The effects of US state income inequality and alcohol policies on symptoms of depression and alcohol dependence
Introduction
The recent resurgence in interest in social epidemiology (Link & Phelan, 2000) has included the identification of contextual variables that emerge only at the level of the group. Such variables are aspects of the environmental context of health, which is omitted if only individual-level data is measured (Susser, 1994). For example, the notion that inequality in the distribution of income may be related to health has recently attracted much attention. Wilkinson's (1992) original report of an association between income inequality and mortality in developed countries has since been replicated for states and metropolitan areas within the US (Lynch et al., 1998; Kaplan, Pamuk, Lynch, Cohen, & Balfour, 1996; Kennedy, Kawachi, & Prothrow-Stith, 1996). These associations, however, do not appear to be as strong in some contexts, as demonstrated by the weaker relationship between income inequality and mortality observed across Canadian provinces (Ross et al., 2000). Further work has examined the extent to which the relation between income inequality and health persists once individual-level factors such as income are controlled for. The relationship has been found to persist in some (Kennedy, Kawachi, Glass, & Prothrow-Stith, 1998; Kahn, Wise, Kennedy, & Kawachi, 2000; Diez-Roux, Link, & Northridge, 2000; Lochner, Pamuk, Makuc, Kennedy, & Kawachi, 2001) but not all studies (Fiscella & Franks, 1997; Daly, Duncan, Kaplan, & Lynch, 1998; Osler et al., 2002; Shibuya, Hashimoto, & Yano, 2002; Sturm & Gresenz, 2002). There is still substantial ongoing debate regarding the presence and strength of an inequality effect, over and above the well-established effect of individual-level income (Wilkinson, 2002; Lynch & Davey Smith, 2002).
Kennedy et al. (1992), Wilkinson (1997a), Wilkinson (1997b) has suggested that income inequality affects both physiology (for example, due to endocrine changes) and stress-related behaviors that affect health, for example, stress-related smoking, drinking, comfort eating and interpersonal violence. It has also been hypothesized that the negative effects of living in an unequal society may be greater in persons of lower income (Kennedy et al., 1998; Kahn et al., 2000; Diez-Roux et al., 2000). To the extent that stress is causally related to common mental health problems (for example, Kendler et al., 1995), this interpretation implies that those living in areas of greater income inequality will have higher rates of mental health problems, particularly those on lower incomes. One study on maternal mental health (Kahn et al., 2000) was consistent with this hypothesis, while two more, on adults in the US (Sturm & Gresenz, 2002) and the UK (Weich, Lewis, & Jenkins, 2001) did not find the predicted relationship. All three of these studies addressed mental health outcomes involving depression or anxiety, but did not include alcohol, specifically suggested by Wilkinson as being one of the health conditions affected by income inequality.
To date, state-level information based on a nationally representative sample has not been used to address the relationship of income inequality to common mental health problems including both depression and alcohol. State-level data offers a number of advantages in addressing this question. According to Wilkinson, income inequality becomes important only in areas large enough to contain the relevant social stratification, while in small neighborhoods with little income gradient, morbidity and mortality are more closely related to the average income. The choice of US state to define the geographic area thus provides large geographic units with considerable social heterogeneity (Wilkinson, 1997a). State-level analysis also allows incorporation of potential state-level confounders, such as state alcohol policies, that may influence mental health outcomes such as alcohol dependence through their effects on consumption. An example of such an alcohol policy is excise tax on alcohol, relevant because studies consistently show an inverse relationship between alcohol price and alcohol consumption (Toomey & Wagenaar, 1999; Chaloupka, Grossman, & Saffer, 2002). Another such policy includes alcohol distribution methods, which may affect consumption by restricting access (Toomey & Wagenaar, 1999). Because variation in both of these can be studied at the state level, US state represents a geographical unit of analysis that is meaningful in many respects.
The National Longitudinal Alcohol Epidemiologic Survey (NLAES) presents an unusual opportunity to study the relationship of income inequality and state alcohol policy to depression and alcohol dependence at the state level because the sample is very large and the response rate very high (Grant et al., 1994). Further, well-validated and reliable measures of depression and alcohol disorders were used (Grant & Hasin, 1992; Grant, Harford, Dawson, Chou, & Pickering, 1995; Hasin et al., 2003). We therefore assessed the contextual effect of state income inequality on symptoms of depression and alcohol dependence, controlling for potential state- and individual-level confounding variables. The gender distribution of major depression and alcohol dependence are almost mirror opposites of each other (Hanna & Grant, 1997). Therefore, we investigated each of these disorders separately for men and women.
Section snippets
Sources of data
Data from the 1992 NLAES (Grant et al., 1994) were used as the source of individual-level data. The NLAES was sponsored by the National Institute of Alcohol Abuse and Alcoholism, with fieldwork conducted by the US Bureau of the Census. It covers the 48 contiguous US States including the District of Columbia except for Nebraska and North Dakota and has a sample of 42,862. A two-stage design was used to ensure a representative sample of non-institutionalized people aged 18 or over. Oversampling
Results
The total unweighted sample size is 42,862. Unweighted sample size by state varies from 47 in New Hampshire to 4993 in California. Women were slightly overrepresented in the unweighted sample (58.4% women vs. 41.6% men). Table 2 shows the results of simple bivariate analyses for men and women, testing the null hypothesis of no association between a binary outcome and a risk factor, for characteristics of the weighted study sample and outcome prevalences by individual and state characteristics.
Discussion
We found no evidence for a positive association between state income inequality and symptoms of depression or DSM-IV Major Depressive Disorder over the last year in men or women. This is contrary to results previously reported regarding the relationship between US State income inequality and maternal mental and physical health (Kahn et al., 2000). However, it is consistent with Sturm and Gresenz (2002) and with Weich et al. (2001), who found no relationship between symptoms of common mental
Acknowledgements
CH is funded by a Medical Research Council special training fellowship in health services research. BGL received support from the Robert Wood Johnson Health Policy Investigator Award. DH is supported by NIAAA grant KO2 AA00151 and DH and XL are supported by R01 AA08159.
References (57)
- et al.
A multilevel analysis of income inequality and cardiovascular disease risk factors
Social Science & Medicine
(2000) - et al.
The Alcohol Use Disorder and Associated Disabilities Schedule (AUDADIS)Reliability of alcohol and drug modules in a general population sample
Drug and Alcohol Dependence
(1995) - et al.
Gender differences in DSM-IV alcohol use disorders and major depression as distributed in the general populationClinical implications
Comprehensive Psychiatry
(1997) - et al.
Dependence symptoms but no diagnosisDiagnostic ‘orphans’ in a community sample
Drug and Alcohol Dependence
(1998) - et al.
Dependence symptoms but no diagnosisDiagnostic ‘orphans’ in a 1992 national sample
Drug and Alcohol Dependence
(1999) - et al.
The relationship of income inequality to mortalityDoes the choice of indicator matter
Social Science & Medicine
(1997) Psychometric evaluation of diagnostic criteriaApplication to a two-dimensional model of alcohol abuse and dependence
Drug and Alcohol Dependence
(1996)- et al.
Using cluster analysis of alcohol use disorders to investigate ‘diagnostic orphans’Subjects with alcohol dependence symptoms but no diagnosis
Drug and Alcohol Dependence
(2000) - et al.
Aggregation and the measurement of income inequalityEffects on morbidity
Social Science & Medicine
(1999) - Alcohol Epidemiology Program. (2000). Alcohol policies in the United States: Highlights from the 50 states....