NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Myers E, Sanders GD, Ravi D, et al. Evaluating the Potential Use of Modeling and Value-of-Information Analysis for Future Research Prioritization Within the Evidence-Based Practice Center Program [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2011 Jun. (Methods Future Research Needs Reports, No. 5.)

Cover of Evaluating the Potential Use of Modeling and Value-of-Information Analysis for Future Research Prioritization Within the Evidence-Based Practice Center Program

Evaluating the Potential Use of Modeling and Value-of-Information Analysis for Future Research Prioritization Within the Evidence-Based Practice Center Program [Internet].

Show details

Case Study: ACEIs and ARBs in Patients with Ischemic Heart Disease

Background

Despite advances in therapy, ischemic heart disease (IHD) remains the most common cause of morbidity and mortality in the United States. The prevalence of IHD is estimated at 16.8 million adults, and the death rate is 278.9 per 100,000 people, with IHD responsible for more than 35 percent of all deaths nationwide.24

Angiotensin-converting enzyme inhibitors (ACEIs) and angiotensin-II receptor blockers/antagonists (ARBs) have been shown to improve clinical outcomes for some patients, including those with heart failure and those with myocardial infarction (MI) and ventricular dysfunction.25–32 However, the comparative effectiveness of ACEIs and ARBs alone or in combination for patients with IHD remains uncertain. Their potential role in the management of the broader population of patients with known IHD or at high risk for IHD is also unclear.

To address this area of uncertainty, a CER project sponsored by AHRQ was awarded to the University of Connecticut EPC. The subsequent CER reviewed data available through July 2009 comparing the benefits and harms of adding ACEIs, ARBs, or both to standard medical therapy in adults with stable IHD or IHD risk equivalents.

The CER found strong evidence that ACEIs reduced total mortality and nonfatal MI in comparison to placebo among adults with stable IHD and preserved ventricular function, but increased the risk for syncope and cough. There was low to moderate evidence that ARBs reduced a composite of cardiovascular endpoints compared to placebo and were well tolerated. The one available study directly comparing the impact of ACEIs and ARBs on cardiovascular outcomes in patients with IHD revealed no significant difference in the rate of cardiovascular outcomes, but demonstrated higher rates of cough and angioedema among patients treated with ACEIs, and higher rates of hypotensive symptoms among patients treated with ARBs.33 The same study compared combination therapy with ACEIs and ARBs to monotherapy with each class of agents and found no difference in vascular outcomes, but a higher discontinuation rate in the combination therapy group due to medication side effects.

Although 41 studies including more than 64,000 randomized patients were evaluated in this CER, the authors identified multiple areas where insufficient evidence existed to answer the key questions regarding the comparative effectiveness of ACEIs and ARBs. While there was a high strength of evidence for ACEIs compared to placebo for total mortality, the evidence was insufficient, low, or moderate for the impact of ACEIs or ARBs on several cardiovascular outcomes, including cardiovascular mortality, nonfatal MI, or stroke, suggesting that future research on the impact of ACEIs or ARBs on cardiovascular outcomes may influence their conclusions.

The Duke EPC was recently tasked with performing a pilot project to explore the future research needs of an existing CER and chose the ACEIs and ARBs in IHD report for this case study. Although the pilot project aimed to produce an actual prioritization of the identified research gaps, the timing of this pilot project in relation to this project on broader issues in methods for prioritization setting allowed us to engage a group of stakeholders and to explore the use of qualitative and quantitative prioritization of research needs.

Methods

Decision Model

We developed a decision analytic framework to explore the underlying uncertainties in the use of ACEIs or ARBs in patients with IHD. Figure 1 provides a schematic of the analytic framework. Patients were assumed to start on either ACEI or ARB monotherapy or both (dual therapy). We explicitly modeled potential side effects from treatment regimens through two mechanisms. First, each month patients could be determined to be nontolerant to their drug regimen. The presence of angioedema was modeled separately from other nontolerance. Patients who experienced angioedema were removed from active therapy. Other nontolerant patients could switch to either the alternative regimen or no therapy. We also explored the impact of additional side effects that did not result in therapy modifications through the use of utilities for the varying drug regimens. We tracked patients’ outcomes over their lifetime and explicitly modeled development of congestive heart failure (CHF), atrial fibrillation, end-stage renal disease (ESRD), diabetes, MI, stroke, and death.

