February 2010  << Back  

  Can't open the PDF? Click here for help.



Assessment of monitoring for glucose and lipid dysregulation in adult Medi-Cal patients newly started on antipsychotics

Mitchell Barnett, PharmD, MS

Touro University-California, College of Pharmacy, Vallejo, CA, USA

Shannon VonMuenster, PharmD

Touro University-California, College of Pharmacy, Vallejo, CA, USA
Center for California Medicaid (Medi-Cal), Sacramento, CA, USA

Heidi Wehring, PharmD

Touro University-California, College of Pharmacy, Vallejo, CA, USA

Sarah Popish, PharmD

Touro University-California, College of Pharmacy, Vallejo, CA, USA

Karna McDonald, PharmD

Touro University-California, College of Pharmacy, Vallejo, CA, USA

Victor M. Walker, RPh

Center for California Medicaid (Medi-Cal), Sacramento, CA, USA

Paul Perry, PhD

Touro University-California, College of Pharmacy, Vallejo, CA, USA

BACKGROUND: Because patients receiving antipsychotics are at increased risk for coronary heart disease, standards of care for such patients now include periodic glucose and lipid testing. The objective of this study was to examine rates of glucose and lipid monitoring among adult Medicaid patients initiated on antipsychotic therapy.

METHODS: California Medicaid (Medi-Cal) claims of 6601 patients identified as “new” antipsychotic users between July 1, 2004 and June 30, 2005 were analyzed. Rates of glucose and lipid testing were compared for 6 months prior to and post–initiation of antipsychotic therapy. Odds ratios (ORs) for testing associated with first-generation antipsychotic (FGA) and second-generation antipsychotic (SGA) use were determined while controlling for patient level factors.

RESULTS: In a multivariate analysis, SGA patients were more likely than FGA patients to undergo glucose testing (OR, 1.38; 95% confidence interval [CI], 1.13 to 1.70; P < .01) and lipid testing (OR, 1.43; 95% CI, 1.14 to 1.81; P < .01), respectively. SGA patients were also more likely than FGA patients to receive both glucose and lipid testing in the 6 months following initiation of antipsychotic treatment (OR, 1.40; 95% CI, 1.11 to 1.79, P < .01).

CONCLUSION: Although increases in glucose and lipid testing rates were observed among Medi-Cal patients after initiation of antipsychotic therapy, recommended monitoring does not appear to occur universally in this population. Interventions to increase monitoring of these patients are warranted.

KEYWORDS: antipsychotic agents, metabolic disorders, type 2 diabetes, dyslipidemia, guideline adherence



All mental disorders are associated with an increased risk for premature death from natural and unnatural causes.1 Public assistance patients diagnosed with major mental illness die at younger ages than do patients with nonmajor mental illness diagnoses. Most such patients die of causes similar to the leading causes of death in the United States, including heart disease, cancer, and cerebrovascular, respiratory, and other lung diseases.2 Patients with schizophrenia have a life expectancy approximately 20% lower than that of the general population, at age 61 vs age 76, respectively.3 More than two-thirds of these patients die of coronary heart disease (CHD), compared with approximately half of the general population. Primary CHD risk factors include cigarette smoking; obesity leading to dyslipidemia, insulin resistance, and diabetes; and hypertension. In the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE), the 10-year risk for CHD among patients diagnosed with schizophrenia who were taking antipsychotic medication was estimated to be elevated among male (9.4%) and female (6.3%) patients, compared with controls. These patients also had higher rates of smoking (68% vs 35%), diabetes (13% vs 3%), hypertension (27% vs 17%), and lower HDL cholesterol levels (43.7 vs 49.3 mg/dL) compared with controls.4 Not surprisingly, the risk for metabolic syndrome among patients diagnosed with schizophrenia was estimated to be 2 to 3 times that of the general population.5 Thus, it would seem to be a reasonable hypothesis that laboratory tests to detect dyslipidemia and type 2 diabetes mellitus (T2DM) are underused in patients diagnosed with schizophrenia and other major mental illness who are being treated with antipsychotics. Studies show that use of preventive medical services such as mammograms, Pap smears, pelvic exams, immunizations, cancer screening, and tobacco screening and counseling is significantly lower among patients receiving antipsychotics compared with the general medical population.6-10

The standard of care for antipsychotic drug monitoring was recently established by a position statement jointly published and endorsed by the American Diabetes Association (ADA), the American Psychiatric Association (APA), the American Association of Clinical Endocrinologists (AACE), and the North American Association for the Study of Obesity (NAASO).11 Regarding clinical laboratory monitoring, fasting plasma glucose (at baseline, 12 weeks, and annually) and a fasting lipid profile (at baseline, 12 weeks, then every 5 years if normal) were recommended. The justification for these recommendations was demonstrated years earlier by the Centers for Disease Control and Prevention (CDC). The CDC used a computer simulation model to demonstrate the effectiveness of early detection and treatment of T2DM.12 The benefits of early detection and treatment result from the postponement of complications and the resulting improvement in quality of life. The clinical implementation of these monitoring recommendations varies widely among institutions. The Louis Stokes Cleveland Veterans Affairs Medical Center recommends that all patients receiving SGAs be monitored for weight gain every 6 months, and have laboratory tests of serum lipids every 6 months and hemoglobin A1C at baseline, 1, 3, and 6 months.13 At the Medical Center, electronic order entry allows for a pop-up box to prompt the prescriber to enter baseline height and weight and select the required laboratory tests when ordering SGAs. In 2004, the US Food and Drug Administration (FDA) required all manufacturers of atypical antipsychotics to change their labeling to include a warning about the risks of hyperglycemia and diabetes. It should be pointed out that, although all atypicals carry the label warning, some evidence suggests that not all atypicals are equal in their effects of glucose and lipids. Despite these explicit requirements and the coupled benefits of glucose and lipid monitoring with atypical antipsychotic use, little data are available to support clinician use of these recommendations.14-15

This study attempted to examine rates of monitoring for dyslipidemia and T2DM in a population of Medicaid patients initiating therapy on antipsychotics and to quantify patient level factors that may affect monitoring rates. This is the first known study to look at the rate of first-generation antipsychotic (FGA) monitoring compared with second-generation antipsychotic (SGA) monitoring. The primary study objectives were: (1) to determine the rate at which California Medicaid (Medi-Cal) patients initiated on antipsychotics are monitored for dyslipidemia and T2DM, as measured by clinical laboratory test claims, and (2) to examine the differences in testing services for dyslipidemia and T2DM among Medi-Cal patients started on SGAs compared with FGAs.


Data sources and elements

This study used computerized patient information from the California Medicaid (Medi-Cal) data files with service dates covering a 24-month period (January 1, 2004 through December 31, 2005). Medi-Cal databases used included: (1) the Medi-Cal eligibility file, and (2) the Medi-Cal claims database, which includes information on inpatient admissions, outpatient visits, procedures (including laboratory tests but not laboratory results), and medications dispensed. Claims and patients were linked using the unique client index number, an identifier similar to a Social Security number but encrypted to protect patient privacy. Patient-specific data elements extracted included age, gender, race, county of residence, and disability status. Claim-specific data elements extracted included date of service (including date of discharge for inpatient visits), International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes, and type of procedure (eg, laboratory test) performed. Specific data elements captured from medication claims included the name, National Drug Code (NDC), therapeutic class, quantity, and days’ supply of all medications dispensed on an inpatient and outpatient basis.


More than 20 million claims generated by 500,385 unique Medi-Cal beneficiaries who were not Medicare–Medi-Cal dually eligible or enrolled in a managed care health plan were queried for antipsychotic use. Patients were required to have been alive throughout the study period, continuously eligible for Medi-Cal benefits for 23 of the 24 study months, and between age 21 and 63 as of January 1, 2004.16 Nearly 70,000 (N = 69,116) patients who received at least a 30-day supply of an antipsychotic medication over the first 18 months (January 1, 2004 through June 30, 2005) of the study period were identified. To alleviate concerns about unreported drug use during institutional stays, 24,673 subjects with an inpatient admission any time during the 24-month study period were excluded.

Further, to restrict the study to patients newly started on antipsychotics, patients receiving an antipsychotic prescription in the first 6 months (January 1, 2004 through June 30, 2004) of the study (n = 37,838) were excluded. Lastly, 9 patients who were identified as beginning new antipsychotic therapy on clozapine were eliminated, as it was thought that these patients were most likely not antipsychotic naïve. These exclusions left a final analytic cohort of 6601 patients who were considered “new” antipsychotic users starting therapy between July 1, 2004 and June 30, 2005. Antipsychotic use could begin at any time during the 12-month period. Patients receiving more than one antipsychotic during the 12 months were categorized according to the first antipsychotic started.

Because the primary outcome construct was the presence of any level of patient glucose or lipid monitoring, a liberal schema with high sensitivity was used to identify patient monitoring. Specifically, glucose monitoring was operationalized as the presence of one of the following provider-ordered tests: comprehensive metabolic panel, glucose assay, glucose blood test, glucose tolerance test, glucose level post–glucose dose, glycosylated hemoglobin (HbA1c), or stick assay of blood glucose. Lipid monitoring was defined as the presence of one of the following tests: assay of triglycerides or a blood serum cholesterol or lipid profile. Rates of glucose and lipid testing were compared in the study cohort for 6 months prior to and 6 months post–initiation of antipsychotic therapy.

Analytic strategy

Comparisons between patients were made using chi-square tests for categorical variables and t tests for continuous variables. To examine the effect of specific patient-level factors on the likelihood of glucose and lipid testing, a multivariate logistic regression analysis was conducted. Patient covariants used in the adjusted logistic regression model included age, gender, race, marital status, being classified as disabled according to the Medi-Cal aid code, and living in Los Angeles (LA) County, as it has been suggested that residents in LA county may have access to different mental health services and utilization rates. Because patients with preexisting T2DM or dyslipidemia would be expected to receive increased glucose and lipid monitoring, a history of T2DM or dyslipidemia was identified using ICD-9-CM codes. Specifically, patients with an ICD-9-CM code (250.02 or 272.xx) associated with a Medi-Cal encounter in the 6 months preceding antipsychotic initiation were classified as having a prior T2DM or dyslipidemia diagnosis, respectively. Further, patients receiving oral hypoglycemic agents or insulin and lipid-lowering agents (eg, statins and fibrates) in the 6 months prior to antipsychotic therapy were also classified as having a history of T2DM and/or dyslipidemia. Use of these medications is thought to indicate the presence of moderate to severe T2DM and lipid dysfunction with a high sensitivity and specificity.17 Actual glucose and lipid testing in the 6 months preceding antipsychotic therapy was also controlled for using a separate indicator variable. Finally, to determine if the likelihood of monitoring varied among patients receiving SGAs compared with FGAs, the OR for the likelihood of testing associated with SGA agents was generated. ORs associated with individual SGAs compared with FGAs were also generated to determine if specific differences among SGAs existed. Ethical approval for this study was obtained from the Touro University Institutional Review Board. All analyses were conducted using SAS for Windows, Version 9.1 (SAS Institute; Cary, NC).


