Using models to predict the future: what to do when the data run out?

Although clinical trials often provide "best evidence" comparing the effectiveness of alternative management strategies, such evidence can be limited in duration or in the results reported, causing clinicians and policy analysts to wonder "what if?" Models of the clinical prognosis-often spanning patients' lifetimes (the "long haul")-are perhaps weaker evidence, but can help answer questions about the management of individual patients and place that best evidence into the context of clinical reality.

Diagnosis of DVT: antithrombotic therapy and prevention of thrombosis: American College of Chest Physicians evidence-based clinical practice guidelines

BACKGROUND: Objective testing for DVT is crucial because clinical assessment alone is unreliable and the consequences of misdiagnosis are serious. This guideline focuses on the identification of optimal strategies for the diagnosis of DVT in ambulatory adults. METHODS: The methods of this guideline follow those described in Methodology for the Development of Antithrombotic Therapy and Prevention of Thrombosis Guidelines: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. RESULTS: We suggest that clinical assessment of pretest probability of DVT, rather than performing the same tests in all patients, should guide the diagnostic process for a first lower extremity DVT (Grade 2B). In patients with a low pretest probability of first lower extremity DVT, we recommend initial testing with D-dimer or ultrasound (US) of the proximal veins over no diagnostic testing (Grade 1B), venography (Grade 1B), or whole-leg US (Grade 2B). In patients with moderate pretest probability, we recommend initial testing with a highly sensitive D-dimer, proximal compression US, or whole-leg US rather than no testing (Grade 1B) or venography (Grade 1B). In patients with a high pretest probability, we recommend proximal compression or whole-leg US over no testing (Grade 1B) or venography (Grade 1B). CONCLUSIONS: Favored strategies for diagnosis of first DVT combine use of pretest probability assessment, D-dimer, and US. There is lower-quality evidence available to guide diagnosis of recurrent DVT, upper extremity DVT, and DVT during pregnancy.

Using observational data to estimate prognosis: an example using a coronary artery disease registry.

With the proliferation of clinical data registries and the rising expense of clinical trials, observational data sources are increasingly providing evidence for clinical decision making. These data are viewed as complementary to randomized clinical trials (RCT). While not as rigorous a methodological design, observational studies yield important information about effectiveness of treatment, as compared with the efficacy results of RCTs. In addition, these studies often have the advantage of providing longer-term follow-up, beyond that of clinical trials. Hence, they are useful for assessing and comparing patients' long-term prognosis under different treatment strategies. For patients with coronary artery disease, many observational comparisons have focused on medical therapy versus interventional procedures. In addition to the well-studied problem of treatment selection bias (which is not the focus of the present study), three significant methodological problems must be addressed in the analysis of these data: (i) designation of the therapeutic arms in the presence of early deaths, withdrawals, and treatment cross-overs; (ii) identification of an equitable starting point for attributing survival time; (iii) site to site variability in short-term mortality. This paper discusses these issues and suggests strategies to deal with them. A proposed methodology is developed, applied and evaluated on a large observational database that has long-term follow-up on nearly 10 000 patients.

When should an effective treatment be used? Derivation of the threshold number needed to treat and the minimum event rate for treatment.

Clinicians and patients must decide when treatment effects are large enough to more than offset the adverse effects and costs of therapy. Calculation of the number of patients one needs to treat (NNT) in order to prevent one patient from having the target event is one tool to help with this decision. Clinicians should treat patients when the NNT is lower than a threshold NNT at which point the therapeutic risk equals the therapeutic benefit. We aimed: (1) to identify the determinants of the threshold NNT, and (2) to derive equations for the explicit estimation of the threshold NNT and of the minimum expected rate of target event, without treatment, above which treatment is justified. We conceived the threshold number needed to treat to prevent one target event as the point at which the benefits of treating that number of patients equal the negative consequences of treating that same number of patients. After identifying the various elements comprising the treatment impact, we equated the benefits and negative consequences and solved the equation for threshold NNT. We then derived the minimum expected rate of target event which would justify treatment. We derived an equation that relates the threshold NNT to the treatment-attributable adverse event rates (AER) and the relative values (RV) of the adverse events compared to that of the target event prevented. Specifically, the threshold NNT, denoted NNT(T) is given by NNT(T) = 1/(AER(1).RV(1) +...+ AER(n).RV(n)). We also derived a more complex equation which addresses the costs incurred by treatment and costs avoided by preventing target events. Solving the equation that includes costs requires specifying both the value of preventing a target event and the values of adverse treatment effects in economic units. The threshold NNT and the relative risk reduction (RRR) for the target event determine the minimum target event rate above which treatment is justified. This minimum event rate for treatment is 1/(NNT(T).RRR). The values that clinicians and patients use determine the threshold NNT and therefore also the minimum target event rate, without treatment, above which treatment is justified. Quantification of the determinants of the threshold NNT and of the minimum event rate to justify treatment can assist clinicians and patients in the explicit use of underlying values when making treatment decisions.

