Dataset Information

Identifying Opioid Use Disorder from Longitudinal Healthcare Data using a Multi-stream Transformer.

ABSTRACT: Opioid Use Disorder (OUD) is a public health crisis costing the US billions of dollars annually in healthcare, lost workplace productivity, and crime. Analyzing longitudinal healthcare data is critical in addressing many real-world problems in healthcare. Leveraging the real-world longitudinal healthcare data, we propose a novel multi-stream transformer model called MUPOD for OUD identification. MUPOD is designed to simultaneously analyze multiple types of healthcare data streams, such as medications and diagnoses, by attending to segments within and across these data streams. Our model tested on the data from 392,492 patients with long-term back pain problems showed significantly better performance than the traditional models and recently developed deep learning models.

SUBMITTER: Fouladvand S

PROVIDER: S-EPMC8861731 | biostudies-literature | 2021

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Identifying Opioid Use Disorder from Longitudinal Healthcare Data using a Multi-stream Transformer.

Fouladvand Sajjad S Talbert Jeffery J Dwoskin Linda P LP Bush Heather H Meadows Amy Lynn AL Peterson Lars E LE Roggenkamp Steve K SK Kavuluru Ramakanth R Chen Jin J

AMIA ... Annual Symposium proceedings. AMIA Symposium 20210101

Opioid Use Disorder (OUD) is a public health crisis costing the US billions of dollars annually in healthcare, lost workplace productivity, and crime. Analyzing longitudinal healthcare data is critical in addressing many real-world problems in healthcare. Leveraging the real-world longitudinal healthcare data, we propose a novel multi-stream transformer model called MUPOD for OUD identification. MUPOD is designed to simultaneously analyze multiple types of healthcare data streams, such as medica ...[more]

PMID: 35308960

Similar Datasets

Project description:AimsTo compare healthcare costs and use between United States (US) Veterans Health Administration (VHA) patients with opioid use disorder (OUD) who experienced an opioid overdose (OD cohort) and patients with OUD who did not experience an opioid overdose (non-OD cohort).DesignThis is a retrospective cohort study of administrative and clinical data.SettingThe largest integrated national health-care system is the US Veterans Health Administration's healthcare systems.ParticipantsWe included VHA patients diagnosed with OUD from October 1, 2017 through September 30, 2018. We identified the index date of overdose for patients who had an overdose. Our control group, which included patients with OUD who did not have an overdose, was randomly assigned an index date. A total of 66 513 patients with OUD were included for analysis (OD cohort: n = 1413; non-OD cohort: n = 65 100).MeasurementsMonthly adjusted healthcare-related costs and use in the year before and after the index date. We used generalized estimating equation models to compare patients with an opioid overdose and controls in a difference-in-differences framework.FindingsCompared with the non-OD cohort, an opioid overdose was associated with an increase of $16 890 [95% confidence interval (CI) = $15 611-18 169; P < 0.001] in healthcare costs for an estimated $23.9 million in direct costs to VHA (95% CI = $22.1 million, $25.7 million) within the 30 days following overdose after adjusting for baseline characteristics. Inpatient costs ($13 515; 95% CI = $12 378-14 652; P < 0.001) reflected most of this increase. Inpatient days (+6.15 days; 95% CI, = 5.33-6.97; P < 0.001), inpatient admissions (+1.01 admissions; 95% CI = 0.93-1.10; P < 0.001) and outpatient visits (+1.59 visits; 95% CI = 1.34-1.84; P < 0.001) also increased in the month after opioid overdose. Within the overdose cohort, healthcare costs and use remained higher in the year after overdose compared with pre-overdose trends.ConclusionsThe US Veterans Health Administration patients with opioid use disorder (OUD) who have experienced an opioid overdose have increased healthcare costs and use that remain significantly higher in the month and continuing through the year after overdose than OUD patients who have not experienced an overdose.

Project description:BACKGROUND:Deploying accurate computable phenotypes in pragmatic trials requires a trade-off between precise and clinically sensical variable selection. In particular, evaluating the medical encounter to assess a pattern leading to clinically significant impairment or distress indicative of disease is a difficult modeling challenge for the emergency department. OBJECTIVE:This study aimed to derive and validate an electronic health record-based computable phenotype to identify emergency department patients with opioid use disorder using physician chart review as a reference standard. METHODS:A two-algorithm computable phenotype was developed and evaluated using structured clinical data across 13 emergency departments in two large health care systems. Algorithm 1 combined clinician and billing codes. Algorithm 2 used chief complaint structured data suggestive of opioid use disorder. To evaluate the algorithms in both internal and external validation phases, two emergency medicine physicians, with a third acting as adjudicator, reviewed a pragmatic sample of 231 charts: 125 internal validation (75 positive and 50 negative), 106 external validation (56 positive and 50 negative). RESULTS:Cohen kappa, measuring agreement between reviewers, for the internal and external validation cohorts was 0.95 and 0.93, respectively. In the internal validation phase, Algorithm 1 had a positive predictive value (PPV) of 0.96 (95% CI 0.863-0.995) and a negative predictive value (NPV) of 0.98 (95% CI 0.893-0.999), and Algorithm 2 had a PPV of 0.8 (95% CI 0.593-0.932) and an NPV of 1.0 (one-sided 97.5% CI 0.863-1). In the external validation phase, the phenotype had a PPV of 0.95 (95% CI 0.851-0.989) and an NPV of 0.92 (95% CI 0.807-0.978). CONCLUSIONS:This phenotype detected emergency department patients with opioid use disorder with high predictive values and reliability. Its algorithms were transportable across health care systems and have potential value for both clinical and research purposes.

