All-payer Claims Database Research Articles

Distribution of care expenditures for men and women with type 2 diabetes treated in primary care in the Netherlands: a case–control study (ZODIAC-59)

ObjectiveThis study aims to provide insight into the distribution of care expenditures for patients with type 2 diabetes mellitus (T2DM)— across multiple healthcare service categories and medical specialties—who receive diabetes...

Pregnancy outcomes among women with endometriosis and fibroids: registry linkage study in Massachusetts

Endometriosis and uterine fibroids are common gynecologic conditions associated with a greater risk for infertility. Previous research has suggested that these conditions are associated with adverse pregnancy outcomes, potentially because of increased utilization of fertility treatments. Our objective was to investigate whether women with a history of endometriosis or fibroids had a greater risk for adverse pregnancy outcomes and whether this risk varied by infertility history and fertility treatment utilization. Deliveries (2013-2017) recorded in Massachusetts' vital records were linked to assisted reproductive technology data, hospital stays, and all-payer claims database. We identified endometriosis and fibroids diagnoses via the all-payer claims database before index delivery. Adjusted relative risks for pregnancy complications were modeled using generalized estimating equations with a log link and Poisson distribution. The influence of subfertility or infertility and assisted reproductive technology was also investigated. Among 91,825 deliveries, 1560 women had endometriosis and 4212 had fibroids. Approximately 30% of women with endometriosis and 26% of women with fibroids experienced subfertility or infertility without utilizing assisted reproductive technology, and 34% of women with endometriosis and 21% of women with fibroids utilized assisted reproductive technology for the index delivery. Women with a history of endometriosis or fibroids were at a greater risk for pregnancy-induced hypertension, preeclampsia, or eclampsia (endometriosis relative risk, 1.17; fibroids relative risk, 1.08), placental abnormalities (endometriosis relative risk, 1.65; fibroids relative risk, 1.38), and cesarean delivery (endometriosis relative risk, 1.22; fibroids relative risk, 1.17) than women with no history of those conditions. Neonates born to women with a history of endometriosis or fibroids were also at a greater risk for preterm birth (endometriosis relative risk, 1.24; fibroids relative risk, 1.17). Associations between fibroids and low birthweight varied by fertility status or assisted reproductive technology (P homogeneity=.01) and were stronger among noninfertile women. Endometriosis or fibroids increased the risk for adverse pregnancy outcomes, possibly warranting differential screening or treatment.

Increased Risk of 90-Day Surgical-Site Infection and Hospital Readmission but Not Reoperation After Open Arthrotomy When Compared With Arthroscopy for Septic Ankle Arthritis

To compare the rates of reoperation and 90-day perioperative complications between open arthrotomy and arthroscopy for the treatment of septic ankle arthritis using a national all-payer claims database. Patients with a diagnosis of septic ankle arthritis who underwent irrigation and debridement through arthroscopy or arthrotomy were identified in a national data set from 2015-2020 through an all-payer claims database. Demographic and comorbidity characteristics including age, sex, infectious etiologies, and Elixhauser comorbidities were obtained. The rate of reoperation, defined as a proxy for failure of initial intervention, was the primary outcome. Secondary outcomes including readmissions, surgical-site infections, amputations, wound complications, and 90-day medical complications were compared between the 2 cohorts. In total, 168 patients undergoing arthroscopy and 794 patients undergoing arthrotomy for septic ankle arthritis were identified. There were no significant differences in reoperation rates between patients who underwent open arthrotomy and those who underwent arthroscopy (P= .997). However, the rates of surgical-site infection (P= .014) and hospital readmission (P < .001) were significantly higher in the open arthrotomy cohort compared with the arthroscopy cohort. Although there was no significant difference in reoperation rates between arthroscopic and open irrigation and debridement for the treatment of septic ankle arthritis, this study showed significantly higher odds of surgical-site infection and hospital readmission in patients who underwent open arthrotomy when compared with arthroscopy. Case-specific patient and technical considerations should guide the surgical decision-making process to limit secondary complications because this study exemplifies similar reoperation rates between the 2 surgical modalities. Level III, nonrandomized cohort analysis.

Cancer Treatment Data in Central Cancer Registries: When Are Supplemental Data Needed?

