COVID-19 vaccine hesitancy and social contact patterns in Pakistan: results from a national cross-sectional study | BMC Infectious Diseases

Respondent characteristics

A total of 3,658 people completed the survey with the planned 50/50 sampling stratification into urban and rural areas and male and female respondents. A total of 8,211 interview attempts were made, 2,192 (27%) had no or no response, and there were 1,360 refusals to participate (17%). 56 interviews were discontinued (1%), 917 (12%) respondents were ineligible for interview because stratification goals had already been met in that area, and 27 (0.3%) interviews were removed due to control issues of quality.

Table 1 shows that the age, gender and geographical distribution of respondents generally mirrored that of Pakistan, although the proportion of respondents aged over 55 and 18–35 was underrepresented in our sample.

Consequences of COVID-19 and control measures

Only 62 (1.6%) of participants reported ever testing positive for COVID-19. One in ten (365) reported that they perceived a high risk of contracting COVID-19, while 45% perceived no or low risk of experiencing COVID-19. As in fig. 2A, only 12% of respondents believe that COVID-19 poses a very big threat to them personally and 11% to their family; however, 30% believe that COVID-19 poses a very big threat to the country and 38% to the world. Just over half (51%) of respondents do not know where to get tested for COVID-19, with some variation by region and individual characteristics (Fig. 2B), and (308) 8% of respondents have ever COVID-19 test. There was a clear gap between self-isolation knowledge and behavior (Fig. 2C) – although 72% of respondents believed that self-isolation was needed more than one week after contracting the coronavirus; 66% of the 62 respondents who had ever tested positive for COVID-19 reported self-isolating for four days or less.

Finally, 17% reported household hunger in the previous four weeks (Fig. 2D), with 60% of people experiencing hunger reporting that the main cause was either a household COVID-19 infection, disease control measures COVID-19, or both. A significant proportion of respondents were affected in various ways: 25% of households experienced job loss, 30% of individuals spent their working time in home schooling or caring for others (Additional file 3 , Fig. 1A). Almost one in five (19%) of those surveyed managed to work or study from home.

Respondents reported significant economic uncertainty due to COVID-19 and containment measures. 77% reported a reduction in their income as a direct result of COVID-19 or containment measures, of these 19% reported a reduction in income of more than half. Almost all (94%) reported seeing a price increase in the past month, and 97% in the past year. To cope with income losses or increases in expenses, 25% of respondents reported borrowing, 15% drawing from savings, 4% receiving gifts from family or friends, and 3% selling assets (Additional file 3 , Fig. 1B ). In the end, the level of household expenditure on various items did not change significantly compared to before COVID-19 (Additional file 3, Fig. 1C).

Attitudes and hesitancy towards vaccines

Nearly half (48%) of participants agreed that they would get a vaccine if offered. Hesitancy about the vaccine was significant and highest in Sindh and Balochistan provinces, where only 14% and 7% of respondents strongly agreed that they would receive a vaccine if available (Additional file 3 , Fig. 2A). There is very strong evidence that swing varies by socioeconomic status (Additional file 3 , Fig. 2B ), with both in strong agreement (nonparametric test for trend p< 0.001) and strongly disagreed that they would receive a vaccine (trend p< 0.001). There is weaker evidence for a trend by age group (Additional file 3 , Fig. 2C) in strong agreement (trend p= 0.05) and strongly disagreed that they would receive a vaccine (trend p= 0.19), with older respondents being significantly more likely to be willing to receive a vaccine. There were no significant gender differences.

Although 51% of all respondents thought vaccines were safe, 46% thought they were effective and 47% thought they were important. These perceptions were much lower in Sindh and Balochistan, the two provinces with the lowest willingness to receive a vaccine if available (Additional file 3 , Fig. 3). Among those who reported hesitancy to seek vaccination, the main reason was concern about side effects for 37% of respondents, while 16% believed vaccines were not effective, 15% believed they were not at risk for COVID-19 , and 13% are against vaccines in general. Another 10% cited time or cost, 4% lack of approval from religious leaders, and 3% fear of causing infertility.

