PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • Morocco’s National Tuberculosis Program is well-established and funded by the Ministry of Health. TB care and medicines are provided free of charge. TB diagnosis, treatment initiation, and follow-up occur at regional public pulmonary clinics (CDTMR). Patients are given TB medications via Directly Observed Therapy (DOT) at local primary care clinics or dispensaries. Study sites included nine CDTMR and one referral hospital in cities with TB “hot spots” (TB incidence of ≥400/100 K): Tangier, Rabat, Salé, Casablanca, Kenitra, and Fez. We conducted a questionnaire-based, case-control study between June, 2010, and October, 2011. Adult patients with definite or probable pulmonary or extrapulmonary TB who either defaulted from TB treatment (cases) or successfully completed it (controls) were enrolled. Treatment default was defined as an interruption in TB treatment for ≥2 consecutive months. Patients identified by review of the registries at study sites were contacted by clinic staff and asked to participate. Patients who defaulted and returned to clinic on their own were also enrolled. Upon enrollment of a case, the next two patients who presented for an end-of-treatment visit with an outcome of treatment success (treatment completion or cure) at that same site were enrolled as controls. To describe risk factors for default, a structured questionnaire was developed based on results of previous studies [5]–[8], [14]. Cases were also asked to describe in their own words the reasons they defaulted. Data collected via direct patient interview were augmented through chart review. A blood sample was collected for HIV testing. A sputum sample was collected from cases for sputum smear evaluation according to the Ziehl-Nielson method. Samples were cultured on Lowenstein-Jensen media at the regional TB laboratory or the National TB Reference Laboratory (LNRT). Drug susceptibility testing (DST) for isoniazid (H), rifampin (R), ethambutol (E) and streptomycin (S) was performed on all positive cultures at LNRT as previously described [16]. Culture data from one city did not meet quality control standards and were excluded from final analyses. Study participants provided written informed consent. This study was approved by the Ethics Committee of the Mohammed V University Faculty of Medicine and Pharmacy of Rabat and by the institutional review board of Johns Hopkins University School of Medicine. Using data from a previous retrospective study [14], we estimated that 80 cases and 160 controls would give us 90% power to detect a difference of 20% or more in the most important risk factors for default. To compare characteristics of cases and controls, we used Pearson’s χ2 or Fisher’s exact tests for categorical variables and student’s t tests for continuous variables. Multivariable logistic regression that included significant risk factors identified in univariate analyses was performed and used to develop a predictive model for treatment default. Variables with a p-value less than 0.2 in univariate analyses were included in the full model. Stepwise backward elimination methods were used to select the variables in the final model. For variables without evidence of multicollinearity, each variable’s significance as a predictor was tested by comparing the residual deviance of the nested model without the variable to that of the full model using the likelihood ratio test [17], [18]. Only those variables that were independently associated with default as indicated by a p-value less than or equal to 0.05 were retained in the final model. In addition, to avoid overfitting, Akaike’s Information Criterion (AIC) was taken into consideration in constructing the final model. In the model, knowledge of treatment duration was treated as a dichotomous variable. Those individuals who correctly stated the expected treatment duration for their TB disease were characterized as knowing treatment duration. Those who did not know or who gave a wrong answer were characterized as not knowing treatment duration. Smoking status was categorized as current, former, or never. In the model, current and never smoking were compared to former smoking. A survey tool to identify patients at high risk of default was developed by assigning points to each risk factor based on its coefficient in the predictive model. Different point cut-offs were tested to obtain the optimal sensitivity and specificity. Goodness of fit was tested using the Hosmer-Lemeshov test, where a p-value of >0.05 indicated that there was no significant difference between the collected data and that predicted by the model [19]. The models’ accuracy was tested by calculating the area under the receiver operator characteristic curve (AUC) and its 95% confidence interval (CI), where AUC that was significantly greater than 0.5 indicated that the model predicted the data better than chance [20]. Raw data were entered into Microsoft Access using EpiInfo. Data analyses were performed in SPSS (SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp) and confirmed in R (Version 3.0.1, The R Foundation for Statistical Computing, Vienna, Austria). For open-ended questions, the relative frequency of each type of response is presented along with representative quotes. Results of the quantitative analysis were compared to patients’ responses and to perspectives of local health care workers with extensive experience caring for patients with TB collected in a parallel study [15]. This mixed methods approach was used to explain and extend the results of the quantitative analysis [21], [22].
rdf:type