Este es un blog medico que nunca podrá sustituir el buen juicio médico en la toma de decisiones.Intentamos compartir con los profesionales nuestras experiencias, conocimientos,lecturas,etc con la finalidad de mejorar la practica clinica.No es un blog para pacientes aunque no rechazamos sus comentarios
Se ha publicado un nuevo metanálisis sobre la relación entre los suplementos de vitamina D y el riesgo de caídas y fracturas, cuyos autores son grandes expertos en la materia (Bolland J, Grey A, Avenell A. Effects of vitamin D supplementation on muskuloskeletal Elath: a systematic review, meta-analysis, and trial sequential análisis. Lancet 2018).
El estudio analiza 81 ensayos clínicos publicados (EC), con 53.537 participantes, sobre el efecto de los suplementos de vitamina D en la incidencia de fracturas, caídas, pero también sobre la densidad mineral ósea. El nuevo metanálisis incluye los más recientes ensayos clínicos publicados, y actualiza de este modo los resultados de metanálisis previos. Pretende determinar si los nuevos EC pudieran cambiar el resultado de los estudios previos, así como comprobar el posible efecto de diferentes dosis de vitamina D en los resultados.
Las conclusiones del metanálisis de Bolland J et al son claras respecto a la necesidad de dejar de prescribir suplementos de Vit-D a las personas sanas con la finalidad de prevenir fracturas, caídas e incluso para mejorar la densidad mineral ósea, ya que su efecto es nulo o ínfimo, tanto a dosis altas como a dosis bajas.
Estas conclusiones parecen definitivas sobre el tema en cuestión, porque como se apunta en la discusión, aunque apareciesen nuevos estudios favorables al uso de suplementos, y aunque estos fueran muy potentes, sus resultados no cambarían el efecto global, ya que su peso se diluiría en el gran volumen de pacientes ya acumulados en el conjunto de estudios previos. Por otra parte, si el resultado fuera discordante, es decir favorable a la suplementación, aumentaría más la ya alta heterogeneididad de los resultados, con lo que su interpretación sería aún más incierta (y ante la incertidumbre, ya se sabe).
La propia revista Lancet publica en el mismo número un editorial acompañante en el que plantea que quizás este estudio pudiera representar el final de la suplementación con Vitamina-D, pero que aún persisten las dudas sobre su posible beneficio en los casos de deficiencia grave (<10 ng/ml). Por otra parte, el editorial destaca que los últimos metanálisis indicaban que los suplementos de Vitamina-D en bolo (semanal, o mensual u otros) se asocian a un mayor riesgo de fracturas y caídas, por lo que se aconseja la vuelta a los suplementos diarios, algo que comenta también nuestro PAPPS 2018, pero que aún no ha acabado de trasladarse a la práctica clínica.
Estos resultados cuestionan por completo la recomendación de algunos organismos, tales como la del PHE (Public Health England)de 2016, que aconseja la toma de un suplemento de 10 mcg/día vitamina D a toda la población general durante el otoño y el invierno, para asegurar un nivel plasmático adecuado.
No obstante, otros organismos ya han modificado sus recomendaciones. Así, por ejemplo, el USPTF (United States Preventive Task Force)en su edición de este año, y en el apartado de actividades preventivas para reducir el riesgo de caídas, ha dejado de recomendar los suplementos de vitamina D a los pacientes mayores de 65 años residentes en la comunidad, ante la falta de evidencia, algo de lo que también se hace eco el PAPPS 2018.
La revista BMJ, no obstante,informa asépticamente sobre el metanálisis de Bolland, y recaba la opinión de algún experto en el tema que afirma que aún es pronto para dejar de recomendar los suplementos de Vitamina-D.
El estudio publicado parece trascendente, sin embargo no aclara si los resultados son diferentes en función de los niveles iniciales de vitamina D, y si en los pacientes con niveles muy bajos pudiera existir beneficio, simplemente porque no lo analiza. Aunque en la mayoría de los estudios incluidos el nivel plasmático de vitamina D era menor de 30 ng/ml, y en más de la mitad de menos de 20 ng/ml, en sólo 4 estudios el nivel era inferior a 10 ng/ml. Sobre la necesidad de tratar los niveles entre 10 y 20 ng/ml tampoco da respuesta el estudio, a pesar de que en la mitad de los pacientes los niveles eran superiores a 20 ng/ml.
Por último merece la pena destacar un editorial publicado por la revista American Family Physician en marzo de este año, titulado: Screnning y suplementación de vitamina D en atención primaria: Es hora de frenar nuestro entusiasmoen el cual alertaba sobre el injustificado aumento tanto en la determinación de los niveles de vitamina D como en los de la población en tratamiento con suplementos. Aconsejamos consultar la traducción de este editorial realizada por Rafa Bravo en su blog Primun Non Nocere porque es altamente clarificador. La inevitablereflexión es que "resulta increíble lo fácilmente que los médicos nos apuntamos acríticamente a determinadas modas sin una sólida base científica, y también lo difícil que resulta dejar de realizar algo una vez se ha introducido en la práctica clínica, del triste desfase entre el conocimiento científico y el día a día".
Finalizamos con una propuesta de buenas prácticas sobre la vitamina D:
·Dejar de recomendar suplementos de vitamina D a la población general, incluida la población institucionalizada.
·En los pacientes que ya estén tomando suplementos de vitamina D y no se puedan suprimir, volver a los preparados de toma diaria.
·Desviar el interés de los pacientes por la vitamina D hacia otras medidas preventivas de carácter general y evidencia probada, como medidas nutricionales, asegurar exposición lumínica,abandono del tabaco y fomento de la actividad física, si las condiciones del paciente lo permiten.
·En los pacientes con alto riesgo de caídas, revisar los factores favorecedores de la mismas: condiciones de la vivienda, necesidad de supervisión o acompañamiento, detección y corrección de problemas visuales, revisión de tratamientos farmacológicos que pudiera aumentar el riesgo de caídas (psicofármacos, hipotensores, anticolinérgicos…).
Sobre a qué pacientes determinar los niveles de vitamina D y sus valores de normalidad discutiremos separadamente en una próxima entrada.
PURPOSE Continuity of care is a defining characteristic of primary care associated with lower costs and improved health equity and care quality. However, we lack provider-level measures of primary care continuity amenable to value-based payment, including the Medicare Quality Payment Program (QPP). We created 4 physician-level, claims-based continuity measures and tested their associations with health care expenditures and hospitalizations.
METHODS We used Medicare claims data for 1,448,952 beneficiaries obtaining care from a nationally representative sample of 6,551 primary care physicians to calculate continuity scores by 4 established methods. Patient-level continuity scores attributed to a single physician were averaged to create physician-level scores. We used beneficiary multilevel models, including beneficiary controls, physician characteristics, and practice rurality to estimate associations with total Medicare Part A & B expenditures (allowed charges, logged), and any hospitalization.
RESULTS Our continuity measures were highly correlated (correlation coefficients ranged from 0.86 to 0.99), with greater continuity associated with similar outcomes for each. Adjusted expenditures for beneficiaries cared for by physicians in the highest Bice-Boxerman continuity score quintile were 14.1% lower than for those in the lowest quintile ($8,092 vs $6,958; β = –0.151; 95% CI, –0.186 to –0.116), and the odds of hospitalization were 16.1% lower between the highest and lowest continuity quintiles (OR = 0.839; 95% CI, 0.787 to 0.893).
CONCLUSIONS All 4 continuity scores tested were significantly associated with lower total expenditures and hospitalization rates. Such indices are potentially useful as QPP measures, and may also serve as proxy resource-use measures, given the strength of association with lower costs and utilization.
INTRODUCTION
The Institute of Medicine labeled continuity of care a defining characteristic of primary care, one that Starfield and others demonstrated as essential to primary care's positive impact on health equity, cost reduction, and improved quality of care.1-4 Described as an implicit contract between physician and patient in which the physician assumes ongoing responsibility for the patient,5 continuity frames the personal nature of medical care, in contrast to the dehumanizing nature of disjointed care.6 Building on the idea that knowledge, trust, and respect have developed between the patient and provider over time, allowing for better interaction and communication,7 continuity at the patient level is associated with a host of benefits.8
Primary care has more measures than any other sector under the federal Quality Payment Program (QPP), yet most of these are disease specific or process measures that do not capture the core primary care functions. Despite a variety of definitions and calculations over the last 40 years, little has been done to operationalize continuity as a quality measure linked to policy-relevant outcomes in the United States or other nations.9 Given current US attention to provider-level, vs practice-level, measures in its value-based purchasing reforms, the objective of our study was to examine the relationship between physician-level continuity and health care expenditures and hospitalizations.
Insulin, Sulfonylureas Raise Cardiovascular Risk in Diabetes
Use of insulin or sulfonylureas as second-line treatment in adults with type 2 diabetes is associated with increased cardiovascular risk, whereas use of newer classes of glucose-lowering drugs is not, new real-world research from the United States indicates.
The findings, from a retrospective analysis of national administrative claims data, were published online published online in JAMA Network Open by Matthew J. O'Brien, MD, of the Division of General Internal Medicine and Geriatrics, Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, and colleagues.
