Alarming trends of Caesarean Section: time to rethink-evidence from large-scale cross-sectional sample survey in India

Background: Caesarean section (CS) delivery rate has increased significantly both globally and within India posing a burden on overstretched health systems. Objective: To analyze the trends of CS delivery from 1998-99 to 2019-21 and to understand the proximate determinants of CS deliveries in India. Methods: Analysis of secondary data (National Family Health Survey) of a nationally representative sample of 230,870 women (year 2019-21) was undertaken to explore the trends, distribution, and determinants of CS deliveries in India and within states. Multivariable analyses were performed to determine the proximate variables associated with CS and elective CS. The relative interaction effect of confounding factors such as number of ANC visit, place of residence and wealth status on cesarean delivery were assessed. A composite index was generated using trust, support and intimate partner violence variables named as partner human capital index (PHI) to study it’s influence on CS deliveries. State wise spatial distribution of most significantly associated factors namely wealth quintile and ANC checkups were also analyzed. Results: Overall prevalence of CS is 21.4% which had risen from 6.6% in 1998-99. The adjusted odds of CS deliveries were significantly higher among women who were highly educated (OR: 7.30; 7.02-7.60), with four-or-more ANC visit (OR: 2.28; 2.15-2.42), belonging to high wealth quintile (OR:7.87; 7.57-8.18), and from urban region. Increasing educational attainment of the head of the household (OR: 3.05; 2.94-3.16) was also found to be a significant


Table of Contents
No, I do not wish to publish my submitted manuscript as a preprint. 2) If accepted for publication in a JMIR journal, would you like the PDF to be visible to the public?
Yes, please make my accepted manuscript PDF available to anyone at any time (Recommended).
Yes, but please make my accepted manuscript PDF available only to logged-in users; I understand that the title and abstract will remain v Yes, but only make the title and abstract visible (see Important note, above). I understand that if I later pay to participate in <a href="http

Original Manuscript
Alarming trends of Caesarean Section: time to rethink-evidence from large scale sample survey in India Background: Caesarean section (CS) is a major life-saving surgical obstetric procedure, highly effective in saving lives of both mother and infant, only for medically indicated causes [1,2]. Over a decade there has been a rapid increase in CS delivery rates across the globe, [3][4][5]. The number of caesarean births recorded each year globally are more than 18 million, accounting for approximately 19.1% of total births. These numbers reportedly have increased from just 7% in 1990 and are projected to increase to nearly third (29%) of all births by 2030 [1,6].
Developed and developing countries fare similarly with respect to prevalence of CS delivery rates (27.2% v/s 20.9%). While in Africa CS accounts for 7.3% of deliveries, the rate rises to 40.5% in Latin America and Caribbean [7]. The highest average annual rate of increase is observed in the regions of Asia (6.4%) [8]. In India, the proportion of caesarean deliveries have dramatically increased to 17% in 2015-16 and 21.5% in 2019-21 from just 3% in 1992-93 [9]. The caesarean rate has always been under high dispute due to its variable need across world that in turn is based on the nature of population, healthcare facility's capacity to handle cases, availability of resources and the clinical management protocols used locally. Based on the available evidences, there is no justification for any region to have a CS rate higher than 10-15% [10], regardless of their complexity or other characteristics. A systematic review concluded that at population level, caesarean rates higher than 10% were not associated with reductions in maternal and newborn mortality rates [2]. Thus, CS must only be undertaken only when medically indicated and in facilities equipped to treat surgical complications [11]. Rather than aiming for a certain rate, the WHO urges that every attempt be made to provide CS to women in need [10].
A Lancet Commission on Surgery and global health states that surgical interventions are essential in bringing down the mortality and morbidity at all stages of life [12]. To achieve target 3.1 Sustainable Development Goals of reducing the global maternal mortality ratio to less than 70 per 100,000 live births, a lot of collaborative efforts still are required [13]. The aim of SDG-3 [14] is to ensure healthy lives and promote wellbeing and hence, it is essential to understand the geographical disparity and to explore the determinants of CS in India.
Literature has found that besides clinical indications, the factors associated with increased rate of CS are demographic changes, social and educational advancements which have given rise to obstetrician's preference for it, financial incentives, women's request for caesarean delivery, women deferring pregnancy until they reach the end of their reproductive years, and inadequate training of physicians in vaginal delivery. Socioeconomic inequities appear to have created a pattern of underuse and overuse of CS based on income and levels of education [15][16][17].
According to the nationally representative survey, one in every five pregnant women in India had a CS even if they didn't require one medically [8].  threshold of 15%, posing a serious public health risk. It is 47.4% in private facilities as compared to 14.3% in public facilities [9]. A study reported that if private sector institutions in India had adopted the World Health Organization's 15 percent caesarean delivery rate standard, the number of preventable caesarean deliveries would have been 1.83 million, with a potential cost savings of $320.60 million [18].
Several studies have investigated factors contributing to CS [3,17,[19][20][21][22]; the socio-economic lopsidedness of CS deliveries in India towards urban and wealthier population is well established but some social factors like partner's human capital comprising of factors like any behavior within intimate relationships, trust, psychological abuse and other controlling behavior have not been explored thus far. Hence, this study is planned to understand the factors associated with the CS rate with an objective to analyze the trends of CS delivery from 1998-99 to 2019-21 and to understand the proximate determinants of CS deliveries in India.