Baseline characteristics for the study cohort (new Medi-Cal antipsychotic users) are shown in TABLE 1, with SGA agents representing over 90% of all new antipsychotic therapy. Patients begun on SGAs (N = 6124) were similar to patients begun on FGAs (N = 477) regarding age, residing in LA County, and previous diagnoses of T2DM and dyslipidemia. However, SGA users were less likely to be male or disabled compared with FGA users (P < .01). Several differences in race were also observed; notably, SGA users were more likely to be white or Hispanic and less likely to be black compared with FGA users (P < .01). The most common FGA was haloperidol, which accounted for nearly half of all FGA initial therapy starts. Other common FGAs included fluphenazine (14%), thioridazine (13%), and chlorpromazine (9%). The most commonly used SGA was quetiapine (33%), followed by risperidone (28%), olanzapine (22%), aripiprazole (11%), and ziprasidone (6%). Among new antipsychotic users, rates of medication use for T2DM or lipid dysregulation during the preceding 6 months were 8% and 13%, respectively.

During the 6-month period prior to antipsychotic initiation, 24% of the study population underwent blood glucose laboratory testing, 39% underwent lipid testing, and 23% underwent testing for both glucose and lipids (FIGURE). During the 6-month period following antipsychotic initiation, 28% of the study population underwent laboratory blood glucose testing, 43% underwent lipid testing, and 27% underwent both glucose and lipid testing. However, these increases were not equally distributed across the FGA and SGA classes. The rates of testing rose only 1% for both glucose and lipid testing in FGA patients, whereas the rates of testing rose 4% for both glucose and lipid testing in SGA patients (TABLE 1). Absolute differences were even greater, with only 34% of FGA patients receiving glucose testing within 6 months post–therapy initiation, compared with 43% of the SGA patients who received glucose testing within 6 months post–therapy initiation (P < .01). A similar absolute difference was observed in lipid testing, with 22% of FGA patients receiving testing within 6 months post–therapy initiation, compared with 29% of SGA patients during a 6-month follow-up period (P < .01).


Baseline characteristics of “new” Medi-Cal antipsychotic users (N = 6601)

Patient characteristic New FGA user (N = 477) New SGA user (N = 6124) P value
Mean age ± SD, y 45.5 ± 10.7 45.0 ± 10.7 .33
Male, No. (%) 260 (54.5%) 2420 (39.5%) <.01
  White, No. (%) 171 (35.8%) 2669 (43.6%) <.01
  Hispanic, No. (%) 50 (10.5%) 861 (14.1%) .03
  Black, No. (%) 131 (27.5%) 1202 (19.6%) <.01
  Asian, No. (%) 54 (11.3%) 574 (9.4%) .16
  Other, No. (%) 71 (14.9%) 818 (13.4%) .35
LA County resident, No. (%) 155 (32.5%) 1961 (32.0%) .83
Disabled, No. (%) 420 (88.0%) 5062 (82.7%) <.01
Previous T2DM medication use 41 (8.6%) 473 (7.8%) .49
Previous lipid-lowering agent use 52 (10.9%) 783 (12.8%) .22
AP use
Haloperidol, No. (%) 236 (49.4%)    
Other FGA, No. (%) 241 (50.5%)    
Aripiprazole, No. (%)   675 (11.0%)  
Olanzapine, No. (%)   1375 (22.4%)  
Quetiapine, No. (%)   1995 (32.6%)  
Risperidone, No. (%)   1718 (28.0%)  
Ziprasidone, No. (%)   361 (5.9%)  
Patient glucose testing
6 mo prior to AP therapy 159 (33.3%) 2284 (38.9%) .02
6 mo post–AP therapy 164 (34.4%) 2655 (43.4%) <.01
Patient lipid testing
6 mo prior to AP therapy 98 (20.5%) 1501 (24.5%) .06
6 mo post–AP therapy 103 (21.6%) 1764 (28.8%) <.01
AP: antipsychotic; FGA: first-generation antipsychotic; LA: Los Angeles; SGA: second-generation antipsychotic; T2DM: type 2 diabetes mellitus.

FIGURE Glucose, lipid, and glucose-lipid testing rates among new Medi-Cal antipsychotic users during 6 months prior to and 6 months post–antipsychotic start date (N = 6601)

Patients were more likely to undergo glucose testing within 6 months of antipsychotic initiation if previously tested during the 6 months prior to initiation (OR, 1.81; 95% CI, 1.63 to 2.01; P < .01) (TABLE 2). Compared with patients beginning FGA therapy, patients initiated on SGA were more likely to undergo glucose testing within the 6 months after initiation (OR, 1.38; 95% CI, 1.13 to 1.70; P < .01). Compared with patients initiated on FGAs, patients were also more likely to receive glucose testing if initiated on any of the individual SGAs, including aripiprazole (OR, 1.32; 95% CI, 1.04 to 1.72; P = .03), olanzapine (OR, 1.35; 95% CI, 1.08 to 1.69; P = .02), quetiapine (OR, 1.35; 95% CI, 1.12 to 1.73; P < .01), risperidone (OR, 1.37; 95% CI, 1.10 to 1.71; P < .01), and ziprasidone (OR, 1.63; CI, 1.23 to 2.19; P < .01).


