Racial/Ethnic Disparities in Monitoring Metabolic Parameters for Patients with Schizophrenia Receiving Antipsychotic Medications Karon L. Phillips, Ph.D., M.P.H., Laurel A. Copeland, Ph.D., M.P.H., John E. Zeber, Ph.D., M.H.A., Eileen M. Stock, Ph.D., Jack Y. Tsan, Ph.D., Andrea A. MacCarthy, B.S.

Objective: Patients with schizophrenia experience risks for metabolic dysregulation from medications and lifestyle behaviors. Although most patients with schizophrenia in the Veterans Health Administration (VA) receive antipsychotics, variation in monitoring metabolic dysregulation by race/ethnicity has not been assessed. This study analyzed differential monitoring of metabolic parameters by minority status. Methods: This retrospective study approximated the five components of metabolic syndrome (fasting glucose, high-density-lipoprotein cholesterol, triglycerides, blood pressure, and large waistline) using archival data, substituting body mass index for waistline. VA patients with schizophrenia age 50 or older were followed from October 1, 2001 through September 2009 (N ¼ 30,258). Covariates included age, gender, race (white, black), Hispanic ethnicity, region, marital status, VA priority status, comorbidity, and antipsychotic type. Repeated-measures analysis assessed the association of race/ethnicity with metabolic monitoring. Results: Average patients age was 59 years (standard deviation: 9; range: 50e101), 97% were men, 70% white, 30% black, and 8% Hispanic. At baseline, 6% were monitored on all five metabolic components; this increased to 29% by 2005. In adjusted models, blacks were less likely to be monitored on all parameters, whereas Hispanics were less likely to have glucose and high-density-lipoprotein cholesterol monitored but more likely to have triglycerides tested. By 2009, lab assays were similar across race and ethnicity. Conclusion: Guideline-concordant monitoring metabolic parameters appear to be equitable but low and somewhat at odds with racial/ethnic risk among older patients with schizophrenia. Physicians should discuss lipids, weight, and glucose with patients at risk for developing heart disease, diabetes, and other sequelae of the metabolic syndrome. (Am J Geriatr Psychiatry 2014; -:-e-)

Received November 21, 2013; revised July 21, 2014; accepted July 24, 2014. From IMPAQ International (KLP), Columbia, MD; Center for Applied Health Research (LAC, JEZ, EMS), Central Texas Veterans Health Care System, jointly with Scott & White Healthcare, Temple, TX; Central Texas Veterans Health Care System (JYT), Temple, TX; and South Texas Veterans Health Care System (AAM), San Antonio, TX. Send correspondence and reprint requests to Laurel A. Copeland, Ph.D., M.P.H., Center for Applied Health Research, 2102 Birdcreek Dr., Temple, TX 76502. e-mail: [email protected] Ó 2014 American Association for Geriatric Psychiatry http://dx.doi.org/10.1016/j.jagp.2014.07.007

Am J Geriatr Psychiatry

-:-, -



Variation in Monitoring Metabolic Dysregulation by Race/Ethnicity

Key Words: African Americans, healthcare quality assessment, Hispanic, metabolic syndrome, schizophrenia, veterans

INTRODUCTION Metabolic syndrome comprises five conditions that are associated with poor health outcomes: high serum glucose, low high-density-lipoprotein (HDL) cholesterol, elevated triglycerides, high blood pressure, and large waistline.1 An estimated 50 million people in America have metabolic dysregulation.1e3 Older adults, in particular, are at risk for developing the condition that is linked to increased risk of cardiovascular disease and premature death.4 In addition, individuals with schizophrenia are more likely to have metabolic syndrome than the general population.5 Consequently, older patients diagnosed with schizophrenia should be closely monitored for metabolic risk factors. A Consensus Statement was issued jointly by several groups, including the American Diabetes Association and the American Psychiatric Association, in February 2004 to address metabolic sequelae of antipsychotic medications.6 The Consensus recommended a raft of monitoring activities for patients prescribed antipsychotics. Second-generation antipsychotics (atypicals) used to treat patients with schizophrenia are associated with weight gain, diabetes, and dyslipidemia. First-generation antipsychotics do not alleviate negative symptoms such as blunted affect and may induce distressing and even permanent side effects (dystonic reactions, drug-induced parkinsonism, akathisias, and tardive dyskinesia).7 Therefore, atypicals are commonly prescribed for the treatment of schizophrenia. Accordingly, monitoring symptoms of metabolic disturbance is important to preventing disease onset and progression in at-risk persons. In a previous study of older patients with schizophrenia receiving care in the Veterans Health Administration (VA) during 2002, about one-third received conventional antipsychotics and two-thirds received second-generation agents. Several patients were treated with drugs from both classes, whereas others were prescribed no antipsychotic medications;


