13.8.2 Diabetes mellitus and psychotic disease

In 2006, investigators compiled data from the public mental health systems of eight states in the USA and compared life expectancy for patients with a major mental illness with general population values. Focusing on states with outpatient as well as inpatient data, this study indicated that individuals with a major mental illness have a mean age at death that is 25–30 years earlier than that observed in the general population over the same years in the same states (1). In this study, ‘major mental illness’ included affective disorders such as major depression and bipolar disorder, attention deficit/hyperactivity disorders, schizophrenia, and schizoaffective disorders. Importantly, these data indicated that the leading cause of death in the mentally ill is coronary heart disease (CHD) and when death due to stroke or cerebrovascular disease is included in a category of cardiovascular disease (CVD), they account for more than 35% of deaths in this population. Suicide, by contrast, was responsible for fewer than 5% of deaths overall. Such observations have led to growing clinical interest in the cardiovascular and metabolic risk factors that contribute to the major causes of morbidity and mortality in patients with psychotic disease, as exemplified by schizophrenia.

Standardized mortality ratios (SMRs) can be useful for quantifying the mortality risk associated with a condition like schizophrenia, relative to the general population, though caution is warranted in their interpretation. SMRs are calculated as the ratio of observed to expected deaths in a specified sample (sometimes × 100). The ‘expected’ mortality rate is the rate seen in the general population, whereas the ‘observed’ mortality rate is that seen in the population in question. It is a measure of relative, not absolute, risk. SMRs can be useful for identifying factors that contribute to the relative risk of death for patients in comparison to the general population. They are less useful, however, when interest is focused on the absolute risk of death. For example, CVD is responsible for the largest number of absolute deaths in patients with schizophrenia (1, 2). However, as rates of death from CVD are also very large in the general population, the SMR for CVD in schizophrenia is only around 2. This is, by contrast, lower than the SMR of approximately 15.7 for males and 19.7 for females for suicide in this population (2). The SMR for suicide is larger than that for CVD, despite a lower number of absolute deaths, (1) because suicide is rare in the general population. In 2007, Saha and colleagues studied the distribution of all-cause SMRs in schizophrenia patients, including data from 37 studies. This analysis yielded a median all-cause SMR of 2.58 compared to the general population. Increased SMRs were found for most causes of death, and overall SMRs were noted to be increasing over the past three decades, suggesting that the gap between mortality rates observed in schizophrenia and rates in the general population is growing wider (3).

In a population-based sample of 7,784 inpatients with schizophrenia analysed by Osby and colleagues, male and female subjects had an SMR of 2.8 and 2.4, respectively, for death due to any cause. All natural causes of death except cancer in men, and nervous system diseases in women, were significantly elevated. Overall, patients were more than twice as likely to die from cardiovascular, cerebrovascular, or respiratory disease (2). In another study by Hansen and colleagues, exploring mortality in 1,998 deinstitutionalized individuals with schizophrenia, all-cause mortality risk in patients was 3.2 times higher in men and 2.4 times higher in women versus the general population (4). The higher SMRs in this outpatient study in comparison to some inpatient samples may be attributable to the study’s very high follow-up rate, which led to the detection of many deaths which may otherwise have been missed, as well as the general potential for outpatients to be even less likely than inpatients to receive appropriate primary or secondary prevention.

Register-based studies comparing mortality in patients with schizophrenia versus the general population over a specified period of years have contributed to our understanding of the clinical importance of medical co-morbidities to long-term health outcomes for patients with this disorder (5–7). One registry-based study by Brown and colleagues of 370 schizophrenia patients over 13 years confirmed that most patients died from natural causes rather than unnatural causes (such as suicide or accidents); in this study, 73% of patients died as a result of medical diseases. The all-cause mortality SMR was 2.98 in the overall study population, with disease-specific SMRs of 1.46 for cancer, 2.08 for lung cancer, 3.19 for respiratory diseases, 26.13 for epilepsy, 2.49 for circulatory diseases, 5.34 for cerebrovascular disease, 1.87 for CVD, 6.14 for diseases of the nervous system, 11.66 for endocrine diseases in general, and 9.96 for diabetes mellitus (5).

