Total cost of care associated with opioid use disorder treatment.

The opioid epidemic in the United States disproportionately affects Medicaid beneficiaries than other groups. This results in a significant financial burden on state Medicaid programs. In this analysis, we investigate the association of medication for opioid use disorder (MOUD) treatment initiation and linkage to ongoing care on overall healthcare costs of Medicaid Fee-for-Service patients. We conducted a retrospective study among adult patients diagnosed with opioid use disorder (OUD) and who had a clinical encounter at a safety-net institution in Denver Colorado in 2020. Three categories of MOUD status of patients were defined: 1) identified with OUD but did not receive MOUD; 2) initiated MOUD but not linked to ongoing treatment and 3) received MOUD and linked to ongoing treatment. Our outcome variable was per-member per-month total healthcare cost. We estimated a multivariable model to test the association between healthcare cost and MOUD status, while controlling for demographic and risk classification variables. We found that in individuals with OUD who initiated MOUD treatment but were not linked to ongoing care had the highest healthcare cost, while those who were linked to ongoing MOUD treatment had the lowest healthcare cost. MOUD treatment is not only effective at addressing the significant morbidity and mortality burden of OUD but also associated with decreased financial cost, which is disproportionately incurred by Medicaid. Additional policy and care delivery changes are needed to focus efforts to improve linkage to ongoing treatment.

Predictors of 007 triphosphate concentrations in dried blood spots in persons with hepatitis C and active drug or alcohol use.

BACKGROUND: Sofosbuvir is converted to its active form, 007 triphosphate (007-TP), within cells. To date, the association between treatment adherence and 007-TP in dried blood spots (DBS) and factors that influence this relationship remain unknown. OBJECTIVES: To examine relationships between adherence and 007-TP concentrations in DBS and identify factors that influence 007-TP in DBS. METHODS: Persons with HCV or HIV/HCV coinfection and self-reported drug and/or alcohol use were randomized to one of two technology-based approaches for monitoring 12 week adherence to once-daily ledipasvir/sofosbuvir. Convenience blood samples were collected every 2 weeks during treatment. 007-TP in DBS was quantified using LC/MS and analysed using mixed-effects models. RESULTS: A total of 337 observations were available from 58 participants (78% male; 21% black; 22% Hispanic/Latino; 26% cirrhotic; 78% HIV-coinfected). The mean half-life of 007-TP in DBS was 142 h (95% CI 127-156) and concentrations increased by 7.3% (95% CI 2.2-12.6) for every 10% increase in between-visit adherence. Geometric mean (95% CI) 007-TP concentrations in DBS were 301 (247-368), 544 (462-639) and 647 (571-723) fmol/punch by adherence categories of ≤50%, >50 to ≤80%, and >80%. Adherence, time on therapy, increasing age and decreased estimated glomerular filtration rate were associated with higher 007-TP, whereas increased time since last dose, male sex, black race and higher BMI were associated with lower 007-TP. CONCLUSIONS: 007-TP has an extended half-life in DBS and concentrations increased with adherence. Further research is needed to examine additional factors that affect 007-TP and the clinical utility of this measure.

Pharmacokinetics and renal safety of tenofovir alafenamide with boosted protease inhibitors and ledipasvir/sofosbuvir.

BACKGROUND: Ledipasvir/sofosbuvir increases tenofovir plasma exposures by up to 98% with tenofovir disoproxil fumarate (TDF), and exposures are highest with boosted PIs. There are currently no data on the combined use of the newer tenofovir prodrug, tenofovir alafenamide (TAF), boosted PIs and ledipasvir/sofosbuvir. OBJECTIVES: To compare the plasma and intracellular pharmacokinetics and renal safety of TAF with ledipasvir/sofosbuvir when co-administered with boosted PIs. METHODS: Persons with HIV between 18 and 70 years and on a boosted PI with TDF were eligible. The study was comprised of four phases: (1) TDF 300 mg with boosted PI; (2) TAF 25 mg with boosted PI; (3) TAF 25 mg with boosted PI and ledipasvir/sofosbuvir; and (4) TAF 25 mg with boosted PI. Pharmacokinetic sampling, urine biomarker collection [urine protein (UPCR), retinol binding protein (RBP) and ß2 microglobulin (ß2M) normalized to creatinine] and safety assessments occurred at the end of each phase. Plasma, PBMCs and dried blood spots were collected at each visit. RESULTS: Ten participants were enrolled. Plasma tenofovir exposures were 76% lower and tenofovir-diphosphate (TFV-DP) concentrations in PBMCs increased 9.9-fold following the switch to TAF. Neither of these measures significantly increased with ledipasvir/sofosbuvir co-administration, nor did TAF plasma concentrations. No significant changes in estimated glomerular filtration rate or UPCR occurred, but RBP:creatinine and ß2M:creatinine improved following the switch to TAF. CONCLUSIONS: Ledipasvir/sofosbuvir did not significantly increase plasma tenofovir or intracellular TFV-DP in PBMCs with TAF. These findings provide reassurance that the combination of TAF, boosted PIs and ledipasvir/sofosbuvir is safe in HIV/HCV-coinfected populations.

