ORIGINAL ARTICLE
Estimating Cost-Effectiveness in Type 2 Diabetes: The Impact of Treatment Guidelines and Therapy Duration Phil McEwan, PhD, Jason Gordon, PhD, Marc Evans, MD, Thomas Ward, MSc, Hayley Bennett, MSc, Klas Bergenheim, PhD
Objectives. Type 2 diabetes mellitus (T2DM) clinical guidelines focus on optimizing glucose control, with therapy escalation classically initiated within a ‘‘failurebased’’ regimen. Within many diabetes models, HbA1c therapy escalation thresholds play a pivotal role, controlling duration of therapy and, consequently, incremental costs and benefits. The objective of this study was to assess the relationship between therapy escalation threshold and time to therapy escalation on predicted cost-effectiveness of T2DM treatments. Methods. This study used the Cardiff Diabetes Model to illustrate the relationship between costs and health outcomes associated with first-, second-, and third-line therapy as a function of time on each. Data from routine clinical practice were used to contrast predicted costs and health outcomes associated with guideline therapy escalation thresholds compared with clinical practice. The impact of baseline HbA1c and therapy escalation thresholds on cost-effectiveness was investigated, comparing a sodium/ glucose cotransporter 2 inhibitor v. sulfonylurea added to
metformin monotherapy. Results. Lower thresholds are associated with a shorter time spent on monotherapy, ranging from 1.1 years (escalation at 6.5%) to 13 years (escalation at 9.0%) and an increase in total lifetime cost of therapy. Treatment-related disutility is minimized with higher thresholds because progression to insulin is delayed. Using metformin combined with either dapagliflozin or a sulfonylurea to illustrate lower baseline HbA1c and/or higher therapy escalation thresholds was associated with increased cost-effectiveness ratios, driven by a longer duration of therapy. Discussion. A marked difference in treatment cost-effectiveness was demonstrated when comparing routine clinical practice with guideline-advocated therapy escalation. This is important to both health care professionals and the wider health economic community with respect to understanding the true cost-effectiveness profile of any particular T2DM therapy option. Key words: decision aids; internal medicine; provider decision making; cost-effectiveness analysis. (Med Decis Making 20XX;XX: XXX–XXX)
T
morbidity and mortality.1,2 The current and expected future health care expenditure associated with managing T2DM is substantial. In the United Kingdom, annual direct health care costs for T2DM are estimated at £8.8 billion,3 and recent studies have reported on the increasing trends in diabetes-specific expenditure associated with treatment costs and acute hospital care.4,5 Of primary concern to health care policy makers is the optimal management of patients with T2DM to minimize health care expenditure and maximize
ype 2 diabetes mellitus (T2DM) is a chronic condition associated with substantial excess
Received 13 January 2014 from the Swansea Centre for Health Economics, Swansea University, Wales, UK (PCM, HB); Health Economics & Outcomes Research Ltd., Wales, UK (PCM, JG, TW); University Health Board, Llandough, Wales, UK (ME); Global Health Economics & Outcomes Research, AstraZeneca, Molndal, Sweden (KB); and Department of Public Health, University of Adelaide, South Australia, Australia (JG). Financial support for this study was provided entirely by a grant from AstraZeneca plc. The funding agreement ensured the authors’ independence in designing the study, interpreting the data, and writing and publishing the report. The following author is employed by the sponsor: Bergenheim K. Revision accepted for publication 23 November 2014. Ó The Author(s) 2014 Reprints and permission: http://www.sagepub.com/journalsPermissions.nav DOI: 10.1177/0272989X14565821
Supplementary material for this article is available on the Medical Decision Making Web site at http://mdm.sagepub.com/supplemental. Address correspondence to Phil McEwan, Health Economics & Outcomes Research Ltd., Singleton Court Business Park, Wonastow Rd, Monmouth, NP25 5JA, UK; e-mail: [email protected].
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
1
MCEWAN AND OTHERS
both patient life expectancy and quality-adjusted life expectancy. T2DM is a condition characterized by declining bcell function and deteriorating glycemic control. Current clinical practice reflects clinical guidelines that are focused on optimizing glucose control, with therapy escalation classically initiated within a ‘‘failurebased’’ regimen.6 These guidelines vary but typically advocate the maintenance of HbA1c 6.5%; this is individualized with less stringent targets (7.5%– 8.0%) for certain patient groups.7 In the United Kingdom, the National Institute of Health and Care Excellence (NICE) advocates the escalation of therapy from monotherapy when HbA1c 6.5% and from dual therapy onwards when HbA1c 7.5%.8 Importantly, the choice of therapy regimen is not just based on glucose-lowering potential; factors such as weight gain, hypoglycemia, cardioprotective benefits, and the durability of these factors over time are all important considerations. Health economic modeling in diabetes has evolved considerably, driven by the requirement to inform health care decision making in an increasingly complex environment, particularly in relation to significant patient heterogeneity and increasing therapeutic options. Health economic evaluations typically focus on assessing a new technology in comparison to current standard of care within local technology appraisal and clinical guidelines. Less common is the evaluation and comparison of treatment regimens over the lifetime of the patient. This is important as therapeutic profiles associated with the management of diabetes can have significant impact on current and downstream costs and benefits. For example, the occurrence of hypoglycemia is a potential barrier to achieving optimal glycemic control,9 and therapy-related weight gain can contribute to treatment dissatisfaction,10 excess downstream costs,11 and decreased quality of life.12,13 Failure to achieve optimal glucose control has implications for both the patient and the health care system: reduced life expectancy and quality-adjusted life expectancy for the patient and excess costs and resource utilization for the health care system. Furthermore, failing to achieve HbA1c clinical guideline targets has an important methodological implication for the evaluation of the cost-effectiveness of new health technologies. This is because future costs and benefits are evaluated within a theoretical context not representative of clinical reality. HbA1c therapy escalation thresholds play a pivotal role in many diabetes models14 by controlling duration of therapy and, consequently, incremental costs and benefits.
2
Due to the complexity of long-term diabetes models, the interrelationship between baseline HbA1c, posttreatment HbA1c, and therapy escalation on predicted cost-effectiveness is often obscured. Therefore, the principle objective of this study was to assess and illustrate the role of therapy escalation thresholds and time to therapy escalation on predicted cost-effectiveness, while using either a clinical guideline or routine clinical practice perspective. Furthermore, motivated by its recent recommendation for use in the United Kingdom, an illustrative evaluation using dapagliflozin was also conducted. METHODS This study used an established diabetes health economic model to illustrate the relationship between costs and health outcomes associated with escalation to mono, dual, and insulin-based therapies as a function of therapy escalation thresholds. We used published longitudinal, patient-level observational data to contrast current UK guidelines with observed clinical practice. Finally, we illustrated the relevance of understanding the relationship between therapy escalation thresholds, duration of therapy, and cost-effectiveness by evaluating the incremental cost-effectiveness of metformin plus sulfonylurea (M 1 S) compared with metformin plus dapagliflozin (M 1 D) in a UK setting, varying baseline HbA1c and therapy escalation levels. Each component of this study is further described below. Model The model used in this study was the Cardiff Diabetes Model.15–18 This model is a fixed time-increment stochastic simulation model designed to evaluate the cost-effectiveness of treatments or treatment sequences used in the management of T2DM. Supplemental Figure S1 shows the model’s flow diagram. Individual patients are simulated, generating cohorts of 1000 patients that progress through the model in 6monthly cycles for up to 40 years. The model fully incorporates the United Kingdom Prospective Diabetes Study 68 (UKPDS 68) event and mortality risk equations and risk factor progression equations.19 Each simulated patient is initialized with a baseline demographic and risk factor profile and progressed through the model. After every cycle, relevant demographic parameters, such as age, are incremented appropriately, and modifiable risk factor values are updated according to treatment and natural
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
progression profiles. For this analysis, initial treatment effects and those associated with therapy escalation were fully applied in the first cycle following initiation/escalation. In all subsequent cycles, risk factor trajectories were updated according to the natural history progression specified by the UKPDS 68 panel equations. The only exception to this relates to the increase in body mass index (BMI) applied for each patient where we applied a 0.040 kg/m2 per year increment.20 In this analysis, when modeling weight trajectories, we assumed that increases/ decreases in weight associated with therapy initiation/escalation were maintained while patients remained on the respective therapy. This assumption was independent of the annual 1 0.1 kg natural weight gain that was applied to all patients for the duration of the simulated time horizon. Health benefit is driven by treatment-induced changes to modifiable risk factors (e.g., HbA1c, systolic blood pressure [SBP], cholesterol, and weight), which affect the risk of occurrence of diabetesrelated vascular events and quality of life (e.g., weight loss and the avoidance of hypoglycemia). The vascular events predicted by the model include ischemic heart disease (IHD), myocardial infarction (MI), coronary heart failure (CHF), stroke, amputation, nephropathy, and blindness. Costs associated with the occurrence of these complications were primarily taken from the UKPDS 65 study21; these were indexed to 2012 values reported in Supplemental Table S1. Quality of life is an important driver of health economic benefit in diabetes, and the health utilities applied in this study were predominantly drawn from the UKPDS 62 study,22 although there were a number of exceptions. These were health utility for end-stage renal disease,23 urinary tract infections or gastrointestinal side effects,24 weight change,25 and hypoglycemia.26 The health values used are presented in Supplemental Table S2 except for hypoglycemia, in which published equations were employed to model the disutility of hypoglycemia as a function for both frequency and severity of episodes. A payer perspective was employed for the analysis, and both costs and health benefits were discounted at 3.5%. HbA1c Thresholds for Therapy Escalation Hyperglycemia is associated with an increased risk of microvascular and macrovascular complications. Consequently, clinical guidelines advocate the escalation of therapy for the management of
T2DM when HbA1c exceeds prespecified levels. In the United Kingdom, according to guidelines issued by NICE, the escalation from metformin monotherapy is advocated when HbA1c exceeds 6.5%, and from dual therapy onwards, a threshold of 7.5% is recommended. Within an economic evaluation, the threshold employed exerts considerable influence over the expected duration of therapy. Figure 1 plots an example HbA1c progression curve over time using the relevant equation from UKPDS 68,19 assuming a baseline HbA1c of 8% and hypothetical treatment-related reduction of 1.1%. This graph illustrates the relationship between HbA1c therapy escalation thresholds and the expected time on therapy; clearly, the higher the threshold, the longer the duration on a specific therapy. Figure 2 expands on this concept by demonstrating that, for a fixed-therapy escalation threshold (here 7.5%), the duration of time spent on a specific therapy is also influenced by baseline HbA1c, with lower baseline HbA1c values being associated with longer therapy duration. Figure 2 also highlights an additional challenge in modeling diabetes therapy progression when very high baseline HbA1c profiles are employed; in this case, the applied treatment effect results in an HbA1c level that remains above target. Within the programmed logic of the model employed, this will result in therapy escalation occurring in the following cycle. Observational Data The observational data used for this study have been published and described in detail.27 In brief, routine UK primary care data from The Health Improvement Network (THIN) were used to obtain the demographic and risk factor profiles of patients with T2DM initiating monotherapy, dual therapy, and insulin-based therapy between 1 January 2005 and 31 December 2009. Within the overall T2DM cohort, patients who had a first prescription for an oral antidiabetic (OAD) agent or who received a first prescription for insulin during the study period were selected. Patients were required to have at least 365 days of follow-up before and after the first prescription date, at least one reading each of SBP, weight, and HbA1c in the 365 days before the first prescription date, and at least 2 readings each of SBP, weight, and HbA1c in the 365 days after the first prescription date. Table 1 reports the demographic and risk factor profiles in relation to initiation of monotherapy, dual therapy, and insulin-based therapy used to inform the modeling analysis.