Among more than 130,000 insured adults with type 2 diabetes who required a second glucose-lowering agent after metformin, use of insulin or sulfonylureas was associated with consistent cardiovascular harm compared with dipeptidyl peptidase 4 (DPP-4) inhibitors, which have been shown to have a neutral cardiovascular effect.
On the other hand, glucagon-like peptide-1 (GLP-1) receptor agonists, sodium-glucose cotransporter 2 (SGLT2) inhibitors, and thiazolidinediones (TZDs) were not associated with cardiovascular harm compared with DPP-4 inhibitors, but they also didn't produce the significant cardiovascular benefit that has been demonstrated in randomized clinical outcome trials of these agents in patients with type 2 diabetes and established cardiovascular disease.
Such high-risk populations have typically been necessary for statistical power in US Food and Drug Administration-mandated cardiovascular outcomes trials (CVOTs), but aren't representative of the adult type 2 diabetes population as a whole, of whom just 18% have established cardiovascular disease. Moreover, CVOTs are conducted on just one drug compared with placebo.
New Study Targets Area of Significant Clinical Uncertainty
"To date, no studies have directly compared the cardiovascular effects of all contemporary [glucose-lowering drug] options among patients starting second-line therapy...By examining cardiovascular outcomes among patients initiating second-line [glucose-lowering drugs] in the real world, this study aimed to complement findings from individual drug trials and further inform [glucose-lowering drug] choices for the broad population of patients currently receiving these medications," the investigators say.
In an accompanying editorial, Alison Callahan, PhD, and Nigam H. Shah, MBBS, PhD, both of the Center for Biomedical Informatics Research, Stanford University School of Medicine, California, praise the study, noting that it "targets an area of significant clinical uncertainty with the potential to inform the treatment of millions of individuals with type 2 diabetes," and in doing so "makes an important contribution to this area."
Callahan and Shah note that the findings agree with their recent study examining the real-world impact of various classes of second-line glucose-lowering agents on glycemic control and complication rates, including myocardial infarction. This new study makes "a valuable contribution" by adding GLP-1 receptor agonists.
Both studies, they note, "leverage observational data that capture details of healthcare processes and patient outcomes for millions of lives, with significant longitudinal coverage."
Basal Insulin, Sulfonylureas Robustly Associated With CV Harm
The current study included 132,737 adults with type 2 diabetes enrolled in commercial or Medicare Advantage health insurance plans during 2011-2015. All had initiated a second-line glucose-lowering drug, mostly along with metformin. The data were analyzed from January 2017 to October 2018.
Overall, 5.5% had a history of cardiovascular events before starting treatment with the index second-line agent.
Of the prescription fills for those agents, 47.6% were sulfonylureas, 21.8% DPP-4 inhibitors, 12.2% basal insulin, 8.6% GLP-1 agonists, 5.6% TZDs, and 4.3% SGLT2 inhibitors.
The investigators established the DPP-4 inhibitor users as the comparison group because data have shown that class to have a neutral effect on cardiovascular outcomes.
Relative to starting treatment with a DPP-4 inhibitor, and following adjustment for patient, prescriber, and health plan characteristics, the risk for composite cardiovascular events was 36% higher in the sulfonylurea group (HR, 1.36) and more than double with basal insulin (HR, 2.03).
This corresponds to numbers needed to harm during 2 years of treatment with sulfonylureas and basal insulin of 103 and 37, respectively.
Increased relative cardiovascular risk associated with use of sulfonylureas or basal insulin was observed across all individual cardiovascular outcomes and remained "robust" in sensitivity analyses, O'Brien and colleagues report.
Of concern, they point out, "Despite the observed cardiovascular harms associated with initiating sulfonylureas and basal insulin, prescriptions for these two [drug] classes were filled by 60% of patients in our nationwide analysis."
Newer Agents Don't Show Harm or Benefit in Real-World T2D Population
Among the newer agents, use of a GLP-1 agonist was associated with a significantly lower adjusted risk of composite cardiovascular events compared with DPP-4 inhibitor use (hazard ratio, 0.78; 95% CI, 0.63 - 0.96). However, that benefit lost significance in several sensitivity analyses.
The CV event rates after starting treatment with either SGLT2 inhibitors or TZDs weren't significantly different from those of DPP-4 inhibitors (HR, 0.81 and HR, 0.92, respectively).
O'Brien and colleagues say their work will be complemented by the ongoing randomized Glycemia Reduction Approaches in Diabetes: A Comparative Effectiveness (GRADE) study, which is comparing long-term glycemic efficacy of a sulfonylurea (glimepiride), a DPP-4 inhibitor (sitagliptin), a GLP-1 agonist (liraglutide), and basal insulin (glargine) added to metformin. Unfortunately, GRADE doesn't include any SGLT2 inhibitors.
In conclusion, the researchers say their new findings "raise concerns about the cardiovascular safety of sulfonylureas and basal insulin," compared with newer glucose-lowering drugs and suggest that short-term cardiovascular outcomes of newer glucose-lowering drug classes may be similar among patients starting second-line treatment.
"Therefore, clinicians may consider prescribing GLP-1 agonists, DPP-4 inhibitors, or SGLT2 inhibitors more routinely after metformin rather than sulfonylureas or basal insulin."
The study was supported through a grant to Northwestern University from UnitedHealthcare Services. O'Brien has reported receiving personal fees from Novo Nordisk outside the submitted work.
JAMA Network Open. Published online December 21, 2018. Abstract,Editorial
For more diabetes and endocrinology news, follow us on Twitter and on Facebook.
Influenza activity in the United States rose slightly, exceeding the national baseline for the first time this season, during the week ending November 24 (week 47), according to a report from the Centers for Disease Control and Prevention (CDC).
It is still too early to tell whether the 2018-2019 season has begun, the CDC notes.
A reduction in routine healthcare visits during the Thanksgiving holidays may be partly responsible for the rise, as well as for similar increases in previous influenza seasons.
The proportion of outpatient visits for influenzalike illness (ILI) rose to 2.3% — just above the national baseline of 2.2%. Five of 10 regions reported ILI "at or above their region-specific baseline level," the CDC states.
The CDC received reports of two influenza-associated pediatric deaths during week 47, bringing the total for the 2018-2019 season to five. Both deaths were associated with influenza A viruses; the death during week 46 was linked to an influenza A(H1N1)pdm09 virus, and the one during week 47 was associated with an influenza A(H3) virus. Additionally, a 6-year-old Montana girl died from influenza B and pneumonia on December 2, according to the Missoulian.
ILI activity was high in Georgia and Louisiana and was moderate in Alabama, Oklahoma, and Utah. Activity was low in New York City, the District of Columbia, Puerto Rico, and eight states, and was minimal in 37 states.
The geographic spread of influenza was regional in Connecticut, Kentucky, Massachusetts, Oregon, and Utah; it was local in 16 states. Sporadic activity was reported in the District of Columbia, Puerto Rico, the US Virgin Islands, and 28 states. Guam and Virginia reported no influenza activity.
The proportion of deaths resulting from pneumonia and influenza during the week ending November 17 (week 46) was 5.8%, below the National Center for Health Statistics Mortality Surveillance System's epidemic threshold of 6.4%.
The Influenza Hospitalization Surveillance Network will begin reporting influenza-associated hospitalization data for the 2018-2019 influenza season later this season.
Influenza A viruses have predominated in the United States since early July, and influenza A(H1N1)pdm09, influenza A(H3N2), and influenza B viruses continue to cocirculate during week 47. The most frequently reported influenza viruses since September 30 were influenza A(H1N1)pdm09 viruses.
Most of the influenza viruses are antigenically and genetically similar to cell-grown reference viruses representing those used in the 2018-2019 Northern Hemisphere influenza vaccines.
All influenza viruses that have been tested since late May were susceptible to the antiviral drugs oseltamivir (Tamiflu, Roche), zanamivir (Relenza, GlaxoSmithKline), and peramivir (Rapivab, BioCryst).
Healthcare professionals should consider the risk of antimicrobial resistance when deciding whether antibiotics are needed for treating or preventing a flare up of symptoms of chronic obstructive pulmonary disease (COPD).
The antimicrobial guidance recommends that antibiotics should be offered to people who have a severe flare up of symptoms, also known as a severe acute exacerbation.
However other factors should be taken into account when considering the use of antibiotics for treating an acute exacerbation that is not severe, such as the number and severity of symptoms.
The guidance notes that acute exacerbations of COPD can be caused by a range of factors including viral infections and smoking. Only around half are caused by bacterial infections, so many exacerbations will not respond to antibiotics.
Paul Chrisp, director of the centre for guidelines at NICE, said: "Evidence shows that there are limited benefits of using antibiotics for managing acute exacerbations of COPD and that it is important to take other options into account before antibiotics are prescribed.
"These recommendations will help healthcare professionals to make responsible prescribing decisions, which will not only help people manage their condition but also reduce the risk of antimicrobial resistance."
The other clinical guideline update, published today, states when to use antibiotics to prevent exacerbations happening in the first place (antibiotic prophylaxis). It recommends that antibiotics used in this way should only be offered to people who are most likely to benefit from them.
Dr Andrew Molyneux, chair of the COPD update committee, said: "COPD is a common and life-threatening illness, causing 115,000 admissions to hospital every year. For some people who have frequent exacerbations, prophylactic antibiotics can help to reduce the frequency of exacerbations and admissions to hospital. However, the benefits of prophylactic antibiotics needs to be balanced against the potential for more antibiotic resistance."