Project description:This study aims to fill three knowledge gaps: (1) unclear role of ecological factors in shaping older adults' risk of opioid use disorder (OUD), (2) a lack of longitudinal perspective in OUD research among older adults, and (3) underexplored racial/ethnic differences in the determinants of OUD in older populations. This study estimates the effects of county-level social isolation, concentrated disadvantage, and income inequality on older adults' risk of OUD using longitudinal data analysis. We merged the 2013-2018 Medicare population (aged 65+) data to the American Community Survey 5-year county-level estimates to create a person-year dataset (N = 47,291,217 person-years) and used conditional logit fixed-effects modeling to test whether changes in individual- and county-level covariates alter older adults' risk of OUD. Moreover, we conducted race/ethnicity-specific models to compare how these associations vary across racial/ethnic groups. At the county-level, a one-unit increase in social isolation (mean = -0.197, SD = 0.511) increased the risk of OUD by 5.5 percent (OR = 1.055; 95% CI = [1.018, 1.094]) and a one-percentage-point increase in the working population employed in primary industry decreases the risk of OUD by 1 percent (OR = 0.990; 95% CI = [0.985, 0.996]). At the individual-level, increases in the Medicare Hierarchical Condition Categories risk score, physical comorbidity, and mental comorbidity all elevate the risk of OUD. The relationship between county-level social isolation and OUD is driven by non-Hispanic whites, while Hispanic beneficiaries are less sensitive to the changes in county-level factors than any other racial ethnic groups. Between 2013 and 2018, US older adults' risk of OUD was associated with both ecological and individual factors, which carries implications for intervention. Further research is needed to understand why associations of individual factors with OUD are comparable across racial/ethnic groups, but county-level social isolation is only associated with OUD among non-Hispanic white beneficiaries.

Project description:ObjectiveTo assess the effectiveness of evidence-based quality improvement (EBQI) as an implementation strategy to expand the use of medications for opioid use disorder (MOUD) within nonspecialty settings.Data sources and study settingWe studied eight facilities in one Veteran Health Administration (VHA) region from October 2015 to September 2022 using administrative data.Study designInitially a pilot, we sequentially engaged seven of eight facilities from April 2018 to September 2022 using EBQI, consisting of multilevel stakeholder engagement, technical support, practice facilitation, and data feedback. We established facility-level interdisciplinary quality improvement (QI) teams and a regional-level cross-facility collaborative. We used a nonrandomized stepped wedge design with repeated cross sections to accommodate the phased implementation. Using aggregate facility-level data from October 2015 to September 2022, we analyzed changes in patients receiving MOUD using hierarchical multiple logistic regression.Data collection/extraction methodsEligible patients had an opioid use disorder (OUD) diagnosis from an outpatient or inpatient visit in the previous year. Receiving MOUD was defined as having been prescribed an opioid agonist or antagonist treatment or a visit to an opioid substitution clinic.Principal findingsThe probability of patients with OUD receiving MOUD improved significantly over time for all eight facilities (average marginal effect [AME]: 0.0057, 95% CI: 0.0044, 0.0070) due to ongoing VHA initiatives, with the probability of receiving MOUD increasing by 0.577 percentage points, on average, each quarter, totaling 16 percentage points during the evaluation period. The seven facilities engaging in EBQI experienced, on average, an additional 5.25 percentage point increase in the probability of receiving MOUD (AME: 0.0525, 95%CI: 0.0280, 0.0769). EBQI duration was not associated with changes.ConclusionsEBQI was effective for expanding access to MOUD in nonspecialty settings, resulting in increases in patients receiving MOUD exceeding those associated with temporal trends. Additional research is needed due to recent MOUD expansion legislation.

Dataset Information

Identifying Opioid Use Disorder from Longitudinal Healthcare Data using a Multi-stream Transformer.

Publications

Identifying Opioid Use Disorder from Longitudinal Healthcare Data using a Multi-stream Transformer.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets