Estimating Costs Associated with Disease Model States Using Generalized Linear Models: A Tutorial.

Estimates of costs associated with disease states are required to inform decision analytic disease models to evaluate interventions that modify disease trajectory. Increasingly, decision analytic models are developed using patient-level data with a focus on heterogeneity between patients, and there is a demand for costs informing such models to reflect individual patient costs. Statistical models of health care costs need to recognize the specific features of costs data which typically include a large number of zero observations for non-users, and a skewed and heavy right-hand tailed distribution due to a small number of heavy healthcare users. Different methods are available for modelling costs, such as generalized linear models (GLMs), extended estimating equations and latent class approaches. While there are tutorials addressing approaches to decision modelling, there is no practical guidance on the cost estimation to inform such models. Therefore, this tutorial aims to provide a general guidance on estimating healthcare costs associated with disease states in decision analytic models. Specifically, we present a step-by-step guide to how individual participant data can be used to estimate costs over discrete periods for participants with particular characteristics, based on the GLM framework. We focus on the practical aspects of cost modelling from the conceptualization of the research question to the derivation of costs for an individual in particular disease states. We provide a practical example with step-by-step R code illustrating the process of modelling the hospital costs associated with disease states for a cardiovascular disease model.

Impact of New Cardiovascular Events on Quality of Life and Hospital Costs in People With Cardiovascular Disease in the United Kingdom and United States.

BACKGROUND: Despite optimized risk factor control, people with prior cardiovascular disease remain at high cardiovascular disease risk. We assess the immediate- and longer-term impacts of new vascular and nonvascular events on quality of life (QoL) and hospital costs among participants in the REVEAL (Randomized Evaluation of the Effects of Anacetrapib Through Lipid Modification) trial in secondary prevention. METHODS AND RESULTS: Data on demographic and clinical characteristics, health-related quality of life (QoL: EuroQoL 5-Dimension-5-Level), adverse events, and hospital admissions during the 4-year follow-up of the 21 820 participants recruited in Europe and North America informed assessments of the impacts of new adverse events on QoL and hospital costs from the UK and US health systems' perspectives using generalized linear regression models. Reductions in QoL were estimated in the years of event occurrence for nonhemorrhagic stroke (-0.067 [United Kingdom], -0.069 [US]), heart failure admission (-0.072 [United Kingdom], -0.103 [US]), incident cancer (-0.064 [United Kingdom], -0.068 [US]), and noncoronary revascularization (-0.071 [United Kingdom], -0.061 [US]), as well as in subsequent years following these events. Myocardial infarction and coronary revascularization (CRV) procedures were not found to affect QoL. All adverse events were associated with additional hospital costs in the years of events and in subsequent years, with the highest additional costs in the years of noncoronary revascularization (£5830 [United Kingdom], $14 133 [US Medicare]), of myocardial infarction with urgent CRV procedure (£5614, $24722), and of urgent/nonurgent CRV procedure without myocardial infarction (£4674/£4651 and $15 251/$17 539). CONCLUSIONS: Stroke, heart failure, and noncoronary revascularization procedures substantially reduce QoL, and all cardiovascular disease events increase hospital costs. These estimates are useful in informing cost-effectiveness of interventions to reduce cardiovascular disease risk in secondary prevention. REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifier: NCT01252953; https://www.Isrctn.com. Unique identifier: ISRCTN48678192; https://www.clinicaltrialsregister.eu. Unique identifier: 2010-023467-18.

Hospital costs associated with adverse events in people with diabetes in the UK.