During the past several decades, CVD mortality has markedly declined in the USA, from more than 50% to approximately 36% of the underlying cause of deaths, but this improvement in public heath outcome has not extended to patients with schizophrenia in the USA or other developed countries (8). CVD is the leading cause of mortality in individuals with schizophrenia, who are even more likely to experience premature cardiovascular mortality than individuals in the general population (1, 9–11). In a retrospective cohort study conducted by Curkendall and colleagues, 3,022 individuals with schizophrenia were compared to a general population cohort. The overall prevalence of CVD was increased in the patients with schizophrenia (10.6% vs 8.7%), as was the incidence of ventricular arrhythmia (odds ratio (OR) 2.3, 95% CI 1.2 to 4.3), stroke (OR 1.5, 95% CI 1.2 to 2.0), diabetes (OR 1.8, 95% CI 1.2 to 2.6), and heart failure (OR 1.6, 95% CI 1.2 to 2.0) (9). Cardiovascular mortality in schizophrenia has also been evaluated in large population-based samples using long periods of observation. Patients with schizophrenia were evaluated for mortality risk over a period of 19 years in a Swedish registry study, analysed by Osby and colleagues. Between 1976 and 1991, death rates due to CVD increased 4.7-fold in men and 2.7-fold in women (2, 12).

The Framingham Heart Study and other large population-based samples have allowed substantial progress in the identification of modifiable risk factors for CVD, and how these risk factors interact (13). The knowledge gained from these studies has important implications for decreasing cardiovascular-related morbidity and mortality in higher risk populations such as people with schizophrenia, as well as the mortality from other major medical conditions associated with these risk factors. Discussed further below, it is important to note that persons with schizophrenia have an increased prevalence of all the key modifiable cardiovascular and metabolic risk factors, including obesity, smoking, hyper-tension, dyslipidaemia, and hyperglycaemia.

In a study by Cohn and colleagues of 240 schizophrenia patients from a Canadian national sample, male patients were found to have a significantly increased 10-year risk of MI vs the general population (8.9% vs 6.3%), as calculated by Framingham criteria (14). Another study, conducted by McCreadie used the Framingham assessment methodology to determine the risk of coronary heart disease (CHD) in 101 patients with schizophrenia compared to 8,127 members of the general population over ten years (15). The impact of diet, weight, smoking and exercise habits on CHD risk was evaluated. Men with schizophrenia were found to have a significantly higher risk of CHD (10.5% vs 6.4%) and stroke (4.2% vs 2.3%) compared with the general population. Risk among women in this study was not found to be statistically higher than that of the general population to a significant degree. However, it is important to contrast this result with cardiometabolic risk data from schizophrenia patients entering the US-based Clinical Antipsychotic Trials in Intervention Effectiveness (CATIE) study (16). McEvoy and colleagues compared the baseline presence of cardiometabolic risk factors in the CATIE patient sample with an age-, gender- and race/ethnicity- matched general population sample from the third National Health and Nutrition Examination Survey (NHANES III). Not only did both sexes in the CATIE population have an elevated prevalence of cardiometabolic risk factors compared to NHANES III controls, women tended to have even more elevated risk than men (16). In another study comparing CATIE patients with schizophrenia to an age-matched sample from NHANES III, Goff and colleagues reported that schizophrenia patients had a significantly higher 10 year risk of CHD than matched controls–9.4% vs 7.0% in men, and 6.3% vs 4.2% in women, respectively (17). Lawrence and colleagues examined excess mortality due to CHD for psychiatric patients in comparison to the general population, and found that men with schizophrenia were only 60% as likely to be admitted to a hospital for CHD, but were 1.8 times more likely to die from CHD (7).

