The study participant/household selection is shown in Fig. 1. Out of 1200 randomly selected, we excluded 437 patients: those who had their phones switched off or not available, 30 with wrong contact numbers, 100 who refused to participate, seven who could not speak in English or Tamil and 22 who had died and their family members declined to participate. Seven hundred and forty-seven households were included for the final analysis.
Household characteristics of COVID-19 patients
The general characteristics of households analysed in this study are shown in Table 1. The number of people per household ranged from 1 to 16, with a median (IQR) of 4 (2). The number of COVID-19 patients in a household ranged from one to seven, with a median (IQR) of one (1). Nearly half (42.3%) of the participating households were from districts with very high HDIs, and only 2.5% were from districts with low HDIs. More than half (51.1%) reported having unstable employment. One-third of the households (66.5%) had only one earning member, and more than a quarter (31.3%) of the households were living on rent. A quarter of the households (25.3%) had a family member over 65 years of age, and one-third of the households (34.3%) had a family member with any comorbidity. Less than half of the households (43.2%) had only one case of COVID-19. Nearly half of the households included had visited health facilities more than once and had had an episode of hospitalization. About one-third (30%) of households have used a private hospital at least once. One in every two households was facing OOPE and distress financing. A severe COVID-19 patient was found in three out of every twenty households. One in eight households self-reported of having some form of insurance towards health; among those, 40% of the households received any co-payments or reimbursements.
COVID-19 patient characteristics
The mean age of the participants was 41.04 ± 17.69 years, and 52% were males. 4% of the males reported smoking tobacco, and 8% reported using alcohol. 16.4% of the patients were asymptomatic. Fever (70%) was the most prevalent symptom, followed by cough (41%), myalgia (34%), sore throat (29%), and runny nose (28%). More than one-third of the patients reported a loss of smell (37%) and loss of taste (39%). Only 13% of patients reported shortness of breath. Diabetes (17.5%) and hypertension (13.2%) were the most commonly reported comorbidities. One-fifth of the patients had been hospitalised at least once in private hospitals (19.5%) and government hospitals (17.35%). The average number of days spent in a private hospital was higher, with a mean hospital duration of 5.86 ± 5.15 days, median (IQR) 3.5 (6) compared to a government hospital with a mean hospital duration of 4.91 ± 3.77, median (IQR) of 4 (6). The mean ICU duration was 6.27 ± 5.40, with a median (IQR) of 5 (4) days.
Out of pocket expenditure (OOPE)
The mean (95% CI) OOPE per household was INR 122,221 (92,744–151,698) [US $1643 (1247–2040)], and the median (IQR) values were INR 3,500 (41,300); [US $47 (555)]. The mean (95% CI) household’s annual income and total health expenditure were INR 397,566 (362,167–432,964), INR 116,622 (87,558–145,686) [US $5346 (4870–5822), $1568 (1177–1959)], and the median (IQR) values were INR 300,000 (276,000), INR4,000 (413000); [US$ 4034 (3711), $54 (5533]. The mean (95% CI) direct health expenditure and direct non-medical expenditure were respectively INR 115,356 (86,143–144,569), INR 8,292 (6,352–10,231); [US$ 1551 (1158–1944), 111 (85–138)] and the median (IQR) values were INR 1,350 (41,000), 0 (0); [US $18 (551), 0].
Catastrophic health expenditure (CHE) and its determinants
Households with CHE due to COVID-19 illness was 25.84% (95% CI: 22.69–29.0), with 36.36% (95% CI: 28.38–44.34%) in the lowest income quintile and 19.84% (95% CI: 12.78–26.90) in the highest income quintile. Households in the rural areas, coming from the lowest income quintile, having only one earning member, had a higher proportion of CHE. Households with the head of households having an unstable employment status or education below the elementary level had a higher proportion of CHE. Households having at least one person over 65 years of age, one person with comorbidity or one person with more than one COVID-19 infection had higher proportions of CHE. Households with at least one episode of hospitalization and at least one private hospital visit experienced higher CHE proportions. (Table 1).
The main drivers for CHE was severe COVID-19 infection, with one severe COVID-19 case per households having an OR of 6.63 [(3.75–11.73), p-value < 0.001], and two or more severe COVID-19 cases having an OR of 12.58 [(7.06–22.42, p-value < 0.001]. Other predictor variables included admission to a private hospital [OR = 15.58 (10.49–23.13), p-value < 0.001], at least one episode of hospitalisation [OR = 11.85 (7.68–18.27), p-value < 0.001], facing OOPE [OR = 11.0 (5.46–15.22), p-value < 0.001], having a family member over 65 years of age [OR = 2.89 (2.03–4.12), p-value < 0.001] and households with at least one comorbid individual [OR = 3.38 (2.41–4.75), p-value < 0.001]. Households that self-reported having a health insurance [OR = 2.75 (1.76–4.28), p-value < 0.001] and households with insurance who have received co-payments/reimbursements also had an increased odd of facing CHE [OR = 5.8 (2.81–11.97), p-value < 0.001] (Table 2).
A multivariate logistic regression model with an enter method was used to determine the effect of number of severe COVID-19 cases in a household facing CHE. The variables: number of COVID-19 patients per household, having a family member over 65 years of age, household with at least one comorbid individual, household’s income quintile, households receiving Insurance payment, and households with at least one private hospital visit was adjusted in the model. The model was statistically significant, with Hosmer-Lemeshow goodness of fit, p > 0.05. The model explained 50.25% (Nagelkerke R2) of the variance in facing CHE and correctly classified 84.2% of cases. In multivariate analysis, having one severe COVID-19 patient increases the risk of CHE by fourfold [AOR = 4.33 (2.13–8.34)] and having more than one COVID-19 patient increases the risk by five times [AOR = 5.10 (2.42–10.74)] (Supp Table 2).
Sensitivity analysis for CHE thresholds
Using 10% and 25% thresholds for subsistence income for calculating CHE, 36.81% (33.34–40.28) and 29% (95%CI: 25.79–32.31) of the households faced CHE. We also found that 196 households (26.24%, 95%CI: 23.3–29.40) had faced CHE due to COVID-19 while considering CHE as total health expenses of more than 10% of total annual income.
Concentration curves and indices
The concentration curve and index value of 0.384 reveal inequality in household income. There was more inequality in household subsistence income when household income was taken into account. This was shown by an index value of 0.456. There was minimal inequality for severe COVID-19 disease in households and for annual or subsistence income. There was relative equality in OOPE based on household income. An index value of 0.576 showed a severe inequality in OOPE in relation to the number of severe COVID-19 patients in the household. 80% of the OOPE was made up of households with a high proportion of severe COVID-19 cases (Supp Fig. 1, Supp Table 1).