High proportion of caesarean section at a rural hospital in south western Uganda: A cross sectional study

Background: Globally, overall prevalence of caesarean section (CS) is estimated at 18.6%, with 27% in high-income countries and 6% in low- and middle-income countries. There is an increase in trends of CS in Uganda from 8.5% in 2012 to 11% in 2016. There have been no studies done to account for the high rates of CS in a rural hospital in Uganda. This study determined the proportion and factors associated with CS delivery at a rural hospital in south western Uganda. Methods: This was a cross sectional study of 321 immediate postnatal women in a rural hospital in south western Uganda. A structured questionnaire and data abstract forms were used to collect information on proportion and factors associated with CS. Eligible participants were enrolled consecutively. Logistic regression analysis was done to identify the factors associated with CS taking into account potential confounders. Results: This study recruited 321 women with mean age of 25.8 ±6.086 years, mean parity 2.6 ±1.673 and mean antenatal care visits of 4.27 ±1.197. The proportion of CS in this study was 38.3% (123/321). Of these, 110 (89.4%) were emergencies and 27(10.6%) electives. Only 8.4% of the respondents were referrals. The commonest indications of CS were fetal distress (28.5%), history of previous CS (18.7%) and poor progress of labour (11.4%). No factors were signicantly associated with CS in this study. Conclusion: There is a high proportion of CS in a rural hospital in Uganda and this is three times higher than the WHO recommended CS rates. Majority of CS are emergencies due to fetal distress and poor progress of labour. There is a need for additional studies exploring the reasons for the much higher than expected CS rates.


Background
Caesarean section is a life-saving procedure that is important in preventing poor obstetric outcomes for the woman & fetus (1,2). There is, however, a growing concern of increasing CS delivery rates and this rise has become a major public health concern (3). Globally, the prevalence of CS is estimated at 18.6%, with 27% in high income countries, 6% in low and middle income countries and 7.3% in Africa (4). In Uganda, the overall rate of CS for live births at facilities was 9.9%, increasing from 8.5% in 2012 to 11% in 2016 (5).
The World Health Organisation (WHO) recommends CS rates of 10-15% to achieve optimum maternal and neonatal bene ts and CS rates less than 10% may indicate inadequate utilization of CS services among women who need them, while more than 15% of CS rates suggests an injudicious use of CS especially without medically indicated reasons (2).
In 2017, the study rural hospital had a total of 2,229 deliveries, to which 675 (30.2%) were CS deliveries (Hospital Records, 2018). The indications for CS deliveries are well documented in the literature including fetal, maternal and placental factors (6)(7)(8)(9). Several studies have attempted to determine the factors that contribute to CS delivery including social demographics such as age, occupation , obstetric factors (number of ANC visits, parity, timing of rst ANC visit) and health system factors like referral status and distance (3,(10)(11)(12)(13)(14). There are few studies which focus on factors associated with CS in Uganda and these are limited to urban settings (9,15).
There are higher rates of CS deliveries in the rural hospital in south western Uganda than the national CS rates & WHO recommended rates. Higher CS rates are associated with poor pregnancy outcomes such as hospital-acquired infections, delayed breastfeeding, long hospitalization & increased costs (1,2,16,17).
Studies have not been done in the rural private facility setting to account for the high rates of CS in Uganda and there is scanty literature focusing on reasons for the high CS rates and associated factors.
Therefore, this study aimed at determining the proportion and factors associated with CS at a rural hospital in south western Uganda.
Speci c objectives 1. To determine the proportion of CS delivery at a rural hospital in south western Uganda.
2. To determine the sociodemographic factors associated with CS delivery at a rural hospital in south western Uganda.
3. To identify the obstetric factors associated with CS delivery at a rural hospital in south western Uganda.
4. To identify the health system factors associated with CS delivery at a rural hospital in south western Uganda.

Methodology
Study design and setting: This was a hospital based cross sectional study that was conducted from December 2018 to March 2019 among 321 women who were immediate postnatal mothers within 72 hours at a rural hospital in south western Uganda. The rural hospital is a missionary not-for-pro t referral hospital that offers general and specialized healthcare and serves a population of 300,000. In 2017, the rural hospital had a total of 2,229 deliveries, with 1,554 (SVD's) and 675 (CS) (Hospital records, 2018). The rural hospital had sta ng of 2 obstetricians, 5 Medical o cers & 2 intern doctors and 23 midwives in 2017 (Hospital records, 2018).
Study population and eligibility criteria: The study included all women who had delivered in the rural hospital within 72hrs. Women who were critically ill, deaf and dumb, had still births and early neonatal deaths were excluded from the study. This study was conducted between December 2018 to March 2019.

