Correlates of Caesarean Section Delivery in West Bengal, India: An Analysis Based on DLHS-3

It has been well recognised that medically unnecessary caesarean section (C-section) delivery could increase morbidity risks for both the mother and her child and also could put strain on both institutional and individual assets mainly in developing countries. The present study tried to assess the variations in C-section delivery rates by women’s background characteristics and to examine the factors associated with C-section delivery in West Bengal—a state of India. Data from the third round of the District Level Household and Facility Survey (DLHS-3) 2007–2008, covering 6447 ever-married women of age 15–49 years, were used. The results reveal that about 12% women delivered their babies by C-section irrespec-tive of place of delivery, but it rose to about 24% in only institutional delivery. It is also found that the rate of C-section delivery was excessively high in private health facilities (55.8%) followed by higher educated women (50.4%) and for health insurance (36.4%), and antenatal care service eight or more times (36%). The results of predicted (adjusted) probability computed from logistic regression reveal that delivery in private health facilities, higher maternal age, lower birth order and higher level of education were the main influential factors of C-section delivery.


Introduction
The operation for caesarean delivery constitutes a major surgical procedure. There are a large number of adverse effects on women and infants after the C-section delivery. Study found that C-section delivery is associated with a higher risk of ureteral tract and vesical damage, hysterectomy, abdominal pain, maternal mortality, uterine rupture in future pregnancies, neonatal respiratory morbidity, placenta previa and foetal death [1]. It was revealed that women who delivered their baby by elective C-section have 2.84 times more chance of maternal death than women who delivered their baby normally [2]. A study from Africa found that C-section delivery is associated with stillbirths, neonatal deaths and neonatal morbidity [3]. C-section delivery individually minimises the overall risk to foetus death from the breech birth presentations, although it raises the risk of severe neonatal and maternal morbidities and mortality in cephalic presentation [4]. Women who delivered their first baby by C-section have slightly higher long-term morbidity

Data and methods
The present analysis was based on the data from the third round of the District Level Household and Facility Survey, carried out during December 2007-2008 in India (DLHS-3, 2007(DLHS-3, -2008. The District Level Household and Facility Survey was a countywide survey covering 601 districts of India [11]. This survey was designed to gather information at the district level on different aspects of women's healthcare utilisation for Reproductive and Child Health (RCH) including accessibility to the health facilities and to evaluate the health facility capacity and readiness regarding infrastructure. DLHS-3 surveyed a sum of 22213 households and 21878 ever-married women in West Bengal. However, this study was based on 6447 ever-married women of age 15-49 years who had given live birth between January 1, 2004, and the survey date. This was the third round of data which were in the public domain.

Outcome variable
The outcome or dependent variable was C-section delivery; a dichotomous variable was coded as "1" for yes and "0" for no, or, simply, those women aged 15-49 years who delivered their last live birth after January 1, 2004, by surgical procedure were coded as "1", and those women aged 15-49 years who delivered their last live birth after January 1, 2004, by natural process/vaginally or with assistance or instrument were coded as "0".

Explanatory variables
The C-section delivery is an outcome of demographic, socioeconomic, insurance status and institutional factors. Among demographic factors, maternal age at last birth (below 20, 20-24, 25-29 and 30+ years) and birth order (first birth order, second birth order and third birth order or above) were taken. The level of mother's education (no schooling, up to 5 years, 6-10 years, 11+ years), household wealth index (poor, middle and rich), religion (Hindus, Muslims and others), caste/tribe (Scheduled Caste (SC), Scheduled Tribe (ST) and others or general) and place of residences (rural and urban residence) were taken from socioeconomic factors. Coverage by health insurance scheme (yes, no) was also included as an explanatory variable. Antenatal care services include the number of ANC visits (up to three times, four to seven times and eight or more times) and places of ANC services (no ANC visits, only public health facilities, only private health facilities, public/private health facilities, and home or elsewhere), and the place of delivery (public health facilities, private health facilities) were taken from institutional factors.

