Independence of diabetes and obesity in adults with serious mental illness: Findings from a large urban public hospital

Abstract

Objective:

There is limited research on metabolic abnormalities in psychotropic-naïve patients with serious mental illness (SMI). Our study examined metabolic conditions in a large, ethnically diverse sample of psychotropic-naïve and non-naïve adults with SMI at an urban public hospital.

Methods:

In this cross-sectional study of 923 subjects, the prevalences of hyperglycemia meeting criteria for type 2 diabetes mellitus (T2DM) based on fasting plasma glucose and obesity defined by BMI and abdominal girth were compared across duration of psychotropic medication exposure. Multiple logistic regression models used hyperglycemia and obesity as dependent variables and age, sex, race/ethnicity, and years on psychotropics as independent variables.

Results:

Psychotropic-naïve patients, including both schizophrenia and non-psychotic subgroups, showed an elevated prevalence of hyperglycemia meeting criteria for T2DM and a decreased prevalence of obesity compared to the general population. Obesity rates significantly increased for those on psychotropic medications more than 5 years, particularly for patients without psychosis (BMI: aOR = 5.23 CI = 1.44–19.07; abdominal girth: aOR = 6.40 CI = 1.98–20.69). Women had a significantly higher obesity rate than men (BMI: aOR = 1.63 CI = 1.17–2.28; abdominal girth: aOR = 3.86 CI = 2.75–5.44). Asians had twice the prevalence of hyperglycemia as whites (aOR = 2.29 CI = 1.43–3.67), despite having significantly less obesity (BMI: aOR = .39 CI = .20–.76; abdominal girth: aOR = .34 CI = .20–.60). Hispanics had a higher rate of obesity by BMI than whites (aOR = 1.91 CI = 1.22–2.99).

Conclusions:

This study showed disparities between obesity and T2DM in psychotropic-naïve patients with SMI, suggesting separate risk pathways for these two metabolic conditions.

1. Introduction

Metabolic syndrome is a cluster of conditions that tend to occur together and increase a patient’s risk of cardiovascular disease, T2DM, stroke, and all-cause mortality (Kaur, 2014). Metabolic syndrome is associated with a three times higher risk of death from coronary heart disease and twice the risk of all-cause mortality (Lakka et al., 2002). T2DM and obesity are chronic medical diseases associated with metabolic syndrome that have increased dramatically in the U.S. population (Barnes, 2011). These trends are particularly alarming for patients with SMI who demonstrate 1.5–2 fold the prevalence of T2DM, dyslipidemia, hypertension, and obesity compared to the general population (Newcomer and Hennekens, 2007), along with less favorable outcomes, which contributes to an estimated 10–20 year decreased life expectancy (Chesney et al., 2014). Factors posited to explain the increased rate of metabolic disorders among those with SMI include psychotropic medications, genetics, unhealthy lifestyles, low socioeconomic status, cigarette smoking, and healthcare inequalities (Lawrence and Kisely, 2010, Padmavati, 2016).

Antipsychotic use is reported to increase the risk of T2DM, although of an uncertain magnitude, and almost all antipsychotics are associated with weight gain (Correll et al., 2015). Mood stabilizers, including lithium and valproic acid, and many antidepressants, such as amitriptyline and mirtazapine, produce lesser weight gain than that associated with antipsychotic use (Correll et al., 2015, Mcknight et al., 2012). Although mood stabilizers also increase the risks for insulin resistance and T2DM (Belcastro et al., 2013, Chien et al., 2012), findings are inconclusive regarding a possible association between antidepressants and T2DM (Correll et al., 2015).

While the relationship between psychotropic medications and metabolic syndrome has long been known, there is emerging evidence of an inherent predisposition to certain metabolic conditions among some patients with SMI. Research on psychotropic-naïve or first-episode patients has primarily focused on patients with schizophrenia. Multiple studies have demonstrated an increased prevalence of impaired fasting glucose tolerance in psychotropic-naïve patients with schizophrenia (Ryan et al., 2003, Spelman et al., 2007), although some studies have not found such elevations in fasting plasma glucose (Padmavati et al., 2010). A recent meta-analysis, however, demonstrated elevations in several indicators of dysglycemia including fasting plasma glucose levels in first-episode patients with schizophrenia compared to matched controls (Pillinger et al., 2017a). While one small study in 2002 with 15 patients found that psychotropic-naïve patients with schizophrenia had increased central obesity compared to matched controls (Thakore et al., 2002); more recent larger studies do not show an increased prevalence of obesity in psychotropic-naïve patients with schizophrenia (Padmavati et al., 2010, Verma et al., 2009).

Effects of sex and race/ethnicity are rarely considered in studies of metabolic parameters in patients with SMI. In the general population, males have a slightly higher rate of diabetes than females, 8.7% vs. 7.7% (CDC, 2009); however, there is limited research on sex differences in T2DM risk among patients with SMI. One community study of 1123 patients with schizophrenia in Canada showed no significant differences (Voruganti et al., 2007). Regarding sex differences in obesity, females with SMI have a significantly higher prevalence of obesity than their male counterparts (Carliner et al., 2014, Jonikas et al., 2016).

