Phillip Crawford

Medication-assisted treatment for substance use disorders within a national community health center research network

ABSTRACT Background: The Affordable Care Act increases access to treatment services for people who suffer from substance use disorders (SUDs), including alcohol use disorders (AUDs) and opioid use disorders (OUDs). This increased access to treatment has broad implications for delivering health services and creates a dramatic need for transformation in clinical care, service lines, and collaborative care models. Medication-assisted treatments (MAT) are effective for helping SUD patients reach better outcomes. This article uses electronic health record (EHR) data to examine the prevalence of EHR-documented SUDs, patient characteristics, and patterns of MAT prescribing and screening for patients within the Community Health Applied Research Network (CHARN), a national network of 17 community health centers that facilitates patient-centered outcomes research among underserved populations. Methods: Hierarchical generalized linear models examined patient characteristics, SUD occurrence rates, MAT prescription, and human immunodeficiency virus (HIV) and hepatitis virus C screening for patients with AUDs or OUDs. Results: Among 572,582 CHARN adult patients, 16,947 (3.0%) had a documented AUD diagnosis and 6,080 (1.1%) an OUD diagnosis. Alcohol MAT prescriptions were documented for 547 AUD patients (3.2%) and opioid MAT for 1,764 OUD patients (29.0%). Among OUD patients, opioid MAT was significantly associated with HIV screening (odds ratio [OR] = 1.31, P < .001) in OUD patients, as was alcohol MAT among AUD patients (OR = 1.30, P = .013). Conclusions: These findings suggest that effective opioid and alcohol MAT may be substantially underprescribed among safety-net patients identified as having OUDs or AUDs.

Health Services Utilization Among Children With and Without Autism Spectrum Disorders

Using data from multiple health systems (2009–2010) and the largest sample to date, this study compares health services use among youth with and without an autism spectrum disorder (ASD)—including preventive services not previously studied. To examine these differences, we estimated logistic and count data models, controlling for demographic characteristics, comorbid physical health, and mental health conditions. Results indicated that youth with an ASD had greater health care use in many categories, but were less likely to receive important preventive services including flu shots and other vaccinations. An improved understanding of the overall patterns of health care use among this population could enable health systems to facilitate the receipt of appropriate and effective health care.

Excess Clinical Comorbidity Among HIV-Infected Patients Accessing Primary Care in US Community Health Centers:

Objectives: As the life expectancy of people infected with human immunodeficiency virus (HIV) infection has increased, the spectrum of illness has evolved. We evaluated whether people living with HIV accessing primary care in US community health centers had higher morbidity compared with HIV-uninfected patients receiving care at the same sites. Methods: We compared data from electronic health records for 12 837 HIV-infected and 227 012 HIV-uninfected patients to evaluate the relative prevalence of diabetes mellitus, hypertension, chronic kidney disease, dyslipidemia, and malignancies by HIV serostatus. We used multivariable logistic regression to evaluate differences. Participants were patients aged ≥18 who were followed for ≥3 years (from January 2006 to December 2016) in 1 of 17 community health centers belonging to the Community Health Applied Research Network. Results: Nearly two-thirds of HIV-infected and HIV-uninfected patients lived in poverty. Compared with HIV-uninfected patients, HIV-infected patients were significantly more likely to be diagnosed and/or treated for diabetes (odds ratio [OR] = 1.18; 95% confidence interval [CI], 1.22-1.41), hypertension (OR = 1.38; 95% CI, 1.31-1.46), dyslipidemia (OR = 2.30; 95% CI, 2.17-2.43), chronic kidney disease (OR = 4.75; 95% CI, 4.23-5.34), lymphomas (OR = 4.02; 95% CI, 2.86-5.67), cancers related to human papillomavirus (OR = 5.05; 95% CI, 3.77-6.78), or other cancers (OR = 1.25; 95% CI, 1.10-1.42). The prevalence of stroke was higher among HIV-infected patients (OR = 1.32; 95% CI, 1.06-1.63) than among HIV-uninfected patients, but the prevalence of myocardial infarction or coronary artery disease did not differ between the 2 groups. Conclusions: As HIV-infected patients live longer, the increasing burden of noncommunicable diseases may complicate their clinical management, requiring primary care providers to be trained in chronic disease management for this population.

