Dataset Information

Assessing the Value of Unsupervised Clustering in Predicting Persistent High Health Care Utilizers: Retrospective Analysis of Insurance Claims Data.

ABSTRACT:

Background

A high proportion of health care services are persistently utilized by a small subpopulation of patients. To improve clinical outcomes while reducing costs and utilization, population health management programs often provide targeted interventions to patients who may become persistent high users/utilizers (PHUs). Enhanced prediction and management of PHUs can improve health care system efficiencies and improve the overall quality of patient care.

Objective

The aim of this study was to detect key classes of diseases and medications among the study population and to assess the predictive value of these classes in identifying PHUs.

Methods

This study was a retrospective analysis of insurance claims data of patients from the Johns Hopkins Health Care system. We defined a PHU as a patient incurring health care costs in the top 20% of all patients' costs for 4 consecutive 6-month periods. We used 2013 claims data to predict PHU status in 2014-2015. We applied latent class analysis (LCA), an unsupervised clustering approach, to identify patient subgroups with similar diagnostic and medication patterns to differentiate variations in health care utilization across PHUs. Logistic regression models were then built to predict PHUs in the full population and in select subpopulations. Predictors included LCA membership probabilities, demographic covariates, and health utilization covariates. Predictive powers of the regression models were assessed and compared using standard metrics.

Results

We identified 164,221 patients with continuous enrollment between 2013 and 2015. The mean study population age was 19.7 years, 55.9% were women, 3.3% had ≥1 hospitalization, and 19.1% had 10+ outpatient visits in 2013. A total of 8359 (5.09%) patients were identified as PHUs in both 2014 and 2015. The LCA performed optimally when assigning patients to four probability disease/medication classes. Given the feedback provided by clinical experts, we further divided the population into four diagnostic groups for sensitivity analysis: acute upper respiratory infection (URI) (n=53,232; 4.6% PHUs), mental health (n=34,456; 12.8% PHUs), otitis media (n=24,992; 4.5% PHUs), and musculoskeletal (n=24,799; 15.5% PHUs). For the regression models predicting PHUs in the full population, the F1-score classification metric was lower using a parsimonious model that included LCA categories (F1=38.62%) compared to that of a complex risk stratification model with a full set of predictors (F1=48.20%). However, the LCA-enabled simple models were comparable to the complex model when predicting PHUs in the mental health and musculoskeletal subpopulations (F1-scores of 48.69% and 48.15%, respectively). F1-scores were lower than that of the complex model when the LCA-enabled models were limited to the otitis media and acute URI subpopulations (45.77% and 43.05%, respectively).

Conclusions

Our study illustrates the value of LCA in identifying subgroups of patients with similar patterns of diagnoses and medications. Our results show that LCA-derived classes can simplify predictive models of PHUs without compromising predictive accuracy. Future studies should investigate the value of LCA-derived classes for predicting PHUs in other health care settings.

SUBMITTER: Ramachandran R

PROVIDER: S-EPMC8663459 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Background Invasive surgical procedures carry risk of harm to patients. In addition to avoidable harm, disparities between patient expectations and the outcome of a procedure may lead to patient injury claims. The follow-up of claims and compensation is an important entity for patient safety. The number of claims should be related to the surgical volume, so that a healthcare provider can benchmark with similar organizations and see if its trends are developing favourably or deteriorating. Our objective was to find out the claims and compensation rates due to surgery in an insurance-based system. Methods Data related to surgical claims and reference volume in the period 2011–2015 were collected from the claim register of the Finnish Patient Insurance Centre and benchmarking community register of Finnish operating departments. The data included age, gender, hospital, year of surgery, surgical code, and outcome of the claim. Results There were 8,901 claims related to the corresponding reference group of 1,470,435 surgical procedures. The claims rate was 0.61% and compensation rate was 0.22%. Trends for claims and compensation rates decreased over the study period. In high volume procedures, a low compensation rate was detected for excision of tonsils and adenoids, Caesarean section and extracapsular cataract operations using the phacoemulsification technique. A high compensation rate was detected for primary prosthetic replacement of the hip and knee joints and decompression of spinal cord and nerve roots. Unreasonable injury (death or permanent deterioration of health) was compensated in 2.4 per 100,000 procedures. Conclusions Register data research in a no-fault patient insurance system revealed a claims rate of 6 per 1,000 procedures and compensation rate of 2 per 1,000 procedures. A decreasing trend in both rates over the study period was detected. Different surgical procedures exhibit varying claims and compensation rates. Supplementary Information The online version contains supplementary material available at 10.1186/s13037-023-00353-0.

Project description:OBJECTIVE:To examine the prevalence, costs and trends (2010-2014) for 21 low-value inpatient procedures in a privately insured Australian patient cohort. DESIGN:We developed indicators for 21 low-value procedures from evidence-based lists such as Choosing Wisely, and applied them to a claims data set of hospital admissions. We used narrow and broad indicators where multiple low-value procedure definitions exist. SETTING AND PARTICIPANTS:A cohort of 376 354 patients who claimed for an inpatient service from any of 13 insurance funds in calendar years 2010-2014; approximately 7% of the privately insured Australian population. MAIN OUTCOME MEASURES:Counts and proportions of low-value procedures in 2014, and relative change between 2010 and 2014. We also report both the Medicare (Australian government) and the private insurance financial contributions to these low-value admissions. RESULTS:Of the 14 662 patients with admissions for at least 1 of the 21 procedures in 2014, 20.8%-32.0% were low-value using the narrow and broad indicators, respectively. Of the 21 procedures, admissions for knee arthroscopy were highest in both the volume and the proportion that were low-value (1607-2956; 44.4%-81.7%).Seven low-value procedures decreased in use between 2010 and 2014, while admissions for low-value percutaneous coronary interventions and inpatient intravitreal injections increased (51% and 8%, respectively).For this sample, we estimated 2014 Medicare contributions for admissions with low-value procedures to be between $A1.8 and $A2.9 million, and total charges between $A12.4 and $A22.7 million. CONCLUSIONS:The Australian federal government is currently reviewing low-value healthcare covered by Medicare and private health insurers. Estimates from this study can provide crucial baseline data and inform design and assessment of policy strategies within the Australian private healthcare sector aimed at curtailing the high volume and/or proportions of low-value procedures.