A review examining racial/ethnic differences in diabetes found that persons with African ancestry and Hispanics appear to be at a higher risk of diabetes than whites with SMI (Carliner et al., 2014), but studies on the association between race/ethnicity and obesity in patients with SMI are limited and inconclusive. Some studies showed increased obesity in those with African ancestry and Hispanics compared to whites with SMI, while others demonstrated no significant differences in obesity by race/ethnicity (Carliner et al., 2014).

Our metabolic screening study at Bellevue Hospital Center, a large public hospital in New York City, provided a unique opportunity to analyze the cross-sectional prevalence of two metabolic conditions, hyperglycemia and obesity, in patients with SMI with respect to duration of psychotropic medication treatment and demographic variables. The primary aim of our study was to look at the prevalence of metabolic conditions in psychotropic-naïve patients with schizophrenia and other SMI as there is a scarcity of studies with large, diverse samples of psychotropic-naïve patients. We also examined associations between the two metabolic conditions and the hypothesized predictors: years on psychotropics, sex, and race/ethnicity. Additionally, this study explored the likelihood of hyperglycemia and obesity by years on psychotropics in psychotic and non-psychotic patients separately in order to draw comparisons between these two groups.

2. Methods

2.1. Data sources

In accordance with FDA recommendations, a protocol was implemented to screen all adult psychiatric inpatients for metabolic syndrome from 2006 to 2009. The human subjects committee at Bellevue Hospital Center approved the protocol. On admission, patients underwent a set of metabolic laboratory tests and physical measurements, and an attending psychiatrist on the inpatient unit obtained additional information through patient interview, medical records, and collateral sources. The data included information on demographic and clinical measures, such as weight, height, body mass index (BMI), abdominal girth, blood pressure, fasting plasma glucose, triglycerides, high-density lipoprotein (HDL) cholesterol, patient history of diabetes, diagnosis, presence or absence of psychosis, psychotropic medications prescribed within six months of admission, and years on psychotropics. Data was entered into a “Metabolic Findings Form” (MFF), developed to capture information about the predictors of metabolic syndrome for future analysis.

2.2. Conditions of metabolic syndrome

Hyperglycemia and obesity were defined in accordance to the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III), American Diabetic Association (ADA), and Center for Disease Control (CDC) guidelines (Huang, 2009, ADA, 2016, CDC, 2016). Patients were categorized with regards to hyperglycemia as normal (fasting plasma glucose < 100 mg/dL), hyperglycemia meeting criteria for pre-diabetes (fasting plasma glucose 100–125 mg/dL), and hyperglycemia meeting criteria for T2DM (fasting plasma glucose ≥126 mg/dL) (ADA, 2016). Those with a history of diabetes were categorized as having hyperglycemia meeting criteria for T2DM. Additionally, patients were categorized by NCEP ATP III criteria for a binary logistic regression analysis. Using this criteria to define hyperglycemia dichotomously, a single hyperglycemia group was defined for fasting plasma glucose ≥100 mg/dL that also included all patients with a history of diabetes (Huang, 2009).

Based on NCEP ATP III criteria, males with abdominal girth > 40 inches and females with abdominal girth > 35 inches at the level of the navel were categorized as obese by abdominal girth (Huang, 2009). Since the NCEP ATP III criteria does not define obesity by BMI, the CDC criteria that categorizes obesity as BMI ≥30 was employed (CDC, 2016).

2.3. Psychiatric diagnosis

Primary diagnoses were made pursuant to DSM-IV criteria by an inpatient academic psychiatrist. They included schizophrenia, schizoaffective disorder, psychosis NOS, bipolar disorder, depressive disorder, mood disorder NOS, psychiatric diagnosis secondary to a primary medical condition, and “other psychiatric diagnosis”. Patients were also categorized as either psychotic or non-psychotic based on symptoms at admission, regardless of primary diagnosis.

2.4. Psychotropic medications

Psychotropic medications prescribed during the past six months prior to admission were categorized as follows: first-generation antipsychotics, second-generation antipsychotics, selective serotonin reuptake inhibitors (SSRIs), mood stabilizers, anticonvulsants, and “other psychiatric medications”. The length of time on any psychotropic medication was recorded as never, < 1, 1 to 5, or > 5 years.

2.5. Demographic categories

Patients were categorized into one of four age groups: ≤24, 25 to 39, 40 to 54, and ≥55 years. Racial/ethnic groups included white, African ancestry, Hispanic, Asian, and “other race/ethnicity”. The Asian group consisted of East Asians and Pacific Islanders. South Asians were categorized in the “other race/ethnicity” group.

2.6. Data analysis

Bivariate comparisons assessed differences in the prevalence of hyperglycemia, obesity by BMI, and obesity by abdominal girth, by sex, race/ethnicity, and years on psychotropics. These categorical variables were evaluated using chi-square tests with an alpha (α) < .05 to define significance. Multiple logistic regression models were performed for each of the following dependent variables: hyperglycemia, obesity by BMI, and obesity by abdominal girth. All models adjusted for age group, sex, race/ethnicity, and years on psychotropics. Regression analysis examining psychotic and non-psychotic patients separately was performed to explore differences between these two groups.