Statistical analysis
The differences in C-section delivery by women's background characteristics were gross differentials and had been obtained through bivariate analysis. As a number of factors were strongly associated with each other, there was the possibility of confounding. Therefore, it was necessarily desirable to detect the net effect. For this purpose, logistic regression model had been used. In this model, the coefficient (B) and odds ratio (Exp B) were estimated. In order to assess the true differences, it was desirable to obtain adjusted probabilities; by that one can see the actual difference in probabilities [35]. The adjusted probabilities were computed from the coefficients of logistic regression analysis for C-section delivery. A p-value of less than and equal to 0.05 was considered as the significant association between independent variable and outcome variable. Table 1 presents the results of bivariate analysis of C-section delivery rates, by the place of delivery, among all deliveries and all institutional deliveries by the women's background characteristics. From Table 1, it was found that among all deliveries, about 12% of women delivered their last birth by C-section, while it was about 24% among all institutional deliveries. And by place of delivery, it was 58.8 and 15.2% for private and public hospitals, respectively. Among all deliveries, it was observed that the proportion of C-section delivery increased with the increases in the maternal age, while the rate of C-section delivery decreased with the increases in birth order. With an increase in the number of ANC visits, the proportion of C-section delivery also increased. The proportions of C-section delivery were higher for receiving ANC services at only private hospitals and for receiving ANC services at both the private and public hospitals than the categories of not receiving any ANC services, receiving it at home and receiving it at public hospitals only. With the increase in the mother's level of education and household's income, the rates of C-section delivery also increased. The percent of C-section delivery was relatively higher for Hindus than that of Muslims and other minor religious groups. Also, the rate of C-section delivery was higher for other categories (general or nondeprived population) than the deprived communities, that is, Scheduled Caste (SC) and Scheduled Tribe (ST). As compared with rural areas, C-section delivery rate was higher for urban areas. Further, the rate of C-section delivery was quite higher for the women who had health insurance than those who had not. However, the rate of C-section delivery was excessively high for the women who delivered their babies in private health facilities (55.8%) followed by the women who attained higher secondary or more education (50.4%), women who had health insurance (36.4%), women who had received antenatal care service eight or more times (36%), women who had received ANC service in only private health facilities (30.4%) and women who lived in urban areas (29.7%). Besides, women who had only one child, received ANC service four to seven times, received ANC service in both public health facilities and private health facilities, attained upper primary or secondary education and delivered their infants in public health facilities present above 15% of C-section delivery rate in West Bengal. On the other hand, the rates of caesarean delivery were very low, which was lower than 5% for the women who had three children, women who did not receive any ANC service, women who received ANC service at home or elsewhere, illiterate women, poor women and tribal women. Table 2 presents the results of the logistic regression analysis and adjusted probabilities which were computed from the coefficients of logistic regression analysis for C-section delivery. The logistic regression analysis included only the women (unweighted no. = 3149) of age 15-49 years who had given live birth in any health facilities since January 1, 2004, in West Bengal because performing of C-section is possible only in health institutions. Women's background characteristics, utilisation of antenatal care and delivery care service were considered as independent variables, and the type of delivery (normal delivery or C-section delivery) was taken as a dependent variable in this analysis. The actual probability of C-section delivery was 24.1% (weighted) for all the women who had given live birth in any health facilities. The results showed that the place of delivery and number of ANC visits were the significant factors of C-section delivery among institutional factors; maternal age and birth order were the significant factors of C-section delivery among demographic factors; and the level of maternal education was the only one factor significantly associated with the C-section delivery among socioeconomic factors. Delivery in private health facility was the strongest predictor of C-section delivery after controlling for other variables. The adjusted probability of having  C-section in private health facilities was 49.6%; that was almost three times higher than public health facilities (18.3%). The place of ANC services did not seem to have a very clear effect on C-section delivery, but the frequency of antenatal visits had a mild effect on C-section delivery; it was mostly found at higher number of ANC visits. The probability of C-section delivery for the older women was higher than younger women after controlling for other variables. With the increase in maternal age, the chances of having C-section delivery also increased. The birth order also was one of the strongest predictors of C-section delivery. With the increase in birth order, the probability of having C-section decreased, which was in the opposite direction to the maternal age. The effect of education was observed,  which was mostly found at the higher level of education. A small variation in the probability of C-section delivery was observed between the rural and urban residences, but it was an insignificant factor after controlling for others. The effect of level of income on C-section delivery was mild, so were the effects of religion and caste. Besides, the insurance coverage did not show any significant effect on C-section delivery in this analysis, although it had a large gross effect on C-section delivery in bivariate analysis.

