Metabolic Risk Factors Among Medicaid Outpatients With Schizophrenia Receiving Second-Generation Antipsychotics

Although the average American anticipates a lifespan of 78 years or more, those with schizophrenia face an actuarial guillotine: premature death stalks those with severe mental illness, deducting years, possibly even decades, from their life expectancy ( 1 , 2 ). Although one-third of this excess mortality is attributable to homicide and suicide, the majority is due to natural causes, principally cardiovascular disease ( 3 ).

Risk factors for cardiovascular disease are proportionately higher among patients with schizophrenia than in the general population: persons with schizophrenia develop diabetes at a rate 1.5 to 3.5 times higher than that of the general population, with similarly elevated prevalence rates for obesity, dyslipidemia, insulin resistance, and hypertension—all of which are associated with increased morbidity and mortality ( 4 , 5 ). Although lifestyle factors such as smoking and poor diet contribute to the development of these cardiovascular risk factors, it is widely accepted that many second-generation antipsychotics, the first-line treatment for schizophrenia, are independently associated with dyslipidemia, weight gain, and increased incidence of diabetes ( 6 ). Ultimately, because the level of excess mortality is so high, identifying the prevalence of these risk factors among persons with schizophrenia is of considerable interest to public health investigators ( 7 ).

This study was derived from a quality improvement project to reduce morbidity and mortality attributable to cardiovascular disease among clients with schizophrenia receiving public mental health services and taking second-generation antipsychotic medications. The project's aims were to identify clients at risk, increase illness awareness among providers, and explore intervention options. Although a number of studies have examined risk factors for cardiovascular disease among persons with serious, chronic mental illness, the exclusion criteria of the studies compromise their generalizability. For example, the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) published metabolic data on 689 participants with schizophrenia who were randomly assigned to receive a second-generation antipsychotic or perphenazine, but the study excluded persons with serious and unstable medical conditions, a history of treatment resistance, or severe cognitive disorders ( 8 ). Similarly, other investigations of the topic excluded individuals with diabetes, alcohol dependence, or certain medical conditions; were limited to specialty groups, such as male veterans; or included second-generation antipsychotics not available in the United States ( 9 , 10 , 11 ). Taken together, these studies canvassed a population less diverse and with fewer psychiatric and medical comorbidities than that treated in the typical public mental health system in the United States.

Here we present screening results for a diverse population with schizophrenia being treated with second-generation antipsychotics at 15 clinics within a large, county public mental health system. Its scale and absence of clinical exclusion criteria provides a comprehensive view of metabolic risk among public mental health outpatients with schizophrenia who receive second-generation antipsychotics and offers a glimpse of the challenges faced in assessing and treating basic medical issues in this population.

Methods

This project collected clinical information regarding body mass index (BMI), hypertension, diabetes or impaired fasting glucose levels, and dyslipidemia from Medicaid patients diagnosed as having schizophrenia and treated with a second-generation antipsychotic medication. Data were collected between May 1, 2006, and October 31, 2007.

Patients were enrolled in one of the 15 outpatient mental health clinics providing publicly funded mental health services in King County, Washington, which includes the greater Seattle area. All participants had a diagnosis of schizophrenia, were Medicaid recipients, and received a second-generation antipsychotic through one of the clinics. Because this was an ongoing quality improvement project that was conducted to meet the requirements of the Center for Medicare and Medicaid Services, institutional review board approval was not initially sought. However, an exemption from review was obtained from the Washington State Human Research Review Board for the purposes of this publication.

A pool of 1,758 potentially eligible participants was drawn from the county's electronic management information system on the basis of enrollment, diagnosis, and Medicaid status. Clinics either screened the individuals listed or provided a reason for exclusion. Clinics also identified an additional 182 eligible patients who entered treatment after the original list was drawn.

Of the 1,940 potentially eligible individuals, 434 (22%) were ineligible for inclusion: 257 clients (13%) were excluded because they were not prescribed a second-generation antipsychotic, 110 (6%) had a prescriber outside the affiliated center, 60 (3%) were not diagnosed as having schizophrenia, five (<1%) had died, and two (<1%) no longer received Medicaid. Of the remaining 1,506 eligible individuals, 189 were discharged from services or were unable to be contacted during the six-month enrollment period. Of the 1,317 persons who were eligible, 498 (38%) refused to participate or did not complete the screening process. A total of 819 individuals completed the initial screen. Of the 819 participants who were screened, 692 (84%) provided fasting glucose samples and 675 (82%) provided both fasting glucose and fasting lipid samples.

