Trends, Regional Variations, and Socioeconomic Disparities in Cesarean Births in India, 2010-2016

Key Points Question What is the current proportion of cesarean births across India? Findings In this cross-sectional study of 699 686 adolescent girls and women aged 15 to 49 years, the cesarean birth rate was 17.2% in 2010 through 2016, with variations ranging from 3% to 70% according to regions and socioeconomic groups. Meaning India is characterized at the same time by a deficit of access to cesarean deliveries in poorer communities and the emerging overuse of cesarean delivery in more affluent regions and social groups.


eMethods 1. Assessment of the NFHS-4 Figures on Cesarean Births
The preliminary analysis of original NFHS-4 data revealed that very high cesarean birth rates were observed in several states in India. Especially noticeable were the states of Andhra Pradesh and Telangana, where rates reach the unusually high levels of 40% and 58% respectively. The NFHS-4 estimates can be further compared with the state-wise estimates drawn from the previous NFHS round conducted in 2005-06 (eTable 1). This comparison points to the very rapid increase observed in the former Andhra Pradesh state, which included then Telangana: the cesarean rate more than doubled in Andhra Pradesh, rising by 25.3 points in a decade.

eTable 1. Population-Based Cesarean Rates in India by
The national in cesarean rates have been significantly slower in India (+8.7%) and in specific regions, notably among other South Indian states such as Kerala (+5.7%), Tamil Nadu (+13.8%) or Karnataka (+8.0%). Thus, in spite of strong congruence between estimates of cesarean rates from the last two NFHS rounds (r²=.81), the proportion of cesarean deliveries seems to have recorded an unusual growth in the states of Andhra Pradesh and Telangana.
In order to test the reliability of the NFHS-4 figure for Andhra Pradesh and Telangana, we conducted two consistency tests at state-and district-level. We used for that the results of the DLHS-4 survey conducted in 2011-12. This is the latest survey-based estimates of cesarean deliveries in India. The DLHS-4 was replaced by the Annual Health Survey (AHS) in nine less advanced poorer states (Assam, Bihar, Chhattisgarh, Jharkhand, Madhya Pradesh, Odisha, Rajasthan, Uttarkhand, and Uttar Pradesh). We use therefore data from the AHS (2011-12) for these states. Cesarean rates from AHS were computed from the rates in government and private institutions mentioned in the respective state AHS publications. In addition, no result has been published by the DLHS-4 for several states and territories including Gujarat. A total of 29 state-wise estimates of cesarean rates are therefore available for comparison between the NFHS-4 and the NFHS-4 and AHS conducted about 2 years earlier (eTable 1).
The values plotted in eFigure 1 demonstrate the very strong correlation (r²=0.94) existing between the two series, in spite of the overall increase in cesarean rates between the two successive surveys. We observe in particular that the NFHS-4 estimates for Telangana and Andhra Pradesh where cesarean rates rose very fast between the last two NFHS rounds are confirmed by the DLHS-4 figures.

eFigure 1. Population-Based Cesarean Rates in India by State and Union Territory According to the DLHS-4 and the NFHS-4 Surveys, 2011-2016
We conducted a further district-level analysis for the former state of Andhra Pradesh (Andhra Pradesh and Telangana). We plotted again the DHLS-4 estimates against the NFHS-4 estimates for the 23 districts that constitute today Andhra Pradesh and Telangana. The correlation coefficient (r²=0.65) between the two series is high and significant at 1%, while lower than the coefficient measured at state-level because of the higher variability of district-level estimates. This strong correlation confirms the consistency of NFHS-4 figures in the two states where cesarean rates recorded the fastest growth during the last ten years.  Based on cesarean rates computed by socioeconomic quintile for the states with more than 2000 sample births. To calculate relative inequality, we divided the CS rate in the richest quintile by the CS rate in the poorest quintile. The same computations were replicated separately for rural and urban areas.
* Relative risk (inter-quintile ratio) computed as the ratio of rates in the richest and poorest quintiles  Values are percentages of cesarean births among all births.  States with fewer than 2000 samples births omitted

eMethods 2. Computation of the Excess, Deficit and Concentration of Cesarean Births
Following WHO latest indications, we use the values of 10 and 15% as lower and upper limits of cesarean births. We consider proportions below or above this range to represent respectively a deficit or an excess of cesarean deliveries.
The easiest way to compute any excess cesarean births would be to use the all-India rate of 17.2% and to estimate the excess above the 15% threshold as 2.1% of deliveries. This, however, ignores the considerable variations observed across regions or social groups that are documented in this paper (Table 2, eTable 2). Bihar for instance recorded a proportion of 6.2%, pointing to the presence of a significant deficit of 3.8 in a major state of India. In contrast, Kerala recorded a cesarean rate of 35.9%, pointing to a potential excess by 20.9 percentage points in this state. Cesarean rates similarly vary along a broad range according to wealth quintile or place of delivery. We also rejected the place of delivery as a criterion since the place of delivery may be as much a cause as a consequence of the delivery methods: in case of pregnancy complications, poor women may for instance be referred to a public hospital instead of delivering at home, while middle-class women in the same situation may opt for private facilities instead.
To estimate the excess or deficit of cesarean deliveries, we therefore decided to divide the NFHS-4 sample into more homogenous subgroups, using states, districts, fertility levels, and socioeconomic quintiles for that purpose. For each subgroup, we counted excess or deficit cesarean births by using the 10-15% threshold. A purely regional division (by states or districts) was abandoned as it proved unable to capture the socioeconomic variations within regional units. It may, for instance, be observed that the richest quintile in Bihar-a state with low cesarean rates-had cesarean rates close to 30%, i.e. distinctly above the 15% threshold. Inversely, the poorest quintile in Gujarat-a state with high cesarean rates-recorded a cesarean rate of 6% below the 10% threshold. The division by fertility levels also proved insufficient. The best disaggregation was obtained by dividing the NFHS-4 sample by states and socioeconomic quintiles. The resulting 180 groups (36 states x 5 quintiles) had an average 1442 births. These subgroups yielded a detailed partition of India into homogenous regional and economic subpopulations. We then computed the deficit or the excess of cesarean births in each of these 180 subgroups by using the 10-15% range.
A majority of these subgroups (97) had cesarean rates above 15%, half of them with rates greater than 30%.
When cumulated over these subgroups, the excess cesarean deliveries amount to an overall excess of cesarean deliveries of 7.0% of all deliveries. In contrast, a substantial number (65) of subgroups recorded on the contrary rates below 10% and half of them less than 5%. The overall shortfall in cesarean births among them affects 2.2% of deliveries. When we restricted our computation to cesarean rates significantly different from the 10-15% range in each subgroup, we obtained an identical number of deficit and excess cesarean births. Using the estimated number of annual births in India during this period derived from United Nations