Figure 1 provides a schematic of the analytic framework. Patients were assumed to start on either ACEI or ARB monotherapy or both (dual therapy). Patients’ outcomes over their lifetime and explicitly modeled development of congestive heart failure (CHF), atrial fibrillation, end-stage renal disease (ESRD), diabetes, MI, stroke, and death.

Figure 1

Schematic of decision model.

We assumed that all therapies were equally effective in reducing MI, stroke, ESRD, diabetes, atrial fibrillation, and development of CHF compared to standard medical therapy, but also evaluated a range of potential differences between ACEIs and ARBs. We also assumed that there was no difference in a patient’s blood pressure for any health state. The model also included estimates of quality of life associated with the different health states. In our base-case analysis we assumed that those patients who were tolerating their given drug regimens did not have an additional disutility associated with therapy. The base-case model assumed a class effect for all ACEIs and ARBs. Table 3 lists some of the key data estimates used in our analysis. Additional details about the model are found in our pilot project report.34

Table 3. Key data inputs for decision model.

Table 3

Key data inputs for decision model.

Prioritization Exercises

The engagement of stakeholders and prioritization exercises used are described in more detail in our pilot project report.34

Briefly, nine stakeholders were selected for participation in this project from a variety of backgrounds and perspectives. They included physicians affiliated with academic institutions, representatives of professional societies with a cardiovascular focus or expertise in comparative effectiveness research, a payer institution, industry representatives, the National Heart Lung and Blood Institute, and a patient representative. In selecting members of the stakeholder group, efforts were made to assemble a balanced group of individuals representing a range of perspectives. Efforts were also made to avoid inclusion of researchers whose participation in the prioritization process might result in an unfair advantage in the development of future research proposals.

Project stakeholders participated in three conference calls and three prioritization exercises (Appendix D). Each prioritization exercise built off the findings of the previous exercise. The call and prioritization exercises occurred in the following order:

  • Conference Call 1: Introduced stakeholders to the project’s objectives and described the key clinical questions, the original CER and its findings, and proposed methods for the prioritization process, including use of a decision model and VOI analyses to quantitatively prioritize future research needs.
  • Prioritization Exercise 1: Stakeholders were asked to rate the importance of future research exploring various characteristics using a 5-point Likert scale via an online tool. They were also asked to rank their top five research priorities from the complete list.
  • Conference Call 2: Used to review and discuss the results of the initial exercise.
  • Prioritization Exercise 2: We distributed additional material to stakeholders, including a list of potential priority-setting criteria to use when considering the appropriate priority for the research questions, the results of the initial survey prioritization, and summary evidence tables from the original CER. Each stakeholder was then asked to rank the 16 research areas from 1 to 16 in order of importance.
  • Conference Call 3: Reviewed the findings of the second prioritization exercise, detailed our search of recently published literature and ongoing trials, described the decision analytic model and its key assumptions and data, discussed the model’s findings, and then provided an opportunity for the group to discuss the existing ranking.
  • Prioritization Exercise 3: Further material was distributed to stakeholders, including the qualitative ranking results and the recently published literature and ongoing trials in each research area. Each stakeholder was then asked to rank the areas from 1 to 16. This final step produced our final ranking.

Each call was recorded and stakeholder feedback elicited both during the call, and through a brief survey sent subsequently to the stakeholders to provide an opportunity for further structured feedback.

Results

Model Results

Table 4 presents the health and economic outcomes for the decision model.

Table 4. Health and economic outcomes.

Table 4

Health and economic outcomes.

In the base-case analysis, treatment with ARBs increases life expectancy by 0.0049 years (1.79 days) or 0.0054 QALYs (1.97 quality-adjusted life days) but costs an additional $277, corresponding to an incremental cost-effectiveness ratio (ICER) of $56,198/life-year (LY) or $51,456/QALY. Use of dual therapy is dominated by both monotherapy options (costs more while not increasing life expectancy).

