Housing Conditions Linked to African Americans' Diabetes Risk

ST. LOUIS, Aug. 14 -- Poor housing conditions may contribute to African Americans' risk of developing diabetes, though how is unclear, researchers said.

Each of several adverse housing factors was significantly associated with the likelihood that middle-age African Americans would develop diabetes, found Mario Schootman, Ph.D., of Washington University here, and colleagues in a population-based study.

But, the investigators came up with no explanation for the effect, which ranged from 78% to 253% per adverse housing factor. None of the factors substantially attenuated the associations, the investigators reported in the Aug. 15 issue of the American Journal of Epidemiology.

"No one believes that the status of one's home is a causative factor in the development of diabetes," commented Richard W. Grant, M.D., M.P.H., of Harvard and the Massachusetts General Hospital, in an accompanying editorial. Rather, more work is needed to discover the underlying factor, Dr. Grant.

"Given that the disparities in disease severity between African Americans and other U.S. racial/ethnic groups begin early and worsen over time, it is imperative to search for the fundamental cause of these differences during the prediabetic period," he said.

Previous research into social and behavioral factors in disparities has often cited the role of the physical environment but not looked at conditions specifically, the researchers said.

To see whether there would be a link with diabetes occurrence, they analyzed data from 644 middle-age African Americans without baseline diabetes in the larger African-American Health Study.

It included a population-based cohort living in either a poor, inner-city area or suburbs of St. Louis and interviewed at home at baseline and three years later. Participants rated their neighborhood as a place to live, their feelings about and attachment to it, and safety.

The investigators rated the external appearance of each participant's neighborhood on the basis of the condition of houses, noise and air pollution, and condition of streets, yards, and sidewalks in front of the participant's home. They also assessed housing conditions during the interview on five factors: cleanliness inside, physical condition inside, condition of furnishings inside, condition of the outside of the building, and overall condition of the dwelling.

About a quarter of the respondents rated fair or poor for the individual measures of neighborhood and housing conditions.

At the three-year follow-up, 65 respondents (10.3%) reported having developed diabetes.

The researchers found no association of any of "objectively" evaluated or perceived neighborhood conditions with incident diabetes.

But, there was a significant association between researcher-rated housing conditions and incident diabetes.

A poor-to-fair rating predicted significantly higher risk of developing diabetes during the study compared with a good-to-excellent rating for all the measures ranging from odds ratios of 1.78 for cleanliness inside the building to 2.53 for physical condition inside the building in unadjusted analysis.

Adjusting for potential confounders had little effect on the odds ratios overall, "suggesting that housing conditions may produce the observed effect by another untested pathway," Dr. Schootman and colleagues wrote. The odds ratio ranges in multivariate analyses were:

• 1.85 to 2.65 when controlling for physical activity, smoking, and risk of alcohol abuse.
• 1.75 to 2.55 when controlling for social support and depression.
• 1.78 to 2.56 when controlling for health factors, including self-rated health, number of severe chronic conditions, lower-body functional limitation, body mass index, and hypertension.
• 1.69 to 2.46 when controlling for healthcare access factors, including health insurance, hospitalization, and self-report of not receiving medical care because of cost.
• 2.08 to 3.00 when controlling for age, sex, income, perceived adequacy of income, education, marital status, employment, length of time at present address, home ownership, and area.
• 1.63 to 2.35 when adjusting for diabetes propensity based on age, body mass index, hypertension, self-rated health, number of severe chronic conditions, and lower-body functional limitations.

The researchers noted that their study did not examine diet or pollution in the physical environment, whether indoor or outdoor. Furthermore, their analysis of only African Americans in a single urban area and of a restricted age range limit the generalizabilty of the findings, they noted.

"Alternative explanations need to be explored," they concluded.

This may "require greater interdisciplinary collaboration and exchange of ideas between researchers from very different backgrounds: geneticists with social epidemiologists, physiologists with health services researchers, economists with nutritionists," Dr. Grant added.