Possible Alternatives to Diagnosis-Based Denominators for Addiction Treatment Quality Measures

Comparison of Substance Use Disorder Diagnosis Rates From Electronic Health Record Data With Substance Use Disorder Prevalence Rates Reported in Surveys Across Sociodemographic Groups in the Veterans Health Administration

Importance

Substance use disorders (SUDs) are major contributors to morbidity and mortality globally, but they are often underrecognized and underdiagnosed, particularly in some sociodemographic subgroups. Understanding the extent to which clinical diagnoses underestimate these conditions within subgroups is imperative to achieving equitable treatment, regardless of race, ethnicity, gender, or age, and to informing and improving performance monitoring.

Objective

To compare clinically documented diagnosis rates of alcohol use disorder (AUD), drug use disorder (DUD), and total SUD (AUD and/or DUD) with the prevalence of these disorders as reported in surveys-based on structured, validated diagnostic assessments-across demographic subgroups.

Design, Setting, and Participants

A telephone-based survey was conducted from January 8, 2018, to April 30, 2019, among 5995 Veterans Health Administration (VHA) outpatients who were randomly sampled from 30 VHA facilities and were 18 years of age or older, could complete the survey in English, and had a valid address and telephone number. Survey data were linked to electronic health record (EHR) data for all participants. Statistical analysis was performed between January 29, 2020, and April 20, 2021.

Exposures

Demographic subgroups based on self-report: gender (male or female), age (18-34, 35-49, 50-64, 65-74, and ≥75 years), and race and ethnicity (Black non-Hispanic, Hispanic, multiracial, other [Asian or Asian-American, American Indian or Alaskan Native, Native Hawaiian or Pacific Islander, and any other race endorsed by the participant], and White non-Hispanic).

Main Outcomes and Measures

Survey-based prevalence rates of AUD, DUD, and SUD were assessed using the Mini International Neuropsychiatric Interview, version 7.0, the only validated instrument available at study outset that measured Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) criteria for past 12-month diagnoses. Clinically documented diagnosis rates were measured using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision diagnoses from VHA EHR data. Analyses compared survey-based prevalence rates of AUD, DUD, and SUD with diagnosis rates using sensitivity and specificity and difference-in-difference analysis. All analyses were weighted with survey weights to account for nonresponse.

Results

Of 5995 participants, 4115 (68.6%) were White non-Hispanic, and 5429 (91.1%) were male; the mean (SD) age was 61.5 (15.3) years. The survey-based prevalence rates of AUD, DUD, and SUD were higher than the diagnosis rates among all patients (AUD, 608 [10.1%] vs 360 [6.0%]; DUD, 282 [4.7%] vs 275 [4.6%]; SUD, 768 [12.8%] vs 515 [8.6%]). Survey-based prevalence rates of AUD and SUD exceeded the diagnosis rates in every demographic subgroup. Gaps between diagnosis rates and survey-based prevalence rates for AUD and SUD were largest among patients aged 18 to 34 years (AUD diagnosis rate, 27 [6.9%; 95% CI, 4.8%-9.9%] vs AUD prevalence rate, 88 [22.4%; 95% CI, 17.3%-28.5%]; SUD diagnosis rate, 41 [10.5%; 95% CI, 8.1%-13.4%] vs SUD prevalence rate, 109 [27.7%; 95% CI, 22.6%-33.3%]) and Hispanic and Latinx patients (AUD diagnosis rate, 31 [7.6%; 95% CI, 5.3%-10.8%] vs AUD prevalence rate, 72 [17.7%; 95% CI, 14.0%-22.1%]; and SUD diagnosis rate, 48 [11.7%; 95% CI, 7.9%-16.9%] vs SUD prevalence rate, 88 [21.6%; 95% CI, 18.0%-25.8%]). For DUD, only patients aged 18 to 34 years had a true prevalence rate that significantly exceeded the diagnosis rate (diagnosis rate, 21 [5.4%; 95% CI, 3.7%-7.8%] vs prevalence rate, 40 [10.1%; 95% CI, 7.2%-14.0%]).

Conclusions and Relevance

The results of this survey study suggest that existing diagnostic procedures and tools are insufficient to capture SUD prevalence rates, particularly among younger patients and Hispanic and Latinx patients. Clinics and health systems should implement standardized SUD assessments to ensure the provision of equitable care and the optimal identification of underlying conditions for performance monitoring.

Prevalence of substance use and substance-related disorders among US Veterans Health Administration patients

BACKGROUND

Substance use and related disorders are common among US Veterans, but the population burden of has never been directly assessed among Veterans Health Administration (VA) patients. We surveyed VA patients to measure substance use and related disorders in the largest US integrated healthcare system.

