Masaquel Anthony, Wells Kenneth, Ettner Susan L
Department of Health Services, UCLA School of Public Health, Los Angeles, CA, USA.
J Ment Health Policy Econ. 2007 Sep;10(3):133-44.
To determine the structural effect of the persistence of depression on continuity and type of health insurance and coverage limits on mental health therapy.
Data came from the Partners in Care study (PIC), a randomized controlled trial examining the effect of quality improvement (QI) programs involving medication or psychotherapy on the outcomes of initially depressed patients in seven managed care settings. The sample included approximately 945 adult patients under the age of 63 years who were primarily depressed and insured at baseline. Single-equation multivariate probit regressions were estimated to determine the association of depression burden days aggregated over the 6 to 24-month period post-baseline with the following dichotomous outcomes: continuous health insurance over 6 to 24 months; continuous private health insurance over 6 to 24 months; any public health insurance over 6 to 24 months; and reporting no insurance limits on mental health therapy coverage at 24 months. Other control variables included baseline insurance status, age, sex, race, marital status, education, income, assets, fixed site effects, and (in sensitivity analyses) number of medical comorbidities, alcohol use and drug use. To address the possibility of endogeneity bias in the relationship between depression and insurance, consistent estimates were derived from instrumental variables (IV) probit regressions and the endogeneity of depression burden days was tested. Potential instruments included the random assignment to intervention and control groups in the PIC study, type of depression at baseline, and baseline Mental Component Summary (MCS) score from the Short Form-12 (SF-12). In sensitivity analyses, data pooled (rather than aggregated) across waves were used to estimate probit and IV probit regressions, using Generalized Estimating Equations methods to adjust for within-person correlation of the error terms.
Evidence was found that depression burden days were exogenous to all of the health insurance outcomes except for coverage limits on mental health therapy. Based on the appropriate estimates (single-equation if exogenous, IV if endogenous), depression burden days appeared to increase the probability of having any public health insurance coverage and decrease the probability of having no coverage limits on mental health therapy. However, these effects were small in magnitude.
Reverse causality may be more of a concern when examining the influence of depression on mental health care coverage than on health insurance in general. Consistent with the government's historical role in financing mental health services, patients whose depression persisted to a greater extent were slightly more likely to have some public health insurance during an 18-month follow-up period. Furthermore, they were slightly more likely to have limits on mental health therapy coverage, suggesting that insurers may be more likely to control access at the level of the benefits structure than at the level of insurance coverage per se. Future analyses should examine the mediating factors in the relationship between depression and limits on mental health therapy coverage, e.g., diminished employment opportunities with large companies that offer more generous benefits.
确定抑郁症持续存在对医疗保险的连续性、类型以及心理健康治疗覆盖范围限制的结构影响。
数据来自“关爱伙伴研究”（PIC），这是一项随机对照试验，研究涉及药物治疗或心理治疗的质量改进（QI）项目对七个管理式医疗环境中初发抑郁症患者结局的影响。样本包括约945名63岁以下的成年患者，他们在基线时主要患有抑郁症且有保险。估计单方程多元概率回归，以确定基线后6至24个月内汇总的抑郁症负担天数与以下二分结局之间的关联：6至24个月的连续医疗保险；6至24个月的连续私人医疗保险；6至24个月的任何公共医疗保险；以及在24个月时报告心理健康治疗覆盖无保险限制。其他控制变量包括基线保险状况、年龄、性别、种族、婚姻状况、教育程度、收入、资产、固定地点效应，以及（在敏感性分析中）医疗合并症数量、酒精使用和药物使用情况。为解决抑郁症与保险之间关系中可能存在的内生性偏差问题，从工具变量（IV）概率回归中得出一致估计，并对抑郁症负担天数的内生性进行了检验。潜在工具变量包括PIC研究中干预组和对照组的随机分配、基线时的抑郁症类型，以及简式12项健康调查（SF - 12）的基线心理成分总结（MCS）得分。在敏感性分析中，使用跨波合并（而非汇总）的数据来估计概率回归和IV概率回归，采用广义估计方程方法来调整误差项的个体内相关性。
发现除心理健康治疗覆盖范围限制外，抑郁症负担天数对于所有医疗保险结局而言是外生的。根据适当的估计（如果是外生的则用单方程，如果是内生的则用IV），抑郁症负担天数似乎增加了拥有任何公共医疗保险覆盖的概率，并降低了心理健康治疗覆盖无限制的概率。然而，这些影响的幅度较小。
在研究抑郁症对心理健康护理覆盖范围的影响时，反向因果关系可能比研究其对一般医疗保险的影响更值得关注。与政府在为心理健康服务提供资金方面的历史作用一致，在18个月的随访期内，抑郁症持续程度较高的患者更有可能拥有某种公共医疗保险。此外，他们更有可能面临心理健康治疗覆盖范围的限制，这表明保险公司可能更倾向于在福利结构层面而非保险覆盖本身层面控制获取机会。未来的分析应研究抑郁症与心理健康治疗覆盖范围限制之间关系的中介因素，例如在提供更优厚福利的大公司就业机会减少的情况。