In addition to metabolic abnormalities, increasing risk of CVD, schizophrenia patients are approximately twice as likely to have hypertension, as are members of the general population. Estimates of the rate of hypertension in schizophrenia range from 19% (18) to roughly 47% (16). To some extent, this may be related to weight gain and insulin resistance (see below). Another risk factor that is more common in patients with schizophrenia than in the general population is smoking, with up to 81% of individuals with schizophrenia smoke cigarettes, as opposed to 28% of the general population (16, 19–21). Schizophrenia patients are also more likely to use alcohol and other stimulants than the general population, which may contribute to cardiovascular risk (20).

Diabetes is now the sixth leading cause of death in the USA, and its impact on health is global. Worldwide, the prevalence of diabetes is expected to increase 72% between 2003 to 2025, according to projections from the US Centers for Disease Control and Prevention (22). In people with schizophrenia, prevalence estimates for type 2 diabetes mellitus range from two to four times higher than in the general population. It has been estimated in multiple studies that roughly 15–18% of patients with schizophrenia have type 2 diabetes mellitus, compared with an overall prevalence of approximately 4% in the general population (23–26). The incidence of diabetes in the general population increases progressively with age, while schizophrenia is associated with an increase in the incidence of diabetes in earlier adult years, with prevalence then maintained at that level over the lifespan (25–27).

The interaction between schizophrenia and diabetes mellitus is complex, and remains incompletely understood (26, 28). Parsimony suggests that the elevated prevalence of a number of risk factors in this population may be more than sufficient to explain the observed increase in disease prevalence; indeed it is now widely assumed that multiple risk factors underlie the increased rate of diabetes among those with schizophrenia. Relevant factors include poverty, urbanization, crowding, psychological stress, and the effects of treatments such as antipsychotic medications, as well as hypothesized but so far unidentified genetic factors. There are limited data from drug-naïve schizophrenia patients which might suggest that increased activation of the hypothalamic–pituitary–adrenal (HPA) axis and the sympathetic nervous system may contribute to at least acute hyperglycaemia in patients with schizophrenia (29). However, the finding of increased adiposity or glucoregulatory impairments in unmedicated patients is not consistently observed (30), and HPA activation is variably associated with schizophrenia, and generally reduced by antipsychotic treatment rather than worsened. Familial, possibly genetic, factors may play some role, as one report found that 18–19% of individuals with schizophrenia had a family history of type 2 diabetes mellitus (23), although, in general, studies of probands with diabetes can be subject to an ascertainment bias that would tend to increase the prevalence of diabetes observed in family members, independent of any effect of schizophrenia. It remains to be seen to what extent the increased prevalence of diabetes in this population cannot be fully explained by increases in the prevalence of overweight and obesity, reductions in the overall level of physical fitness, and other effects related to lifestyle.

Under-recognition of schizophrenia as a marker for risk for diabetes may contribute to the impact of diabetes-related risk on the overall mortality observed in schizophrenia, in the absence of compensatory efforts at primary or secondary prevention (22, 26). Of note, the Canadian Diabetes Association and the American Association of Clinical Endocrinologists have added schizophrenia as a risk factor for diabetes in their screening guidelines. Currently, it is too soon to determine what effects this change might have on practice patterns including especially the low rate of screening for abnormalities in plasma glucose or lipids in this population (31).

Key risk factors for diabetes and CVD that form the basis of public health efforts in the general population are overweight and obesity, smoking, hypertension, dyslipidaemia, and hyperglycaemia. All of these factors are more prevalent among those with schizophrenia than in the general population. The success of efforts to address the excess mortality observed in persons with schizophrenia is highly likely to depend on the success of primary and secondary prevention efforts targeting modifiable risks in this population (8).

Obesity is approximately twice as prevalent in people with severe mental illnesses such as schizophrenia, in comparison to the general population (32, 33). Evidence from a number of studies illustrates this problem. In a study of Scottish schizophrenia patients conducted by McCreadie, a total of 73% of the sample were found to be overweight, while 86% of women with schizophrenia were either overweight or obese (15). A study of North American patients with schizophrenia or schizoaffective disorder, conducted by Cohn and colleagues, found that 31% of men and 43% of women were obese (14). A study of psychiatric inpatients with schizophrenia by Cormac and colleagues showed that the rate of obesity was 36% in men and 75% in women versus obesity rates of 17% and 22%, respectively, in the reference population (21).