COPD is a broad term that covers several lung conditions that make breathing difficult. Some people experience flare-ups where their symptoms are particularly severe, these are called exacerbations.
COPD affects approximately 3 million people in the UK, 2 million of which are undiagnosed.
This news story was updated on 5 December 2018 - it was originally published in July when draft guidelines were released.
Hace unos días, como cada mes, José acudió al centro de salud a «mirarse el sintrom». Tiene 68 años y, entre sus antecedentes, presenta una fibrilación auricular no valvular (FANV) desde hace 6 años.Acude regularmente también a su mutua privada y en esta ocasión su cardiólogo le prescribió un anticoagulante oral directo (ACOD). Se lo comentó a la enfermera, la cual nos hizo llegar la receta privada. Le indicamos que viniera a vernos cuando pudiera, que deseábamos también poder hablar con él.
En Catalunya los ACOD deben ser autorizados y justificados. Recordamos que hacía unas semanas nos habían enviado unas pautas donde se detallaban los criterios para la prescripción de los ACOD.
Lo primero que hicimos fue verificar el tiempo en rango terapéutico (TRT) de los últimos 6 meses. En el módulo de anticoagulación disponemos de este dato actualizado. El TRT del paciente de los últimos 6 meses era de 83%.
El documento que comentamos es Pautes d'harmonització (agosto 2018, CatSalut)1. En él se detallan los motivos de indicación de los ACOD. Resumimos:
Los antagonistas de la vitamina K (AVK) (acenocumarol y warfarina) son los anticoagulantes orales priorizados para pacientes con fibrilación auricular (FA) con indicación de anticoagulación.
Los ACOD son de elección en pacientes con FANV en las siguientes situaciones:
Pacientes con hipersensibilidad conocida o contraindicación específica a los AVK.
Pacientes con antecedentes de hemorragia intracraneal en los que está indicada la anticoagulación.
Pacientes con ictus isquémico que presentan criterios clínicos y de neuroimagen de alto riesgo de hemorragia intracraneal (leucoaraiosis extensa y/o microsangrados corticales múltiples y un HAS-BLED ≥ 3) en los que se considera beneficiosa la anticoagulación.
Pacientes en tratamiento con AVK que sufren episodios tromboembólicos arteriales, a pesar del tratamiento con AVK con valores de INR en rango terapéutico.
Pacientes en tratamiento con AVK en los que no es posible mantener un buen control del índice internacional normalizado (INR) (rango 2-3), a pesar de una buena adherencia al tratamiento. Se considera buen control un TRT ≥ 65% calculado por el método de Rosendaal o un TRT ≥ 60% calculado por el método directo durante los últimos 6 meses, excluyendo los INR del primer mes en caso de ajuste inicial de la dosis, y los períodos de ajuste de la pauta de los AVK debidos a intervenciones quirúrgicas, dentales u otros procedimientos invasivos.
Pacientes con imposibilidad para acceder a los controles de INR convencionales.
El paciente no cumplía ninguno de los seis puntos.
Imprimimos las páginas 1 y 17 para entregárselas al paciente, como material escrito, el día de la visita.
En el documento existen otras dos indicaciones relacionadas con la cardioversión:
Pacientes no anticoagulados previamente y programados para cardioversión electiva: iniciar el tratamiento anticoagulante con un AVK o un ACOD según las características y las preferencias del paciente.
Pacientes no anticoagulados previamente y programados para cardioversión precoz: iniciar el tratamiento anticoagulante con heparina o con un ACOD antes del procedimiento.
La entrevista con el paciente fue relajada. Le explicamos que no se le podía hacer la receta prescrita por su cardiólogo porque no cumplía ninguno de los criterios de elección de los ACOD.
Además, le informamos de los pros y contras de los AVK y ACOD:
Y le entregamos también por escrito el material del documento de consenso donde se detalla la información que deben recibir los pacientes (p. 52)1.
En nuestro caso, el paciente no pensaba que los ACOD eran también anticoagulantes. A él le preocupaba el riesgo de sangrado ante una herida o una fractura grave. Esta información sirvió para tranquilizarle y reducir las expectativas del nuevo tratamiento.
En general, se debe explicar a los pacientes:
Para qué sirven los ACOD, cuál es el objetivo del tratamiento y que se trata de un tratamiento crónico.
Dosis, horario, relación con alimentos o sin alimentos, qué hacer en caso de olvido de alguna dosis.
Es fundamental informarles de la importancia de la adherencia al tratamiento y de que este no se debe interrumpir sin consultar con el médico.
Tanto los pacientes tratados con AVK como con ACOD deben seguir controles periódicos y ajustar las dosis al INR, en el caso de los AVK, y a la función renal, en el caso de los ACOD
La importancia de no automedicarse y tener especial precaución con el uso de antinflamatorios no esteroideos (AINE), ya que pueden aumentar el riesgo de tener una hemorragia y, en el caso de los ACOD, debe reducirse la dosis si el tratamiento con AINE se prolonga en el tiempo.
Qué hacer si presenta sangrados o hematomas significativos.
Y se les ha de recordar a los pacientes que deben informar a todos los profesionales sanitarios que están tratados con un anticoagulante oral (ACO).
¿Cuál sería la última hora en anticoagulantes orales en la práctica clínica real?
Recientemente se han publicado interesantes metanálisis que revisan estudios basados en población de la práctica clínica habitual, que utiliza ACO en el contexto de la FANV2-4, y en ellos se aprecian diferencias sustanciales en las conclusiones, debido a la gran heterogeneidad de los estudios que se han revisado: poblaciones distintas, seguimientos variables, cumplimentación y persistencia en el tratamiento a veces no explorada. A pesar de ello, tienen el valor de utilizar poblaciones de la práctica real con cumplimentaciones y seguimientos reales y no los controlados de los ensayos clínicos aleatorizados (ECA) que pueden no reflejar la realidad cotidiana.
La conclusión es que, en la práctica clínica habitual, con problemas frecuentes e importantes de mala cumplimentación y seguimientos deficientes, se puede afirmar que los ACOD y los AVK son similares en cuanto eficacia respecto a la prevención del ictus isquémico y el embolismo sistémico, pero con un número menor de hemorragias intracraneales con ACOD, aunque con más sangrados gastrointestinales con algunos de estos.
Cabe decir que la reducción del riesgo relativo, para algunas de las variables estudiadas, es realmente pequeña y algunos autores opinan que sería conveniente valorar también el número necesario de pacientes a tratar (NNT) para conseguir el efecto deseado5.
Otro punto clave en la ACO es la cumplimentación del tratamiento. Hay bastantes estudios respecto a la cumplimentación de los AVK y la permanencia adecuada en el rango terapéutico del INR (TRT) que reflejan la disparidad de resultados en la práctica clínica común6-8, siendo manifiestamente mejorable y con el peligro de que la percepción, por parte de sus médicos, de que los pacientes están mal controlados es inexistente o tardía8. La permanencia adecuada en el rango terapéutico (TRT) en Atención Primaria varía entre 52,7 y 63,3% (INR: 2-3) hasta el 88,1% (INR: 1,8-3,2).
En un metanálisis con 64.661 pacientes de una base de datos US, el cumplimiento adecuado con AVK fue de un 40,2% y de un 47,5% con los ACOD (dabigatrán, rivaroxabán y apixabán)9.
Otro problema detectado respecto al tratamiento con ACO es el sobretratamiento y el infratratamiento9,10. En 2014, en el Reino Unido, el 34% de los pacientes con FA y CHA2DS2-VASc > 2 no recibieron ACO. También el sobretratamiento con ACO en pacientes con CHA2DS2-VASc de 0 y 1 es un problema, porque el riesgo de hemorragias mayores es claramente superior al beneficio de la prevención de ictus en esta población de bajo riesgo.
Actualmente, hay numerosos estudios de coste-eficacia comparando, en la práctica clínica habitual, el uso de AVK y de ACOD4,11-18. La conclusión general de estos estudios es que los ACOD son costo-efectivos respecto a los AVK cuando el TRT con AVK es ≤ 60% y los AVK son más costo-efectivos cuando el TRT es ≥ 70%.
Si tuviéramos que pasar de un AVK a un ACOD, deberíamos suspender el AVK e iniciar el ACOD cuando el valor del INR sea < 2-3. Para pasar de enoxaparina (HBPM) a un ACOD, se debería hacer al día siguiente a la misma hora o 2 horas antes de la siguiente administración de la HBPM y, para iniciar un ACOD en un paciente no anticoagulado, no es preciso el uso de HBPM, únicamente se debe ajustar la dosis al filtrado glomerular, edad y al peso, en algunos pacientes. Pueden administrarse antes o después de la comida.
Con los ACOD, aunque se suspendan de forma temporal en el período previo de cirugía u otros procedimientos cruentos, habitualmente no se precisa terapia anticoagulante puente (ver anexo 7 de la GT semFYC). Puede ser necesaria en caso de que en el posoperatorio no se pueda utilizar la vía oral20.
Puntos clave:
El tratamiento de elección, para prevenir el tromboembolismo en la FANV, siguen siendo los AVK, pero siempre que se haga un correcto seguimiento y que, si se detecta que el TRT no es el adecuado (< 60% en los últimos 6 meses), se realice, sin inercias terapéuticas, una revisión de cuál es el problema, y si este no se puede corregir, se debe cambiar a un ACOD.
