Borislava Mihaylova

Review of statistical methods for analysing healthcare resources and costs.

We review statistical methods for analysing healthcare resource use and costs, their ability to address skewness, excess zeros, multimodality and heavy right tails, and their ease for general use. We aim to provide guidance on analysing resource use and costs focusing on randomised trials, although methods often have wider applicability. Twelve broad categories of methods were identified: (I) methods based on the normal distribution, (II) methods following transformation of data, (III) single-distribution generalized linear models (GLMs), (IV) parametric models based on skewed distributions outside the GLM family, (V) models based on mixtures of parametric distributions, (VI) two (or multi)-part and Tobit models, (VII) survival methods, (VIII) non-parametric methods, (IX) methods based on truncation or trimming of data, (X) data components models, (XI) methods based on averaging across models, and (XII) Markov chain methods. Based on this review, our recommendations are that, first, simple methods are preferred in large samples where the near-normality of sample means is assured. Second, in somewhat smaller samples, relatively simple methods, able to deal with one or two of above data characteristics, may be preferable but checking sensitivity to assumptions is necessary. Finally, some more complex methods hold promise, but are relatively untried; their implementation requires substantial expertise and they are not currently recommended for wider applied work. Copyright

Cost-effectiveness of simvastatin in people at different levels of vascular disease risk: economic analysis of a randomised trial in 20,536 individuals.

BACKGROUND Statin therapy reduces the rates of heart attack, stroke, and revascularisation among a wide range of individuals. Reliable assessment of its cost-effectiveness in different circumstances is needed. METHODS 20,536 adults (aged 40-80 years) with vascular disease or diabetes were randomly allocated 40 mg simvastatin daily (10,269) or placebo (10,267) for an average of 5 years. Comparisons were made of hospitalisation and statin costs (2001 UK prices) during the scheduled treatment period between all simvastatin-allocated versus all placebo-allocated participants. Cost-effectiveness was estimated among different categories of participant. FINDINGS Allocation to simvastatin was associated with a highly significant 22% (95% CI 16-27; p<0.0001) proportional reduction in hospitalisation costs for all vascular events, with similar proportional reductions in every subcategory of participant studied. During an average of 5 years, estimated absolute reductions in vascular event costs per person allocated 40 mg simvastatin daily ranged from UK 847 pounds sterling (SE 137) in the highest risk quintile studied to 264 pounds sterling (48) in the lowest. Mean excess cost of statin therapy among participants allocated simvastatin was 1497 pounds sterling (8), with similar absolute increases in every subcategory. Costs of preventing a major vascular event with 40 mg simvastatin daily ranged from 4500 pounds sterling (95% CI 2300-7400) among participants with a 42% 5-year major vascular event rate to 31,100 pounds sterling (22,900-42,500) among those with a 12% rate (corresponding to 5-year major coronary event rates of 22% and 4%, respectively). INTERPRETATION Statin therapy is cost effective for a wider range of individuals with vascular disease or diabetes than previously recognised (particularly with lower-priced generics). It would be appropriate to consider reducing the estimated level of vascular event risk at which statin therapy is recommended.

Lifetime cost effectiveness of simvastatin in a range of risk groups and age groups derived from a randomised trial of 20,536 people.

Objectives To evaluate the cost effectiveness of 40 mg simvastatin daily continued for life in people of different ages with differing risks of vascular disease. Design A model developed from a randomised trial was used to estimate lifetime risks of vascular events and costs of treatment and hospital admissions in the United Kingdom. Setting 69 hospitals in the UK. Participants 20 536 men and women (aged 40-80) with coronary disease, other occlusive arterial disease, or diabetes. Interventions 40 mg simvastatin daily versus placebo for an average of 5 years. Main outcome measures Cost effectiveness of 40 mg simvastatin daily expressed as additional cost per life year gained. Major vascular event defined as non-fatal myocardial infarction or death from coronary disease, any stroke, or revascularisation procedure. Results were extrapolated to younger and older age groups at lower risk of vascular disease than were studied directly, as well as to lifetime treatment. Results At the April 2005 UK price of �4.87 (€7;

The impact of diabetes-related complications on healthcare costs: new results from the UKPDS (UKPDS 84).