AIM: To estimate the annual hospital costs associated with a range of adverse events for people with diabetes in the UK. METHODS: Annual hospital costs (2019/2020) were derived from 15 436 ASCEND participants from 2005 to 2017 (120 420 person-years). The annual hospital costs associated with cardiovascular events (myocardial infarction, coronary revascularization, transient ischaemic attack [TIA], ischaemic stroke, heart failure), bleeding (gastrointestinal [GI] bleed, intracranial haemorrhage, other major bleed), cancer (GI tract cancer, non-GI tract cancer), end-stage renal disease (ESRD), lower limb amputation and death (vascular, non-vascular) were estimated using a generalized linear model following adjustment for participants' sociodemographic and clinical factors. RESULTS: In the year of event, ESRD was associated with the largest increase in annual hospital cost (£20 954), followed by lower limb amputation (£17 887), intracranial haemorrhage (£12 080), GI tract cancer (£10 160), coronary revascularization (£8531 if urgent; £8302 if non-urgent), heart failure (£8319), non-GI tract cancer (£7409), ischaemic stroke (£7170), GI bleed (£5557), myocardial infarction (£4913), other major bleed (£3825) and TIA (£1523). In subsequent years, most adverse events were associated with lasting but smaller increases in hospital costs, except for ESRD, where the additional cost remained high (£20 090). CONCLUSIONS: Our study provides robust estimates of annual hospital costs associated with a range of adverse events in people with diabetes that can inform future cost-effectiveness analyses of diabetes interventions. It also highlights the potential cost savings that could be derived from prevention of these costly complications.

Performance of the UK Prospective Diabetes Study Outcomes Model 2 in a Contemporary UK Type 2 Diabetes Trial Cohort.

OBJECTIVES: The UK Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS-OM) developed using 30-year (1977-2007) data from the UKPDS is widely used for health outcomes' projections and economic evaluations of therapies for patients with type 2 diabetes (T2D). Nevertheless, its reliability for contemporary UK T2D populations is unclear. We assessed the performance of version 2 of the model (UKPDS-OM2) using data from A Study of Cardiovascular Events in Diabetes (ASCEND), which followed participants with diabetes in the UK between 2005 and 2017. METHODS: The UKPDS-OM2 was used to predict the occurrence of myocardial infarction (MI), other ischemic heart disease, stroke, cardiovascular (CV) death, and other death among the 14 569 participants with T2D in ASCEND, all without previous CV disease at study entry. Calibration (comparison of predicted and observed year-on-year cumulative incidence over 10 years) and discrimination (c-statistics) of the model were assessed for each endpoint. The percentage error in event rates at year 7 (mean duration of follow up) was used to quantify model bias. RESULTS: The UKPDS-OM2 substantially overpredicted MI, stroke, CV death, and other death over the 10-year follow-up period (by 149%, 42%, 269%, and 52%, respectively, at year 7). Discrimination of the model for MI and other ischemic heart disease (c-statistics 0.58 and 0.60, respectively) was poorer than that for other outcomes (c-statistics ranging from 0.66 to 0.72). CONCLUSIONS: The UKPDS-OM2 substantially overpredicted risks of key CV outcomes and death in people with T2D in ASCEND. Appropriate adjustments or a new model may be required for assessments of long-term effects of treatments in contemporary T2D cohorts.

Decrements in health-related quality of life associated with adverse events in people with diabetes.

AIM: To estimate the decrements in health-related quality of life (QoL) associated with a range of adverse events to inform assessments of the effects of diabetes treatments on QoL in contemporary clinical practice. METHODS: Participants' QoL utility measures were derived from the five-level EuroQoL five-dimensional (EQ-5D-5L) questionnaires completed by 11 683 ASCEND participants (76% of 15 480 recruited). EQ-5D utility decrements associated with cardiovascular (myocardial infarction, coronary revascularization, transient ischaemic attack [TIA], ischaemic stroke, heart failure), bleeding (gastrointestinal [GI] bleed, intracranial haemorrhage, other major bleed), cancer (GI tract cancer, non-GI tract cancer), and microvascular events (end-stage renal disease [ESRD], amputation) were estimated using a linear regression model following adjustment for participants' sociodemographic and clinical risk factors. RESULTS: Amputation was associated with the largest EQ-5D utility decrement (-0.206), followed by heart failure (-0.185), intracranial haemorrhage (-0.164), GI bleed (-0.091), other major bleed (-0.096), ischaemic stroke (-0.061), TIA (-0.057), and non-GI tract cancer (-0.026). We were unable to detect decrements in EQ-5D utility associated with myocardial infarction, coronary revascularization, GI tract cancer, or ESRD. EQ-5D utility was lower at older age, independent of other factors. CONCLUSION: These estimated decrements in QoL associated with cardiovascular, bleeding, cancer, and other adverse events can inform assessments of the overall value of treatments in patients with diabetes.

How valid are projections of the future prevalence of diabetes? Rapid reviews of prevalence-based and Markov chain models and comparisons of different models' projections for England.

OBJECTIVES: To examine validity of prevalence-based models giving projections of prevalence of diabetes in adults, in England and the UK, and of Markov chain models giving estimates of economic impacts of interventions to prevent type 2 diabetes (T2D). METHODS: Rapid reviews of both types of models. Estimation of the future prevalence of T2D in England by Markov chain models; and from the trend in the prevalence of diabetes, as reported in the Quality and Outcomes Framework (QOF), estimated by ordinary least squares regression analysis. SETTING: Adult population in England and UK. MAIN OUTCOME MEASURE: Prevalence of T2D in England and UK in 2025. RESULTS: The prevalence-based models reviewed use sample estimates of past prevalence rates by age and sex and projected population changes. Three most recent models, including that of Public Health England (PHE), neither take account of increases in obesity, nor report Confidence Intervals (CIs). The Markov chain models reviewed use transition probabilities between states of risk and death, estimated from various sources. None of their accounts give the full matrix of transition probabilities, and only a minority report tests of validation. Their primary focus is on estimating the ratio of costs to benefits of preventive interventions in those with hyperglycaemia, only one reported estimates of those developing T2D in the absence of a preventive intervention in the general population.Projections of the prevalence of T2D in England in 2025 were (in millions) by PHE, 3.95; from the QOF trend, 4.91 and by two Markov chain models, based on our review, 5.64 and 9.07. CONCLUSIONS: To inform national policies on preventing T2D, governments need validated models, designed to use available data, which estimate the scale of incidence of T2D and survival in the general population, with and without preventive interventions.

Impact of achieving primary care targets in type 2 diabetes on health outcomes and healthcare costs.

AIMS: In England and Wales, the National Diabetes Audit (NDA) assesses the quality of management of type 2 diabetes (T2D) in primary care using treatment targets for HbA1c ≤58 mmol/mol, total cholesterol <5 mmol/L and blood pressure ≤140/80 mm Hg. We quantified the impact of variation in achieving these targets on health outcomes and healthcare costs across general practitioners' (GP) practices. METHODS: Summary of characteristics of T2D patients from the 2015-2016 NDA were used to generate representative populations of T2D patients. The UKPDS Outcomes Model 2 was used to estimate long-term health outcomes and healthcare costs. The effects of achieving treatment targets on these outcomes were evaluated using regression models. RESULTS: Achieving more of the HbA1c, cholesterol and blood pressure targets led to a lower incidence of diabetes-related complications. Approximately 0.5 (95% CI, 0.4-0.6) quality-adjusted life years (QALYs) and 0.6 (95% CI, 0.4-0.7) years of life (LYs) were gained by T2D patients over a lifetime for each additional target met. The projected healthcare cost savings arising from fewer diabetes-related complications as the result of achieving one, two or three targets compared to none were £859 (95% CI, £553-£1165), £940 (95% CI, £485-£1395) and £1037 (95% CI, £414-£1660) over a patient's lifetime. A typical GP practice in the lowest performing decile (average, 371 T2D patients per practice, with 27% achieving all targets) is projected to gain 201 (95% CI, 123-279) QALYs and 231 (95% CI, 133-329) LYs, if all T2D patients achieved all three targets. CONCLUSIONS: Substantial gains in health outcomes and reductions in healthcare costs could be achieved with further improvements in attainment of HbA1c, cholesterol and blood pressure targets for T2D patients.