Stigma and its associated factors among patients with COVID-19 in Dhaka City: evidence from a cross-sectional investigation

Study design and participants

This cross-sectional study was conducted among COVID-19 survivors living in Dhaka city in Bangladesh from 19 January to 30 June 2021. Study participants included patients with COVID-19 aged 18 years or older who had been hospitalised or had stayed at home and were tested negative 15 days to 6 months before data collection. Patients with speech and/or hearing impairments were excluded so that we might not have difficulty gathering information from those patients.

Procedures

For this study, we recruited two types of patients with COVID-19–hospitalised and non-hospitalised, who had recovered over 15 days to 6 months, from 1^st November 2020 to 15^th April 2021. We chose hospitalised and non-hospitalised patients to see differences in the prevalence rate of stigma between the groups. Because of administrative difficulties and confidentiality issues, we could collect the medical records of two hospitals in Dhaka City. One was a government COVID-19 dedicated hospital, and the other was a private hospital with a COVID-19 unit. We collected the medical records of 500 inpatients from these two hospitals. We selected 192 patients from these 500 hospitalised patients following the simple random sampling technique and the other 192 non-hospitalised patients using the snowball sampling technique. We communicated with all the selected patients over the phone and fixed the interview time at least 3 days before the interview. As some respondents were not interested in joining the in-person interview for fear of contracting COVID-19, we also employed the telephone interview mode. We recruited data collectors with medical science and social science backgrounds and gave them 2-day training before data collection. However, data collection was done through in-person and telephone interviews (282 vs. 102) using a semi-structured survey questionnaire. Each interview lasted for approximately 20 minutes. Almost all the in-person interviews were conducted in the respondents’ houses, and a few took place in their offices. Safety measures such as physical distancing and face mask use were strictly maintained during the time of in-person interviews. However, the respondents were not given any financial benefits for participating in the study.

Measures

COVID-19-related stigma scale

In this study, stigma was measured using a 15-item stigma scale questionnaire. Dar et al. (2020) adapted the questionnaire from the Ebola-related stigma scale (Overholt et al., 2018) for use among patients with COVID-19 in Kashmir, India. The Ebola-related stigma scale was derived from Berger’s HIV stigma scale (Wright et al., 2007), a validated measure of stigma for people living with HIV. This stigma scale includes four subscales-enacted stigma (three items), disclosure concerns (two items), internalised stigma (three items) and perceived external stigma (seven items). Each item is rated on a 4-point Likert scale (0 = strongly disagree, 1 = disagree, 2 = agree, 3 = strongly agree), with total scores ranging from 0 to 45. Higher scores indicate greater experiences of stigma. The respondents scoring greater than or equal to the median score of 23 were classified as having experienced stigma. In contrast, those scoring less than the median score of 23 were classified as not having experienced stigma (Sima, Belachew & Abebe, 2019). The final COVID-19-related stigma questionnaire that we used in our study is shown in Table 1. The English version questionnaire was translated, back-translated into and from Bengali and pretested among 20 COVID-19 survivors for our research. The Bengali version COVID-19-related stigma scale had good reliability (Cronbach’s alpha = 0.857).

Sample size calculation

Probability sampling was used to calculate the sample size of this study. The sample size was calculated using the following formula:

n = $\frac{z^{2}pq}{d^2}$; n = $\frac{{1.96}^2\times (1-0.50)}{{0.05}^2}$ = 384.16 ≈ 384

Here,

n = Desired sample size

z = Standard normal deviate = 1.96 at 95% confidence interval

p = Prevalence estimate = 50% = 0.50 as there was no reported prevalence of stigma among COVID-19 patients in Dhaka city at the time when this study was designed.

q = 1 − p = 1 − 0.50 = 0.50

d = Precision of the prevalence estimate = 5% = 0.05

The calculated sample size of this study was 384.

Statistical analysis

Data analysis was performed using SPSS statistical software version 25. For descriptive statistics, frequency and percentage were used to describe the basic characteristics of the respondents. Median and interquartile range (IQR) were calculated for the stigma scale and subscales. We checked for normality of the stigma scale and subscale scores using the Shapiro-Wilk test, and the data did not meet normality (p < 0.05). Therefore, we used the nonparametric Mann-Whitney U test and Kruskal-Wallis test to compare stigma scale and subscale score differences between groups. Binary logistic regression analyses were done with variables associated with the total stigma score at p < 0.20 in the univariate analysis. These variables included sex, age, education, occupation, monthly family income, and hospitalisation status. The backward stepwise logistic regression method was used to identify the predictors of stigma. The associations between predictors and outcome variables were presented as adjusted odds ratios (ORs) and 95% CIs. The regression model met the requirements of both the omnibus test (p-value < 0.05) and the Hosmer-Lemeshow test for goodness of fit (p-value > 0.05). All statistics were tested using a two-sided test, and a p-value of <0.05 was considered statistically significant.

Ethical considerations

Formal ethical approval was obtained from the national research ethics body, Bangladesh Medical Research Council (BMRC) (Ref: BMRC/HPNSP-Research Grant l2020-2021 I 52(1-47). We did not disclose the identities of the two hospitals because the authorities assumed that this could put these hospitals at risk of stigmatization. Written informed consent was taken from those who joined the in-person interview, and verbal informed consent was sought from the respondents participating in the phone interview. Confidentiality of the information given by the respondents was strictly maintained. The informed consent form clearly explained the aims and procedures of the study, risks and benefits associated with participation, their right to voluntary participation and their right to withdraw from the study, and anonymity and confidentiality of their data.