modest increases in monitoring metabolic parameters were noted over time.8 A unique subpopulation that provides an additional context for studying metabolic syndrome is racial/ethnic minorities. Some evidence shows of differences in the diagnosis and treatment of schizophrenia among racial/ethnic minorities.9e11 Racial/ethnic minority patients also have higher rates of metabolic syndrome when compared with the general population.2,3 Thus, older racial/ethnic minorities with schizophrenia taking atypicals may be expected to have very high risk for developing metabolic syndrome. Information on Hispanics may be particularly lacking, and the VA treats a large numbers of Hispanic patients. Reducing potential inequities through closer attention to clinical guidelines represents both an overall public health priority and important mission for the VA and other healthcare systems. Monitoring the components of metabolic syndrome should be at least as common among racial/ ethnic minority patients as among their nonminority counterparts. Studying this issue helps provide information on the quality of care for older racial/ ethnic minorities and permits a test of the hypothesis that care was equivalent for minority race/ethnic patients relative to white patients on antipsychotics and reveals changes over time in monitoring patients on these drugs. The purpose of this study was to assess whether changes in the monitoring of metabolic parameters from 2002 to 2009 differed by minority versus nonminority race/ethnic status among older VA patients with schizophrenia prescribed antipsychotic medications.

METHODS Study Sample VA patients aged 50 or older and diagnosed with schizophrenia in fiscal year 2002 (FY02, October 2001 to September 2002) were identified from a larger study

Am J Geriatr Psychiatry

-:-, -


Phillips et al. of late-life patients; the original study included patients with schizophrenia and/or diabetes, whereas the current analysis examined only patients with schizophrenia.8,12 Schizophrenia was defined by diagnosis of schizophrenia (International Classification of Diseases, Ninth Revision code 295.x excluding 295.5) on (1) one inpatient date or (2) two or more outpatient dates of care in FY02.13 Late life was defined as age 50 or older because schizophrenia significantly shortens life by 10e25 years relative to the life span of persons without schizophrenia.14e17 Inclusion criteria were (1) receipt of antipsychotic medication during the baseline year (FY02), (2) absence of diabetes at baseline, (3) valid data on race/ethnicity, and (4) no hypoglycemic medications prescribed during baseline. A sample of 30,258 veterans met criteria for study inclusion. Measures Dependent variables. Dependent measures included presence of A1c/fasting glucose tests, lipid tests, blood pressure readings, and height-to-weight data to calculate body mass index (BMI). These measures approximate the five components of the metabolic syndrome: fasting glucose, HDL cholesterol, triglycerides, blood pressure, and obesity.18 Patients with missing data on any of these measures were coded as not monitored for those parameters. Patients who had died the prior year(s) were coded as missing on subsequent years’ monitoring indicators. Monitoring was assessed each year for each parameter. Serum glucose monitoring was assessed each year. Because patients were not diagnosed with diabetes at study entry, A1c assessments were rare. In addition, fasting status was not attached to glucose test results. Therefore, we used a proxy for fasting glucose developed in a previous study: a random glucose test paired with a low-density-lipoprotein cholesterol test; annual receipt was indicated (yes or no).19 Having a valid triglycerides test was assessed each year, as was having a valid HDL cholesterol lab result. Having at least one valid blood pressure reading was assessed each year. BMI was ascertained from the VA Corporate Data Warehouse, which includes heights and weights recorded in the VA’s electronic medical records system. Multiple measures were aggregated per methods outlined by Noel et al.,20 and the baseline BMI was essentially the median of the quarterly median weights divided by the modal height squared (in English units,

Am J Geriatr Psychiatry

-:-, -


[pounds  703]/[inches-squared]). Height was taken from the first available record, because it was frequently missing. Having a valid weight indicated that obesity was monitored (yes or no; each year FY02eFY05). Noting that lab testing was both less common and more variable than monitoring BMI and blood pressure, we also examined 2009 follow-up test rates for serum glucose, HDL cholesterol, and triglycerides among patient surviving through FY09. To determine whether monitoring metabolic syndrome components was similar to other monitoring, we also assessed receipt of three liver function tests (g-glutamyltransferase [GGTP], alanine aminotransferase [ALT], and aspartate aminotransferase [AST]) each year. Liver function tests are often ordered for patients taking hepatotoxic drugs or otherwise at risk for liver damage and deterioration. Independent variables. Demographic measures included age, race operationalized as most commonly recorded value from all available healthcare records (white, black), Hispanic ethnicity (yes or no), gender, baseline marital status (married versus other), and VA priority status from enrollment data. Priority status is determined by a veteran’s disability from health disorders that are related to military service, poverty status, and special wartime experiences (such as Purple Heart recipient) and is associated with both severity of illness and socioeconomic status.21e23 Priority 1 veterans have no copays, Priority 2e6 generally have copayments for pharmacy only, and Priority 7e8 have agreed to copayments for care and pharmacy. From these eight categories assigned by VA, we made indicators of highly service-connected (50%e100% disabled by service-connected disorders; Priority 1), very low income (Priority 5), and a combination of catastrophically disabled (Priority 4) or other status. Census region, with values of Northeast, Midwest, South, West, and Puerto Rico/Virgin Islands, was determined from each patient’s most commonly used healthcare facility at baseline. Use of VA care each year was also assessed. Case-mix adjustment was addressed by the Selim Physical Comorbidity Score, assessing 30 chronic conditions validated for construction from VA administrative data in prior studies.23,24 Some patients were only on first-generation (conventional) antipsychotics, considered to have less metabolic impact; therefore, an indicator of this baseline status was created. Behavioral disorders other than schizophrenia were assessed: bipolar disorder, post-traumatic stress