Patients with schizophrenia are likely to be at a higher risk of becoming overweight or obese due to a constellation of clinical, physiologic, psychosocial, environmental factors, and possibly additional genetic factors. The negative symptoms of schizophrenia, including apathy and social withdrawal, as well as poverty can be hypothesized to contribute to the sedentary lifestyle and poor diet of this patient population. In addition, there is substantial evidence to indicate that treatment with psychotropic medications, particularly some antipsychotic medications, can induce clinically significant increases in weight and adiposity. Drugs with greater antagonism for histamine (H₁) receptors, and to some extent α₁ adrenoceptors, have been implicated in greater degrees of antipsychotic-induced weight gain (34). In general, drug treatments that decrease energy expenditure, via sedation or reduced motor activity, or increase caloric intake, via increase appetite or reduced satiety, can potentially increase body weight. Rates of clinically significant weight gain, generally defined as a 7% or more increase in body weight, are available from pooled registration trial data reported in the US package insert for available medications. For example, the reported rates of incident weight gain during registration trials using this definition are 29% for olanzapine (vs 3% for placebo), 23% for quetiapine (vs 6% for placebo), 18% for risperidone (vs 9% for placebo), 10% for ziprasidone (vs 4% for placebo), and 8% for aripiprazole (vs 3% for placebo), (35, 36)

Longer-term data show a similar pattern. Over 1 year of treatment, again using pooled registration clinical trial data, olanzapine produces a mean 12 kg of weight gain at doses in the range of 12.5 to 17.5 mg/day (37), while quetiapine produces a mean 3.2 kg increase (38), risperidone 2.2 kg, and ziprasidone and aripiprazole approximately 1 kg (35). It should be noted that the 1-year data for quetiapine may not be directly comparable to data for the other drugs, as the quetiapine data are based on ‘observed case’ analyses where all patients completed the 1 year exposure, in contrast to a commonly used ‘last observation carried forward’ method of analysis that tends to underestimate drug effects on treatment-related weight change.

Dyslipidaemia is another risk factor for CVD that is found at increased prevalence in those with schizophrenia. When a group of patients with schizophrenia were studied by Heiskanen and colleagues to determine the frequency of the metabolic syndrome, 31% were found to have serum triglyceride levels higher than 1.68 mmol/l (149 mg/dl), meeting the hypertriglyceridaemia criterion for that syndrome. In addition, 58% of men had low serum high-density lipoprotein (HDL) cholesterol levels, and 25% of women had low HDL levels (39). Among patients entering the CATIE study, 50.7% were found by McEvoy and colleagues to have elevated triglyceride levels, and 48.9% had low HDL levels (15).

Some antipsychotic medications are associated with risk for developing clinically significant increases in serum lipid concentrations, generally—but not exclusively—in association with their propensity to produce weight gain. In phase 1 of the CATIE study, olanzapine was associated with the largest worsening of lipid parameters from baseline. Patients treated with olanzapine experienced an exposure-adjusted mean triglyceride increase of 0.46 mmol/l (40.5 mg/dl) and a mean increase in total cholesterol of 0.24 mmol/l (9.4 mg/dl), in contrast for example to exposure-adjusted mean decreases of 0.19 mmol/ l (16.5 mg/dl) and 0.21 mmol/l (8.2 mg/dl), respectively, for patients treated with ziprasidone (40). Notably, 23% of those patients randomized to olanzapine had already been taking olanzapine prior to the study, and therefore they may have already experienced some drug-induced effect on plasma lipids, attenuating observed change during the study. In CATIE phase 2, where the study methodology prevented assignment to a treatment arm that the patient was already taking in phase 1, drug effects on lipids were more apparent. For example, patients in phase 2 who were assigned to olanzapine showed greater increases in exposure-adjusted mean triglycerides (1.06 mmol/l or 94.1 mg/dl) and total cholesterol (0.45 mmol/l or 17.5 mg/dl) than seen in phase 1 (41).

