Faculty Bibliography

OBJECTIVES: To test the performance of new variants of models to identify people at risk of an emergency hospital admission. We compared (1) the impact of using alternative data sources (hospital inpatient, A&E, outpatient and general practitioner (GP) electronic medical records) (2) the effects of local calibration on the performance of the models and (3) the choice of population denominators. DESIGN: Multivariate logistic regressions using person-level data adding each data set sequentially to test value of additional variables and denominators. SETTING: 5 Primary Care Trusts within England. PARTICIPANTS: 1 836 099 people aged 18-95 registered with GPs on 31 July 2009. MAIN OUTCOME MEASURES: Models to predict hospital admission and readmission were compared in terms of the positive predictive value and sensitivity for various risk strata and with the receiver operating curve C statistic. RESULTS: The addition of each data set showed moderate improvement in the number of patients identified with little or no loss of positive predictive value. However, even with inclusion of GP electronic medical record information, the algorithms identified only a small number of patients with no emergency hospital admissions in the previous 2 years. The model pooled across all sites performed almost as well as the models calibrated to local data from just one site. Using population denominators from GP registers led to better case finding. CONCLUSIONS: These models provide a basis for wider application in the National Health Service. Each of the models examined produces reasonably robust performance and offers some predictive value. The addition of more complex data adds some value, but we were unable to conclude that pooled models performed less well than those in individual sites. Choices about model should be linked to the intervention design. Characteristics of patients identified by the algorithms provide useful information in the design/costing of intervention strategies to improve care coordination/outcomes for these patients.

OBJECTIVE: to assess the impact of telecare on the use of social and health care. Part of the evaluation of the Whole Systems Demonstrator trial.Participants and setting: a total of 2,600 people with social care needs were recruited from 217 general practices in three areas in England. DESIGN: a cluster randomised trial comparing telecare with usual care, general practice being the unit of randomisation. Participants were followed up for 12 months and analyses were conducted as intention-to-treat.Data sources: trial data were linked at the person level to administrative data sets on care funded at least in part by local authorities or the National Health Service.Main outcome measures: the proportion of people admitted to hospital within 12 months. Secondary endpoints included mortality, rates of secondary care use (seven different metrics), contacts with general practitioners and practice nurses, proportion of people admitted to permanent residential or nursing care, weeks in domiciliary social care and notional costs. RESULTS: 46.8% of intervention participants were admitted to hospital, compared with 49.2% of controls. Unadjusted differences were not statistically significant (odds ratio: 0.90, 95% CI: 0.75-1.07, P = 0.211). They reached statistical significance after adjusting for baseline covariates, but this was not replicated when adjusting for the predictive risk score. Secondary metrics including impacts on social care use were not statistically significant. CONCLUSIONS: telecare as implemented in the Whole Systems Demonstrator trial did not lead to significant reductions in service use, at least in terms of results assessed over 12 months.International Standard Randomised Controlled Trial Number Register ISRCTN43002091.

OBJECTIVE: To test whether two hospital-avoidance interventions altered rates of hospital use: "intermediate care" and "integrated care teams." DATA SOURCES/STUDY SETTING: Linked administrative data for England covering the period 2004 to 2009. STUDY DESIGN: This study was commissioned after the interventions had been in place for several years. We developed a method based on retrospective analysis of person-level data comparing health care use of participants with that of prognostically matched controls. DATA COLLECTION/EXTRACTION METHODS: Individuals were linked to administrative datasets through a trusted intermediary and a unique patient identifier. PRINCIPAL FINDINGS: Participants who received the intermediate care intervention showed higher rates of unscheduled hospital admission than matched controls, whereas recipients of the integrated care team intervention showed no difference. Both intervention groups showed higher rates of mortality than did their matched controls. CONCLUSIONS: These are potentially powerful techniques for assessing impacts on hospital activity. Neither intervention reduced admission rates. Although our analysis of hospital utilization controlled for a wide range of observable characteristics, the difference in mortality rates suggests that some residual confounding is likely. Evaluation is constrained when performed retrospectively, and careful interpretation is needed.