Hay que tener presente la importancia de la información verbal y escrita para el paciente con ACO (tanto con los ACOD como con los AVK).
En el caso de prescribir un ACOD, es muy necesario también planificar controles periódicos en función de las características personales del paciente, para revisar la indicación, dosificación y la cumplimentación adecuada del mismo, así como para efectuar el seguimiento de la función renal.
En este momento se dispone de interesantes documentos como los del CatSalut1, AEMPS19 y el artículo del Grupo Catalán de Trombosis para el tratamiento de los pacientes que reciben ACOD21, que pretenden establecer y homogeneizar las directrices de actuación clínica en el paciente anticoagulado.
Escobar C. Anticoagulantes orales directos frente a antagonistas de la vitamina K en pacientes con fibrilación auricular de la práctica clínica: revisión sistemática y metanálisis. Rev Esp Cardiol. [Internet.] 2018. Disponible en: https://doi.org/10.1016/j.recesp.2018.02.023
Ntaios G, Papavasileiou V, Makaritsis K, Vemmos K, Michel P, Lip GYH. Real-World Setting Comparison of Nonvitamin-K Antagonist Oral Anticoagulants Versus Vita-min-K Antagonists for Stroke Prevention in Atrial Fibrillation: A Systematic Review and Meta-Analysis. Stroke. 2017;48:2494-503.
López-López JA. Oral anticoagulants for prevention of stroke in atrial fibrillation: systematic review, network meta-analysis, and cost-effectiveness analysis BMJ. [Internet.] 2017;359:5058. Disponible en: http://dx.doi.org/10.1136/bmj.j5058
Sorigue M, Sarrate E y Orna E. Número de pacientes necesario a tratar con anticoagulantes orales directos con respecto a warfarina en la fibrilación auricular. Med Clin (Barc). [Internet.] 2017;149(6):275-6. Disponible en: http://dx.doi.org/10.1016/j.medcli.2017.03.039
Alonso R, Figueroa CA, Mainar de Paz V, Arribas MP, Sánchez L, Rodríguez R, et al. Grado de control del tratamiento anticoagulante oral en los centros de Atención Primaria de la Comunidad de Madrid: estudio CHRONOS-TAO. Med Clin (Barc). [Internet.] 2015;145:192-7. Disponible en: http://dx.doi.org/10.1016/j.medcli.2014.09.023
Barrios V, Escobar C, Prieto L, Osorio G, Polo J, Lobos JM, et al. Control de la anticoagulación en pacientes con fibrilación auricular no valvular asistidos en atención primaria en España. Estudio PAULA. Rev Esp Cardiol. 2015;68:769-76.
Boned-Ombuena A, Pérez-Panadés J, López-Maside A. Prevalencia de la anticoagulación oral y calidad de su seguimiento en el ámbito de la atención primaria: estudio de la Red Centinela Sanitaria de la Comunitat Valenciana. Aten Primaria. [Internet.] 2017;49(9): 534-48. Disponible en: http://dx.doi.org/10.1016/j.aprim.2016.11.015
Yao X, Abraham NS, Alexander GC. Effect of Adherence to Oral Anticoagulants on Risk of Stroke and Major Bleeding Among Patients with Atrial Fibrillation. J Am Heart Assoc. 2016; 5: e003074 doi: 10.1161/JAHA.115.003074
Andreea D. Ceornodolea, Roland Bal, Johan L. Severens. Epidemiology and Management of Atrial Fibrillation and Stroke: Review of Data from Four European Countries. Stroke Research and Treatment. [Internet.] Volume 2017, Article ID 8593207, 12 pages. Disponible en: https://doi.org/10.1155/2017/8593207
Hernandez I, Smith KJ, Zhang Y. Cost-effectiveness of non-vitamin K antagonist oral anticoagulants for stroke prevention in patients with atrial fibrillation at high risk of bleeding and normal kidney function. Thromb Res. 2017; 150:123-130. doi: 10.1016/j.thromres.2016.10.006
Liu CY, Chen HC. Cost-Effectiveness Analysis of Apixaban, Dabigatran, Rivaroxaban, and Warfarin for Stroke Prevention in Atrial Fibrillation in Taiwan. Clin Drug Investig. 2017 Mar;37(3):285-93. doi: 10.1007/s40261-016-0487-7.
Shah A, Shewale A, Hayes CJ, Martin BC. Cost-Effectiveness of Oral Anticoagulants for Ischemic Stroke Prophylaxis Among Nonvalvular Atrial Fibrillation Patients. Stroke. 2016;47:1555-61. DOI: 10.1161/STROKEAHA.115.012325
Janzic A, Kos M. Cost effectiveness of novel oral anticoagulants for stroke prevention in atrial fibrillation depending on the quality of warfarin anticoagulation control. Pharmacoeconomics. 2015; 33: 395-408. doi:10.1007/s40273-014-0246-7.
Hospodar AR, Smith KJ, Zhang Y, Hernandez I. Comparing the Cost Effectiveness of Non-vitamin K Antagonist Oral Anticoagulants with Well-Managed Warfarin for Stroke Prevention in Atrial Fibrillation Patients at High Risk of Bleeding. Am J Cardiovasc Drugs. 2018. doi: 10.1007/s40256-018-0279-y. [Epub ahead of print]
Hasan SS, Kow CS, Curley LE, Baines DL, Babar ZU. Economic evaluation of prescribing conventional and newer oral anticoagulants in older adults. Expert Rev Pharmacoecon Outcomes Res. 2018 May 9:1-7. doi: 10.1080/14737167.2018.1474101. [Epub ahead of print]
Gilligan AM, Franchino-Elder J, Song X. Comparison of all-cause costs and healthcare resource use among patients with newly-diagnosed non-valvular atrial fibrillation newly treated with oral anticoagulants. Curr Med Res Opin. 2018 Feb;34(2):285-95. doi: 10.1080/03007995.2017.1409425.
Amin A, Bruno A, Trocio J, Lin J, Lingohr-Smith M. Comparison of differences in medical costs when new oral anticoagulants are used for the treatment of patients with non-valvular atrial fibrillation and venous thromboembolism vs warfarin or placebo in the US. J Med Econ. 2015 Jun;18(6):399-409. doi: 10.3111/13696998.2015.1007210
Vivasa D, Roldan I, Ferrandis R. Manejo perioperatorio y periprocedimiento del tratamiento antitrombotico: documento de consenso de SEC, SEDAR, SEACV, SECTCV, AEC, SECPRE, SEPD, SEGO, SEHH, SETH, SEMERGEN, SEMFYC, SEMG, SEMICYUC, SEMI, SEMES, SEPAR, SENEC, SEO, SEPA, SERVEI, SECOT y AEU. Rev Esp Cardiol. 2018;71(7):553-64.
Olivera P. Recomendaciones del Grupo Catalán de Trombosis (Tromboc@t Working Group) para el tratamiento de los pacientes que reciben anticoagulantes orales directos. Med Clin (Barc). [Internet.] 2017. Disponible en: https://doi.org/10.1016/j.medcli.2018.01.022
José Carlos Pérez Villarroya yJordi Espinàs Boquet. Especialistas en Medicina Familiar y Comunitaria. Miembros del Comité Editorial de la Guía Terapéutica en Atención Primaria de la semFYC.
Timing of Renal-Replacement Therapy in Patients with Acute Kidney Injury and Sepsis
Abstract
Background
Acute kidney injury is the most frequent complication in patients with septic shock and is an independent risk factor for death. Although renal-replacement therapy is the standard of care for severe acute kidney injury, the ideal time for initiation remains controversial.
Methods
In a multicenter, randomized, controlled trial, we assigned patients with early-stage septic shock who had severe acute kidney injury at the failure stage of the risk, injury, failure, loss, and end-stage kidney disease (RIFLE) classification system but without life-threatening complications related to acute kidney injury to receive renal-replacement therapy either within 12 hours after documentation of failure-stage acute kidney injury (early strategy) or after a delay of 48 hours if renal recovery had not occurred (delayed strategy). The failure stage of the RIFLE classification system is characterized by a serum creatinine level 3 times the baseline level (or ≥4 mg per deciliter with a rapid increase of ≥0.5 mg per deciliter), urine output less than 0.3 ml per kilogram of body weight per hour for 24 hours or longer, or anuria for at least 12 hours. The primary outcome was death at 90 days.
Results
The trial was stopped early for futility after the second planned interim analysis. A total of 488 patients underwent randomization; there were no significant between-group differences in the characteristics at baseline. Among the 477 patients for whom follow-up data at 90 days were available, 58% of the patients in the early-strategy group (138 of 239 patients) and 54% in the delayed-strategy group (128 of 238 patients) had died (P=0.38). In the delayed-strategy group, 38% (93 patients) did not receive renal-replacement therapy. Criteria for emergency renal-replacement therapy were met in 17% of the patients in the delayed-strategy group (41 patients).
Conclusions
Among patients with septic shock who had severe acute kidney injury, there was no significant difference in overall mortality at 90 days between patients who were assigned to an early strategy for the initiation of renal-replacement therapy and those who were assigned to a delayed strategy. (Funded by the French Ministry of Health; IDEAL-ICU ClinicalTrials.gov number, NCT01682590.)