3. Results

The analytic sample included 923 adults with SMI (age range: 17–90, mean: 43.9, SD: ± 14.9 years) (Table 1). There were 150 psychotropic-naïve adults (age range: 17–90, mean: 45.1, SD: ± 18.9 years), 95 psychotropic-naïve psychotic adults (age range: 19–83, mean: 43, SD: ± 18.3 years), 32 psychotropic-naïve adults with schizophrenia (age range: 19–75, mean: 45.7, SD: ± 17.9 years), and 55 psychotropic-naïve non-psychotic adults (age range: 19–90, mean: 48.8, SD: ± 19.4 years). The sample was predominantly male (63.9%), with racial/ethnic groups of African ancestry (32.3%), white (25.2%), Hispanic (21.8%), Asian (16.6%), and other race/ethnicity (4.1%). The most prevalent psychiatric diagnoses were schizophrenia (34.2%), schizoaffective disorder (18.2%), bipolar disorder (12.4%), psychosis NOS (10.5%), and depressive disorder (8.7%). Most patients had psychosis (70%). A greater proportion of patients were on second-generation antipsychotics (41.4%) than first-generation antipsychotics (8.8%). The most frequently prescribed antipsychotics were risperidone (19.2%), olanzapine (11.7%), and quetiapine (9.8%). The most frequently prescribed non-antipsychotic medications were mood stabilizers (19.3%) and SSRIs (8.1%). Patients were categorized into one of four categories based on their years of exposure to psychotropic medication: never (16.3%), < 1 year (12.7%), 1 to 5 years (30.9%), and > 5 years (40.0%). Overall, 17.4% of patients had hyperglycemia meeting criteria for T2DM, and 17.6% had hyperglycemia meeting criteria for pre-diabetes; 22.3% of patients had obesity by BMI and 30% had obesity by abdominal girth.

Table 1.

Frequency of demographic and clinical characteristics in adults age 17 to 90 with serious mental illness admitted to Bellevue Hospital from 2006 to 2009.

	Number	Percent	P-Value
Total Sample	923	100
Age Group
≤24 years	101	10.9	< .001
25–39 years	269	29.1
40–54 years	319	34.6
≥55 years	234	25.4
Sex
Male	590	63.9	< .001
Female	333	36.1
Race/Ethnicity
White	233	25.2	< .001
African Ancestry	298	32.3
Hispanic	201	21.8
Asian	153	16.6
Other Race/Ethnicity	38	4.1
Primary Diagnosis
Schizophrenia	316	34.2	< .001
Schizoaffective	168	18.2
Psychosis NOS	97	10.5
Bipolar Disorder	114	12.4
Depressive Disorder	80	8.7
Mood Disorder NOS	42	4.6
Psychiatric Diagnosis Secondary to a Primary Medical Condition	58	6.3
Other Psychiatric Diagnosis	48	5.2
Psychosis
Non-Psychotic	277	30	< .001
Psychotic	646	70
Medications (Prescribed Within Six Months of Admission)
Risperidone	177	19.2
Olanzapine	108	11.7
Quetiapine	90	9.8
Aripiprazole	50	5.4
Clozapine	29	3.1
Ziprasidone	13	1.4
First-Generation Antipsychotic	81	8.8
Second-Generation Antipsychotic	382	41.4
Lithium	47	5.1
Valproic Acid	140	15.2
Mood Stabilizer	178	19.3
Anti-Convulsant	19	2.1
SSRI	75	8.1
Other Psychiatric Medication	118	12.8
Years on Psychotropic Medication
Never Used	150	16.3	< .001
Less than 1 year	117	12.7
1 to 5 years	284	30.9
Greater than 5 years	368	40.0
Hyperglycemia (Fasting Blood Sugar)
< 100mg/dL (Not Diabetic)	600	65.0	< .001
100–125 mg/dL (Pre-Diabetic)	162	17.6
≥126mg/dL or History of Diabetes (Diabetic)	161	17.4
Obesity by Body Mass Index (Weight/Height 2 )
≤30 (Not Obese)	717	77.7	< .001
> 30 (Obese)	206	22.3
Obesity by Abdominal Girth
Males ≤ 40 inches (Not Obese)	573	70.0	< .001
Females ≤ 35 inches (Not Obese)
Males > 40 inches (Obese)	246	30.0
Females > 35 inches (Obese)

The prevalence of hyperglycemia meeting criteria for T2DM was 20% in psychotropic-naïve patients (22.1% in the psychotic subgroup, 18.8% in the schizophrenia subgroup, and 16.4% in the non-psychotic subgroup). The prevalence of obesity by BMI was 14.7% in psycho-tropic-naïve patients (20% in the psychotic subgroup, 21.9% in the schizophrenia subgroup, and 5.5% in the non-psychotic subgroup). The prevalence of obesity by abdominal girth was 19.4% in psychotropic-naïve patients (24.4% in the psychotic subgroup, 24.1% in the schizophrenia subgroup, and 9.1% in non-psychotic subgroup).

