The Effects of Language Barriers on the treatment of HIV/AIDS in Namibia

Thesis (Master's)--University of Washington, 2013Background: HIV/AIDS continues to be a devastating disease of great global importance despite the fact that antiretroviral therapy (ART) is a proven effective treatment. In the last decade, there have been massive efforts to improve access to antiretroviral drugs (ARVs) around the world. Beyond the logistics of obtaining access to ART, many personal, social, cultural, and economic factors influence an HIV+ person's ability or willingness to initiate and adhere to ART. This study investigates one such factor, the effects that language barriers between the patient and provider potentially have on the treatment of HIV in Namibia. Methods: Patients analyzed (n=387) were a subset of a larger (n=590) prospective study at four ART clinics around Namibia. All patients were enrolling in ART for the first time between Jan 2012 and Jan 2013. Background demographic information was collected at enrollment and their consultation with their provider was audio recorded and later coded using Roter interaction analysis system (RIAS) methodology. Language barrier, the exposure of interest, was defined as the patient and provider speaking at least one language in common and was assessed for 387 subjects. Adherence to treatment was measured by timing of attendance to their first follow-up appointment after their initial treatment visit. Low patient positive affect and low provider positive affect were calculated from global affect scores, which are a part of the RIAS coding, and split into binary variables based on expected results. The number of questions asked by the patient was also calculated from the RIAS data and excluded questions checking patients own understanding or asking for repetition as they are likely caused by language barriers. The outcome measured was the length of the consultation as determined by the length of the audio file. Poisson regression was used to obtain relative risks (RRs) for the binary outcomes and linear regression was used to analyze the continuous outcomes. Multivariable analyses adjusted for clinic, patient gender, age, education, and marital status and included robust variance estimates that adjusted for within-provider correlation. A set of secondary analyses on the effects of translators on these same outcomes were also performed. Results: Language barrier was not significantly associated with adherence (RR 1.05, 95% CI 0.96, 1.15), low provider positive affect RR=1.74 (0.78, 3.89), or low patient positive affect RR= 1.41 (0.96, 2.07). The coefficients relating language barrier to consultation length and number of questions were not meaningful or significant. Amongst those with language barrier, having a translator was associated with an increase in the mean consultation length by 2.65 min (p= 0.047). Conclusion: We found that patient provider language barriers were not associated with a change in likelihood of patient adherence and were associated with low provider and patient positive affect score though they were not significant at the α=0.05 level. Furthermore, the presence of a translator increases the mean length of consultation but does not affect adherence or positive affect of the provider or patient. This was the first study to look into the effects of language barriers on HIV/AIDS treatment in a developing country, and further research is needed to verify and further explore this relationship