Modelo predictivo de abandono del tratamiento antituberculoso para la Región Metropolitana de Chile

Objetivos. Identificar las variables asociadas al abandono, determinar el perfil del paciente que abandona y construir un modelo predictivo de riesgo de abandono del tratamiento antituberculoso. Método. Encuesta al ingresar en el Programa y en el domicilio a los abandonos, a pacientes ingresados en el Programa Nacional de Control de Tuberculosis entre octubre de 1998 y marzo de 1999. Región Metropolitana de Chile. Seguimiento hasta octubre de 1999 (n = 570). Revisión de fichas clínicas, tarjeta de tratamiento, hoja informe mensual de cada paciente, formulario de notificación obligatoria. Resultados. Catorce de 28 variables estudiadas mostraron una asociación estadísticamente significativa con el abandono. Variables del perfil: paciente sin previsión, alcohólico, drogadicto, que vive solo, con antecedente de abandonos anteriores (p < 0,0001). Las combinaciones de variables entre el 0,1 y el 20% de probabilidad de abandono se denominan de "bajo riesgo", con una puntuación de hasta 24; las combinaciones con probabilidad de abandono entre 20,1 y 40% se denominan de "mediano riesgo", con una puntuación de 25-44, y sobre el 40% de probabilidad de abandono se denomina de "alto riesgo", con una puntuación de 45 o más. Puntuación de cada una de las variables del perfil: alcoholismo, 10 puntos; sin previsión, 15 puntos; vive solo, 20 puntos; drogadicto, 25 puntos; abandonos anteriores, 35 puntos. Conclusiones. El paciente con un 95,1% de probabilidades de abandonar el tratamiento corresponde al perfil de paciente sin previsión, alcohólico, drogadicto, que vive solo y con antecedentes de abandonos anteriores. La puntuación de riesgo de abandono se empezó a implementar en los distintos servicios de salud del país en el año 2002.

Objectives. To identify the variables associated with discontinuance of antituberculosis treatment, to determine the profile of patients who discontinue this therapy, and construct a predictive model for risk of discontinuance. Method. Patients enrolled in the national tuberculosis control program between October 1998 and March 1999 were surveyed on entering the program. Those who discontinued treatment were interviewed at home. This study was performed in the Metropolitan Region of Chile. Patients were followed-up until October 1999 (n = 570). Patients' clinical charts, treatment cards, the monthly report for each patient, and the obl igatory notification form were reviewed. Results. Fourteen of the 28 variables studied showed a statistically significant association with discontinuance. The profile variables were: lack of access to healthcare, alcoholism, drug addiction, living alone, and a history of previous discontinuance (p < 0.0001). The combinations of variables with a probability of discontinuance of between 0.1% and 20% were classified as low risk with a score of up to 24 points; combinations with a probability of discontinuance of between 20.1% and 40% were classified as medium risk with a score of between 25 and 44 points and those with a probability of discontinuance of above 40% were classified as high risk with a score of 45 points or more. The following scores for each of the profile variables were: alcoholism 10 points; lack of access to healthcare 15 points; living alone 20 points; drug addiction 25 points; prior discontinuance 35 points. Conclusions. The profile of patients with a 95.1% probability of discontinuing treatment consisted of alcoholic, drug-addicted patients who were living alone, without access to healthcare, and with a history of discontinuance. The risk of discontinuance score began to implemented in the country's various health centers in 2002.