Odds for glucose testing done within 6 months after beginning new antipsychotic therapy

Patient characteristic OR 95% CI P value
Age 1.02 1.02 to 1.03 <.01
Male (referent: female) 0.74 0.67 to 0.82 <.01
Race (referent: white)
  Hispanic 1.11 0.98 to 1.34 .08
  Black 0.90 0.78 to 1.04 .19
  Asian 1.12 0.96 to 1.38 .12
  Other race 1.07 0.67 to 0.82 .43
LA County resident (referent: not an LA County resident) 1.03 0.92 to 1.15 .59
Disabled (referent: nondisabled) 0.85 0.74 to 0.98 .03
Glucose test in 6 months prior to starting AP therapy (referent: no prior test) 1.81 1.63 to 2.01 <.01
Prior T2DM diagnosis (referent: no prior T2DM diagnosis) 2.59 2.12 to 3.19 <.01
AP use
SGA (referent: FGA initiation) (N = 6124) 1.38 1.13 to 1.70 <.01
Individual SGA started (referent: FGA initiation)      
  Aripiprazole (n = 675) 1.32 1.04 to 1.72 .03
  Olanzapine (n = 1375) 1.35 1.08 to 1.69 .02
  Quetiapine (n = 1995) 1.35 1.12 to 1.73 <.01
  Risperidone (n = 1718) 1.37 1.10 to 1.71 <.01
  Ziprasidone (n = 361) 1.63 1.23 to 2.19 <.01
AP: antipsychotic; CI: confidence interval; FGA: first-generation antipsychotic LA: Los Angeles; OR: odds ratio; SGA: second-generation antipsychotic; T2DM: type 2 diabetes mellitus.
Multivariate model controlling for age, gender, race, disability status, residence in LA County, glucose testing in 6 months prior to beginning therapy, and prior T2DM and dyslipidemia diagnosis (N = 6601).

Similarly, patients were more likely to undergo lipid testing within 6 months of antipsychotic initiation if previously tested during the 6 months prior to beginning therapy (OR, 1.49; 95% CI, 1.31 to 1.70; P < .01) (TABLE 3). Compared with patients initiated on FGAs, patients initiated on SGAs were more likely to undergo lipid testing within the 6 months after beginning therapy (OR, 1.43; 95% CI, 1.14 to 1.81; P < .01). Compared with FGA starters, patients were more likely to have lipid testing if initiated on any of the individual SGAs, including aripiprazole (OR, 1.36; 95% CI, 1.02 to 1.81; P=.04), olanzapine (OR, 1.46; 95% CI, 1.13 to 1.89; P < .01), quetiapine (OR, 1.48; 95% CI, 1.15 to 1.90; P < .01), risperidone (OR, 1.35; 95% CI, 1.05 to 1.73; P=.02), or ziprasidone (OR, 1.56; 95% CI, 1.12 to 2.16; P < .01).


Odds for clinical lipid testing done within 6 months after beginning new antipsychotic therapy

Patient characteristic OR 95% CI P value
Age 1.03 1.03 to 1.04 <.01
Male (referent: female) 0.82 0.73 to 0.92 <.01
Race (referent: white)
  Hispanic 1.18 0.99 to 1.41 .06
  Black 0.99 0.80 to 1.17 .97
  Asian 1.43 1.19 to 1.73 <.01
  Other 1.10 0.92 to 1.30 .30
LA County resident (referent: not an LA County resident) 1.19 1.06 to 1.35 <.01
Disabled (referent: nondisabled) 0.94 0.80 to 1.11 .48
Lipid test in 6 months prior to starting AP therapy (referent: no prior test) 1.49 1.31 to 1.70 <.01
Prior dyslipidemia diagnosis (referent: no prior dyslipidemia diagnosis) 2.59 2.20 to 3.06 <.01
Ap use
SGA (referent: FGA initiation) (N = 6124) 1.43 1.14 to 1.81 <.01
Individual SGA started (referent: FGA initiation)
  Aripiprazole (n = 675) 1.36 1.02 to 1.81 .04
  Olanzapine (n = 1375) 1.46 1.13 to 1.89 <.01
  Quetiapine (n = 1995) 1.48 1.15 to 1.90 <.01
  Risperidone (n = 1718) 1.35 1.05 to 1.73 .02
  Ziprasidone (n = 361) 1.56 1.12 to 2.16 <.01
AP: antipsychotic; CI: confidence interval; FGA: first-generation antipsychotic; LA: Los Angeles; OR: odds ratio; SGA: second-generation antipsychotic; T2DM: type 2 diabetes mellitus.
Multivariate model controlling for age, gender, race, disability status, residence in LA County, lipid testing in 6 months prior to beginning therapy, and prior T2DM and dyslipidemia diagnosis (N = 6601).