METHODS

We surveyed N = 6000 outpatients from 30 geographically-representative VA healthcare systems. We assessed substance use (lifetime, past 12-month, daily in past 3 months) and past 12-month disorders following DSM-5 criteria and estimated the association with Veteran characteristics (age, gender, race/ethnicity, socioeconomic status, VA utilization).

RESULTS

Alcohol was the most commonly-reported substance (24% used past 12 months, 11% daily in past 3 months, 10% met criteria for alcohol use disorder), followed by cannabis (42% lifetime use, 12% use in past 12 months, 5% daily use in past 3 months, 3% met criteria for cannabis use disorder). Overall, 5% met criteria for non-alcohol drug use disorder (13% for substance use disorder (SUD)). SUD prevalence was highest for young Veterans and those who were unemployed or otherwise not employed for wages. Past 12-month cannabis use was common, even among older adults (65-74 years: 10%; 75 and older: 2%).

CONCLUSIONS

Prevalence data are important inputs into decisions around population health monitoring, treatment capacity, and quality measurement strategies. Substance use and SUD are more prevalent than previously reported, and VA may need to screen for non-alcohol drugs to identify patients who need care. More tailored assessment may be needed for cannabis use, high-prevalence subgroups, and older adults.

Opioid Agonist Therapy During Hospitalization Within the Veterans Health Administration: a Pragmatic Retrospective Cohort Analysis

BACKGROUND

Hospitalization of patients with opioid use disorder (OUD) is increasing, yet little is known about opioid agonist therapy (OAT: methadone and buprenorphine) administration during admission.

OBJECTIVE

Describe and examine patient- and hospital-level characteristics associated with OAT receipt during hospitalization in the Veterans Health Administration (VHA).

PARTICIPANTS

A total of 12,407 unique patients, ≥ 18 years old, with an OUD-related ICD-10 diagnosis within 12 months prior to or during index hospitalization in fiscal year 2017 from 109 VHA hospitals in the continental U.S.

MAIN MEASURE

OAT received during hospitalization.

KEY RESULTS

Few admissions received OAT (n = 1914; 15%) and when provided it was most often for withdrawal management (n = 834; 7%). Among patients not on OAT prior to admission who survived hospitalization (n = 10,969), 2.0% (n = 203) were newly initiated on OAT with linkage to care after hospital discharge. Hospitals varied in the frequency of OAT delivery (range, 0 to 43% of qualified admissions). Patients with pre-admission OAT (adjusted odds ratio [AOR] = 15.30; 95% CI [13.2, 17.7]), acute OUD diagnosis (AOR = 2.3; 95% CI [1.99, 2.66]), and male gender (AOR 1.52; 95% CI [1.16, 2.01]) had increased odds of OAT receipt. Patients who received non-OAT opioids (AOR 0.53; 95% CI [0.46, 0.61]) or surgical procedures (AOR 0.75; 95% CI [0.57, 0.99]) had decreased odds of OAT receipt. Large-sized (AOR = 2.0; 95% CI [1.39, 3.00]) and medium-sized (AOR = 1.9; 95% CI [1.33, 2.70]) hospitals were more likely to provide OAT.

CONCLUSIONS

In a sample of VHA inpatient medical admissions, OAT delivery was infrequent, varied across the health system, and was associated with specific patient and hospital characteristics. Policy and educational interventions should promote hospital-based OAT delivery.

Facility-level changes in receipt of pharmacotherapy for opioid use disorder: Implications for implementation science

BACKGROUND

The U.S. is facing an opioid epidemic, but despite mandates for pharmacotherapy for opioid use disorder to be available at Veterans Health Administration (VHA) facilities, the majority of veterans with opioid use disorder do not receive these medications. In implementation research, facilities are often targeted for qualitative inquiry or quality improvement efforts based on quality measure performance during a one-year period. However, sites that experience quality performance changes from one year to the next may be highly informative because mechanisms that impact facility change may be more discoverable. The current study examined changes in receipt of pharmacotherapy for opioid use disorder in a national healthcare system to determine the extent to which sites fluctuated in performance over a two-year period and illustrate how changes in quality measures over time may be useful for implementation research and healthcare surveillance of quality measures.

METHODS

Using national VHA data from Fiscal Years (FY) 2016 and 2017, we calculated quality measure performance as the number of patients who received pharmacotherapy for opioid use disorder (i.e., methadone, buprenorphine, and naltrexone) divided by the number of patients with a current non-remitted opioid use disorder diagnosis for each FY at each facility (n = 129) and examined change from FY16 to FY17.

