Seasonal influenza risk in hospital healthcare workers is more strongly associated with household than occupational exposures: results from a prospective cohort study in Berlin, Germany, 2006/07

The German standing commission for immunisation, along with other authorities [1–3] currently recommends that healthcare workers (HCWs) be vaccinated against seasonal influenza. Two reasons are cited: firstly that HCWs can be a source of infection for vulnerable people under their care [1–3], and secondly that HCWs are at increased risk for contracting influenza [1]. Vaccination of HCWs should theoretically reduce the risk of influenza infection both in themselves and their patients. There is evidence to support the first reason for vaccination, the protection of patients. HCWs can transmit influenza to those under their care during both outbreaks and non-outbreak situations [4–7]. Low vaccination rates in HCWs have been associated with nosocomial outbreaks [4, 8], and higher vaccination rates with reduced nosocomial influenza incidences [7].

HCWs may transmit influenza to those under their care, but is there evidence that their occupational exposures (to patients, relatives, colleagues and the hospital environment) confer an increased risk of influenza compared to the general population? Influenza serological attack rates in HCWs of 23% (single season, [9]) and 14% (an average of two seasons, [10]) have been documented. However, as serological attack rates of influenza may vary considerably from season to season as well as from location to location, without the inclusion of a comparison group of non-HCWs neither study could demonstrate an increased risk.

Another aspect of influenza in HCWs is that many HCWs argue that they withdraw from work when they become ill with influenza-like illness to reduce their risk of transmitting influenza to their patients. However, not all serologically diagnosed influenza infections experience an influenza-like illness, and a proportion will be asymptomatic. The frequency of asymptomatic influenza infection in HCWs has been assessed in volunteer studies (33%, [11]), a cohort study (28%, [9]) and one randomised controlled study (42%, [10]).

The objective of this study was to address the question of whether HCWs in the acute care hospital setting have a higher risk of serologically confirmed influenza infections (SCII) than non-HCWs, and to assess the proportion of individuals with SCII who experience either any respiratory symptoms or an influenza-like illness.

We conducted our study in people living or working in Berlin during the influenza season of 2006/07, using a multicentre, prospective cohort design.

There were 11 study sites: three hospitals, two administrative centres (the Robert Koch Institute and Vivantes Healthcare administrative centre), four blood-donation centres and two colleges. HCWs and some non-HCWs were sought from the hospitals, but only non-HCWs were sought from all other sites.

We recruited participants through occupational health services in the hospitals and one administrative centre; through direct recruitment at blood donation centres; and through active recruitment during site visits at the other study sites (non-healthcare workers from one administrative centre and two colleges).

On recruitment before the influenza season, consenting participants gave a single serum sample and completed an exposure questionnaire. Details recorded included age, sex, type of employment, risk factors for influenza, smoking status, and vaccination status. Where participants had been vaccinated fewer than 14 days before the sample was taken, or were found to have been vaccinated shortly after the initial sample was taken, a second sample, taken at least 14 days after vaccination, was sought through direct recall or through occupational health.

Blood samples were refrigerated, then transported within three days to the national reference laboratory for influenza at the Robert Koch Institute in Berlin, where they were centrifuged and frozen.

We found no significant association between being a healthcare worker in acute hospital care and serogically confirmed influenza infection. Instead we identified household contacts, in particular children, and car ownership, as important risk factors for SCII. 30% of participants with SCII reported no symptoms, and only around one-quarter of those reporting influenza-like illness also had SCII.

We did not demonstrate a strongly increased risk of influenza infection in HCWs in acute hospital care. As HCWs do appear to have an increased risk for ARI (OR = 1.3) it would be plausible to expect an effect of similar order for influenza infections. Although study limitations might have led to a failure to detect a true increase, it is unlikely that we have missed a large difference in the risk of influenza.