Finally, patients were more likely to undergo both glucose and lipid testing within 6 months of antipsychotic initiation if previously monitored for lipids (OR, 1.25; 95% CI, 1.05 to 1.49; P < .01) but not for glucose (OR, 1.11; 95% CI, 0.95 to 1.30; P =.19) (TABLE 4). Compared with patients initiated on FGAs, patients initiated on SGAs were more likely to undergo both glucose and lipid testing after beginning therapy (OR, 1.40; 95% CI, 1.11 to 1.79; P < .01). Patients were also more likely to have both glucose and lipid testing if initiated on any of the individual SGAs, including aripiprazole (OR, 1.37, 95% CI, 1.02 to 1.83; P=.03), olanzapine (OR, 1.44; 95% CI, 1.11 to 1.87; P < .01), quetiapine (OR, 1.44; 95% CI, 1.11 to 1.87; P < .01), risperidone (OR, 1.31; 95% CI, 1.01 to 1.69; P=.04), or ziprasidone (OR, 1.57; 95% CI, 1.12 to 2.18; P =.04), compared with FGAs.


Odds for both clinical T2DM and lipid testing done within 6 months after beginning new antipsychotic therapy

Patient characteristic OR 95% CI P value
Age 1.02 1.01 to 1.03 <.01
Male (referent: female) 0.81 0.72 to 0.91 <.01
Race (referent: white)
  Hispanic 1.15 0.96 to 1.37 .12
  Black 0.99 0.85 to 1.18 .97
  Asian 1.42 1.16 to 1.72 <.01
  Other 1.05 0.88 to 1.25 .59
LA County resident (referent: not an LA County resident) 1.25 1.10 to 1.41 <.01
Disabled (referent: nondisabled) 0.97 0.81 to 1.14 .71
T2DM test in 6 months prior to starting AP therapy (referent: no prior test) 1.11 0.95 to 1.30 .19
Prior T2DM diagnosis (referent: no prior T2DM diagnosis) 1.61 1.31 to 1.96 <.01
Lipid test in 6 months prior to starting AP therapy (referent: no prior test) 1.25 1.05 to 1.49 <.01
Prior dyslipidemia diagnosis (referent: no prior dyslipidemia diagnosis) 2.14 1.81 to 2.54 <.01
AP use
SGA (referent: FGA initiation) (N = 6124) 1.40 1.11 to 1.79 <.01
Individual SGA started (referent: FGA initiation)
  Aripiprazole (n = 675) 1.37 1.02 to 1.83 .03
  Olanzapine (n = 1375) 1.44 1.11 to 1.87 <.01
  Quetiapine (n = 1995) 1.44 1.11 to 1.87 <.01
  Risperidone (n = 1718) 1.31 1.01 to 1.69 .04
  Ziprasidone (n = 361) 1.57 1.12 to 2.18 .04
AP: antipsychotic; CI: confidence interval; FGA: first-generation antipsychotic; LA: Los Angeles; OR: odds ratio; SGA: second-generation antipsychotic; T2DM: type 2 diabetes mellitus.
Multivariate model controlling for age, gender, race, disability status, residence in LA County, T2DM and lipid testing in 6 months prior to beginning therapy, and prior T2DM and dyslipidemia diagnosis (N = 6601).

For each logistic regression analysis, model calibration (ie, goodness of fit) was assessed by the Hosmer-Lemeshow statistic, which compares the observed and predicted values for the regression models in ordered deciles.18 The Hosmer-Lemeshow statistics for each of the multivariate models were nonsignificant (P > .05), indicating that all models had excellent calibration. In addition, model discrimination was assessed by the C statistic, which determines the proportion of times patients who received a laboratory test had a higher predicted rate of testing than patients who did not receive testing.19 The C statistics ranged from 0.67 to 0.69, indicating moderately good model discrimination.


Antipsychotic medications are widely used for treating a variety of neuropsychiatric and behavioral problems in the adult population. Although used for decades, the use of FGAs was greatly reduced after the introduction of SGAs in the 1980s and 1990s. The sharp reduction in use of FGAs was driven, at least in part, by the relative decrease in the risk of serious adverse movement effects associated with SGAs compared with FGAs. With their comparatively minimal extrapyramidal adverse effect profile, SGAs quickly became the preferred treatment option.20,21 The widespread use of SGAs as first-line agents, however, has been called into question in recent years. Precipitating the scrutiny of SGA use are studies suggesting a lack of improved efficacy of SGAs compared with FGAs22 as well as possible increased risks for weight gain, T2DM, lipid disorders,23 and cerebrovascular events,24 among other symptoms in SGA users compared with FGA users.

As a direct result of these concerns, clinical monitoring guidelines for patients receiving antipsychotic therapy have been published by numerous organizations, including ADA and the APA. Cohn et al reviewed the 6 sets of currently available antipsychotic monitoring guidelines.25 There is consensus among the guidelines that antipsychotic monitoring, but not necessarily medical treatment of metabolic disorders, falls within the scope of psychiatric practice and should include monitoring for metabolic disturbances as well as tracking the effects of antipsychotic treatment. The ADA-APA guidelines recommend testing of fasting plasma glucose levels (at baseline, 12 weeks, then annually) and a fasting lipid profile (at baseline, 12 weeks, then every 5 years if normal). Despite these recommendations by expert organizations, the clinical implementation of monitoring varies widely among patient populations and has yet to be fully quantified. Of particular concern are rates of glucose and lipid testing among public assistance patients receiving antipsychotics, as these patients appear to have greater comorbidity and shorter life spans compared with private sector patients.2 This study is the first attempt to describe the level of glucose and lipid testing in a cohort of public assistance patients beginning antipsychotic therapy.