3
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
Figure 1 Illustrating the relationship between HbA1c progression, therapy escalation thresholds, and expected duration of therapy.
Figure 2 Illustrating the relationship between baseline HbA1c, progression, National Institute of Health and Care Excellence (NICE) therapy escalation threshold, and expected duration of therapy.
4
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
Table 1 Baseline Demographics and Modifiable Risk Factor Profile Applied in the Cost-Effectiveness Analysis Mean/Proportion (Nauck, 2011)28
Age (y) Sex (proportion male) Duration diabetes (y) Height (m) Proportion AC Proportion smokersa HbA1c (%) Total cholesterol (mg/dL)b HDL cholesterol (mg/dL)b SBP (mm Hg) Weight (kg)
58.4 0.551 6.32 1.67 0.062 0.176 7.72 182.54 45.87 133.3 88.02
AC, Afro-Caribbean; HDL, high-density lipoprotein; SBP, systolic blood pressure. a. Bergenheim K. 52-Week clinical study report for dapagliflozin. Study code: D1690C00004; 1 (2010). b. Total cholesterol (mmol/L): 4.75, HDL cholesterol (mmol/L): 1.19.
sulfonylurea, annual cost £51.36), and insulin therapy (metformin 1 basal insulin, annual cost £193.69). Therapy costs were obtained from the British National Formulary.31 HbA1c and Cost-Effectiveness To demonstrate how baseline HbA1c profile and HbA1c therapy escalation thresholds can affect the assessment of cost-effectiveness, we initiated the Cardiff model with the clinical profiles shown in Table 2 and evaluated the incremental cost-effectiveness of metformin plus sulfonylurea v. metformin plus dapagliflozin with insulin escalation as rescue therapy. Baseline HbA1c and HbA1c therapy escalation thresholds were varied between 6.5% and 9.0%. The efficacy and safety profiles used for this analysis are presented in Table 3. RESULTS
Scenarios Modeled
Therapy Line and Treatment Escalation
Therapy escalation and therapy line
Figure 3 shows the relationship between therapy escalation threshold (shown on the x-axis of each plot) and duration of therapy, total therapy cost, treatment disutility, and the number of major diabetesrelated complications predicted, stratified by mono, dual, and insulin therapy. These plots highlight a number of intuitive observations:
We used results from the observational analysis of THIN data (Table 1) initiating the Cardiff model with the monotherapy baseline profile and applying the observed effects of therapy initiation on modifiable risk factors (HbA1c, weight, SBP, and cholesterol). For this analysis, we sought to illustrate how therapy escalation thresholds affect the time spent on monotherapy, dual therapy, and insulin-based therapy and consequently total therapy costs, predicted event rates, and quality-adjusted life years. We allowed HbA1c therapy escalation thresholds to vary between 6.5% and 9% and used the same threshold to control escalation to dual and insulin-based therapy. Upon escalation to dual or insulin therapy, we applied the associated change in risk factor values reported in Table 1 for HbA1c, weight, SBP, and cholesterol. As rates of hypoglycemia are not adequately captured in routine data sources, we used published rates of hypoglycemia from alternative sources. We applied a nonsevere hypoglycemia episode rate of 0.02 and an annual probability of severe hypoglycemia of 0.001 from a systematic review of metformin monotherapy in type 2 diabetes.29 Hypoglycemia rates for dual therapy (metformin plus sulfonylurea) were taken from Nauck et al. and for insulin therapy from Monami et al.; these are reported in Table 2. Therapy costs applied assumed metformin monotherapy (annual cost £23.46), dual therapy (metformin 1
A lower escalation threshold is associated with a shorter time spent on monotherapy and more time spent on dual or insulin therapy. Average time on monotherapy ranged from 1.1 years (escalation at 6.5%) to 13 years (escalation at 9.0%). A lower escalation threshold is associated with an increase in total lifetime cost of therapy, due to shorter durations of inexpensive monotherapy and longer durations of more expensive rescue therapies, such as insulin. Total therapy-related expenditure decreased from £4630 (escalation at 6.5%) to £566 (escalation at 9.0%). Treatment-related disutility is minimized with higher thresholds because progression to insulin, with its associated impact on weight gain and hypoglycemia, is delayed. Total lifetime treatment disutility ranged from 0.352 (escalation at 6.5%) to 0.022 (escalation at 9.0%). Total macrovascular and microvascular complication rates are minimized with a lower escalation threshold.
Modeling therapy escalation at thresholds in line with those observed in clinical practice, compared with clinical guidelines, results in predictions that increase the amount of time spent on later stage
5
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
Table 2
Treatment Effects and Adverse Event Parameters Applied in the Cost-Effectiveness Analysis
HbA1c (%) Weight (kg) Total cholesterol (mg/dL) HDL cholesterol (mg/dL) SBP (mm Hg) Probability of discontinuationb Number of NSHEc Probability SHEc Probability of UTIc Probability of GIc Annual cost (£)
M 1 S (Nauck, 201128)
M 1 D (Nauck, 201128)
M 1 Ins (Monami, 200830)
20.52 1.44 20.028 20.002 0.8 0.059 0.408 0.00735 0.064 0.027 51.36
20.52 23.22 0.071 0.07 24.3 0.091 0.035 0 0.108 0.123 500.38
21.1 1.084 0a 0a 0a 0a 0.0108 0.037 0a 0a 193.69d
GI, genitourinary infection; HDL, high-density lipoprotein; M 1 D, metformin plus dapagliflozin; M 1 Ins, metformin plus insulin; M 1 S, metformin plus sulfonylurea; NSHE, nonsevere hypoglycemia event; SBP, systolic blood pressure; SHE, severe hypoglycemia event; UTI, urinary tract infection. a. No estimate available and/or zero value assumed. b. Probability of treatment discontinuation due to adverse events was applied during the first model cycle (first 6 months). c. Probabilities of adverse events were applied during every model cycle. d. Based on a patient baseline weight of 88kg and unit cost of £0.0053 per kg/day.
Table 3 Demographic and Risk Factor Profiles Observed in Clinical Practice, Obtained from The Health Improvement Network, Prior to Therapy Escalation and Change in Risk Factor Profile following Escalation Patient Characteristic
Mono
Number of patients 23,626 Age (y) 62.75 (12.59) Male sex (%) 56.68 Lipid-lowering therapy (%) 54.38 Blood pressure therapy (%) 65.91 Baseline risk factor values (prior to therapy escalation) HbA1c (%) 8.03 (1.24) Total cholesterol (mmol/L) 4.79 (1.16) SBP (mm Hg) 139.47 (16.95) Weight (kg) 91.32 (89.65) Change in risk factor values following therapy escalation HbA1c (%) 20.93 (0.17) Total cholesterol (mmol/L) 20.52 (0.02) SBP (mm Hg) 21.96 (0.19) Weight (kg) 21.67 (0.32)
Dual
Insulin
7230 62.91 (11.80) 60.39 80.07 74.92
4474 61.40 (12.85) 56.97 79.21 76.82
8.48 (1.28) 4.23 (0.93) 136.37 (15.67) 90.05 (19.17)
9.78 (1.94) 4.41 (1.18) 135.32 (17.39) 87.12 (19.21)
21.1 (0.02) 20.17 (0.02) 0.43 (0.32) 1.00 (0.46)
21.47 (0.04) 20.28 (0.03) 1.07 (0.39) 2.12 (0.48)
Values are presented as mean (SD) unless otherwise indicated. Mono: 89.6% Met, 9.8% SU, 0.6% TZD, and 0.1% other OAD. Dual: 55.3% Met 1 SU, 21.4% SU 1 Met, 20.8% Met 1 TZD, and 2.5% other OAD combination. Insulin: 14.4% single OAD, 53.0% dual OAD, and 30.4% triple OAD. Met, metformin; OAD, oral antidiabetic agent; SBP, systolic blood pressure; SD, standard deviation; SU, sulfonylurea; TZD, thiazolidinedione.