RESULTS

The mean rate of receipt of pharmacotherapy for opioid use disorder was 38% (facility range = 3% to 74%) in FY16 and 41% (facility range = 2% to 76%) in FY17. The average facility-level change in performance was 3% and ranged from -19% to 26%. There were 32 facilities that decreased in provision of pharmacotherapy, 12 facilities with no change, and 85 facilities that increased.

CONCLUSIONS

For facilities with average or high performance, it was difficult to maintain their performance over time. Identifying and learning from facilities with recent fluctuations may be more informative to guide the design of future quality improvement efforts than studying facilities with stable high or low performance.

Alcohol-related and mental health care for patients with unhealthy alcohol use and posttraumatic stress disorder in a National Veterans Affairs cohort

OBJECTIVE

Unhealthy alcohol use and posttraumatic stress disorder (PTSD) frequently co-occur. Patients with both conditions have poorer functioning and worse treatment adherence compared to those with either condition alone. Therefore, it is possible that PTSD, when co-occurring with unhealthy alcohol use, may influence receipt of evidence-based alcohol-related care and mental health care. We evaluated receipt of interventions for unhealthy alcohol use and receipt of mental health follow-up care among patients screening positive for unhealthy alcohol use with and without PTSD in a national sample from the Veterans Health Administration (VA).

METHODS

National clinical and administrative data from VA's electronic medical record were used to identify all patients who screened positive for unhealthy alcohol use (AUDIT-C score≥5) between 10/1/09-5/30/13. Unadjusted and adjusted Poisson regression models were fit to estimate the relative rate and prevalence of receipt of: brief interventions (advice to reduce or abstain from drinking≤14days after positive screening), specialty addictions treatment for alcohol use disorder (AUD; documented visit≤365days after positive screening), pharmacotherapy for AUD (filled prescription≤365days after positive screening), and mental health care ≤14days after positive screening for patients with and without PTSD (documented with ICD-9 CM codes). In secondary analyses, we tested effect modification by both severity of unhealthy alcohol use and age.

RESULTS

Among 830,825 patients who screened positive for unhealthy alcohol use, 140,388 (16.9%) had documented PTSD. Of the full sample, 71.6% received brief interventions, 10.3% received specialty AUD treatment, 3.1% received pharmacotherapy for AUD, and 24.0% received mental health care. PTSD was associated with increased likelihood of receiving all types of care. Adjusted relative rates were 1.04 (95% CI 1.03-1.05) for brief interventions, 1.06 (1.05-1.08) for specialty AUD treatment, 1.35 (1.31-1.39) for AUD pharmacotherapy, and 1.82 (1.80-1.84) for mental health care. Alcohol use severity modified effects of PTSD for specialty AUD treatment, AUD pharmacotherapy, and mental health care such that effects were maintained at lower severity but attenuated among patients with severe unhealthy alcohol use. Age modified all effects with the strength of the association between PTSD and care outcomes being strongest for younger (18-29years) and older veterans (65+ years) and weaker or non-significant for middle-aged veterans (30-44 and 45-64years).

CONCLUSIONS

In this large national sample of patients with unhealthy alcohol use, PTSD was associated with increased likelihood of receiving alcohol-related and mental health care. PTSD does not appear to be a barrier to care among VA patients with unhealthy alcohol use.

Trends in Opioid Use Disorder Diagnoses and Medication Treatment Among Veterans With Posttraumatic Stress Disorder

OBJECTIVE

Despite long-standing interest in posttraumatic stress disorder (PTSD) and opioid use disorder comorbidity, there is a paucity of data on the prevalence of opioid use disorder in patients with PTSD. Therefore, there is limited understanding of the use of medications for opioid use disorder in this population. We determined the prevalence of diagnosed opioid use disorder and use of medications for opioid use disorder in a large cohort of patients with PTSD.

METHODS

We obtained administrative and pharmacy data for veterans who initiated PTSD treatment in the Department of Veterans Affairs (VA) between 2004 and 2013 (N = 731,520). We identified those with a comorbid opioid use disorder diagnosis (2.7%; n = 19,998) and determined whether they received a medication for opioid use disorder in the year following their initial clinical PTSD diagnosis (29.6%; n = 5,913). Using logistic regression, we determined the predictors of receipt of opioid use disorder medications.