Background:We evaluated treatment concordance between the Colorado All Payer Claims Database (APCD) and the Colorado Central Cancer Registry (CCCR) to explore whether APCDs can augment registry data. We compare treatment concordance for breast cancer, an extensively studied site with an inpatient reporting source and select leukemias that are often diagnosed outpatient.Methods:We analyzed concordance by cancer type and treatment, patient demographics, reporting source, and health insurance, calculating the sensitivity, specificity, positive predictive values (PPV) and Kappa statistics. We estimated an adjusted logistic regression model to assess whether the APCD statistically significantly reports additional cancer-directed treatments.Results:Among women with breast cancer, 14% had chemotherapy treatments that were absent from the CCCR. Missing treatments were more common among women younger than age 50 (15%) and patients aged 75 and older (19%), rural residents (17%), and when the reporting source was outpatient (22%). Similar and more pronounced patterns for people with leukemia were observed. Concordance for oral treatments was lower for each cancer. Sensitivity and PPVs were high, with moderate Kappa statistics. The APCD was 5.3 percentage points less likely to identify additional treatments for breast cancer patients and 10 percentage points more likely to identify additional treatments when the reporting source was an outpatient facility.Conclusion:A robust data infrastructure is needed to investigate research questions that require population-level analyses, particularly for questions seeking to reduce health inequity and comparisons across payers, including Medicare Advantage and fee-for-service. APCD data are a step toward creating an infrastructure for cancer, particularly for patients who reside in rural areas and/or receive care from outpatient centers.

Abstract PO-230: Racial differences in the adoption of opportunistic salpingectomy for ovarian cancer prevention in the United States

Abstract Introduction: Opportunistic salpingectomy (OS), the removal of the fallopian tubes during benign gynecologic surgery, has been rapidly adopted by clinicians in the U.S. as a strategy for ovarian cancer prevention. However, little is known about racial differences in OS adoption. As with other medical innovations, surgical innovations in gynecology may be adopted differentially across racial groups, exacerbating disparities in care quality. We examined racial differences in OS adoption in a population-based sample of inpatient and outpatient surgeries. Methods: A cohort of 809,911 women aged 18-50 years undergoing hysterectomy with ovarian conservation or surgical sterilization from 2011-2018 was identified using the Premier Healthcare Database, a large geographically diverse all-payer insurance claims database including over 600 acute care hospitals across the U.S. OS adoption was examined by race/ethnicity (non-Hispanic white [NHW], non-Hispanic Black [NHB], Hispanic, other) using multivariable-adjusted log-binomial regression models accounting for hospital-level clustering. Models included a race/ethnicity by year of surgery (2011-2013 [before guideline] vs. 2014-2018 [after guideline]) interaction term to test whether racial differences in OS adoption changed with the release of national guidelines supporting OS use. Results: From 2011-2018, 104,276 hysterectomy and OS (NHW: 60.1%, NHB: 18.8%, Hispanic: 11.4%, other: 9.7%) and 28,015 OS for sterilization (NHW: 65.4%, NHB: 10.0%, Hispanic: 10.6%, other: 14.0%) were performed. The proportion of benign hysterectomy procedures involving OS increased from 6.4% in 2011 to 62.3% in 2018, and the proportion of sterilization procedures involving OS increased from 0.7% in 2011 to 19.1% in 2018. After adjusting for demographic, clinical, procedural, and hospital/provider characteristics, NHB (risk ratio [RR]: 0.97; 95% CI: 0.95-0.99) and Hispanic women (RR: 0.84; 95% CI: 0.77-0.92) were significantly less likely to undergo hysterectomy and OS than NHW women. A significant interaction between race/ethnicity and year of surgery was noted in NHBs compared to NHWs, with a lower likelihood of hysterectomy and OS in NHBs undergoing surgery before but not after national guideline release (RR2011-2013: 0.82; 95% CI: 0.75-0.89; RR2014-2018: 0.99; 95% CI: 0.96-1.01). Results were similar in stratified analyses by hysterectomy route (abdominal, laparoscopic, vaginal). NHBs were also significantly less likely to undergo OS for sterilization than NHWs (RR: 0.84; 95% CI: 0.81-0.88), with no differences by year of surgery. Conclusions: Although OS for ovarian cancer prevention has been rapidly adopted into clinical practice in the U.S., our findings suggest that its adoption has not been equitable across racial groups. NHB and Hispanic women are less likely to undergo OS than NHW women even after adjusting for demographic, clinical, procedural, and hospital/provider characteristics. If these differences in OS adoption persist, disparities in ovarian cancer incidence may worsen in the future. Citation Format: Pritesh S. Karia, Yongmei Huang, Parisa Tehranifar, Kala Visvanathan, Jason D. Wright, Jeanine M. Genkinger. Racial differences in the adoption of opportunistic salpingectomy for ovarian cancer prevention in the United States [abstract]. In: Proceedings of the AACR Virtual Conference: 14th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2021 Oct 6-8. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2022;31(1 Suppl):Abstract nr PO-230.

