[Validity of predictive power of the Adjusted Morbidity Groups (AMG) with respect to others population stratification tools.] / Validación del poder predictivo de los Grupos de Morbilidad Ajustada (GMA) respecto de otras herramientas de estratificación de la población.

OBJECTIVE: This work was performed in order to get objective elements of judgment that support the improvement of a national population morbidity grouper based in the Adjusted Morbidity Groups (AMG). The study compared the performance in terms of predictive power on certain health and resource outcomes, in between the AMG and several existing morbidity groupers (ACG®, Adjusted Clinical Groups and CRG®, Clinical Risk Group) used in some Autonomous Regions in Spain (Aragón, Canarias y Castilla y León). METHODS: Cross-sectional analytical study in entitled/insured population with respect to rights of healthcare. Predictive capacity of the complexity weight obtained with the different stratification tools in the first year of the study period was evaluated using a simple classification method that compares the areas under the curves ROC for the following outcomes that occurred in the second year of the study period: Probability of death; probability of having at least one urgent hospital admission; total number of visits to hospital emergencies; total number of visits to primary care; total number of visits to hospital care and spending in pharmacy. RESULTS: The results showed that AMG complexity weight were good predictors for almost all the analyzed outcomes (AUC ROC>0.7; p<0.05), for the different Autonomous Regions and compared to ACG® or CRG®. Only for the outcome of visits to hospital emergencies in Aragon and Canarias; and visits to specialized care in Aragon, the predictive power was weak for all the compared stratification tools. CONCLUSIONS: GMA® is a population stratification tool adequate and as useful as others existing morbidity groupers.

OBJETIVO: Este trabajo se realizó con el objetivo de conseguir elementos objetivos de juicio que apoyasen la evolución de un estratificador de la población nacional desarrollado en base a los Grupos de Morbilidad Ajustada (GMA). Para ello se validó el poder predictivo de esta herramienta de estratificación sobre determinadas variables de resultado, mediante comparación con otros estratificadores como ACG® (Adjusted Clinical Groups) y CRG® (Clinical Risk Group), utilizados en algunas comunidades autónomas (CCAA) como Aragón, Canarias y Castilla y León. METODOS: Se realizó un estudio analítico transversal en la población con derecho a la asistencia sanitaria. Se evaluó la capacidad predictiva del peso de complejidad obtenido con cada una de las herramientas de estratificación en el primer año, mediante un método de clasificación simple que comparó las áreas bajo las curvas ROC sobre las siguientes variables de resultado que sucedieron en el año siguiente: probabilidad de muerte; probabilidad de tener al menos un ingreso hospitalario urgente; número total de asistencias a urgencias hospitalarias; número total de visitas a Atención Primaria (AP); número total de consultas externas de Atención Hospitalaria (AH) y gasto farmacéutico. RESULTADOS: Los resultados obtenidos mostraron que los GMA® fueron buenos predictores de casi todas las variables analizadas (Resultados Curvas ROC AUC>0,7; p<0,05) para las distintas comunidades autónomas, al comparar con los ACG® o los CRG®. Únicamente para la variable de asistencia a urgencias hospitalarias en el caso de Aragón y Canarias, y las derivaciones a AH en el caso de Aragón, la capacidad predictiva no fue adecuada con ninguna de las herramientas de estratificación comparadas. CONCLUSIONES: La herramienta GMA® es un sistema de estratificación de la población adecuado y tan útil como otras alternativas existentes.

Relationship between glycemic control and chronic obstructive pulmonary disease in patients with type 2 diabetes: A nested case-control study.

