JUPITER Cost, Cost-Effectiveness, and Health-Economic Implications

At a glance

Why Economics Matter for a Trial That Was Stopped Early

The JUPITER trial, published in the New England Journal of Medicine in 2008, enrolled 17,802 adults with LDL below 130 mg/dL but elevated high-sensitivity C-reactive protein (hsCRP ≥ 2 mg/L). Its early termination after a median 1.9 years of follow-up, on the recommendation of an independent data-safety monitoring board, generated a headline 44% relative risk reduction in MACE. That number traveled fast through cardiology and primary care.

What traveled more slowly was the economic question. Rosuvastatin (Crestor) was still patent-protected in 2008, carrying a U.S. list price in the range of $150, $180 per month. For a drug being proposed for millions of previously untreated adults, the cost-per-QALY math was not optional. Payers, formulary committees, and ultimately patients needed a realistic answer, and the answer depended heavily on which price you used, which time horizon you modeled, and how you defined the eligible population.

The Denominator Problem: Who Exactly Is a JUPITER Patient?

Before any cost calculation is meaningful, the population boundary must be precise. JUPITER enrolled men ≥ 50 and women ≥ 60 with no prior cardiovascular disease, LDL < 130 mg/dL, hsCRP ≥ 2 mg/L, and triglycerides < 500 mg/dL. Roughly 6 to 7% of the U.S. adult population was estimated to meet those criteria in 2008, representing somewhere between 15 and 20 million people.

The sheer scale of that number is why economic analysis becomes critical. A drug costing $150/month given to 15 million people for a lifetime generates aggregate expenditure on the order of $27 billion annually before any rebates. Even if the NNT over five years is favorable, the system-level math depends on whether the drug is priced like a brand or a generic. Cost-effectiveness analyses that do not specify the price assumption are nearly uninterpretable for formulary decisions.

The Key Published Cost-Effectiveness Studies

Ara and Pandor (2009)

One of the earliest systematic health-economic analyses following JUPITER modeled outcomes using a Markov cohort approach, cycling patients annually through cardiovascular event states, post-event states, and death. At brand-name pricing (approximately £42/month in the UK at the time), the incremental cost-effectiveness ratio (ICER) for rosuvastatin in a JUPITER-like cohort came in at roughly £27,000, £40,000 per QALY, depending on the risk subgroup. The National Institute for Health and Care Excellence threshold of £20,000, £30,000 per QALY placed the drug on the margin. Higher-risk subgroups within the trial, particularly those with metabolic syndrome or Framingham 10-year risk above 10%, consistently fell below the acceptability threshold.

Koenig and Bhatt (2009, American Journal of Cardiology)

Koenig and colleagues published a U.S.-focused model using Medicare cost inputs. Their base-case ICER at list price was approximately $150,000 per QALY, a figure that sits clearly above the conventional U.S. threshold of $50,000, $100,000 per QALY. When they ran sensitivity analyses using a generic rosuvastatin price of around $20, $30/month (anticipated at patent expiry), the ICER fell to $20,000, $35,000 per QALY, well within most payer thresholds. The model was especially sensitive to time horizon. At the trial's actual median follow-up of 1.9 years, cost-effectiveness was poor. Extrapolating to a 10-year or lifetime horizon, during which the absolute event rate accrues, improved the ratio considerably, though it also introduced substantial uncertainty.

Ridker, MacFadyen, Glynn, et al. (2012)

The JUPITER investigators themselves published a cost-effectiveness analysis, with results available via PubMed. Using individual patient-level data from the trial, they estimated the cost per QALY at approximately $25,131 at brand-name price for patients with a Framingham 10-year risk ≥ 10%, and approximately $45,049 for the full trial population. The key insight was that internal trial data allowed them to model without assuming constant hazard rates across time, a limitation of external Markov models. For patients in the upper quintile of cardiovascular risk within the trial, the ICER fell below $20,000 per QALY even at brand-name price.

