Of the 3,000 questionnaires distributed, 1,032 (34.4%) were returned complete, and 70 were returned blank, indicating a desire not to receive a reminder questionnaire. No statistically significant difference in age, gender or job category was found between respondents and non-respondents (Table ), suggesting that the responses received reflected the demographic composition of the wider survey population. Response rates varied by occupational group, ranging from 23.5% from GPs (n = 141) to 50.3% from managers (n = 151).
Survey response rates by occupational category and analysis of non-response bias (age, gender)
The study population (Table ) comprised more females than males (ratio 2.18:1), and 57.9% of survey respondents were aged 41 or above (n = 597). The largest employment category was ancillary workers, who constituted 17.3% of respondents (n = 179); the smallest was doctors (11.8%; n = 122). HCWs who indicated some form of caring responsibility (for children aged under 16 and/or elderly dependents) constituted just over half of responses received (50.2%; n = 518), and full-time workers outnumbered part-time workers (ratio 3:1; n = 767 vs. n = 260).
Demographic characteristics of the study population and likelihood of reporting to work
Likelihood of working
Figure shows the proportion of respondents who indicated that they would be 'likely' to work in a given circumstance. Factors with the greatest potential impact on likelihood of working were illness to children (13% of respondents would continue to work in this circumstance; n = 134) and illness to partner (23%; n = 237). The potential need to work more hours than normal (60%; n = 619) or working with untrained volunteers (63%; n = 650) were reported to have the lowest potential impact on likelihood of working.
Proportion of HCWs who indicated they would be 'likely' to work in a given circumstance.
The mean likelihood score (i.e. the mean percentage of circumstances which may arise during a pandemic under which individuals stated they would be willing to work) for all respondents was 59.3%. Only 149 respondents (14.4%) indicated that they would be likely to work in all (individually relevant) circumstances (a likelihood score of 100%) (Table ). Females were significantly less likely to work during a pandemic than males (bivariate OR: 0.6; CI 0.4 to 0.9), as were part-time employees compared to full-time workers (bivariate OR: 0.4; CI 0.2 to 0.6), and HCWs with caring responsibilities compared to those without children or elderly dependents (bivariate OR: 0.4; CI 0.3 to 0.6). Across employment categories, nurses, ancillary workers and community HCWs had the lowest reported likelihood of working (bivariate OR: 0.3; CI 0.1 to 0.7; 0.5; CI 0.2 to 0.9 and 0.5; CI 0.2 to 0.9 respectively).
The number of individually relevant circumstances in which respondents stated they would be likely to work
Multivariate analysis showed that those who lived alone or who shared with friends were significantly more likely to report to work than those in households with children (multivariate OR: 3.7; CI 1.6 to 9.0 and 4.7; CI 1.5 to 15.5 respectively). GPs indicated a higher likelihood of working during a pandemic than other categories of HCW, along with respondents aged 41–50 years (multivariate OR 2.3; 1.2 to 4.4).
Predictors of reported likelihood of working under different circumstances
Table shows the demographic and employment characteristics associated with whether an individual would be potentially 'likely' or 'unlikely' to work under a given scenario. Infection of self and/or family was most strongly associated with different respondent characteristics. Gender was a significant predictor of the likelihood of working if children fell ill: females were significantly less likely to continue to work under this circumstance than males (OR 0.4; CI 0.3 to 0.7). Work type was also a significant predictor of likelihood of working under this scenario, with part-time workers less likely to work than full-time HCWs (OR 0.4; CI 0.2 to 07).
Predictors of likelihood of working under different circumstances
In the case of personal infection risk, doctors and GPs were the most likely to continue working despite the risk, compared with HCWs in other job categories (OR: 2.6; CI 1.3 to 5.4 for doctors, and OR: 4.8; CI 2.2 to 10.4 for GPs). Similarly, GPs were also significantly more likely to work despite the risk of infecting family members (OR: 3.8; CI 2.1 to 7.0) or if a partner fell ill (OR: 3.0; CI 1.6 to 5.6).
For all circumstances relating to working conditions rather than personal or family risk, job type was the only significant predictor of whether an individual was likely to agree with the survey statements. Compared to those in other HCW categories, nurses were particularly reluctant to work if they had to take on duties for which they had not received training (OR 0.5; CI 0.3 to 0.9). They were similarly unlikely to attend if asked to work at a different site to normal (OR 0.2; CI 0.1 to 0.3), as were ancillary workers (OR: 0.3; CI 0.2 to 0.5). Working with untrained volunteers or those brought out of retirement produced a mixed response on the basis of job type – nurses and ancillary workers would be reluctant to work under this scenario (OR: 0.5; CI 0.2 to 0.9 and OR: 0.5; CI 0.3 to 0.9 respectively), whereas GPs would be particularly likely to continue working on this basis (OR: 4.0; CI 1.4 to 11.9).