Real-world analysis of insurance churn among young adults with schizophrenia using the Colorado All-Payer Claims Database.

BACKGROUND: Most patients with schizophrenia are diagnosed in their early twenties and often have commercial insurance at diagnosis. These young adults can experience changes in insurance coverage, that is, "churn," which can lead to disruptions in care. OBJECTIVE: To examine the frequency, speed, and type of insurance churn events in a young adult schizophrenia population with commercial insurance coverage at diagnosis. METHODS: The Colorado All-Payer Claims Database, containing insurance claims data from commercial and public insurers for Colorado residents, was used for the study. Eligible patients were required to have at least 1 inpatient or 2 outpatient claims for schizophrenia or schizoaffective disorder, be of age 18-34 years at index, have previous insurance coverage for 12 consecutive months, and have commercial insurance at diagnosis. These patients were 1:5 propensity score matched (PSM) with nonschizophrenia members. Percentages of members on different insurance types were calculated monthly to assess churn events. Cohorts were compared using descriptive statistics, Cox proportional hazards, and generalized estimating equation models. RESULTS: The matched schizophrenia and nonschizophrenia cohorts comprised 501 and 2,510 members, respectively. Before PSM, cohorts were imbalanced (schizophrenia cohort had a younger median age and higher proportion of males). After matching, the cohorts were similar in terms of the matched baseline characteristics. Previous mental health disorders were more common in the schizophrenia cohort (75%) than in the nonschizophrenia cohort (26%). The proportion of members with at least 1 churn event for the schizophrenia and nonschizophrenia cohorts, respectively, were 53.8% vs 36.5% after 12 months and 84.6% vs 69.2% after 48 months. Time to first churn event was significantly shorter in the schizophrenia cohort (16 months) than the nonschizophrenia cohort (23 months; P < 0.001). Schizophrenia cohort members had 64.1 and 56.8 churn events per 1,000 members per month vs 43.0 (P ≤ 0.001) and 42.8 (P = 0.011) churn events for nonschizophrenia cohort members in the first and second 6-month periods, respectively. Proportions of members in the schizophrenia and nonschizophrenia cohorts on public insurance, respectively, were 22.9% vs 6.9% after 12 months and 52.4% and 10.7% after 48 months. In the schizophrenia cohort, the most common churn event type was from commercial to public insurance rather than to a different commercial insurance; notably, 41% of members were still on a commercial plan 4 years after diagnosis. CONCLUSIONS: Young adults with schizophrenia experienced churn events more rapidly and more frequently than those without schizophrenia for the first 4 years studied after the index date. These disruptions may be associated with reduced access to care and treatment gaps in this vulnerable patient population. DISCLOSURES: This research was sponsored by Janssen Scientific Affairs, LLC. Pesa, Benson, and Patel are employees of Janssen Scientific Affairs, LLC, and are stockholders of Johnson & Johnson. Potluri, Rotter, and Papademetriou are employees of SmartAnalyst Inc, and their work on this study was funded by Janssen Pharmaceuticals. A version of this study was presented as a poster at the Psych Congress 2020 Virtual Experience, September 10-13, 2020.