therapies and, in this example, lower overall treatment costs and treatment-related disutility. However, these improvements in cost and disutility associated with treatment are offset by higher rates of long-term complication incidence. Baseline HbA1c, Therapy Escalation, and CostEffectiveness Predicted cost-effectiveness, estimated as part of the illustrative economic evaluation, combines the
6
relationships between baseline HbA1c and escalation thresholds and predicted treatment and complication outcomes. Figure 4 summarizes the incremental costeffectiveness profile of metformin plus dapagliflozin v. metformin plus sulfonylurea, using the clinical profiles reported in Tables 1 and 2, as a function of baseline HbA1c and therapy escalation threshold. Evaluated at a baseline HbA1c of 7.5% and therapy escalation level of 7.5%, the predicted incremental cost-effectiveness ratio (ICER) associated with dapagliflozin is £3063. Increasing the therapy escalation
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
Figure 3 Illustrating the relationship between therapy escalation threshold and duration of therapy, event incidence, therapy cost by treatment line, and treatment-related disutility, based on The Health Improvement Network data. (Highlighted results relate to the therapy escalation threshold recommended by the National Institute of Health and Care Excellence from dual therapy onwards [blue] and those observed in clinical practice [red].)
threshold to 8.5% and 9.0%, levels closer to clinical practice observations, lead to ICERs of £8649 and £12,443, respectively. At a fixed baseline HbA1c of 6.5%, the ICER ranges from £2679 to £12,223 as a function of an increasing therapy escalation threshold; at a fixed threshold of 7.5%, the ICER decreases from £5662 (baseline HbA1c of 6.5%) to £79 (baseline HbA1c of 8.5%). Figure 4 illustrates how decreasing the therapy escalation threshold and/or increasing the baseline HbA1c decreases the overall ICER. This is due to a reduction in time spent on dual therapy, which ultimately limits both the incremental costs and benefits associated with dapagliflozin. Conversely, as baseline HbA1c and/or therapy escalation thresholds increase, patients are exposed to dapagliflozin for longer periods, thereby incurring greater therapy-related cost; it
is the favorable weight and hypoglycemia profile associated with dapagliflozin that ensures costeffectiveness is maintained across all ranges. While it is noteworthy that the cost-effectiveness of dapagliflozin is maintained across all ranges, the variability observed here in the incremental cost-effectiveness ratio is solely attributable to varying these 2 parameters (threshold and baseline HbA1c).
DISCUSSION This study has sought to emphasize an important feature of health economic modeling in T2DM: the interrelationship between baseline HbA1c and therapy escalation thresholds on predicted costs, qualityadjusted life years, and cost-effectiveness. Specifically,
7
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
Figure 4 Illustrating the relationship between baseline HbA1c and therapy escalation threshold on the predicted cost-effectiveness of dapagliflozin v. sulfonylurea when added to metformin. (Highlighted cost-effectiveness results relate to therapy escalation thresholds observed in clinical practice [red] and recommended by the National Institute of Health and Care Excellence from dual therapy onwards [blue]).
it is baseline HbA1c and choice of therapy escalation threshold that directly influence time on therapy, and it is this latter component that is invariably the primary focus of interest in most cost-effectiveness applications. The clinical scenarios typically evaluated within diabetes economic analyses use a combination of clinical trial data coupled with an idealized view of glucoselowering therapy based on advocated treatment guidelines. Importantly, patient management within routine clinical practice is often different from both the clinical trial setting and the advocated treatment escalation guidelines. Consequently, based on the discrepancies between model input values reflecting guidelines and current clinical practice, there is a significant potential for cost-effectiveness analyses in diabetes to lack face validity and offer a distorted perspective on expected value for money.
8
Despite evidence of the beneficial effects of achieving early tight glycemic control (UKPDS 80), studies from routine clinical practice suggest patients with type 2 diabetes are exposed to significant excess glycemic exposure. In a large retrospective study of more than 80,000 type 2 diabetes patients in the United Kingdom, median time to therapy intensification in those patients with suboptimal glycemic control was in excess of 7 years.32 This observation held for those treated with oral or insulin-based therapies. Mean HbA1c at therapy intensification for those taking 1, 2, or 3 oral therapies was 8.7%, 9.1%, and 9.7%, respectively, consistent with those reported in this study. Concerns are often expressed regarding the external validity of randomized clinical trials. A recent study assessing the generalizability of glycemic
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
control trials to the population of people living in Scotland with type 2 diabetes concluded that age was the most common factor that limited external validity.33 While clinical trial inclusion/exclusion criteria can lead to the selection of cohorts not representative of the broad spectrum of patients seen by clinicians in routine clinical practice, this would appear not to be so with baseline glycemic control. Meta-analyses from glucose-lowering therapy trials indicate that HbA1c levels within clinical trials are more consistent with those seen in routine clinical practice. Meta-analysis of 59 clinical trials in type 2 diabetes reported a weighted mean (SD) baseline HbA1c of 8.5% (1.3%).34 In a similar study focusing on insulin analogues in type 2 diabetes, median baseline HbA1c levels ranged from 8.35% (basal-bolus) to 9.1% (biphasic insulin).35 Our study demonstrates that there is a marked difference in treatment cost-effectiveness when comparing clinical practice with guideline-advocated therapy escalation, and we have used metformin combination with either dapagliflozin or a sulfonylurea to illustrate this concept. In this example, higher therapy escalation thresholds are associated with increased cost-effectiveness ratios, driven by a longer duration of therapy. A similar relationship between lower baseline HbA1c and cost-effectiveness was also demonstrated, highlighting the importance of sampling baseline patient profiles as part of such evaluations. This study highlights the important role that baseline HbA1c and therapy escalation thresholds have on predicted costs and health outcomes in diabetes models. Currently, sensitivity analysis in health economics typically focuses on the assessment of parameter influence, via deterministic sensitivity analysis (DSA) and the joint estimation of parameter uncertainty via probabilistic sensitivity analysis (PSA). In the simulation literature, the use of mean values instead of input probability distributions is generally discouraged. This is because the variance of input parameters as well as the mean of the input distribution jointly determines model output. Certainly, the point estimates derived from PSA will not equal those derived when using mean values because sampled cohorts’ duration of therapy will differ significantly from those attributed to an average cohort. Consequently, we would suggest that economic evaluations in diabetes that are evaluated at the mean should use extensive sensitivity analysis in relation to baseline HbA1c and therapy escalation thresholds to ensure that model predictions are able to robustly inform cost-effectiveness decisions.
