Ambulatory patients with latent TB infection and active extrapulmonary disease with Mycobacterium tuberculosis, as well as those with atypical mycobacterial disease, were surveyed for satisfaction with counselling services. Patients with active pulmonary TB were excluded because air exchange in the pharmacy was inadequate (as defined by the Canadian Standards Association). Each patient was receiving at least one antituberculosis medication (rifampin, isoniazid, ethambutol, pyrazinamide, levofloxacin, ciprofloxacin, or pyridoxine).
Once a physician at the BCCDC prescribes antituberculosis medication, the patient is given a 1-month supply, either by the nurse at the TB Clinic (on Mondays and Tuesdays) or by the pharmacist in the BCCDC pharmacy (on Wednesdays through Fridays). Counselling about TB medications is conducted at that time. Thus, during this study, patients receiving their drugs on Monday or Tuesday received medication counselling from a nurse and those who received their drugs on Wednesday, Thursday, or Friday were counselled by a pharmacist. Patients normally return for a follow-up visit with the physician within 2 to 4 weeks. The nurses in the TB Clinic administered the satisfaction questionnaire to the patients at the time of follow-up.
Questionnaires are an effective vehicle for obtaining feedback on patients’ counselling experiences when direct observation is not feasible. The feedback from these questionnaires can help to identify differences in levels of care and can assist in pinpointing areas for future improvement. The VSQ-9 (Visit-Specific Satisfaction Questionnaire) by RAND Health Surveys (Santa Monica, California) was modified to focus on elements related to medication counselling. Patients used a 5-point Likert scale to rate their level of satisfaction on 6 aspects of medication counselling (Appendix 1). Content validity was established through an interactive review process involving a panel of 5 reviewers (2 physicians, 2 pharmacists, and 1 nurse, all from within the facility), who were asked to evaluate the clarity and conciseness of each item in the satisfaction questionnaire. In addition to questions about satisfaction with counselling services, the survey included questions to determine the patient’s knowledge about 5 aspects of the drug regimen, for which the patient’s response was coded as known or unknown (Appendix 2). The reliability of the scales used for the patient satisfaction questionnaire was tested by application of Cronbach’s a coefficient (a < 0.70).
Data were analyzed by means of SPSS 10 statistical software (SPSS, Inc, Chicago, Illinois). Statistics on the questionnaire items were computed, including means, standard deviations, 95% confidence intervals, and significance values (p < 0.05). Mean scale scores were computed by summing the scores for individual questions and dividing by the number of responses. The mean scale scores for each question and the overall scores were compared between groups. The significant differences in satisfaction and patient knowledge between groups were assessed using ordinal logistic regression. An ordinal logistic regression model was used to predict the odds of achieving a higher satisfaction score. The natural log e (the base of natural logarithms) was raised to an exponent equal to the logit parameter estimate to calculate the odds ratio. This was an appropriate statistical test because the data were ordinal in nature and the results were more meaningful when expressed in terms of estimated probability of a higher score. Because of the small sample size, Fisher’s exact test was used to test the differences in knowledge ratings between the 2 groups. The level of significance was set at p < 0.05. Variables derived from test instruments are declared to be reliable only when they provide stable and reliable responses with repeated administration of the test. Cronbach’s a is used to test the reliability of the variables derived from summated scales. In addition, this test shows that if the same questions were readministered to the same respondents, the results would be similar. The reliability of the patient satisfaction questionnaire was tested by applying the Cronbach’s a coefficient calculated by SPSS and setting the benchmark a level at less than 0.70.