Case-control studies of the effectiveness of vaccines: validity and assessment of potential bias.

BACKGROUND

Because case-control studies of the effectiveness of vaccines are nonexperimental, it is difficult to assure that bias does not affect the validity of the results.

METHODS

A case-control study of the effectiveness of vaccines against Haemophilus influenzae type b (Hib) was replicated with a "sham" study. Cases were children > or =18 months of age with invasive infection caused by either Hib (original study) or Streptococcus pneumoniae (sham study) between January 1988 and December 1990. Controls were matched to the cases by both date and town of birth.

RESULTS

Overall 34% of the 29 cases with invasive Hib infections and 64% of matched controls had received Hib vaccine. The effectiveness of Hib vaccines against infection with Hib was 88% (95% confidence interval, 57 to 97%; P < 0.01). In the sham study 74% of the 62 cases with invasive pneumococcal infections and 74% of matched controls had received Hib vaccine. The effectiveness of Hib vaccines against pneumococcal infection was 0% (P = 0.9).

CONCLUSION

With the use of a virtually identical study design, vaccines against Hib were shown to be highly effective in preventing invasive Hib infections but were not effective in preventing invasive infections due to S. pneumoniae. Case-control studies are a valid method of assessing the effectiveness of vaccines.