Cost-effectiveness of 24 or 48 weeks of interferon alpha-2b alone or with ribavirin as initial treatment of chronic hepatitis C. International Hepatitis Interventional Therapy Group.

OBJECTIVE: Initial therapy with ribavirin and interferon alpha-2b results in a higher sustained virological response than interferon alone, but this regimen is expensive. We aimed to examine the cost-effectiveness of 24- or 48-wk initial treatment with combination therapy versus interferon alone for patients who have chronic hepatitis C. METHODS: Data from recent randomized clinical trials comparing combination therapy to interferon alone were applied to a previously published computer cohort simulation to project lifelong clinical and economic outcomes. Natural history and economic estimates were based on published literature, expert panel estimates, and actual variable cost and reimbursement data. RESULTS: Using treatment stopping rules, sustained viral negative response rates would be 33.1% and 39.8% for patients receiving 24 versus 48 wk of ribavirin/interferon, compared with 14.3% for 48 wk of interferon alone. Compared to the interferon alone strategy, 24 or 48 wk of combination therapy should prolong life expectancy by 1.4 to 2.0 yr at marginal cost-effectiveness ratios of $4,400 to $5,400 per discounted quality-adjusted life-year (DQALY) gained. Compared to 24 wk of combination therapy, 48 wk of combination therapy should prolong life expectancy by 0.6 yr at a marginal cost-effectiveness ratio of $7,700 per DQALY gained. The results were robust, with 24 or 48 wk of combination therapy remaining preferred and cost-effective in sensitivity analysis compared with interferon alone. CONCLUSION: For patients with chronic hepatitis C, 24 or 48 wk of ribavirin and interferon should prolong life and be cost-effective when compared with 48 wk, of interferon alone.

Cost effectiveness of ribavirin/interferon alfa-2b after interferon relapse in chronic hepatitis C.

PURPOSE: Many patients with chronic hepatitis C who are treated with interferon suffer a relapse after an initial response. About half of these patients have a sustained virological response to retreatment with the combination of ribavirin and interferon alfa-2b. The aim of this study was to estimate the cost effectiveness of retreatment with combination therapy versus interferon alone for patients who have previously relapsed after interferon. SUBJECTS AND METHODS: Data from a randomized trial among 345 relapsed patients that compared combination therapy with interferon alone were used to project lifelong clinical and economic outcomes. Natural history and economic estimates (discounted at 3% per year) were based upon published literature, expert panel estimates, and cost and reimbursement data. RESULTS: Compared with retreatment with interferon alone, combination therapy should prolong life expectancy by about 2 discounted quality-adjusted life years (3 life years, undiscounted) while increasing costs modestly. The results were robust, maintaining an advantage to combination therapy in sensitivity analysis for all subgroups and with reasonable variations in all model parameters. CONCLUSION: For patients with chronic hepatitis C who relapse after an initial response to interferon alone, retreatment with the combination of ribavirin and interferon alfa-2b should prolong life and be cost effective.

Cytomegalovirus immune globulin after liver transplantation: a cost-effectiveness analysis.

OBJECTIVE: Cytomegalovirus (CMV) immune globulin (CMVIG) has been shown to significantly reduce severe CMV-associated disease complicating orthotopic liver transplant (OLT). We evaluated the economic impact of severe CMV-associated disease and calculated the marginal cost-effectiveness (C/E) of routine prophylaxis with CMVIG after OLT. DESIGN: C/E analysis. SETTING: Four teaching hospitals in Boston. PATIENTS: Patients who underwent OLT from January 1988 through June 1990. MEASUREMENTS: We gathered actual cost data of hospital care for patients enrolled in a clinical trial of CMVIG prophylaxis in OLT. We calculated average outpatient expenses from a separate group of patients undergoing OLT and developed a regression model to estimate costs during the first year post-transplant (R2 = 0.77). Based on this model, we calculated variable costs (in 1999 US dollars) for all patients in the randomized trial. From the published literature we obtained the probability of CMV outcomes and of long-term survival after OLT. We then developed a decision analytical model to determine an incremental C/E ratio, using a Markov simulation to estimate long-term survival and long-term costs. We discounted costs and life-years at 3% and 5% per yr. RESULTS: Based on the efficacy rate of 54% in the controlled trial, we estimate that CMVIG will increase life expectancy by 0.65 discounted years at an additional cost of $11,600, providing a marginal C/E ratio of $17,900/yr life saved. Examining the confidence limits of efficacy, we estimate that CMVIG will have a marginal C/E ratio of $66,200 gained/yr at an efficacy of 11% and $14,000 gained/yr at an efficacy of 83%. CONCLUSION: After OLT, prophylactic CMVIG has an incremental C/E ratio comparable to that of other well-accepted medical therapies and should be used routinely in these patients.

