Potential Role of Network Meta-Analysis in Value-Based Insurance Design

, , , ,
The American Journal of Managed Care, August 2014, Volume 20, Issue 8

This study illustrates that where clinical trials are lacking, network meta-analysis can provide valuable insights into the potential clinical and economic benefits of value-based insurance design.


Value-based insurance design (V-BID) has emerged as an approach to improve health outcomes and contain healthcare costs by encouraging use of high-value care. We estimated the impact of a V-BID for osteoporosis treatments using comparative effectiveness evidence and real-world data from a California health insurance plan to estimate the benefits of the design’s implementation.


This study consisted of 4 steps. First, we reviewed randomized clinical trials including osteoporosis treatments—alendronate, ibandronate, risedronate, raloxifene, and teriparatide—reported in a recent Agency for Health Research Quality systematic review. Second, we performed a network meta-analysis to synthesize data from the clinical trials and estimate the comparative effectiveness of included treatments. Third, we implemented a V-BID by removing co-payments for the most effective treatments. Fourth, using a Monte Carlo simulation, we estimated the impact of the V-BID in terms of fracture reduction and cost-savings.


Thirty-two randomized controlled trials were included in the network meta-analysis. We estimated that alendronate, risedronate, and teriparatide have the highest probability of being most effective across each fracture type—vertebral, hip, and nonvertebral/ nonhip. After eliminating co-payments, (ie, reducing them to zero), for these treatments, we estimated the health plan would experience a 7% (n = 287) decrease in fractures and an 8% ($6.8 million) decrease in costs.


Our study illustrates the benefits of comparative effectiveness evidence in V-BID development. We show that where clinical trials are lacking, network meta-analysis can provide valuable insights into the potential clinical and economic benefits of V-BID.

Am J Manag Care. 2014;20(8):641-648

In this study, we use network meta-analysis to generate comparative effectiveness evidence and rank osteoporosis treatments in order of efficacy. Using claims data from a large California health plan, we illustrate that where clinical trials are lacking, network meta-analysis can provide valuable insights into the potential clinical and economic benefits of value-based insurance design (V-BID). This study:

  • Emphasizes the importance of taking an evidence-based approach to pharmacy benefit design.
  • Illustrates the potential value of network meta-analysis in the absence of appropriate clinical trial evidence.
  • Provides a framework for using evidence synthesis methods in V-BID.

It is estimated that more than 10 million Americans 50 years or older have osteoporosis and more than 43 million have low bone mass.1 A chronic condition characterized by low bone mass and deterioration of bone microarchitecture, osteoporosis increases patients’ risk of bone fracture. Most common fractures include those of the vertebrae, hip, and wrist, and result in substantial morbidity and medical and hospital costs. It is estimated that approximately 2.05 million osteoporosis-related fractures occur annually in the United States, costing about $16.9 billion.2 Osteoporosis management focuses on reducing fracture risk, and includes lifestyle modification (eg, smoking cessation and alcohol moderation), weight-bearing exercises, and treatment with pharmaceuticals. Various classes of pharmaceuticals are indicated for osteoporosis treatment, among them bisphosphonates (including alendronate, ibandronate, and risedronate), recombinant parathyroid hormone (teriparatide), selective estrogen receptor modulator (raloxifene), and monoclonal antibodies (denosumab). osteoporosis therapies—alendronate, ibandronate, risedronate, raloxifene, teriparatide—are generally available through a health plan’s pharmacy benefit.

Historically, drug co-payments and coinsurance have been largely based on drug cost, and did not take treatment benefit into account. However, this paradigm is changing, and a move toward value-based insurance design (V-BID) is gathering momentum.3-5 Essentially, the aim of V-BID is to improve healthcare and reduce costs by encouraging high-value care— care that offers clinical benefits at a reasonable cost—and discouraging low-value care. First proposed more than a decade ago, various value-based health insurance programs are now established and subject to much discussion and evaluation in the medical literature. V-BID has been applied to many indications, including diabetes, hyperlipidemia, and asthma.6-9 In contrast to existing approaches to V-BID in which copayments are reduced to encourage use of broad categories of high-value care (eg, statins for hypercholesterolemia or hyperglycemic medications for type 2 diabetes mellitus), we sought to use comparative effectiveness evidence to prioritize treatments within a drug class, using drugs that treat osteoporosis as a case study.7,10 However, as osteoporosis treatments have not been adequately compared against one another in head-to-head studies, we could not rely on clinical trial data to inform our approach, and therefore we used network meta-analysis to synthesize the necessary comparative evidence.