In the context of a cost-effectiveness analysis, very high baseline HbA1c levels and an applied treatment effect that is insufficient to bring HbA1c levels below a given target will typically invoke an immediate therapy escalation. Within any economic evaluation, baseline HbA1c cannot be considered in isolation from treatment effect. Meta-analyses of randomized clinical trials have demonstrated a clear relationship between baseline HbA1c and the magnitude of HbA1c decrease in patients with type 2 diabetes treated with either oral or insulin-based therapy regimens.34 However, the covariance between baseline risk factors and treatment effects is rarely published in clinical studies. This represents a challenge when sampling treatment effects, and it is not uncommon to see probabilistic sensitivity analysis conducted where changes in HbA1c and other modifiable risk factors are sampled independent of baseline values. Clinical guidelines advocate therapy escalation based on perceived best clinical practice, but the observations modeled in this study reaffirm the degree of clinical inertia in clinical practice that exists across the glycemic therapy-lowering paradigm. A consequence of this is that few patients actually achieve guideline-advocated targets for glycemic control. While our data demonstrate a differential cost-effectiveness profile of glucose-lowering therapies between clinical guidelines and current practice, it is worthwhile noting there is a considerable need to optimize current treatment approaches with respect to glycemic control to address this issue of therapy inertia and excess glycemic exposure. There are a number of limitations to this study. While we have attempted to reconcile clinical guidelines and routine practice, we have not conducted a full probabilistic analysis as suggested within best practice disease modeling guidelines.36,37 In addition, we have chosen to exemplify the differential cost-effectiveness profile based on modeling routine practice thresholds as opposed to clinical guideline thresholds using dapagliflozin and sulfonylurea metformin combination as an example, but alternative therapy options, such as DPP-4 or GLP-1 metformin combinations, would also provide enlightening results as to the true economic values and cost-effectiveness profile of these agents as evaluated from the perspective of routine clinical practice. The use of dapagliflozin in this analysis was motivated by its recent recommendation for use in the United Kingdom.38 A further limitation of this analysis is that the routine clinical practice data used to punctuate this analysis largely reflect older
9
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
therapy classes, such as metformin, sulfonylurea, thiazolidinedione, and exogenous insulin. Newer technologies, such as the incretin-based agents, were not included in this data analysis. The primary motivation for focusing on the older therapy options is that in many health economies, these treatment options are perceived as offering best value for money, based on a combination of drug acquisition cost and broadly equivalent glucose-lowering potential.39,40 Finally, cost-effectiveness results presented in this analysis are dependent on the time-varying exposure to the therapeutic profiles of the agents considered, consequently resulting in greater exposure to hypoglycemia risk and body weight gain, for example, in patients treated with the sulfonylurea and metformin combination. Importantly, these results are conditional on the specific choice of costs, utilities, and other necessary modeling assumptions applied and detailed in the Methods and supplementary material. Within a conventional economic evaluation, it would be usual to vary these components, both individually and jointly, to assess their influence on cost-effectiveness. However, within this analysis, we have deliberately avoided doing this because our principal objective was to demonstrate the important role that HbA1c and timing of therapy escalation has on modeled results. Consequently, the results presented in this analysis should be interpreted with this limitation in mind. In conclusion, our analysis demonstrates an important discrepancy between guideline-derived, as well as routine clinical practice–derived, health economic output. This is important to both health care professionals and the wider health economic community with respect to understanding the true cost-effectiveness profile of any particular therapy option for people with T2DM, since the estimated value of any particular treatment option may widely differ between a guideline-based analysis compared with real-world clinic practice. An understanding of this issue is therefore essential to more completely understand the cost-effectiveness of different treatment strategies, particularly in an environment increasing economic constraints and a proliferation of new glucose-lowering technologies. REFERENCES 1. Seshasai SRK, Kaptoge S, Thompson A, et al. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med. 2011;364(9):829–41. 2. Sarwar N, Gao P, Seshasai SRK, et al. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease:
10
a collaborative meta-analysis of 102 prospective studies. Lancet. 2010;375(9733):2215–22. 3. Hex N, Bartlett C, Wright D, Taylor M, Varley D. Estimating the current and future costs of type 1 and type 2 diabetes in the UK, including direct health costs and indirect societal and productivity costs. Diabetes Med. 2012;29(7):855–62. 4. Currie CJ, Gale EAM, Poole CD. Estimation of primary care treatment costs and treatment efficacy for people with type 1 and type 2 diabetes in the United Kingdom from 1997 to 2007. Diabetes Med. 2010;27(8):938–48. 5. Morgan CL, Peters JR, Dixon S, Currie CJ. Estimated costs of acute hospital care for people with diabetes in the United Kingdom: a routine record linkage study in a large region. Diabetes Med. 2010;27(9):1066–73. 6. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes: a patient-centered approach: position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care. 2012;35(6):1364–79. 7. Paschou SA, Leslie RD. Personalizing guidelines for diabetes management: twilight or dawn of the expert? BMC Med. 2013;11: 161. 8. National Institute for Health and Care Excellence. Type 2 diabetes. CG87. London, UK: National Institute for Health and Care Excellence; 2014. 9. Choudhary P, Amiel SA. Hypoglycaemia: current management and controversies. Postgrad Med J. 2011;87(1026):298–306. 10. Marrett E, Stargardt T, Mavros P, Alexander CM. Patientreported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain. Diabetes Obes Metab. 2009;11(12):1138–44. 11. Watson L, Wilson BP, Alsop J, Kumar S. Weight and glycaemic control in type 2 diabetes: what is the outcome of insulin initiation? Diabetes Obes Metab. 2011;13(9):823–31. 12. Grandy S, Fox KM, Bazata DD; SHIELD Study Group. Healthrelated quality of life association with weight change in type 2 diabetes mellitus: perception vs. reality. Int J Clin Pract. 2013;67(5): 455–61. 13. Bagust A, Beale S. Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data. Health Econ. 2005;14(3):217–30. 14. Palmer AJ; Mount Hood 5 Modeling Group. Computer modeling of diabetes and its complications: a report on the Fifth Mount Hood Challenge Meeting. Value Health. 2013;16(4):670–685. 15. McEwan P, Peters JR, Bergenheim K, Currie CJ. Evaluation of the costs and outcomes from changes in risk factors in type 2 diabetes using the Cardiff stochastic simulation cost-utility model (DiabForecaster). Curr Med Res Opin. 2006;22(1):121–9. 16. McEwan P, Bergenheim K, Yuan Y, Tetlow AP, Gordon JP. Assessing the relationship between computational speed and precision: a case study comparing an interpreted versus compiled programming language using a stochastic simulation model in diabetes care. Pharmacoeconomics. 2010;28(8):665–74. 17. McEwan P, Evans M, Kan H, Bergenheim K. Understanding the inter-relationship between improved glycaemic control, hypoglycaemia and weight change within a long-term economic model. Diabetes Obes Metab. 2010;12(5):431–6.
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
18. McEwan P, Evans M, Bergenheim K. A population model evaluating the costs and benefits associated with different oral treatment strategies in people with type 2 diabetes. Diabetes Obes Metab. 2010;12(7):623–30. 19. Clarke PM, Gray AM, Briggs A, et al. A model to estimate the lifetime health outcomes of patients with type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68). Diabetologia. 2004;47(10):1747–59.
29. Saenz A, Fernandez-Esteban I, Mataix A, et al. Metformin monotherapy for type 2 diabetes mellitus. Cochrane Database Syst Rev. 2005;20(3):CD002966. 30. Monami M, Marchionni N, Mannucci E. Long-acting insulin analogues versus NPH human insulin in type 2 diabetes: a metaanalysis. Diabetes Res Clin Pract. 2008;81(2):184-9. 31. Joint Formulary Committee. British National Formulary. 65th ed. London: BMJ Group and Pharmaceutical Press; 2013.
20. Ara R, Blake L, Gray L, et al. What is the clinical effectiveness and cost-effectiveness of using drugs in treating obese patients in primary care? A systematic review. Health Technol Assess. 2012; 16(5):iii–xiv, 1–195.
32. Khunti K, Wolden ML, Thorsted BL, Andersen M, Davies MJ. Clinical inertia in people with type 2 diabetes: a retrospective cohort study of more than 80,000 people. Diabetes Care. 2013; 36(11):3411–7.
21. Clarke P, Gray A, Legood R, Briggs A, Holman R. The impact of diabetes-related complications on healthcare costs: results from the United Kingdom Prospective Diabetes Study (UKPDS Study No. 65). Diabetic Med. 2003;20:442–50.
33. Saunders C, Byrne CD, Guthrie B, et al; Scottish Diabetes Research Network Epidemiology Group. External validity of randomized controlled trials of glycaemic control and vascular disease: how representative are participants? Diabetes Med. 2013; 30(3):300–8.
22. Clarke P, Gray A, Legood R, Holman R. Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62). Med Decis Making. 2002;2(4):340–349. 23. Currie CJ, McEwan P, Peters JR, Patel TC, Dixon S. The routine collation of health outcomes data from hospital treated subjects in the Health Outcomes Data Repository (HODaR): descriptive analysis from the first 20,000 subjects. Value Health. 2005;8:581–90. 24. Barry H, Ebell M, Hickner J. Evaluation of suspected urinary tract infection in ambulatory women: a cost-utility analysis of office-based strategies. J Family Pract. 1997;44(1):49–60. 25. Lane S, Levy AR, Mukherjee J, Sambrook J, Tildesley H. The impact on utilities of differences in body weight among Canadian patients with type 2 diabetes. Curr Med Res Opin. 2014;30(7): 1267–73. 26. Currie CJ, Morgan CL, Poole CD, Sharplin P, Lammert M, McEwan P. Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes. Curr Med Res Opin. 2006; 22(8):15–23. 27. Bennett H, McEwan P, Bergenheim K, Gordon J. Assessment of unmet clinical need in type 2 diabetic patients on conventional therapy in the UK [published online 4 September 2014]. Diabetes Ther. 28. Nauck MA, Del Prato S, Meier JJ, Dura´n-Garcı´a S, Rohwedder K, Elze M, et al. Dapagliflozin versus glipizide as add-on therapy in patients with type 2 diabetes who have inadequate glycemic control with metformin: a randomized, 52-week, double-blind, active-controlled noninferiority trial. Diabetes Care. 2011;34(9): 2015-22.
34. DeFronzo RA, Stonehouse AH, Han J, Wintle ME. Relationship of baseline HbA1c and efficacy of current glucose-lowering therapies: a meta-analysis of randomized clinical trials. Diabetes Med. 2010;27(3):309–17. 35. Giugliano D, Maiorino M, Bellastella G, Chiodini P, Esposito K. Relationship of baseline HbA1c, HbA1c change and HbA1c target of \ 7% with insulin analogues in type 2 diabetes: a meta-analysis of randomised controlled trials. Int J Clin Pract. 2011;65(5):602–12. 36. National Institute of Health and Care Excellence. Guide to the methods of technology appraisal. 2013. Available from: http://publications.nice.org.uk/pmg9 37. Briggs AH, Weinstein MC, Fenwick EA, Karnon J, Sculpher MJ, Paltiel AD; ISPOR-SMDM Modeling Good Research Practices Task Force. Model parameter estimation and uncertainty analysis: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force Working Group-6. Med Decis Making. 2012;32(5):722–32. 38. NICE technology appraisal guidance 288. Available from: www.guidance.nice.org.uk/ta288 39. McIntosh B, Cameron C, Singh SR, et al. Second-line therapy in patients with type 2 diabetes inadequately controlled with metformin monotherapy: a systematic review and mixed-treatment comparison meta-analysis. Open Med. 2011;5(1):e35-e48. 40. McIntosh B, Cameron C, Singh SR, Yu C, Dolovich L, Houlden R. Choice of therapy in patients with type 2 diabetes inadequately controlled with metformin and a sulphonylurea: a systematic review and mixed-treatment comparison meta-analysis. Open Med. 2012;6(2):e62–74.