Method & Data source:
This study utilized the data on cesarean delivery from large-scale health survey: the National Family Health Survey (NFHS The generated scores were categorized using percentile value into 'low' indicating abusive, suspicious, and inhuman partner, 'moderate', and high which implies to highly supportive and caring partner. A total of 24,216 women provided the information on the included variables for PHI and thus a separate stepwise reverse regression analysis was done keeping the sample as 24,216 to identify the determinants. husband/partner's actions; ever had eye injuries, sprains, dislocations or burns because of husband/partner; respondent ever physically hurt husband/partner when he was not hurting her; husband/partner: person who hurt respondent during a pregnancy; respondent afraid of husband/partner most of the time, sometimes or never. A composite index was created using these dichotomous variables. The generated scores were categorized into low, moderate, and high. It was done using percentile value. The Cronbach alpha is 0.80.

Women's educational attainment
The women's educational attainment was recoded into 4 categories such as 0= no education; 1 = primary education; 2= secondary educated; 3 = higher and above

Place of residence
Missing values: Of 232,920 observations of women who had an institutional delivery during the 5 years preceding the survey, in 24% of cases, number of ANC visits were missing. We assumed that the number of ANC visits is an important parameter that is necessary to detect high risk pregnancy and also influence the decision for CS. Hence, we used the single imputation technique. The missing information on ANC visits were imputed based on baseline non-missing background characteristics of women namely caste and region.

Statistical analysis:
Analyses were done to observe the trend in frequency and distribution of CS over the past 15 years. This was done with regards to place of residence, type of health facilities and need of CS. CS for year 2005-06, 2015-16 and 2019-21 data were graphically plotted to assess state wise distribution. We also studied the state wise spatial distribution of two most significantly associated factors namely 'Wealth quintile' (which was categories as poor, middle, and richest using percentile values) and 'ANC checkups' (which was categories as No ANC and one or More than 4 ANC using percentile), that have emerged from the multivariable analyses. Analyses were done to explore the association of primary outcome and the explanatory variables. The initial bivariate analysis was done with chi-squared test for ordered categorical variables. The list of confounding factors for unadjusted and adjusted regression models were screened based on bivariate analysis if the differences among categories were higher than 5 percent. Those with a significant difference (p<0.05) and those biologically plausible were selected for the adjusted analysis. Multivariable analyses were performed to determine the proximate variables associated with CS and later to ascertain the proximate variables associated with elective CS. The regression results are presented as odds ratio (OR) at 95% CI. The step wise reverse regression models were used to maintain the same sample size for variable with the smaller number of cases, such as information available at partner' level. The list of variables for assessing the interaction were chosen based on the differences among categories(p<0.05). The data are available in public domain which could be accessed post registration on the website and hence there are no ethical implications [23].