Network meta-analysis is a statistical approach to synthesizing comparative effectiveness evidence and is a generalization of meta-analysis, combining head-tohead clinical trial evidence with statistically inferred indirect comparisons across treatments not studied within head-to-head clinical studies.11,12 The approach requires a connected network of clinical trials. For example, in a data set consisting of pairwise comparisons (eg, A compared with B, and A compared with C), the relative efficacy can be inferred for those comparisons not studied directly (eg, in the previous example, B compared with C through common comparator A). Thus, through a combination of direct and indirect evidence, the network meta-analysis provides the relative efficacy of the whole network. Network meta-analysis has 2 principal roles: first, to strengthen inference of relative effectiveness between a pair of treatments through the combination of direct and indirect evidence; and second, to infer relative efficacies between treatments that have not been evaluated in a head-to-head study. Furthermore, the approach allows estimation of the probability that each included treatment is most effective, an important consideration for V-BID. Network meta-analysis methods are increasingly used and are promoted by health technology assessment agencies including the National Institute for Health and Clinical Excellence (NICE) and the Canadian Agency for Drugs and Technology in Health (CADTH).13-15

The objectives of this study were to construct a V-BID for osteoporosis treatments using comparative effectiveness evidence synthesized from a network meta-analysis; and to illustrate the potential of this approach through estimation of the number of avoided osteoporosis-related fractures and of cost savings using a simulation model. These estimates were specific to a large California-based health plan. We considered the osteoporosis treatments alendronate, ibandronate, risedronate, raloxifene, and teriparatide in this research, as these drugs are commonly available through drug formularies. We excluded the intravenously administered osteoporosis treatments denosumab and zoledronic acid, as they are not typically part of a tiered drug formulary and not subject to the same co-payment structure.


We used claims obtained from a large California-based private health insurance plan. These data specified the number of patients with an osteoporosis diagnosis, and, for these patients, which one of the included treatments they received. We also used this plan data as the source of drug acquisition cost and patient co-payments.

This study consisted of 4 steps. First, we identified clinical trials of osteoporosis treatments that were included in a 2012 Agency for Healthcare Research and Quality (AHRQ) systematic review. The trials that met our inclusion criteria became part of our study. Second, using these studies, we performed a network meta-analysis to synthesize the baseline fracture risk for vertebral, hip, and nonvertebral/nonhip fracture, and the comparative effectiveness of competing treatments. Third, we simulated a V-BID by adjusting the co-payments of the existing pharmacy benefit in accordance with the synthesized comparative effectiveness data, with co-payments eliminated, ie, reduced to zero, for the most effective treatments. Fourth, we simulated the impact of co-payment adjustment in terms of fracture reduction and cost savings for the health plan. The details of each step follow.


Step 1: Study Identification and Review. We relied on studies reported in the 2012 AHRQ report Treatment To Prevent Fractures in Men and Women With Low Bone Density or Osteoporosis: Update of a 2007 Report.16 Included studies were limited to those that: (i) included adults with low bone density or with osteoporosis; (ii) examined a pharmacological intervention reported within the private health insurance plan’s claims data; (iii) reported vertebral, hip, and/or total fractures; (iv) lasted a minimum of 6 months; and (v) were randomized controlled trials. Eligibility criteria of each of the included studies are listed in the (available at www.ajmc.com). Pairs of reviewers read each article to confirm reported counts for vertebral, hip, nonvertebral, and nonvertebral/nonhip fractures, and in some instances contacted the original authors for clarification.

