key recommended measure. In fact, there was in 1994 and there remains today little controlled research documenting the independent effects of these elements in preventing transmission of M. tuberculosis.4 Instead, evidence of the effectiveness of tuberculosis control measures comes primarily from case reports, analyses of survey responses, and a few studies of specific precautions. For the most part, the case for the CDC recommendations and the proposed OSHA rule rests on these sources supplemented by logic, biologic plausibility, theoretical arguments, animal studies, laboratory simulations, and mathematical modeling.
The committee identified several published reports on the experiences of hospitals with protective measures that were newly or more vigorously implemented after outbreaks of tuberculosis. Some of the limitations of case reports have already been described above and in Chapter 5. As already noted in the discussion of implementation, most studies identified by the committee describe steps taken before the release of the 1994 guidelines. It is often difficult to tell from published reports how well the specifics of control measures matched the recommendations in either the 1990 or the 1994 CDC guidelines.
Strategies for implementing the guidelines have often involved the nearly simultaneous implementation of multiple precautions. This makes judgments about the effectiveness of individual measures difficult. Although several reports present time series data, many institutions had such rudimentary tuberculin skin testing programs before the adoption of new measures that they could not report data on the preintervention period. In addition, reports vary in the way that they define time periods for study and sometimes report rates for unequal time periods. Some, for example, compare a preintervention period with an intervention period. Others compare an intervention period with a postintervention period, or they compare different periods during which different interventions were adopted.
With outbreaks of infectious disease, another concern is that subsequent decreases in disease rates might reflect not the result of implementation of control measures but rather the natural waning of infection after
The kinds of rigorous scientific studies needed to document the effect on tuberculosis or health of hazard reduction strategies in the workplace often would be operationally infeasible, requiring very large numbers of test subjects followed for a very long period under relatively stable conditions. They would also likely raise ethical and political objections.