Testing for Lipid Levels and Diabetes among Autistic Youth who are and are not Prescribed Anti-Psychotics: A Cohort Study

Abstract

Introduction:

Autistic youth have higher rates of hyperlipidemia and diabetes than non-autistic youth and thus need age-appropriate monitoring for hyperlipidemia and diabetes. Little is known about how frequently autistic youth are monitored for these conditions.

Methods:

This analysis assessed monitoring for hyperlipidemia and diabetes among 230 autistic youth ages 16 to 30 years (113 were prescribed anti-psychotics, 117 were not) between January 2011 and May 2020. Outcomes assessed included the proportion of patients who had ANY testing for hyperlipidemia and diabetes in both groups, proportion of prescriptions monitored for hyperlipidemia and diabetes in the last year, and identification of patient factors associated with monitoring.

Results:

A significantly higher proportion of autistic youth prescribed anti-psychotics had testing for both hyperlipidemia and diabetes during the study period than autistic youth who were not (73% vs. 49%, p< 0.001). While most autistic youth who were prescribed anti-psychotics had some monitoring done, of the 1538 prescriptions for anti-psychotics (new and renewal) identified, 847 (55%) were considered unmonitored. Having other bloodwork done was significantly associated with higher odds of testing for hyperlipidemia or diabetes (OR 1.45, 95% CI [1.37, 1.56]), but not other factors assessed.

Conclusions:

Most autistic youth prescribed anti-psychotics in this cohort underwent some monitoring for hyperlipidemia and diabetes, but not at a level consistent with guidelines. Many autistic youth not prescribed anti-psychotics are not getting testing for hyperlipidemia or diabetes. Providers should consider adding lipid and diabetes testing to other bloodwork, as this was positively associated with monitoring in this analysis.

Introduction

Diabetes is more common among autistic adults than non-autistic adults,^1–3 with one study finding that autistic young adults ages 22-27 have higher odds of having diabetes compared to their non-autistic peers (OR 2.67, 95% CI = 2.33-3.06).¹ Dyslipidemia is more common in autistic adults than those who are not autistic.^3,4 While the rates of dyslipidemia are low in autistic young adults (2.6% of 22 year olds), they are much higher than non-autistic peers (0.3%).⁴

The higher prevalence of diabetes and dyslipidemia in autistic young adults is likely driven by a number of risk factors, including higher rates of obesity in autistic youth^2,5,6 as well as higher prevalence of use of anti-psychotic medication.^7–9 Anti-psychotics can induce dyslipidemia and diabetes as part of their side effect profile in adults¹⁰ and in children and adolescents.¹¹ Guidelines recommend monitoring for diabetes and hyperlipidemia annually among adults with risk factors,^12,13 every 3-5 years for those without risk factors,^12,13 and also that all youth get their lipid levels checked at least once between the ages of 17 and 21, whether or not risk factors are present.¹⁴

Studies have looked at adherence to these guidelines generally, but not specifically at adherence to guidelines for autistic youth with risk factors. One study found that, among a cohort of adults with a serious mental illness who were prescribed antipsychotics, fewer than half of adults ages 18 to 29 years (43%) and fewer than half of autistic adults (49%) were screened for diabetes in a given year.¹⁵ The study did not look specifically at autistic adults ages 18 to 29 years. Another study found that 38% of adults ages 18 to 27 years on Medicaid with serious mental illness who were prescribed anti-psychotics did not get any diabetes testing.¹⁶ The study did not specifically assess screening rates among autistic people.

Thus, there is a need to understand rates of screening for diabetes and hyperlipidemia specifically in autistic adolescents and young adults (hereafter called autistic youth). Study of screening in this age group is particularly important because autistic youth are particularly vulnerable to disruptions in care during the transition from pediatric to adult health services.^17–21 The aim of this study is to address this gap using analysis of data from a cohort of autistic youth who were in the midst of their transition from pediatric to adult health care.

