| ObjectiveMedication adherence is critical in Tuberculosis (TB) treatment success, but existing tools are inadequate in identifying non-adherents, reasons for non-adherence or interventions to improve adherence. This study intended to fill the gap by screening a series of potential factors for non-adherence and developing a valid TB medication adherence scale (TBMAS).MethodsWe conducted comprehensive reviews of literature on patient adherence and organized an advisory panel to discuss the literature and relevant factors in TB medication adherence. Then we designed the preliminary questionnaire. According to the result of pilot-testing, we modified the preliminary questionnaire. And finally, the validation study of the TBMAS was conducted from September1,2010, to August31,2011, in23Community Health Centers (CHCs) in two districts of Wuhan City (Jianghan District and Qiaokou District). All participating patients were interviewed face-to-face using TBMAS by general practitioners at the moment they visited the CHCs for a refill of TB medication. To establish external criteria for validating TBMAS, the pharmacy records of the438participating patients were reviewed and the Continuous Multiple-Interval Medication Gaps (CMG) was calculated. In the present study, descriptive analysis was carried out for demographics data and potential risk factors of medication adherence. Multivariate logistic regression analysis was employed to explore the risk factors for non-adherents. Discrimination coefficient, spearman correlation and exploratory factor analysis were used to screen redundant or non-informative items. Cronbach’s alpha, test-retest reliability and split-half reliability were calculated to measure the reliability of TBMAS. Content validity, construct validity, convergent validity and criterion-related validity was used to assess the validity of TBMAS. Receiver operating characteristics (ROC) analysis was employed to identify the TBMAS cut-off point in identifying non-adherents and estimated the predictive value. The sensitivity, specificity, positive predictive value and negative predictive values of TBMAS were also calculated by examining the concordance and discordance between non-adherents identified by TBMAS with113as cut-off and CMG, using CMG as gold standard. Finally, multivariate logistic regression analysis and pathway analysis were used to explore the correlation between medication adherence and TBMAS subscale.The database was set up by Epidata3.0by two individuals separately. We conducted confirmatory factor analysis and pathway analysis using Amos20.0software from SPSS. And SAS9.2was used for other data analysis. Concerning missing data imputation for the inventories employed in present study, we used non-missing data to predict the values of missing data. All missing values were inputted by the median of the non-missing item scores. Results1. The analysis of current situation and influencing factors of TB patients’ medication adherence.438TB patients were investigated in our study and they had a mean age of39years old. Of the respondents,68.04%were males. The CMG measurement identified181(41.3%) of the438patients as non-adherents. The result of multivariate logistic regression analysis indicated that there was no significant association between medication adherence and some socio-demographic characteristics such as age, educational level and marital status. Compared with patients younger than20-year-old, patients aged40～60years old were more likely to be non-adherents (OR=0.25,95%CI:0.09～0.71). Access to healthcare, communication with healthcare providers, patients’ mood disorder, confidence in curing and social supports were statistically significantly associated with medication adherence.2. The development and validation of TBMAS. The TBMAS is a scale of9sub-scales including30items. The nine sub-scales are:(1) communication with healthcare providers,(2) personal traits,(3) confidence in curing TB,(4) social support,(5) mood disorders,(6) lifestyle and habits,(7) coping style,(8) forgetfulness, and (9) access to healthcare. Cronbach’s alpha was0.87for the entire TBMAS, and0.88,0.78,0.73,0.75,0.67,0.78,0.77,0.51and0.52for the nine factors, respectively; test-retest reliability was0.83; and split-half reliability was0.85; all of which suggest TBMAS’s robust reliability. TBMAS’s content validity was supported by the statistically significant correlations between the total TBMAS scores and the nine factor scores when measured by Pearson correlation analysis, and TBMAS’s convergent validity and criterion-related validity were supported by the statistically significant correlations between CMG scores and the total TBMAS scores as well as the nine individual factor scores. According to confirmatory factor analysis, the factor loading of nine factors of30-item TBMAS are all greater than0.4. The goodness-of-fit test of structural equation modeling was fine indicated by Goodness of Fit Index (GIF), Criterion Fit Index (CFI) and Root Mean Square Residual (RMR). All above showed TBMAS’s robust construct validity. The area under the ROC curve was0.82(95%CI:0.77-0.86), indicating TBMAS’s high predictive value. The cut-off point for labeling non-adherent was identified where the total TBMAS score is at113. Using CMG as gold standard, TBMAS has sensitivity of82.9%, specificity of69.3%, a positive predict value of65.5%, and a negative predict value of85.2%in identifying non-adherents.3. The correlation between TBMAS sub-scales and medication adherence. According to multivariate logistic regression analysis, the nine factors were protective factors when TBMAS sub-scales scores are greater than their corresponding cut-off points, otherwise they are risk factors. Result of pathway analysis indicated that confidence in curing TB, mood disorders and lifestyle and habits had only direct effects on medication adherence, and their standardized coefficient were0.36,0.16and0.11, respectively. Communication with healthcare providers and social support had both direct and indirect effects on medication adherence, and their standardized total effects were0.30and0.27. Personal traits, coping style, forgetfulness and assess to healthcare had only indirect effects on medication adherence, and their standardized coefficient were0.19,0.18,0.14and0.09, respectively. Conclusion and suggestionMedication adherence among TB patients was relative serious. TBMAS demonstrated satisfactory internal consistency, reliability and validity in identifying TB patients with poor adherence and potential causes for non-adherence. Age and marital status are not significant associated with medication adherence, while there is positive correlation between education level and medication adherence. Patients’aged40～60years old are more likely to be non-adherents than their counterparts. Patients’confidence in curing TB has the most important effect on medication adherence, followed by communication with healthcare providers and social support. The other factors indirectly influence medication adherence through the above three factors.It is necessary to conduct further studies to evaluate the reliability and validity of TBMAS in different patient populations with different validity criteria. In the implementation of DOTS, TB medical professionals should pay attention to TB patients’ medication adherence, and manage them differently according their medication adherence. In addition, healthcare providers should enhance communication with patients to improve their confidence in curing TB.Highlights and limitationsThis study closely tracked the hot research field of TB control and prevention strategy, but it did not duplicate the work of previous research. Starting from the international research frontier and the TB patient adherence, we aimed to explore influencing factors of adherence and develop a powerful tool for measuring medication adherence. It may provide scientific evidence for deepening and improving DOTS strategy. In addition, our research successfully develops the first TBMAS in the world, which not only can identify non-adherents but also delve into the possible reasons for non-adherence, and it is a great pioneering try. In terms of the health administration research, our study deepened management strategies research from health policy to disease control and prevention, and it is a pioneering exploration.Several limitations of this study need to be noted. First, the validation study was conducted in one urban area, and may not be representative of TB patients from rural areas or other socio-economically different areas. These limitations could be addressed in the future when TBMAS is used in the field and evaluated in different patient populations. Second, we used CMG as gold standard to validate TBMAS, while CMG based on pharmacy record review has its own limitations in measuring medication adherence although it was accepted and used extensively in researches. For example, pharmacy refill records offer an objective observation on whether patients get their refills on time, but there is no guarantee that obtained medicine are actually taken, therefore, CMG may over-estimate medication adherence and consequently bias our estimates of TBMAS’s validity. So it is needed to employ more objective and more accurate validity criteria to assess TBMAS’s validity in the future. |
PDF Full Text Request