A bivariate analysis did not reveal significant sex differences in the prevalence of hyperglycemia, although females were significantly more likely to have obesity by both BMI and abdominal girth (Table 2). There was likewise no significant effect on hyperglycemia by race/ethnicity, although Hispanics were most likely and Asians were least likely to have obesity by BMI, while whites were most likely and Asians were least likely to have obesity by abdominal girth (Table 3).

Table 2.

Distribution of metabolic conditions by sex in adults age 17 to 90 with serious mental illness admitted to Bellevue Hospital from 2006 to 2009.

Metabolic Condition	Male		Female		P-Value
	Number	Percent	Number	Percent
Total
Hyperglycemia
Not Diabetic	391	66.3	209	62.8	.56
Pre-Diabetic	100	16.9	62	18.6
Diabetica	99	16.8	62	18.6
Obesity by Body Mass Index
Not Obese	471	79.8	246	73.9	< .05
Obese	119	20.2	87	26.1
Obesity by Abdominal Girth
Not Obese	418	79.0	155	53.4	< .001
Obese	111	21.0	135	46.6

^aDetermined based on fasting blood sugar and history of diabetes.

Table 3.

Distribution of metabolic conditions by race/ethnicity in adults age 17 to 90 with serious mental illness admitted to Bellevue Hospital from 2006 to 2009.

Metabolic Condition	White		African Ancestry		Hispanic		Asian		Other Race/Ethnicity		P-Value
	Number	Percent	Number	Percent	Number	Percent	Number	Percent	Number	Percent
Hyperglycemia
Not Diabetic	154	66.1	198	66.4	127	63.2	93	60.8	28	73.7	.78
Pre-Diabetic	38	16.3	55	18.5	37	18.4	28	18.3	4	10.5
Diabetic	41	17.6	45	15.1	37	18.4	32	20.9	6	15.8
Obesity by Body Mass Index
Not Obese	183	78.5	225	75.5	137	68.2	139	90.8	33	86.8	< .001
Obese	50	21.5	73	24.5	64	31.8	14	9.2	5	13.2
Obesity by Abdominal Girth
Not Obese	128	61.2	183	69.8	112	67.5	125	82.2	25	83.3	< .001
Obese	81	38.8	79	30.2	54	32.5	27	17.8	5	16.7

There was no significant difference in the prevalence of hyperglycemia by years on psychotropics, even though greater years on psychotropics significantly predicted increased obesity by BMI and abdominal girth (Table 4). Psychotic patients who were psychotropic-naïve had similar rates of hyperglycemia meeting criteria for T2DM as those on psychotropics > 5 years, while obesity defined by abdominal girth significantly increased with longer years on psychotropics (although not obesity defined by BMI) in psychotic patients. Psychotic patients on psychotropics > 5 years had the highest prevalence and those on psychotropics < 1 year had the lowest prevalence of obesity by abdominal girth. Nonpsychotic cases had similar rates of hyperglycemia meeting criteria for T2DM over durations of psychotropic use. Compared to psychotic patients, non-psychotic patients had an even higher increase in obesity prevalence with longer psychotropic medication exposure, whether defined by BMI or abdominal girth.

Table 4.

Distribution of metabolic conditions by years on psychotropic medication in the total sample, psychotic adults, and non-psychotic adults age 17 to 90 with serious mental illness admitted to Bellevue Hospital from 2006 to 2009.

Metabolic Condition	Never Used Psychotropic Medication		Less than 1 Year		1 to 5 Years		Greater than 5 Years		P-Value
	Number	Percent	Number	Percent	Number	Percent	Number	Percent
Total Sample
Hyperglycemia
Not Diabetic	89	59.3	74	63.2	199	70.1	234	63.6	.18
Pre-Diabetic	31	20.7	26	22.2	44	15.5	61	16.6
Diabetic	30	20.0	17	14.5	41	14.4	73	19.8
Obesity by Body Mass Index
Not Obese	128	85.3	99	84.6	209	73.6	278	75.5	< .01
Obese	22	14.7	18	15.4	75	26.4	90	24.5
Obesity by Abdominal Girth
Not Obese	108	80.6	80	80.8	174	68.8	209	63.5	< .001
Obese	26	19.4	19	19.2	79	31.2	120	36.5
Psychotic Patients
Hyperglycemia
Not Diabetic	54	56.8	42	63.6	149	74.1	181	64.2	.01
Pre-Diabetic	20	21.1	17	25.8	28	13.9	45	16.0
Diabetic	21	22.1	7	10.6	24	11.9	56	19.9
Obesity by Body Mass Index
Not Obese	76	80.0	56	84.8	149	74.1	212	75.2	.25
Obese	19	20.0	10	15.2	52	25.9	70	24.8
Obesity by Abdominal Girth
Not Obese	68	75.6	48	81.4	125	67.9	166	64.3	< .05
Obese	22	24.4	11	18.6	59	32.1	92	35.7
Non-Psychotic Patients
Hyperglycemia
Not Diabetic	35	63.6	32	62.7	50	60.2	53	61.6	1.00
Pre-Diabetic	11	20.0	9	17.6	16	19.3	16	18.6
Diabetic	9	16.4	10	19.6	17	20.5	17	19.8
Obesity by Body Mass Index
Not Obese	52	94.5	43	84.3	60	72.3	66	76.7	< .01
Obese	3	5.5	8	15.7	23	27.7	20	23.3
Obesity by Abdominal Girth
Not Obese	40	90.9	32	80.0	49	71.0	43	60.6	< .01
Obese	4	9.1	8	20.0	20	29.0	28	39.4