Integrated Telehealth and Telecare for Monitoring Frail Elderly with Chronic Disease
Objective:To investigate the potential of an integrated care system that acquires vital clinical signs and habits data to support independent living for elderly people with chronic disease.
Materials and Methods:We developed an IEEE 11073 standards-based telemonitoring platform for monitoring vital signs and activity data of elderly living alone in their home. The platform has important features for monitoring the elderly: unobtrusive, simple, elderly-friendly, plug and play interoperable, and self-integration of sensors. Thirty-six (36) patients in a primary care practice in the United Kingdom (mean [standard deviation] age, 82 [10] years) with congestive heart failure (CHF) or chronic obstructive pulmonary disease (COPD) were provided with clinical sensors to measure the vital signs for their disease (blood pressure [BP] and weight for CHF, and oxygen saturation for COPD) and one passive infrared (PIR) motion sensor and/or a chair/bed sensor were installed in a patient's home to obtain their activity data. The patients were asked to take one measurement each day of their vital signs in the morning before breakfast. All data were automatically transmitted wirelessly to the remote server and displayed on a clinical portal for clinicians to monitor each patient. An alert algorithm detected outliers in the data and indicated alerts on the portal. Patient data have been analyzed retrospectively following hospital admission, emergency room visit or death, to determine whether the data could predict the event.
Results:Data of patients who were monitored for a long period and had interventions were analyzed to identify useful parameters and develop algorithms to define alert rules. Twenty of the 36 participants had a clinical referral during the time of monitoring; 16 of them received some type of intervention. The most common reason for intervention was due to low oxygen levels for patients with COPD and high BP levels for CHF. Activity data were found to contain information on the well-being of patients, in particular for those with COPD. During exacerbation the activity level from PIR sensors increased slightly, and there was a decrease in bed occupancy. One subject with CHF who felt unwell spent most of the day in the bedroom.
Conclusions:Our results suggest that integrated care monitoring technologies have a potential for providing improved care and can have positive impact on well-being of the elderly by enabling timely intervention. Long-term BP and pulse oximetry data could indicate exacerbation and lead to effective intervention; physical activity data provided important information on the well-being of patients. However, there remains a need for better understanding of long-term variations in vital signs and activity data to establish intervention protocols for improved disease management.
Background
The increase of the aging population presents challenges to social care and healthcare, in particular, the prevalence of chronic disease among the elderly and need for long-term management is increasing healthcare costs. In addition, the level of independence of the elderly may fall due to disability resulting from aging, a disease, or cognitive ability,1 all of which may undermine their autonomy and make them dependent on care providers and social services. Combined with decrease in young population in developed countries, a need for new care plans that require less human resource and that combine health and social care services has emerged. Telemonitoring technologies have been considered for care delivery in the elderly with a high level of need and who require long-term care,2 thereby extending the period of independent living through timely intervention when deterioration in their well-being is detected. Ideally, timely intervention would result in averting hospitalization, speedy recovery, improved outcome and quality of life, and decrease in cost of treatment.3
The potential of telemonitoring technologies to improve management of chronic diseases and reduce cost to the healthcare system has been extensively researched over the last three decades.3–6 Most of these studies have focused on congestive heart failure (CHF), diabetes, hypertension, stroke, and chronic obstructive pulmonary disease (COPD), as timely intervention for these diseases can significantly improve the outcome of intervention and reduce cost of care.3,7 Vital signs that have been monitored include electrocardiogram (ECG), blood pressure (BP), blood glucose, pulse, oxygen saturation (SpO2), weight, and body temperature.8 Most studies have reported positive effects of telemonitoring.9
Changes in daily activity level and habits can provide vital information in relation to functional capabilities, deterioration in well-being, progress of an existing chronic disease, and loss of autonomy.10 Acknowledging this, over the last two decades, many studies have been conducted to investigate the potential of telemonitoring activity profiles of subjects to detect deterioration in their well-being and changes in lifestyle.11,12 These studies did not necessarily target the elderly with chronic disease; they reported results of technology development and evaluation of technological feasibility, and only a few of the studies associated changes detected in activity profiles with well-being of the subjects being monitored. Results that associated changes in activity and well-being included increased bathroom visits due to urinary tract infection and increased level of nocturnal activities was thought to be sign of deteriorating cognitive abilities. Approaches and sensors used to acquire activity data varied, and included activity-log records, passive infrared (PIR) motion sensors, electricity used by appliances and accelerometer-based wearable sensors. Parameters monitored included activity level, bed restlessness, bathroom visits, forgotten stove burner, body movements, and posture (walking, running, standing, and fall).11
Few studies have monitored physiological parameters together with activity data.11 Most of the projects were restricted to using volunteers to test the feasibility of their systems; only a few of them involved elderly with chronic disease(s) with the aim to predict key medical events that required intervention or changes in habits profile that were associated with deterioration in well-being of elderly with CHF.13 Only a few have investigated the association between changes in clinical and activity data, with results being encouraging for the relevance of monitoring activity data of subjects with chronic disease(s).14
Acknowledging the growing demand for independent living among elderly in developed countries, a research project, entitled Integrated Network for completely assisted Senior citizen's Autonomy (inCASA) was developed to demonstrate the concept of integrated health and social services for the frail elderly living alone.15 Since there were no commercial systems available to support integrated telehealth and telecare, an integrated platform for telemonitoring of vital signs and habits data was developed for the United Kingdom pilot. The platform was used to manage 36 frail elderly who were registered with Chorleywood Health Centre (UK), received care from social services, had a chronic disease, and were living alone. The telemonitoring system was purpose designed for the elderly, having several important features, some of which are unique to the system:
Plug and play installation; self-integration of sensors to the system
Monitors both activity and physiological data
Online analysis of data and alert
This article presents results from the U.K. pilot where habits and vital signs of 36 frail elderly with chronic disease(s) were monitored. Our aim was to investigate:
Feasibility of the concept of integrating health and social care on a single platform.
Habits profiles of elderly and rules to notify professionals when there is deviation from normal patterns.
Whether change in habits profile is associated with patient's well-being.
Advantages of sharing and exchanging information between the primary care and social services.
Following a brief description of the study design and monitoring system, we present results of the data analysis and discuss findings.
Materials and Methods
Participant Identification and Recruitment
Subjects for the pilot were selected from patients registered with Chorleywood Health Centre, United Kingdom, using the following criteria:
Over the age of 65 years
Have at least one chronic disease
Living alone
Determined to be "Frail" as defined by the Edmonton Frailty Score15
Had an unplanned hospital admission in the past 6 months or two in the past 12 months.
One hundred five (n = 105) patients were identified as meeting the inclusion criteria, and were informed of the study and invited to participate. Ethical approval for the study was gained from the local research ethics committee.
A total of 44 patients initially gave informed consent to participate, and 36 were recruited into the study from October 2012 onward. After the end of the pilot phase on May 31, 2013, some remained in the service till March 2014, which enabled us to obtain monitoring data for longer than a year.
The service team was made up of clinical nurses, general practitioners, nonclinical researchers, social service workers, administrators, and technical support; the service provided guidelines for self-management and the communication channel (mainly phone) between patients and their nurse care managers.
Telemonitoring System—Sensors; Home Gateway; Remote Server; Clinician Portal
The Home Monitoring Platform was designed and deployed to participant's house.16 The platform comprised of sensors to acquire patient's habits and clinical data, a home gateway, a remote server to store patient data, and a clinician portal to view and manage patient data and records (Fig. 1). The platform used a standards based approach for data communication that enabled many different types of devices, habits, and health, to be deployed to patients with co-morbidities. The IEEE 11073 medical device standards17 were used for communication from the sensor to the gateway; IHE PCD-01,18 a profile of HL719 was used for data communication from the gateway to the server. All data are automatically transmitted wirelessly from the sensor devices to the home gateway using the ZigBee Healthcare Profile,20 and then wirelessly to the remote server using GPRS. The activity and clinical sensors (Fig. 2) were obtained off-the-shelf and modified to take our IEEE 11073 radio modules to allow wireless data transmission to the gateway.
The Gateway
The gateway was designed to be simple, unobtrusive, and self-contained so that it required no configuration for installation and being based on cell-phone technology (GPRS) avoided the need for patients to have existing internet connectivity or landline. It had no user interface other than an LED to indicate connection to the server, and its installation was as simple as "plug it into a mains socket and watch for the green light indicating connection to the remote server." As there was no user interface, sensors could be installed anywhere in the home. The installation of the telemonitoring equipment was carried out mostly by the nurses. Devices were located by taking into account both the preference of the patient and the quality of wireless connections.
Patients were given training during the installation, which included how to operate their clinical sensor(s) and observe the light on each device to confirm successful data transmission. They were also given the contact numbers of the clinical team, whom they could contact in case of a concern.
Data were transmitted from the gateway to the clinical server over a secure private mobile network; the clinical server was located in the secure data center. Patients were assured that data would be managed securely and kept private, and any data would be published anonymously.