OBJECTIVE: To assess the effect of home based telehealth interventions on the use of secondary healthcare and mortality. DESIGN: Pragmatic, multisite, cluster randomised trial comparing telehealth with usual care, using data from routine administrative datasets. General practice was the unit of randomisation. We allocated practices using a minimisation algorithm, and did analyses by intention to treat. SETTING: 179 general practices in three areas in England. PARTICIPANTS: 3230 people with diabetes, chronic obstructive pulmonary disease, or heart failure recruited from practices between May 2008 and November 2009. INTERVENTIONS: Telehealth involved remote exchange of data between patients and healthcare professionals as part of patients' diagnosis and management. Usual care reflected the range of services available in the trial sites, excluding telehealth. MAIN OUTCOME MEASURE: Proportion of patients admitted to hospital during 12 month trial period. RESULTS: Patient characteristics were similar at baseline. Compared with controls, the intervention group had a lower admission proportion within 12 month follow-up (odds ratio 0.82, 95% confidence interval 0.70 to 0.97, P = 0.017). Mortality at 12 months was also lower for intervention patients than for controls (4.6% v 8.3%; odds ratio 0.54, 0.39 to 0.75, P < 0.001). These differences in admissions and mortality remained significant after adjustment. The mean number of emergency admissions per head also differed between groups (crude rates, intervention 0.54 v control 0.68); these changes were significant in unadjusted comparisons (incidence rate ratio 0.81, 0.65 to 1.00, P = 0.046) and after adjusting for a predictive risk score, but not after adjusting for baseline characteristics. Length of hospital stay was shorter for intervention patients than for controls (mean bed days per head 4.87 v 5.68; geometric mean difference -0.64 days, -1.14 to -0.10, P = 0.023, which remained significant after adjustment). Observed differences in other forms of hospital use, including notional costs, were not significant in general. Differences in emergency admissions were greatest at the beginning of the trial, during which we observed a particularly large increase for the control group. CONCLUSIONS: Telehealth is associated with lower mortality and emergency admission rates. The reasons for the short term increases in admissions for the control group are not clear, but the trial recruitment processes could have had an effect. TRIAL REGISTRATION NUMBER: International Standard Randomised Controlled Trial Number Register ISRCTN43002091.

OBJECTIVES: To develop an algorithm for identifying inpatients at high risk of re-admission to a National Health Service (NHS) hospital in England within 30 days of discharge using information that can either be obtained from hospital information systems or from the patient and their notes. DESIGN: Multivariate statistical analysis of routinely collected hospital episode statistics (HES) data using logistic regression to build the predictive model. The model's performance was calculated using bootstrapping. SETTING: HES data covering all NHS hospital admissions in England. PARTICIPANTS: The NHS patients were admitted to hospital between April 2008 and March 2009 (10% sample of all admissions, n=576 868). MAIN OUTCOME MEASURES: Area under the receiver operating characteristic curve for the algorithm, together with its positive predictive value and sensitivity for a range of risk score thresholds. RESULTS: The algorithm produces a 'risk score' ranging (0-1) for each admitted patient, and the percentage of patients with a re-admission within 30 days and the mean re-admission costs of all patients are provided for 20 risk bands. At a risk score threshold of 0.5, the positive predictive value (ie, percentage of inpatients identified as high risk who were subsequently re-admitted within 30 days) was 59.2% (95% CI 58.0% to 60.5%); representing 5.4% (95% CI 5.2% to 5.6%) of all inpatients who would be re-admitted within 30 days (sensitivity). The area under the receiver operating characteristic curve was 0.70 (95% CI 0.69 to 0.70). CONCLUSIONS: We have developed a method of identifying inpatients at high risk of unplanned re-admission to NHS hospitals within 30 days of discharge. Though the models had a low sensitivity, we show how to identify subgroups of patients that contain a high proportion of patients who will be re-admitted within 30 days. Additional work is necessary to validate the model in practice.