Discussion
This study showed that the actual probability of C-section delivery was about 12% among all deliveries and 24% among all institutional deliveries in West Bengal. The results of logistic regression revealed that the place of delivery, the number of ANC visits, maternal age, birth order and the level of maternal education were the significant factors associated with the C-section delivery. Delivery in private health facilities was the strongest predictor of C-section delivery as expected. This finding is consistent with the findings of previous studies [14,17,20,28,29,36,37]. This finding could be explained in various ways. Firstly, the proprietors of private health facilities are revenue oriented, and they always try to encourage doctors to perform C-section delivery instead of normal delivery because it brings more revenue; secondly, many doctors are also financially motivated and, therefore, advise patients to have C-section; thirdly, generally doctors are very busy persons, engaged in multiple tasks, and, thus, often they perform C-section even before the arrival of the delivery's labour pain, so as to avoid patient call; and fourthly, both doctors and proprietors of private health facilities do not take risks regarding delivery, so doctors perform C-section before the arrival of the actual delivery's labour pain for avoiding any risks. The higher maternal age was also another important significant factor of C-section delivery. This finding is found to be significant in almost all the previous studies [18,26,31,38,39]. The higher age of women is much more associated with the prolonged labour, unable to progress at the time of birth and foetal distress which could lead to C-section delivery. Birth order (parity) was also another significant factor of C-section delivery. This finding is similar to a large number of studies [15,27,[40][41][42][43]. The pregnancy and delivery complications are higher among the primiparous women or women of lower birth order than women of higher birth order which leads to higher chances of C-section delivery. On the other hand, maternal age and birth order are highly correlated with each other. The probability of having C-section of lower birth order is higher, but once the birth order is controlled, then higher age has greater chances of C-section delivery. So, women of higher age with the low birth order have higher chances to have C-section delivery. Another most important factor of C-section delivery was the level of woman's education. This finding is also consistent with a large number of previous studies [25,32,34,42,44,45]. In general, highly educated women are more aware of maternal and child health and quality of care which would lead women to prefer to go to private health facilities for delivering and ultimately lead to have C-section delivery. The higher number of antenatal visits was the significant factor of C-section delivery as expected though the effect was mild. This finding is also consistent with other studies [22,26,29,30,46]. The higher number of ANC visits might be the result of pregnancy complications which indicates the surgical operation to deliver a baby. The place of residence was not a significant factor in this study. A similar finding has been observed in the study of Kerala, India [31], and in Jordon [47]. These studies argue that well connectivity and availability of health facilities across the state might be the possible reasons for this finding. The level of income, religion, caste systems and insurance coverage did not show a significant effect on C-section delivery.

Conclusions
From the above analysis, the present study revealed that women's demographic, socioeconomic background characteristics, antenatal care service and delivery care can have an effect on C-section delivery. From the findings of the present study, it could be recommended that there are some steps which may help to reduce or stop the medically unnecessary C-section delivery for the betterment of women and child health and appropriate use of resources. First, it is found that the rates of C-section delivery were almost three times higher in private health sectors than the public health sectors. Therefore, universal guidelines, protocols and medical audit on C-section should be implemented. Further, the public health system should take steps to monitor the reasons of C-section delivery. The results revealed that women at higher age were at more risk for C-section delivery. The results also found that higher educated women were more tend to have C-section delivery. Thus, the maternal and child health-related educational programme should be implemented for educated women as well as uneducated women. Finally, the community health workers should be trained to circulate the awareness about risks and benefits of C-section delivery, so that medically unnecessary C-section deliveries are not requested or demanded by women and their families. One major limitation of this study is that, in the data source (DLHS, 2007(DLHS, -2008, there is no information on whether the C-section delivery was medically indicated or not. Thus, further studies are needed to examine the factors for medically indicated C-section delivery and medically unindicated C-section delivery separately.