Methods

This study was approved by the Nationwide Children’s Hospital Institutional Review Board without requirement for consent due to the retrospective nature of the data collection and analyses. The data are considered protected health information, and so not publicly available, but interested parties can contact the corresponding author regarding data access.

Population

Eligible youth for this study were identified and included based on the following eligibility criteria: 1) they had been seen at least once at the Center for Autism Services and Transition (CAST), a primary care-based program established to provide primary care to autistic adults that is adapted to the unique needs of this population and with particular attention to their transition to adult care^22,23 and 2) they had at least one encounter at Nationwide Children’s Hospital (NCH) prior to their first CAST appointment. NCH and CAST are located less than 15 miles apart, so many CAST patients have been referred to CAST by an NCH clinician. Thus, CAST patients are a group well-suited for studying the transition to adult health care. Patients had to have one BMI measurement recorded, which could be at NCH or CAST, in order to assess for risk factors of diabetes and hyperlipidemia appropriately.

In addition to the above criteria, this analysis was limited to patients between the ages of 16 and 30 years at the time of their first appointment with the CAST clinic.

All patients enrolled in CAST have their diagnosis of autism confirmed via review of medical and psychological records at the time of enrollment with CAST. Although the DSM-5 introduced three levels of severity to differentiate autistic people’s support needs, the diagnoses in the electronic medical record (EMR) where the data was obtained did not consistently include severity levels, so there was no means to differentiate patients based on severity or support needs.

Measures

To acquire the data used in this analysis, people seen as patients in CAST were cross-referenced with people seen as patients at NCH and matched by name and date-of-birth by an honest broker (a data expert who was not part of the study team) who then provided a deidentified dataset to the study team. For all matched patients, data from the EMR from both institutions for the dates January 2011 to May 2020 were pulled and linked so analyses could be performed across both institutions over time. Analyses were completed using this linked data set.

The variables used in this analysis were:

• Prescriptions for anti-psychotics – Anti-psychotics are a group of medications used for management of schizophrenia and other disorders associated with psychosis.²⁴ In addition, they are used to treat multiple symptoms associated with autism, including self-injury, irritability, and insomnia.²⁵ There are two main categories: first-generation and second-generation.²⁴ While certain medications (ex. olanzapine, clozapine) have stronger associations with metabolic side effects, all drugs in the class carry increased risk for hyperlipidemia and diabetes.^10,11 For this study, all outpatient prescriptions for both first-generation anti-psychotics (ex. haloperidol) and second-generation anti-psychotics (ex. risperidone) were identified in the EMR, including the date the medication was prescribed. Inpatient and emergency department orders for these medications were not included to remove those who may have received these medications as sedation in the inpatient and/or emergency department setting but were not taking them regularly and so would not meet criteria for monitoring.

• Hyperlipidemia testing – All measurements of low-density lipoprotein

  (LDL) -- calculated or directly measured – were included as measures of monitoring for hyperlipidemia.

• Diabetes testing – All hemoglobin A1C and glucose measurements in the EMR were considered monitoring for diabetes testing for two main reasons. First, certain insurances, including Medicare, will only cover hemoglobin A1C testing for diagnosis and monitoring of diabetes, but not for screening for diabetes. Secondly, there was no reliable way to differentiate fasting from non-fasting glucose levels.

• Body Mass Index (BMI) – BMI was calculated per the standard formula – weight in kilograms / height in meters squared. Either the average of the last 3 BMI measurements (if done in the last year) or the next BMI value determined after the date of the prescription were used in the regression analysis. Because analyses were limited to those between the ages of 16 and 30, the BMI percentiles used in pediatric care were not employed, as the percentiles are not used in adult care and most of the patients in the study were adults.

• Age – Age at the time of the first CAST appointment was used for the purpose of generating descriptive statistics. For the logistic regression, age was normalized such that the age input into the regression was the number of years above age 16 for each patient for each year after their first prescription for an anti-psychotic. For example, if someone had a first prescription for an anti-psychotic at age 17 years, and were now 22 years old, then the normalized age for the linear regression would be 5 because 22 minus 17 is 5.