Individualizing therapy to prevent long-term consequences of estrogen deficiency in postmenopausal women.

BACKGROUND: Alendronate sodium and raloxifene hydrochloride were recently approved for the prevention of postmenopausal osteoporosis, but data on their clinical efficacy are limited. We compared these drugs with hormone replacement therapy (HRT) to help women and physicians guide postmenopausal treatment decisions. OBJECTIVE: To help physicians understand how they can best help women choose the most beneficial therapy after menopause based on their individual risk profile. METHODS: We developed a decision analytic Markov model to compare the effects of alendronate therapy, raloxifene therapy, and HRT on risks of hip fracture, coronary heart disease (CHD), breast cancer, and life expectancy. Regression models linked individual risk factors to future disease risks and were modified by drug effects on bone density, lipid levels, and associated breast cancer effects. RESULTS: Hormone replacement therapy, alendronate therapy, and raloxifene therapy have similar predicted efficacies in preventing hip fractures (estimated relative risk, 0.57, 0.54, and 0.58, respectively). Hormone replacement therapy should be more than 10 times more effective than raloxifene therapy in preventing CHD, but raloxifene therapy may not induce breast cancer. Women at low risk for hip fracture, CHD, and breast cancer do not benefit significantly from any treatment. Among women at average risk, HRT was preferred unless raloxifene therapy could reduce the risk of breast cancer by at least 66%, compared with a 47% increase for HRT. Women at high risk for CHD benefit most from HRT; women at high risk for breast cancer but low risk for CHD benefit most from raloxifene therapy, but only if it lowers the risk of breast cancer. CONCLUSION: Because of significant differences in the impact of these drugs, treatment choice depends on an individual woman's risk for hip fracture, CHD, and breast cancer.

Cost-effectiveness of therapies for patients with nonvalvular atrial fibrillation.

BACKGROUND: The most appropriate treatment(s) for patients with atrial fibrillation remains uncertain. OBJECTIVE: To examine the cost-effectiveness of anti-thrombotic and antiarrhythmic treatment strategies for atrial fibrillation. METHODS: We performed decision and cost-effectiveness analyses using a Markov state transition model. We gathered data from the English-language literature using MEDLINE searches and bibliographies from selected articles. We obtained financial data from nationwide physician-fee references, a medical center's cost accounting system, and one of New England's larger managed care organizations. We examined strategies that included combinations of cardioversion, antiarrhythmic therapy with quinidine, sotalol hydrochloride, or amiodarone, and anticoagulant or antiplatelet therapy. RESULTS: For a 65-year-old man with nonvalvular atrial fibrillation, any intervention results in a significant gain in quality-adjusted life years (QALYs) compared with no specific therapy. Use of aspirin results in the largest incremental gain (1.2 QALYs). Cardioversion followed by the use of amiodarone and warfarin together is the most effective strategy, yielding a gain of 2.3 QALYs compared with no specific therapy. The marginal cost-effectiveness ratios of cardioversion followed by aspirin, with or without amiodarone, are $33800 per QALY and $10800 per QALY, respectively. Cardioversion followed by amiodarone and warfarin has a marginal cost-effectiveness ratio of $92400 per QALY compared with amiodarone and aspirin. Strategies that include cardioversion followed by either quinidine or sotalol are both more expensive and less effective than competing strategies. CONCLUSIONS: Cardioversion of patients with nonvalvular atrial fibrillation followed by the use of aspirin alone or with amiodarone has a reasonable marginal cost-effectiveness ratio. While cardioversion followed by the use of amiodarone and warfarin results in the greatest gain in quality-adjusted life expectancy, it is expensive (ie, has a high marginal cost-effectiveness ratio) compared with aspirin and amiodarone. Finally, for patients who are bothered little by symptoms of atrial fibrillation, cardioversion followed by either aspirin or warfarin without subsequent antiarrhythmic therapy is the treatment of choice.