Two clinically relevant questions were put forth in this study. The first, regarding the rate at which Medi-Cal outpatients started on antipsychotics are monitored for T2DM and dyslipidemia, was answered by comparing rates of testing of glucose, lipid, and glucose-lipid levels 6 months after beginning therapy vs 6 months prior to therapy. A 4% increase was found for each testing measure, namely 39% vs 43%, 24% vs 28%, and 27% vs 23%, for glucose, lipid, and glucose-lipid testing, respectively. The second question, regarding differences in monitoring between patients started on FGAs or SGAs, found that patients started on SGAs were approximately 40% more likely to receive post-therapy testing compared with patients started on FGAs in the multivariate model controlling for patient-specific factors. Further, these findings were relatively consistent when comparing the likelihood of testing for each of the individual agents vs FGAs. Somewhat curious was the observation that testing was approximately 20% less likely among male patients. Although it is unknown if this simply reflects unmeasured comorbidity or other factors related to males in this population, this finding deserves future investigation.

Although the findings of the current study may appear low, the observed testing rates for Medi-Cal patients are actually similar to the 2 published private sector studies.15,26 More specifically, while looking only at 97 HMO patients who were newly prescribed select SGAs (risperidone, quetiapine, olanzapine, or aripiprazole) during a similar time period as the current study, Olson et al found that only 7% and 17% of patients received baseline (90 days prior to or 7 days after antipsychotic initiation) fasting blood glucose and fasting lipoprotein chemistries, respectively. Although the primary outcome of the Olson et al study was testing rates occurring over a 6-month time frame, an analysis of the 6601 Medi-Cal individuals revealed that 26.0% received a glucose test, whereas 15.3% received a lipid test either within 90 days of or 7 days after starting antipsychotic therapy. This comparison suggests similar rates for lipid monitoring and better rates for glucose monitoring among the Medi-Cal population beginning antipsychotic therapy compared with private sector HMO patients.

The second study examined rates of glucose and lipid monitoring among antipsychotic users, utilizing claims data from 85 health plans.15 The authors found rates of baseline glucose and lipid testing to be 7.8% and 20.6%, respectively, prior to the publication of the 2004 consensus statement, with only a slight improvement to 8.5% and 22.5% following its publication. Unlike the current research, the authors reported that follow-up rates declined after antipsychotic treatment was initiated, but they have not yet published the actual numbers.

Two large non–private sector studies have recently been published. The first, a retrospective cohort study of 1826 Veterans Affairs (VA) patients from 2001 to 2003 found that nearly 39% of SGA patients had at least one lipid-monitoring procedure prior to an SGA switch, whereas 59% had a lipid-monitoring procedure within 12 months after the switch.27 The same study found that 57% of VA SGA patients had at least one glucose-monitoring procedure prior to an SGA switch, whereas 80% had a glucose test performed within 12 months after the switch. It should be noted, however, that this study was limited to VA patients diagnosed with schizophrenia who had been maintained on an SGA for a minimum of 90 days and who were subsequently switched to another SGA. In addition, this study was conducted using data collected before the current guidelines and before widespread dissemination of the FDA warning. Further, unlike the current study, the authors did not look at rates of glucose or lipid monitoring among FGA patients.

The second large retrospective study was conducted on 55,436 Medicaid patients from 4 states between 1998 and 2003 by Morrato et al.28 Over the 6-year study period, the authors noted that only 19% of all patients initiating SGA therapy received baseline glucose monitoring, whereas only 6% received baseline lipid monitoring. Baseline monitoring was defined as 14 days prior to or 28 days after the start of SGA therapy. Expanding the definition of baseline to 4 months before and after initiation, the authors noted an 11% increase in glucose monitoring and a 3% increase in lipid testing. No attempt was made by the authors to follow SGA patients beyond 4 months. Like the current study, Morrato et al noted an increase of greater than twofold in the rate of glucose and lipid testing in patients with preexisting T2DM or dyslipidemia. Similar to the previously discussed VA study, the Morrato Medicaid study was limited to SGA patients and conducted on data before the release of the current guidelines and the FDA warning.

The current study highlights that, despite recent guideline recommendations and FDA warnings, new users of antipsychotic medications are not receiving simple metabolic monitoring tests. Of particular concern are patients without a prior history of T2DM or dyslipidemia, as these new users appear to be less than twice as likely to receive appropriate monitoring. Also disconcerting are the rates of monitoring for new FGA users. As noted previously, new FGA users were significantly less likely to receive post-therapy metabolic testing than were patients starting on SGA therapy. This observation verifies an apparent clinical misperception among clinicians prescribing antipsychotics. These findings suggest that many clinicians associate antipsychotic weight gain and metabolic changes only with SGAs. However, from a pharmacologic perspective, individual antipsychotic histamine-1 receptor affinities (antihistaminic potency) are the most robust predictor of weight gain, regardless of whether the antipsychotic is an FGA or an SGA.29 In a recent study using contingency table analysis, the antihistaminic/weight-positive antipsychotics were accurately separately from the nonantihistaminic/weight-neutral antipsychotics (Fisher exact test, P =.003). The antihistaminic antipsychotics included chlorpromazine, clozapine, olanzapine, perphenazine, quetiapine, thioridazine, and thiothixene. The nonantihistaminic antipsychotics included aripiprazole, fluphenazine, haloperidol, molindone, pimozide, thiothixene, trifluoperazine, and ziprasidone. Loxapine and risperidone did not fit the categorization paradigm. These data lead to the conclusion that clinicians need to routinely monitor the metabolic activity of patients taking any antipsychotic, as antihistaminic antipsychotics include both FGA and SGA agents.30 Indeed, this is the recommendation of the ADA-APA consensus guidelines.