Increased tenofovir monoester concentrations in patients receiving tenofovir disoproxil fumarate with ledipasvir/sofosbuvir.

BACKGROUND: Intracellular tenofovir diphosphate concentrations are markedly increased in HIV/HCV coinfected individuals receiving tenofovir disoproxil fumarate (TDF) with sofosbuvir-containing treatment. Sofosbuvir may inhibit the hydrolysis of TDF to tenofovir, resulting in increased concentrations of the disoproxil or monoester forms, which may augment cell loading. We sought to quantify tenofovir disoproxil and monoester concentrations in individuals receiving TDF with and without ledipasvir/sofosbuvir. METHODS: HIV/HCV coinfected participants receiving TDF-based therapy were sampled pre-dose and 1 and 4 h post-dose prior to and 4 weeks after initiating ledipasvir/sofosbuvir. Tenofovir disoproxil was not detectable. Tenofovir monoester in plasma and tenofovir diphosphate in PBMC and dried blood spots (DBS) were quantified using LC-MS/MS. Geometric mean ratios (week 4 versus baseline) and 95% CIs were generated for the pharmacokinetic parameters. P values reflect paired t-tests. RESULTS: Ten participants had complete data. At baseline, geometric mean (95% CI) tenofovir monoester plasma concentrations at 1 and 4 h post-dose were 97.4 ng/mL (33.0-287.5) and 0.74 ng/mL (0.27-2.06), respectively. With ledipasvir/sofosbuvir, tenofovir monoester concentrations at 4 h post-dose were 5.02-fold higher (95% CI 1.40-18.05; P = 0.019), but did not significantly differ at 1 h post-dose (1.72-fold higher, 95% CI 0.25-11.78; P = 0.54), possibly due to absorption variability. Tenofovir diphosphate in PBMC and DBS were increased 2.80-fold (95% CI 1.71-4.57; P = 0.001) and 7.31-fold (95% CI 4.47-11.95; P < 0.0001), respectively, after 4 weeks of ledipasvir/sofosbuvir. CONCLUSIONS: Tenofovir monoester concentrations were increased in individuals receiving TDF with ledipasvir/sofosbuvir, consistent with inhibition of TDF hydrolysis. Additional studies are needed to determine the clinical relevance of this interaction.

The Association Between Stimulant, Opioid, and Multiple Drug Use on Behavioral Health Care Utilization in a Safety-Net Health System.

BACKGROUND: Prior studies show an association between drug use and health care utilization. The relationship between specific drug type and emergent/urgent, inpatient, outpatient, and behavioral health care utilization has not been examined. We aimed to determine if multiple drug use was associated with increased utilization of behavioral health care. METHODS: To assess health care utilization, we conducted a retrospective cohort study of patients who accessed health care at a safety-net medical center and affiliated clinics. Using electronic health records, we categorized patients who used stimulants, opioids, or multiple drugs based on urine toxicology screening tests and/or International Classification of Diseases, 9th Revision (ICD-9). Remaining patients were categorized as patients without identified drug use. Health care utilization by drug use group and visit type was determined using a negative binomial regression model. Associations were reported as incidence rate ratios. Utilization was described by rates of health care-related visits for inpatient, emergent/urgent, outpatient, and behavioral health care among patients who used drugs, categorized by drug types, compared with patients without identified drug use. RESULTS: Of 95,198 index visits, 4.6% (n=4340) were by patients who used drugs. Opioid and multiple drug users had significantly higher rates of behavioral health care visits than patients without identified drug use (opioid incidence rate ratio [IRR]=7.2; 95% confidence interval [CI]: 3.8-13.8; multiple drug use IRR=5.6, 95% CI: 3.3-9.7). Patients who used stimulants were less likely to use behavioral health services (IRR=1.3, 95% CI: 0.9-2.0) when compared with opioid and multiple drug users, but were more likely to use inpatient (IRR=1.6, 95% CI: 1.4-1.8) and emergent/urgent care (IRR=1.4, 95% CI: 1.3-1.5) services as compared with patients without identified drug use. CONCLUSIONS: Integrated medical and mental health care and drug treatment may reduce utilization of costly health care services and improve patient outcomes. How to capture and deliver primary care and behavioral health care to patients who use stimulants needs further investigation.