Contact models

A total of 56,455 contacts were reported, of which 14,786 (26%) were household contacts. One in three participants (29%) reported nine or fewer contacts, so we have complete information on 16,357 contacts, 77% of which were household contacts. The median number of contacts reported was 15 (median 5, IQR 3–9). Participants reported an average of 4 household contacts (median 3, IQR 2–5) and 11 non-household contacts (median 2, IQR 0–4). As shown in Figure 3, there was no significant difference in the number of contacts by socioeconomic status, gender, participant age, education level, province, or whether the respondent lived in an urban or rural area. As expected, the number of reported contacts increased with household size, as 35% (14,786/41,669) of contacts were reported within the household.

Average number of direct contacts (physical and non-physical) by (A) socio-economic status, b gender, ° С age of the respondent, e level of education, e household size, Well living in an urban or rural area, and (Z) province. Each panel shows the median, hinges (25th and 75th percentiles), and whiskers representing upper and lower neighbors. Outliers are not shown in scale boxes, they are plotted in (h), showing the distribution of the number of reported direct contacts

Figure 4 summarizes the characteristics of the 16,357 (28%) contacts for whom we have detailed information because respondents reported fewer than ten contacts in each setting, including 12,540 (85%) household contacts and 3,817 (9 %) contacts outside the household. Within the household, roughly half of the contacts are physical and are split equally between the sexes. Outside the household, 70% of all contacts are men. There was significant gender assortativeness, with men accounting for 97% of male respondents’ external household contacts, compared to 29% of female respondents’ extrahousehold contacts. Gender assortativeness is lower for household contacts, where only 46% of male respondents’ contacts are male and 59% of female respondents’ contacts are female (p< 0.001). Almost half (49%) of male respondents' contacts were physical, compared to 44% of female respondents' (p< 0.001); 92% of external contacts in the household were carried out without wearing masks, and 43% were carried out in a residential property. Among those reporting detailed contacts, most (33%) contacts lasted between 5 and 14 minutes, with a further 27% lasting between 15–59 minutes.

The characteristics of (A) household and (b) non-domestic contacts for which complete information was collected

We examine how mean reported contacts vary across respondents reporting varying degrees of vaccine hesitancy in Fig. 5. First, we find that those who strongly agree that they would receive a vaccine report a greater number of contacts (16.7 compared to 9.9, t-test p-value < 0.01), which is beneficial for case reduction by vaccination, as the reduction in disease risk from vaccination will counteract the greater number of contacts. However, we find that people who strongly agree that vaccines have positive features have significantly fewer contacts, specifically that vaccines are important (14.4 compared to 19.6, p= 0.01), effective (14.6 compared to 18.5, p= 0.07), or encouraged by their religion (14.2 compared to 22.9, p< 0.01). These sources of variation may be important because lower vaccination rates among those with more contacts will mean that existing models will overestimate the impact of vaccination.

Average number of contacts from those who answered that they strongly agreed with questions about vaccine hesitancy. Asterisks represent the difference in t-tests of mean contacts between groups: *** p< 0.01, ** p< 0.05, *p< 0.1

Figure 6 shows age-specific contact matrices, where panel A shows an asymmetric matrix directly estimated from all contacts in this study without adjusting for demographics. Uncorrected site-specific contact matrices for all contacts and detailed contacts are shown in Additional file 3 , Fig. 4 and s5 respectively. Figure 5B uses pre-COVID-19 synthetic contact matrices for Pakistan to impute contacts between children, and panel C adjusts this imputed matrix for age distribution and symmetry. Compared to the synthetic matrices of Prem et al. [12], we estimate a 9% reduction in contacts in this study. Among the contacts for which we have detailed data, we observe some evidence of age assortativity (Additional file 3 , Fig. 5C) in non-household and household contacts and parent-child interactions within the household.

Age-stratified mean number of contacts reported by survey respondents, where (A) is the uncorrected contact matrix, b the mixing matrix obtained when the estimates from Prem et al. are used to impute child-to-child contacts, and reported adult-child contacts are used to impute child-to-adult contacts, and (° С) the mixing matrix obtained at matrix (b) is adjusted for reciprocity using the age structure of Pakistan in 2020