Medical Conditions in the First Years of Life Associated with Future Diagnosis of ASD in Children

This study examines medical conditions diagnosed prior to the diagnosis of autism spectrum disorder (ASD). Using a matched case control design with 3911 ASD cases and 38,609 controls, we found that 38 out of 79 medical conditions were associated with increased ASD risk. Developmental delay, mental health, and neurology conditions had the strongest associations (ORs 2.0–23.3). Moderately strong associations were observed for nutrition, genetic, ear nose and throat, and sleep conditions (ORs 2.1–3.2). Using machine learning methods, we clustered children based on their medical conditions prior to ASD diagnosis and demonstrated ASD risk stratification. Our findings provide new evidence indicating that children with ASD have a disproportionate burden of certain medical conditions preceding ASD diagnosis.

PS1-13: Variation in the Incidence and Prevalence of Autism from Multiple Health Systems: Findings from the Mental Health Research Network Autism Registry Study

Background/Aims Autism spectrum disorders (ASD) are characterized by impairments in social interaction and communication, as well as restricted, stereotyped interests and behaviors. A recent study found that approximately 1 in 88 children in the U.S. were diagnosed with an ASD and that prevalence varied widely among different demographic groups. The goals of this study were to obtain accurate prevalence and incidence statistics for ASD across several large, diverse health systems and to describe the variation of these statistics across demographic factors. Methods All members within the five participating health systems born between January 1, 1993 and December 31, 2008 with electronic claims, enrollment, or medical record information were included in the study. Information on member demographics and ASD subtypes were collected from earliest available records at each site through the end of December 31, 2010. Individuals with an ASD diagnosis from an ASD specialist or two or more ASD diagnoses from non-specialists were defined as valid cases. Results A preliminary examination of data from one site (N = 1,271,823) found 10,114 individuals <18 years ever diagnosed with an ASD. Of those 10,114 ASD cases, 8,085 met the validation criteria and were included in final analyses. Prevalence of all ASDs in children ≤8 years old was 1.1/1000 in 2001 (1 in 909 children) and increased steadily to 7.1/1000 in 2010 (1 in 141 children). Prevalence specifically for autistic disorder (AD), a more severe subtype, in children ≤8 years old was 0.3/1000 in 2001 and increased to 1.9/1000 in 2010. Similar secular increases were noted for incidence. Prevalence and incidence varied greatly among demographic groups. Prevalence of all ASDs in 2010 was 8.4/1000 among Whites, 7.1/1000 among Blacks, and 10.6/1000 among Asians. Prevalence of ASDs among females was lower than among males in all years (2010 males: 11.2/1000, 2010 females: 2.8/1000). Conclusions This study provides up-to-date prevalence and incidence information from a group of large, diverse, community-based settings. Incidence and prevalence differed across racial groups and sex status. Strong increasing trends in the diagnosis of ASDs in general, as well as the AD subtype, were observed.

C-B1-01: Psychotherapeutic Prescription Patterns Across Health Plans in the Mental Health Research Network

Background The need for effectiveness and dissemination research is at least as great in mental health as in other areas of health care. The Mental Health Research Network aims to use the VDW for multi-site studies of mental disorder epidemiology, treatment patterns, or treatment effectiveness. We expect that the existing infrastructure is satisfactory in some areas, while others will need improvement or expansion. We present early results from validation studies of enrollment and pharmacy data on patterns of psychotherapeutic drug use. Methods We identified the relevant denominator populations using existing VDW variables indicating continuous enrollment with any drug coverage in 2009. After developing National Drug Code (NDC) lists for drugs used in the treatment of depression, bipolar disorder, attention-deficit disorder, and other mental health conditions, we calculated rates of use of these drugs across ten health plans in 2009. Results Antidepressant exposure rates vary across sites, although our data include no diagnostic criteria and will include uses for other conditions. Initial results indicate overall antidepressant use rates vary more than two-fold between ten sites, from 62 to 162 exposures per thousand members. In addition, the preferred member of a class (for example SSRIs), and the rate of any pharmacy utilization also differs across sites, although to a lesser degree. Conclusions Data validation is an iterative process where results are compared to what is known and expected and discrepancies are investigated. The considerable variation in overall antidepressant exposure rates at different sites is unexpected, based on reports of the incidence of depression (ref) and warrants further exploration. Possible explanations include use of antidepressants for other conditions, cultural attitudes toward mental health treatment in different populations, or delivery system practices. Alternatively, they may be artifacts of information systems or incorrect interpretations of the underlying data. These possibilities can be explored with a combination of local expertise from each delivery system and further analysis of VDW data.

