The Tuberculosis Clinic: Methods

The Tuberculosis Clinic: MethodsMethods
The medical records from 1978 to 1987 for all patients with a documented diagnosis of pulmonary tuberculosis were reviewed. The diagnosis was made by either a report from the laboratory of Mycobacterium tuberculosis growing in culture or a physicians note of a positive sputum smear or culture. The following were compiled from the medical record: the year of diagnosis, the presence of alcohol abuse, the completion of treatment, the drugs given and whether the regimen followed American Thoracic Society guidelines, the follow-up, whether the patient was alive at the end of treatment, and whether the patient was referred to a treatment source outside of the medical center.

The second episode of tuberculosis in patients who had more than one episode was analyzed as a separate new case and was omitted from the analysis. In one regard, omitting an episode of tuberculosis would be better because more than one episode occurring in the same individual violates the statistical assumption of independence. However, each episode required separate treatment and follow-up, and these patients often were more difficult to treat and more important to the question of successful treatment.
An additional possible confounding factor was that once the tuberculosis clinic was established, most patients were followed up there. The patients in the general medicine clinics were seen primarily from 1978 to 1982 at a time when they were less likely to be treated with the shorter course rifampin regimens. Therefore, the two groups also were analyzed by their treatment. Patients who received rifampin and isoniazid for nine months were analyzed separately from patients who received isoniazid and ethambutol for 18 months.
The cases treated in the tuberculosis clinic were compared with those treated in the general clinics by Fishers exact test. Since there was no matching or randomization, paired testing was not done. We did not conclude significance unless the p value was less than 0.01. To further check the confounding data, we performed a logistic regression with “completed effective therapy” as the dependent variable and clinic, year of treatment, abuse of alcohol, and the use of isoniazid and rifampin as predicting factors.


So Many Advances in Medicine, So Many Yet to Come