Sensors and Parameters Extracted
The first generation gateway could support up to 3 sensors connected concurrently, and the second generation gateway could support up to 10 sensors. In general, three sensors were installed in the home of each patient, depending on the disease. Patients with CHF were given a weight scale and a BP meter, and those with COPD were given a pulse oximeter (Table 1). All patients were given a PIR motion sensor in the living room, and those with a single health device were given a bed or chair occupancy sensor.
Table 1. Sensors Deployed
SENSOR
REASON
DATA COLLECTION FREQUENCY
MONITORING FOR
BP
BP in CHF
Daily
Exceed defined target >140/80 mmHg
SpO2
Oxygen saturation in COPD
Daily
Exceed defined target <85%
Weight
Fluid retention in CHF
Daily
Change of >1 kg in 24 h or 1.4 kg over 3 days
Motion sensor
Habits monitoring
Continuous
Movement variance from normal
Bed sensor
Habits monitoring
Continuous
Unusual time for bed occupancy; number of times out of bed during night
Chair sensor
Habits monitoring
Continuous
Unusual time in chair; excessive time in chair
All the data from the sensors were sent automatically to the gateway and from the gateway to the remote server without user intervention, where they were used by the clinical team for management of the patient. Automatic transmission of data eliminated reporting bias of manual entry or confirmation,21 and was a very useful feature for the elderly due to the likelihood of physical and/or intellectual limitation.1,22 An alert algorithm, normally based on default thresholds as shown in Table 1, or customized limits, was applied to all incoming data to provide visual alerts for high BP, low SpO2, or significant change in weight on the clinical portal.
Clinical Sensors
All the sensors were clinically validated, and were modified to take our ZigBee radio module to allow wireless data transmission to the gateway and then on to the remote server. Patients were instructed to take at least one measurement each day, where possible first thing in the morning before breakfast. With the BP meter and pulse oximeter, they were instructed to take their measurement after sitting quietly for 5 min and while their arm was resting on a table or the armrest of a chair.22
The BP meter was an upper arm cuff meter and patients were instructed to use it with the upper arm leveled with the heart.22 We chose a finger Pulse oximeter that provided a noninvasive estimation of arterial hemoglobin oxygen saturation (SpO2).
Occasionally patients took two or three clinical measurements on a single day. With SpO2 the higher reading was chosen as the reading for that day, as this approach is used by clinicians23; the median value of multiple readings on a day was used as the representative value for BP and weight readings.
Habits Sensors
Two types of sensor were deployed to patients' homes to monitor habits: (1) PIR motion sensors (Fig. 2c) to detect movement in a location and (2) pressure sensors (Fig. 2d) to detect bed or chair occupancy. Our aim was to define a daily habits profile for the elderly person in their home to determine when there was deviation that might be indicative of change of well-being.
PIR sensors
The location of the PIR sensor was determined so as to capture and profile important and relevant daily activities. The sensors were typically located in the living room in a position to capture the significant movements within the home, such as from living room to/from the kitchen, bathroom, or bedroom, but not to capture movements while sitting in the chair or sofa.
Bed/chair occupancy sensors
These sensors were calibrated pressure sensors located underneath the mattress or the chair cushion and were configured to send a message for both "usage started" and "usage ended." To avoid glitches in the sensor data, a change in state of usage message was only sent after the sensor had remained in its new state for 30 s.
A few patients asked for their chair-sensor to be removed as they found it uncomfortable; and a few of the sensors were found to be sensitive to changes in the room temperature and gave unreliable data. About 6 months into the monitoring period, the bed/chair sensors were replaced with PIR sensors in the bedroom due to comfort or reliability issues.
Parameters Extracted from Activity Data
We analyzed the data to define a normal profile for each of the parameters, typically formed from the moving average of data (defined for each parameter), and from this we determined deviations from the normal profile to investigate whether deviations are associated with the well-being of the patient. Some of the algorithms and parameters were used to provide alerts on the clinical portal; others were used for retrospective analysis. Following a clinical intervention, we retrospectively analyzed the data to identify patterns or parameters that might predict the oncoming event. The following parameters were derived from the habits sensor data:
Number of sensor events in a given period
Mean of hourly movement counts
Time of first movement in early morning and last movement in the evening
Time to next sensor event
Bed/chair occupancy in a given period.
The parameters were derived as follows:
Number of movements detected by PIR sensors and usage triggers in different time periods
The number of movements detected in each hour was counted. These were accumulated to determine the number for different time periods in the day and for the whole day. Similarly, the number of usage triggers from usage sensors was counted and accumulated. We used the variation from the normal value for the whole day from both PIR and bed/chair sensors to raise alerts on the clinical portal. We found the binning period of 1 h sufficiently short to determine the times of activities, but sufficiently long to filter out short-term daily variations in the times of activities.
Mean of hourly movement count
Data from a PIR sensor across 20 days were used to determine the activity profile of a subject across the day (e.g., Fig. 3). Depending on the location of the PIR motion sensor, it would be possible to estimate the time for: getting out of bed, breakfast, lunch, dinner, and going to bed. For example, from Figure 3, we could infer that the subject got up at around 7 am, the high level activity around 8 am might correspond to breakfast, at around 5 pm to dinner, and the subject leaving the living room at 8 pm was going to bed.
Time for first movement in early morning and last movement in the evening
Observing the mean of the number of movements in each hour in Figure 3 highlights that this patient has a clear "bed time" and "wake-up time" routine; they get up between 4:00 and 6:00 and go to bed by 22:00. Using this knowledge and a simple algorithm, times of first movement in the morning and last movement in the evening can be estimated, and can be used as an estimate of bedtime routine, in particular, when no bed sensor is used, as in Figure 4.
Time to next move: time to next sensor event
We obtained the time intervals between consecutive sensor events; this gave us a time series of time intervals between consecutive events. We then determined the 90th quantile and median value of the time interval for days, where possible, with 30 movements or more.
Bed/chair occupancy
The bed/chair pressure sensor provides a time stamped event to indicate a change in state of occupancy. Using times of consecutive "usage started" and "usage ended," we could calculate the length of occupancy for each usage, and then total bed/chair occupancy in a day by accumulating the individual occupancies.
Raising Alerts from the Data
A simple algorithm was implemented to detect deviations from the norm for the habits data. For the clinical data, two types of thresholds were used: an absolute threshold taken from the clinical assessment protocols given in Table 1; and subject specific thresholds (mean ±2 standard deviation [SD]).
On the presence of an alert on the portal, further steps were taken: in the case of an activity alert the nurse would contact the patient to determine the reason for the alert, and if the alert persisted, a visit to the patient was planned. If the nurse considered that a change in treatment or medication was required, they would refer the patient to the doctor.
Initially alerts were generated from the activity data by dividing a day into four periods: (1) 00:00–06:00, (2) 06:00–12:00, (3) 12:00–18:00, and (4) 18:00–24:00. The mean and SD of the number of sensor events for each period was determined by using a moving window of 15 days. Period-specific thresholds were calculated as mean ± 2SD. If the number of movements in a period fell outside the threshold values for that period, a red flag was shown on the clinicians' portal. However, after 6 months, the four time periods were deemed to be giving rise to too many false alarms; for example, the absence of a patient for part of a time period could easily result in an underactivity alert, or a visitor in the afternoon to an overactivity alert for that time period. From our experience and discussion with the clinical team, three time periods were found to be more relevant to well-being of a subject: (1) all-day (mid-night to mid-night), (2) night-time (22:00–06:00), and (3) morning (06:00–10:00). Instead of generating alerts for all time periods, we decided to generate alerts on the portal only for all-day.
Clinical Portal
A clinical portal was developed to (1) visualize patients' data and alerts, (2) allow the clinician to view, manage patients' data, and edit patient records for the project, and (3) allow the research team to download the patients' data.
Alerts were displayed on the clinical portal to notify (draw the attention of) the clinicians to patients that may require intervention. The clinicians' portal was reviewed daily by a nurse to determine whether alerts had occurred and intervention might be required and to examine the data of specific patients to monitor progress (e.g., after change of treatment).
Results
Thirty-six (n = 36) patients were enrolled in the service (mean age 82 years [SD = 10], 38% male, 56% average frail, and 27% very frail). The majority of the patients enrolled in the study were not familiar with new technologies. Acceptance of habits monitoring was an issue for about 15% of the patients for a number of reasons, including intrusiveness of the technology; did not want stigmatization of being "frail"; and did not feel that the technology was for them, as they did not consider themselves as frail. Generally most did not give a specific reason for declining participation, stating only that "they did not want to."
Compliance rate with daily readings of BP and SpO2 was generally over 60%. Only a few patients had low compliance; two with CHF were very frail and only occasionally would weigh themselves due to safety concerns of standing on the weigh scales. One patient took only 17 BP measurements over 75 days and one patient took only 3 SpO2 readings over 110 days. We did not investigate the reasons for low compliance rate for BP and SpO2.
Fifty-five percent (20) of patients were referred for investigation during the time that they were being monitored; 44% (16) received some type of intervention. The most common reason for intervention was due to low oxygen levels for patients with COPD; these patients were referred to community pulmonary services. We illustrate our preliminary results with four case studies.
Patient 1—Chf and Associated Hypertension
This patient had CHF and associated hypertension and was monitored for 353 days. A PIR sensor in the living room, a chair sensor, and BP meter were deployed to this subject. However, the subject asked for the chair-sensor to be removed due to discomfort issues. Later in the monitoring period (day 244), a PIR sensor was installed in the bedroom.