Step 2: Network Meta-Analysis. Using the extracted data, we performed a Bayesian network meta-analysis to estimate 2 pieces of information—underlying fracture risk, ie, the fracture risk in untreated patients, and comparative effectiveness of the various agents both to the reference treatment (placebo) and to one another.

We conducted the network meta-analysis using a random- effect, binomial logit-linked model implemented in WinBUGS 1.4. We used a binomial distribution because of the binomial nature of fractures, and the logit-link assumes a linearity of effects on the logit scale. We determined to use a random effects model after our inspection of the deviance information criterion showed no significant difference in goodness of fit between fixed and random effects models. The networks of studies for each end point (ie, vertebral fractures, hip fractures, and nonvertebral/nonhip fractures) shared the same structure, with individual studies comparing treatments to placebo, though the individual patient counts varied across the outcome networks. Characteristics of the studies populating these networks are provided in the eAppendix. For each end point, ie, vertebral, hip, and nonvertebral/ nonhip fractures, we estimated the underlying fracture risk from the placebo arms of each study. We estimated the comparative effectiveness of the various agents to one another (relative risk and 95% credible interval) indirectly through the reference treatment, ie, placebo. We additionally examined a synthetic end point—total fractures—as this provided a summary efficacy end point with which to rank treatments.

Step 3: Implementation of a Value-Based Insurance Design. In accordance with the results of the network metaanalysis, we modeled the effect of implementing a V-BID by eliminating co-payments for the most effective drugs, ie, alendronate, risedronate and teriparatide. We assumed that co-payment reduction would result in a proportion of patients shifting toward these treatments. How consumers respond to price is typically quantified using the elasticity of demand, which provides an estimate of how utilization of a product is influenced by price. For example, if a product’s price decreases 10% and the elasticity of demand is —0.5, then the population will consume 5% more of the product. Based on estimates from the literature, we simulated an elasticity of demand of –0.2 to –0.6 using a uniform distribution.17 Additionally, as a sensitivity analysis, we estimated the effects of decreasing utilization of the least effective drugs by 50% (range 10%-90%), and increasing utilization of the most effective drugs, accordingly. We used a 50% decrease in utilization of the least effective drugs as we considered it infeasible, in practice, to shift all patients.

Table 1

Table 2

Step 4: Estimation of Fracture Reduction and Cost-Savings. We used a simulation model to estimate the aggregate reduction in fractures and cost savings associated with implementing the V-BID using a 3-year time horizon. The model compares estimated fracture incidence and health plan costs of an existing distribution of osteoporosis treatments among a cohort of patients with an osteoporosis diagnosis covered by a large, private California health plan (n = 13,777) with an alternative distribution of treatments among this same cohort resulting from the V-BID implementation. We relied on the California health insurance plan claims data as the source of baseline drug utilization, annual drug cost, and annual patient drug co-pay, and on Shi et al (2009) for fracture cost ().18 Using a Monte Carlo simulation approach (Microsoft Excel via Visual Basic), we iterated the model 1000 times, each time using a different set of random values from each input parameter probability function (Table 1 and ). Using this method, uncertainties in model inputs are propagated into uncertainties in model outputs, ie, mean estimate and 95% CI.



Thirty-two studies met our inclusion criteria and were included in the network meta-analysis, including 8 randomized, placebo-controlled trials pertaining to alendronate, 4 to ibandronate, 3 to raloxifene, 13 to risedronate, and 4 to teriparatide (details of each study provided in the available at www.ajmc.com). Results from the network meta-analyses are shown in Table 2. For vertebral, hip, and nonvertebral/nonhip fractures, alendronate, risedronate, and teriparatide were consistently estimated to have the 3 highest probabilities of being the most effective treatment, although rank order varied by end point. Teriparatide and risedronate had the highest probabilities of being the best treatment for vertebral and nonvertebral/nonhip fractures with alendronate ranked third. Alendronate was ranked second to risedronate for hip fracture. To confirm efficacy order and provide an aggregate efficacy end point, we constructed the synthetic end point of total fractures. Consistent with the other end points, alendronate, risedronate, and teriparatide were estimated to have the highest probability of being the most effective. Across each end point, ibandronate and raloxifene were consistently estimated to have the lowest probability of being most effective. Consistent with these findings, we constructed a V-BID by eliminating copayments for alendronate, risedronate, and teriparatide, while maintaining the existing co-payment for ibandronate and raloxifene.