AIMS: To evaluate the relationship between glycemic control and plasma glycated hemoglobin (HbA1c) levels in patients with type 2 diabetes mellitus (T2D) and the risk of chronic obstructive pulmonary disease (COPD). METHODS: We conducted a population-based, retrospective, nested, case-control study involving 124,876 patients with DM2 from the Canary Islands, Spain. From the cohort, we selected all COPD cases and, for each case, five control subjects who were COPD free. We analyzed the association between glycemic control, HbA1c level and incident COPD. RESULTS: A total of 1320 incidence cases of COPD (1.06%) were identified and matched individually with 6600 controls according to age and sex. After multivariate adjustment, the COPD risk increased among patients with poor glycemic control compared to patients with good glycemic control [HbA1c levels <7% (53 mmol/mol)] (OR 1.18; 95% CI: 1.03-1.36). In comparison with patients exhibiting HbA1c levels <7% (53 mmol/mol), the risk of COPD was higher among people with HbA1c levels of 7-8% (53-64 mmol/mol) (OR 1.24; 95% CI: 1.05-1.47) and 8-9% (64-75 mmol/mol) (OR 1.31; 95% CI: 1.04-1.66). CONCLUSIONS: Poor glycemic control reveals a weak association with increased risk of COPD in T2D patients.

Validación del poder predictivo de los Grupos de Morbilidad Ajustada (GMA) respecto de otras herramientas de estratificación de la población / Validity of predictive power of the Adjusted Morbidity Groups (AMG) with respect to others population stratification tools

OBJETIVO: Este trabajo se realizó con el objetivo de conseguir elementos objetivos de juicio que apoyasen la evolución de un estratificador de la población nacional desarrollado en base a los Grupos de Morbilidad Ajustada (GMA). Para ello se validó el poder predictivo de esta herramienta de estratificación sobre determinadas variables de resultado, mediante comparación con otros estratificadores como ACG® (Adjusted Clinical Groups) y CRG® (Clinical Risk Group), utilizados en algunas comunidades autónomas (CCAA) como Aragón, Canarias y Castilla y León. MÉTODOS: Se realizó un estudio analítico transversal en la población con derecho a la asistencia sanitaria. Se evaluó la capacidad predictiva del peso de complejidad obtenido con cada una de las herramientas de estratificación en el primer año, mediante un método de clasificación simple que comparó las áreas bajo las curvas ROC sobre las siguientes variables de resultado que sucedieron en el año siguiente: probabilidad de muerte; probabilidad de tener al menos un ingreso hospitalario urgente; número total de asistencias a urgencias hospitalarias; número total de visitas a Atención Primaria (AP); número total de consultas externas de Atención Hospitalaria (AH) y gasto farmacéutico. RESULTADOS: Los resultados obtenidos mostraron que los GMA® fueron buenos predictores de casi todas las variables analizadas (Resultados Curvas ROC AUC>0,7; p < 0,05) para las distintas comunidades autónomas, al comparar con los ACG® o los CRG®. Únicamente para la variable de asistencia a urgencias hospitalarias en el caso de Aragón y Canarias, y las derivaciones a AH en el caso de Aragón, la capacidad predictiva no fue adecuada con ninguna de las herramientas de estratificación comparadas. CONCLUSIONES: La herramienta GMA® es un sistema de estratificación de la población adecuado y tan útil como otras alternativas existentes

OBJECTIVE: This work was performed in order to get objective elements of judgment that support the improvement of a national population morbidity grouper based in the Adjusted Morbidity Groups (AMG). The study compared the performance in terms of predictive power on certain health and resource outcomes, in between the AMG and several existing morbidity groupers (ACG®, Adjusted Clinical Groups and CRG®, Clinical Risk Group) used in some Autonomous Regions in Spain (Aragón, Canarias y Castilla y León). METHODS: Cross-sectional analytical study in entitled/insured population with respect to rights of healthcare. Predictive capacity of the complexity weight obtained with the different stratification tools in the first year of the study period was evaluated using a simple classification method that compares the areas under the curves ROC for the following outcomes that occurred in the second year of the study period: Probability of death; probability of having at least one urgent hospital admission; total number of visits to hospital emergencies; total number of visits to primary care; total number of visits to hospital care and spending in pharmacy. RESULTS: The results showed that AMG complexity weight were good predictors for almost all the analyzed outcomes (AUC ROC>0.7; p < 0.05), for the different Autonomous Regions and compared to ACG® or CRG®. Only for the outcome of visits to hospital emergencies in Aragon and Canarias; and visits to specialized care in Aragon, the predictive power was weak for all the compared stratification tools. CONCLUSIONS: GMA® is a population stratification tool adequate and as useful as others existing morbidity groupers

