آیا بیمار خودمدیریتی می تواند شیب سلامت را توضیح دهد؟ خودمدیریتی گلدمن و اسمیت می تواند کمکی به توضیح شیب سلامت SES کند؟"

In their much-cited paper, “Can patient self-management help explain the SES health gradient?” Goldman and Smith (2002) use samples of diabetic and HIV+ patients in the United States to conclude that disease self-management is an important explanation for the much-documented positive gradient in education and health outcomes. In this paper, I revisit their analysis and point to some fundamental difficulties in interpreting their results as conclusive evidence in favor of self-management. I also argue that for individuals for whom self-management might be expected to matter –i.e. populations of patients managing complex conditions – economic factors such as resource availability and insurance access might be a more important mechanism behind the gradient than medical compliance. The impact of self-management, though it might matter, is likely to be small.

There is a large literature documenting the positive association between education and health outcomes, known as the education-health gradient. The association holds across countries and over time, and it holds for different measures of health outcomes, such as mortality, morbidity, physical functioning, health behaviors and self-reported health (Cutler and Lleras-Muney, 2007, Cutler and Lleras-Muney, 2008 and Grossman, 2006). The continuing existence of the gradient indicates that education-driven health inequalities persist even as societies become healthier and more educated overall (Cutler & Lleras-Muney, 2008). If policy is to effectively alleviate such inequalities, we must first understand the different mechanisms driving the gradient and their relative importance. A large body of work has devoted itself to this task (Grossman, 2006 and Mirowsky and Ross, 2003). One group of theories focuses on the role of economic and labor market-related factors in driving the gradient (Cutler and Lleras-Muney, 2008 and Mirowsky and Ross, 2003). Higher education facilitates full-time employment and the obtainment of better jobs (Pencavel, 1991 and Ross and Wu, 1995). Better jobs have higher wages, leading to higher income, prevention of material deprivation and better access to quality healthcare (Andrulis, 1998, Card, 2001, Fiscella et al., 2000, Marmot, 2002 and Pencavel, 1991). Better jobs also typically entail better working conditions and more opportunities for productive self-expression (Brown, 1980, Kohn, 1976 and Lazear and Oyer, 2007). The availability of resources can alleviate the stress associated with economic hardship (Fremont and Bird, 2000 and Ross and Huber, 1985). Finally, economic stability can help to ease time and money constraints which impede the adoption of healthy practices (Cutler & Lleras-Muney, 2007). All of these factors have a positive effect on health (Bindman et al., 1995, Karasek et al., 1988, Ross and Mirowsky, 1995, Ross and Van Willigen, 1997 and Zhang et al., 2008). Another set of theories invokes behavioral factors to explain the gradient. Social scientists have proposed that ‘learned effectiveness’ plays an important role in motivating behaviors that lead to healthier outcomes (Mirowsky & Ross, 2003). One aspect of learned effectiveness is cognitive: the ability to seek and use information to successfully achieve one's goals. Another aspect, such as ‘personal control’, is non-cognitive: the belief that personal actions are responsible for outcomes, and the confidence in one's ability to affect the same. Education enhances both cognitive and non-cognitive aspects of learned effectiveness by imparting specific knowledge and skills such as critical thinking (Cutler and Lleras-Muney, 2007, de Walque, 2004, de Walque, 2005, Lleras-Muney and Lichtenberg, 2002, Rosenzweig and Schultz, 1989, Spandorfer et al., 1995 and Williams et al., 1998), and by increasing exposure to and confidence in problem solving thereby generating a sense of control (Fremont and Bird, 2000, Mirowsky and Ross, 2003, Ross and Wu, 1995, Seeman and Seeman, 1983 and Seeman et al., 1988). Both the economic and behavioral mechanisms described above assume that good health is a universally desirable goal. But economists have argued that personal goals are themselves determined by individual preferences (Cutler and Lleras-Muney, 2007 and Cutler and Lleras-Muney, 2008). Education could alter these preferences. Specifically, education could make individuals more averse to risk; it could also make individuals value the future more or lower their discount rates (Becker and Mulligan, 1997, Fuchs, 1982 and Leigh, 1990). This could lead to greater investment in health, and hence healthier outcomes. Early-childhood or intrauterine environment has been suggested as another individual-specific factor that could be responsible for health outcomes (Barker, 1995, Barker, 1997, Case et al., 2005 and Ravelli et al., 1998). Yet another group of theories suggests that education affects health outcomes through its social impact. Education enables access to networks which provide financial, psychological and emotional support (Berkman, 1995, Berkman and Syme, 1979 and Cutler and Lleras-Muney, 2008). Societal rank and peer effects may also influence health outcomes (Cutler and Glaeser, 2007 and Marmot, 2002). These explanations are not mutually exclusive. For example, economic factors (e.g. job characteristics) or social factors (e.g. social rank) could drive the gradient by enhancing personal control (Ross & Mirowsky, 1992) or a related concept, self-efficacy (Berkman, 1995 and Mirowsky and Ross, 2003). These and other linkages have been extensively investigated, but even so, much of the education-health gradient remains unaccounted for (Cutler and Lleras-Muney, 2007, Cutler and Lleras-Muney, 2008, Kenkel, 1991, Lahelma et al., 2004, Meara, 2001 and Mirowsky and Ross, 2003). In a much-cited paper, Goldman and Smith (2002) argue that education impacts health via its effect on disease management. The argument is based on the cognitive advantage hypothesis of social scientists (Mirowsky & Ross, 2003). Education enhances cognitive skills and is hence a good proxy for the ability to comprehend and execute complex treatment regimens. Education is also a marker for the ability to internalize the future outcomes of poor behavior. Education thus leads to better self-management behavior, and hence to better health outcomes. Goldman and Smith test their hypothesis for patients from different samples suffering from two chronic conditions – insulin-dependent diabetes and HIV – known for the complexity of their treatment regimens. Treatments for these conditions are potentially efficacious, but often involve intricate combinations of oral and injected medications, constant monitoring, and meticulous coordination. As such, careful disease management is vital to better health outcomes. In this paper, I revisit Goldman and Smith's (2002) analysis of diabetics from the Health and Retirement Study (HRS), and highlight several difficulties in interpreting their results as evidence that self-management drives the gradient. Furthermore, I argue that in patients for whom self-management of conditions might matter – viz. individuals with chronic conditions requiring complex treatments – economic factors such as resources and access to healthcare are likely to matter more. While cognitive ability is an advantage for successful disease self-management, it is not sufficient; a necessary condition for such management to be effective is the affordability of treatment. The availability of resources ensures that long-term complex treatments remain affordable; it also gives individuals the potential to bypass the need for self-management by hiring caregivers. Results using Goldman and Smith's sample of HIV+ patients are not presented here, but point to the same qualitative conclusions, both regarding the flaws in the argument and the relative importance of economic factors. Specifically, I find that resource availability and insurance access are the two most important individual factors explaining the gradient in the HRS sample of diabetics. Controlling for these factors reduces the gradient by 34% and 27%, respectively. Cognitive ability is the third most important factor, although the magnitude of its impact (11%) is only a third that of economic resources. Other factors that explain the gradient, albeit by very little, are health behaviors (smoking, drinking, exercise) and risk preferences. Considered jointly, all of these factors explain close to 60% of the gradient in education and health. This leaves 40% of the gradient to be explained by mechanisms beyond those considered here. Some of these mechanisms may be even more important than economic factors; future research must identify these. However, my results strongly suggest that economic factors may matter more for explaining the education-health gradient than disease self-management, in populations where the latter might be most important.