Dataset Information

Body Mass Index Variable Interpolation to Expand the Utility of Real-world Administrative Healthcare Claims Database Analyses.

ABSTRACT:

Introduction

Administrative claims data provide an important source for real-world evidence (RWE) generation, but incomplete reporting, such as for body mass index (BMI), limits the sample sizes that can be analyzed to address certain research questions. The objective of this study was to construct models by implementing machine-learning (ML) algorithms to predict BMI classifications (? 30, ? 35, and ? 40 kg/m²) in administrative healthcare claims databases, and then internally and externally validate them.

Methods

Five advanced ML algorithms were implemented for each BMI classification on a random sampling of BMI readings from the Optum PanTher Electronic Health Record database (2%) and the Optum Clinformatics Date of Death (20%) database, while incorporating baseline demographic and clinical characteristics. Sensitivity analyses with oversampling ratios were conducted. Model performance was validated internally and externally.

Results

Models trained on the Super Learner ML algorithm (SLA) yielded the best BMI classification predictive performance. SLA model 1 utilized sociodemographic and clinical characteristics, including baseline BMI values; the area under the receiver operating characteristic curve (ROC AUC) was approximately 88% for the prediction of BMI classifications of ? 30, ? 35, and ? 40 kg/m² (internal validation), while accuracy ranged from 87.9% to 92.8% and specificity ranged from 91.8% to 94.7%. SLA model 2 utilized sociodemographic information and clinical characteristics, excluding baseline BMI values; ROC AUC was approximately 73% for the prediction of BMI classifications of ? 30, ? 35, and ? 40 kg/m² (internal validation), while accuracy ranged from 73.6% to 80.0% and specificity ranged from 71.6% to 85.9%. The external validation on the MarketScan Commercial Claims and Encounters database yielded relatively consistent results with slightly diminished performance.

Conclusion

This study demonstrated the feasibility and validity of using ML algorithms to predict BMI classifications in administrative healthcare claims data to expand the utility for RWE generation.

SUBMITTER: Wu B

PROVIDER: S-EPMC7889527 | biostudies-literature | 2021 Feb

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Body Mass Index Variable Interpolation to Expand the Utility of Real-world Administrative Healthcare Claims Database Analyses.

Wu Bingcao B Chow Wing W Sakthivel Monish M Kakade Onkar O Gupta Kartikeya K Israel Debra D Chen Yen-Wen YW Kuruvilla Aarti Susan AS

Advances in therapy 20210111 2

<h4>Introduction</h4>Administrative claims data provide an important source for real-world evidence (RWE) generation, but incomplete reporting, such as for body mass index (BMI), limits the sample sizes that can be analyzed to address certain research questions. The objective of this study was to construct models by implementing machine-learning (ML) algorithms to predict BMI classifications (≥ 30, ≥ 35, and ≥ 40 kg/m<sup>2</sup>) in administrative healthcare claims databases, and then internall ...[more]

PMID: 33432543

Similar Datasets

Project description:INTRODUCTION:In addition to the considerable patient and societal burdens, the financial burdens of ankylosing spondylitis (AS) are substantial. Understanding both all-cause and AS-specific direct costs in patients with AS is important if we are to understand the financial impact on patients with AS and payers in the United States. This study assessed both all-cause and AS-specific healthcare utilization and direct costs in US patients with AS using administrative claims data. METHODS:Adults aged ≥ 18 years enrolled in the MarketScan® Commercial and Medicare databases with ≥ 1 inpatient or ≥ 2 non-rule-out outpatient diagnoses of AS between January 1, 2013, and December 31, 2013, were included. Patients had continuous enrollment with medical and pharmacy benefits for ≥ 12 months before and after the index date (first diagnosis). Non-AS controls were matched up to 5:1 to patients with AS on age, geographic region, index calendar year, and sex. All-cause and AS-specific healthcare utilization and direct costs were measured during the follow-up period and reported as per patient per year. RESULTS:Patients with AS (N = 6679) had significantly higher rates of total all-cause inpatient admission (12% vs 6%), emergency department visits (23% vs 15%), nonhospital-based outpatient visits (100% vs 84%), hospital-based outpatient visits (68% vs 46%), other outpatient services (97% vs 81%), and medication use (97% vs 82%) compared with matched controls (N = 19,951). Patients with AS had approximately tenfold higher median total healthcare costs than matched controls ($24,978 vs $2139 per patient per year), largely driven by increased outpatient and pharmacy costs; P < 0.05 for all comparisons. The median (IQR) total AS-specific healthcare costs were $10,250 ($774, $28,824). CONCLUSION:In this analysis of claims data, increased outpatient and pharmacy costs were key contributors to higher all-cause total healthcare costs in US patients with AS. FUNDING:Novartis Pharmaceuticals Corporation, East Hanover, NJ.