Why might HCWs in hospital not have a higher risk of influenza? The prevalence of infectious influenza in patients may have been low, either due to the absence of influenza patients or because patients admitted later in the course of illness might have been less infectious. Also, infection control measures such as use of personal protective equipment and individual or cohort isolation might have reduced the risks of infection in HCWs. Lastly, prior immunity may have played a role. HCWs were less likely than non-HCWs to be susceptible (preseason titer of less than 40) to A/California/07/2004 (H3N2), which was the A/H3N2 strain of the previous season and related to A/Wisconsin/67/2005 (H3N2), the dominant strain in the 2006/07 season.

As most influenza patients admitted to hospitals are children or older people, one might expect to find an increased risk for SCII in HCWs who work with these patient groups. Our finding of an elevated odds ratio for HCWs with more frequent contact with child patients, albeit not significant in this setting, would be consistent with this supposition. Another study focusing on this hypothesis is needed to investigate it further.

Household contacts, and in particular children in the home, were the main significant risk factors identified in this study. This finding is similar to that in another droplet-transmitted infection, meningitis, where household contacts constitute by far the most important risk group [13]. We found a strong dose-response relationship for child household contacts. The role of children as the main sources of influenza transmission has been suggested in several studies [14–17]. Adult household contacts may play a role analogous, but possibly less marked, to that of children.

Whilst public transport usage was not associated with SCII, car ownership was a significant risk factor, albeit only in households without children. This was unexpected, as car ownership was only included as a counterpart question to public transport usage. For this reason, further details on passengers, frequency or duration of car use were not obtained in the study.

Possession of a car has not previously been recognised as a risk factor for influenza acquisition. However, sharing a car does involve prolonged close contact in an enclosed space and has been linked to transmission of one airborne pathogen [18]. In one observational study on risk factors for SARS in China "using a taxi more than once a week" was identified as independent risk factor for SARS infection with an increased odds ratio close to significance (p = 0.07). Furthermore, this variable was kept in the multivariate model while "riding a bus" and "taking the subway" (who were significant in bivariate analysis) were not [19]. Ownership of a car may also be a marker for having a greater number of social contacts, perhaps including children. Car users are also exposed to air pollution from other vehicles, and this may predispose users to respiratory infections. Despite the unexpected nature of this finding, it is possible that car usage is indeed a risk factor and should be explored through further studies.

We did not find an increased risk of influenza in the 58% of participants who were regular users of public transport. Public transport usage has been cited as a possible risk factor for influenza infection, particularly in the context of pandemic planning [20]. In a recent international survey of precautionary behaviour for pandemic influenza, 75% of respondents said that they would avoid public transport [21]. Both rail and air travel have been associated with respiratory infection transmission [22, 23]. Whilst it is possible to contract influenza on public transport, our results suggest that, at least in Berlin, using public transport does not increase a person's risk of influenza. In Berlin, crowding on public transport is infrequent and not extreme, so this result may not be generalisable to cities with different conditions.

Vaccine uptake was relatively high both in HCWs (42%) and non-HCWs (32%). Reasons for this high HCW uptake include the participation of a hospital from former East Germany where vaccination uptake in hospitals is still found to be higher than in West German hospitals. In non-HCWs, the participation of a federal public health institute where influenza vaccination is freely offered to employees may account for the high vaccination rate. High vaccination uptake limited the number of SCII and reduced the study power. Vaccination effectiveness was 50% against SCII and reached 73% for a more severe outcome, SCII with reported ILI.

Only 28% of participants with SCII reported an ILI, the majority reporting no symptoms or more minor respiratory illness. The proportion of asymptomatic infections (30% or more) was substantial and similar to that reported by Elder (28%,[9] and consistent with an estimated proportion of 33% obtained from pooled challenge studies [10]. In addition, only 26% of persons with an ILI had evidence of an influenza infection. Thus, the ILI-syndrome was a poor marker for influenza infections. Unvaccinated HCWs who intend to withdraw from work when they become ill with influenza-typical symptoms may overlook many (symptomatic or asymptomatic) influenza infections, putting patients at unnecessary risk.