Comparación de modelos predictivos para la selección de pacientes de alta complejidad / Comparison of predictive models for the selection of high-complexity patients

Objetivo: Comparar la concordancia de los pesos de complejidad entre los estratificadores Clinical Risk Groups (CRG) y los grupos de morbilidad ajustada (GMA), determinar cuál de ellos es el mejor predictor de ingreso hospitalario y optimizar el método para seleccionar el 0,5% de pacientes de más alta complejidad que se incluirán en un protocolo de intervención. Método: Estudio analítico transversal en 18 zonas de salud de Canarias, con una población a estudio de 385.049 personas, usando variables sociodemográficas procedentes de la tarjeta sanitaria, los diagnósticos y el uso de los recursos asistenciales obtenidos de la historia electrónica de salud de atención primaria (HSAP) y del conjunto mínimo básico de datos hospitalario, el estado funcional registrado en la HSAP y los fármacos prescritos en el sistema de receta electrónica. A partir de esos datos se estimó la concordancia entre estratificadores, se evaluó la capacidad de cada estratificador para predecir ingresos y se construyeron modelos para optimizar la predicción. Resultados: La concordancia entre los pesos de complejidad de los estratificadores fue fuerte (rho = 0,735) y la concordancia entre categorías de complejidad fue moderada (Kappa ponderado = 0,515). El peso de complejidad GMA predice el ingreso hospitalario mejor que el del CRG (área bajo la curva [AUC]: 0,696 [0,695-0,697] vs. 0,692 [0,691-0,693]). Se añadieron otras variables predictivas al peso GMA, obteniendo la mejor AUC (0,708 [0,707-0,708]) el modelo compuesto por GMA, sexo, edad, escalas de Pfeiffer y Barthel, existencia de reingreso y número de grupos terapéuticos prescritos. Conclusiones: Se constató una fuerte concordancia entre estratificadores y una mayor capacidad predictiva de los ingresos por parte de los GMA, que puede aumentarse añadiendo otras dimensiones

Objective: To compare the concordance of complexity weights between Clinical Risk Groups (CRG) and Adjusted Morbidity Groups (AMG). To determine which one is the best predictor of patient admission. To optimise the method used to select the 0.5% of patients of higher complexity that will be included in an intervention protocol. Method: Cross-sectional analytical study in 18 Canary Island health areas, 385,049 citizens were enrolled, using sociodemographic variables from health cards; diagnoses and use of healthcare resources obtained from primary health care electronic records (PCHR) and the basic minimum set of hospital data; the functional status recorded in the PCHR, and the drugs prescribed through the electronic prescription system. The correlation between stratifiers was estimated from these data. The ability of each stratifier to predict patient admissions was evaluated and prediction optimisation models were constructed. Results: Concordance between weights complexity stratifiers was strong (rho = 0.735) and the correlation between categories of complexity was moderate (weighted kappa = 0.515). AMG complexity weight predicts better patient admission than CRG (AUC: 0.696 [0.695-0.697] versus 0.692 [0.691-0.693]). Other predictive variables were added to the AMG weight, obtaining the best AUC (0.708 [0.707-0.708]) the model composed by AMG, sex, age, Pfeiffer and Barthel scales, re-admissions and number of prescribed therapeutic groups. Conclusions: strong concordance was found between stratifiers, and higher predictive capacity for admission from AMG, which can be increased by adding other dimensions

[Comparison of predictive models for the selection of high-complexity patients]. / Comparación de modelos predictivos para la selección de pacientes de alta complejidad.