Tenofovir, emtricitabine, and darunavir/ritonavir pharmacokinetics in an HIV-infected patient after Roux-en-Y gastric bypass surgery.

OBJECTIVE: To determine antiretroviral (ARV) pharmacokinetics in a patient who previously underwent Roux-en-Y gastric bypass (RYGB) surgery. CASE SUMMARY: We describe a 38-year-old Hispanic man who tested positive for human immunodeficiency virus (HIV) 11 months following RYGB surgery. When the patient presented for care of his HIV, his HIV-1 RNA was 146 138 copies/mL (5.20 log) and his CD4 T cell count was 320 cells/mm(3) (25%). He was initiated on tenofovir disoproxil fumarate (TDF) 300 mg once daily, emtricitabine (FTC) 200 mg once daily, and darunavir/ritonavir (DRV/r) 600/100 mg twice daily. ARV concentrations were similar to historical data. Six months following ARV initiation, HIV-1 RNA was <48 copies/mL and CD4 count had increased to 562 cells/mm(3) (39%). DISCUSSION: Bariatric surgery has been successfully performed in obese persons infected with the HIV, but data are limited on ARV drug selection and pharmacokinetics in this group. Optimal suppression of HIV replication requires appropriate concentrations of ARV drugs, and in a patient who has undergone RYGB, this can be challenging not only because of a decreased absorptive surface area but also because of an increased intragastric pH. CONCLUSION: We found that once daily TDF/FTC and twice daily DRV/r produced trough concentrations similar to historic data in a patient who previously underwent RYGB with virologic suppression and immunologic recovery.

Using artificial intelligence to predict choledocholithiasis: can machine learning models abate the use of MRCP in patients with biliary dysfunction?

BACKGROUND: Prompt diagnosis of choledocholithiasis is crucial for reducing disease severity, preventing complications and minimizing length of stay. Magnetic resonance cholangiopancreatography (MRCP) is commonly used to evaluate patients with suspected choledocholithiasis but is expensive and may delay definitive intervention. To optimize patient care and resource utilization, we have developed five machine learning models that predict a patients' risk of choledocholithiasis based on clinical presentation and pre-MRCP investigation results. METHODS: Inpatients admitted to the Royal Hobart Hospital from 2018 to 2023 with a suspicion of choledocholithiasis were included. Exclusion criteria included prior hepatobiliary surgery, known hepatobiliary disease, or incomplete records. Variables related to clinical presentation, laboratory testing, and sonographic or CT imaging were collected. Four machine learning techniques were employed: logistic regression, XGBoost, random forest, and K-nearest neighbours. The three best performing models were combined to create an ensemble model. Model performance was compared against the American Society for Gastrointestinal Endoscopy (ASGE) choledocholithiasis risk stratification guidelines. RESULTS: Of the 222 patients included, 113 (50.9%) had choledocholithiasis. The most successful models were the random forest (accuracy: 0.79, AUROC: 0.83) and ensemble (accuracy and AUROC: 0.81). Every model outperformed the ASGE guidelines. Key variables influencing the models' predictions included common bile duct diameter, lipase, imaging evidence of cholelithiasis, and liver function tests. CONCLUSION: Machine learning models can accurately assess a patient's risk of choledocholithiasis and could assist in identifying patients who could forgo an MRCP and proceed directly to intervention. Ongoing validation on prospective data is necessary to refine their accuracy and clinical utility.

Using 42 CFR part 2 revisions to integrate substance use disorder treatment information into electronic health records at a safety net health system.

