Still Looking for Health Outcomes in All the Wrong Places? Misinterpreted Observational Evidence, Medication Adherence Promotion, and Value-Based Insurance Design

Abstract

Above all, the right question needs to be framed."1 To those who would argue that sophisticated statistical modeling obviates the need to consider confounding factors and alternative explanations when interpreting observational associations, Freedman's answer was simple: "The technology is relatively easy to use. … However, the appearance of methodological rigor can be deceptive."1 More contemporary guidelines for research reporting acknowledge the same understanding: when interpreting associations, multivariate modeling is often appropriate and valuable, but is still potentially vulnerable to the effects of confounding and bias and is therefore no substitute for a randomized design. 2,3,4 Lessons Learned From Confusing Association With Causation in Chasing BiomarkersCurtiss and Fairman observed in the July/August 2008 issue of JMCP that "evidence-based" interventions targeted to biomarkers frequently do not produce the end point "outcomes we love," such as reductions in hospitalization rates or mortality.5 Notably, most of the instances cited were characterized by a single common pattern: the usurping of lower-quality evidence based on observational associations with higher-quality evidence garnered from experimental testing of hypothesized causal factors.For example, the Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation (ADVANCE) trial was conducted because observational research had documented associations of blood glucose and HbA1c (A1c) levels with cardiovascular events. Researchers randomized 11,140 patients with type 2 diabetes to standard glucose control or intensive glucose control, targeted to achieve an A1c level of 6.5% or less.6 During a median 5 years of follow-up, the patients randomized to the intensive glucose control intervention did not have a significantly lower risk of major macrovascular events (hazard ratio [HR] = 0.94, 95% confidence interval [CI] = 0.84-1.06), cardiovascular mortality (HR = 0.88, 95% CI = 0.74-1.04), or all-cause mortality (HR = 0.93, 95% CI = 0.83-1.06), but did have an increased risk of severe hypoglycemia (2.7% intensive control vs. 1.5% standard control; HR = 1.86, 95% CI = 1.42-2.40). Similarly, the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial was initiated because of high-quality prospective observational evidence suggesting that after "adjusting for other risk factors," each 1% decrease in A1c was associated with a 21% decrease in the risk of diabetes-related mortality, a 14% decrease Naturally, there is a strong desire to substitute intellectual capital for labor. That is why investigators often try to base causal inference on statistical models.

What Do We Really Know About VBID? Quality of the Evidence and Ethical Considerations for Health Plan Sponsors

“…57 We have also noted that the "case" for the negative effect of typical copayment increases on adherence has been made primarily with research employing observational designs in which association was sometimes erroneously equated with causation; quasiexperimental (pre vs. post with comparison group) studies of typical copayment increases in commercially insured groups have generally found positive outcomes for patients and plan sponsors. 8,[58][59][60][61] Unfortunately, much of the literature on cost sharing has tended to overstate or misstate previous research findings, suggesting that cost sharing had a greater influence on medication adherence than was actually documented in the original work. For example, in discussions of cross-sectional analyses comparing higher versus lower cost sharing levels, one sometimes sees references to "price responsiveness" to " [copayment] changes" when no copayment change was measured 58 or to "adherence" outcomes from studies that included no measures of adherence; 57 and seemingly small but often substantively important discrepancies are common.…”

“…65,66 In contrast to the commentary's portrayal, RHIE authors even argued in 1987 and 1992 that their finding of "enormous potential savings" from cost sharing, with "little apparent health impact on the kind of people who typically are covered under employer health insurance," may have prompted plan sponsors to increase coinsurance and deductibles in the years immediately following the RHIE's publication, resulting in large health care cost savings nationwide. 65,67 New Observational Evidence About VBID Copayment Reduction Since publication of our earlier observations about deficiencies in research on VBID, 8,57 new copayment reduction studies using nonrandomized comparison groups have been published, 29,62,63,[68][69][70][71] and an additional study 31 measured the association between medication adherence and all-cause health What Do We Really Know About VBID? Quality of the Evidence and Ethical Considerations for Health Plan Sponsors not contain the patient subgroup counts necessary to calculate unbiased discontinuation rates for the study groups.…”

Nonadherence, Clinical Inertia, or Therapeutic Inertia?

“…39 While the recent report "Thinking Outside the Pillbox" (August 2009), also trumpeted the now-familiar song, there was recognition of the need for a systemwide approach to improving medication adherence in patients with chronic disease, citing factors such as the patient's inability to navigate the health care system, cognitive impairment, and imperfect drug regimens. 40 We may indeed be stuck with some authors using the term "therapeutic inertia," but we suggest that when this term is used that authors be specific and not attribute failure of patients to meet therapeutic goals solely to clinicians' failure to intensify treatment in a timely manner.…”

Association Between Prescription Cost Sharing and Adherence to Initial Combination Antiretroviral Therapy in Commercially Insured Antiretroviral-Naïve Patients with HIV

“…First, as noted in editorial critiques of studies that have examined the association between cost sharing and adherence, 41,42 this study is limited by its observational (nonrandomized) nature and therefore can only be interpreted as suggesting an association between cART prescription cost sharing and adherence to behavior. Eighth, race, actual income, biometric information, and mortality are unavailable within the data due to privacy protections.…”