Surgical Treatment for Septic Ankle Arthritis: A Comparison of Arthrotomy and Arthroscopy

Category:Ankle; ArthroscopyIntroduction/Purpose:Ankle joint septic arthritis can result in substantial morbidity involving cartilage destruction, bone erosion, osteomyelitis and loss of function. Operative intervention is the preferred method of irrigation and debridement, and options include both open arthrotomy and arthroscopic debridement. Prior studies have established both ankle arthroscopy and arthrotomy as effective management strategies for the treatment of ankle septic arthritis. However, no prior research has compared open arthrotomy verses arthroscopy to treat septic arthritis of the ankle, and it is unclear if one approach is superior to the other. The purpose of this study is to compare the rate of reoperation and 90-day perioperative complications between arthroscopy and open arthrotomy for the treatment of septic ankle arthritis using a national all-payer claims database.Methods:Patients who were diagnosed with septic arthritis in a native ankle and underwent irrigation and debridement through arthroscopy or arthrotomy were identified in the 2010-2019 PearlDiver database using International classification of diseases (ICD) and current procedural (CPT) codes. Demographic characteristics including age, gender and Elixhauser comorbidities were obtained for each cohort. The rate of reoperation, defined as a proxy for failure of initial intervention, was the primary outcome. Secondary outcomes including readmission, surgical site infection (SSI) and other 90-day complications were compared between the two cohorts. Univariate analysis using chi-square tests and student T-tests were performed to analyze any differences in patient demographics, comorbidities and complications. Multivariate analysis through logistic regression was subsequently conducted to account for any confounding variables and covariates.Results:In total, 331 patients undergoing arthroscopy and 886 patients undergoing arthrotomy for septic ankle arthritis were identified. On univariate analysis, the incidence of reoperation was higher in the open arthrotomy cohort (12.30%) compared to the arthroscopy cohort (9.37%), however this result was not statistically significant (p=.153). There was also no statistically significant difference between the two cohorts in superficial SSI incidence (3.95% vs. 2.42%, p=.197), readmission rates (36.79% vs. 34.14%, p=.0.391). However, the incidence of renal failure was higher in the open arthrotomy cohort (17.83%) compared with the arthroscopy cohort (12.39%) (p=.022). On multivariate analysis, the incidence of reoperation was higher for those undergoing ankle arthrotomy (OR 1.421; 95% CI 0.937-2.255; p=.108) relative to arthroscopy, however this result was not statistically significant (Table 1). The incidence of renal failure was higher in the open arthrotomy cohort after multivariate analysis (OR 1.615; 95% CI 1.113-2.393; p=0.014).Conclusion:This study demonstrated similar 90-day rates of reoperation, postoperative complications and readmissions, suggesting that arthrotomy and arthroscopy have similar efficacy in treating septic ankle arthritis. Although renal failure was more prevalent in the cohort undergoing open arthrotomy, differences in the primary endpoint of reoperation/repeat washout rates were non-significant. Based on these findings, deciding between these two operative approaches may be determined by patient specific risk factors and factors not elicited by this study including surgeon comfort with the technique, rather than the treatment modality itself.

Increased Medical Complications Following Primary Total Hip Arthroplasty in Patients With Solid Organ Transplant: A Matched Cohort Analysis.

As patient longevity increases following solid organ transplantation (SOT), more transplant patients are undergoing total hip arthroplasty (THA). The purpose of this study is to compare 90-day postoperative complications and 2-year surgical complications following primary THA in patients with and without a history of SOT. Patients with a history of SOT with subsequent primary THA between 2010 and 2018 were identified in a national all-payer claims database (PearlDiver Technologies). This SOT cohort was propensity-matched with a control cohort (no history of SOT) based on age, gender, Charlson Comorbidity Index, and obesity with bivariate analysis to compare outcomes between cohorts. Following matching, 3103 patients were included in the SOT cohort and 6196 patients in the control cohort. The cohorts were successfully matched, with no differences in demographics or comorbidities. Relative to the control cohort, patients with a history of SOT were at significantly increased risk of renal failure (P < .001), anemia (P < .001), arrhythmia with and without atrial fibrillation (P < .001), blood transfusion (P < .001), cellulitis (P= .048), myocardial infarction (P < .001), pneumonia (P= .036), heart failure (P < .001), and sepsis (P= .038) at 90 days postoperatively. There were no significant differences between the cohorts in 2-year surgical revisions, regardless of indication. Following primary THA, patients with a history of SOT are at increased risk of 90-day medical complications but not 2-year surgical complications or revisions relative to patients without SOT. Clinicians should be mindful of the increased risk for cardiopulmonary, renal, hematologic, and infectious complications when counseling and managing this patient population. Level IV-Retrospective Database Study.