In the multiple logistic regression analysis, females were significantly more likely to have obesity by BMI (aOR = 1.63 CI = 1.17–2.28) and almost fourfold more likely to have obesity by abdominal girth compared to males (aOR = 3.86 CI = 2.75–5.44) (Table 5). In comparison to whites, Asians were twice as likely to have hyperglycemia (aOR = 2.29 CI = 1.43–3.67) despite being significantly less likely to have obesity by BMI (aOR = .39 CI = .20–.76) or abdominal girth (aOR = .34 CI = .20–.60). Hispanics were nearly twofold more likely to have obesity by BMI compared to whites (aOR = 1.91 CI = 1.22–2.99).

Table 5.

Multiple logistic regressions with metabolic conditions as dependent variables and age group, sex, race/ethnicity, and years on psychotropic medication as independent variables in the total sample, psychotic adults, and non-psychotic adults age 17 to 90 with serious mental illness admitted to Bellevue Hospital from 2006 to 2009.

Metabolic Condition	All Patients			Psychotic Patients			Non-Psychotic patients
	Hyperglycemia	Obesity by Body Mass Index	Obesity by Abdominal Girth	Hyperglycemia	Obesity by Body Mass Index	Obesity by Abdominal Girth	Hyperglycemia	Obesity by Body Mass Index	Obesity by Abdominal Girth
Age Group
≤24 years	–	–	–	–	–	–	–	–	–
25–39 years	2.04* (1.08–3.87)	2.37* (1.13–4.96)	1.16 (.59–2.27)	1.93 (.90–4.15)	1.86 (.81–4.23)	1.30 (.58–2.88)	2.38 (.71–7.98)	8.33* (1.01–68.47)	.93 (.24–3.66)
40–54 years	4.63* (2.47–8.69)	2.92* (1.42–6.03)	1.68 (.87–3.23)	4.55* (2.13–9.75)	2.39* (1.07–5.34)	1.88 (.86–4.10)	5.81* (1.79–18.91)	9.89* (1.22–80.10)	1.48 (.41–5.38)
≥55 years	8.89* (4.66–16.97)	2.71* (1.28–5.71)	2.37* (1.22–4.61)	8.14* (3.70–17.87)	2.26* (.98–5.19)	2.91* (1.32–6.43)	13.55* (4.08–45.03)	9.32* (1.13–76.93)	1.75 (.48–6.42)
Sex
Male	–	–	–	–	–	–	–	–	–
Female	1.02 (.76–1.38)	1.63* (1.17–2.28)	3.86* (2.75–5.44)	1.19 (.83–1.71)	1.78* (1.19–2.65)	4.31* (2.87–6.47)	.62 (.35–1.10)	1.14 (.59–2.23)	2.84* (1.43–5.64)
Race/Ethnicity
White	–	–	–	–	–	–	–	–	–
African Ancestry	1.32 (.89–1.94)	1.28 (.84–1.95)	.74 (.49–1.13)	1.24 (.78–1.99)	1.13 (.69–1.85)	.76 (.46–1.24)	1.88 (.91–3.91)	1.97 (.82–4.74)	.62 (.25–1.52)
Hispanic	1.48 (.97–2.26)	1.91* (1.22–2.99)	.86 (.54–1.36)	1.25 (.72–2.15)	1.90* (1.10–3.28)	1.00 (.57–1.78)	2.11* (1.06–4.21)	2.33* (1.03–5.26)	.70 (.30–1.61)
Asian	2.29* (1.43–3.67)	.39* (.20–.76)	.34* (.20–.60)	2.24* (1.28–3.91)	.31* (.15–.65)	.28* (.14–.53)	2.89* (1.04–8.03)	.83 (.20–3.38)	.72 (.23–2.25)
Other Race/Ethnicity	1.06 (.47–2.38)	.59 (.21–1.64)	.28* (.10–.83)	1.13 (.42–3.09)	.41* (.11–1.49)	.25* (.07–.86)	.76 (.18–3.26)	1.12 (.21–6.06)	.30 (.03–2.63)
Years on Psychotropic Medication
Never Used	–	–	–	–	–	–	–	–	–
Less than 1 year	.91 (.54–1.56)	1.04 (.52–1.08)	1.21 (.60–2.46)	.79 (.40–1.59)	.74 (.31–1.80)	1.03 (.42–2.52)	1.41 (.58–3.40)	3.43 (.83–14.22)	2.44 (.64–9.34)
1 to 5 years	.66 (.42–1.02)	2.17* (1.26–3.73)	2.61* (1.51–4.50)	.45* (.26–.78)	1.45 (.78–2.71)	2.11* (1.12–3.97)	1.64 (.75–3.60)	7.16* (1.96–26.14)	4.90* (1.46–16.43)
Greater than 5 years	.72 (.47–1.09)	1.64 (.96–2.78)	2.64* (1.56–4.46)	.60 (.36–1.01)	1.10 (.61–2.04)	1.93* (1.05–3.55)	1.19 (.55–2.56)	5.23* (1.44–19.07)	6.40* (1.98–20.69)

Adjusted Odds Ratio (95% Confidence Interval).