Basic characteristics of the respondents

The distribution of respondents by basic characteristics is shown in Table 2. Most respondents (57.8%) were males, whereas 42.2% were females. 43.0% of the respondents belonged to the 18–39 age group, 35.7% belonged to the 40–59 age group, and 21.4% belonged to the 60+ age group. The vast majority of the respondents (80.2%) were married, 18.2% were unmarried and only 1.6% were divorced/widowed. Overall, the educational status of the respondents was very high. More than three-fourths of the respondents (78.9%) completed Bachelor’s degree and more education, 12.0% completed higher secondary education, 4.9% completed secondary education, and 4.2% had no education or completed primary education. Most respondents were service holders (41.1%), followed by housewives (22.4%). By religion, 92.7% of the respondents were Muslims, 6.5% were Hindus, and a few were Buddhists. About half of the respondents (50.3%) had a monthly family income of BDT <50,000, 33.1% had a monthly family income BDT 50,000 to 99,000, and 16.7% had a monthly income of BDT 100,000 and above. By hospitalisation status, 50.0% of the respondents were hospitalised and the remaining 50.0% were non-hospitalised during the time they were COVID-19 positive.

Comparison of stigma levels between different groups

Table 3 shows the median scores of stigma scale and subscales among the respondents by basic characteristics. The median (IQR) scores of stigma, enacted stigma, disclosure concerns, internalised stigma and externalised stigma were 23 (21–26), 5.5 (4–6), 3 (2–4), 4 (3–5) and 12 (10–13), respectively. More females than their male counterparts experienced stigma (median = 24, IQR = 22–28), enacted stigma (median = 6, IQR = 5–6), disclosure concerns (median = 3, IQR = 2–4), internalised stigma (median = 4, IQR = 3–5) and externalised stigma (median = 12, IQR = 11–14). Stigma, enacted stigma and disclosure concerns were most prevalent among the respondents from the 18–39 age group (median = 25, IQR = 22–27; median = 6, IQR = 5–6 and median = 3, IQR = 2–4, respectively), whereas internalised stigma was reported highest by the respondents from both the 18–39 age group (median = 4, IQR = 3–5) and the 40–59 age group (median = 4, IQR = 3–5), and externalised stigma was most prevalent among the respondents from the 18–39 age group (median = 12, IQR = 11–13), the 40–59 age group (median = 12, IQR = 10–13) and the 60+ age group (median = 12, IQR = 10–13). By the education category, stigma was highest among those who completed higher secondary education (median = 23, IQR = 21–25.2) and Bachelor’s degree and above (median = 23, IQR = 21–26). In contrast, enacted stigma was experienced most by those who completed Bachelor’s degree and above (median = 6, IQR = 5–6). By occupation, stigma and externalised stigma were reported highest by health professionals (median = 25, IQR = 22–28.7 and median = 12.5, IQR = 11–14, respectively). On the other hand, enacted stigma was faced most by service holders (median = 6, IQR = 4–6), housewives (median = 6, IQR = 5–6.2) and health professionals (median = 6, IQR = 5–6). Stigma was experienced most by the respondents whose monthly family income was BDT 100,000 and above (median = 24.5, IQR = 21–27), and enacted stigma was reported highest by the respondents whose monthly family income was BDT 50,000–99,000 (median = 6, IQR = 5–6) and BDT 100,000 and above (median = 6, IQR = 5–6). By hospitalisation status, more non-hospitalised patients compared to their hospitalised counterparts were found to experience stigma (median = 24, IQR = 22–27), enacted stigma (median = 6, IQR = 5–6), disclosure concerns (median = 3, IQR = 2–4) and externalised stigma (median = 12, IQR = 11–13).

Individual responses to stigma-related items

Individual responses of the respondents to stigma-related items are shown in Fig. 1. 84.9% and 72.9% of the respondents agreed/strongly agreed with two out of three items of the enacted stigma subscale (‘hurt by how people reacted to my coronavirus disease’ and ‘I have stopped socialising with some people’, respectively). 86.2%, 79.7%, 74.5%, 68.0% and 62.8% of the respondents agreed/strongly agreed with five out of seven items of the externalised stigma subscale. On the other hand, the least proportion of respondents agreed/strongly agreed with two out of three items of the internalised stigma subscale– ‘a person with coronavirus disease is disgusting’ (18.8%) and ‘I feel that I am not as good a person as others’ (20.6%).

Experience of stigma among the respondents

As shown in Fig. 2, 53.1% (n = 204) of the respondents experienced some level of stigma when they were COVID-19 positive, whereas 46.9% (n = 180) of the respondents did not experience stigma when they were COVID-19 positive.

Predictors of stigma

The binary logistic regression analysis results presented in Table 4 describe that the females had a 3.24 times higher chance of experiencing stigma than their male counterparts (95% CI [2.039–5.163]; p < 0.001) when they were COVID-19 positive. The respondents from the 60+ age group and the 40–59 age group were 63.0% (95% CI [0.199–0.696]; p = 0.015) and 48.0% (95% CI [0.311–0.880]; p = 0.002) less likely to experience stigma than those from the 18–39 age group. By hospitalisation status, the non-hospitalised patients had 1.67 times higher odds of facing stigma than the hospitalised patients (95% CI [1.033–2.727]; p-value = 0.037).