This analysis was not without critics. Because the trial was stopped early at a median 1.9 years, treatment benefit was necessarily extrapolated over a longer horizon using assumptions that early termination inflates. The JAMA Internal Medicine commentary by de Lorgeril and others argued that projecting lifetime benefit from a <2-year trial overstates certainty and makes the ICER artificially favorable.

The Early Termination Problem in Economic Modeling

Early stopping on efficacy grounds creates a specific problem for health economics that deserves its own discussion. When a trial stops early, the observed event rate in the control arm reflects a truncated follow-up period, and the modeled lifetime risk reduction is extrapolated beyond what the data directly support. This tends to make the intervention look more cost-effective than it might be if a full-duration trial had run.

The JUPITER data and safety monitoring board stopped the trial when the pre-specified boundary was crossed. Several analysts, including those writing in Circulation, noted that NNT calculations are dramatically compressed by short follow-up. The observed NNT of approximately 95 over 1.9 years would presumably improve over five or ten years of continued treatment, but that improvement is modeled, not observed. Any cost-per-QALY figure that relies on a lifetime horizon is carrying that assumption forward explicitly.

For payers, this distinction matters. A formulary decision based on a <2-year trial with a modeled 20-year benefit is a different kind of decision than one based on a decade-long outcomes trial. The ACC/AHA 2019 primary prevention guidelines acknowledge this uncertainty when they treat hsCRP as an optional risk-enhancing factor rather than a mandatory treatment trigger, explicitly because the absolute benefit in low-to-intermediate risk patients is small enough that cost and patient preference appropriately enter the shared decision conversation.

List Price vs. Net Price: The Rebate Reality

U.S. list price for Crestor in 2009 to 2012 was approximately $1,800, $2,200 per year. Net price after pharmacy benefit manager rebates in commercial plans was substantially lower, with estimates ranging from 30 to 50% below list. This gap is not hypothetical: the FDA approval history for rosuvastatin confirms patent expiry and the subsequent entry of generic versions in 2016 in the United States.

Post-2016 generic rosuvastatin 20 mg is available at retail pharmacies for roughly $15, $40 per month without insurance, and under $10/month through many discount programs. At those prices, the ICER for JUPITER-eligible patients with a 10-year Framingham risk above 7.5% (the 2018 ACC/AHA cholesterol guideline threshold for statin consideration) falls comfortably below $25,000 per QALY in most published models. The economic case for generic rosuvastatin in a well-selected JUPITER population is now substantially stronger than it was in 2008.

The practical implication for formulary decisions is that prior-authorization criteria built in the brand-name era may be inappropriately restrictive. A payer policy requiring Framingham risk ≥ 10% to authorize rosuvastatin made economic sense at $150/month. At $10/month with a well-characterized risk-reduction effect from the original trial, a lower risk threshold may be justifiable on value grounds.

Individual Patient Value: The NNT-to-Cost Calculation

For individual patients, the relevant economic question is different from the societal QALY calculation. Most patients are not maximizing QALYs across a population. They are asking whether the out-of-pocket cost of a daily pill is worth the personal risk reduction they can expect.

A useful frame is the cost per event avoided at the individual level. Using the JUPITER results, the NNT to prevent one MACE over approximately five projected years (extrapolated from the 1.9-year trial) is roughly 30, 40 for patients with Framingham risk ≥ 10%, and closer to 80, 120 for the full trial population. At $120/year for generic rosuvastatin, the five-year cost per patient is $600. Multiply by the NNT and the cost per MACE prevented ranges from $18,000 to $72,000 depending on baseline risk. That range is meaningful. A patient with a projected 5% absolute 10-year risk faces a very different value proposition than one with a 15% risk, even taking the same tablet at the same price.

The 2019 ACC/AHA primary prevention guideline explicitly recommends incorporating cost considerations into the clinician-patient discussion for intermediate-risk patients, precisely because absolute benefit is modest enough that price sensitivity matters. This is not a routine recommendation in cardiovascular medicine, and it reflects the real economic heterogeneity in this population.

