Using prescription claims to detect aberrant behaviors with opioids: comparison and validation of 5 algorithms

Rough, Kathryn; Huybrechts, Krista F.; Hernández–Dı́az, Sonia; Desai, Rishi; Patorno, Elisabetta; Bateman, Brian T.

doi:10.1002/pds.4443

Cited by 15 publications

(10 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thirdly, to reduce the potential for exposure misclassification, we excluded women with opioid filling patterns that might suggest misuse or diversion. 33 34 …”

Section: Methodsmentioning

confidence: 99%

Association of first trimester prescription opioid use with congenital malformations in the offspring: population based cohort study

Bateman

Hernández–Dı́az

Straub

et al. 2021

BMJ

Self Cite

View full text Add to dashboard Cite

Objective To evaluate the risk of first trimester exposure to prescription opioids for major congenital malformations, previously reported to be associated with such exposure. Design Population based cohort study. Setting Nationwide sample of publicly and commercially insured pregnant women linked to their liveborn infants, nested in the Medicaid Analytic eXtract (MAX, 2000-14) and the MarketScan Research Database (MarketScan, 2003-15). Participants 1 602 580 publicly insured (MAX) and 1 177 676 commercially insured (MarketScan) pregnant women with eligibility from at least three months before pregnancy to one month after delivery; infants with eligibility for at least three months after birth. Interventions Use of prescription opioids was ascertained by requiring two or more dispensations of any opioid during the first trimester. Main outcomes measures Major malformations overall, cardiac malformations overall, ventricular septal defect, secundum atrial septal defect/patent foramen ovale, neural tube defect, clubfoot, and oral cleft, defined based on validated algorithms. Propensity score stratification was used to adjust for potential confounders and/or proxies for confounders. Estimates from each database were combined using meta-analysis. Results 70 447 (4.4%) of 1 602 580 publicly insured and 12 454 (1.1%) of 1 177 676 commercially insured pregnant women had two or more dispensations of an opioid during the first trimester. Absolute risk of malformations overall was 41.0 (95% confidence interval 39.5 to 42.5) per 1000 pregnancies exposed to opioids versus 32.0 (31.7 to 32.3) per 1000 unexposed pregnancies in the MAX cohort, and 42.6 (39.0 to 46.1) and 37.3 (37.0 to 37.7) per 1000, respectively, in the MarketScan cohort. Pooled unadjusted relative risk estimates were raised for all outcomes but shifted substantially toward the null after adjustment; for malformations overall (relative risk 1.06, 95% confidence interval 1.02 to 1.10), cardiovascular malformations (1.09, 1.00 to 1.18), ventricular septal defect (1.07, 0.95 to 1.21), atrial septal defect/patent foramen ovale (1.04, 0.88 to 1.24), neural tube defect (0.82, 0.53 to 1.27), and clubfoot (1.06, 0.88 to 1.28). The relative risk for oral clefts remained raised after adjustment (1.21, 0.98 to 1.50), with a higher risk of cleft palate (1.62, 1.23 to 2.14). Conclusions Prescription opioids used in early pregnancy are not associated with a substantial increase in risk for most of the malformation types considered, although a small increase in the risk of oral clefts associated with their use is possible.

show abstract

“…Thirdly, to reduce the potential for exposure misclassification, we excluded women with opioid filling patterns that might suggest misuse or diversion. 33 34 …”

Section: Methodsmentioning

confidence: 99%

Association of first trimester prescription opioid use with congenital malformations in the offspring: population based cohort study

Bateman

Hernández–Dı́az

Straub

et al. 2021

BMJ

Self Cite

View full text Add to dashboard Cite

show abstract

“…Death certificate data can capture some overdoses not receiving medical attention. Furthermore, recent studies indicate the shortcomings of opioid risk prediction tools in current use and recommend the development of more advanced models to better identify individuals who are at risk (or no risk) of an opioid overdose [ 5 , 11 – 14 ]. Machine-learning techniques may improve opioid-overdose risk prediction because of its capabilities handling a large number of variables and complex interactions [ 6 , 15 – 17 ], as we have recently demonstrated among Medicare beneficiaries.…”