Figure 5 illustrates the BP readings. The patient measured the BP for 281 days of the 351 monitoring days. BP readings varied considerably over the monitoring period: they were mainly over 150 mmHg at the beginning of the monitoring, which lead to BP assessment and a medication change on day 15. The medication change helped to reduce the BP level. Around day 220, the BP value fell with the patient complaining of dizziness, which led to a further medication change on day 235. The subject complained of swollen ankles, which led to a further medication change on day 294, followed by another on day 309. The subject felt worse on day 316, and there was a nighttime event on day 327.
The variability of the systolic BP and pulse increased significantly from day 270 onward, and the patient was diagnosed with atrial fibrillation (AF) on day 284.
Figure 6 illustrates the number of movements detected by the PIR motion sensor in the living room for the whole day and the nighttime (22:00 pm–6:00 am) periods. The living room PIR event counts for the whole day exhibited a fluctuating trend with a period of about 90–120 days; however, we cannot offer any physiological explanation for this. Although there are other fluctuations, we cannot find any association with clinical events.
Nighttime activities resulted in an alert on day 15; when contacted the patient reported having a heavy cold and so was up and down all night. After this event, the nighttime activity level remained zero except for two occasions.
The results for the PIR sensor in the bedroom complement those of the living room and are given in Figure 7. After day 325, there is a slight drop in bedroom activity level for whole day; after day 327 no activity was detected during nighttime.
We further analyzed the living room PIR data by calculating the time between successive events to determine the time to next move. We then ranked the time to next move values for the whole day (Fig. 8). Values of the 10th quantile could give information on the length of time small tasks took to perform, such as going to the kitchen/bathroom and coming back. Those for the 90% values will give information about the length of prolonged periods with no movement; the longer this value the less likely the patient would want to move. The values for the 50th percentile will indicate general tendency for time-to-next-move values. The value for the 50th quartile has episodes where the value increases, notably around day 20, between day 85 and 140, and day 340.
Figure 3 illustrates the combination of the average number of PIR events for each hour for the living room (white bar) and bedroom (black bar) for the days from 251 to 270 to determine the pattern of behavior around the house throughout the day and night. The subject usually has no nighttime activity in the living room and has a very predictable pattern for first and last movements detected in the living room each day. This subject habitually first enters the living room at around 5:00 am and leaves before 22:00 pm.
We therefore observed the daily times for the first and last movements detected in the living room, as shown in Figure 4.
To demonstrate, Figure 9 shows the hourly activity graph for the living room and bedroom for day 349 (left) and day 350 (right). Each exhibits an unusually early last movement, with the subject last leaving the living room in the early afternoon (e.g., 14:00 day 349) and the bedroom activity confirming they spent the remainder of the day in the bedroom. There is a corresponding late first movement the next day in the living room (8:00). We therefore investigated the other days with unusually early last movement for the period with two PIR motion sensors (one in living room, the other in bedroom), and found that the patient spent the remainder of the day in the bedroom. In contrast, there was no significant difference (i.e., no indication of such a pattern) in the total daily and nighttime PIR event counts for the living room (Fig. 9, left) and bedroom (Fig. 9, right) to generate an alarm.
We observed an increase in the number of these incidences after day 300, and these correlated with the problems seen with their BP (Fig. 5) and diagnosis of AF, and would also be associated with the patient reporting on day 316 that they were feeling worse.
Patient 2—Chf and Associated Hypertension
This patient had CHF and associated hypertension and was monitored for 495 days. At the start of the study, the patient was given a BP meter, and a PIR sensor (in the living room) and a bed sensor were deployed. Due to reliability issues, the bed sensor was removed and replaced by a PIR sensor in the bedroom on day 253.
Figure 10 shows the 105 BP readings that were taken by the patient during the monitoring period. Typical systolic BP was around 145 mmHg or higher for the first 70 days, which generated alerts on the clinical portal by being above the threshold of 140 mmHg and led to medication change on day 74. Systolic BP fell to around 120 mmHg after the medication change and remained below the threshold for the remainder of monitoring.
Figure 11 shows the number of movements detected for whole day by the PIR sensor in the living room. The patient had several visits by the nurse during the first 40 days due to alerts on the portal for high BP, as seen by the increased number of movements on certain days. However, there was a trend of decreasing number of movements in daily PIR activities after day 50, which led to underactivity alerts on the portal. When the patient was contacted by phone on day 65, they said that they had stayed in bed longer after a recent fall. These incidents on day 65 and 68 are marked by a circle.
A nurse visited the patient on day 74 and found that the patient had cellulitis. The visit by the nurse resulted in a peak in living room PIR activities. In contrast, the patient movements were becoming fewer after the fall event (Fig. 11) and the median values of time-to-next-move were increasing (Fig. 12). Note that the median values of time-to-next-move are less sensitive to visits.
Inspection of the median (50th quartile) values of time-to-next-move (Fig. 12) appears to indicate that the patient continued to have health issues until around day 300 at which time the value returned to one comparable to the beginning of the monitoring period when the patient was considered in good health. We have no clinical events to corroborate.
Patient 3—Copd
This patient had COPD and was monitored for 212 days. The patient was given a pulse oximeter, a PIR sensor in the living room, and a bed sensor. The bed sensor was only used for the first 62 days; therefore, no results are presented for this sensor. This patient had two major clinical events during the 212 days; hospital admission on day 120 for 3 days and a chest infection on day 204 (thick vertical dashed lines in Fig. 13). There are also notes indicating clinical concerns around day 22, 30, and 88 (light vertical dashed lines in Fig. 13). The patient, when contacted, did not report any change in condition on day 14 or 22; and believed that their breathing had improved around day 14. On day 88, the patient was diagnosed with a cold, and the condition continued to deteriorate until the patient was admitted to hospital on day 120. Our earlier work on analysis of daily readings of SpO223 (Fig. 13) showed that the short-term and long-term trends and residuals closely followed the condition of the patient, that is, decreasing level in trends and increasing SD of residuals during periods of clinical events, and returning to their usual levels following the interventions.
The number of movements detected by the PIR motion sensor in the living room and the times of first and last detected movement in each day are given in Figures 14and15 respectively. There is a slight increase in movement counts for the whole day between monitoring days 85 and 110, from about 60 movements to 70 movements (Fig. 14). The reason for this may be discomfort or that they had to pause to take a breath or were walking more slowly, both of which would have led to extra PIR detection events while they were walking through the detection zone. From around day 90 onward, more movements were detected in the afternoon. These increases in number of movements coincided with very low SpO2 levels. The patient also appears to be particularly restless on some nights (50, 85 and 130). This patient started to get up slightly earlier after day 80 (Fig. 15), which coincides with the summer daylight savings time change, and is not significant. There is 1 day (50) when the patient appears to have retired to bed earlier than usual.
Figure 16 presents the 10th, 50th, and 90th quantiles for time-to-next-move for days with 30 movements or more. The values for the 90th quantile became slightly higher before day 120 and after day 200, that is, toward the hospitalization. The reason for this may be that the subject would tend to walk more slowly due to difficulty in walking during exacerbations (when SpO2 values are low)21 and/or was taking longer to complete a task.
Patient 4—Copd
This patient had pulmonary fibrosis, and died at home on day 135. The patient was given a pulse oximeter, a PIR sensor in the living room, and a bed sensor. We had data from the bed sensor for almost the whole monitoring period, but data from PIR motion sensor for only the first 57 days; therefore, we do not present results from the motion sensor. Figure 17 shows the SpO2 readings. The patient commenced oxygen therapy on day 34, and reported that therapy was helping. The long-term trend indicated a steady decline in the condition of the patient over the monitoring period.
The bed data provided important information on the well-being of the patient. Figure 18 shows the periods of bed occupancy over the monitoring period as stacked vertical lines. Midnight appears at the top and bottom of the Figure 18, with periods of bed occupancy in each day shown as black vertical lines.
Figure 18 shows the bedtime routine and behavior over the period of monitoring. The patient retires to bed at a fairly constant time each day between 22:00 and 24:00; however, the time of waking varies and indicates a significant drop in bed occupancy between days 80 and 90 during an exacerbation, and day 125 onward as death approaches. This change may be due to being unable to sleep or experiencing difficulty with staying in the bed due to breathing difficulties.
Close inspection of Figure 18 shows that the sleep pattern is broken during each day. We therefore identify each period of bed occupancy during the day and order them according to decreasing length. Figure 19 shows the stacked bar graph of the lengths of the five longest occupancies each day during the final 90 days. The five longest occupancies using a stacked bar graph (Fig. 19) can provide information on the length of each period of uninterrupted sleep, and thus the sleep quality and how comfortable they are when sleeping in bed. The longer the period of uninterrupted sleep, the more likely it was deep sleep and therefore beneficial for well-being or, in the case of subjects with COPD, that they were comfortable in bed. Although Figure 19 does not show any specific pattern, it clearly shows that bed occupancy drops drastically between days 80 and 90, and after day 125 due to the deterioration in the condition of the patient. As we do not have full data for the PIR in the living room, we are unable to determine whether they slept in a chair in preference. However, the change in habit remains significant.