9) per 28 day pack of generic 40 mg simvastatin, lifetime treatment was cost saving in most age groups and vascular disease risk groups studied directly. Gains in life expectancy and cost savings decreased with increasing age and with decreasing risk of vascular disease. People aged 40-49 with 5 year risks of major vascular events of 42% and 12% at start of treatment gained 2.49 and 1.67 life years, respectively. Treatment with statins remained cost saving or cost less than �2500 per life year gained in people as young as 35 years or as old as 85 with 5 year risks of a major vascular event as low as 5% at the start of treatment. Conclusions Treatment with statins is cost effective in a wider population than is routinely treated at present.

Cost effectiveness of perindopril in reducing cardiovascular events in patients with stable coronary artery disease using data from the EUROPA study

To estimate the immediate and long‐term inpatient and non‐inpatient costs for Type 2 diabetes‐related complications.

Impact of renal function on the effects of LDL cholesterol lowering with statin-based regimens: a meta-analysis of individual participant data from 28 randomised trials

Background: The EUropean trial on Reduction Of cardiac events with Perindopril in stable coronary Artery disease (EUROPA) trial has recently reported. Objective: To assess the cost effectiveness of perindopril in stable coronary heart disease in the UK. Methods: Clinical and resource use data were taken from the EUROPA trial. Costs included drugs and hospitalisations. Health-related quality of life values were taken from published sources. A cost-effectiveness analysis is presented as a function of the risk of a primary event (non-fatal myocardial infarction, cardiac arrest or cardiovascular death) in order to identify people for whom treatment offers greatest value for money. Results: The median incremental cost of perindopril for each quality-adjusted life year (QALY) gained across the heterogeneous population of EUROPA was estimated as £9700 (interquartile range £6400–£14 200). Overall, 88% of the EUROPA population had an estimated cost per QALY below £20 000 and 97% below £30 000. For a threshold value of cost effectiveness of £30 000 per QALY gained, treatment of people representing the 25th, 50th (median) and 75th centiles of the cost effectiveness distribution for perindopril has a probability of 0.999, 0.99 and 0.93 of being cost effective, respectively. Cost effectiveness was strongly related to higher risk of a primary event under standard care. Conclusions: Whether the use of perindopril can be considered cost effective depends on the threshold value of cost effectiveness of healthcare systems. For the large majority of patients included in EUROPA, the incremental cost per QALY gained was lower than the apparent threshold used by the National Institute for Health and Clinical Excellence in the UK.

The effect of diabetes complications on health-related quality of life: the importance of longitudinal data to address patient heterogeneity.

BACKGROUND Statin therapy is effective for the prevention of coronary heart disease and stroke in patients with mild-to-moderate chronic kidney disease, but its effects in individuals with more advanced disease, particularly those undergoing dialysis, are uncertain. METHODS We did a meta-analysis of individual participant data from 28 trials (n=183 419), examining effects of statin-based therapy on major vascular events (major coronary event [non-fatal myocardial infarction or coronary death], stroke, or coronary revascularisation) and cause-specific mortality. Participants were subdivided into categories of estimated glomerular filtration rate (eGFR) at baseline. Treatment effects were estimated with rate ratio (RR) per mmol/L reduction in LDL cholesterol. FINDINGS Overall, statin-based therapy reduced the risk of a first major vascular event by 21% (RR 0·79, 95% CI 0·77-0·81; p<0·0001) per mmol/L reduction in LDL cholesterol. Smaller relative effects on major vascular events were observed as eGFR declined (p=0·008 for trend; RR 0·78, 99% CI 0·75-0·82 for eGFR ≥60 mL/min per 1·73 m(2); 0·76, 0·70-0·81 for eGFR 45 to <60 mL/min per 1·73 m(2); 0·85, 0·75-0·96 for eGFR 30 to <45 mL/min per 1·73 m(2); 0·85, 0·71-1·02 for eGFR <30 mL/min per 1·73 m(2) and not on dialysis; and 0·94, 0·79-1·11 for patients on dialysis). Analogous trends by baseline renal function were seen for major coronary events (p=0·01 for trend) and vascular mortality (p=0·03 for trend), but there was no significant trend for coronary revascularisation (p=0·90). Reducing LDL cholesterol with statin-based therapy had no effect on non-vascular mortality, irrespective of eGFR. INTERPRETATION Even after allowing for the smaller reductions in LDL cholesterol achieved by patients with more advanced chronic kidney disease, and for differences in outcome definitions between dialysis trials, the relative reductions in major vascular events observed with statin-based treatment became smaller as eGFR declined, with little evidence of benefit in patients on dialysis. In patients with chronic kidney disease, statin-based regimens should be chosen to maximise the absolute reduction in LDL cholesterol to achieve the largest treatment benefits. FUNDING UK Medical Research Council, British Heart Foundation, Cancer Research UK, European Community Biomed Programme, Australian National Health and Medical Research Council, Australian National Heart Foundation.