Serum samples were collected at secondary sites (such as phlebotomy laboratories) unless recent medical records (less than six months old) were available. Participants were asked for a fasting venous blood sample, but nonfasting samples were accepted and analyzed separately. Blood work was processed by accredited laboratories for glucose level, Hb1Ac, and lipids (total cholesterol, high-density lipoprotein [HDL], low-density lipoprotein [LDL], and triglycerides). Information on second-generation antipsychotic usage was provided by the client's prescribing physician. Height, weight, blood pressure, laboratory data, and second-generation antipsychotic prescribed were recorded at the patient's mental health clinic on a standardized form. Clinics received an incentive of $150 for each participant who completed the screening process.

Obesity was defined as a BMI of 30 or more, and overweight was defined as a BMI between 25 and 29.9 ( 12 ). Hypertension was defined as either taking antihypertensive medication or having a systolic blood pressure in excess of 140 mm Hg or a diastolic blood pressure higher than 90 mm Hg: if the patient had been diagnosed with diabetes, hypertension was defined as having a systolic blood pressure in excess of 130 mm Hg or having a diastolic blood pressure higher than 80 mm Hg ( 13 ). Diabetes was defined as a diagnosis of either type I or type II diabetes or a fasting plasma glucose level ≥126 mg/dL, and impaired fasting glucose level was defined as a fasting glucose level between 100 and 125 mg/dL ( 14 ). Dyslipidemia was defined as receiving medications for dyslipidemia or having suboptimal fasting levels of HDL (<40 mg/dL for men and <50 mg/dL for women), triglycerides (>200 mgs/dL), or LDL (>160 mg/dL for those without diabetes or >100 mg/dL for those with diabetes) ( 15 ).

It was anticipated that there would be difficulties obtaining fasting serum samples in this population, and three primary care physicians and the medical directors of the participating mental health clinics established nonfasting thresholds for glucose intolerance (glucose levels ≥200 mg/dL or Hb1Ac levels >7%) and elevated lipids (total cholesterol divided by HDL >6.8 for women and >5.6 for men).

The full sample of 819 persons was used in analyses of the prevalence of obesity and hypertension. The subsample of 692 patients providing fasting glucose samples was used to determine the prevalence of diabetes and impaired fasting glucose. The slightly smaller subsample of 675 patients providing fasting samples for both glucose and lipid testing was used to determine the prevalence of dyslipidemia (because determining the appropriate lipid threshold required an accurate diagnosis of diabetes). This subgroup of 675 was also used to calculate the prevalence of multiple risk factors. Persons providing nonfasting samples were evaluated separately, because nonfasting values are not directly comparable to the existing literature. Chi square analyses and t tests were used to examine sample generalizability and were carried out using SPSS, version 11.1.

Results

Among the 819 participants completing the screening, 246 were women (30%). The participants' mean±SD age was 47.1±12.2 years. A total of 512 (63%) were Caucasian, 148 (18%) were African American, 114 (14%) were Asian or Pacific Islander, 22 (3%) were Hispanic, 15 (2%) were Native American, and ten (1%) were listed as "other." The 819 individuals who completed the screening did not differ from the 498 people who were eligible but did not complete the screening with respect to age, gender, race, or ethnicity.

All second-generation antipsychotics approved for use in the United States in 2007 were represented among medications prescribed to study participants: risperidone (either depot or oral) (N=313, 38%), olanzapine (N=241, 29%), clozapine (N= 147, 18%), quetiapine (N=121, 15%), aripiprazole (N=95, 12%), ziprasidone (N=63, 8%), and paliperidone (N=11, 1%). Percentages add to more than 100% because 168 participants were taking two second-generation antipsychotics and two were taking three second-generation antipsychotics.

The 819 participants had a mean BMI of 29.3±7.0. A total of 591 participants (72%) were either overweight (N=249, 30%) or obese (N= 342, 42%) ( Table 1 ).

Enlarge table

Thirty-nine percent had an indicator of hypertension: antihypertensives were prescribed to 26%, and an additional 13% were not receiving antihypertensives but had elevated blood pressure ( Table 1 ).

Of the 692 individuals providing fasting glucose samples, 133 (19%) were found to have diabetes: 15% had an existing diagnosis of diabetes (94 persons, or 14%, had type II diabetes; five, or 1%, had type I diabetes; and four, or 1%, did not specify the type), and an additional 4% were said to have diabetes on the basis of the screening threshold for fasting glucose levels (≥126 mg/dL). Another 18% had fasting glucose levels between 100 and 125 mgs/dL, indicating impaired fasting glucose. Therefore, of the 692 individuals providing fasting glucose samples, 260 (37%) had either diabetes or impaired fasting glucose.

