A state representative cross-sectional survey was conducted in UP during December 2020–February 2021. This was an integrated survey in which data from households with women in the reproductive age range 15–49 years were linked to data from care providers from public facilities of the catchment area of Primary Sampling Unit (PSU) as well as data from community-level front line workers (FLWs) such as Accredited Social Health Activist (ASHA) and Auxiliary Nurse Midwives (ANM). UP has 18 administrative divisions, 75 districts, 820 blocks and more than 97,000 inhabited villages making it a large and complex setting to implement any public health program due to the diversity in the culture as well as socio-demographic and economic composition of the population.
Sample size and sampling design
The sample size for this study was determined using the divisional-level modern contraceptive prevalence rate (mCPR) of NFHS-4 (2015–16). The primary respondents for the study were CMW aged 15–49 years and a total of 12,700 CMW were required for the study. The required sample size in each division was proportionally allocated to both rural and urban areas in proportion to the population distribution within the division as per Census of India 2011. Altogether, 508 PSUs (394 rural and 114 urban) were selected across the state. Within each division, a two-stage sampling technique was adopted. In the first stage, the required number of PSU were selected using the Probability Proportional to Size (PPS) approach using a list of habituated census villages in rural areas and Census Enumeration Blocks (CEBs) in urban areas. The area covered by an ASHA, who is a female health activist responsible for providing outreach health-related services to the catchment area with a population of about 1000–1500 individuals across 250–300 households , was considered as the PSU in rural areas. If the selected village covered less than 300 households in the ASHA area, then all the households in that PSU were listed. However, in any selected PSU with more than 300 households and which was served by more than one ASHA, then one ASHA catchment area was selected randomly. Further, if there were more than 300 households served by only one ASHA in the selected village, then segments of 150–200 households were made and a maximum of two segments were selected using PPS. The same approach of segmentation was adopted in urban areas if there were more than 300 households in the selected CEB. In the second stage, approximately, 27 households in each PSU were randomly selected using a systematic random sampling approach. A household listing was done in all selected PSUs before the main survey which provided a sampling frame for the selection of households. In total 12,200 CMW participated in the study and were interviewed with a response rate of 96%.
In addition, 496 public health facilities (Primary Healthcare Centres and higher-level facilities) catering to the selected PSU were selected for a facility readiness assessment for FP services. One doctor and one staff nurse, usually engaged in providing FP services in the selected facilities, were also interviewed. A total of 476 doctors and 451 staff nurses were included in the study. The study interviewed all the available FP counsellors (n = 223) in the state. The study also interviewed FLWs-419 ASHAs and 370 ANMs serving the selected rural PSUs providing community-based FP services. Also, the study observed Village Health and Nutrition Day (VHND) in half of the selected rural PSUs. VHND is a community-based platform where outreach services like routine immunization, antenatal care, and family planning services are provided by ANM in an ASHA catchment area once a month.
Written consent was obtained from all adult participants. Assent was taken from the CMW aged 15–17 years with written consent from their husband/head of the household. All consenting CMW aged 15–49 years within a household, who stayed in the selected household on the night before the survey, were interviewed. Written consent was obtained either from the chief medical superintendent / chief medical officer / medical officer-in-charge in the selected facilities before observing the facility. Written consent was also obtained from the care providers who participated in the study.
A household questionnaire was administered to an adult member or selected women participant in the households selected under the study. Socio-economic information about the household along with the details of family members were captured in the household questionnaire. Women's questionnaire was administered to all CMW from the selected household to capture information on their demographic characteristics, reproduction, marriage and cohabitation, contraception use, fertility preferences, and program exposure. In addition, domestic violence, decision making, spousal communication on FP, self-efficacy related to FP, myths and misconceptions related to contraceptive methods, and mass media exposure were captured using globally validated standard tools [8, 14,15,16]. A contraceptive and fertility event calendar that captured month by month history of contraceptive and reproductive events including live birth, stillbirth, miscarriage and abortion for the 36 months calendar period preceding to survey (starting from January 2018) was also administered. The contraceptive and fertility event calendar also covered information on the source of obtaining the contraceptive methods or source of services for reproductive events and reasons for discontinuing contraceptive methods.