Table 3

We estimated that plan beneficiaries receiving treatments under the current benefit structure would suffer an estimated 3668 total fractures and incur roughly $72.9 million in osteoporosis-related costs, with $67.8 million related to fracture costs (). Implementing the V-BID resulted in an estimated 3381 total fractures and costs of approximately $66.1 million, with $60.9 million related to fracture costs. Therefore, compared with the status quo, the V-BID was estimated to result in 287 fewer fractures (a 7% reduction) while reducing health plan costs by approximately $6.8 million (an 8% reduction). Percentage reduction in costs was greater than the percentage reduction in fractures as the majority of avoided fractures were of the hip (221, 77% of total fractures), the fracture type associated with the highest average cost ($26,545). Indeed, hip fracture reduction accounted for the majority of cost savings, approximately $6.2 million (approximately 90% of total fracture- related cost saving). Implementing the V-BID reduced co-payment revenue by approximately $113,000 (7%), but this was more than offset by cost savings resulting from reduced hospitalizations of approximately $6.9 million (8%), and reduced drug costs of approximately $63,000 (1%).

We simulated a scenario in which 50% (range 10% to 90%) of individuals using the least effective treatments were shifted to the most effective treatments. In this scenario, the number of total fractures was reduced by 8.5% (range 1.8% to 15.5%; corresponding to a 10% and 90% shift, respectively) and total costs for the plan were reduced by 9.0% (range 1.9% to 16.0%).


Osteoporosis is a burdensome disease that will become increasingly prevalent as the population ages.2 Alongside lifestyle changes, pharmaceuticals are an integral part of disease management. This study builds on a growing body of literature showing how V-BID can lead to improved drug adherence, health outcomes, and cost savings.6,19,20 We took a novel approach to constructing a V-BID formulary and estimating potential health benefits and cost savings using real-world data from a major Californiabased private health plan. Our approach incorporated available randomized clinical trial evidence reported in an AHRQ systematic review, and used established statistical approaches to develop a V-BID. We estimated that alendronate, risedronate, and teriparatide were the most efficacious treatments, with ibandronate and raloxifene performing unfavorably across the considered end points. These findings are consistent with other network metaanalyses evaluating these treatments.21,22 We estimated that implementation of the V-BID would avoid 287 fractures (7%), including 19 vertebral fractures, 221 hip fractures, and 47 nonvertebral nonhip fractures, and would result in $6.8 million (8%) in cost savings for 1 California-based health plan. However, as our approach is associated with inherent uncertainties, and assumptions regarding adherence and therapeutic substitution, study results should be considered illustrative. Nevertheless, our findings support a burgeoning body of literature showing that V-BIDs can lead to both clinical and economic benefits.

Limitations. One limitation of our study is that it does not account for drug adherence, an important aspect of care for which V-BIDs have been shown to have a positive impact.6,19,20 Another limitation is that we based our V-BID solely on efficacy data, which is only 1 input into prescribing decisions, and did not account for other factors that play a role, including side effect profile and drug interactions. Patient and physician preferences also likely play a role. For example, unlike the other considered orally administered treatments, teriparatide is administered via subcutaneous injection, which may affect patient treatment preference.23 In addition, the FDA-approved labels for teriparatide and raloxifene include a black box warning for potential increase in the incidence of osteosarcoma, and increased risk of venous thromboembolism and death from stroke, respectively, which may also affect patient and physician choice.1,23 Further, we did not consider whether physicians account for treatment cost in prescribing decisions; evidence as to whether physicians account for the cost of technology, or patients’ ability to pay, in prescribing decisions is conflicting.24-26 These limitations may have affected estimated health gain and cost savings.