Prevalence of caesarean delivery and associated attributes:
A total of 230,870 women delivering in a period spanning 5 years from the date of survey were included in the study nested within 707 districts and 36 States/UTs of India. Of these, 21.4% (n=49,634) delivered through CS which had risen from 6.6% in 1998-99 (Figure 2).
The prevalence of CS deliveries was more in urban areas [32.3%; n=30,203 v/s rural 7.6%; n= 14,539] and in private facilities [47.0%; n= 29175 v/s 14.3%; n= 20,459 in public facilities]. Also, it is evident from figure 2 that CS deliveries have increased manifolds at private facilities as against the stagnant rates in public healthcare facilities. Of the total deliveries reported in India during 2019-21, 12% (n=28512) were elective CS while 9% (n=21122) were emergency CS. (Figure 2) A state wise comparison (Figure 3) shows an overall change in the CS deliveries over the years across all the States/UTs with a slight skewness towards the southern and extreme northern states. It is also evident that 27% (101 out of 707) of the districts during 2019-21 have CS rate more than WHO's recommended cut-off, compared to <15% (101 out of 641 districts) during 2015-16.
{Insert figure 3 here} The prevalence of CS was higher among those who were more than 25 years (70.4%), those who were more educated (40%), with low family size (25.6%), from urban areas (32.3%) and general caste (28.4%). Probability of CS amongst women who are residing in the northern and southern part of India was higher (70.9%) which is also evident from the spatial distribution of CS as shown in Figure 3. Probability was more among women with one parity (31.9%), those having 4 or more ANC visits (27%), those with tall (24%) stature, who were overweight or obese (81.83%), those with mild (21%) or no anaemia (24%) and those belonging to higher wealth quintile (Richest-39.1%) ( Table 2). Logit regression adjusted for caste, religion, and region was performed which found that the odds of CS were significantly higher among women having secondary education (OR: 3.51; 95% CI 3.39-3.64) and higher (OR: 7.30; 95% CI 7.02-7.60) compared to those who were illiterate. Increasing educational attainment of the head of the household was also found to increase the odds of having CS (OR: 3.05; 95% CI 2.94-3.16). The odds of cesarean delivery among women with four or more ANC visits (OR: 2.28; 95% CI 2.15-2.42), those belonging to high wealth quintile (OR:7.87; 95% CI 7.57-8.18), and from urban region were higher compared to their counterparts with less than 4 ANC visits, poor and residing in rural region, respectively (p<0.001) ( Table 2). information on other predictors of CS deliveries.   The dynamics of confounding factors was clearer after including the partner's characteristics to cesarean delivery indicating closer association with the outcome. The null model of reverse regression indicate that the odds of cesarean delivery was 0.32; 95% CI 0.31 -0.33 (p<0.001). The odds of outcome were significantly higher among the women with moderate (OR: 1.46; 95% CI 1.36-1.56) and high (OR: 1.61; 95% CI 1.49-1.74) PHI. After adjusting for other background characteristics of women, the strength of the association reduced (OR: 1.24; 95% CI 1.14-1.35) but remained statistically significant. (Table 3). Note: Exponentiated coefficients; * p < 0.05, ** p < 0.01, *** p < 0.001; 1 religion, caste, region and household size were controlled The available information on type of CS in terms of elective and emergency were analyzed as presented in appendix 1. The odds of elective cesarean delivery were higher among women belonging to higher wealth status (OR: 1.66; 95% CI 1.25-2.21), women belonging to Christian religion (OR: 1.67; 95% CI 1.14-2.43) and those with lower parity (Appendix-1).
The relative interaction effect of confounding factors such as number of ANC visits, place of residence, wealth status and PHI on cesarean delivery were estimated and shown in The odds of CS were 3.14 (95% CI 2.56-3.85) among women with 4 or more ANC visits and higher PHI compared to their reference category (i.e., low PHI and No ANC visits). The combined effect of place of residence and educational attainment remained statistically significant and indicated that the urban educated women were more likely to have CS than rural educated women. Amongst those women with 4 or more ANC visits, the odds of the cesarean delivery starkly increased among women with 12 or more years of education specially in urban region (OR: 10 Categorizing state wise distribution of CS deliveries by ANC visits reiterated the regional disparities. It was found that women from southern and northern states have higher CS (higher than the WHO's cutoff) irrespective of the ANC status (Figure 4a). The same could be seen as regardless of the region, women belonging to the wealthier households were found undergoing CS deliveries ( Figure  4b).
{Insert figure 4a and 4b here} Note: N= 28,305; Exponentiated coefficients; * p < 0.05, ** p < 0.01, *** p < 0.001; 1 religion, caste, and region were controlled. The findings are in line with a study reporting an upward trajectory of CS deliveries globally and in India [3,24]. State-specific prevalence vary from as low of 5.2% in Nagaland to as high of 60.7% in Telangana state within India. Specifically, our study reported increasing prevalence across all the state/UT amongst the wealthier population, which is similar to already reported figures [24]. Also, southern states with better health indicators have a greater preponderance of CS [6].