Section: Introductionmentioning

confidence: 99%

Integrating human services and criminal justice data with claims data to predict risk of opioid overdose among Medicaid beneficiaries: A machine-learning approach

et al. 2021

View full text Add to dashboard Cite

Health system data incompletely capture the social risk factors for drug overdose. This study aimed to improve the accuracy of a machine-learning algorithm to predict opioid overdose risk by integrating human services and criminal justice data with health claims data to capture the social determinants of overdose risk. This prognostic study included Medicaid beneficiaries (n = 237,259) in Allegheny County, Pennsylvania enrolled between 2015 and 2018, randomly divided into training, testing, and validation samples. We measured 290 potential predictors (239 derived from Medicaid claims data) in 30-day periods, beginning with the first observed Medicaid enrollment date during the study period. Using a gradient boosting machine, we predicted a composite outcome (i.e., fatal or nonfatal opioid overdose constructed using medical examiner and claims data) in the subsequent month. We compared prediction performance between a Medicaid claims only model to one integrating human services and criminal justice data with Medicaid claims (i.e., integrated model) using several metrics (e.g., C-statistic, number needed to evaluate [NNE] to identify one overdose). Beneficiaries were stratified into risk-score decile subgroups. The samples (training = 79,087, testing = 79,086, validation = 79,086) had similar characteristics (age = 38±18 years, female = 56%, white = 48%, having at least one overdose = 1.7% during study period). Using the validation sample, the integrated model slightly improved on the Medicaid claims only model (C-statistic = 0.885; 95%CI = 0.877–0.892 vs. C-statistic = 0.871; 95%CI = 0.863–0.878), with small corresponding improvements in the NNE and positive predictive value. Nine of the top 30 most important predictors in the integrated model were human services and criminal justice variables. Using the integrated model, approximately 70% of individuals with overdoses were members of the top risk decile (overdose rates in the subsequent month = 47/10,000 beneficiaries). Few individuals in the bottom 9 deciles had overdose episodes (0-12/10,000). Machine-learning algorithms integrating claims and social service and criminal justice data modestly improved opioid overdose prediction among Medicaid beneficiaries for a large U.S. county heavily affected by the opioid crisis.

show abstract

“…First, our work reemphasizes the importance of incorporating opioid prescriptions when defining the spectrum of opioid misuse and OUD. Prior algorithms have incorporated opioid prescriptions to identify opioid misuse (e.g., Calcaterra et al, 2018;Canan et al, 2017;Rough et al, 2019), but additional efforts could place patients on a spectrum of problematic opioid use behaviors. For example, to identify individuals at risk for developing OUD, phenotype risk scores (Bastarache et al, 2019;Ruderfer et al, 2020) could be constructed by agnostically training and testing a risk model using diagnosis codes (for OUD and other relevant predictors) and prescriptions.…”

Section: Discussionmentioning

confidence: 99%

Identifying high-risk comorbidities of short and long-term opioid prescription use

Jennings

Lee

Rocha³

et al. 2021

Preprint

View full text Add to dashboard Cite

BackgroundElectronic health records (EHR) are useful tools for understanding complex medical phenotypes, but they have been underutilized for opioid use disorders (OUD). Patterns of prescription opioid use might provide an objective measure of OUD risk.MethodsWe extracted data for over 2.6 million patients across three health registries (Vanderbilt University Medical Center, Mass General Brigham, Geisinger) between 2005 and 2018. We defined three groups based on levels of opioid exposure: No Prescription, Minimal Exposure (2 prescriptions within 90 days at least once, but never 3 prescriptions <90 days apart), and Chronic Exposure (≥10 opioid prescriptions in a year), and compared them to the full registries and to patients with OUD diagnostic codes. We extracted demographic and clinical characteristics known to co-occur with OUD, including psychiatric and substance use disorders, pain-related diagnoses, HIV, and hepatitis C.ResultsThe prevalence of substance (alcohol, tobacco, cannabis) use disorders was higher in patients with OUD and Chronic Exposure than those with No Prescription or Minimal Exposure. Patients in the OUD and Chronic Exposure groups had more psychiatric (anxiety, depression, schizophrenia, bipolar disorder) and medical comorbidities (pain, hepatitis C, HIV) than those in the Minimal Exposure group. Notably, patients in the Minimal Exposure group had different comorbidity profiles (higher rates of substance use and psychiatric disorders, more pain conditions) than those in the Unscreened or No Prescription groups, highlighting the value of including opioid exposure in studies of OUD.ConclusionsLong-term opioid prescription use may serve as an additional tool to characterize OUD risk.

show abstract

Using prescription claims to detect aberrant behaviors with opioids: comparison and validation of 5 algorithms

Abstract: In 2 large databases, algorithms applied to prescription data had varying accuracy in identifying increased risk of adverse opioid-related events.

Cited by 15 publications

References 16 publications

Association of first trimester prescription opioid use with congenital malformations in the offspring: population based cohort study

Association of first trimester prescription opioid use with congenital malformations in the offspring: population based cohort study

Integrating human services and criminal justice data with claims data to predict risk of opioid overdose among Medicaid beneficiaries: A machine-learning approach

Identifying high-risk comorbidities of short and long-term opioid prescription use

Contact Info

Product

Resources

About