Moreover, network meta-analyses do not possess the validity of head-to-head clinical trials and it is important to recognize the approach’s limitations. By extrapolating the available data, network meta-analysis makes indirect comparisons between treatments not subject to head-to head clinical trials. As is the case for traditional meta-analysis, because randomization only holds within and not across the clinical trials included in a network meta-analysis, there is a risk of consistency violations, ie, that patients included in different comparisons are dissimilar. Therefore, a firm conclusion can be best drawn when a network meta-analysis is restricted to well-conducted, adequately powered randomized trials including similar patient populations.

These caveats notwithstanding, evidence suggests that results from well-conducted network meta-analyses are typically consistent with those from head-to-head randomized clinical trials.27 Further, while adequately powered prospective randomized clinical trials comparing all competing treatments provide the most robust information, they are infeasible in practice. Network meta-analysis approaches likely will become increasingly recognized and accepted as the source of this evidence, since they allow decision makers to immediately compare treatments as new evidence and treatments are produced.

We did not include osteoporosis treatments administered in an outpatient setting (eg, denosumab and zoledronic acid), because they are typically not part of the pharmacy benefit. While including outpatient drugs in the analysis would more comprehensively account for osteoporosis therapeutic options, the study would need to be redesigned and expanded beyond the scope of a pharmacy benefit.

Osteoporosis was particularly amenable to this study as fracture occurrence is the predominant cause of disease- related morbidity and costs. Minimizing fractures leads to unambiguous health benefits for patients, and the avoidance of costly hospitalizations for the health plan. While this approach could be in theory extended to other indications, complex chronic diseases with multiple health states would require a more elaborate disease model and simulation approach.


Our approach illustrates the potential of using existing systematic review evidence and established evidence synthesis techniques for V-BID. It is a novel evidencebased approach that relies on comparative effectivenes evidence, and allows all available clinical trial evidence to be accounted for. While pharmacy benefit design should be informed by well-structured randomized controlled trials when available, we show that applying properly done network meta-analysis can provide insights where trials alone are lacking, and a realistic means of simulating the clinical and economic benefits of V-BID.


The authors are grateful to Peter J. Neumann, ScD, for his helpful comments on an earlier draft of this manuscript.

Author Affiliations: All authors are from The Center for the Evaluation of Value and Risk in Health, Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, Boston, MA, except Michael J. Cangelosi, who is now with Boston Scientific.

Funding Source: This research was based in part on support from the California Health Care Foundation.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (JDC, AW, NO, MJC); acquisition of data (JDC, AW, YZ, NO, MJC); analysis and interpretation of data (JDC, AW, YZ); drafting of the manuscript (JDC, AW, YZ, MJC); critical revision of the manuscript for important intellectual content (JDC, AW, YZ, NO, MJC); statistical analysis (JDC, AW, YZ, MJC); administrative, technical, or logistic support (YZ, MJC); and supervision (JDC).

Address correspondence to: James D. Chambers, PhD, MPharm, MSc, assistant professor, The Center for the Evaluation of Value and Risk in Health, Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, 800 Washington St, Boston, MA 02111. E-mail: JChambers@tuftsmedicalcenter.org.

1. FDA. Prescribing information for EVISTA (raloxipene hydrocholride). September 13, 2007. http://www.accessdata.fda.gov/drugsatfda_docs/label/2007/022042/bi.pdf.

2. Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, Tosteson A. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005-2025. J Bone Miner Res. 2007;22(3):465-475.

3. Fendrick AM, Chernew ME, Levi GW. Value-based insurance design:embracing value over cost alone. Am J Manag Care. 2009;15(10 suppl):S277-S283.

4. Neumann PJ, Auerbach HR, Cohen JT, Greenberg D. Low-value services in value-based insurance design. Am J Manag Care. 2010;16(4):280-286.

5. Chernew ME, Rosen AB, Fendrick AM. Value-based insurance design. Health Aff (Millwood). 2007;26(2):w195-w203.

6. Farley JF, Wansink D, Lindquist JH, Parker JC, Maciejewski ML. Medication adherence changes following value-based insurance design. Am J Manag Care. 2012;18(5):265-274.