A significant increase in the CS deliveries is being reported from private facilities (from 20% in 1998-99 to 47% in 2019-20). Literature reports similar findings [25][26][27][28] which could be associated with willingness to pay, especially amongst women belonging to wealthier status. It is reported that an increase in CS rates could be attributed to physician's choices in private sector [15]. It is now established that the probability of CS delivery and elective CS increases with better wealth quintile [28][29][30][31][32][33]. Our study also echoes similar findings. This is in contrast to a global study across 57 developed and developing countries that states that there is a poor correlation between income inequality and absolute wealth related inequality in CS deliveries [34].
Further, affluent women, have a greater likelihood of CS by choice, probably because of perceived lower risks [35]. This corroborates with our findings wherein wealth status independently influences the odds of having CS deliveries manifolds irrespective of the place of residence, ANC check-up status or PHI. Similarly better access to health services promotes CS. This is evident from an increase in the likelihood CS births with increase in the number of ANC visits or early initiation of ANC checkups [36,37], similar to our study.
However, there are conflicting results from studies that looked at the relationship between maternal education level and CS delivery. Although most of the studies from Bangladesh [25], Thailand [38], Pakistan [39] and India [40,24] have reported a strong association of formal education with CS deliveries, the one from Egypt found no significant association [41]. Improved autonomy and capacity to take decisions probably explains increased CS among educated women [41,42]. Along similar lines, education levels among heads of household also influence decisions for CS. Concerns around medical malpractices or viewing CS a measure to prevent any mishappening could be the possible reasons that motivates families to support CS [43][44][45].
Access to equitable health services has generated a lot of discussions around social factors. Available studies have explored the relationship of relation with partner and violence and negative health outcomes [44][45][46]; however we could not find any literature on its association with CS. To explore it further, we developed an index that comprehensively captures the attitude and behavioral issues towards the partners. This measure assumes importance when we attempt to understand the reproductive health issues beyond the medical lens.
Data from large nationally representative sample collected using a scientific methodology is a strength of this analysis. Although there are several reports already available on CS, this study provides the most recent data of a country as large as India. The findings underscore the growing concerns around CS. Also, it highlights an important parameter of Partner Human Capital, a composite index encompassing attitudinal and behavioral factors, that is rarely discussed in the context of CS. Imputation of a key exposure variable, 4 ANC visits enabled us to overcome the gap resulting from missing data, thus paving the way for a more robust analysis.
Despite the fact that our study has numerous methodological and conceptual strengths, it also has certain limitations. The study captures information for a reference period of five years preceding the survey. Though self-reports on CS might not have been affected, yet other variables such as number of ANC visits pertaining to CS births, and reporting of partner human capital may have suffered from a long recall period. Although we sought to adjust for individual risk factors for CS, even after imputing for the missing values and excluding variables with more than 25% missing values, there are possibilities that we may have missed some essential predictors or determinants.
To conclude, the study reiterates the increasing trend of CS deliveries across India, thereby raising concerns. Better education, wealth and social factors have been incriminated as the contributory factors. There is a need to institute proper monitoring mechanisms to assess the need of CS, especially when performed electively. Improved awareness about the obstetric dangers and post-partum complications of Caesarean deliveries over normal deliveries along with strategical implementation of Govt initiatives can help us take a rational decision on CS deliveries. As we prioritize increasing access to services for better health gains, this turns out to be one example where better access predisposes to overmedicalization. While low CS in underdeveloped communities can be a concern, the potential for medically unnecessary overuse of CS delivery as income, education and higher PHI raises a different set of issues that demands for a targeted health policy interventions to achieve more appropriate use of CS among richer and more educated women.

Figures
Spatial distribution of percentage of caesarean section delivery across Indian States.