Skip to main content

Table 5 Description of study variable: Definition/Coding

From: Out-of-pocket expenditure on maternity care for hospital births in Uttar Pradesh, India

Name of the variable Definition/Coding
Outcome variables
Maternity expenditures Maternity expenditures were measured as a linear variable using three broad categories: ANC expenditures, delivery expenditures, and total maternity expenditures. Each of these broad categories was derived from the five disaggregated expenditure components: doctor’s fees, medication costs, transportation costs, the cost of hospitalization and room rent.
Catastrophic maternity expenditure [CME] CME was measured in relation to income. However, there is no accepted single measure of catastrophic spending in the health financing literature. Some studies measure catastrophic spending in relation to the budget share (Wagstaff and van Doorslaer 1993; Russell 2004; Pradhan and Prescott 2002); while others argue that catastrophic spending should be measured in relation to capacity to pay, such as household expenditure net of food spending (Xu et al. 2003; Garg and Karan 2009; Raban et al. 2013). Nonetheless, all measures suggest that when households expand a large proportion of their budget on health care, they often forgo other goods and services, which can have the negative implications for their living standards (Raban et al. 2013). We use the most popular approach, which defines the medical spending as “catastrophic” if it exceeds some fraction of a household’s income or total expenditure in a given period, typically 1 year (Berkiw 1986; Wagstaff and van Doorslaer 1993; Wagstaff and van Doorslaer 2003; Russell 2004). We use income rather than consumption expenditure in the denominator. Since, there is no acceptable definition for cut-offs, we calculated catastrophic spending for different cut-off points (i.e. less than 15% level, 15 to 25% level and above 25%). This assessment of catastrophic spending at multiple cut-offs provides us with both the incidence as well as the intensity of catastrophic spending (O’Donnell et al. 2005).
Predictor variables: Socio-economic
Age (in years) Age of the women is categorised into three groups: less than 25 years, 25–29 years and above 30 years. This classification was done by keeping in the mind both the distribution of the sample across the ages and also considering the ideal ages of childbearing for better pregnancy outcomes. While for regression analyses we used age as linear variable.
Education level of women The educational status of women is coded into three categories: up to high school, intermediate and under graduation and above. These groups are classified in such as way that they have a distinct effect on the nature of health care spending.
Religion The presence of other religions in Uttar Pradesh is nearly negligible which is also reflected in our sample. Therefore, we have classified our sample into Hindu and Muslim.
Social group The social groups are recoded into three groups: Scheduled Caste (SC)/Schedule Tribe (ST), Other Backward Castes (OBCs) and General Castes. A system that allows social hierarchal division of people in India.
Family’s annual per capita income The collection of income of the household is always a challenging exercise. In the case of this survey, it is, even more, difficult because it was at the hospital setting. However, the 82% of our sample are coming from Urban and Semi-urban areas and more than 70% of the sample is from the non-primary sector as a principal occupation. Within primary sector (30%), 18% of them are daily wage labourers. Therefore, in total 88% (around 202 out 230) of our respondents have not faced any problem in reporting their daily or monthly or annual income. However, for those who stated their husbands/her own occupation as cultivation and business, we have asked women to take the help of family members (who were present with her at the time of survey, mostly the husband) in reporting the annual share of couples’ income in the total income of the household in past 12 months if in case they are residing in joint families. Thus, we have collected daily (a reference to last working day) or the monthly (reference to the past 1 month)/Yearly (a reference to the past 1 year) income of women and her husband but later it is aggregated to estimate the annual per capita income of the family members. Based on the distribution of family’s annual per capita income, we have categorised the income into four groups: Below Rupees (Rs.) 24,000 (Below $390), Rs. 24gmht000 to 60,000 ($390 to $976), Rs. 60,001 to 100,000 ($976.1 to $1626), Rs. Above 100,000 ($1626). The first category is near to below poverty line according to the World Bank definition of poverty line prior to 2015 i.e. less than 1.00$ per day. While for regression analyses we used income as a linear variable.
Place of residence Place of residence is recoded into Urban and Rural area.
Social network We collected information on social networks of the family with any medical person working in the hospital they have visited. The answer was recorded as “Yes” if they have social connection otherwise coded as “No”.
Any mass media exposure Mass media exposure is a composite variable. It is computed based on women’s exposure to print media (newspaper/magazine), and electric media (television, radio, and cinema). Exposure to any of these media sources was denoted “Yes” Otherwise “No”.
Predictors (Demographic/Public health/Policy)
Number of previous pregnancies The number of previous pregnancies is a continuous variable that was recoded into 0 “for the first time pregnant” 1 “if the current pregnancy is the second” 2 “for more than two time pregnant women”. While for regression analyses we used it as linear variable.
Current pregnancy registered with the ANM If the women were registered, their current pregnancy with ANM was coded as “Yes” otherwise “No”.
Number of ANC visits The number of ANC visits is a continuous variable that was recoded into less than 3, 3 to 8, and 9 and above ANCs. Minimum three ANC visits are a part of World Health Organization standards of Full ANC. In case of regression analyses we used it as linear variable.
Quality of ANCs The quality of the ANC is a composite indicator computed from the information on medical checkups conducted during ANCs. ANC cost is very sensitive to the kind of medical tests conducted during ANC visits that are in turn indicating the quality of medical check-up during ANC visits. We consider six check-ups and advice: weighing, blood pressure, blood test, urine test, abdomen check-up and ultrasound test, advice on food and personal care. Out of seven, no or less than 3 checkups is considered as “low quality” ANC, 3 to 4 checkups is considered as medium quality ANC, 4 and above checkups is considered as a high-quality ANC check-up.
Place of ANC Place of the ANC is coded as Government health center/ Hospital and Private clinic/ hospital.
Distance to ANC clinic/ hospital Distance to ANC hospital has a huge role in expenditure on transportation. We coded distance to ANC clinic/hospital into less than 3 km (km), 3.1 to 5 km, 6 km and above. For regression analyses we used it as linear variables.
Decision on institutional delivery The decision on institutional delivery is coded as self/family planned, the doctor advised, rushed to the hospital due to Emergency Obstetric Care (EmOC). The rationale behind this classification is that if the delivery is pre-planned by a woman or her family often incurs less expenditure than doctor advised or rushed to a hospital in EmOC.
Type of delivery Type of delivery has a huge impact on delivery cost. It is recoded as normal and caesarean or forceps.
Type of hospital for delivery Type of hospital for delivery is coded into (I) Government Hospital, which is purely government, (II) a Government-aided hospital, which is government hospital, but charges a nominal fee, and (iii) a Private maternity hospital.
Received JSY If women received JSY, it is coded as “Yes” Otherwise “No”.