Variation in Monitoring Metabolic Dysregulation by Race/Ethnicity

TABLE 1. Veterans with Schizophrenia Age 50D Years with No Diabetes in FY02 (N [ 30,258) White (N [ 21,180)

Total Sample

Age (range: 50e101) Selim Physical Comorbidity Score (range: 0e13) Race White Black Hispanic Female Married Region Midwest Northeast South West Puerto Rico/Virgin Islands Priority status 1: 50% service-connect disability 5: very low income, no copay All other statuses Baseline diagnoses Hypertension Dyslipidemia Typical antipsychotics only Mental disorders Bipolar PTSD MDD SUD Died during study period

Black (N [ 9,078)

Mean or N

SD or %

Mean or N

SD or %

Mean or N

SD or %

59.0 1.7

8.9 1.6

59.6 1.7

9.1 1.7

57.5 1.7

8.3 1.6

21,180 9,078 2,469 1,051 7,381

70 30 8.2 3.5 24.4

1,250 769 5,002

5.9 3.6 23.6

1,219 282 2,379

13.4 3.1 26.2

6,702 6,661 10,563 5,180 1,152

22.2 22.0 34.9 17.1 3.8

5,014 5,232 6,359 4,006 569

23.7 24.7 30.0 18.9 2.7

1,688 1,429 4,204 1,174 583

18.6 15.7 46.3 12.9 6.4

15,842 5,665 8,751

52.4 18.7 28.9

10,816 4,118 6,246

51.1 19.4 29.5

5,026 1,547 2,505

55.4 17.0 27.6

11,228 7,440 6,760

37.1 24.6 22.3

7,396 5,659 4,607

34.9 26.7 21.8

3,832 1,781 2,153

42.2 19.6 23.7

2,630 2,490 2,153 2,955 4,275

8.7 8.2 7.1 9.8 14.1

2,083 1,547 1,534 1,654 3,243

9.8 7.3 7.2 7.8 15.3

547 943 619 1,301 1,032

6.0 10.4 6.8 14.3 11.4

Notes: SD: standard deviation; MDD: major depression.

disorder (PTSD), major depression, and substance use disorders (SUD). Death during the study period was determined from VA’s aggregated mortality data, which has 98% sensitivity and is based on Social Security Administration, Centers for Medicare & Medicaid Services, VA Death Benefits claims, and VA inpatient death data.25 Analysis Plan After presenting descriptive statistics for the total sample and for minority versus nonminority groups, we assessed the association of minority race (black versus white) and minority ethnicity (Hispanic versus non-Hispanic) with repeated measures of each of the five metabolic monitoring measures controlling for patient characteristics (age in decades, female gender, married at baseline, priority status, Selim Physical Comorbidity Score), comorbid mental disorders (bipolar disorder, PTSD, major depression, SUD), use of conventional antipsychotics only at


baseline, use of VA each year, and region. We used a logistic random-effects model for analyzing incomplete binary repeated-measures data that allowed us to account for death during the 4-year observation period, operationalized in SAS GLIMMIX (SAS Institute, Cary, NC). To determine long-term sustainment of metabolic monitoring, we assessed association of race and ethnicity with available FY09 parameters (receipt of serum glucose, HDL cholesterol, and triglyceride testing) among patients surviving through September 2009 and modeled receipt using logistic regression adjusting for the factors noted above. Criterion alpha was set at 0.01.

RESULTS Demographic Characteristics Characteristics of the sample are presented in Table 1. The sample comprised mostly men (97%).

Am J Geriatr Psychiatry

-:-, -


Phillips et al.

FIGURE 1. Metabolic monitoring by race in patients with schizophrenia, FY02eFY05 (N [ 30,258).

Most patients were white (70%), whereas 30% of the patients were black. Eight percent (8%) of the sample were identified as Hispanic. Patient mean age was 59  9 years. A high proportion of the sample (35%) resided in the southern region of the United States. Most black patients (46%) lived in the south. Most of the sample (52%) had a VA Priority 1 status, whereas one-fifth (19%) qualified for VA care per very low income. The overall rates of glucose and lipid monitoring were less than 60% in all years (range: 9%e58%). Blood pressure and weight were consistently monitored in more than 80% of patients with schizophrenia (range: 84%e87%). All five parameters were monitored in 6% of the sample in the baseline year and in 29% by the final year. Initial comparisons revealed some differences between whites and blacks in the sample (Table 1). White patients tended to be older (mean age 59.6  9.1 versus 57.5  8.3 years; Mann-Whitney z ¼ 20.7, p

Ethnic disparities in monitoring metabolic parameters for patients with schizophrenia receiving antipsychotic medications.

Patients with schizophrenia experience risks for metabolic dysregulation from medications and lifestyle behaviors. Although most patients with schizop...
542KB Sizes 3 Downloads 13 Views