Differences in patient outcomes after outpatient GI endoscopy across settings: a statewide matched cohort study

Outpatient GI endoscopy has been shifting from hospital outpatient departments (HOPDs) to ambulatory surgery centers (ASCs) in recent years. However, evidence on whether patient outcomes after endoscopic procedures are comparable across settings is limited. This study compares the incidence of unplanned hospital visits after GI endoscopy performed in ASCs versus HOPDs. We conducted a retrospective cohort study examining unplanned hospital visits after outpatient GI endoscopy performed in Massachusetts during 2014 to 2017 using Massachusetts All-Payer Claims Database and Medicare fee-for-service claims. We identified screening colonoscopy, nonscreening colonoscopy, and esophagogastroduodenoscopies (EGDs) performed in ASCs or HOPDs and estimated unplanned hospital visit rates within 7 and 30 days after these procedures. To compare rates between ASCs and HOPDs, we constructed procedure-specific, propensity score-matched samples and used multilevel logistic regressions adjusting for patient, procedure, and facility characteristics. Seven-day unplanned hospital visit rates were 10.6, 18.3, and 38.9 per 1000 procedures for screening colonoscopy, nonscreening colonoscopy, and EGD, respectively, with significant variation across facilities. ASC patients consistently had fewer postprocedure hospital encounters. The relative risk of having 7-day hospital visits after screening colonoscopy performed in ASCs was .88 (95% confidence interval [CI], .79-.98) compared with HOPDs. The estimates were .84 (95% CI, .75-.94) for nonscreening colonoscopy and .57 (95% CI, .50-.65) for EGD. Thirty-day visits showed similar patterns. Unplanned hospital visits after outpatient GI endoscopy were not uncommon. However, ASC patients consistently had less frequent hospital-based acute care encounters, indicating that GI endoscopy could be performed safely in ASCs for select patients.

Association of Medicaid Expansion in Arkansas With Postpartum Coverage, Outpatient Care, and Racial Disparities

Non-Hispanic Black individuals are disproportionally covered by Medicaid during pregnancy and, compared with non-Hispanic White individuals, have higher rates of postpartum coverage loss and mortality. Expanded Medicaid coverage under the Affordable Care Act may have increased continuity of coverage and access to care in the critical postpartum period in expansion states. To examine the association of Medicaid expansion in Arkansas with continuous postpartum coverage, postpartum health care use, and change in racial disparities in the study outcomes. This cohort study with a difference-in-differences analysis compared persons with Medicaid and commercially financed childbirth, stratified by race, using Arkansas' All-Payer Claims Database for persons with a childbirth between 2013 and 2015. Race and ethnicity from birth certificate data were classified as Hispanic, non-Hispanic Black (hereafter Black), non-Hispanic White (hereafter White), and other (including Asian, Native American or Alaska Native, and Pacific Islander) or unknown race. Data were analyzed between June 2020 and August 2021. Medicaid-paid childbirth after January 1, 2014. Continuous health insurance coverage and the number of outpatient visits during the first 6 months postpartum. A total of 60 990 childbirths (mean [SD] age of birthing person, 27 [5.3] years; 67% White, 22% Black, and 7% Hispanic) were included, among which 72.3% were paid for by Medicaid and 27.7% were paid for by a commercial payer. Medicaid expansion in Arkansas was associated with a 27.8 (95% CI, 26.1-29.5) percentage point increase in continuous insurance coverage and an increase in outpatient visits of 0.9 (95% CI, 0.7-1.1) during the first 6 months postpartum, representing relative increases of 54.9% and 75.0%, respectively. Racial disparities in postpartum coverage decreased from 6.3 (95% CI, 3.9-8.7) percentage points before expansion to -2.0 (95% CI, -2.8 to -1.2) percentage points after expansion. However, disparities in outpatient care between Black and White individuals persisted after Medicaid expansion (preexpansion difference, 0.4 [95% CI, 0.2-0.6] visits; postexpansion difference, 0.5 [95% CI, 0.4-0.6] visits). In this cohort study with a difference-in-differences analysis of 60 990 childbirths, Medicaid expansion was associated with higher rates of postpartum coverage and outpatient visits and lower racial and ethnic disparities in postpartum coverage. However, disparities in outpatient visits between Black and White individuals were unchanged. Additional policy approaches are needed to reduce racial and ethnic disparities in postpartum care.

