The United States Centers for Disease Control and Prevention advocate annual influenza immunisation for all healthcare workers in contact with vulnerable patients.1 The health departments in the United Kingdom, however, advise immunising only people with risk factors.2 Little evidence exists to support or refute a policy of immunisation for such healthcare workers, and, although influenza outbreaks have been documented, epidemiological data concerning influenza in healthcare workers are lacking. We aimed principally to determine the incidence of influenza in a cohort of healthcare workers. As prevention of cross infection is one of the main arguments in favour of immunisation of healthcare workers, we also estimated the proportion of asymptomatic infection by linking recall of illness with serological results.
Subjects, methods, and results
The study population consisted of all 970 healthcare workers at four acute hospitals in Glasgow who had serum stored for a routine post-vaccination test for antibody to hepatitis B between 1 February and 26 October 1993. This group was likely to be representative of healthcare workers in contact with patients as internal audits have shown that over 80% of targeted staff complete the hepatitis B vaccination programme. After the influenza epidemic (late October 1993 to end of January 1994) we invited these healthcare workers to provide a further blood sample and complete a questionnaire on their history of influenza and respiratory infection between the end of October 1993 and 1 February 1994. In all, 163 subjects were excluded as they had resigned or were on long term sick leave or maternity leave. Of the remaining 807 subjects, 602 (75%) agreed to enter the study and provided the blood sample during six weeks beginning 1 February 1994.
Analysis of the 602 subjects showed that their age, sex, and occupation were consistent with those of staff offered hepatitis B vaccination, and over 90% of the subjects had regular contact with patients. Further exclusions from analysis were due to influenza vaccination (20 subjects), insufficient serum (25), and inability to trace first serum sample (39).
We matched the remaining 518 samples with baseline stored serum samples and tested for antibodies to influenza A and B by single radial haemolysis using the method of the National Institute for Biological Standards and Control, with antigens derived from the 1993–4 season. This test is known to compare favourably with the standard haemagglutination inhibition test3; a 50% increase in reactivity between two samples is diagnostic of infection. Questionnaire responses and serological findings were analysed with the χ2 test.
Overall, 120 samples (23.2% (95% confidence interval 19.6% to 26.8%)) had a significant rise in titre due to influenza. Type A influenza occurred in 107 samples (20.7% (17.2% to 24.2%)) and type B in 18 samples (3.5% (1.4% to 5.1%)), with both type A and type B occurring in five samples (1.0%). No significant associations were found between serological result and age, sex, occupation, or hospital site.
Table 1 shows the serological results correlated with questionnaire recall. Only 49/161 (30%) subjects recalling influenza had positive serological results, implying a high rate of self misdiagnosis. Of the 120 subjects with a seropositive result, 71 (59%) could not recall influenza and 32 (28%) could not recall any respiratory infection. Recall of sick leave owing to influenza (P = 0.0005, relative risk of seropositivity 1.91) and a doctor-diagnosis of influenza (P = 0.0009, 1.97) had the strongest associations with a positive serological result. In all, 42/518 (8%) subjects both had a seropositive result and recalled sick leave owing to influenza (median duration four days); this approximately represents the time lost from work that potentially could have been prevented by vaccination.