^*P ≤ .05.

Psychotropic-naïve patients served as the reference group for years on psychotropic medication comparisons. An analysis including the total sample of patients showed no significant difference in the prevalence of hyperglycemia for greater years of psychotropic treatment exposure. Patients on psychotropics between one and five years were more than twofold more likely to have obesity by BMI (aOR = 2.17 CI = 1.26–3.73) or abdominal girth (aOR = 2.61 CI = 1.51–4.50) (Table 5). Patients on psychotropics > 5 years also demonstrated increased obesity by abdominal girth (aOR = 2.64 CI = 1.56–4.46), with a similar, but insignificant, increase in obesity by BMI (aOR = 1.64 CI = .96–2.78).

Among only psychotic patients, those on psychotropics between one and five years were significantly less likely to have hyperglycemia than psychotropic-naïve patients (aOR = .45 CI = .26–.78), while those on psychotropics > 5 years showed no difference in likelihood of hyperglycemia (aOR = .60 CI = .36–1.01) (Table 5). While there was no significant difference in obesity by BMI for psychotic patients with longer duration on medication, those on psychotropics between one and five years (aOR = 2.11 CI = 1.12–3.97) and > 5 years (aOR = 1.93 CI = 1.05–3.55) had approximately twice the rate of obesity by abdominal girth.

Finally, an examination of only non-psychotic patients showed no significant difference in hyperglycemia prevalence over years of psychotropic exposure (Table 5). Conversely, those treated between one and five years were sevenfold more likely to have obesity by BMI (aOR = 7.16 CI = 1.96–26.14) and almost fivefold more likely to have obesity by abdominal girth (aOR = 4.90 CI = 1.46–16.43), and those treated > 5 years were fivefold more likely to have obesity by BMI (aOR = 5.23 CI = 1.44–19.07) and sixfold more likely to have obesity by abdominal girth (aOR = 6.40 CI = 1.98–20.69).

4. Discussion

Our study found an elevated prevalence of hyperglycemia meeting criteria for T2DM and a decreased prevalence of obesity in psycho-tropic-naïve patients with SMI compared to the general population at the time of the study. This finding held true across all subgroups of psychotropic-naïve patients. The rate of hyperglycemia meeting criteria for T2DM was 20% in psychotropic-naïve patients (22.1% in the psychotic subgroup, 18.8% in the schizophrenia subgroup, and 16.4% in the non-psychotic subgroup). The national rate of diabetes in the general population was 8% in 2009, but it was 11% among adults age 45 to 64, which was used to make conservative comparisons (CDC, 2009). Although, the mean and median age of our sample of psychotropic-naïve patients was at the youngest end of this age range, the rate of hyperglycemia meeting criteria for T2DM was several percentage points higher in all subgroups of psychotropic-naïve patients compared to the rate of diabetes in the general population.

A case-control study in Nigeria involving 250 psychotropic-naive patients also demonstrated an elevated rate of diabetes in psychotropic-naive patients with mental illness. The Nigerian study found that 12.8% of cases with mental illness had diabetes compared to 5.6% of controls without mental illness (Olatunbosun et al., 2015). Other individual studies on dysglycemia in early mental illness involved small sample sizes and focused only on patients with schizophrenia. Supporting our findings, a rigorous meta-analysis of 14 case-control studies comprising 1345 participants demonstrated elevated fasting plasma glucose levels, elevated plasma glucose levels after an oral glucose tolerance test, elevated fasting plasma insulin levels, and greater insulin resistance in first-episode patients with schizophrenia compared to healthy controls (Pillinger et al., 2017a).

The elevated prevalence of hyperglycemia meeting criteria for T2DM in psychotropic-naïve patients suggests that factors other than psychotropic medication and obesity contribute to glucose dysregulation in patients with SMI. These factors may involve lifestyle, inflammatory, and genetic mechanisms. People with SMI are more likely to be sedentary, smoke, and have unhealthy diets, which are known risk factors for T2DM (Holt and Mitchell, 2015). However an unhealthy lifestyle would likely raise the risk of T2DM and obesity concurrently, so the finding of a decreased prevalence of obesity in psychotropic-naïve patients may indicate that inflammatory and genetic factors play a more prominent role in the elevated rate of hyperglycemia in these patients. Schizophrenia is associated with inflammatory changes and chronic inflammation has been suggested as a precursor to both T2DM and SMI (Holt and Mitchell, 2015). There is a growing body of literature on the role of inflammatory pathways in both T2DM and schizophrenia, and these two diseases share many cytokine pathways and inflammatory genetic components (Perry et al., 2016). A high number of people with SMI (17–50%) have first-degree relatives with T2DM, which further supports the hypothesis of a shared genetic link between T2DM and SMI (Gough and O’Donovan, 2005).