Among the 127 participants providing only a nonfasting serum sample, 20 (16%) were receiving treatment for diabetes and two (2%) met the nonfasting thresholds for glucose intolerance (glucose levels ≥200 mg/dL or Hb1Ac levels >7%). Adding these 22 persons to the 260 who met the fasting thresholds for diabetes or impaired fasting glucose suggests that an estimated 282 (34%) of the screened participants had diabetes or impaired fasting glucose ( Table 1 ).

Of the 675 participants providing both fasting lipid samples and fasting glucose samples, 28% were being treated for dyslipidemia. In addition, 39% had at least one of the following: HDL levels below the recommended value, elevated triglycerides, or elevated LDL levels. Thus the total incidence of dyslipidemia for those providing both fasting glucose and fasting lipid samples was 452 (67%).

Of the 144 clients who did not provide both fasting glucose and serum lipid samples, 24 (17%) were receiving treatment for dyslipidemia and 21 (15%) exceeded the nonfasting lipid threshold (total cholesterol divided by HDL >6.8 for men and >5.6 for women). Adding these 45 to those who provided fasting serum lipid samples suggests that an estimated 497 (61%) of all screened participants (N=819) had dyslipidemia ( Table 1 ).

Of the 675 participants who provided fasting serum samples suitable for both glucose and lipid analysis, 90% had at least one risk factor (obesity or overweight, hypertension, diabetes or impaired fasting glucose, or dyslipidemia). Specifically, 137 participants (20%) had only one risk factor, 208 (31%) had two risk factors, 175 (26%) had three risk factors, and 93 (14%) had all four risk factors. Recalculating using the full screened sample of 819, including the additional nonfasting subset, did not change the rates of participants who experienced one, two, three, or four risk factors by more than 2%.

Discussion

The prevalence of metabolic risk factors was nearly ubiquitous for our community-based population, with 607 (90%) of the 675 who provided fasting serum samples having at least one risk factor and over two-thirds having two or more risk factors. When modified protocols were used for the additional 144 participants who provided nonfasting serum samples, the prevalence of at least one metabolic risk factor for the entire group of 819 screened was only slightly lower (N=720, 88%).

The most common metabolic risk factor, experienced by over two-thirds of the participants, was elevated BMI, followed closely by dyslipidemia. About one-third of the participants showed impaired fasting glucose or diabetes, and one-third evidenced hypertension. [An appendix showing how these incidence rates compare with those of a general population of low-income adults is available as an online supplement at ps.psychiatry.online.org .]

Although this project looked at an unusually large cohort of clients with schizophrenia who received care through community health centers and who qualified for Medicaid, its generalizability was diminished because approximately one-third of individuals did not complete the screening. The greatest number of incomplete screenings was due to a failure to complete the blood work, undoubtedly attributable to a combination of patient factors, including aversion to an unpleasant procedure, lack of motivation to test for nonsymptomatic illnesses such as dyslipidemia, and the need to travel to a laboratory for the procedure. Participation could be improved by sharing the $150 financial incentive with patients who complete the laboratory tests and by decreasing barriers to laboratory services through providing transportation or on-site phlebotomists.

Another limitation of the project was the use of a single blood pressure measurement to identify hypertension: this lacks the sensitivity and specificity of serial measurements across time, and there is insufficient information in the field to estimate the validity of this particular diagnostic technique.

Conclusions

Given the excess mortality among persons with the most serious mental illness and the tremendous burden of metabolic disease found in the inclusive sample of community mental health patients in this study, psychiatrists must not only be assiduous in referring patients to primary care but also ensure that their patients have access to the preventive services and ongoing care needed to address these chronic metabolic illnesses—either as part of their own clinical services or with reliable and available community partners. Obviously, this will require increasing the pool of primary care providers willing to treat Medicaid patients by addressing the financial disincentives caused by poor Medicaid reimbursement rates.

Such medical follow-up must include effective medications as well as interventions addressing the crucial role played by poor diet, inadequate exercise, and smoking in the development of the metabolic disorders discussed in this brief report. Our mission must not only be to help patients achieve remission from their mental illness but also to work with the larger medical system to ensure that they receive the care required in order to live to enjoy their recovery for years to come.

Acknowledgments and disclosures

Dr. Ries is on the speakers bureau for Eli Lilly and Company, Janssen, AstraZeneca, and Pfizer. The other authors report no competing interests.