7. Frank MB, Fendrick AM, He Y, et al. The effect of a large regional health plan’s value-based insurance design program on statin use. Med Care. 2012;50(11):934-939.

8. Gibson TB, Wang S, Kelly E, et al. A value-based insurance design program at a large company boosted medication adherence for employees with chronic illnesses. Health Aff (Millwood). 2011;30(1):109-117.

9. Fendrick AM, Smith DG, Chernew ME, Shah SN. A benefit-based copay for prescription drugs: patient contribution based on total benefits, not drug acquisition cost. Am J Manag Care. 2001;7(9):861-867.

10. Kim YA, Loucks A, Yokoyama G, Lightwood J, Rascate K, Serxner SA. Evaluation of value-based insurance design with a large retail employer. Am J Manag Care. 2011;17(10):682-690.

11. Lu G, Ades AE. Combination of direct and indirect evidence in mixed treatment comparisons. Stat Med. 2004;23(20):3105-3124.

12. Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple treatments: combining direct and indirect evidence. BMJ. 2005;331(7521):897-900.

13. National Institute for Health and Clinical Evidence (NICE). Guide to the Methods of Technology Appraisal 2013. London, UK: NICE; 2013.

14. Canadian Agency for Drugs and Technology in Health (CADTH). Indirect Evidence: Indirect Treatment Comparisons in Meta-Analysis. Ottawa, Canada: CADTH, 2009.

15. Salanti G. Indirect and mixed-treatment comparison, network, or multiple-treatments meta-analysis: many names, many benefits, many concerns for the next generation evidence synthesis tool. Res Synth Methods. 2012;3(2):80-97.

16. Newberry SJ, Crandall CJ, Gellad WG, et al. Treatment To Prevent Fractures in Men and Women With Low Bone Density or Osteoporosis: Update of a 2007 Report. Comparative Effectiveness Reviews, No. 53. Rockville, MD: Agency for Health Research and Quality; 2012.

17. Goldman DP, Joyce GF, Zheng Y. Prescription drug cost sharing: associations with medication and medical utilization and spending and health. JAMA. 2007;298(1):61-69.

18. Shi N, Foley K, Lenhart G, Badamgarav E. Direct healthcare costs of hip, vertebral, and non-hip, non-vertebral fractures. Bone. 2009;45(6):1084-1090.

19. Choudhry NK, Fischer MA, Avorn J, et al. At Pitney Bowes, valuebased insurance design cut copayments and increased drug adherence. Health Aff (Millwood). 2010;29(11):1995-2001.

20. Maciejewski ML, Farley JF, Parker J, Wansink D. Copayment reductions generate greater medication adherence in targeted patients. Health Aff (Millwood). 2010;29(11):2002-2008.

21. Freemantle N, Cooper C, Diez-Perez A, et al. Results of indirect and mixed treatment comparison of fracture efficacy for osteoporosis treatments: a meta-analysis. Osteoporos Int. 2013;24(1):209-217.

22. Murad MH, Drake MT, Mullan RJ, et al. Clinical review. Comparative effectiveness of drug treatments to prevent fragility fractures: a systematic review and network meta-analysis. J Clin Endocrinol Metab. 2012;97(6):1871-1880.

23. FDA. Prescribing information for FORTEO (teriparatide [rDNA origin] injection) for subcutaneous use [package insert]. Indianapolis, IN: Lilly USA, LLC; 2012.

24. Li S, Laxminarayan R. Are physicians’ prescribing decisions sensitive to drug prices? evidence from a free-antibiotics program. Health Econ. 2013. doi.10.1002/hec.3008.

25. Glied S, Zivin JG. How do doctors behave when some (but not all) of their patients are in managed care? J Health Econ. 2002;21(2):337-353.

26. Wang R, Pauly M. Spillover effects of restrictive drug formularies on physician prescribing behavior: evidence from Medicaid. Journal of Economics and Management Strategy. 2005;14(3):755-773.

27. Song F, Altman DG, Glenny AM, Deeks JJ. Validity of indirect comparison for estimating efficacy of competing interventions: empirical evidence from published meta-analyses. BMJ. 2003;326(7387):472.