The prevalence of insulin resistance in patients with schizophrenia has been estimated at 1.5–2 times the general population prevalence (25). When individuals with insulin resistance progress to diabetes, they are at a greater risk for increased morbidity and mortality due to acute metabolic complications, as well as chronic microvascular and macrovascular complications. Antipsychotics have been implicated as a risk factor for hyperglycaemia in patients with schizophrenia, but some evidence also suggests that the increased tendency towards hyperglycaemia in this population may be related to the disease state (29). Lifestyle factors can affect outcomes, with poor diet, inactivity, and lack of exercise commonly associated with overweight and obesity and insulin resistance in untreated non-psychiatric samples as well as in treated patients (29, 35, 42).

Different antipsychotic medications are associated with different levels of risk for insulin resistance and hyperglycaemia, generally proportional to drug-induced weight gain, although additional adiposity-independent effects may contribute to risk with some medications. In phase 1 of CATIE, plasma glucose levels increased by a mean 0.8 ± 0.2 mmol/l (15.0 ± 2.8 mg/dl) in patients randomized to olanzapine, in comparison for example to 0.1 ± 0.2 mmol/l (2.3 ± 3.9 mg/dl) in patients randomized to ziprasidone, although neither change was statistically significant (40). In general, the ability to detect drug-induced risk for hyperglycaemia during the routine duration of most clinical trials is limited by the natural course of a condition like type 2 diabetes.

While antipsychotics medications confer most of their associated risk for worsening insulin resistance through weight gain, evidence from controlled studies suggests the existence of some additional, adiposity-independent treatment effects on blood glucose, possibly via drug effects on glucose transporter function (42–44). Certain antipsychotic agents, such as clozapine and olanzapine, achieve relatively high intracellular concentrations where they can interact with intracellular glucose transporter proteins and reduce glucose transport from extracellular to intracellular space, potentially contributing to risk for hyperglycaemia (45).

Studies indicate that the prevalence of metabolic syndrome in patients with schizophrenia ranges from 37% (39) to over 50% (46) While the prevalence of the metabolic syndrome increases with age in the general population, this tends not to be the case in patients with schizophrenia, where a higher prevalence of metabolic syndrome tends to be achieved earlier in life (14, 39). One study showed a prevalence rate of 43.8% in individuals with schizophrenia who were under age 45, while schizophrenia patients over 45 had a similar prevalence rate of 45.8% (14). Another study found that the baseline rate of metabolic syndrome in patients entering the CATIE study was roughly 41% (16). They also studied the prevalence of individual metabolic syndrome criteria at baseline in CATIE, comparing the results with an age-matched general population sample from NHANES III, and found that all criteria for the metabolic syndrome were present at elevated rates in CATIE patients of both genders, with the sole exception of hyperglycaemia in male patients (16).

Antipsychotic medications are associated with risk of the metabolic syndrome to differing degrees, again generally in proportion to weight gain. Pooled data from two 26-week randomized double-blinded studies, comparing aripiprazole with placebo (47) and comparing aripiprazole with olanzapine, (48) showed incidences of metabolic syndrome of 19.2 ± 4.0% for olanzapine, 12.8 ± 4.5% for placebo, and 7.6 ± 2.3% for aripiprazole.