11
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
Estimating Cost-Effectiveness in Type 2 Diabetes: The Impact of Treatment Guidelines and Therapy Duration Phil McEwan, PhD, Jason Gordon, PhD, Marc Evans, MD, Thomas Ward, MSc, Hayley Bennett, MSc, Klas Bergenheim, PhD
Objectives. Type 2 diabetes mellitus (T2DM) clinical guidelines focus on optimizing glucose control, with therapy escalation classically initiated within a ‘‘failurebased’’ regimen. Within many diabetes models, HbA1c therapy escalation thresholds play a pivotal role, controlling duration of therapy and, consequently, incremental costs and benefits. The objective of this study was to assess the relationship between therapy escalation threshold and time to therapy escalation on predicted cost-effectiveness of T2DM treatments. Methods. This study used the Cardiff Diabetes Model to illustrate the relationship between costs and health outcomes associated with first-, second-, and third-line therapy as a function of time on each. Data from routine clinical practice were used to contrast predicted costs and health outcomes associated with guideline therapy escalation thresholds compared with clinical practice. The impact of baseline HbA1c and therapy escalation thresholds on cost-effectiveness was investigated, comparing a sodium/ glucose cotransporter 2 inhibitor v. sulfonylurea added to
metformin monotherapy. Results. Lower thresholds are associated with a shorter time spent on monotherapy, ranging from 1.1 years (escalation at 6.5%) to 13 years (escalation at 9.0%) and an increase in total lifetime cost of therapy. Treatment-related disutility is minimized with higher thresholds because progression to insulin is delayed. Using metformin combined with either dapagliflozin or a sulfonylurea to illustrate lower baseline HbA1c and/or higher therapy escalation thresholds was associated with increased cost-effectiveness ratios, driven by a longer duration of therapy. Discussion. A marked difference in treatment cost-effectiveness was demonstrated when comparing routine clinical practice with guideline-advocated therapy escalation. This is important to both health care professionals and the wider health economic community with respect to understanding the true cost-effectiveness profile of any particular T2DM therapy option. Key words: decision aids; internal medicine; provider decision making; cost-effectiveness analysis. (Med Decis Making 20XX;XX: XXX–XXX)
T
morbidity and mortality.1,2 The current and expected future health care expenditure associated with managing T2DM is substantial. In the United Kingdom, annual direct health care costs for T2DM are estimated at £8.8 billion,3 and recent studies have reported on the increasing trends in diabetes-specific expenditure associated with treatment costs and acute hospital care.4,5 Of primary concern to health care policy makers is the optimal management of patients with T2DM to minimize health care expenditure and maximize
ype 2 diabetes mellitus (T2DM) is a chronic condition associated with substantial excess
Received 13 January 2014 from the Swansea Centre for Health Economics, Swansea University, Wales, UK (PCM, HB); Health Economics & Outcomes Research Ltd., Wales, UK (PCM, JG, TW); University Health Board, Llandough, Wales, UK (ME); Global Health Economics & Outcomes Research, AstraZeneca, Molndal, Sweden (KB); and Department of Public Health, University of Adelaide, South Australia, Australia (JG). Financial support for this study was provided entirely by a grant from AstraZeneca plc. The funding agreement ensured the authors’ independence in designing the study, interpreting the data, and writing and publishing the report. The following author is employed by the sponsor: Bergenheim K. Revision accepted for publication 23 November 2014. Ó The Author(s) 2014 Reprints and permission: http://www.sagepub.com/journalsPermissions.nav DOI: 10.1177/0272989X14565821
Supplementary material for this article is available on the Medical Decision Making Web site at http://mdm.sagepub.com/supplemental. Address correspondence to Phil McEwan, Health Economics & Outcomes Research Ltd., Singleton Court Business Park, Wonastow Rd, Monmouth, NP25 5JA, UK; e-mail: [email protected].
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
1
MCEWAN AND OTHERS
both patient life expectancy and quality-adjusted life expectancy. T2DM is a condition characterized by declining bcell function and deteriorating glycemic control. Current clinical practice reflects clinical guidelines that are focused on optimizing glucose control, with therapy escalation classically initiated within a ‘‘failurebased’’ regimen.6 These guidelines vary but typically advocate the maintenance of HbA1c 6.5%; this is individualized with less stringent targets (7.5%– 8.0%) for certain patient groups.7 In the United Kingdom, the National Institute of Health and Care Excellence (NICE) advocates the escalation of therapy from monotherapy when HbA1c 6.5% and from dual therapy onwards when HbA1c 7.5%.8 Importantly, the choice of therapy regimen is not just based on glucose-lowering potential; factors such as weight gain, hypoglycemia, cardioprotective benefits, and the durability of these factors over time are all important considerations. Health economic modeling in diabetes has evolved considerably, driven by the requirement to inform health care decision making in an increasingly complex environment, particularly in relation to significant patient heterogeneity and increasing therapeutic options. Health economic evaluations typically focus on assessing a new technology in comparison to current standard of care within local technology appraisal and clinical guidelines. Less common is the evaluation and comparison of treatment regimens over the lifetime of the patient. This is important as therapeutic profiles associated with the management of diabetes can have significant impact on current and downstream costs and benefits. For example, the occurrence of hypoglycemia is a potential barrier to achieving optimal glycemic control,9 and therapy-related weight gain can contribute to treatment dissatisfaction,10 excess downstream costs,11 and decreased quality of life.12,13 Failure to achieve optimal glucose control has implications for both the patient and the health care system: reduced life expectancy and quality-adjusted life expectancy for the patient and excess costs and resource utilization for the health care system. Furthermore, failing to achieve HbA1c clinical guideline targets has an important methodological implication for the evaluation of the cost-effectiveness of new health technologies. This is because future costs and benefits are evaluated within a theoretical context not representative of clinical reality. HbA1c therapy escalation thresholds play a pivotal role in many diabetes models14 by controlling duration of therapy and, consequently, incremental costs and benefits.
2
Due to the complexity of long-term diabetes models, the interrelationship between baseline HbA1c, posttreatment HbA1c, and therapy escalation on predicted cost-effectiveness is often obscured. Therefore, the principle objective of this study was to assess and illustrate the role of therapy escalation thresholds and time to therapy escalation on predicted cost-effectiveness, while using either a clinical guideline or routine clinical practice perspective. Furthermore, motivated by its recent recommendation for use in the United Kingdom, an illustrative evaluation using dapagliflozin was also conducted. METHODS This study used an established diabetes health economic model to illustrate the relationship between costs and health outcomes associated with escalation to mono, dual, and insulin-based therapies as a function of therapy escalation thresholds. We used published longitudinal, patient-level observational data to contrast current UK guidelines with observed clinical practice. Finally, we illustrated the relevance of understanding the relationship between therapy escalation thresholds, duration of therapy, and cost-effectiveness by evaluating the incremental cost-effectiveness of metformin plus sulfonylurea (M 1 S) compared with metformin plus dapagliflozin (M 1 D) in a UK setting, varying baseline HbA1c and therapy escalation levels. Each component of this study is further described below. Model The model used in this study was the Cardiff Diabetes Model.15–18 This model is a fixed time-increment stochastic simulation model designed to evaluate the cost-effectiveness of treatments or treatment sequences used in the management of T2DM. Supplemental Figure S1 shows the model’s flow diagram. Individual patients are simulated, generating cohorts of 1000 patients that progress through the model in 6monthly cycles for up to 40 years. The model fully incorporates the United Kingdom Prospective Diabetes Study 68 (UKPDS 68) event and mortality risk equations and risk factor progression equations.19 Each simulated patient is initialized with a baseline demographic and risk factor profile and progressed through the model. After every cycle, relevant demographic parameters, such as age, are incremented appropriately, and modifiable risk factor values are updated according to treatment and natural
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
progression profiles. For this analysis, initial treatment effects and those associated with therapy escalation were fully applied in the first cycle following initiation/escalation. In all subsequent cycles, risk factor trajectories were updated according to the natural history progression specified by the UKPDS 68 panel equations. The only exception to this relates to the increase in body mass index (BMI) applied for each patient where we applied a 0.040 kg/m2 per year increment.20 In this analysis, when modeling weight trajectories, we assumed that increases/ decreases in weight associated with therapy initiation/escalation were maintained while patients remained on the respective therapy. This assumption was independent of the annual 1 0.1 kg natural weight gain that was applied to all patients for the duration of the simulated time horizon. Health benefit is driven by treatment-induced changes to modifiable risk factors (e.g., HbA1c, systolic blood pressure [SBP], cholesterol, and weight), which affect the risk of occurrence of diabetesrelated vascular events and quality of life (e.g., weight loss and the avoidance of hypoglycemia). The vascular events predicted by the model include ischemic heart disease (IHD), myocardial infarction (MI), coronary heart failure (CHF), stroke, amputation, nephropathy, and blindness. Costs associated with the occurrence of these complications were primarily taken from the UKPDS 65 study21; these were indexed to 2012 values reported in Supplemental Table S1. Quality of life is an important driver of health economic benefit in diabetes, and the health utilities applied in this study were predominantly drawn from the UKPDS 62 study,22 although there were a number of exceptions. These were health utility for end-stage renal disease,23 urinary tract infections or gastrointestinal side effects,24 weight change,25 and hypoglycemia.26 The health values used are presented in Supplemental Table S2 except for hypoglycemia, in which published equations were employed to model the disutility of hypoglycemia as a function for both frequency and severity of episodes. A payer perspective was employed for the analysis, and both costs and health benefits were discounted at 3.5%. HbA1c Thresholds for Therapy Escalation Hyperglycemia is associated with an increased risk of microvascular and macrovascular complications. Consequently, clinical guidelines advocate the escalation of therapy for the management of
T2DM when HbA1c exceeds prespecified levels. In the United Kingdom, according to guidelines issued by NICE, the escalation from metformin monotherapy is advocated when HbA1c exceeds 6.5%, and from dual therapy onwards, a threshold of 7.5% is recommended. Within an economic evaluation, the threshold employed exerts considerable influence over the expected duration of therapy. Figure 1 plots an example HbA1c progression curve over time using the relevant equation from UKPDS 68,19 assuming a baseline HbA1c of 8% and hypothetical treatment-related reduction of 1.1%. This graph illustrates the relationship between HbA1c therapy escalation thresholds and the expected time on therapy; clearly, the higher the threshold, the longer the duration on a specific therapy. Figure 2 expands on this concept by demonstrating that, for a fixed-therapy escalation threshold (here 7.5%), the duration of time spent on a specific therapy is also influenced by baseline HbA1c, with lower baseline HbA1c values being associated with longer therapy duration. Figure 2 also highlights an additional challenge in modeling diabetes therapy progression when very high baseline HbA1c profiles are employed; in this case, the applied treatment effect results in an HbA1c level that remains above target. Within the programmed logic of the model employed, this will result in therapy escalation occurring in the following cycle. Observational Data The observational data used for this study have been published and described in detail.27 In brief, routine UK primary care data from The Health Improvement Network (THIN) were used to obtain the demographic and risk factor profiles of patients with T2DM initiating monotherapy, dual therapy, and insulin-based therapy between 1 January 2005 and 31 December 2009. Within the overall T2DM cohort, patients who had a first prescription for an oral antidiabetic (OAD) agent or who received a first prescription for insulin during the study period were selected. Patients were required to have at least 365 days of follow-up before and after the first prescription date, at least one reading each of SBP, weight, and HbA1c in the 365 days before the first prescription date, and at least 2 readings each of SBP, weight, and HbA1c in the 365 days after the first prescription date. Table 1 reports the demographic and risk factor profiles in relation to initiation of monotherapy, dual therapy, and insulin-based therapy used to inform the modeling analysis.
3
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
Figure 1 Illustrating the relationship between HbA1c progression, therapy escalation thresholds, and expected duration of therapy.
Figure 2 Illustrating the relationship between baseline HbA1c, progression, National Institute of Health and Care Excellence (NICE) therapy escalation threshold, and expected duration of therapy.
4
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
Table 1 Baseline Demographics and Modifiable Risk Factor Profile Applied in the Cost-Effectiveness Analysis Mean/Proportion (Nauck, 2011)28
Age (y) Sex (proportion male) Duration diabetes (y) Height (m) Proportion AC Proportion smokersa HbA1c (%) Total cholesterol (mg/dL)b HDL cholesterol (mg/dL)b SBP (mm Hg) Weight (kg)
58.4 0.551 6.32 1.67 0.062 0.176 7.72 182.54 45.87 133.3 88.02
AC, Afro-Caribbean; HDL, high-density lipoprotein; SBP, systolic blood pressure. a. Bergenheim K. 52-Week clinical study report for dapagliflozin. Study code: D1690C00004; 1 (2010). b. Total cholesterol (mmol/L): 4.75, HDL cholesterol (mmol/L): 1.19.
sulfonylurea, annual cost £51.36), and insulin therapy (metformin 1 basal insulin, annual cost £193.69). Therapy costs were obtained from the British National Formulary.31 HbA1c and Cost-Effectiveness To demonstrate how baseline HbA1c profile and HbA1c therapy escalation thresholds can affect the assessment of cost-effectiveness, we initiated the Cardiff model with the clinical profiles shown in Table 2 and evaluated the incremental cost-effectiveness of metformin plus sulfonylurea v. metformin plus dapagliflozin with insulin escalation as rescue therapy. Baseline HbA1c and HbA1c therapy escalation thresholds were varied between 6.5% and 9.0%. The efficacy and safety profiles used for this analysis are presented in Table 3. RESULTS
Scenarios Modeled
Therapy Line and Treatment Escalation
Therapy escalation and therapy line
Figure 3 shows the relationship between therapy escalation threshold (shown on the x-axis of each plot) and duration of therapy, total therapy cost, treatment disutility, and the number of major diabetesrelated complications predicted, stratified by mono, dual, and insulin therapy. These plots highlight a number of intuitive observations:
We used results from the observational analysis of THIN data (Table 1) initiating the Cardiff model with the monotherapy baseline profile and applying the observed effects of therapy initiation on modifiable risk factors (HbA1c, weight, SBP, and cholesterol). For this analysis, we sought to illustrate how therapy escalation thresholds affect the time spent on monotherapy, dual therapy, and insulin-based therapy and consequently total therapy costs, predicted event rates, and quality-adjusted life years. We allowed HbA1c therapy escalation thresholds to vary between 6.5% and 9% and used the same threshold to control escalation to dual and insulin-based therapy. Upon escalation to dual or insulin therapy, we applied the associated change in risk factor values reported in Table 1 for HbA1c, weight, SBP, and cholesterol. As rates of hypoglycemia are not adequately captured in routine data sources, we used published rates of hypoglycemia from alternative sources. We applied a nonsevere hypoglycemia episode rate of 0.02 and an annual probability of severe hypoglycemia of 0.001 from a systematic review of metformin monotherapy in type 2 diabetes.29 Hypoglycemia rates for dual therapy (metformin plus sulfonylurea) were taken from Nauck et al. and for insulin therapy from Monami et al.; these are reported in Table 2. Therapy costs applied assumed metformin monotherapy (annual cost £23.46), dual therapy (metformin 1
A lower escalation threshold is associated with a shorter time spent on monotherapy and more time spent on dual or insulin therapy. Average time on monotherapy ranged from 1.1 years (escalation at 6.5%) to 13 years (escalation at 9.0%). A lower escalation threshold is associated with an increase in total lifetime cost of therapy, due to shorter durations of inexpensive monotherapy and longer durations of more expensive rescue therapies, such as insulin. Total therapy-related expenditure decreased from £4630 (escalation at 6.5%) to £566 (escalation at 9.0%). Treatment-related disutility is minimized with higher thresholds because progression to insulin, with its associated impact on weight gain and hypoglycemia, is delayed. Total lifetime treatment disutility ranged from 0.352 (escalation at 6.5%) to 0.022 (escalation at 9.0%). Total macrovascular and microvascular complication rates are minimized with a lower escalation threshold.
Modeling therapy escalation at thresholds in line with those observed in clinical practice, compared with clinical guidelines, results in predictions that increase the amount of time spent on later stage
5
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
Table 2
Treatment Effects and Adverse Event Parameters Applied in the Cost-Effectiveness Analysis
HbA1c (%) Weight (kg) Total cholesterol (mg/dL) HDL cholesterol (mg/dL) SBP (mm Hg) Probability of discontinuationb Number of NSHEc Probability SHEc Probability of UTIc Probability of GIc Annual cost (£)
M 1 S (Nauck, 201128)
M 1 D (Nauck, 201128)
M 1 Ins (Monami, 200830)
20.52 1.44 20.028 20.002 0.8 0.059 0.408 0.00735 0.064 0.027 51.36
20.52 23.22 0.071 0.07 24.3 0.091 0.035 0 0.108 0.123 500.38
21.1 1.084 0a 0a 0a 0a 0.0108 0.037 0a 0a 193.69d
GI, genitourinary infection; HDL, high-density lipoprotein; M 1 D, metformin plus dapagliflozin; M 1 Ins, metformin plus insulin; M 1 S, metformin plus sulfonylurea; NSHE, nonsevere hypoglycemia event; SBP, systolic blood pressure; SHE, severe hypoglycemia event; UTI, urinary tract infection. a. No estimate available and/or zero value assumed. b. Probability of treatment discontinuation due to adverse events was applied during the first model cycle (first 6 months). c. Probabilities of adverse events were applied during every model cycle. d. Based on a patient baseline weight of 88kg and unit cost of £0.0053 per kg/day.
Table 3 Demographic and Risk Factor Profiles Observed in Clinical Practice, Obtained from The Health Improvement Network, Prior to Therapy Escalation and Change in Risk Factor Profile following Escalation Patient Characteristic
Mono
Number of patients 23,626 Age (y) 62.75 (12.59) Male sex (%) 56.68 Lipid-lowering therapy (%) 54.38 Blood pressure therapy (%) 65.91 Baseline risk factor values (prior to therapy escalation) HbA1c (%) 8.03 (1.24) Total cholesterol (mmol/L) 4.79 (1.16) SBP (mm Hg) 139.47 (16.95) Weight (kg) 91.32 (89.65) Change in risk factor values following therapy escalation HbA1c (%) 20.93 (0.17) Total cholesterol (mmol/L) 20.52 (0.02) SBP (mm Hg) 21.96 (0.19) Weight (kg) 21.67 (0.32)
Dual
Insulin
7230 62.91 (11.80) 60.39 80.07 74.92
4474 61.40 (12.85) 56.97 79.21 76.82
8.48 (1.28) 4.23 (0.93) 136.37 (15.67) 90.05 (19.17)
9.78 (1.94) 4.41 (1.18) 135.32 (17.39) 87.12 (19.21)
21.1 (0.02) 20.17 (0.02) 0.43 (0.32) 1.00 (0.46)
21.47 (0.04) 20.28 (0.03) 1.07 (0.39) 2.12 (0.48)
Values are presented as mean (SD) unless otherwise indicated. Mono: 89.6% Met, 9.8% SU, 0.6% TZD, and 0.1% other OAD. Dual: 55.3% Met 1 SU, 21.4% SU 1 Met, 20.8% Met 1 TZD, and 2.5% other OAD combination. Insulin: 14.4% single OAD, 53.0% dual OAD, and 30.4% triple OAD. Met, metformin; OAD, oral antidiabetic agent; SBP, systolic blood pressure; SD, standard deviation; SU, sulfonylurea; TZD, thiazolidinedione.
therapies and, in this example, lower overall treatment costs and treatment-related disutility. However, these improvements in cost and disutility associated with treatment are offset by higher rates of long-term complication incidence. Baseline HbA1c, Therapy Escalation, and CostEffectiveness Predicted cost-effectiveness, estimated as part of the illustrative economic evaluation, combines the
6
relationships between baseline HbA1c and escalation thresholds and predicted treatment and complication outcomes. Figure 4 summarizes the incremental costeffectiveness profile of metformin plus dapagliflozin v. metformin plus sulfonylurea, using the clinical profiles reported in Tables 1 and 2, as a function of baseline HbA1c and therapy escalation threshold. Evaluated at a baseline HbA1c of 7.5% and therapy escalation level of 7.5%, the predicted incremental cost-effectiveness ratio (ICER) associated with dapagliflozin is £3063. Increasing the therapy escalation
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
Figure 3 Illustrating the relationship between therapy escalation threshold and duration of therapy, event incidence, therapy cost by treatment line, and treatment-related disutility, based on The Health Improvement Network data. (Highlighted results relate to the therapy escalation threshold recommended by the National Institute of Health and Care Excellence from dual therapy onwards [blue] and those observed in clinical practice [red].)
threshold to 8.5% and 9.0%, levels closer to clinical practice observations, lead to ICERs of £8649 and £12,443, respectively. At a fixed baseline HbA1c of 6.5%, the ICER ranges from £2679 to £12,223 as a function of an increasing therapy escalation threshold; at a fixed threshold of 7.5%, the ICER decreases from £5662 (baseline HbA1c of 6.5%) to £79 (baseline HbA1c of 8.5%). Figure 4 illustrates how decreasing the therapy escalation threshold and/or increasing the baseline HbA1c decreases the overall ICER. This is due to a reduction in time spent on dual therapy, which ultimately limits both the incremental costs and benefits associated with dapagliflozin. Conversely, as baseline HbA1c and/or therapy escalation thresholds increase, patients are exposed to dapagliflozin for longer periods, thereby incurring greater therapy-related cost; it
is the favorable weight and hypoglycemia profile associated with dapagliflozin that ensures costeffectiveness is maintained across all ranges. While it is noteworthy that the cost-effectiveness of dapagliflozin is maintained across all ranges, the variability observed here in the incremental cost-effectiveness ratio is solely attributable to varying these 2 parameters (threshold and baseline HbA1c).
DISCUSSION This study has sought to emphasize an important feature of health economic modeling in T2DM: the interrelationship between baseline HbA1c and therapy escalation thresholds on predicted costs, qualityadjusted life years, and cost-effectiveness. Specifically,
7
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
Figure 4 Illustrating the relationship between baseline HbA1c and therapy escalation threshold on the predicted cost-effectiveness of dapagliflozin v. sulfonylurea when added to metformin. (Highlighted cost-effectiveness results relate to therapy escalation thresholds observed in clinical practice [red] and recommended by the National Institute of Health and Care Excellence from dual therapy onwards [blue]).
it is baseline HbA1c and choice of therapy escalation threshold that directly influence time on therapy, and it is this latter component that is invariably the primary focus of interest in most cost-effectiveness applications. The clinical scenarios typically evaluated within diabetes economic analyses use a combination of clinical trial data coupled with an idealized view of glucoselowering therapy based on advocated treatment guidelines. Importantly, patient management within routine clinical practice is often different from both the clinical trial setting and the advocated treatment escalation guidelines. Consequently, based on the discrepancies between model input values reflecting guidelines and current clinical practice, there is a significant potential for cost-effectiveness analyses in diabetes to lack face validity and offer a distorted perspective on expected value for money.
8
Despite evidence of the beneficial effects of achieving early tight glycemic control (UKPDS 80), studies from routine clinical practice suggest patients with type 2 diabetes are exposed to significant excess glycemic exposure. In a large retrospective study of more than 80,000 type 2 diabetes patients in the United Kingdom, median time to therapy intensification in those patients with suboptimal glycemic control was in excess of 7 years.32 This observation held for those treated with oral or insulin-based therapies. Mean HbA1c at therapy intensification for those taking 1, 2, or 3 oral therapies was 8.7%, 9.1%, and 9.7%, respectively, consistent with those reported in this study. Concerns are often expressed regarding the external validity of randomized clinical trials. A recent study assessing the generalizability of glycemic
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
control trials to the population of people living in Scotland with type 2 diabetes concluded that age was the most common factor that limited external validity.33 While clinical trial inclusion/exclusion criteria can lead to the selection of cohorts not representative of the broad spectrum of patients seen by clinicians in routine clinical practice, this would appear not to be so with baseline glycemic control. Meta-analyses from glucose-lowering therapy trials indicate that HbA1c levels within clinical trials are more consistent with those seen in routine clinical practice. Meta-analysis of 59 clinical trials in type 2 diabetes reported a weighted mean (SD) baseline HbA1c of 8.5% (1.3%).34 In a similar study focusing on insulin analogues in type 2 diabetes, median baseline HbA1c levels ranged from 8.35% (basal-bolus) to 9.1% (biphasic insulin).35 Our study demonstrates that there is a marked difference in treatment cost-effectiveness when comparing clinical practice with guideline-advocated therapy escalation, and we have used metformin combination with either dapagliflozin or a sulfonylurea to illustrate this concept. In this example, higher therapy escalation thresholds are associated with increased cost-effectiveness ratios, driven by a longer duration of therapy. A similar relationship between lower baseline HbA1c and cost-effectiveness was also demonstrated, highlighting the importance of sampling baseline patient profiles as part of such evaluations. This study highlights the important role that baseline HbA1c and therapy escalation thresholds have on predicted costs and health outcomes in diabetes models. Currently, sensitivity analysis in health economics typically focuses on the assessment of parameter influence, via deterministic sensitivity analysis (DSA) and the joint estimation of parameter uncertainty via probabilistic sensitivity analysis (PSA). In the simulation literature, the use of mean values instead of input probability distributions is generally discouraged. This is because the variance of input parameters as well as the mean of the input distribution jointly determines model output. Certainly, the point estimates derived from PSA will not equal those derived when using mean values because sampled cohorts’ duration of therapy will differ significantly from those attributed to an average cohort. Consequently, we would suggest that economic evaluations in diabetes that are evaluated at the mean should use extensive sensitivity analysis in relation to baseline HbA1c and therapy escalation thresholds to ensure that model predictions are able to robustly inform cost-effectiveness decisions.
In the context of a cost-effectiveness analysis, very high baseline HbA1c levels and an applied treatment effect that is insufficient to bring HbA1c levels below a given target will typically invoke an immediate therapy escalation. Within any economic evaluation, baseline HbA1c cannot be considered in isolation from treatment effect. Meta-analyses of randomized clinical trials have demonstrated a clear relationship between baseline HbA1c and the magnitude of HbA1c decrease in patients with type 2 diabetes treated with either oral or insulin-based therapy regimens.34 However, the covariance between baseline risk factors and treatment effects is rarely published in clinical studies. This represents a challenge when sampling treatment effects, and it is not uncommon to see probabilistic sensitivity analysis conducted where changes in HbA1c and other modifiable risk factors are sampled independent of baseline values. Clinical guidelines advocate therapy escalation based on perceived best clinical practice, but the observations modeled in this study reaffirm the degree of clinical inertia in clinical practice that exists across the glycemic therapy-lowering paradigm. A consequence of this is that few patients actually achieve guideline-advocated targets for glycemic control. While our data demonstrate a differential cost-effectiveness profile of glucose-lowering therapies between clinical guidelines and current practice, it is worthwhile noting there is a considerable need to optimize current treatment approaches with respect to glycemic control to address this issue of therapy inertia and excess glycemic exposure. There are a number of limitations to this study. While we have attempted to reconcile clinical guidelines and routine practice, we have not conducted a full probabilistic analysis as suggested within best practice disease modeling guidelines.36,37 In addition, we have chosen to exemplify the differential cost-effectiveness profile based on modeling routine practice thresholds as opposed to clinical guideline thresholds using dapagliflozin and sulfonylurea metformin combination as an example, but alternative therapy options, such as DPP-4 or GLP-1 metformin combinations, would also provide enlightening results as to the true economic values and cost-effectiveness profile of these agents as evaluated from the perspective of routine clinical practice. The use of dapagliflozin in this analysis was motivated by its recent recommendation for use in the United Kingdom.38 A further limitation of this analysis is that the routine clinical practice data used to punctuate this analysis largely reflect older
9
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
MCEWAN AND OTHERS
therapy classes, such as metformin, sulfonylurea, thiazolidinedione, and exogenous insulin. Newer technologies, such as the incretin-based agents, were not included in this data analysis. The primary motivation for focusing on the older therapy options is that in many health economies, these treatment options are perceived as offering best value for money, based on a combination of drug acquisition cost and broadly equivalent glucose-lowering potential.39,40 Finally, cost-effectiveness results presented in this analysis are dependent on the time-varying exposure to the therapeutic profiles of the agents considered, consequently resulting in greater exposure to hypoglycemia risk and body weight gain, for example, in patients treated with the sulfonylurea and metformin combination. Importantly, these results are conditional on the specific choice of costs, utilities, and other necessary modeling assumptions applied and detailed in the Methods and supplementary material. Within a conventional economic evaluation, it would be usual to vary these components, both individually and jointly, to assess their influence on cost-effectiveness. However, within this analysis, we have deliberately avoided doing this because our principal objective was to demonstrate the important role that HbA1c and timing of therapy escalation has on modeled results. Consequently, the results presented in this analysis should be interpreted with this limitation in mind. In conclusion, our analysis demonstrates an important discrepancy between guideline-derived, as well as routine clinical practice–derived, health economic output. This is important to both health care professionals and the wider health economic community with respect to understanding the true cost-effectiveness profile of any particular therapy option for people with T2DM, since the estimated value of any particular treatment option may widely differ between a guideline-based analysis compared with real-world clinic practice. An understanding of this issue is therefore essential to more completely understand the cost-effectiveness of different treatment strategies, particularly in an environment increasing economic constraints and a proliferation of new glucose-lowering technologies. REFERENCES 1. Seshasai SRK, Kaptoge S, Thompson A, et al. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med. 2011;364(9):829–41. 2. Sarwar N, Gao P, Seshasai SRK, et al. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease:
10
a collaborative meta-analysis of 102 prospective studies. Lancet. 2010;375(9733):2215–22. 3. Hex N, Bartlett C, Wright D, Taylor M, Varley D. Estimating the current and future costs of type 1 and type 2 diabetes in the UK, including direct health costs and indirect societal and productivity costs. Diabetes Med. 2012;29(7):855–62. 4. Currie CJ, Gale EAM, Poole CD. Estimation of primary care treatment costs and treatment efficacy for people with type 1 and type 2 diabetes in the United Kingdom from 1997 to 2007. Diabetes Med. 2010;27(8):938–48. 5. Morgan CL, Peters JR, Dixon S, Currie CJ. Estimated costs of acute hospital care for people with diabetes in the United Kingdom: a routine record linkage study in a large region. Diabetes Med. 2010;27(9):1066–73. 6. Inzucchi SE, Bergenstal RM, Buse JB, et al. Management of hyperglycemia in type 2 diabetes: a patient-centered approach: position statement of the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetes Care. 2012;35(6):1364–79. 7. Paschou SA, Leslie RD. Personalizing guidelines for diabetes management: twilight or dawn of the expert? BMC Med. 2013;11: 161. 8. National Institute for Health and Care Excellence. Type 2 diabetes. CG87. London, UK: National Institute for Health and Care Excellence; 2014. 9. Choudhary P, Amiel SA. Hypoglycaemia: current management and controversies. Postgrad Med J. 2011;87(1026):298–306. 10. Marrett E, Stargardt T, Mavros P, Alexander CM. Patientreported outcomes in a survey of patients treated with oral antihyperglycaemic medications: associations with hypoglycaemia and weight gain. Diabetes Obes Metab. 2009;11(12):1138–44. 11. Watson L, Wilson BP, Alsop J, Kumar S. Weight and glycaemic control in type 2 diabetes: what is the outcome of insulin initiation? Diabetes Obes Metab. 2011;13(9):823–31. 12. Grandy S, Fox KM, Bazata DD; SHIELD Study Group. Healthrelated quality of life association with weight change in type 2 diabetes mellitus: perception vs. reality. Int J Clin Pract. 2013;67(5): 455–61. 13. Bagust A, Beale S. Modelling EuroQol health-related utility values for diabetic complications from CODE-2 data. Health Econ. 2005;14(3):217–30. 14. Palmer AJ; Mount Hood 5 Modeling Group. Computer modeling of diabetes and its complications: a report on the Fifth Mount Hood Challenge Meeting. Value Health. 2013;16(4):670–685. 15. McEwan P, Peters JR, Bergenheim K, Currie CJ. Evaluation of the costs and outcomes from changes in risk factors in type 2 diabetes using the Cardiff stochastic simulation cost-utility model (DiabForecaster). Curr Med Res Opin. 2006;22(1):121–9. 16. McEwan P, Bergenheim K, Yuan Y, Tetlow AP, Gordon JP. Assessing the relationship between computational speed and precision: a case study comparing an interpreted versus compiled programming language using a stochastic simulation model in diabetes care. Pharmacoeconomics. 2010;28(8):665–74. 17. McEwan P, Evans M, Kan H, Bergenheim K. Understanding the inter-relationship between improved glycaemic control, hypoglycaemia and weight change within a long-term economic model. Diabetes Obes Metab. 2010;12(5):431–6.
MEDICAL DECISION MAKING/MON–MON XXXX
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
ESTIMATING COST-EFFECTIVENESS IN TYPE 2 DIABETES
18. McEwan P, Evans M, Bergenheim K. A population model evaluating the costs and benefits associated with different oral treatment strategies in people with type 2 diabetes. Diabetes Obes Metab. 2010;12(7):623–30. 19. Clarke PM, Gray AM, Briggs A, et al. A model to estimate the lifetime health outcomes of patients with type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68). Diabetologia. 2004;47(10):1747–59.
29. Saenz A, Fernandez-Esteban I, Mataix A, et al. Metformin monotherapy for type 2 diabetes mellitus. Cochrane Database Syst Rev. 2005;20(3):CD002966. 30. Monami M, Marchionni N, Mannucci E. Long-acting insulin analogues versus NPH human insulin in type 2 diabetes: a metaanalysis. Diabetes Res Clin Pract. 2008;81(2):184-9. 31. Joint Formulary Committee. British National Formulary. 65th ed. London: BMJ Group and Pharmaceutical Press; 2013.
20. Ara R, Blake L, Gray L, et al. What is the clinical effectiveness and cost-effectiveness of using drugs in treating obese patients in primary care? A systematic review. Health Technol Assess. 2012; 16(5):iii–xiv, 1–195.
32. Khunti K, Wolden ML, Thorsted BL, Andersen M, Davies MJ. Clinical inertia in people with type 2 diabetes: a retrospective cohort study of more than 80,000 people. Diabetes Care. 2013; 36(11):3411–7.
21. Clarke P, Gray A, Legood R, Briggs A, Holman R. The impact of diabetes-related complications on healthcare costs: results from the United Kingdom Prospective Diabetes Study (UKPDS Study No. 65). Diabetic Med. 2003;20:442–50.
33. Saunders C, Byrne CD, Guthrie B, et al; Scottish Diabetes Research Network Epidemiology Group. External validity of randomized controlled trials of glycaemic control and vascular disease: how representative are participants? Diabetes Med. 2013; 30(3):300–8.
22. Clarke P, Gray A, Legood R, Holman R. Estimating utility values for health states of type 2 diabetic patients using the EQ-5D (UKPDS 62). Med Decis Making. 2002;2(4):340–349. 23. Currie CJ, McEwan P, Peters JR, Patel TC, Dixon S. The routine collation of health outcomes data from hospital treated subjects in the Health Outcomes Data Repository (HODaR): descriptive analysis from the first 20,000 subjects. Value Health. 2005;8:581–90. 24. Barry H, Ebell M, Hickner J. Evaluation of suspected urinary tract infection in ambulatory women: a cost-utility analysis of office-based strategies. J Family Pract. 1997;44(1):49–60. 25. Lane S, Levy AR, Mukherjee J, Sambrook J, Tildesley H. The impact on utilities of differences in body weight among Canadian patients with type 2 diabetes. Curr Med Res Opin. 2014;30(7): 1267–73. 26. Currie CJ, Morgan CL, Poole CD, Sharplin P, Lammert M, McEwan P. Multivariate models of health-related utility and the fear of hypoglycaemia in people with diabetes. Curr Med Res Opin. 2006; 22(8):15–23. 27. Bennett H, McEwan P, Bergenheim K, Gordon J. Assessment of unmet clinical need in type 2 diabetic patients on conventional therapy in the UK [published online 4 September 2014]. Diabetes Ther. 28. Nauck MA, Del Prato S, Meier JJ, Dura´n-Garcı´a S, Rohwedder K, Elze M, et al. Dapagliflozin versus glipizide as add-on therapy in patients with type 2 diabetes who have inadequate glycemic control with metformin: a randomized, 52-week, double-blind, active-controlled noninferiority trial. Diabetes Care. 2011;34(9): 2015-22.
34. DeFronzo RA, Stonehouse AH, Han J, Wintle ME. Relationship of baseline HbA1c and efficacy of current glucose-lowering therapies: a meta-analysis of randomized clinical trials. Diabetes Med. 2010;27(3):309–17. 35. Giugliano D, Maiorino M, Bellastella G, Chiodini P, Esposito K. Relationship of baseline HbA1c, HbA1c change and HbA1c target of \ 7% with insulin analogues in type 2 diabetes: a meta-analysis of randomised controlled trials. Int J Clin Pract. 2011;65(5):602–12. 36. National Institute of Health and Care Excellence. Guide to the methods of technology appraisal. 2013. Available from: http://publications.nice.org.uk/pmg9 37. Briggs AH, Weinstein MC, Fenwick EA, Karnon J, Sculpher MJ, Paltiel AD; ISPOR-SMDM Modeling Good Research Practices Task Force. Model parameter estimation and uncertainty analysis: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force Working Group-6. Med Decis Making. 2012;32(5):722–32. 38. NICE technology appraisal guidance 288. Available from: www.guidance.nice.org.uk/ta288 39. McIntosh B, Cameron C, Singh SR, et al. Second-line therapy in patients with type 2 diabetes inadequately controlled with metformin monotherapy: a systematic review and mixed-treatment comparison meta-analysis. Open Med. 2011;5(1):e35-e48. 40. McIntosh B, Cameron C, Singh SR, Yu C, Dolovich L, Houlden R. Choice of therapy in patients with type 2 diabetes inadequately controlled with metformin and a sulphonylurea: a systematic review and mixed-treatment comparison meta-analysis. Open Med. 2012;6(2):e62–74.
11
Downloaded from mdm.sagepub.com at VITERBO UNIV on January 30, 2015