BACKGROUND: Regulations put in place to protect the privacy of individuals receiving substance use disorder (SUD) treatment have resulted in an unintended consequence of siloed SUD treatment and referral information outside of the integrated electronic health record (EHR). Recent revisions to these regulations have opened the door to data integration, which creates opportunities for enhanced patient care and more efficient workflows. We report on the experience of one safety-net hospital system integrating SUD treatment data into the EHR. METHODS: SUD treatment and referral information was integrated from siloed systems into the EHR through the implementation of a referral order, treatment episode definition, and referral and episode-related tools for addiction therapists and other clinicians. Integration was evaluated by monitoring SUD treatment episode characteristics, patient characteristics, referral linkage, and treatment episode retention before and after integration. Satisfaction of end-users with the new tools was evaluated through a survey of addiction therapists. RESULTS: After integration, three more SUD treatment programs were represented in the EHR. This increased the number of patients that could be tracked as initiating SUD treatment by 250%, from 562 before to 1,411 after integration. After integration, overall referral linkage declined (74% vs. 48%) and treatment episode retention at 90-days was higher (45% vs. 74%). Addiction therapists appreciated the efficiency of having all SUD treatment information in the EHR but did not find that the tools provided a large time savings shortly after integration. CONCLUSIONS: Integration of SUD treatment program data into the EHR facilitated both care coordination in patient treatment and quality improvement initiatives for treatment programs. Referral linkage and retention rates were likely modified by a broader capture of patients and changed outcome definition criteria. Greater preparatory workflow analysis may decrease initial end-user burden. Integration of siloed data, made possible given revised regulations, is essential to an efficient hub-and-spoke model of care, which must standardize and coordinate patient care across multiple clinics and departments.

Environmental enrichment protects against the acquisition of cocaine self-administration in adult male rats, but does not eliminate avoidance of a drug-associated saccharin cue.

One of the most menacing consequences of drug addiction is the devaluation of natural rewards (e.g. food, sex, work, money, caring for one's offspring). However, evidence also suggests that natural rewards, such as an enriched environment, can devalue drugs of abuse. Thus, this study used a rodent model to test whether exposure to an enriched environment could protect adult rats from acquiring cocaine self-administration and from the resultant drug-induced devaluation of a natural saccharin reward cue. Adult male Sprague-Dawley rats were implanted with intravenous jugular catheters. Rats were then separated into two housing conditions: an enriched condition, including social companions(four/cage) and novel objects (e.g. balls, polyethylene tubes, paper, etc.), and a nonenriched condition where the rats were singly housed with no novel objects. During testing, the rats were given 5-min access to 0.15% saccharin, followed by 1 h to self-administer saline or cocaine (0.167 mg/infusion) on fixed ratio and progressive ratio schedules of reinforcement. The results showed that rats that were singly housed in the nonenriched environment fell into two groups: low drug-takers (n=34) and high drug-takers (n=12). In comparison, only one out of the 22 rats housed in the enriched environment was a high drug-taker. Thus, all rats in the enriched environment, except one, behaved like low drug-takers under the nonenriched condition. As such, these rats self-administered almost no drug on either the fixed ratio or the progressive ratio schedule of reinforcement and were extremely slow to self-administer their first cocaine infusion. Interestingly, despite their very low levels of drug self-administration, low-drug-taking rats housed in the enriched environment continued to avoid intake of the drug-associated saccharin cue. Taken together, these data suggest that the enriched environment itself served as a salient natural reward that reduced cocaine seeking and cocaine taking, but had little impact on avoidance of the cocaine-paired taste cue. The protective effects of the enriched environment were robust and, as such, have important implications for the methods used in the study of drug addiction in animal models and for the prevention, and possibly the treatment, of the disease in adult humans.

Demographics and treatment of the American family.

There have been numerous changes to the US family over the past several decades. Traditional family roles have changed, and the conception of what Americans consider a 'family' has likewise shifted with differing societal views regarding gender, gender roles, race, and ethnicity. This review examines demographics of the American family as well as a number of family therapies that have been historically and are presently used to treat family problems. We expect that with the changes present in US society, family therapies will need to continue to be sensitive and adaptive to these shifts in order to be effective.

The Impact of a PDMP-EHR Data Integration Combined With Clinical Decision Support on Opioid and Benzodiazepine Prescribing Across Clinicians in a Metropolitan Area.