COVID-19-related adaptations to the implementation and evaluation of a clinic-based intervention designed to improve opioid safety.

The United States faces an opioid crisis with an unprecedented and increasing death rate from opioid overdose. Successfully reducing the rates of opioid use disorder (OUD) and overdose will require the engagement of frontline clinicians to prescribe opioids more safely and to build their capacity to treat patients with OUD using evidence-based approaches. The COVID-19 pandemic has created significant challenges for patients, clinicians and health systems and has been associated with increasing risks of overdoses and deaths. Herein, we review a multidisciplinary project designed to implement and evaluate clinic-based interventions in Oregon, USA, to improve pain management, opioid prescribing and treatment of OUD. The intervention, called Improving PaIn aNd OPiOId MaNagemenT in Primary Care (PINPOINT), combines practice facilitation, academic detailing and education through the Oregon ECHO Network. Implementation of PINPOINT has occurred across the Oregon Rural Practice-based Research Network and has involved 49 clinic sites to date. To evaluate the impact of the intervention, the research team created the Provider Results of Opioid Management and Prescribing Training (PROMPT), a dataset that links information from the state prescription drug monitoring program, all-payer claims database, emergency medical services, vital records and substance use disorder treatment system. The PROMPT dataset will allow evaluation of the impact of the intervention at both the clinician and clinic levels. Due to the constraints of the COVID-19 pandemic, elements of both implementation and evaluation required significant adaptations to continue to meet the original project goals.

Delineating the Type 2 Diabetes Population in the Netherlands Using an All-Payer Claims Database: Specialist Care, Medication Utilization and Expenditures 2016-2018.

ObjectivesThe aim of this study was to describe the healthcare utilization and expenditures related to medical specialist care and medication of the entire type 2 diabetes population in the Netherlands in detail.MethodsFor this retrospective, observational study, we used an all-payer claims database. Comprehensive data on specialist care and medication utilization and expenditures of the type 2 diabetes population (n = 900,522 in 2018) were obtained and analyzed descriptively. Data were analyzed across medical specialties and for various types of diabetes medication (or glucose-lowering drugs [GLDs]) and other medication.ResultsSpecialist care utilization was diverse: different medical specialties were visited by a considerable fraction of the type 2 diabetes population. Total expenditures on specialist care were €2498 million in 2018 (i.e., 10.6% of the national specialist care expenditures). In total, 97.8% of patients used other medication (not GLDs) and 81.8% used GLDs; 25.6% of medication expenditures were for GLDs. For both specialist care and medication, mean expenditures per treated patient were higher than median expenditures, indicating a skewed distribution of spending.ConclusionUse of and expenditures on specialist care and medication of the type 2 diabetes population is diverse. These heterogeneous healthcare use patterns are likely caused by the presence of comorbidities. Additionally, we found that a small fraction of the population is responsible for a large share of the expenditures. A shift towards more patient-centered care could lead to health improvements and a reduction in overall costs, subsequently promoting the sustainability of healthcare systems.

Identifying and delineating the type 2 diabetes population in the Netherlands using an all-payer claims database: characteristics, healthcare utilisation and expenditures

ObjectivesWe aimed to identify and delineate the Dutch type 2 diabetes population and the distribution of healthcare utilisation and expenditures across the health system from 2016 to 2018 using an...

Improving suicide risk prediction via targeted data fusion: proof of concept using medical claims data.

Reducing suicidal behavior among patients in the healthcare system requires accurate and explainable predictive models of suicide risk across diverse healthcare settings. We proposed a general targeted fusion learning framework that can be used to build a tailored risk prediction model for any specific healthcare setting, drawing on information fusion from a separate more comprehensive dataset with indirect sample linkage through patient similarities. As a proof of concept, we predicted suicide-related hospitalizations for pediatric patients in a limited statewide Hospital Inpatient Discharge Dataset (HIDD) fused with a more comprehensive medical All-Payer Claims Database (APCD) from Connecticut. We built a suicide risk prediction model for the source data (APCD) and calculated patient risk scores. Patient similarity scores between patients in the source and target (HIDD) datasets using their demographic characteristics and diagnosis codes were assessed. A fused risk score was generated for each patient in the target dataset using our proposed targeted fusion framework. With this model, the averaged sensitivities at 90% and 95% specificity improved by 67% and 171%, and the positive predictive values for the combined fusion model improved 64% and 135% compared to the conventional model. We proposed a general targeted fusion learning framework that can be used to build a tailored predictive model for any specific healthcare setting. Results from this study suggest we can improve the performance of predictive models in specific target settings without complete integration of the raw records from external data sources.

