds of all maternal deaths occurred in sub-Saharan Africa, where the maternal mortality ratio was 542 deaths per 100,000 live births. Although this region has achieved significant progress in lowering maternal mortality since 2000, maternal mortality is still almost 78 times higher than in Australia and New Zealand, which has the lowest ratio of any region. Major efforts are needed to bring maternal mortality under 70 deaths per 100,000 live births by 2030, as prescribed by Sustainable Development Goals 3.1. 8 PE remains a global problem. 9 It is common problem in developing countries because of illiteracy, poor antenatal care (ANC) follow-up, and lack of health awareness. 10 In Ethiopia access to health care is limited, PE is a leading cause of maternal mortality, with estimates of 16% maternal deaths per year. 11 Despite preventive measures such as the expansion of health facilities, maternal waiting homes, and training of health professionals are undertaken by the government of Ethiopia to reduce maternal and perinatal mortality, maternal and perinatal mortality related to PE is still on an increase. [12][13][14] This indicates that there are several risk factors associated with their occurrence and the survival of this disease. Therefore, this study attempted on assessing the risk factors on time to develop PE among pregnant women follow-up service.

| Study area
The research area is located in Minjar Shenkora district in North Shewa, Ethiopia. This town is found 132 km away from the capital city of Ethiopia, Addis Ababa.

| Study design
A retroactive follow-up study of pregnant women was implemented at Arerti Primary Hospital between September 2018 and June 2019.
The survival data were taken from the pregnant women's follow-up charts which contain important factors and covariates.

| Study population
The study population covered all the pregnant women who followed up in the given time period at Arerti Primary Hospital. The pilot survey was conducted to determine a suitable hypothesized sample size.
A systematic probability sampling technique was implemented for choosing a desirable sample size listed in the sampling frame of pregnant women cards which contains their coded numbers and names. From a total of pregnant women followed up in the survey time, the random interval (K) was computed by allocating the total pregnant women followed up in the ward from September 2018 to June 2019. The value of K can be computed by dividing the total number of participants who get service in the hospital during the study time period by the total sample size which equals 5, that is, K = 1220/244 = 5. A random number was selected from one to five by using a random table and in each fifth pregnant woman followed up in the ward, we selected the participants.

| Inclusion and exclusion criteria
All pregnant women on ANC follow-up service at the booking visit who have two and more than two visits were extracted and followed up until delivery or development of the outcome included in the study.

| The outcome variable
The response variable for this study is time to develop PE during the follow-up period of pregnant women attending ANC service at Arerti Primary Hospital between September 2018 and June 2019.

| Method of data analysis
Survival data analysis was employed to address the aim of this study.

| Ethics and consent
The University of Gondar committee on the research initially granted permission to conduct the study. Second, it has permission from the medical director's office of Arerti Primary Hospital to study the research. To ensure the participants' anonymity, the data collector has not related their names and identity numbers with the respondents' information. Furthermore, all collected data were treated as confidential and were not used beyond the scope of the study.

| Descriptive analysis
The study included 201 pregnant women who followed in the ANC service at Arerti Primary Hospital on the study period September 2018 to June 2019. The descriptive results of the study indicate that out of the total pregnant women who were examined, more than 80% of the women had no consecutive previous abortion. The median age and the weight of pregnant women were above 25 years and 50 kg, correspondingly. Concerning the prior information of PE, more than 85% of the pregnant women had no prior history of PE.
Majority of the pregnant women, more than 85% of them were married and more than 10% of pregnant women were unmarried; and out of married women, more than 98% of them were censored, the remaining of them were event occurred, and out of unmarried women more than 41% of them were censored and the remaining were event occurred. Regarding the current pregnancy women status, more than 85% of them were singleton pregnancy and the remaining of them were twin; and among singleton pregnant women more than 95% of them were censored, the remaining were event occurred, and from the total of twin pregnant women more than 46% were censored, the remaining were eventoccurred.

| Result of survival data analysis
The total follow-up period of pregnant women was 10 months.

| Results of the log-rank test
To investigate the significance of the observed difference in the Kaplan-Meier estimates of the survivor functions among different categories of the factors, we applied the log-rank test. The log-rank test result indicated that there were substantial differences in survival probability of pregnant women in covariates of marital status, abortion, diabetes, PE, substance, multiplicity, gravidity, and parity at 5% of the significance level.

| Cox proportional hazards (PH) model
The classical point of view was used to select the variables to be included in the survival process. That is, to determine the variables to be included in the survival model, an automatic variable selection method was used. Then, the command revealed that the variables age, weight, PE, parity, multiplicity, and marital status were considered as candidate independent variables to be included in event times.

| Proportional hazard assumption checking
The proportional hazards assumption asserts that the hazard ratios are constant over time and it is important to use a fitted proportional hazards model. The risk of developing PE must be the same no matter how long subjects have been followed. To test this assumption, using the Cox-PH function in R, scaled Schoenfeld residual tests are used.
The hypothesis of no correlation is tested using the chi-square test statistic. For this case Table 1, all covariates are insignificant with (p > 0.05) and the Global test is not significant indicates that the PH assumptions were met. Figure 1 shows that the scaled Schoenfeld residuals are randomly distributed and a smoothened curve does not exhibit much departure from the horizontal line suggesting that the proportional hazards assumption was not violated.