• Receipt of other blood work – Patients who had and had not gotten other blood laboratory testing were identified within the EMR to determine if getting other blood testing was associated with receipt of hyperlipidemia or diabetes testing.

• Seizure medications – As has been noted previously,²⁶ two of the authors classified all medications prescribed to patients in this cohort into 32 higher level groupings to categorize the medications prescribed according to common classes and indications. A custom classification system was used in lieu of standard medical taxonomies to ensure that the groupings were salient to autistic people. Certain medications are used in off-label ways among autistic people, which is better-captured using this custom classification system. The only additional grouping from that system used here was the “seizure medication” grouping, which was included in the logistic regression because seizure medications also frequently require laboratory monitoring and so may influence monitoring rates.

• Site of care – Each year that a patient contributed to the regression analysis was identified as occurring either before or after the first visit to CAST in order to see if there were differences between pediatric and adult care in monitoring.

Statistical Analysis

Descriptive statistics of patients who were prescribed anti-psychotics and patients who were not were generated and compared using Fisher’s exact test and Mood’s median test as appropriate.

Monitoring of hyperlipidemia and diabetes at the patient level was assessed both for those prescribed anti-psychotics and those who were not. For those who were prescribed anti-psychotics, further descriptive analysis was done to determine how many prescriptions (new and renewal) were monitored. A prescription was considered monitored for hyperlipidemia if the patient had an LDL done within a year of the date of the prescription. A prescription was considered as being monitored for diabetes if the patient had a hemoglobin A1C or glucose level within 1 year of the date of the prescription. A prescription was considered as “being monitored” if it was monitored for both hyperlipidemia and diabetes.

Logistic regression was conducted to determine which patient characteristics (e.g. Age, BMI, obtaining other blood work, taking seizure medications, transfer to the CAST clinic) were associated with increased odds of a prescription being monitored. For each patient, the dates of their first and last anti-psychotic prescription between the ages of 16 and 30 were calculated. Then, starting with the first prescription date, the characteristic variables for that year were calculated and used as a single datapoint. This process was repeated for each year until either the patient turned 30 or reached a year past their last prescription. This year-by-year method prevented overcounting prescriptions that happen very close together, as some patients will have many prescriptions in a one-year period. To account for prescriptions from the same individual not being independent and having within-patient correlation, a patient-specific intercept was included for each individual, which ensured that all observations from the same patient shared a unique baseline term.²⁷

Results

Table 1 shows the demographic information for the cohort, stratified by receipt of anti-psychotic prescriptions (or not). Mean ages, proportion of male patients, and proportion of white patients are similar in both groups (mean age of 19.6 years for those prescribed anti-psychotics vs. 19.8 years for those who were not; 82% male patients in both groups; 65% of patients were white in the group who were prescribed anti-psychotics vs. 70% of patients who were in white in the group that was not). A higher percentage of patients prescribed anti-psychotics had public insurance (85%) compared to those who were not (70%) [p = 0.011].

Table 1:

Demographics

	Patients Prescribed an Anti-Psychotic (n= 113)	Patients not Prescribed an Anti-Psychotic (n = 117)	p-value
Mean age at the time of the first adult visit (IQR)a	19.6 (2.32)	19.8 (3.97)	0.283
Male gender (%)	93 (82.3%)	96 (82.1%)	1.000
Median BMI at first adult visit (kg/m2)	27.7 (n = 89)	27.3 (n = 86)	1.000
Race
Black	19 (16.8%)	16 (13.7%)	0.583
White (%)	74 (65.5%)	83 (70.9%)	0.398
Insurance
Public (%)	96 (85.0%)	83 (70.9%)	0.011
Private (%)	68 (60.2%)	84 (71.8%)	0.079

^a.Study was limited to those between the ages of 16 and 30

Table 2 shows the proportion of patients who got lab monitoring for hyperlipidemia and/or diabetes over the study period. A higher percentage of those who were prescribed anti-psychotics had undergone testing for hyperlipidemia, diabetes, and both of these conditions, with only 9% having no monitoring. On the other hand, among those who were not prescribed anti-psychotics, 34% had undergone no monitoring for either condition over the study period.