Dental care utilization in Massachusetts before and after initiation of medication for opioid use disorder: A cross-sectional study of a state all-payer claims database.

Individuals with opioid use disorder (OUD) are at higher risk of poor oral health. Medication for opioid use disorder (MOUD) has been shown to improve outcomes for patients with OUD, but it is unknown how initiation of MOUD affects access to oral health services. This was a retrospective analysis of all individuals in the Massachusetts All-Payer Claims Database prescribed oral buprenorphine-naloxone or injectable naltrexone from 2013 to 2016. We evaluated dental utilization in the year before and after beginning MOUD. A logistic regression predicting dental utilization was conducted. Among the 54,791 individuals, rates of dental utilization were low both before and after MOUD (10.5% and 10% with a dental visit, respectively). Of those who did not have a dental visit in the year before starting MOUD, 95.1% did not have a dental visit in the year after. Rates of various procedure types were comparable before and after MOUD. In a logistic regression, a prior dental visit was associated with 9.82 times the odds (95% CI 9.14-10.55) of having a dental visit after starting MOUD; increasing age, being prescribed naltrexone, having a mood disorder or HIV, year of initiation or being on Medicaid were also associated with having a dental visit. Male patients and those with Medicare or private insurance were less likely to have a dental visit. Initiating MOUD did not substantially result in increased dental access or substantial changes in dental procedures received. Patients receiving treatment for OUD may require additional support to access dental care.

The role of all-payer claims databases to expand central cancer registries: Experience from Colorado.

To evaluate the quality of a multiyear linkage between the Colorado all-payer claims database (APCD) and the Colorado Central Cancer Registry. Secondary 2012-2017 data from the APCD and the Colorado Cancer Registry. Descriptive analysis of the proportion of cases captured by the linkage in relation to the cases reported by the registry. We used probabilistic linkage to combine records from both data sources for all patients diagnosed with cancer. We successfully linked 93% of the 146,884 patients in the registry. Approximately 63% of linked patients were perfect matches on five identifiers. Of partial matches, 81.6% were matched on four identifiers with missing or partial Social Security Numbers. The linkage rate was lower for uninsured patients at diagnosis (74.7%) or patients with private plans (89.4%) but close to 100% for Medicare and Medicaid enrollees. Most of the 29% of patients who did not have claims at the time of diagnosis were covered by private plans that may not submit claims. APCD-registry linkages are a promising source of data to conduct population-based research from multiple payers. However, not all payers submit claims, and the quality of the data may vary by state.

Understanding racial disparities in severe maternal morbidity using Bayesian network analysis.

Previous studies have evaluated the marginal effect of various factors on the risk of severe maternal morbidity (SMM) using regression approaches. We add to this literature by utilizing a Bayesian network (BN) approach to understand the joint effects of clinical, demographic, and area-level factors. We conducted a retrospective observational study using linked birth certificate and insurance claims data from the Arkansas All-Payer Claims Database (APCD), for the years 2013 through 2017. We used various learning algorithms and measures of arc strength to choose the most robust network structure. We then performed various conditional probabilistic queries using Monte Carlo simulation to understand disparities in SMM. We found that anemia and hypertensive disorder of pregnancy may be important clinical comorbidities to target in order to reduce SMM overall as well as racial disparities in SMM.

A comparative analysis of attribution methods for newly diagnosed cancer patients to hospitals in oncology care.