The rate of obesity by BMI was 14.7% in psychotropic-naïve patients (20% in the psychotic subgroup, 21.9% in the schizophrenia subgroup, and 5.5% in the non-psychotic subgroup). The rate of obesity by abdominal girth was 19.4% in psychotropic-naïve patients (24.4% in the psychotic subgroup, 24.1% in the schizophrenia subgroup, and 9.1% in non-psychotic subgroup). These rates were lower than the national rate of obesity in the general population in 2009–2010, 35.7% (CDC, 2009–2010).

These findings are supported by a 2010 study involving 51 patients which showed that antipsychotic-naïve patients with chronic schizophrenia had a significantly lower mean BMI compared to controls (Padmavati et al., 2010). One Asian study examined metabolic indices in 160 drug-naïve patients with schizophrenia in Singapore. The study found a significantly lower mean BMI and weight in drug-naïve cases, along with a significantly higher rate of diabetes in drug-naïve cases compared to controls (Verma et al., 2009). Multiple cohort studies have found that young adult men who would later develop schizophrenia had lower mean BMI and weight compared to peers, demonstrating an inverse relationship between early adulthood BMI and risk of developing schizophrenia (Zammit et al., 2007, Sørensen et al., 2006). Mood disorders such as depression have generally been associated with obesity, but some studies have also found a negative association between BMI and depression (De Wit et al., 2009), and research on psychotropic-naïve patients with mood disorders is sorely lacking.

Genetic, environmental, and disease symptoms are possible explanations for the decreased obesity in psychotropic-naïve patients with SMI. Researchers have suggested that genetic factors may contribute to obesity resistance, and a genetic predisposition to schizophrenia may influence growth patterns (Sørensen et al., 2006). Malnutrition during critical early development may limit adult growth potential and increase the risk of schizophrenia (Zammit et al., 2007). Finally, prodromal symptoms of schizophrenia may include depression with decreased appetite (Sørensen et al., 2006).

Although metabolic syndrome is a cluster of associated conditions, this study and others demonstrate that the components of metabolic syndrome are dissociable in SMI (Pillinger et al., 2017b). A recent meta-analysis found decreased total and LDL cholesterol levels in individuals with first episode psychosis compared with healthy controls even after matching for BMI, while hypertriglyceridemia, which is a feature of type 2 diabetes, was increased in patients with first episode psychosis compared to controls (Pillinger et al., 2017b). These findings corroborate the idea of separate, unique metabolic pathways in SMI.

Our study found no significant increase in the rate of hyperglycemia with greater years of psychotropic exposure in psychotic and non-psychotic patients despite increasing obesity. There is strong literature suggesting that certain second-generation antipsychotics, particularly clozapine and olanzapine, and to a lesser extent quetiapine and risperidone, are associated with an increased risk of glucose dysregulation and T2DM (Correll et al., 2015). However, aripiprazole has not been shown to increase T2DM risk (ADA, 2004). In our study, the most frequently prescribed antipsychotic was risperidone (19.2%), followed by olanzapine (11.7%), quetiapine (9.8%) and aripiprazole (5.4%), while clozapine (3%) was prescribed sparingly. Therefore, our findings may be a reflection of the particular antipsychotics prescribed to our sample, and they should not be interpreted to mean that antipsychotics have no effect on the development of T2DM. Furthermore, patients with mood disorders may not have a high burden of antipsychotic usage despite many years on other psychotropic medications.

Psychotropic medication use was associated with a higher rate of obesity. Notably, the increase in obesity prevalence with longer psychotropic exposure was substantially greater for non-psychotic cases compared to psychotic cases. A combination of medication and lifestyle factors may explain this finding. The majority of the non-psychotic patients had diagnoses of bipolar disorder or depressive disorder, and they were taking a variety of psychotropic medications, including antipsychotics, mood stabilizers, and SSRIs. Mood stabilizers and certain antidepressants have been associated with weight gain, but weight gain is generally more modest compared to antipsychotic use (Correll et al., 2015, Mcknight et al., 2012). Lifestyle factors may be a significant contributor to the greater rise in obesity prevalence in non-psychotic patients compared to psychotic patients in our study. Mood disorders can entail overeating, reduced physical activity, and elevated cortisol levels that can lead to obesity over the course of the disease (Goldstein et al., 2011, Mann and Thakore, 1999). Regardless of the etiology, our findings emphasize the importance of monitoring non-psychotic patients, in addition to psychotic ones, for weight gain.