Sample size and Sampling Procedure
The sample size was calculated using Kish-Leslie formula. The 30.2% proportion of CS was used in this study (Hospital Records, 2018), 95% con dence interval and gave a sample size of 324 women. Consecutive sampling was used to select the participants until the required sample was got.

Study variables
The outcome variable was CS while the independent variables included; social demographic, obstetric and health system factors. Caesarean section was de ned as an operation performed under regional or general anaesthesia to deliver the fetus, placenta, and membranes through an incision in the abdominal and uterine wall (7,8). Age was collected in complete years from date of birth. Age was then categorised as <18yrs, 18-34yrs and ≥35yrs. Parity and number of pregnancies one had had before were collected in complete numbers and both were categorized as 1, 2-4 and ≥5. Number of ANC visits a mother had attended during that pregnancy was collected in numbers and categorized as <4 and ≥4.
With education level, no formal education was collected as those who had never gone to school, primary education was collected as those who had had 1-7 years in school, secondary education was collected as 1-6 years in secondary school and tertiary education was collected as university or diploma.
Data collection procedure and tools: A structured questionnaire was the tool for data collection for both independent and dependent variables. A data abstract form was a tool used to gather additional information from the participant's medical le. The questionnaire was used to collect data on social demographic variables such as age, educational level, tribe, family income, religion, type of marriage, marital status and distance from home to the nearest health facility. The questionnaire had the obstetric variables such as parity, number of pregnancies one had ever had, number of ANC visits and timing of 1st ANC visit. The data abstract form was used to collect more information on obstetric variables such as gestational age at delivery, reasons for CS and the neonates' birth weight. The data abstract form was used to collect data on some health system variables such as referral status, referral distance and level of referring health facility. The questionnaire was also used to collect a health system variable like community based insurance.
Quality control: The questionnaire and data abstract forms were pretested at a rural hospital in Uganda.
Data were checked for completeness, errors, and omissions. We recruited four research assistants and were trained on research protocols and data collection procedures.

Data analysis
Data were entered, cleaned, analyzed using SPSS computer package version 23.0. Continuous variables were summarized as medians, means, and standard deviation. The proportion of CS was determined by dividing the number of postnatal women who had delivered by CS by the total number of women who had delivered from the rural hospital during the study period. Bivariate analysis was done to determine the factors associated with CS. Odds ratios and 95% con dence intervals were computed to measure association and a P<0.05 indicated statistical signi cance. All variables with P values <0.05 at bivariate analysis were entered into a multivariate logistic regression model to determine independent associations with CS.

Social demographic characteristics
The mean age of the study participants was 25.8 (±6.086), median age 25.0 years and majority 275 (85.7%) of the respondents were aged 18-34 years. The social demographic characteristics of participants are shown in (Table 1) Obstetric characteristics of study participants More than half of the respondents 180 (56.1%) had had 2-4 pregnancies and majority of the participants 211 (65.7%) had attended <4 ANC visits. The obstetric characteristics of participants are shown in ( Table   2) Health system of study participants More than half of the respondents 168 (52.3%) had community based insurance, 27 (8.4%) were referrals and 12 (44.4%) of the respondents had travelled a distance of ≥20km to a rural hospital in south western Uganda. The health system characteristics of participants are shown in (Table 3).

Proportion of caesarean section
Out of the 321 women who delivered at the rural hospital, almost two thirds 198 (62.0%) of them had SVD and 123 (38.0%) had CS. Of the 123 women who had CS, 110 (89.4%) were emergencies and 13 (10.6%) had electives.

The indications of caesarean section
Out of 123 respondents who had had CS delivery, slightly over one in four (29.9%) was due to fetal distress, (18.2%) was due to previous CS and (10.7%) was due to poor progress of labour.

