Search

BackgroundThere are large amounts of unstructured, free-text information about quality of health care available on the Internet in blogs, social networks, and on physician rating websites that are not captured in a systematic way. New analytical techniques, such as sentiment analysis, may allow us to understand and use this information more effectively to improve the quality of health care. ObjectiveWe attempted to use machine learning to understand patients’ unstructured comments about their care. We used sentiment analysis techniques to categorize online free-text comments by patients as either positive or negative descriptions of their health care. We tried to automatically predict whether a patient would recommend a hospital, whether the hospital was clean, and whether they were treated with dignity from their free-text description, compared to the patient’s own quantitative rating of their care. MethodsWe applied machine learning techniques to all 6412 online comments about hospitals on the English National Health Service website in 2010 using Weka data-mining software. We also compared the results obtained from sentiment analysis with the paper-based national inpatient survey results at the hospital level using Spearman rank correlation for all 161 acute adult hospital trusts in England. ResultsThere was 81%, 84%, and 89% agreement between quantitative ratings of care and those derived from free-text comments using sentiment analysis for cleanliness, being treated with dignity, and overall recommendation of hospital respectively (kappa scores: .40–.74, P

[phrase omitted] Introduction Health organizations have a responsibility to learn from health-care-associated harm. In 2002, the World Health Assembly called for action to reduce the scale of preventable deaths and [...], Primary care lags behind secondary care in the reporting of, and learning from, incidents that put patient safety at risk. In primary care, there is no universally agreed approach to classifying the severity of harm arising from such patient-safety incidents. This lack of an agreed approach limits learning that could lead to the prevention of injury to patients. In a review of research on patient safety in primary care, we identified 21 existing approaches to the classification of harm severity. Using the World Health Organization's (WHO's) International Classification for Patient Safety as a reference, we undertook a framework analysis of these approaches. We then developed a new system for the classification of harm severity. To assess and classify harm, most existing approaches use measures of symptom duration (11/21), symptom severity (11/21) and/or the level of intervention required to manage the harm (14/21). However, few of these approaches account for the deleterious effects of hospitalization or the psychological stress that may be experienced by patients and/or their relatives. The new classification system we developed builds on WHO's International Classification for Patient Safety and takes account not only of hospitalization and psychological stress but also of so-called near misses and uncertain outcomes. The constructs we have outlined have the potential to be applied internationally, across primary-care settings, to improve both the detection and prevention of incidents that cause the most severe harm to patients. Les soins primaires ont du retard sur les soins secondaires en ce qui concerne l'etablissement de rapports sur les incidents qui menacent la securite des patients et les enseignements qui en decoulent. Dans le cas des soins primaires, il n'existe pas de methode universellement acceptee pour classifier la gravite des dommages resultant d'incidents lies a la securite des patients. L'absence d'une telle methode limite les enseignements qui pourraient favoriser la prevention des traumatismes chez les patients. Dans le cadre d'une analyse documentaire sur la securite des patients en matiere de soins primaires, nous avons repere l'existence de 21 methodes de classification de la gravite des dommages. En prenant comme reference la Classification internationale pour la securite des patients de l'Organisation mondiale de la Sante (OMS), nous avons entrepris une analyse du cadre de ces methodes. Nous avons ensuite concu un nouveau systeme de classification de la gravite des dommages. Pour evaluer et classifier les dommages, la plupart des methodes existantes utilisent des mesures portant sur la duree des symptomes (11/21), la gravite des symptomes (11/21) et/ou le niveau d'intervention requis pour prendre en charge les dommages (14/21). Neanmoins, rares sont celles qui tiennent compte des effets deleteres de l'hospitalisation ou du stress psychologique que peuvent ressentir les patients et/ou leurs proches. Le nouveau systeme de classification que nous avons elabore repose sur la Classification internationale pour la securite des patients de l'OMS et tient compte non seulement de l'hospitalisation et du stress psychologique, mais aussi de ce qu'il est convenu d'appeler les accidents evites de justesse et des resultats incertains. Les concepts que nous avons definis peuvent etre appliques dans les etablissements de soins primaires du monde entier pour ameliorer la detection et la prevention des incidents qui provoquent les plus graves dommages pour les patients. La atencion primaria queda por debajo de la atencion secundaria en la notificacion y el aprendizaje de incidentes que ponen en riesgo la seguridad del paciente. En la atencion primaria, no existe un enfoque universalmente aceptado para clasificar la gravedad del dano que surge de tales incidentes que afectan a la seguridad del paciente. Esta falta de un enfoque consensuado limita el aprendizaje que podria conducir a la prevencion de lesiones a los pacientes. En una revision de la investigacion sobre la seguridad del paciente en la atencion primaria, se identificaron 21 enfoques existentes para la clasificacion de la gravedad del dano. Con la Clasificacion Internacional para la Seguridad del Paciente de la Organizacion Mundial de la Salud (OMS) como referencia, se llevo a cabo un analisis del marco de estos enfoques. A continuacion, se desarrollo un nuevo sistema para la clasificacion de la gravedad del dano. Para evaluar y clasificar el dano, la mayoria de los enfoques existentes usan medidas de la duracion de los sintomas (11/21), la gravedad de los sintomas (11/21) y/o el nivel de intervencion necesario para gestionar el dano (14/21). Sin embargo, pocos de estos enfoques explican los efectos nocivos de la hospitalizacion o el estres psicologico que pueden experimentar los pacientes y/o sus familiares. El nuevo sistema de clasificacion desarrollado se basa en la Clasificacion Internacional para la Seguridad del Paciente de la OMS y tiene en cuenta no solo la hospitalizacion y el estres psicologico, sino tambien los denominados casi accidentes y los resultados Inciertos. Los constructos descritos tienen el potencial de aplicarse internacionalmente, en entornos de atencion primaria, para mejorar tanto la deteccion como la prevencion de incidentes que causan los danos mas graves a los pacientes.