While there was no significant difference in hyperglycemia prevalence by sex, females were significantly more likely to have obesity by BMI and fourfold more likely to have obesity by abdominal girth. This finding corroborates prior research showing sex differences in obesity among outpatients with mental illness (Jonikas et al., 2016). Biological, lifestyle, and environmental factors may explain this difference in obesity. Women have increased sensitivity to insulin, along with greater adiposity and less lean body mass (Azarbad and Gonder-Frederick, 2010). Furthermore, pregnancy and menopause in women contributes to weight gain (Azarbad and Gonder-Frederick, 2010). Additionally, physical activity is more common in men than women according to a surgeon general report, and more than 25% of American women do not engage in any type of physical activity (CDC, 1996). Finally, childhood trauma has been found to be associated with obesity (Alvarez et al., 2007), and childhood trauma is more prevalent among women than men with mental illness (Jonikas et al., 2016).

Disparities by race/ethnicity were also found in our sample. Asians were twofold more likely to have hyperglycemia compared to whites. Prior studies comparing diabetes in patients with SMI by race/ethnicity have shown that those with African ancestry and Hispanics have the highest prevalence of diabetes (Carliner et al., 2014). However, there is evidence that Asians have a higher prevalence of diabetes compared to whites in patients with SMI (Mangurian et al., 2017). The particularly high rate of hyperglycemia observed among Asians in our sample may also be related to the fact that the Asian psychiatric patients at Bellevue Hospital Center were predominantly immigrants of low socioeconomic background. Clinicians must take care in monitoring Asian patients for T2DM because at the same BMI Asians have a higher prevalence of T2DM than Caucasians (Huxley et al., 2008).

Compared to whites, Asians were significantly less likely to have obesity by both BMI and abdominal girth, and Hispanics were almost twofold more likely to have obesity by BMI. The lack of a significant difference between Hispanics and whites in abdominal girth may be due to the relatively smaller stature of Hispanics who had a two inch shorter median height in our sample. Prior literature has inconsistently shown a higher prevalence of obesity in persons with African ancestry and Hispanics with SMI (Carliner et al., 2014). A variety of factors including physical inactivity, cultural attitudes and norms, and access to affordable healthy foods may contribute to the elevated rate of obesity in Hispanics (Adler and Stewart, 2009, Millstein et al., 2008).

Prior studies on psychotropic-naïve patients faced several limitations including small sample sizes, focusing only on patients with schizophrenia, lacking a diverse patient sample, or examining only a single metabolic condition. However, our study addressed these limitations, while carrying additional strengths. This study had a large sample of patients with SMI based on acute admission to a public psychiatric hospital, and there was a large subgroup of psychotropic-naïve patients. The sample was ethnically diverse, permitting comparison across groups. Academic faculty made diagnoses and obtained comprehensive data through targeted interviews, collateral sources, and rigorous records review. Both hyperglycemia and obesity were examined, and they were defined in accordance to standard guidelines through laboratory measures, physical measures, and patient history.

Our study had several limitations. First, we did not have matched controls for comparison of prevalence data. Our sample was ethnically diverse and predominately low socioeconomic status; therefore, direct comparison with the general population was imperfect. Second, as this was a cross-sectional study causality could not be definitively determined. Third, reporting and recall biases may influence data obtained from patient interviews. Fourth, we did not obtain socioeconomic information; however, the cases admitted to this large public psychiatric service commonly lack insurance or receive public assistance, so social class differences are unlikely to contribute to much variability in the analysis. Fifth, our study was not designed to distinguish between type 1 and type 2 diabetes. Finally, we only had information on psychotropic medications prescribed within six months of admission, so we did not know how many years patients were on specific psychotropic medications.

5. Conclusion

The findings from our study suggest separate pathways for the development of T2DM and obesity in patients with SMI and demonstrate important sex and race/ethnicity effects. Certain groups, including non-psychotic patients, females, Hispanics, and Asians with SMI, may be particularly vulnerable to obesity or T2DM; therefore, careful selection of psychotropic medications is essential, along with healthy lifestyle interventions and regular monitoring of metabolic conditions. Future research should further elucidate genetic, inflammatory, and environmental mechanisms underlying these metabolic conditions for the purpose of generating optimal treatment and services.

Biography

Langston Sun is a medical student at NYU School of Medicine, Mara Getz is a research coordinator at NYU School of Medicine, Sulaima Daboul, M.D. is a research assistant at NYU School of Medicine, Melanie Jay, M.D., M.S. is an Associate Professor in DGIMCI, Scott Sherman, M.D. is an Associate Professor in the Department of Population Health at NYU School of Medicine, Erin Rogers, MPH, Ph.D is an Assistant Professor in the Department of Population Health at NYU School of Medicine, Nicole Aujero is a research coordinator at NYU School of Medicine, Raymond Goetz, Ph.D is a Research Scientist and an Assistant Professor at the New York Psychiatric Institute and Columbia University, Judith Weissman, Ph.D, JD is an Assistant Professor of Psychiatry and Population Health at the Icahn Medical School at Mt. Sinai, Dolores Malaysian, M.D., MPH was Professor of Psychiatry at NYU Langone Medical Center and the Director of the Institute for Social and Psychiatric Initiatives before relocating to the Icahn School of Medicine at Mt. Sinai, NY, to develop a psychosis program, and Samoon Ahmed, M.D. is a Clinical Associate Professor of Psychiatry at NYU School of Medicine and Unit Chief Psychiatry Inpatient at Bellevue Hospital.