You are here:
PARTICIPANT BLINDING AND GASTROINTESTINAL ILLNESS IN A RANDOMIZED, CONTROLLED TRIAL OF AN IN-HOME DRINKING WATER INTERVENTION
Citation:
Colford, J. M., J. R. Rees, T. J. Wade, A. Khalakdina, J. Hilton, I. J. Ergas, S. Burns, A. Benker, C. Ma, C. Bowen, D. Mills, S. Abbott, D. Vugia, D. D. Juranek, B. Mackenzie, AND D. A. Levy. PARTICIPANT BLINDING AND GASTROINTESTINAL ILLNESS IN A RANDOMIZED, CONTROLLED TRIAL OF AN IN-HOME DRINKING WATER INTERVENTION. EMERGING INFECTIOUS DISEASES. Center for Disease Control, 8(1):29-36, (2002).
Impact/Purpose:
To determine if a large study of a randomized home drinking water intervention could be undertaken to estimate of the burden of waterborne infectious disease
Description:
Background. There is no consensus about the level of risk of gastrointestinal illness posed by consumption of drinking water that meets all regulatory requirements. Earlier drinking water intervention trials from Canada suggested that 14% - 40% of such gastrointestinal illness could be attributed to consumption of properly treated municipal drinking water. These earlier intervention trials were randomized but the participants were not blinded. In the United States, a provision in the Safe Drinking Water Act of 1996 requires CDC and USEPA to provide an estimate of the burden of waterborne infectious disease. The trial now presented was a pilot study to determine if a much larger study of a randomized home drinking water intervention could be undertaken while successfully blinding participants.
Methods. We conducted a randomized, triple-blinded home drinking water intervention trial in a community in Northern California. The households were randomized 50:50 to use externally identical active or sham treatment devices in their homes. We measured the effectiveness of blinding of participants using a previously published blinding index. The index is a weighted kappa statistic constrained between 0.0 and 1.0. Values greater than 0.5 are consistent with successful blinding and the study was powered to detect a blinding index value > 0.5. The principal health outcome measured, "highly credible gastrointestinal illness, " (HCGI) was based on the earlier Canadian studies.
Results. Participants (n=236) from 77 households were successfully blinded to their treatment assignment. At the end of the study the blinding index was 0.66 (95% CI, 0.59 ? 0.73). There were 103 episodes of HCGI during 10,790 person-days at risk in the sham group. In the active treatment group there were 82 episodes during 11,380 person-days at risk. The relative rate of disease (adjusted by Poisson regression for the clustered sampling) was 1.32 (95% CI 0.75, 2.33).
Conclusions. These data confirm that it is possible for participants to be successfully blinded to treatment group assignment during a randomized trial of an in-home drinking water intervention. The relative rate of illness in the sham group compared to the active group was not statistically different, nor was the study powered to detect a difference of the magnitude observed in HCGI. The point estimate of the overall relative rate in our trial, however, is very similar to that of the earlier Canadian studies.