In the general US population during the past decade, most of the reductions in CVD mortality are attributable to improvements in the treatment of acute cardiovascular events, and in long-term secondary prevention. For example, the decrease in the fatality rate for hospitalized myocardial infarction has been attributed to the utilization of aspirin, thrombolytics, β-blockers, and angiotensin converting enzyme (ACE) inhibitors. Longer-term post-myocardial infarction use of aspirin, β-blockers, ACE inhibitors, and statins, as well as therapeutic lifestyle changes, have made additional contributions to reductions in myocardial infarction-related mortality. Unfortunately, patients with major mental illnesses such as schizophrenia who experience an acute myocardial infarction are significantly less likely than the general population to receive therapies of proven benefit, including thrombolytics, aspirin, β-blockers, and ACE inhibitors (49). Patients with schizophrenia are also less likely than members of the general population to undergo cardiac catheterizations and receive emergency angioplasties or coronary artery bypass grafts. In a study of over 88 000 Medicare patients hospitalized for myocardial infarction, mortality in the follow-up period was increased by 19% in the presence of any mental disorder, and by 34% in persons with schizophrenia, with these increases in mortality attributable to reductions in the quality of care (48).

However, smoking rates in the general population have declined considerably over the past several decades, from over 50% in the 1950s to around 25% at present (50). Among patients with diagnosable mental illness, 50–80% are smokers and consume 34–44% of all cigarettes in the USA (51). In the USA, while patients with major mental illness are overrepresented in state programmes such as Medicaid, some states do not cover any form of smoking cessation treatment and only a few states cover all the smoking cessation treatments recommended in the US Preventive Services Task Force guidelines.

Some portion of the high prevalence of modifiable cardiometabolic risk factors in the schizophrenia population can be explained by underdiagnosis and undertreatment. Nasrallah and colleagues studied the rates of treatment for existing modifiable risk factors among schizophrenia patients entering the CATIE study. Among the approximately 1,500 patients entering the CATIE study, which was conducted at 57 US sites spanning a range of academic and public sector treatment settings, 88% of those patients with dyslipidaemia were receiving no lipid-lowering pharmacotherapy, 30% of those with diabetes mellitus were receiving no antidiabetic medications, and 62% with hypertension were receiving no antihypertensive medication (52).

Screening for dyslipidaemia and hyperglycaemia occurs at very low rates in patients with schizophrenia. This phenomenon is observed even during treatment with antipsychotic medications associated with risk for disturbances in glucose and lipid metabolism (35, 40, 53). A large cohort study by Morrato and colleagues, involving Medicaid claims data for 55 436 enrollees with mental illness from several US states, showed that in the four months prior to and after a new antipsychotic prescription, less than one-third of patients overall received any plasma glucose measurement and less than 10% received any plasma lipid measurement (31). When patients with an existing diagnosis of diabetes were excluded from the analysis, glucose testing dropped to 10–15% of patients, and lipid measurements decreased to only 5% of patients. Therefore, prevention-oriented screening was found to be especially uncommon. Such low levels of screening may contribute to the observed low levels of diagnosis and treatment for modifiable CVD risk factors in patients with schizophrenia. National Cholesterol Education Program (NCEP) guidelines recommend that patients with diabetes mellitus be treated as aggressively as patients with prior MI or stroke who do not have diabetes. However, patients with diabetes and schizophrenia are less likely than non-mentally ill diabetes patients to receive standard-of-care treatments. Frayne and colleagues conducted a study of over 300 000 patients with diabetes in the Veteran’s Administration system, with approximately 25% of the sample also having a mental illness. They found that the presence of mental illnesses such as schizophrenia and bipolar disorder significantly increased the risk of not receiving appropriate elements of care such as eye exams, plasma lipid testing, and HbA_1c monitoring (54).

Based on a consensus development statement from the American Diabetes Association, drug-naïve patients initiating antipsychotic therapy, in addition to those changing antipsychotic treatment, are appropriate targets for a baseline assessment of family and personal medical history, in addition to an assessment of metabolic risk indicators such as weight (body mass index (BMI)) and/or waist circumference, blood pressure, fasting plasma glucose (FPG), and fasting lipid measurements (total, low-density lipoprotein (LDL) and HDL cholesterol, and triglyceride), as well as possibly HbA_1c, with repeat measurements of weight at every visit thereafter. Regular measurements of weight can provide important information regarding treatment-related risk, requiring only a scale. Laboratory tests should be repeated approximately 3 months into initial treatment and at least annually thereafter, or more frequently in the setting of increased risk (25). At least every 6 months, all patients with schizophrenia should have waist circumference, BMI, and blood pressure recorded (17), with more frequent screening for higher-risk patients. For patients with hypertension, for example, the National Heart, Lung, and Blood Institute (NHLBI) recommends monthly monitoring of blood pressure following the initiation of antihypertensive therapy, with follow-up every 3–6 months once blood pressure has been stabilized.

