Seven percent of the American population has diabetes.3 Research shows that patients with schizophrenia are at a higher risk of developing diabetes. Rates of type 2 diabetes mellitus in patients with schizophrenia are two to four times that in the general population.4–7 In a recent report on the New York State Office of Mental Health psychiatric hospital system, the prevalence of patients with diabetes doubled from 6.9% of 10,091 patients in 1997 to 14.5% of 7,420 patients in 2004 irrespective of antipsychotic use.8
Theories About the Schizophrenia-Diabetes Link
Several theories have been suggested to account for the apparent connection between diabetes and schizophrenia. One is the potential genetic link between the heritability of diabetes and the heritability of schizophrenia. Patients with schizophrenia have a higher rate of family history of diabetes than the general population. Studies have reported that 17 to 50% of people with schizophrenia have a family history of type 2 diabetes,9,10 and a positive family history of diabetes may increase the risk of developing diabetes in individuals with schizophrenia and other serious and persistent mental illnesses up to threefold.11
We have known, even before the availability of antipsychotics, that patients with schizophrenia were prone to develop hyperglycemia. A link between schizophrenia and diabetes was postulated in the 1920s,12 and a temporal association between first-generation antipsychotic drug treatment and hyperglycemia was first reported in the 1950s.13
Patients with schizophrenia often are disadvantaged economically, so they may not be able to afford healthier eating habits. These patients also may not be aware of the risks for diabetes or have access to adequate primary care services.
Large-scale epidemiologic studies14,15 have spawned an increased interest in an association between second-generation antipsychotics and diabetes. Clinicians have begun to be more concerned about diabetes in patients with schizophrenia, because it has been suggested that the second- generation antipsychotics we are using today are putting our patients at a higher risk for developing diabetes.
Insulin resistance can take more than a decade to manifest as diabetes. Someone destined to become diabetic often has alterations in his or her ability to handle glucose that have existed for years—alterations that may remain asymptomatic in patients who remain unaware of having this problem until they eventually have their blood sugar levels checked. Thus, a patient with schizophrenia may have diabetes or prediabetes even before treatment with antipsychotics. Because of this and many other factors, the potential contribution of antipsychotic treatment toward the risk for developing diabetes mellitus is unclear.14,15
The Different Ways to Define “Risk”
Different study designs have been used to estimate the contribution of antipsychotic treatment toward the risk for developing diabetes in a patient with schizophrenia. All of the large epidemiologic studies have been retrospective analyses of large claims databases. These studies offer the advantage of large numbers and longitudinal data. Before discussing study design, I will discuss the types of risk these studies try to estimate.
The odds ratio is the odds that a patient with a disease was exposed to a risk factor versus the odds that a control (unaffected by the disease) was exposed to the same risk factor. The relative risk ratio represents the increased (or decreased) likelihood of developing a disease on exposure to one risk factor compared with exposure to something else other than that risk factor.15 Odds ratios and relative risk ratios help us compare the potential risk of developing a disease in patients treated with one treatment versus another, but by themselves they don't place the results into a clinically meaningful context and often are misunderstood. Odds and relative risk ratios do not tell us if the event rarely occurs overall, or if the absolute difference in risk for patients treated with different agents is small.16
Let's go over an example to highlight the problem of just looking at relative risk. Let's say the incidence of green toes occurring in patients on drug A is 0.1%, and for patients treated with drug B, 0.13%. Effectively this means that for every 10,000 patients treated with drug A, we would expect 10 to develop green toes; for drug B we would expect 13. The relative risk ratio is the incidence of drug B divided by the incidence of drug A (0.13/0.10 = 1.3). This relative risk might be interpreted as “Drug B is 30% more risky for the development of green toes than drug A.” This is a provocative statement because it is out of clinical context. The extra number of cases we would expect (the attributable risk in patients treated with drug B instead of drug A) is 3 per 10,000 patients treated (see table, p. 30). To boil it down, relative risk helps us understand the comparative likelihood that something will occur between two groups of exposures. However, too often, people read “30% more risk” as “30% more of the total number treated” rather than “30% more of the number who developed the outcome.”