OBJECTIVES: To develop a formula for allocating resources for commissioning hospital care to all general practices in England based on the health needs of the people registered in each practice DESIGN: Multivariate prospective statistical models were developed in which routinely collected electronic information from 2005-6 and 2006-7 on individuals and the areas in which they lived was used to predict their costs of hospital care in the next year, 2007-8. Data on individuals included all diagnoses recorded at any inpatient admission. Models were developed on a random sample of 5 million people and validated on a second random sample of 5 million people and a third sample of 5 million people drawn from a random sample of practices. SETTING: All general practices in England as of 1 April 2007. All NHS inpatient admissions and outpatient attendances for individuals registered with a general practice on that date. SUBJECTS: All individuals registered with a general practice in England at 1 April 2007. MAIN OUTCOME MEASURES: Power of the statistical models to predict the costs of the individual patient or each practice's registered population for 2007-8 tested with a range of metrics (R(2) reported here). Comparisons of predicted costs in 2007-8 with actual costs incurred in the same year were calculated by individual and by practice. RESULTS: Models including person level information (age, sex, and ICD-10 codes diagnostic recorded) and a range of area level information (such as socioeconomic deprivation and supply of health facilities) were most predictive of costs. After accounting for person level variables, area level variables added little explanatory power. The best models for resource allocation could predict upwards of 77% of the variation in costs at practice level, and about 12% at the person level. With these models, the predicted costs of about a third of practices would exceed or undershoot the actual costs by 10% or more. Smaller practices were more likely to be in these groups. CONCLUSIONS: A model was developed that performed well by international standards, and could be used for allocations to practices for commissioning. The best formulas, however, could predict only about 12% of the variation in next year's costs of most inpatient and outpatient NHS care for each individual. Person-based diagnostic data significantly added to the predictive power of the models.

BACKGROUND: This retrospective study will assess the extent to which multidisciplinary case management in the form of virtual wards (VWs) leads to changes in the use of health care and social care by patients at high risk of future unplanned hospital admission. VWs use the staffing, systems and daily routines of a hospital ward to deliver coordinated care to patients in their own homes. Admission to a VW is offered to patients identified by a predictive risk model as being at high risk of unplanned hospital admission in the coming 12 months. STUDY DESIGN AND DATA COLLECTION METHODS: We will compare the health care and social care use of VW patients to that of matched controls. Controls will be drawn from (a) national, and (b) local, individual-level pseudonymous routine data. The costs of setting up and running a VW will be determined from the perspectives of both health and social care organizations using a combination of administrative data, interviews and diaries. METHODS OF ANALYSIS: Using propensity score matching and prognostic matching, we will create matched comparator groups to estimate the effect size of virtual wards in reducing unplanned hospital admissions. CONCLUSIONS: THIS STUDY WILL ALLOW US TO DETERMINE RELATIVE TO MATCHED COMPARATOR GROUPS: whether VWs reduce the use of emergency hospital care; the impact, if any, of VWs on the uptake of primary care, community health services and council-funded social care; and the potential costs and savings of VWs from the perspectives of the national health service (NHS) and local authorities.

BACKGROUND: the costs of delivering health and social care services are rising as the population ages and more people live with chronic diseases. OBJECTIVES: to determine whether predictive risk models can be built that use routine health and social care data to predict which older people will begin receiving intensive social care. DESIGN: analysis of pseudonymous, person-level, data extracted from the administrative data systems of local health and social care organisations. SETTING: five primary care trust areas in England and their associated councils with social services responsibilities. SUBJECTS: people aged 75 or older registered continuously with a general practitioner in five selected areas of England (n = 155,905). METHODS: multivariate statistical analysis using a split sample of data. RESULTS: it was possible to construct models that predicted which people would begin receiving intensive social care in the coming 12 months. The performance of the models was improved by selecting a dependent variable based on a lower cost threshold as one of the definitions of commencing intensive social care. CONCLUSIONS: predictive models can be constructed that use linked, routine health and social care data for case finding in social care settings.