Ideally, medical comorbidities that might arise during treatment should be prevented rather than treated after the fact. Clinicians should strongly consider taking cardiometabolic risk into consideration when selecting psychotropic agents, and recognize that agents associated with a higher risk of weight gain may be less appropriate for patients with elevated risk at baseline. It is noteworthy that in October of 2007, the US Food and Drug Administration (FDA) prescribing information for olanzapine was altered to encompass warnings about higher risk of weight gain, hyperglycaemia, and hyperlipidaemia compared to other agents in the class, and further analyses of risk for medications used in this population are likely to be forthcoming.

When elevated cardiometabolic risk is discovered, secondary causes of risk should be addressed if possible, according to the NCEP. This might be achieved, for example, by switching a patient from an agent with a higher to a lower risk of weight gain and dyslipidaemia. Behavioural therapy can also be an effective method for overweight or obese patients with schizophrenia to manage weight gain (55). If these interventions fail, treatment of risk factors such as dyslipidaemia with a medication such as an HMG-CoA reductase inhibitor or other lipid-lowering therapy should be initiated.

Given the differential weight gain associated with different anti-psychotic medications, one could predict that reductions in weight might be associated with a switch from an agent with high risk of weight gain to one with lower risk, but not with switches between two treatments of similar weight gain liability (36). Indeed, these predictions have been confirmed in clinical trials involving switches to aripiprazole and to ziprasidone (35, 56). In a recent example of a longer-term study, patients were switched to ziprasidone from risperidone, olanzapine, or a high-potency first-generation antipsychotic. After 52 weeks of treatment with ziprasidone, patients switched from olanzapine had a mean 9.8 kg weight loss compared with 6.9 kg in those switched from risperidone and negligible change in the switch from agents such as haloperidol, providing an important proof of concept concerning the prediction of weight change during therapeutic substitutions (57).

Investigations are taking place into disparate strategies to address obesity in this population and/or to reverse iatrogenic weight gain associated with antipsychotic medications. A 2007 Cochrane database review by Faulkner and colleagues studied the effects of pharmacological and non-pharmacological intervention strategies. It is noteworthy that the strategy of switching antipsychotic medications was not included in the review. Adjunctive pharmacotherapy with various agents was found to yield generally under-impressive results, based on limited data, typically from one or two studies for each agent with small sample sizes. No single agent was found to yield consistently superior results. The various adjunctive agents studied were amantadine, d-fenfluramine, dextroamphetamine sulfate, nizatidine, phenylpropanolamine, sibutramine, topiramate, and fluoxetine. Among non-pharmacological interventions, cognitive behavioural therapy yielded encouraging results in two studies, with significantly greater weight reduction than standard care.

Currently, switching antipsychotic medications may be one of the most effective strategies for dealing with antipsychotic-associated weight gain. Weight loss has been found to extend at least over a one-year time frame, with concordant improvements in indices of cardiovascular risk such as fasting lipid profiles. Additional data are currently being developed, including the CATIE investigators’ Comparison of Antipsychotics for Metabolic Problems (CAMP) study, which randomly assigns overweight patients (BMI ≥27) with elevated non-HDL cholesterol to either stay on their existing antipsychotic medication (olanzapine, risperidone, or quetiapine) or switch to aripiprazole.

Appropriate monitoring and intervention can decrease long-term medical costs, improve general medical outcomes, and enhance quality of life among patients with schizophrenia. In 2000, Dixon and colleagues showed that more schizophrenia patients who were being treated for type 2 diabetes mellitus were satisfied with their lives than were schizophrenia patients with the same comorbid status who were not receiving treatment for diabetes (23). Concern for the physical health of schizophrenia patients has also been hypothesized to enhance the therapeutic alliance and potentially increase the effects of, and compliance with, psychiatric treatments.