Changes in Coding of Suicide Attempts or Self-Harm With Transition From ICD-9 to ICD-10

Coding of self-harm or possible suicidal behavior changed significantly with the transition from ICD-9-CM to ICD-10-CM. Health systems that use diagnosis codes to monitor rates of suicide attempts should consider the impact of this change to distinguish real trends from coding artifacts. We used data from ten health systems in the Mental Health Research Network (MHRN) to track rates of injury or poisoning diagnoses, classified as either self-inflicted or of undetermined intent, across the October 2015 transition. In ICD-9-CM, self-harm was indicated by recording additional cause-of-injury diagnoses in the range E950–E958 (definitely selfinflicted) or E980–E988 (unknown if accidental or self-inflicted). For example, intentional benzodiazepine overdose was indicated by a primary diagnosis of 969.4 and an additional diagnosis of E950.0. In ICD-10-CM, self-harm is indicated within the primary diagnosis by a required causeof-injury code, typically found in the fifth or sixth position. For example, T42.4X2 indicates intentional benzodiazepine overdose, whereas T42.4X4 indicates overdose of undetermined intent. Diagnoses were extracted from federated data warehouses containing electronic medical record and insurance claims data (1) from April 2014 through March 2016 at each participating site: Group Health Cooperative (Washington), HealthPartners (Minnesota), Essentia Health (Wisconsin), Henry Ford Health System (Michigan), and five Kaiser Permanente regions (Colorado, Georgia, Hawaii, Southern California, and Oregon). Monthly rates of persons receiving a diagnosis of interest were calculated relative to total monthly enrollment. Figure 1 shows monthly rates of injury or poisoning per 10,000 enrolledmembers bywhether the injury or poisoning was intentionally self-inflicted or of undetermined intent. Diagnoses of self-inflicted injury or poisoning appeared to increase abruptly with the coding transition, and this pattern was consistent across health systems [see online appendix]. Diagnoses of injury or poisoning of undetermined intent appeared to decrease with the coding transition, but this pattern varied considerably across health care systems [see online appendix]. Marked changes in coding of intent for injuries and poisonings during the fall of 2015 almost certainly represent artifacts of coding changes rather than true changes in suicidal behavior. Therefore, the transition from ICD-9-CM to ICD10-CM may affect research, surveillance, and quality improvement related to preventing suicidal behavior and should be considered when interpreting trends during this period. Effects of coding changes may vary across electronic health records systems or with different terminology mapping implemented in those systems. Across the ten MHRN health systems, we observed similar patterns for diagnoses of injury or poisoning with intentional self-harm and variable patterns for diagnoses of injury or poisoning of undetermined intent. Our specific findings may not generalize to other health systems, but they underline the importance of close attention to coding changes in all U.S. health care settings.

Implications of American College of Cardiology/American Heart Association (ACC/AHA) Cholesterol Guidelines on Statin Underutilization for Prevention of Cardiovascular Disease in Diabetes Mellitus Among Several US Networks of Community Health Centers

Background Little is known about statin underutilization among diabetes mellitus patients cared for in community health centers, which tend to serve socioeconomically disadvantaged populations. Implications of the American College of Cardiology/American Heart Association (ACC/AHA) guidelines on preexisting gaps in statin treatment in this population are unclear. Methods and Results We included 32 440 adults (45% male, 63% nonwhite, 29% uninsured/Medicaid) aged 40 to 75 years with diabetes mellitus who received care within 16 community health center groups in 11 states in the Community Health Applied Research Network during 2013. Statin prescribing was analyzed as a function of concordance with the National Cholesterol Education Program Adult Treatment Panel 2001 and ACC/AHA 2013 guidelines. More patients’ treatments were concordant with the ACC/AHA (52.8%) versus the National Cholesterol Education Program Adult Treatment Panel (36.2%) guideline. Female sex was associated with lower concordance for both (odds ratio [OR] 0.90, CI 0.85‐0.94; and OR 0.84, CI 0.80‐0.88, respectively). Being insured, an Asian/Pacific Islander, or primarily Spanish speaking were associated with greater concordance for both guidelines: 35.5% (11 526/32 440) were concordant with neither guideline, the majority (79.7%) having no statin prescribed; 28.2% (9168/32 440) were concordant with ACC/AHA but not the National Cholesterol Education Program Adult Treatment Panel. 8.7% of these patients had a low‐density lipoprotein cholesterol >160 mg/dL despite having a moderate‐ or high‐intensity statin prescribed. And 11.6% (3772/32 440) were concordant with the National Cholesterol Education Program Adult Treatment Panel but not with ACC/AHA. Most of these patients had a low‐density lipoprotein cholesterol between 70 and 99 mg/dL with no or a low‐intensity statin prescribed. Conclusions Opportunities exist to improve cholesterol management in diabetes mellitus patients in community health centers. Addressing care gaps could improve cardiovascular disease prevention in this high‐risk population.