INTRODUCTION: Despite inconclusive evidence that prescription drug monitoring programs (PDMP) reduce opioid-related mortality, guidelines recommend PDMP review with opioid prescribing. Some reported barriers to use include time-consuming processes to obtain data and workflow disruptions. METHODS: We provided access to a PMDP-electronic health record (EHR) integrated program to 123 clinicians in one healthcare system. Remaining clinicians within the healthcare system and metropolitan area did not receive PDMP-EHR integration program access. We identified changes in opioid prescribing by linking prescription data available in the state PMDP database to individual clinicians. The primary outcome was change in receipt of high dose opioid prescriptions (>90 mg morphine equivalents) by Colorado residents before and after program integration. Secondary outcomes included changes in long-acting opioid receipt and overlapping opioid and benzodiazepine prescription days. Next, we surveyed clinicians to assess their perspectives on PDMP data acquisition before and after PDMP-EHR integration program access. RESULTS: High-dose opioid receipt decreased significantly across all 3 clinician groups [PDMP-EHR integration program access (27.6%, to 6.9%, P < 0.001); no program access in the same healthcare system (4.8% to 2.9%, P < 0.001), and no program access across the metropolitan area (13.5% to 6.1%, P < 0.001)]. Clinicians reported improved access to PDMP data using the PDMP-EHR integrated program compared to the state PDMP website (98.6%). CONCLUSIONS: Further study of PDMP-EHR integration programs on patient and clinician outcomes may illuminate the role of this technology in public health and in clinical practice.

Barriers to Engagement in Opioid Use Disorder Treatment After Buprenorphine Induction.

OBJECTIVES: Expanded access to buprenorphine induction, including via emergency departments, increases the likelihood of treatment engagement for patients with opioid use disorder (OUD). However, longer-term retention among these patients remains a challenge. In this study, we aimed to identify barriers to engaging and retaining patients with OUD in care and additional services that might improve retention. METHODS: We surveyed counselors at an urban safety net addictions treatment clinic. RESULTS: Twenty-five of 27 (93%) eligible counselors responded. Counselors described patients who were homeless, had no prior treatment history, or lacked health insurance as hardest to retain in treatment. Housing assistance, residential treatment placement, regular access to a phone, and mental health services were thought to be most beneficial for improving retention. Respondents most often reported that screening for services should happen at intake, and almost all respondents agreed that "retention of patients receiving treatment for OUD would improve with a dedicated case manager and/or more coordinated case management services." CONCLUSIONS: Engagement in OUD treatment would be improved with interventions to mitigate the significant social and psychiatric comorbidities of addiction. Community- and emergency department-initiated buprenorphine is a promising intervention whose full promise cannot be realized without interventions to improve treatment retention.

Qualitative exploration of public health vending machines in young adults who misuse opioids: A promising strategy to increase naloxone access in a high risk underserved population.

Background: Take home naloxone (THN) programs have been shown to effectively reverse opioid overdose events with limited adverse events, yet often miss young adults who use opioids. To identify opportunities for naloxone expansion, we conducted interviews with young adults who had used opioids. We explored young adults' experience with current THN programs, and perspectives on ideal THN programs and emerging naloxone public health vending machine (PHVM) programs shown to increase access to sterile syringes in young adults. Methods: We interviewed 16 young adults receiving substance treatment services within an integrated safety net healthcare system. Participants were 18-30 years of age with a history of nonmedical prescription opioid use. Interviews obtained the patient perspective of current THN, ideal THN and PHVM programs. Interviews were transcribed and coded by team-based methods. Themes were developed using an inductive-deductive iterative approach and defined through consensus. Results: Treatment was often the first exposure to naloxone. Participants recommended easy to access programs for ideal naloxone distribution and had overall positive feedback on PHVMs. Three key themes were identified to improve naloxone uptake: knowledge, convenience, and privacy. Participants identified safety, lack of police presence, and low costs as important vending machine features. Conclusions: Our results identified implementation opportunities to increase naloxone uptake including convenient location and hours, privacy, and using trusted sources of information to improve program awareness. PHVMs present an opportunity to maximize these opportunities and increase access to naloxone in young adults.

Adherence to Direct-Acting Antiviral Therapy in People Actively Using Drugs and Alcohol: The INCLUD Study.