Schizophrenia patients have elevated rates of medical comorbidities including especially heart disease and diabetes. Addressing the treatment of medical comorbidities is essential to improving health outcomes among patients in this population. There is a need for strategies that can provide clear guidance to administrators and medical directors on how to incorporate changes into already existing medical and psychiatric systems.

Care providers have a particular responsibility to address those components of medical risk that are iatrogenic in origin, e.g. the contribution of medication side effects. The Institute of Medicine has suggested that all psychiatric and medical systems involved in the care of patients with schizophrenia—including mental health, substance abuse treatment, general health care, and other services—need to effectively collaborate in an effort to coordinate care of their patients and eliminate gaps and redundancies in necessary services (59). This collaboration should include the establishment of structured and routine comorbidity risk assessments, in addition to scheduled monitoring of antipsychotic treatment for side effects and adherence issues, and medical comorbidities of schizophrenia should receive the same established interventions that are provided to the general population (59). Implementation of such screening before the initiation of antipsychotic treatment, and during treatment, will likely improve outcomes among individuals with schizophrenia.

Although there is no one perfect model for healthcare delivery to the mentally ill, one proposed model would create an integrated clinic in which medical care providers are on site and interacting with psychiatrists. Such an approach can be hypothesized to improve medical outcomes without increasing total costs. It could include programmes designed to modify the behaviour of both patients and health care providers, in order to increase access to effective medical care. This approach will involve increases in primary prevention, as well as diagnostic and treatment programmes that specifically target this population. Regardless of the structure of health care delivery, the psychiatrist will need to play a central role in a schizophrenia patient’s overall care. Such a role may involve coordination of care with a patient’s primary care physician.

Data indicate a crucial need for new paradigms in the prevention and treatment of medical illness in patients with schizophrenia, including closer attention to choice of psychotropic drug treatment regimens, and more aggressive use of monitoring and interventions to identify and reduce risk. However, efforts to improve coordination of services face short- and long-term challenges, ranging from fiscal concerns to lack of awareness among health care providers who might play key roles. In the short term, the existing mental care system will have to put forth a significant effort, reallocating existing resources to coordinate screening, interventions including needed referrals, and follow-up monitoring. Mental health care providers and systems will need to improve working relationships with primary care and specialty collaborators, including proactive efforts to facilitate evaluation and follow-up for patients. Without future collaboration between primary health care providers, medical specialists, and psychiatrists, the large burden of avoidable premature mortality in patients with schizophrenia is likely to continue, and the disparities between schizophrenia patients and the general population are likely to increase in severity.

John W. Newcomer has no significant financial conflict of interest in compliance with the Washington University School of Medicine Conflict of Interest Policy. Dr Newcomer has received research grant support from the National Institute of Mental Health, the National Alliance for Research on Schizophrenia and Depression, the Sidney R. Baer Jr. Foundation, Janssen Pharmaceuticals, Bristol-Myers Squibb, Wyeth Pharmaceuticals Inc., and Pfizer Inc.; he has served as a consultant to AstraZeneca Pharmaceuticals, Bristol-Myers Squibb, GlaxoSmithKline, Janssen Pharmaceuticals, Pfizer Inc., Solvay, Otsuka Pharmaceuticals, Wyeth Pharmaceuticals Inc., Forest, Sanofi, Lundbeck, Tikvah, Otsuka and Vanda; he has been a member of Data and Safety Monitoring Boards for Organon, Schering Plough, Dainippon Sumitomo, and Vivus; he has been a consultant to litigation; finally, he has received royalties from Compact Clinicals/Jones and Bartlett Publishing for a metabolic screening form.

Dr John Newcomer would like to thank Glennon M. Floyd, Managing Editor, Department of Psychiatry, Washington University School of Medicine, for editorial assistance on this chapter.