OBJECTIVE: To compare the concordance of complexity weights between Clinical Risk Groups (CRG) and Adjusted Morbidity Groups (AMG). To determine which one is the best predictor of patient admission. To optimise the method used to select the 0.5% of patients of higher complexity that will be included in an intervention protocol. METHOD: Cross-sectional analytical study in 18 Canary Island health areas, 385,049 citizens were enrolled, using sociodemographic variables from health cards; diagnoses and use of healthcare resources obtained from primary health care electronic records (PCHR) and the basic minimum set of hospital data; the functional status recorded in the PCHR, and the drugs prescribed through the electronic prescription system. The correlation between stratifiers was estimated from these data. The ability of each stratifier to predict patient admissions was evaluated and prediction optimisation models were constructed. RESULTS: Concordance between weights complexity stratifiers was strong (rho = 0.735) and the correlation between categories of complexity was moderate (weighted kappa = 0.515). AMG complexity weight predicts better patient admission than CRG (AUC: 0.696 [0.695-0.697] versus 0.692 [0.691-0.693]). Other predictive variables were added to the AMG weight, obtaining the best AUC (0.708 [0.707-0.708]) the model composed by AMG, sex, age, Pfeiffer and Barthel scales, re-admissions and number of prescribed therapeutic groups. CONCLUSIONS: strong concordance was found between stratifiers, and higher predictive capacity for admission from AMG, which can be increased by adding other dimensions.

Relación entre diagnósticos de enfermería y uso de recursos sanitarios / The connection between nursing diagnosis and the use of healthcare resources

Objetivo: El sistema sanitario está invirtiendo un 75% de sus recursos en la atención a pacientes crónicos, en los que el foco de asistencia debe ser el cuidado y no la curación. El personal de enfermería centra su trabajo en dicho cuidado. El cuidado requiere de un registro a través de los lenguajes estandarizados en las historias de salud. Estos registros permiten diversos análisis útiles para la toma de decisiones sanitarias y organizativas. Se propone conocer cómo los diagnósticos de enfermería se asocian a un mayor gasto sanitario total. Método: Estudio transversal observacional analítico basado en los datos de las historias de salud de atención primaria (Drago-AP), altas hospitalarias (CMBD-AH) y dispensaciones de recetas (REC-SCS) en pacientes mayores de 50años durante el periodo de 2012-2013 en la comunidad canaria. Se realiza análisis descriptivo, bivariante y multivariante para generar un modelo predictivo del uso de recursos. Variables independientes: Sociodemográficas (edad, sexo, tipo de afiliación, tipo de aportación farmacéutica) y diagnósticos de enfermería (DdE) existentes a finales de 2012. Variables dependientes: Recursos sanitarios consumidos durante el año 2013. Resultados: Un total de 582.171 pacientes cumplían criterios de inclusión. Un 53,0% eran mujeres, y la edad media era de 64,3 años (DE: 10,8 años); un 53,2% eran pensionistas. El 49,3% de la población incluida tienen algún DdE, con una media de 2,1 DdE por paciente incluido. El coste medio anual por paciente fue de 1.824,62€, con una mediana de 827,5€ y unos percentiles 25 y 75 de 264,1 y 1.824,7€, respectivamente. En el análisis bivariante este gasto se correlacionó significativamente con todas las variables demográficas y aumentó en presencia de diagnósticos de enfermería de forma significativa (ρ de Spearman=0,37: a mayor número de diagnósticos, mayor gasto). En el análisis multivariante, un primer modelo de regresión lineal conformado por las variables sociodemográficas explica el 13,70% (R2=0,137) de la variabilidad del logaritmo de los costes totales. Si a ese modelo le añadimos la existencia y el número de diagnósticos de enfermería, llegamos a una capacidad explicativa del 19,77% (R2=0,1977). Conclusión: Comparado con un modelo que consta solo de variables sociodemográficas, los diagnósticos de enfermería pueden elevar la capacidad explicativa del uso de recursos sanitarios (AU)

