Disparities in defining disparities: Statistical conceptual frameworks

Department of Psychiatry, Columbia University, New York, NY 10032, USA.
Statistics in Medicine (Impact Factor: 1.83). 09/2008; 27(20):3941-56. DOI: 10.1002/sim.3283
Source: PubMed

ABSTRACT Motivated by the need to meaningfully implement the Institute of Medicine's (IOM's) definition of health care disparity, this paper proposes statistical frameworks that lay out explicitly the needed causal assumptions for defining disparity measures. Our key emphasis is that a scientifically defensible disparity measure must take into account the direction of the causal relationship between allowable covariates that are not considered to be contributors to disparity and non-allowable covariates that are considered to be contributors to disparity, to avoid flawed disparity measures based on implausible populations that are not relevant for clinical or policy decisions. However, these causal relationships are usually unknown and undetectable from observed data. Consequently, we must make strong causal assumptions in order to proceed. Two frameworks are proposed in this paper, one is the conditional disparity framework under the assumption that allowable covariates impact non-allowable covariates but not vice versa. The other is the marginal disparity framework under the assumption that non-allowable covariates impact allowable ones but not vice versa. We establish theoretical conditions under which the two disparity measures are the same and present a theoretical example showing that the difference between the two disparity measures can be arbitrarily large. Using data from the Collaborative Psychiatric Epidemiology Survey, we also provide an example where the conditional disparity is misled by Simpson's paradox, whereas the marginal disparity approach handles it correctly.

Download full-text


Available from: Margarita Alegria, Oct 10, 2014
27 Reads
  • Source
    • "Our approach is to leave the first factor and the third factor untouched but adjust the second factor from that of the minority group to that of the reference group (non-Latino white). The full rationale of this methodology is described in Duan et al. (2008). Intuitively, we leave the first factor untouched because if the impact of the covariates (either allowable or non-allowable) on the use of service is different between the minority group and for the non-Latino whites, then the issue of disparity is far more complex than merely the differences in the distributions of non-allowable covariates (e.g., income level). "
    [Show abstract] [Hide abstract]
    ABSTRACT: To estimate whether racial/ethnic behavioral health service disparities are likely to be reduced through insurance expansion coverage expected through the Affordable Health Care Act. Pooled data from the nationally representative NIMH Collaborative Psychiatric Epidemiological Studies (2001-2003). We employ a novel reweighting method to estimate service disparities in the presence and absence of insurance coverage. Access to care was assessed by whether any behavioral health treatment was received in the past year. Need was determined by presence of prior year psychiatric disorder, psychiatric diagnoses, physical comorbidities, gender, and age. Improving patient education and availability of community clinics, combined with insurance coverage reduces service disparities across racial/ethnic groups.However, even with expanded insurance coverage, approximately 10 percent fewer African Americans with need for behavioral health services are likely to receive services compared to non-Latino whites while Latinos show no measurable disparity. Expansion of insurance coverage might have different effects for racial/ethnic groups, requiring additional interventions to reduce disparities for all groups.
    Health Services Research 06/2012; 47(3 Pt 2):1322-44. DOI:10.1111/j.1475-6773.2012.01403.x · 2.78 Impact Factor
  • Source
    • " terms for each retained predictor were similarly introduced and removed. From the final models we calculated white versus black adjusted odds ratios (AORs) for significant interactions to examine moderation of race effects. We used published equations to calculate disparity ratios with 95% CIs for access and use from main and interaction effects.(Duan et al., 2008) Probability estimates for whites used the covariate distribution of the white sample; estimates for blacks used the non-allowable covariate distribution of the black sample and allowable covariate distribution of the whites."
    [Show abstract] [Hide abstract]
    ABSTRACT: Black Americans with depression were less likely to receive electroconvulsive therapy (ECT) than whites during the 1970s and 80s. This pattern was commonly attributed to treatment of blacks in lower quality hospitals where ECT was unavailable. We investigated whether a racial difference in receiving ECT persists, and, if so, whether it arises from lesser ECT availability or from lesser ECT use within hospitals conducting the procedure. Black or white inpatient stays for recurrent major depression from 1993 to 2007 (N=419,686) were drawn from an annual sample of US community hospital discharges. The marginal disparity ratio estimated adjusted racial differences in the probabilities of (1) admission to a hospital capable of conducting ECT (availability), and (2) ECT utilization if treated where ECT is conducted (use). Across all hospitals, the probability of receiving ECT for depressed white inpatients (7.0%) greatly exceeded that for blacks (2.0%). Probability of ECT availability was slightly greater for whites than blacks (62.0% versus 57.8%), while probability of use was markedly greater (11.8% versus 3.9%). The white versus black marginal disparity ratio for ECT availability was 1.07 (95% confidence interval 1.06-1.07) and stable over the period, while the ratio for use fell from 3.2 (3.1-3.4) to 2.5 (2.4-2.7). Depressed persons treated in outpatient settings or receive no care are excluded from analyses. Depressed black inpatients continue to be far less likely than whites to receive ECT. The difference arises almost entirely from lesser use of ECT within hospitals where it is available.
    Journal of Affective Disorders 12/2011; 136(3):359-65. DOI:10.1016/j.jad.2011.11.026 · 3.38 Impact Factor
  • Source
Show more