Showing papers by "Stephen G. Pauker published in 1999"

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

Abstract: 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.

Pretreatment evaluation of chronic hepatitis C - Risks, benefits, and costs

Abstract: Context.—Chronic hepatitis C (CHC) infection affects nearly 4 million people in the United States. Treatment with interferon alfa-2b has been limited by its cost and low likelihood of long-term response.Objective.—To examine the cost-effectiveness of alternative pretreatment management strategies for patients with CHC.Design.—Decision and cost-effectiveness analysis using a Markov model to examine prevalence of genotypes, viral load, and histological characteristics in relation to the sustained response rate with treatment. Data were based on a previously published decision model and a MEDLINE literature search for hepatitisC, biopsy, and liver from 1966 to 1996.Patients.—A hypothetical population of patients with CHC infection and elevated serum alanine aminotransferase level.Interventions.—Combinations of liver biopsy, genotyping, and quantitative viral load determination prior to a single 6-month course of interferon alfa-2b; empirical interferon treatment; and conservative management.Main Outcome Measures.—Proportion of sustained responders, lifetime costs, life expectancy, and quality-adjusted life expectancy.Results.—Strategies involving hepatitis C virus (HCV) RNA testing had marginal cost-effectiveness ratios up to $4400 per discounted quality-adjusted life-year gained but would miss up to 36% of sustained responders. Empirical interferon treatment had a marginal cost-effectiveness ratio of $12,400 per discounted quality-adjusted life-year gained and reached all potential sustained responders. Strategies involving liver biopsy were more expensive and would miss 6% of sustained responders and yield slightly lower life expectancies.Conclusions.—Routine liver biopsy before treatment with interferon increases the cost of managing patients with CHC without improving health outcomes. Using quantitative HCV RNA testing to guide therapy misses some potential sustained responders. Empirical interferon treatment has a marginal cost-effectiveness ratio within the bounds of other commonly accepted therapies and misses none of the sustained responders.

A Little Math Helps the Medicine Go Down

Abstract: Stage A 44-year-old, previously healthy man presented with a three-week history of fatigue and temperatures as high as 38.2°C (100.8°F). The man was a medical resident at a municipal hospital. Response Despite the long list of causes of fever and fatigue, I would be most worried about occupationally related illnesses, such as primary human immunodeficiency virus (HIV) infection and hepatitis in its prodromal phase. Stage The fevers usually occurred at the end of the day and were associated with myalgias and chills. The patient began to take acetaminophen. He was seen at a local emergency room for epigastric pain, where . . .

Does DEALE-ing stack the deck?

Abstract: Almost two decades ago, we proposed a simple approximation to life expectancy that could be used both at the bedside and to understand the effects of competing forces of mortality in clinical problems: the declining exponential approximation of life expectancy, or DEALE.1 At that time, software for stochastic modeling was virtually unavailable; our \"gold standard\" was a hand-run Markov model using a programmable calculator. Furthermore, in the early 1980s the concept of life expectancy or average survival was poorly understood as a measure of clinical outcome and as a metric for comparing alternative strategies. Over the ensuing years, life expectancy has achieved a prominent role both in our clinical vocabulary and in the formal process of medical decision making. At the same time, software has become available to estimate survival in relatively complex Markov models, while the DEALE formalism has been expanded to handle discounting,' multiple periods,3 and different functional forms.4 In this issue of Medical Decision Making, Holland and colleagues describe their comparison of survival estimates using the DEALE with a new, complex approach-the nested Markov model.5 Their method is not easily accomplished, even now, using available desktop decision-analysis software, and therefore the investigators built a custom set of models in Microsoft Access. Their nested Markov approach is conceptually similar to that of Weinstein's Coronary Heart Disease Policy Model, first reported a decade ago.6 Not surprisingly, the nested Markov model proves to be more accurate than constant hazard rates, although its adoption as an uncontested \"gold standard\" might be questioned. The DEALE was designed to be simple and logical and to demonstrate the effects of competing mortality. It appears to make three basic assumptions, each so simplistic that it is hard to believe after 18 years that it ever had any rational basis. First, it seems to assume that the disease-specific and comorbidity-specific mortality rates are constant. Second, it treats the demographic-associated mortality rate (as specified by patient age, sex, and race) as constant, even though the original paper acknowledges that this is wrong.1 Third, it considers no interactions among the demographic-associated mortality rate and the diseaseand comorbidity-specific mortality rates. In fact, the DEALE makes no such assumptions. Rather, it assumes that each mortality rate can be replaced by a functionally equivalent average mortality rate, in order to reduce the calculation of overall life expectancy to simple algebra. Over time, as the DEALE came to be used in ways that violated its original design and validation, many inconsistencies and problems became evident. A constant, average demographic-specific mortality rate was too high in the early periods of the model and too low in the later periods. Many diseases had several phases, sometimes with higher mortalities in the early years (until selection weeded out less hardy patients) and sometimes with higher mortality in later years (when more advanced disease had a higher force of mortality). These observations led to slightly more complex approximations-the two-part DEALE and then the multi-part DEALE to allow for such changes in rates. Other methodologists characterized the types of bias that might result from using constant mortality forces.7 Given the inaccuracies with the DEALE approximation, one might ask, \"Why bother?\" \"Why not do it 'right' with a complex stochastic model?\" Perhaps the answer lies in the ability to apply the DEALE quickly at the bedside. Perhaps the answer lies in the ability of the DEALE to account for patients with more than one disease, in contrast to the nested Markov model used by Holland and colleagues, which does not model the clinical complexity of competing mortality. Perhaps the value of the DEALE is that it offers the clinician and the student of medicine an intuitive grasp of the effects of competing mortality rates and of treatments that affect only one of those competing forces. Perhaps the benefit of the DEALE is its availability to anyone with a hand-held calculator, rather than to a decision scientist with special-purpose software. Of course, Holland and his colleagues are correct in their analysis. The DEALE is simplistic and is especially problematic when time horizons are long

Finding what you seek: analyzing therapies for nonvalvular atrial fibrillation.

Abstract: Over the past decade, the management of patients with nonvalvular atrial fibrillation has been the subject of more papers than any clinician has time to digest. In this issue, Catherwood and collea...