Table 2:

Proportions of Patients with ANY Lab Monitoring for Hyperlipidemia and/or Diabetes

	Patients Prescribed an Anti-Psychotic (n=113)	Patients not Prescribed an Anti-Psychotic (n= 117)	p-value
Had at least one LDL measurement n(%)	101 (89%)	70 (60%)	< 0.001
Had at least one A1C or glucose check n(%)	84 (74%)	64 (55%)	0.003
Had both LDL and A1C or glucose check at least once n(%)	82 (73%)	57 (49%)	< 0.001
Had no checks of LDL or A1C or glucose n(%)	10 (9%)	40 (34%)	< 0.001

Note: Boldface indicates statistical significance (p< 0.05)

Table 3 shows the proportion of prescriptions that were considered monitored, meaning the patient completed monitoring labs within a year of the date the prescription was written. There were a total of 1538 prescriptions across all patients. For over half (55%) of the prescriptions, the patient had no monitoring within a year, with an additional 22% that had monitoring for one condition, but not the other. Only 23% of the prescriptions were considered fully monitored.

Table 3:

Proportion of Prescriptions for Anti-Psychotics with Appropriate Monitoring for Hyperlipidemia and Diabetes

	Diabetes Monitored	Diabetes not Monitored	Total
Hyperlipidemia Monitored n(%)	359 (23%)a	61 (4%)b	420 (27%)
Hyperlipidemia not Monitored n(%)	271 (18%)c	847 (55%)d	1118 (73%)
Total n(%)	630 (41%)	908 (59%)	1538 (100%)

^a.Patient got LDL AND either hemoglobin A1c or glucose within one year of date prescription was written.

^b.Patient got an LDL checked within one year of the date of the prescription, but not a hemoglobin A1C or glucose.

^c.Patient got a hemoglobin A1C or glucose within one year of the date of the prescription, but not an LDL.

^d.Patient got no testing for LDL, hemoglobin A1C, or glucose within 1 year of the prescription.

Table 4 shows the logistic regression results. There was a total of 701 per-patient non-overlapping year-long periods following a prescription across the cohort. The only significant factor associated with getting lab monitoring was getting other blood work (OR 1.45, 95% CI = 1.37 – 1.56). Age at the time of the prescription, BMI at the time of the prescription, and transfer to adult care all had odds ratios close to 1. Taking seizure medications had an OR of 1.24, but this was not statistically significant (95% CI = 0.98 – 1.59)

Table 4:

Logistic Regression of Patient-Level Factors Associated with Obtaining Appropriate Lab Monitoring Among those Prescribed Anti-Psychotics (n=701)

Variable	OR (95% CI)
Age at time of prescription	1.00 ([0.98, 1.02])
BMI at time of prescription	1.01 ([0.99, 1.02])
Other blood work obtained	1.45 ([1.37, 1.56])
Taking seizure medications	1.24 ([0.98, 1.59])
Transferred to CAST	1.03 ([0.94, 1.13])

Discussion

Among a cohort of autistic youth, most youth had undergone some monitoring for hyperlipidemia and/or diabetes over the course of their transition to adult health care. Additionally, those autistic youth who were prescribed anti-psychotics more frequently completed monitoring than those who did not. Among those prescribed anti-psychotics, the major factor associated with undergoing monitoring was having other lab work done.

It is also notable that while most patients had testing done at some point over the study period, the majority of prescriptions had no monitoring within a year of the prescription. This suggests that monitoring is done, but not as consistently as guidelines recommend. Possible reasons for this inconsistency include delays due to difficulty with blood draws for autistic youth, priority being given to other components of care, and lack of provider knowledge about the need for regular testing. The association between getting other blood work done and getting monitoring for hyperlipidemia / diabetes suggests an opportunity to improve monitoring overall by identifying and incorporating strategies to have testing for hyperlipidemia / diabetes added to other needed testing. Quality improvement work to fill this gap is needed.