Trained female research investigators administered the questionnaire in the local language (Hindi). Handheld mobile devices with Open Data Kit (ODK) based (Android) applications were used for data collection. In addition, as part of the quality assurance mechanism, female supervisors were appointed to monitor and supervise the fieldwork, including daily spot/back-check of interviews.
We measured the current contraceptive method used through the question “Are you (or your husband) currently doing something or using any method to delay or avoid getting pregnant?”. If responded ‘no’ then they were considered as nonusers and those who responded ‘yes’ were then asked, “which method(s) are you currently using?”. Response options were female sterilization, male sterilization, IUCD- Copper-T/ Loop, depot medroxy progesterone acetate (“Antara”), oral contraceptives including levonorgestrel & ethinyloestradiol, and ormeloxifene (also known as Centchroman or “Chhaya”), levonorgestrel (known as emergency contraception), male condom, female condom, lactational amenorrhea method, other modern methods (MM) such as diaphragm, foam or jelly, and traditional methods including rhythm method, withdrawal and other traditional methods. The CMW were classified into three groups based on the current use of contraceptive methods: MM users (female sterilization, male sterilization, IUCD- Copper-T/ Loop, Antara, oral contraceptive pills, Chhaya/ Centchroman, emergency contraception pills, male condom, female condom, lactational amenorrhea method, other modern methods), traditional method users (rhythm/withdrawal /other traditional methods) and non-users. The respondents were also asked about their initial contraceptive method use by asking “Which method did you first use to delay or avoid getting pregnant?”. The CMW were also classified as those who started with any modern method, started with a traditional method and those who never used any method. We computed unmet need for family planning as the percentage of currently married women who either want to space their next birth or stop childbearing entirely but are not using contraception according to the revised definition of unmet need as described in Demography Health Surveys (DHS) globally .
The analysis included residence (rural, urban), caste group (Scheduled Caste (SC)/Scheduled Tribe (ST), Other Backward Castes (OBC) and others), religion (Hindu, non-Hindu), wealth quintile (poorest, poor, middle, rich and richest), age of the respondent (15–24, 25–29, 30–34, 35–39, and 40–49), parity (0, 1, 2, 3, and 4 +), education of the respondent and husband (< 5 years of schooling, 5–9 years of schooling, 10–11 years of schooling and 12 years of schooling or more), husband’s occupation (non-agricultural labour, salaried, cultivator/ agricultural labour, business and others), husband’s frequency of home visit (lives at home, visit home once in 1–3 months, visit home once in 4–6 months and visit home once in a year or later), fertility preferences (no more child, wanted to have another child within 2 years and wanted to have another child after 2 years), attainment of desired sex composition of the children (no sex preference, didn’t achieve desired composition for either sex, achieved desired number of boys but not girls, achieved desired number of girls but not boys, achieved desired sex composition), and FLWs interaction (frequency and discussion on FP). In addition, correct knowledge of modern reversible methods and traditional methods, reasons for continuing the method, and sources of information on the traditional methods were also included in the analysis. Definitions used to determine correct knowledge of MM and TM are provided in Additional file 1: Table S1.
Descriptive analyses such as percent distribution, mean, standard deviation (SD), and median with interquartile range were used to describe the sample characteristics. The bivariate analysis presented the association of independent variables and the use of contraceptive methods. A comparison was made in the initiation, continuation and switching of methods and future preferences among TM and MM users. Consistency/switching in the use of TM and modern reversible methods were analyzed using the 3-year contraceptive and fertility event calendar data among a cohort of TM users (n = 2796) and modern reversible methods users (n = 1406) at the beginning of the calendar month and presented using Sankey diagram. Appropriate sampling weights were used in all the analyses except for Sankey diagram and availability of family planning methods in public health facilities. The "don’t know" response and missing data were considered as a separate group while describing sample distribution. However, these two categories were not included in the bivariate tables. There were 41 (0.3%) cases for which caste, religion, and household wealth quintile information were missing and 15 (0.1%) cases of "don’t know" category in the caste variable. Preferred timing of next birth was missing for 165 (1.6%) cases. All analyses were conducted using STATA 16 .