The association of HCW status and ARI may reflect the fact that HCWs communicate with or care for a large number of people including patients, colleagues, and relatives, and are thus exposed to a wide variety of respiratory pathogens. The reduced risk for ARI in people aged over 50 years might be explained by their cumulative immune experience or a lower contact rate. The apparent protective effect of smoking could be due to lower ascertainment of ARI in this group, where a background of smoking-related symptoms may have masked the onset of an additional, mild infectious respiratory illness. Conversely, respiratory symptoms due to non-infectious causes may be over-reported in non-smokers.

What significance does this study have for the current H1N1v pandemic? As pandemic influenza should result in a higher attack rate than seasonal influenza, the prevalence of influenza in hospital inpatients and staff is likely to be higher than for seasonal influenza, so increasing the risk of exposure and infection for HCWs compared to that described here. For the same reasons, the risk in HCWs and non-HCWs from their household contacts would also be higher than for seasonal influenza. Therefore even if a repeat study during pandemic conditions identified a significant occupational risk for HCWs, household exposures might still be more strongly associated with influenza infection.

This study was subject to a number of limitations. As recruiting at the hospital and one administrative study site was done through the occupational health departments, only a small proportion (fewer than 10%) of the HCWs at these sites were enrolled in the study. Selection bias could have concealed any actual relationship between HCWs status and influenza infection, if HCWs with a lower risk of influenza than their non-recruited colleagues were to have enrolled, or if non-HCWs with a higher risk of influenza enrolled. It is possible that HCWs with a higher number of patient contacts and, therefore in theory a higher risk of influenza, would be less likely to participate due to pressure of work. Alternatively, HCWs who were concerned about their higher risk of influenza may have protected themselves better and may have been more likely than other HCWs to participate, which would tend to decrease the relative risk in this group.

Loss of recruited participants, mostly due to losses to follow up (despite repeated attempts to contact) and to the timing of vaccination, reduced the effective power of the study and may have worsened selection bias, although excluded participants did not differ significantly with respect to variables found to influence the outcome.

Although there were differences between the two comparison groups with respect to sex, recent immunisation, smoking and use of public transport, the logistic regression methods should have adjusted for these where they had an effect on the outcome (in particular immunisation). The lower average age and higher proportion of females in the HCW group is likely to be due to recruitment among trainee nurses in the larger hospitals.

Serological testing for influenza infection alone, rather than molecular testing methods, may have underestimated the true number of influenza infections [24]. This, along with the relatively high vaccination uptake, will have reduced the number of SCII and thus the effective study power. With an attack rate of 10% in non-healthcare workers, the sample size of 736 analysed would not have been sufficient to detect a relative risk below 1.8 (80% power, 5% significance).

We explored the possibility of influenza clustering due to site-level group influences or localized influenza outbreaks. Analysis of symptom onset dates revealed no evidence of separate site-specific outbreaks, but instead the epidemic curves at each study site followed the overall trend. In addition the multilevel analysis suggested that there were no significant group-level influences on the outcome.

Our study results suggest that HCWs in an acute hospital care setting are at no higher risk of influenza than the general public, or that if they are, the increased risk is modest. Household contacts, particularly children, play an important role with individuals with three or more children being at highest risk. Use of public transport does not increase the risk of influenza, whereas (in the absence of household children) car ownership does seem to significantly increase the risk for influenza infection, although the mechanism is unclear. Further research would help to clarify the role of household contacts of different age groups, the relevance of car ownership, and whether subgroups of hospital HCWs - or HCWs in other settings, such as in primary care, - are at increased risk for influenza infection. Finally, the ILI-syndrome is a poor marker for influenza infection, suggesting that HCWs cannot rely on this syndrome if they wish to confidently protect their patients but, should instead be vaccinated.