Factors associated with caesarean section
In this study, multivariate analysis was not done because almost all the factors associated with CS were not statistically signi cant after bivariate analysis. Number of ANC visits attended was the only factor found to be associated with CS after bivariate analysis in this study.       The rural hospital is a not for pro t private hospital with general and specialized health care services and serves the surrounding population of Rukungiri District and neighboring districts. The results of this study may be explained by the fact that the rural hospital is known in the western region for offering high quality, affordable obstetric services, which may attract many women to deliver from there either as self-referrals or as referrals from other public and private facilities within Rukungiri district and beyond.
The other reason for the high proportion of CS in this study, maybe attributed to the fact that most of the rural health centres have inadequate resources and the shortage of healthcare staff such including midwives, nurses, general doctors and obstetricians (Dusabe et al., 2018), and this may lead to women preferring to deliver from this rural hospital compared to other health facilities within the region. This is because the study rural hospital has 2 obstetricians, 5

Factors associated with caesarean section delivery
In this study, the number of ANC visits was associated with CS delivery. Women who had attended ≥4 ANC visits were more likely to have CS compared to their counterparts who had attended <4 ANC visits.
This nding may be attributed to the poor quality of ANC services offered to women by the healthcare providers, where the pregnancy deviations may not be detected early. The nding in this study is coherent with similar studies done in Uganda, Kenya and Bangladesh where attendance of ≥4 ANC visits was associated with increased CS (Dusabe et al., 2018; Oweya, 2010; Shiblee, 2017). Another study done in public hospitals in Brazil found that having ≥6 consultations during ANC visits was associated with increased CS (Vieira et al., 2015). These ndings are contrary to those of a study done in Southwestern Ethiopia which found that respondents who had attended one ANC visit were more likely to have CS compared to their counterparts who had attended more than one ANC visit (Dadi & Mihrete, 2018).
Findings in this study show that women who had parity of ≤4 were more likely to have CS compared to their counterparts who had parity of ≥5. This is not surprising as information from literature resources In this study, respondents with the secondary and above level of education were more likely to have CS compared to their counterparts who had primary and below level of education. Women who were referred were more likely to have CS compared to the non-referrals in this study. This is not surprising because they already had delivery complication as explained by the high percentages of indications of CS in this study. Although referral status was not signi cantly associated with CS in this study, a study conducted in the Democratic Republic of Congo found that referral status was associated with CS delivery (Philémon et al., 2017). Similarly, in another study conducted in Tanzania, the level of CS among medically referred women was higher (Sørbye et al., 2011).
Findings in this study show that women who were on community-based insurance were more likely to have CS compared to their counterparts who were not insured. This may be due to the frequent ANC visits attended by women on insurance, as they have already planned costs and attendance of more ANC visits is found associated with CS in this study. Although there was no statistical signi cance between health system factors such as community-based insurance, referral status, and level of the referring health facility with CS in this study, previous studies done in the Democratic Republic of Congo, Kenya and Tanzania identi ed a relationship between some of these factors and having CS (Oweya, 2010; Philémon et al., 2017;Sørbye et al., 2011). On the other hand, a systematic review and meta-analysis found that women with private insurance were more likely to have CS compared to their counterparts with public insurance (Hoxha, Braha, Syrogiannouli, Goodman, & Jüni, 2019).

Study limitation
There were inadequate numbers of some variables during analysis for the different categorizes. This was a cross-sectional study and therefore it cannot derive the causal relationship from the cross-sectional analysis. Consecutive sampling was used and thus introduced a form of bias.

Recommendations
There is the need for caesarean section audits to check whether these CS deliveries were medically indicated at a rural hospital and this may help the hospital to reduce the unnecessary CS if identi ed.
There is a need to assess the labour monitoring process by health care providers in a rural hospital as fetal distress was the commonest indication of CS delivery and this would help to rule out the issue of over diagnosing or misdiagnosing fetal distress by the staff.

Availability of data and materials
The data set used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Ethical approval and consent to participate Ethical approval was sought from the School of Health Sciences Research and Ethics Committee at Makerere University College of Health Sciences (CHS). Administrative approval was sought from the Medical Superintendent, rural hospital in south western Uganda. The study procedures, purpose, risks, and bene ts were explained to participants before obtaining informed consent. The participants were also informed about con dentiality and privacy measures were put in place for the information collected. Filling in of questionnaires was done in one of the o ces in the postnatal ward to ensure privacy.

Consent for publication
Not applicable