RESULTS

Comorbid opioid use disorder diagnoses increased from 2.5% in 2004 to 3.4% in 2013. Patients with comorbid opioid use disorder used more health services and had more comorbidities than other patients with PTSD. Among patients with PTSD and comorbid opioid use disorder, use of medications for opioid use disorder increased from 22.6% to 35.1% during the same time period. Growth in the use of buprenorphine (2.0% to 22.7%) was accompanied by relative decline in use of methadone (19.3% to 12.7%). Patients who received buprenorphine were younger and more likely to be rural, White, and married. Patients who received methadone were older, urban, unmarried, from racial and ethnic minorities, and more likely to see substance abuse specialists. While use of naltrexone increased (2.8% to 8.6%), most (87%) patients who received naltrexone also had an alcohol use disorder. Controlling for patient factors, there was a substantial increase in the use of buprenorphine, a substantial decrease in the use of methadone, and no change in use of naltrexone across years.

CONCLUSIONS

Opioid use disorder is an uncommon but increasing comorbidity among patients with PTSD. Patients entering VA treatment for PTSD have their opioid use disorder treated with opioid agonist treatments in large and increasing numbers. There is a need for research both on the epidemiology of opioid use disorder among patients with PTSD and on screening for opioid use disorder.

Quality of care measures for the management of unhealthy alcohol use

There is a paucity of quality measures to assess the care for the range of unhealthy alcohol use, ranging from risky drinking to alcohol use disorders. Using a two-phase expert panel review process, we sought to develop an expanded set of quality of care measures for unhealthy alcohol use, focusing on outpatient care delivered in both primary care and specialty care settings. This process generated 25 candidate measures. Eight measures address screening and assessment, 11 address aspects of treatment, and six address follow-up. These quality measures represent high priority targets for future development, including creating detailed technical specifications and pilot testing them to evaluate their utility in terms of feasibility, reliability, and validity.

Specifying and Pilot Testing Quality Measures for the American Society of Addiction Medicine's Standards of Care

OBJECTIVES

In 2013, the American Society of Addiction Medicine (ASAM) approved its Standards of Care for the Addiction Specialist Physician. Subsequently, an ASAM Performance Measures Panel identified and prioritized the standards to be operationalized into performance measures. The goal of this study is to describe the process of operationalizing 3 of these standards into quality measures, and to present the initial measure specifications and results of pilot testing these measures in a large health care system. By presenting the process rather than just the end results, we hope to shed light on the measure development process to educate, and also to stimulate debate about the decisions that were made.

METHODS

Each measure was decomposed into major concepts. Then each concept was operationalized using commonly available administrative data sources. Alternative specifications examined and sensitivity analyses were conducted to inform decisions that balanced accuracy, clinical nuance, and simplicity. Using data from the US Veterans Health Administration (VHA), overall performance and variation in performance across 119 VHA facilities were calculated.

RESULTS

Three measures were operationalized and pilot tested: pharmacotherapy for alcohol use disorder, pharmacotherapy for opioid use disorder, and timely follow-up after medically managed withdrawal (aka detoxification). Each measure was calculable with available data, and showed ample room for improvement (no ceiling effects) and wide facility-level variability.

CONCLUSIONS

Next steps include conducting feasibility and pilot testing in other health care systems and other contexts such as standalone addiction treatment programs, and also to study the specification and predictive validity of these measures.

Are Improvements in Measured Performance Driven by Better Treatment or "Denominator Management"?

BACKGROUND

Process measures of healthcare quality are usually formulated as the number of patients who receive evidence-based treatment (numerator) divided by the number of patients in the target population (denominator). When the systems being evaluated can influence which patients are included in the denominator, it is reasonable to wonder if improvements in measured quality are driven by expanding numerators or contracting denominators.

OBJECTIVE

In 2003, the US Department of Veteran Affairs (VA) based executive compensation in part on performance on a substance use disorder (SUD) continuity-of-care quality measure. The first goal of this study was to evaluate if implementing the measure in this way resulted in expected improvements in measured performance. The second goal was to examine if the proportion of patients with SUD who qualified for the denominator contracted after the quality measure was implemented, and to describe the facility-level variation in and correlates of denominator contraction or expansion.

DESIGN

Using 40 quarters of data straddling the implementation of the performance measure, an interrupted time series design was used to evaluate changes in two outcomes.

PARTICIPANTS

All veterans with an SUD diagnosis in all VA facilities from fiscal year 2000 to 2009.

MAIN MEASURES

The two outcomes were 1) measured performance-patients retained/patients qualified and 2) denominator prevalence-patients qualified/patients with SUD program contact.

KEY RESULTS

Measured performance improved over time (P < 0.001). Notably, the proportion of patients with SUD program contact who qualified for the denominator decreased more rapidly after the measure was implemented (p = 0.02). Facilities with higher pre-implementation denominator prevalence had steeper declines in denominator prevalence after implementation (p < 0.001).

CONCLUSIONS

These results should motivate the development of measures that are less vulnerable to denominator management, and also the exploration of "shadow measures" to monitor and reduce undesirable denominator management.