Discussion
We have presented results from one of the first projects to deploy both environmental and physiological sensors to patients. Our results demonstrate the feasibility of integrating both types of sensor on the same interoperable platform to support health and social care together. Simple algorithms were used to generate alerts to the clinicians on the portal to indicate those patients that may require attention, prompting intervention for patients with high BP with medication change, COPD patients with low SpO2 for referral for pulmonary assessment and O2 therapy.
The habits data were capable of generating alerts for events such as a patient who had fallen, and patients with increased level of nighttime activity due to a heavy cold or exacerbation. None of these could have been identified without the integrated platform, unless the patients contacted the professionals with relevant complaints.
We have estimated habits patterns for patients, and detected deviations from normal behavior, which includes under- or overactivity alerts. Both under- and overactivity provided important information about the well-being of a subject. For example, patient 1 spent most of the day in the bedroom when they felt unwell; a sudden drop in activity level in patient 2 was found to be due to a fall and cellulitis, and later due to a leg ulcer. Previous telemonitoring studies have reported that changes in activity levels can relate to changes in well-being.11 For example, sleeping in a chair instead of the bed was observed with one CHF patient; frequent bathroom visits due to urinary tract infection 24; and decrease in activity levels due to increased depression level.25 Underactivity or longer time in bed could be due to depression, and overactivity due to discomfort or onset of dementia.
Our selection of features would concur well with others,26 who used regression analysis to determine the correlation between features of activities of daily living and self-administered health metric scores. However, in contrast, the study investigated only predictions of long-term changes in health rather than the short-term prediction of exacerbation as in this study.26 Neither did the study include independent clinical information or physiological data to corroborate.
We also observed association between habits data and vital signs of patients. For example, for one patient, bed occupancy dropped significantly when SpO2 levels fell below 85%; we believe this may be due to discomfort from dyspnea. For another patient, although we observed a slight increase in all day activity level, there was an increase in the duration of periods of inactivity. The reason for the increase in periods of inactivity was thought to be the unwillingness to move due to decreased exercise capacity or dyspnea. The increase in activity levels may be due to a slower walking pace or having to stop to catch their breath during exacerbation21,27 which may have resulted in two sensor events instead of one during the course of the completion of a task. These hypotheses could be tested in future studies by using accelerometers or position technology, and PIR sensors, to determine walking speed. Future studies should also collect symptoms in addition to vital signs, both of which are useful for evaluating risk of future exacerbations.28
Our previous work shows that changes in long-term vital signs data may have prognostic value and could be used to determine where there is need for intervention.23 In this study, only BP and SpO2 had useful information: low values of SpO2 led to referral for pulmonary assessment and O2 therapy; high values of BP, or low values when accompanied by dizziness, led to medication change and diagnosis of other conditions, such as AF. For some, the medication change was successful in establishing the desired level of BP, but for others the BP values continued to vary outside thresholds, requiring further medication change. We also noted that some BP and SpO2 records fluctuated with a periodic form; for example, the systolic BP for patient 1 varied between 120 and 150 mmHg with a period of 3 months.
It is clear that long-term vital signs data can provide information on the progress of the condition of a patient; however, there is currently a lack of well-defined procedures regarding how to deal with the long-term changes and trends in data, and this undermines the prognostic value of such data. For example, we observed clear indication of the progress of illness in patients with COPD in the long-term SpO2 readings.23 There is a need to determine approaches and gain knowledge to better use the long-term data to understand and manage the condition of patients. Without such approaches, effective use of all the information that is available from vital signs data is lost, and clinical trials to determine the effectiveness of telemonitoring systems are misleading, as not all the available information is being utilized.
However, we have seen that the number of clinical events in our patient population is small and so any approach to determine the effectiveness of the use of long-term vital signs data will require large, long-term observational studies.
In general, patients were very compliant and satisfied with use of the system. However, due to safety concerns regards balancing on the weigh scales, patients with CHF who had scored highly on the frailty scale did not weigh themselves often enough for reliable use of the alert algorithm or to enable management of their condition. In future studies it would be advisable to select weigh scales that are more appropriate for frail patients, such as including grab-on handles. It may also be necessary to collect other vital signs in addition to weight to have a better picture of a patient with CHF; this might include ECG, SpO2, and BP.29
Strengths
Due to the ease of use and unobtrusive features of the platform, we managed to collect telemonitoring data for longer than a year for most patients, which provided us with a significant amount of data and experience to understand: what was a useful set sensors; best sensor locations; issues on user acceptance; what works and what does not with elderly frail subjects; and the areas that can be improved.
The main benefit of the design of the technology was that it was easy to install and use (reduced training for patient); required no user interface (reduced complexity of use and increased acceptance); no (or very low) maintenance (reduced resource requirement from the service); self-contained (did not require broadband so could be installed in any home); and unobtrusive—patients could use the devices anywhere in the home (reduced stigmatization for the patient). The platform and devices worked seamlessly: the patients were measuring their vital signs as normal without need for additional steps (such as entering the reading in a logbook or website, or having to go to a base unit to take measurements); the technology was present but not noticeable or unduly disturbing to their daily routine. These factors improved overall usability and resulted in patients accepting a monitoring period for longer than 1 year. However, there were technical issues at the beginning of the monitoring period, primarily related to the bed sensor, due to discomfort of the sensor under the mattress and the high rate of false alerts. The nurses adjusted their response to the alerts and would only take action following several consecutive alerts of the same type.
The potential for integrated telemonitoring platforms with reliable alerts is significant. The advantages of remote patient monitoring for reducing hospitalization and well-being of the patient are well documented3; however, this work demonstrates that habits monitoring may provide as valuable information on detection of exacerbation as monitoring vital signs, so that the two may complement detection, and together may increase the accuracy of prediction.
Challenges
The high rate of false alarms is an issue for many telemonitoring systems23 and it is essential that reliable alarm algorithms are developed. However, development of such algorithms for habits data in particular, has been challenging for many reasons including type and number of sensors used; location of the sensors; house layout; and the presence of visitors. Developing robust algorithms and metrics that work reliably and effectively in the various settings and conditions is necessary for systems to be usable and deployed at scale. In addition, having multiple sensors and applying combined decision rules that use multiple parameters can further eliminate issues and reduce the rate of false alerts.
There is also a need to understand how best to analyze the habits data and present information to the professionals in a useful and manageable way. Recommendations for improved presentation of the data on the portal were made by the health professionals, which included the ability to view different levels of analysis of the data.30 For example, having seen an alert for habits on the portal, the health professionals wanted to see further details by clicking on a link, including a figure illustrating the long-term data (Figs. 5–7), hourly movement counts for each hour from all sensors (Figs. 3and9) and bed-time routines (Figs. 4and15) if possible.
There is a tendency for the attention of the clinician to be drawn to the clinical data rather than the habits data. Habits data can easily be ignored by the clinicians, especially when there is no reliable algorithm and the worth of the data is yet to be proven. This may undermine the effectiveness of habits monitoring. On the other hand, habits data may be of interest to a close relative or care provider, so that they might, for example, be reassured whether the patient is up and about.
The position of the gateway to ensure good signal to all devices was problematic in some homes. In such cases, a signal strength meter was used to identify appropriate locations, or repeaters installed to extend range.
Limitations
One of the limitations of this study is that we were unable to account for confounding factors and bias due to patient selection; doctors might have chosen subjects with severe conditions or at a late stage of the disease so that deterioration over short monitoring periods could be observed or intervention could take place. On the other hand, telemonitoring services appear to be more effective and beneficial for patients whose disease is at an advanced stage, as they are more likely to suffer from severe adverse events and they may need medical and social interventions7; this was the reason why the inCASA project focused on frail elderly with at least one chronic disease(s), and why many telemonitoring projects focus on these groups.2,31,32
Recommendations
Based on the challenges faced and lessons learned from our experience we are able to make some recommendations.
It is advantageous to have several PIR sensors; their location around the house would be, in descending order of importance: (1) living room, (2) bedroom, (3) bathroom, and (4) kitchen. PIR sensors in these locations not only monitor movements in the house, but can determine bathroom use and meal preparation.
A reliable bed sensor can provide vital information on the well-being of the subject, including bedtimes and bed occupancy. For example, bed occupancy results for a patient with COPD showed clear indication of the struggle to stay in bed when the condition was deteriorating (see patient 4). The bed sensors need to be designed to be comfortable and reliable, and to sense presence over a larger area of the bed than the pressure sensor used in this project.
Conclusion
We have collected and analyzed the data from combined habits and health monitoring of 36 frail elderly participants. We have detected deviations from their normal activity profile and in the physiological data. Long-term changes in activity profile and bed occupancy were associated with the condition of the patient. For example, from changes such as bedtimes and time between activities, we could clearly observe progress of the condition and response to intervention in patients with COPD and CHF. This association between the clinical condition of patients and their behavioral data is promising, but needs to be verified with a large study.
Although BP and SpO2 readings were found to be very useful, simple thresholds were problematic in generating too many false alerts, and the prognostic value of these can be improved with improved algorithms and well-defined protocols on how to deal with long-term data.
Our results also showed the importance of having a simple and unobtrusive telemonitoring platform and devices for use by frail elderly patients to achieve prolonged monitoring periods and acceptance.
Acknowledgments
The inCASA project was funded by the European Commission. The software for our devices was checked using the CodeSonar static analysis tool from Grammatech.