In interpreting our findings, it is important to consider several potential limitations. First, the study involved only adult Medi-Cal patients. The generalizability to the larger population of all patients receiving antipsychotics, especially those age 65 and older, may be in question. Second, although diagnosis-based methods have been found to explain a large proportion of resource utilization, such methods are dependent on the accuracy of captured claims from patient encounters and may be subject to systematic variations between practitioners and facilities.31 In addition, pharmacy data have their own set of potential limitations.16,32 For example, the data did not allow for an accurate dose measurement, so factoring in a dose-response relationship with likelihood of monitoring was not possible. Specifically, prescribers may have discounted monitoring in patients receiving short-term, low-dose quetiapine for sleep. However, a sensitivity analysis removing quetiapine users found similar rates of pre- and postantipsychotic therapy monitoring for both glucose and lipids.

Third, there was no way of appropriately determining if the laboratory tests performed were indicated specifically because the patients began antipsychotic therapy, or that they were ordered to monitor other preexisting medical conditions, as a standardized procedure for regular routine monitoring, or some combination of these possibilities. Further, neither of the 6-month pre- or post–6-month time periods analyzed conforms to actual recommended time frames (eg, baseline, 12 weeks, annually); rather, the pre- and post–6-month time frames were designed to estimate the likelihood that initiation of antipsychotic therapy would increase monitoring of glucose and lipid values in this patient population and to discern relevant patient level factors that influence monitoring. It was also impossible to determine if all of the patients identified as new antipsychotic users were truly new users or, rather, patients receiving antipsychotics through other agencies who newly began receiving antipsychotics through Medi-Cal. Lastly, there was no way to determine if the clinical laboratory values associated with the monitoring claim were read, properly interpreted, and acted upon, if necessary, by providers. As mentioned previously, the data contained laboratory claims for these patients but did not contain the results of the tests that were billed.


Although the mental health patient population receiving public assistance has poorer outcomes and higher resource utilization of medical services, the underlying reasons for these differences remain unknown. However, the use of preventive monitoring services, especially among patients taking antipsychotics, would appear to be a particularly critical therapeutic drug monitoring intervention. Current therapeutic drug monitoring data indicate that educational strategies are needed to target specific cohorts of antipsychotic providers. Specifically, interventions may be needed to ensure that patients started and maintained on antipsychotics receive adequate monitoring for T2DM and dyslipidemia. Ideally, these targeted interventions will ensure that any testing and monitoring services are closely integrated with a patient’s mental health care. Because antipsychotic prescriptions are generally dispensed at 1-month intervals, one such approach could involve monitoring by dispensing pharmacists already trained in medication therapy management (MTM) services. In this scenario, MTM pharmacists would receive a reminder generated at the time of electronic claim transmission that the patient requires glucose and/or lipid testing, and would then arrange for the patient to receive a fingerstick test at the pharmacy. Although current fingerstick technology is not robust enough to allow an actual diagnosis, both blood glucose and lipid fingersticks are inexpensive, relatively easy to perform, and accurate enough to be used as valuable testing tools. Properly trained pharmacists would be the ideal health care professionals to perform periodic testing procedures, as diabetes and lipid dysregulation can develop at any time during antipsychotic treatment, and patients interact with a pharmacist on a regular basis when prescription refills are dispensed.32

ACKNOWLEDGEMENTS: The authors would like to acknowledge Danielle M. Richardson, BS, PharmD candidate, for her assistance with the preparation of this manuscript.

DISCLOSURES: Drs. Barnett, McDonald, Popish, Von-Muenster, and Wehring and Mr. Walker report no financial relationship with any company whose products are mentioned in this article, or with manufacturers of competing products. Dr. Perry is a speaker for Teva Pharmaceuticals.


