*3.3. Negative Binomial Regression of COVID-19 Prevention Behaviors*

The results from the negative binomial regression analysis (Table 4) showed that after controlling for the other HBM constructs and sociodemographic factors (also referred to as modifying factors in the HBM), only one construct predicted the prevention behavior. The "perceived benefits of preventative actions" score was positively associated with the dependent variable, the individual behaviors/actions concerning COVID prevention (incidence rate ratio, (IRR), 1.16; confidence interval, CI, 1.061–1.276; *p* < 0.001). The IRR 1.16 indicates the estimated ratio of change in prevention behavior score, given a one unit increase in the perceived benefits of prevention score after controlling for all other variables in the model. The other constructs of the HBM, including perceived susceptibility to disease, perceived severity of the disease, perceived benefits of preventative actions, perceived barriers of preventative action, and cues to action had no significant association at *p*≤ 0.05 with the individual behaviors/actions concerning COVID-19 prevention.


**Table 4.** Negative binomial regression of the COVID-19 prevention behaviors (scale score) \*, Depalpur City, Pakistan, 2020.

\* The COVID-19 preventive behaviors/actions scale score consisted of a sum of 12 items, each coded as yes = 1 and no/don't know = 0, with potential ranges from 0 to 12. Abbreviations: IRR, Incidence Rate Ratios; CI, Confidence Interval; LL, Lower Limit; UL, Upper Limit; RC, Reference Category. Note: Bolded font for *p* indicates the significance of differences in mean at *p* < 0.05.