'Persuadability' of those with a <100% reported likelihood of working
Isolating those respondents who reported a lower probability of working on the basis of responses to the 'likelihood' circumstances (n = 883), Table shows how 'persuadable' these respondents may be towards overcoming their unwillingness or inability to work if changes to working conditions were introduced during a pandemic. The mean 'persuadability' score for the selected respondents was 69.91% (i.e. that for the suggested policy interventions relevant to them, nearly 70% of these interventions would persuade HCWs to continue working). Demographic groups with the highest persuadability scores were those in the 16–30 age group (n = 158; mean score 75.24%); community HCWs (n = 141; mean score 75.49%); HCWs living in households without children (n = 418; mean score 71.84%), and those living with parents or relatives (n = 64; mean score 74.23%). Groups with the lowest mean persuadability scores were nurses (n = 126; mean score 66.23%) and those living with friends (n = 20; mean score 54.96%).
'Persuadability' score by demographic characteristics, and likelihood of being 'persuaded' to work in an influenza pandemic.
Few of the bivariate or multivariate Odds Ratios were significant at the 0.05 level. Those that did show significant results related to households without children being over twice as persuadable to work than households with children (multivariate OR: 2.2; CI 1.1 to 4.3), and HCWs with caring responsibilities for children and/or elderly dependents reported particularly low levels of persuadability in comparison to those without dependents (multivariate OR: 0.5; CI 0.3 to 0.9).
Acceptability of policy interventions to increase attendance at work
The potential impact of each of the suggested changes to working conditions in increasing HCWs' likely attendance at work varied considerably (Figure ). Of the scenarios relevant to the 883 respondents identified as having a lower initial likelihood of reporting to work during a pandemic, the least influential interventions were shown to be the possibility of working nearer to home (50.1% agreed, n = 443); the provision of accommodation (43.1%, n = 381); the provision of transport (54.6%, n = 482), and the provision of childcare (60.8%, n = 261). The most influential interventions were the provision of vaccination for oneself and one's family (83.4%, n = 736 and 83.1%, n = 734 respectively).
Proportion of HCWs who indicated an intervention would make them 'more likely' to work.
These were followed by the provision of personal protective equipment (PPE) (77.8%, n = 687) and having employers share emergency plans with their employees (77.2%, n = 682). Interventions which would provide incentives or employee safeguards were also recognised as potentially beneficial: having employers accept liability for any mistakes made (68.6%, n = 589); being able to work flexible hours (66.7%, n = 589), receiving a top-up salary commensurate with the level of extra duties an individual may be asked to perform (66.8%, n = 590), and the provision of life/disability insurance (62.9%, n = 555).
Few overriding demographic or employment characteristics emerged as predictors of agreement or disagreement with specific policy interventions outlined in the survey (Table ). Of the 12 potential interventions investigated, none would have encouraged all HCWs within a given demographic or employment group to work. In the case of the provision of facilities to prevent infection spread, significant associations were most often found within groups who indicated a low potential acceptance of such measures. For example, ancillary workers were significantly less likely to view the possibility of working at a site nearer to their homes as acceptable in comparison to other categories of HCW (OR 0.6; CI 0.3 to 0.9). Similarly, whilst those without caring responsibilities and HCWs who lived with parents indicated a significant increase in their likelihood of working if transport was provided (OR 2.2; CI 1.1 to 4.2 and OR 2.6; CI 1.1 to 6.4 respectively), nurses in particular were identified as being significantly less likely to respond positively to this intervention when compared to other job groups (OR 0.5; CI 0.3 to 0.8).
Predictors of the influence of policy interventions on HCWs' potential decisions about working
Nurses were also significantly less likely to choose to work (compared to other types of HCW) if their employers accepted liability for mistakes (OR 0.3; CI 0.2 to 0.6), if they were offered more flexible working hours (OR 0.4; CI 0.2 to 0.7) or were paid a top-up salary (OR 0.4; CI 0.2 to 0.8). Flexibility of working hours was noted as potentially beneficial to respondents without children in comparison to those with children (OR 2.1; CI 1.0 to 4.4 and OR 0.5; CI 0.2 to 0.9 respectively). The interventions with the most significant positive associations were those where employers would offer some form of incentive to work during a pandemic.