BACKGROUND: Hepatitis C virus treatment in persons who use drugs (PWUD) is often withheld due to adherence and reinfection concerns. In this study, we report treatment outcomes, technology-based adherence data, and adherence predictors in PWUD and/or alcohol. METHODS: INCLUD was a prospective, open-label study of ledipasvir/sofosbuvir for 12 weeks in PWUD aged 18-70 years. Participants were randomized to wireless (wirelessly observed therapy) or video-based directly observed therapy (vDOT). Drug use was assessed every 2 weeks. Sustained virologic response (SVR) was examined by intention-to-treat and as-treated. Factors associated with missing ≥1 dose(s) between visits were examined using generalized linear models. RESULTS: Sixty participants received ≥1 ledipasvir/sofosbuvir dose (47 human immunodeficiency virus [HIV]/hepatitis C virus [HCV], 13 HCV only; 78% male; 22% black; 25% cirrhotic). Substance use occurred at 94% of person-visits: 60% marijuana, 56% alcohol, 37% methamphetamine, 22% opioids, 17% cocaine, and 20% injection drug use. The SVR by intention-to-treat was 86.7% (52 of 60) and as-treated was 94.5% (52 of 55). Confirmed failures included 1 relapse, 1 reinfection, and 1 unknown (suspected reinfection). Median total adherence was 96% (interquartile range [IQR], 85%-100%; range, 30%-101%), and between-visit adherence was 100% (IQR, 86%-100%; range, 0%-107%). The odds of missing ≥1 dose between visits increased with HIV coinfection (2.94; 95% confidence interval [CI], 1.37-6.32; P = .006), black race (4.09; 95% CI, 1.42-11.74; P = .009), methamphetamine use (2.51; 95% CI, 1.44-4.37; P = .0.001), and cocaine use (2.12; 95% CI, 1.08-4.18; P = .03) and decreased with marijuana use (0.34; 95% CI, 0.17-0.70; P = .003) and vDOT (0.43; 95% CI, 0.21-0.87; P = .02). CONCLUSIONS: Persons who use drugs achieved high SVR rates with high, but variable, ledipasvir/sofosbuvir adherence using technology-based methods. These findings support efforts to expand HCV treatment in PWUD.

Monitoring opioid addiction and treatment: Do you know if your population is engaged?

BACKGROUND: Assessment of people affected by opioid-related problems and those receiving care is challenging due to lack of common definitions and scattered information. We sought to fill this gap by demonstrating a method to describe a continuum of opioid addiction care in a large, public safety-net institution. METHODS: Using 2017 clinical and administrative data from Denver Health (DH), we created operational definitions for opioid use disorder (OUD), opioid misuse (OM), and opioid poisoning (OP). Six stages along a continuum of patient engagement in opioid addiction care were developed, and operational definitions assigned patients to stages for a specific time point of analysis. National data was used to estimate the Denver population affected by OUD, OM and OP. RESULTS: In 2017, an estimated 6688 people aged ≥12 years were affected by OUD, OM, or OP in Denver; 48.4% (3238/6688) were medically diagnosed in DH. Of those, 32.5% (1051/3238) were in the medication assisted treatment stage, and, of those, 59.8% (629/1051) in the adhered to treatment stage. Among that latter group, 78.4% (493/629) adhered at least 90 days and 52.3% (329/629) for more than one year. Among patients who received medication assisted treatment, less than one third (31.3%, 329/1051) were adherent for more than one year. CONCLUSIONS: A health-system level view of the continuum of opioid addiction care identified improvement opportunities to better monitor accuracy of diagnosis, treatment capacity, and effectiveness of patient engagement. Applied longitudinally at local, state and national levels, the model could better synergize responses to the opioid crisis.

For many patients who use large amounts of health care services, the need is intense yet temporary.

Patients who accumulate multiple emergency department visits and hospital admissions, known as super-utilizers, have become the focus of policy initiatives aimed at preventing such costly use of the health care system through less expensive community- and primary care-based interventions. We conducted cross-sectional and longitudinal analyses of 4,774 publicly insured or uninsured super-utilizers in an urban safety-net integrated delivery system for the period May 1, 2011-April 30, 2013. Our analysis found that consistently 3 percent of adult patients met super-utilizer criteria and accounted for 30 percent of adult charges. Fewer than half of super-utilizers identified as such on May 1, 2011, remained in the category seven months later, and only 28 percent remained at the end of a year. This finding has important implications for program design and for policy makers because previous studies may have obscured this instability at the individual level. Our study also identified clinically relevant subgroups amenable to different interventions, along with their per capita utilization and costs before and after being identified as super-utilizers. Future solutions include improving predictive modeling to identify individuals likely to experience sustained levels of avoidable utilization, better classifying subgroups for whom interventions are needed, and implementing stronger program evaluation designs.