  1. Harris EC, Barraclough B. Excess mortality of mental disorder. Br J Psychiatry. 1998;173:11–53.
  2. Colton CW, Manderscheid RW. Congruencies in increased mortality rates, years of potential life lost, and causes of death among public mental health clients in eight states. Prev Chronic Dis. 2006;3:A42. [Epub March 15, 2006.] www.cdc.gov/pcd/issues/2006/apr/05_0180.htm.
  3. Hennekens CH, Hennekens AR, Hollar D, et al. Schizophrenia and increased risks of cardiovascular disease. Am Heart J. 2005;150:1115–1121.
  4. Goff DC, Sullivan LM, McEvoy JP, et al. A comparison of ten-year cardiac risk estimates in schizophrenia patients from the CATIE study and matched controls. Schizophr Res. 2005;80:45–53.
  5. Lamberti JS, Olson D, Crilly JF, et al. Prevalence of the metabolic syndrome among patients receiving clozapine. Am J Psychiatry. 2006;163:1273–1276.
  6. Druss BG, Rosenheck RA. Use of medical services by veterans with mental disorders. Psychosomatics. 1997;38:451–458.
  7. Daumit GL, Crum RM, Guallar E, et al. Receipt of preventive medical services at psychiatric visits by patients with severe mental illness. Psychiatr Serv. 2002;53:884–887.
  8. Dickerson FB, Pater A, Origoni AE. Health behaviors and health status of older women with schizophrenia. Psychiatr Serv. 2002;53:882–884.
  9. Druss BG, Rosenheck RA, Desai MM, et al. Quality of preventive medical care for patient with mental disorders. Med Care. 2002;40:129–136.
  10. Lindamer LA, Buse DC, Auslander L, et al. A comparison of gynecological variables and service use among older women with and without schizophrenia. Psychiatr Serv. 2003;54:902–904.
  11.  American Diabetes Association, American Psychiatric Association, American Association of Clinical Endocrinologists, et al. Consensus development conference on antipsychotic drugs and obesity and diabetes. Diabetes Care. 2004;27:596–601.
  12.  The CDC Diabetes Cost-Effectiveness Study Group. The cost-effectiveness of screening for type 2 diabetes. JAMA. 1998;280:1757–1763.
  13. Roche-Desilets J, Fuller MA, Konicki PE. Incorporating clinical monitoring into electronic orders for antipsychotics. Psychiatr Serv. 2004;55:455.
  14. Newcomer JW. Metabolic considerations in the use of antipsychotic medications: a review of recent evidence. J Clin Psychiatry. 2007;68:S20–S27.
  15. Rosack J. Clinicians urged to better monitor drug-related side effects. Psychiatr News. 2006;41:1–34.
  16. Crystal S, Akincigil A, Bilder S, et al. Studying prescription drug use and outcomes with Medicaid claims data: strengths, limitations, and strategies. Med Care. 2007;45:S58–S65.
  17. Ray K, Barnett MJ, Kaboli PJ, et al. Use of VA prescription drug data to identify DM-II patients with previous comorbid diseases. Paper presented at: VA Health Services and Research and Development National Meeting; February 16-18, 2005; Baltimore, MD.
  18. Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol. 1982;115:92–106.
  19. Ash AS, Schwartz M. Evaluating the performance of risk-adjustment methods: dichotomous measures. In: Iezzoni LI, ed. Risk adjustment for measuring health care outcomes. 2nd ed. Chicago, IL: Health Administration Press; 1997:427-470.
  20.  Expert Consensus Panel for Dementia. The expert consensus guideline series. Treatment of dementia and its behavioral disturbances. Postgrad Med. 2005;Jan:1–111.
  21. Carson S, McDonagh MS, Peterson K. A systematic review of the efficacy and safety of atypical antipsychotics in patients with psychological and behavioral symptoms of dementia. J Am Geriatr Soc. 2006;54:354–361.
  22. Lieberman JA, Stroup TS, McEvoy JP, et al. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med 2005;353:1209–1223.
  23. Lund BC, Perry PJ, Brooks JM, et al. Clozapine use in patients with schizophrenia and the risk of diabetes, hyperlipidemia, and hypertension: a claims-based approach. Arch Gen Psychiatry. 2001;58:1172–1176.
  24. Barnett MJ, Wehring H, Perry PJ. Comparison of risk of cerebrovascular events in an elderly VA population with dementia between antipsychotic and non-antipsychotic users. J Clin Psychopharmacol. 2007;6:595–601.
  25. Cohn TA, Sernyak MJ. Metabolic monitoring for patients treated with antipsychotic medications. Can J Psychiatry. 2006;51:492–501.
  26. Olson KL, Delate T, Duagn DJ. Monitoring of patients given second-generation antipsychotic agents [letter]. Psychiatr Serv. 2006;57:1045–1046.
  27. Hsu C, Ried LD, Bengtson MA, et al. Metabolic monitoring in veterans with schizophrenia-related disorders and treated with second-generation antipsychotics: findings from a Veterans Affairs-based population. J Am Pharm Assoc (2003). 2008;48:393–400.
  28. Morrato EH, Newcomer JW, Allen RR, et al. Prevalence of baseline serum glucose and lipid testing in users of second-generation antipsychotic drugs: a retrospective, population-based study of Medicaid claims data. J Clin Psychiatry. 2008;69:316–322.
  29. Kroeze WK, Hufeisen SJ, Popadak BA, et al. H1-histamine receptor affinity predicts short-term weight gain for typical and atypical antipsychotic drugs. Neuropsychopharmacology. 2003;28:519–526.
  30. Allison DB, Mentore JL, Heo M, et al. Antipsychotic-induced weight gain: a comprehensive research synthesis. Am J Psychiatry. 1999;156:1686–1696.
  31. Iezzoni LI. The risks of risk adjustment. JAMA. 1997;278:1600–1607.
  32. Kaboli PJ, McClimon BJ, Hoth A, et al. Assessing the accuracy of computerized medication histories. Am J Manag Care. 2004;11:872–877.

CORRESPONDENCE: Paul J. Perry, PhD, Professor and Chair, Pharmacy Practice Department, Touro-University California, College of Pharmacy, 1310 Johnson Lane, Mare Island, Vallejo, CA 94592 USA, E-MAIL: pjperry2@comcast.net