The use of the model allowed us to explore how sensitive our findings were to the data uncertainties—and specifically the data which corresponded to identified potential research areas. To determine the potential benefit of prioritizing specific research areas for further study, we used the model to explore the impact of reducing uncertainties in the comparative effectiveness of ACEIs and ARBs in patients with ischemic heart disease. Specifically, we explored the impact of uncertainty on new diagnoses, quality of life, cardiovascular outcomes, renal insufficiency, non-angioedema adverse events, and angioedema.

Figure 2 displays a tornado diagram which demonstrates the sensitivity of the model’s findings of these key uncertainties and corresponding evidence gaps. This figure demonstrates that the model is most sensitive to the uncertainty surrounding patients’ quality of life and the presence of new diagnoses. Uncertainty related to angioedema does not impact the model’s findings significantly. Ranges used for the listed variables are found in Table 3.

Figure 2 displays a tornado diagram that demonstrates the sensitivity of the model’s findings of these key uncertainties and corresponding evidence gaps. Variables listed below were varied over the range described in Table 3 for uncertainties related to identified research gaps. These represented variables targeting evidence concerning quality of life on ACEI or ARB therapy, new diagnoses of atrial fibrillation, congestive heart failure, or diabetes, occurrence of MIs or stroke, development of renal insufficiency, and angioedema and nonangioedema adverse events.

Figure 2

Sensitivity of model findings to key uncertainties. Variables listed below were varied over the range described in Table 3 for uncertainties related to identified research gaps. These represented variables targeting evidence concerning quality of life (more...)

We presented the model, the underlying data and assumptions, and the findings of our analyses to our stakeholder group. The stakeholders had access to these findings when they prepared their final prioritization of the potential research areas. Note that the use of the decision modeling framework allowed us to explore the relative importance of the underlying uncertainties/research needs concerning the comparative effectiveness of ACEI and ARB therapy for patients with ischemic heart disease—it does not, however, allow a quantitative comparison of these uncertainties with uncertainties in other clinical domains and therefore more formal ranking of these research gaps against those that might be competing for similar resources.

Results of Prioritization Exercise

Detailed rankings for each of our prioritization exercises are described in our pilot report (see also Appendix E). Most of the rankings remained consistent between the second (qualitative) and third (findings from the decision model) exercises. Notable exceptions included the ranking of research into the incidence of new diagnoses (such as diabetes, atrial fibrillation, or CHF with or without preserved left ventricular [LV] function), which fell from second to sixth. It was instead replaced by an emphasis on research into medication adherence. This change was most likely influenced by the relatively large number of recently published studies (n = 6) and ongoing clinical trials (n = 5) related to new diagnoses that were presented to the stakeholders at this point in the project and the scarcity of research (no new studies, and one potentially relevant clinical trial) related to medication adherence. This change also emphasizes the importance of providing sufficient clinical/methodological background information to the stakeholders before the prioritization exercises to limit the changes in their rankings based on the gathering of this knowledge later in the process. Of interest, the decision analytic model of ACEI and ARB therapy in IHD patients indicated that uncertainty related to new diagnoses had a significant impact on the model’s findings.

Although the overall ranking did not change substantially from the second to the third prioritization exercise, the consensus among the stakeholders in their rankings did improve. The variance in the rankings was greatly reduced, there was much more consistency among the stakeholders and their rankings of the top and bottom five areas. Further research is needed to determine whether this greater consistency was related to incorporation of the decision analytic framework, the additional information provided concerning ongoing trials, or the discussions amongst the investigative team and stakeholder group.

Stakeholder Feedback

In feedback provided during the conference calls and via written comments, all stakeholders found the decision analytic modeling exercise useful in thinking about the prioritization of research areas. Stakeholders did not feel that the modeling results and quantitative prioritization process should replace the qualitative prioritization process, but rather felt that these findings should be conveyed to the stakeholders either in advance or in parallel with the qualitative process. The stakeholders felt that the greatest benefit to the prioritization process came from the opportunity to discuss the model and its findings with the analytic team and other stakeholders; however, they also ranked as valuable the quantitative description of key areas of uncertainty, the rank ordering of priorities, and the details of the underlying model. All of the respondents felt that additional background material on the decision analytic framework and VOI analyses, either as a briefing document or as an online resource, would have been helpful.

PubReader format: click here to try

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (1.0M)
  • Disable Glossary Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...