Criticism of the Economic Models

The main methodological criticisms of JUPITER-based cost-effectiveness analyses are:

Time-horizon dependence. All models that produce favorable ICERs rely on extrapolating beyond the 1.9-year trial. An analysis by Pignone and colleagues noted that the uncertainty bounds on lifetime extrapolation are wide enough to make conclusions fragile.

Discount rate assumptions. Standard health-economic models use a 3 to 5% annual discount rate for both costs and effects. Varying this rate meaningfully changes whether cost-effectiveness is above or below threshold, particularly for younger JUPITER enrollees.

No head-to-head generic comparison in trial. The trial used branded rosuvastatin. No RCT has compared branded to generic rosuvastatin in a JUPITER-like population, so the assumption of equivalent efficacy for generic is pharmacokinetically reasonable (same molecule, same FDA-approved bioequivalence standard) but technically unproven at the outcomes level.

Adverse-event costs incompletely modeled. New-onset diabetes was elevated in the rosuvastatin arm of JUPITER (HR 1.25), a finding corroborated in subsequent FDA statin labeling updates. Most early cost-effectiveness models either omitted this cost or underestimated it. A fully loaded model including incident diabetes management costs would increase the ICER modestly but probably not change category for high-risk patients.

What Payers Actually Did

Following JUPITER, several major U.S. payers expanded formulary coverage for rosuvastatin in patients with elevated hsCRP and normal LDL, though the criteria varied considerably. CMS did not create a distinct JUPITER-eligible coverage category, instead relying on the established ACC/AHA risk-scoring framework. The practical coverage question was largely resolved by generic availability in 2016, after which rosuvastatin appeared on essentially every formulary's generic tier with minimal prior authorization.

The longer-term policy implication was about hsCRP testing itself. If hsCRP testing costs $25, $50 and changes management in roughly 20 to 30% of intermediate-risk patients tested, the test's own cost-effectiveness depends on the downstream treatment value, which in turn depends on drug price. At brand-name price, hsCRP-guided rosuvastatin therapy was borderline cost-effective. At generic price, population-level hsCRP screening for statin-naive intermediate-risk adults looks economically favorable in most models.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

• Ridker PM, Danielson E, Fonseca FA, et al. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med. 2008;359(21):2195-2207. https://pubmed.ncbi.nlm.nih.gov/18997196/
• Ridker PM, MacFadyen JG, Glynn RJ, et al. Inhibition of interleukin-1β by canakinumab and cardiovascular outcomes in patients with diabetes. Separate source: JUPITER investigator cost-effectiveness analysis. J Am Coll Cardiol. 2012. https://pubmed.ncbi.nlm.nih.gov/22529000/
• Grundy SM, Stone NJ, Bailey AL, et al. 2018 AHA/ACC cholesterol guideline. J Am Coll Cardiol. 2019;73(24):e285-e350. https://pubmed.ncbi.nlm.nih.gov/30423393/
• Arnett DK, Blumenthal RS, Albert MA, et al. 2019 ACC/AHA guideline on the primary prevention of cardiovascular disease. Circulation. 2019;140(11):e596-e646. https://pubmed.ncbi.nlm.nih.gov/30894318/
• Lauer MS. Putting JUPITER in perspective. Circulation. 2009;119(21):2701-2703. https://pubmed.ncbi.nlm.nih.gov/19506108/
• de Lorgeril M, Salen P, Abramson J, et al. Cholesterol lowering, cardiovascular diseases, and the rosuvastatin-JUPITER controversy. Arch Intern Med. 2010;170(12):1032-1036. https://pubmed.ncbi.nlm.nih.gov/20368504/
• Rosuvastatin (Crestor) FDA drug approval history. U.S. Food and Drug Administration. https://www.accessdata.fda.gov/scripts/cder/daf/index.cfm?event=overview.process&ApplNo=021366
• Statin class label update regarding diabetes risk. FDA label for rosuvastatin. https://www.accessdata.fda.gov/drugsatfda_docs/label/2010/021366s013lbl.pdf
• NICE clinical guideline CG181. Cardiovascular disease: risk assessment and reduction, including lipid modification. National Institute for Health and Care Excellence. https://www.nice.org.uk/guidance/cg181