Aim: The health service invests up to 75% of its resources on chronic care where the focus should be on caring rather than curing. Nursing staff focuses their work on such care. Care requires being redorded in health histories through the standardized languages. These records enable useful analyses to organisational and healthcare decision-making. Our proposal is to know the association of between nursing diagnosis and a higher total expenditure on health. Method: An observational cross-sectional analytical study was performed based on data from electronic health records in Primary Care (Drago-AP), hospital discharges (CMBD-AH) and prescriptions (REC-SCS) of patients over 50 from 2012-2013 in the Canary Islands. A descriptive, bivariate and multivariate analysis was undertaken to create a predictive model on the use of resources. Independent variables: Sociodemographic (age, sex, type of health-care affiliation, type of prescription charge) and nursing diagnosis (ND) recorded in late 2012. Dependent variables: Resources consumed in 2013. Results: 582,171 patients met the criteria for inclusion. 53.0% of them were women with an average age of 64.3 years (SD 10.8 years). 53.2% were pensioners. 49% of the included population had an ND, with an average of 2.1ND per patient. The average costs per patient were 1824.62€ (with a median of 827.5€) 25 and 27 percentiles of 264.1€ and 1824.7€, respectively. The bivariate analysis showed a significant correlation between these expenses and all the demographic variables; the expenses increased when a nursing diagnosis has been made (Spearman’s rank=0.37: the more diagnoses, the more expenses). In the multivariate analysis, a first linear regression with the sociodemographic variables as independent variables explains 13.7% of the variability of the logarithm of the full costs (R2=0.137). If we add to this model the presence of nursing diagnoses, the explanatory capacity reaches 19.77% (R2=0.1977). Conclusion: Compared with a model that only consists of sociodemographic variables, nursing diagnoses can enhance the explanatory capacity of the use of healthcare resources (AU)

The connection between nursing diagnosis and the use of healthcare resources. / Relación entre diagnósticos de enfermería y uso de recursos sanitarios.

AIM: The health service invests up to 75% of its resources on chronic care where the focus should be on caring rather than curing. Nursing staff focuses their work on such care. Care requires being redorded in health histories through the standardized languages. These records enable useful analyses to organisational and healthcare decision-making. Our proposal is to know the association of between nursing diagnosis and a higher total expenditure on health. METHOD: An observational cross-sectional analytical study was performed based on data from electronic health records in Primary Care (Drago-AP), hospital discharges (CMBD-AH) and prescriptions (REC-SCS) of patients over 50 from 2012-2013 in the Canary Islands. A descriptive, bivariate and multivariate analysis was undertaken to create a predictive model on the use of resources. INDEPENDENT VARIABLES: Sociodemographic (age, sex, type of health-care affiliation, type of prescription charge) and nursing diagnosis (ND) recorded in late 2012. Dependent variables: Resources consumed in 2013. RESULTS: 582,171 patients met the criteria for inclusion. 53.0% of them were women with an average age of 64.3 years (SD 10.8years). 53.2% were pensioners. 49% of the included population had an ND, with an average of 2.1ND per patient. The average costs per patient were 1824.62€ (with a median of 827.5€) 25 and 27 percentiles of 264.1€ and 1824.7€, respectively. The bivariate analysis showed a significant correlation between these expenses and all the demographic variables; the expenses increased when a nursing diagnosis has been made (Spearman's rank=0.37: the more diagnoses, the more expenses). In the multivariate analysis, a first linear regression with the sociodemographic variables as independent variables explains 13.7% of the variability of the logarithm of the full costs (R2=0.137). If we add to this model the presence of nursing diagnoses, the explanatory capacity reaches 19.77% (R2=0.1977). CONCLUSION: Compared with a model that only consists of sociodemographic variables, nursing diagnoses can enhance the explanatory capacity of the use of healthcare resources.