While most autistic youth who were prescribed anti-psychotics had undergone monitoring for hyperlipidemia and diabetes in this cohort, only about half of autistic youth who were not prescribed anti-psychotics had done so. This is concerning because autistic adults have increases in their lipid levels with age, whether or not they are taking anti-psychotics.²⁸ The relative lack of screening for hyperlipidemia and diabetes in this cohort may represent missed opportunities to address cardiovascular prevention in this population, though this cannot be assessed in the current study. There are no specific guidelines on recommended monitoring for hyperlipidemia or diabetes in autistic youth. Guidelines for monitoring for these conditions in adolescents and young adults differ. The American Academy of Pediatrics recommends lipid screening for all adolescents and young adults at least once between the ages of 17 and 21 years.¹⁴ The American Heart Association recommends lipid testing every 5 years for adults ages 20 to 39 years.¹³ So, if the general population guidelines were being followed, then one would expect that most of this cohort should have had monitoring for hyperlipidemia at least once in the study period, whether they were prescribed anti-psychotics or not. The American Diabetes Association recommends screening for youth and young adults (ages 10 to 35 years) with overweight or obesity and have at least one other risk factor, which includes having high blood pressure or having a first degree relative with diabetes.¹² It was not possible, given the data available, to identify precisely an expected number of autistic youth in the cohort who would be expected to have diabetes screening. Based on the available BMI data in the dataset, the median BMI as of the first adult visit among those who were not prescribed anti-psychotics was 27, suggesting over half of the patients were considered to have overweight and so should have been considered for screening, depending upon other risk factors.

It is important to note that both hemoglobin a1c and glucose were considered eligible for monitoring for diabetes. However, since a glucose level is obtained as part of a basic metabolic panel, it is likely that some of the glucose tests considered to be monitoring tests may only have been ordered for other purposes (e.g., hyperglycemia or simply within a basic metabolic panel). In these data, approximately 1 in 5 prescriptions had monitoring performed for diabetes but not for hyperlipidemia, supporting this possibility. The study was not powered to explore this possibility in further detail, but this is an important area of future study.

It is possible that some screening occurred just outside of the 12-month window that was set for appropriate monitoring of prescriptions. At the same time, with 55% of prescriptions being considered unmonitored, the lack of monitoring remains notable, even accounting for some fraction of testing being just beyond the time limit used in this study.

Limitations

This study has limitations. The cohort was seen at one children’s hospital and one clinic specifically for autistic adults, and so many not be reflective of the care of autistic youth broadly. In fact, the comprehensive nature of care at CAST means that it’s possible that screening rates for these conditions are lower for autistic people generally than the rates found here. Most of the patients were male and white. The proportion of male patients is consistent with gender breakdown for autistic people, but limits the ability to compare male and female autistic people. Similarly, the racial breakdown of patients in this cohort is similar to Ohio generally, but the findings may not be reflective of those who are not white.

A person’s functional abilities and level of support needs likely influences the frequency of monitoring, and so these findings may not apply evenly to all autistic people. However, data regarding level of support needs was not available in the EMR. Thus, the lack of data regarding the level of support needs for the participants in this study is also a limitation. Additionally, the indication for starting or continuing treatment with anti-psychotics in this cohort is not clearly delineated in the record, and those indications may also influence the frequency of screening, but that influence couldn’t be assessed for here.

Conclusions

While most autistic youth are getting some screening for hyperlipidemia and diabetes during the transition to adult health care, most prescriptions for anti-psychotics among autistic youth are not getting the level of monitoring recommended by guidelines, which represents an area for improvement in the care of this population.

Highlights.

• Autistic youth have higher rates of cardiovascular disease than non-autistic peers.

• Autistic youth also take anti-psychotic medications frequently.

• Thus, autistic youth should be getting regular lipid and diabetes monitoring.

• Monitoring of autistic youth is happening, but not at recommended intervals.