Project description:BackgroundTrifluridine/tipiracil (FTD/TPI) and regorafenib (REG) are standard therapies for refractory metastatic colorectal cancer (mCRC). No results of large real-world data directly comparing FTD/TPI + bevacizumab (BEV) with FTD/TPI or REG monotherapy have been reported. We evaluated the efficacy and safety of FTD/TPI + BEV in a real-world setting.Materials and methodsThis retrospective study used a Japanese claims database provided by Medical Data Vision Co., Ltd. (Tokyo, Japan). Eligible patients were aged 20 years and over with a diagnosis of mCRC, and received their first dose of FTD/TPI or REG from 2014 to 2021. The primary endpoint was overall survival (OS) in a propensity score matching (PSM) population in which PSM was carried out by matching using a 1 : 1 ratio for the FTD/TPI + BEV group and the control group (FTD/TPI or REG) by propensity score. To enhance robustness, sensitivity analyses of OS were carried out using the inverse probability treatment weighted (IPTW) approach and the analysis in the all eligible population. Secondary endpoints included time to treatment discontinuation (TTD), incidence of adverse events, and post-treatment.ResultsEligible population was 2369 for the FTD/TPI + BEV group and 9318 for the control group. The PSM population was 1787 for each group. Median OS (mOS) was longer in the FTD/TPI + BEV group compared to the control group [17.0 versus 11.6 months, hazard ratio (HR) = 0.70, P < 0.001] in the PSM population. Similarly, mOS was longer for the FTD/TPI + BEV group compared to that for the control group in IPTW analyses and in the all eligible population (both HRs = 0.68). Median TTD was 3.3 months for the FTD/TPI + BEV group and 1.8 months for the control group in the PSM population (P < 0.001).ConclusionsReal-world data showed that FTD/TPI + BEV was significantly associated with OS and TTD compared to FTD/TPI or REG. In clinical practice, FTD/TPI + BEV can be a favorable regimen for refractory mCRC.

Project description:The prevalence of nonalcoholic fatty liver disease (NAFLD) is increasing. The health care burden resulting from the multidisciplinary management of this complex disease is unknown. We assessed the total health care cost and resource utilization associated with a new NAFLD diagnosis, compared with controls with similar comorbidities. We used OptumLabs Data Warehouse, a large national administrative claims database with longitudinal health data of over 100 million individuals enrolled in private and Medicare Advantage health plans. We identified 152,064 adults with a first claim for NAFLD between 2010 and 2014, of which 108,420 were matched 1:1 by age, sex, metabolic comorbidities, length of follow-up, year of diagnosis, race, geographic region, and insurance type to non-NAFLD contemporary controls from the OptumLabs Data Warehouse database. Median follow-up time was 2.6 (range 1-6.5) years. The final study cohort consisted of 216,840 people with median age 55 (range 18-86) years, 53% female, 78% white. The total annual cost of care per NAFLD patient with private insurance was $7,804 (interquartile range [IQR] $3,068-$18,688) for a new diagnosis and $3,789 (IQR $1,176-$10,539) for long-term management. These costs are significantly higher than the total annual costs of $2,298 (IQR $681-$6,580) per matched control with similar metabolic comorbidities but without NAFLD. The largest increases in health care utilization that may account for the increased costs in NAFLD compared with controls are represented by liver biopsies (relative risk [RR] = 55.00, 95% confidence interval [CI] 24.48-123.59), imaging (RR = 3.95, 95% CI 3.77-4.15), and hospitalizations (RR = 1.87, 95% CI 1.73-2.02). Conclusion: The costs associated with the care for NAFLD independent of its metabolic comorbidities are very high, especially at first diagnosis. Research efforts shouldfocus on identification of underlying determinants of use, sources of excess cost, and development of cost-effective diagnostic tests.

Project description:ObjectiveTo derive and validate a comorbidity-based delirium risk index (DRI) to predict postoperative delirium.Data source/study settingData of 506 438 hip fracture repair surgeries from 2006 to 2016 were collected to derive DRI and perform internal validation from the Premier Healthcare Database, which provided billing information on 20-25 percent of hospitalizations in the USA. Additionally, data of 1 130 569 knee arthroplasty surgeries were retrieved for external validation.Study designThirty-six commonly seen comorbidities were evaluated by logistic regression with the outcome of postoperative delirium. The hip fracture repair surgery cohort was separated into a training dataset (60 percent) and an internal validation (40 percent) dataset. The least absolute shrinkage and selection operator (LASSO) procedure was applied for variable selection, and weights were assigned to selected comorbidities to quantify corresponding risks. The newly developed DRI was then compared to the Charlson-Deyo Index for goodness-of-fit and predictive ability, using the Akaike information criterion (AIC), Bayesian information criterion (BIC), area under the ROC curve (AUC) for goodness-of-fit, and odds ratios for predictive performance. Additional internal validation was performed by splitting the data by four regions and in 4 randomly selected hospitals. External validation was conducted in patients with knee arthroplasty surgeries.Data collectionHip fracture repair surgeries, knee arthroplasty surgeries, and comorbidities were identified by using ICD-9 codes. Postoperative delirium was defined by using ICD-9 codes and analyzing billing information for antipsychotics (specifically haloperidol, olanzapine, and quetiapine) typically recommended to treat delirium.Principal findingsThe derived DRI includes 14 comorbidities and assigns comorbidities weights ranging from 1 to 6. The DRI outperformed the Charlson-Deyo Comorbidity Index with better goodness-of-fit and predictive performance.ConclusionsDelirium risk index is a valid comorbidity index for covariate adjustment and risk prediction in the context of postoperative delirium. Future work is needed to test its performance in different patient populations and varying definitions of delirium.

Dataset Information

Body Mass Index Variable Interpolation to Expand the Utility of Real-world Administrative Healthcare Claims Database Analyses.

Introduction

Methods

Results

Conclusion

Publications

Body Mass Index Variable Interpolation to Expand the Utility of Real-world Administrative Healthcare Claims Database Analyses.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets