Cost-effectiveness of treating type 2 diabetes with angiotensin-converting enzyme inhibitors: Should all patients receive this agent?

Original Citation

Golan L, Birkmeyer JD, and Gilbert H. The cost-effectiveness of treating all patients with type 2 diabetes with angiotensin-converting enzyme inhibitors.

Overall Study Question

The objective of this study was to determine the cost-effectiveness of treating all newly diagnosed type 2 diabetic patients with ACE inhibitors.  A Markov model was used to simulate the progression of diabetic nephropathy in a hypothetical cohort of type 2 diabetics. Patients were assumed to be 50 years old at the time of their diagnoses and were followed through their lifetime.

Lifetime cost, quality adjusted life years (QALY) and marginal cost effectiveness ratios were calculated for the 3 interventions including:

1) a “treat-all” category where patients were not screened and start receiving ACE inhibitor therapy at the time of diagnosis of diabetes;

2) screen for microalbuminuria-patients intervention where patients were screened for microalbuminuria and treatment is attempted in all patients whose tests result is positive; and,

3) a screen for gross proteinuria > intervention where patients are screened and treated if results are positive.

Are the Results of the Study Valid?

Did the analysis provide a full economic comparison of health care strategies?

Yes.  This study provided a full economic evaluation and considered the cost and outcomes for each intervention, from the societal perspective, and included both a cost effectiveness and a cost utility analysis.  The analysis included Quality Adjusted Life Years (QALYs) and a cost utility measure (cost/QALYs gained).  Incremental analyses were performed, and the three interventions were compared to each other.

Were the costs and outcomes properly measured and valued? 

Yes.  The costs considered were the health care costs associated with ACE inhibitor therapy, screening, and treatment of end stage renal disease (ESRD).  The authors assumed that other health-related costs of diabetes were otherwise the same for the three intervention strategies. The costs seem to have been taken from appropriate U.S. sources (e.g., Medicare Clinical Diagnostic Fee Schedule), but the economic evaluation was taken from a Canadian perspective and therefore the costs may not be appropriate.  Since so few clinical studies have been conducted on this topic, and since good clinical data would require a large cohort followed over a long period of time, the Markov model used in this study was appropriate.  The main outcome measures were average lifetime cost; life expectancy; and quality adjusted life expectancy.  From this, the marginal (i.e. incremental) cost-effectiveness was calculated for each strategy.  Health related QOL was taken from literature and included diabetes-specific QOL scores, therefore seems appropriate.  A discount rate of 3% was used for all costs and health benefits.  The cost-effectiveness data was limited to type 2 diabetes, and the data from the Markov model used was based on the results of only 1 clinical trial that was conducted on Type 1 diabetics.  The authors were as specific as possible without doing a primary clinical study, and recognized the potential limitations of the model.

Was appropriate allowance made for uncertainties in the analysis?

Yes.  Extensive sensitivity analyses were performed.  The authors indicate that the results were sensitive to cost of drugs, effectiveness of interventions, and estimates of HRQOL impact, but not sensitive to the costs of treating ESRD.  The ranges tested in the sensitivity analyses were fairly wide and therefore the sensitivity analyses were appropriate.

Are estimates of costs and outcomes related to the baseline risk in the treatment population? 

Yes.  A sub-group analysis was not explicitly done, but within the sensitivity analyses, the age at diagnosis and the relative risk of progression to microalbuminuria were varied to determine the impact of those > risk factors on cost and QALYs gained. The benefit of treating all patients decreased and the marginal cost-effectiveness ratio increased when ACE inhibitors became less efficacious. Therefore, the earlier the age of diagnosis, the more “economically attractive” the “treat all” strategy was shown to be.

What are the Results?

What were the incremental costs and outcomes of each strategy?

STRATEGY: Screen for gross proteinuria: average cost: $19520 life years: 15.38 QALY: 11.59 Cost Effectiveness: dominated.  STRATEGY: Screen for microalbuminuria average cost: $14940 life years: 15.59 QALY: 11.78 Cost Effectiveness: base case.  STRATEGY: Treat all average cost: $15240 life years: 15.63 QALY: 11.82 Cost Effectiveness: $7500/QALYs gained.  Since the “screen for microalbuminuria” intervention had the lowest cost, it  was used as the reference case, and compared to this intervention, the treat-all strategy was more expensive but was associated with the highest life expectancy (i.e. it was dominated).

Do incremental costs and outcomes differ between sub-groups?

Yes.  Again, under the sensitivity analyses, the cost-effectiveness of the “treat all” strategy compared to the screening for microalbuminuria strategy was sensitive to age at diagnosis of diabetes, cost of ACE inhibitors, relative risk for progression to microalbuminuria and HRQOL adjustment.  The incremental costs and outcomes increased with increasing age.  The ratio for the treat all strategy exceeded $20,000/QALY if the patients age at diagnosis was 55 years or older.  However, if the age of diagnosis is 44 years or younger, the treat all intervention becomes the dominant strategy (least expensive, highest benefit).  For the very elderly patients, older than 85 years, the screen for proteinuria was the dominant intervention.

How much does allowance for uncertainty change the results?

Unable to determine.  There is considerable impact of uncertainty on results.  The cost-utility analysis was sensitive to drug costs, risk of progression of nephropathy and HRQOL estimates.  This is not surprising, however, given the wide range of the sensitivity analysis and the wide ranges used.

Will the results help me in caring for my patients?

Are the treatment benefits worth the harms and costs?

Yes.  The base estimate for the cost of ACE inhibitors used was $320/yr, which resulted in a cost-effectiveness ratio of $7500/QALY.  There is an incremental cost effectiveness in the treat all category if the cost of the ACE inhibitor is less that $260/yr.  If the annual cost of ACE exceeds $420/yr, the marginal cost-effectiveness ratio of the treat all strategy exceeds $20000/QALY.  These are all relative values of treating all at the time of diagnosis.  Screening for gross proteinuria was more costly and less effective. It was a dominated strategy.  The findings in this study were not sensitive to screening adherence or treatment discontinuation.  However, if adherence to screening was 100%, and if no patient discontinued treatment, the treat-all strategy was still cost effective.

Could my patients expect similar health outcomes? 

This study was a model of a hypothetical cohort of type 2 diabetics, based in  part on the results of one clinical study conducted on Type 1 diabetics.  Like all economic analyses, the results are not relevant to the individual patient, but is relevant for the entire patient population.  Nonetheless, the model allows for variation in adherence to screening guidelines and treatment discontinuation, age at diagnosis, costs, different degrees of efficacy of ACE inhibitor therapy, and mortality (lifetime outcomes).  The study patients were type 2 diabetics and since an ideal study would require a large number of patients with a long follow-up, this study models such an ideal trial.  It accounted for treatment discontinuation but assumed perfect screening as per National guidelines. However, there are new diagnostic criteria for diabetes and the new criteria may allow for earlier diagnosis and therefore improved cost-effectiveness.

Could I expect similar costs?

Yes, more or less.  Costs seem to have been taken from appropriate “normal” U.S. sources (literature and fee guides) but the model was taken from an American perspective and costs may differ and therefore the results may not be directly relevant.  Costs in Canada may differ, and the different provinces in Canada may have different drug costs.  This is especially important because the model was shown to be sensitive to drug costs.  Therefore, whether or not we could expect similar cost effectiveness depends on the local drug costs.  In Alberta, for example, the average wholesale price per year of lisinopril (as used in the study) is $295.36/yr.  This value falls within the range of the sensitivity analysis, and also falls close to the base annual cost used in this study ($320).  Also, the study used 1998 costs of drugs, and 1996 cost for dialysis, but the range of the sensitivity analysis was very wide for costs, so we could therefore expect costs to be in the range of the sensitivity analysis.

How can the results be put into the context of previous knowledge and clinical practice? What are any important methodological issues with the study?  How should the study findings be applied to the field of clinical pharmacotherapy?

The results of this study were based on a Markov model of a hypothetical cohort of type 2 diabetic patients.  In order for a clinical trial to weigh the costs of treating patients with ACE inhibitors against the progression of nephropathy, a large number of patients with a very long follow-up period would be required, and therefore, the modeling of progression might be the most appropriate type of study.  There is limited, but mounting, previous clinical knowledge regarding treatment with ACE inhibitors in type 2 diabetes.  The current Canadian clinical practice guidelines recommend that all patients with type 1 or type 2 diabetes be annually screened for microalbuminuria (grade D consensus). The CPGs also recommend treatment with an ACE inhibitor for type 1 diabetics with elevated microalbuminuria (30-299 mg/d) to decrease the albuminuria (grade A consensus).  They currently do not recommend treating all type 2 diabetic patients with ACE inhibitors, but suggest that ACE inhibitor treatment may be beneficial (grade B).  ACE inhibition is also recommended for type 1 diabetics with overt nephropathy (grade A consensus).  The model of this study suggests that all newly diagnosed type 2 diabetic patients should be treated with ACE inhibitors to prevent the progression of nephropathy and that treating all patients would be cost effective.  However, CPGs are more conservative since clinical evidence is still lacking and thus cannot provide a strong recommendation with respect to ACE inhibitor treatment in all diabetic patients.

Latest Posts