60 Background: Attribution is the term that describes how payers and employers determine which provider is responsible for a member’s care, when prospective patient selection is not available. Several claims-based attribution models exist for primary care. The challenges of attribution become salient in oncology because cancer care is often multidisciplinary—involving medical oncologists, surgeons, and radiation oncologists—rendering it difficult to discern which practice should be held accountable. Given the uniqueness of the population the generic primary care attribution model does not fit well. Methods: The objective of this study was to propose and compare methods that attribute patients to hospitals using administrative databases. The models were defined as pre-specified rules that determine the specific patients, types of health care services, and the duration of care for which providers are responsible. Both National Medicare database and Statewide All Payers Claims Database (2014-2015) were used for the purpose to analyze Medicare and commercial population. Two different methods with 6 and 12-months episodes were compared. Method 1 defined episode trigger as first diagnosis of cancer and Method 2 as first treatment of cancer, both with no prior 12 months of cancer diagnosis. Patients were attributed to a hospital based on plurality of claims (including both outpatient and inpatient) with a minimum threshold of 2 claims from the same hospital. Ties were broken with the most recent visit, if not, the highest cost. Success measure was defined as highest attribution rate vs. lowest feasible unassigned rate. Results: A total of 1.7 million patients were included in the Medicare cohort and 98,005 from All Payers Claims database (APCD). Results suggested for a 6 months episode, Method 1 vs. Method 2 attributed 94% vs. 98% to a hospital. For a 12-months episode, Method 1 vs. Method 2 attributed 96% vs. 98% to a hospital. Similar results were evident from APCD. It was evident attribution of patients were higher towards their first diagnosed hospital. Also, longer the duration of care, better the attribution. The outcome of the study was a tool in tableau. Conclusions: Attribution is not a problem to be solved and left alone; it requires ongoing work, enhancements. This study results in a framework for attribution that can be used as a mechanism to link indicators of patient-level health care quality and spending to specific providers for the purpose of profiling and accountability. Better systems will seek to identify specific care for a condition based on the types of doctors a patient is seeing (chemotherapy), and identify who is delivering most of that care vs. who is delivering different types of care (surgery, radiation, primary care).

Patient, Provider, and Health Care System Characteristics Associated With Overuse in Bronchiolitis.

The American Academy of Pediatrics recommends against the routine use of β-agonists, corticosteroids, antibiotics, chest radiographs, and viral testing in bronchiolitis, but use of these modalities continues. Our objective for this study was to determine the patient, provider, and health care system characteristics that are associated with receipt of low-value services. Using the Virginia All-Payers Claims Database, we conducted a retrospective cross-sectional study of children aged 0 to 23 months with bronchiolitis (code J21, International Classification of Diseases, 10th Revision) in 2018. We recorded medications within 3 days and chest radiography or viral testing within 1 day of diagnosis. Using Poisson regression, we identified characteristics associated with each type of overuse. Fifty-six percent of children with bronchiolitis received ≥1 form of overuse, including 9% corticosteroids, 17% antibiotics, 20% β-agonists, 26% respiratory syncytial virus testing, and 18% chest radiographs. Commercially insured children were more likely than publicly insured children to receive a low-value service (adjusted prevalence ratio [aPR] 1.21; 95% confidence interval [CI]: 1.15-1.30; P < .0001). Children in emergency settings were more likely to receive a low-value service (aPR 1.24; 95% CI: 1.15-1.33; P < .0001) compared with children in inpatient settings. Children seen in rural locations were more likely than children seen in cities to receive a low-value service (aPR 1.19; 95% CI: 1.11-1.29; P < .0001). Overuse in bronchiolitis remains common and occurs frequently in emergency and outpatient settings and rural locations. Quality improvement initiatives aimed at reducing overuse should include these clinical environments.

Statewide Analysis Reveals Period of Well-Child Visit Attendance for Earlier Diagnosis of Autism Spectrum Disorder

To explore the relationship between well-child visit (WCV) attendance during early childhood and age at autism spectrum disorder (ASD) diagnosis using data drawn from a statewide all-payer claims database. We used a correlational study design with longitudinal data drawn from the Virginia All-Payer Claims Database. All children born in 2011 with a diagnosis of ASD were included (n=253). Survival analysis determined the impact of WCV attendance on ASD diagnosis at each American Academy of Pediatrics-recommended early childhood visit, and the 5-year visit. Survival analysis revealed a significant impact of WCV attendance at the 24-month, 3-, and 4-year visits on earlier ASD diagnosis. Children who attended the 24-month visit were diagnosed nearly 10months earlier than those who did not. Overall, children with ASD attended fewer than 50% of visits during early childhood. Promoting consistent WCV attendance during early childhood is an actionable strategy for improving early identification of ASD. Further exploration is needed to determine barriers to visit attendance and the impact of patterns of early childhood WCV attendance on age of ASD diagnosis. Development and implementation of interventions to promote adherence to the American Academy of Pediatrics-recommended visits is needed.