The effect of diabetes complications on health-related quality of life

We estimate the impact of six diabetes-related complications (myocardial infarction, ischaemic heart disease, stroke, heart failure, amputation and visual acuity) on quality of life, using seven rounds of EQ-5D questionnaires administered between 1997 and 2007 in the UK Prospective Diabetes Study. The use of cross-sectional data to make such estimates is widespread in the literature, being less expensive and easier to collect than repeated-measures data. However, analysis of this dataset suggests that cross-sectional analysis could produce biased estimates of the effect of complications on QoL. Using fixed effects estimators, we show that variation in the quality of life between patients is strongly influenced by time-invariant patient characteristics. Our results highlight the importance of studying quality-of-life changes over time to distinguish between time-invariant determinants of QoL and the effect on QoL of specific events such as diabetes complications.

What is the impact of chronic kidney disease stage and cardiovascular disease on the annual cost of hospital care in moderate-to-severe kidney disease?

We estimate the impact of six diabetes-related complications (myocardial infarction, ischaemic heart disease, stroke, heart failure, amputation and visual acuity) on quality of life, using seven rounds of EQ-5D questionnaires administered between 1997 and 2007 in the UK Prospective Diabetes Study. The use of cross-sectional data to make such estimates is widespread in the literature, being less expensive and easier to collect than repeated-measures data. However, analysis of this dataset suggests that cross-sectional analysis could produce biased estimates of the effect of complications on QoL. Using fixed effects estimators, we show that variation in the quality of life between patients is strongly influenced by time-invariant patient characteristics. Our results highlight the importance of studying quality-of-life changes over time to distinguish between time-invariant determinants of QoL and the effect on QoL of specific events such as diabetes complications.

Cost-Effectiveness of Duloxetine:The Stress Urinary Incontinence Treatment (SUIT) Study

BackgroundReliable estimates of the impacts of chronic kidney disease (CKD) stage, with and without cardiovascular disease, on hospital costs are needed to inform health policy.MethodsThe Study of Heart and Renal Protection (SHARP) randomized trial prospectively collected information on kidney disease progression, serious adverse events and hospital care use in a cohort of patients with moderate-to-severe CKD. In a secondary analysis of SHARP data, the impact of participants’ CKD stage, non-fatal cardiovascular events and deaths on annual hospital costs (i.e. all hospital admissions, routine dialysis treatments and recorded outpatient/day-case attendances in United Kingdom 2011 prices) were estimated using linear regression.Results7,246 SHARP patients (2,498 on dialysis at baseline) from Europe, North America, and Australasia contributed 28,261 years of data. CKD patients without diabetes or vascular disease incurred annual hospital care costs ranging from £403 (95% confidence interval: 345-462) in CKD stages 1-3B to £525 (449-602) in CKD stage 5 (not on dialysis). Patients in receipt of maintenance dialysis incurred annual hospital costs of £18,986 (18,620-19,352) in the year of initiation and £23,326 (23,231-23,421) annually thereafter. Patients with a functioning kidney transplant incurred £24,602 (24,027-25,178) in hospital care costs in the year of transplantation and £1,148 (978-1,318) annually thereafter. Non-fatal major vascular events increased annual costs in the year of the event by £6,133 (5,608-6,658) for patients on dialysis and by £4,350 (3,819-4,880) for patients not on dialysis, and were associated with increased costs, though to a lesser extent, in subsequent years.ConclusionsRenal replacement therapy and major vascular events are the main contributors to the high hospital care costs in moderate-to-severe CKD. These estimates of hospital costs can be used to inform health policy in moderate-to-severe CKD.