A4-1: Validating Electronic Health Record Demographic Data Using Self-reported Data from the Autism Registry Survey

Background/Aims The use of administrative patient data via the electronic health record (EHR) is very important in research. It’s critical we have a sense of the validity and relative accuracy of key data in this widely available data source. The HMORN Virtual Data Warehouse (VDW) is an example of a large repository of administrative data used in numerous studies. The recent Mental Health Research Network (MHRN) Autism Spectrum Disorder (ASD) Registry Survey was hosted by four HMORN sites inviting Kaiser Permanente members to complete a Web-based questionnaire on behalf of their child identified as having an ASD diagnosis in the EHR. Methods In addition to an extensive battery of questions regarding ASD, the survey collected data on a number of demographic data elements also available in the VDW. These are: age, gender, race/ethnicity (child), income (household) and education (household adult). A total of 1155 adults responded to the ASD Registry Survey. These records were matched with demographic data from the VDW for the same child. Results Preliminary examination of race-ethnicity and gender data shows that there‘s a good to excellent level of agreement between the two sources when data are non-missing in both data sources. A total of 992 records had non-missing race/ethnicity data from both data sources. In 81.1% (805/992) of the cases, both data sources agreed (kappa = 0.68, CI = 0.64–0.72). The categories that had the highest level of agreement are White, Black, and Asian, while the Hispanic and multi-racial groupings had a comparatively much lower level of agreement (46.9%). For gender, level of agreement was very high (99.0%, 1118/1129, kappa = 0.97, CI = 0.95–0.99). Conclusions In this study, race in the electronic health record was a very accurate measure for major race categories but was far less accurate when reporting the emergent and important multi-racial category and Hispanic ethnicity. Gender had an excellent level of agreement between the two sources. Age, gender, and race-ethnicity are key covariates for many studies analyzing EHR data. Income and education can serve to illuminate socioeconomic factors very relevant to health care research. Having a sense of the validity and accuracy of these data is crucial to the research process.

PS1-14: Health-Related Quality of Life in Children with Autism Spectrum Disorders: Findings from the Mental Health Research Network Autism Registry Web Survey

Background/Aims Approximately 1 in 88 children in the U.S. is diagnosed with Autism Spectrum Disorder (ASD). ASD is a complex disorder characterized by impairment in social skills, communication, and cognitive and behavioral functioning. In order for policy makers and clinical managers to evaluate new approaches to treating and managing ASD, they need brief comprehensive outcome measures. One approach that could be useful in this context is measurement of health-related quality of life (HR-QOL), which provides a comprehensive picture of health status including an individual’s psychosocial, emotional, and physical wellbeing. This comprehensive approach is particularly important in conditions such as ASD that have multiple impacts on a person’s health. Few previous studies have examined HR-QOL in persons with ASD, and most of these studies have used small samples. The purpose of this analysis is to examine HR-QOL in a group of geographically- and racially/ethnically-diverse children with ASD who are enrolled in the Mental Health Research Network (MHRN) Autism Registry. Methods A Web-based survey of parents of children with ASD was implemented at four MHRN Autism Registry sites, including children’s HR-QOL, measured by the Pediatric Quality of Life Inventory (PedsQLTM). The PedsQLTM provides an overall score, as well as subscales for important domains including physical health, psychosocial health, emotional functioning, social functioning, and school functioning. Results To date, recruitment letters have been mailed to approximately 8800 parents and 800 surveys have been completed. Preliminary analyses of respondents indicate that HR-QOL is lower in children with ASD compared to national norms. We will present the final results from the survey, which will conclude in November 2012. The presentation will examine the overall scores, scores on subscales, and scores by subgroup (e.g., age, gender, race) and will compare these scores to national norms. Conclusions We successfully implemented a Web-based survey of parents of children with ASD across four MHRN sites. With 800 completed surveys (recruitment will continue through November 2012), this is the largest known population-based survey on children with ASD to date. The current study will help to confirm results from smaller samples and will allow for more refined analyses of subgroups.