Health care costs associated with unrecognized progression to late-stage kidney disease.

OBJECTIVES: Many individuals with chronic kidney disease (CKD) are undiagnosed or unaware of the disease and at risk of not receiving services to manage their condition and of "crashing" into dialysis. Past studies report higher health care costs among patients with delayed nephrology care and suboptimal dialysis initiation, but they are limited because they focused on patients undergoing dialysis and did not evaluate costs associated with unrecognized disease for patients "upstream," or patients with late-stage CKD. We compared costs for patients with unrecognized progression to late-stage (stages G4 and G5) CKD and end-stage kidney disease (ESKD) with costs for individuals with prior CKD recognition. STUDY DESIGN: Retrospective study of commercial, Medicare Advantage, and Medicare fee-for-service enrollees 40 years and older. METHODS: Using deidentified claims data, we identified 2 groups of patients with late-stage CKD or ESKD, one group with prior evidence of CKD diagnosis and the other without, and compared total and CKD-related costs in the first year following late-stage diagnosis between the 2 groups. We used generalized linear models to determine the association between prior recognition and costs and used recycled predictions to calculate predicted costs. RESULTS: Total and CKD-related costs were 26% and 19% higher, respectively, for patients without prior diagnosis compared with those with prior recognition. Total costs were higher both for unrecognized patients with ESKD and unrecognized patients with late-stage disease. CONCLUSIONS: Our findings indicate that costs associated with undiagnosed CKD extend to patients not yet requiring dialysis and highlight potential savings from earlier disease detection and management.

Designing Equitable Health Care Outreach Programs From Machine Learning Patient Risk Scores.

There is growing interest in ensuring equity and guarding against bias in the use of risk scores produced by machine learning and artificial intelligence models. Risk scores are used to select patients who will receive outreach and support. Inappropriate use of risk scores, however, can perpetuate disparities. Commonly advocated solutions to improve equity are nontrivial to implement and may not pass legal scrutiny. In this article, we introduce pragmatic tools that support better use of risk scores for more equitable outreach programs. Our model output charts allow modeling and care management teams to see the equity consequences of different threshold choices and to select the optimal risk thresholds to trigger outreach. For best results, as with any health equity tool, we recommend that these charts be used by a diverse team and shared with relevant stakeholders.

Procedure for Organizing a Post-FDA-approval Evaluation of Antidepressants.

PURPOSE: The study reports the construction of a cohort used to study the effectiveness of antidepressants. METHODS: The cohort includes experiences of 3,678,082 patients with depression in the United States on antidepressants between January 1, 2001, and December 31, 2018. A total of 10,221,145 antidepressant treatment episodes were analyzed. Patients who had no utilization of health services for at least two years, or who had died, were excluded from the analysis. Follow-up was passive, automatic, and collated from fragmented clinical services of diverse providers. RESULTS: The average follow-up was 2.93 years, resulting in 15,096,055 person-years of data. The mean age of the cohort was 46.54 years (standard deviation of 17.48) at first prescription of antidepressant, which was also the enrollment event (16.92% were over 65 years), and most were female (69.36%). In 10,221,145 episodes, within the first 100 days of start of the episode, 4,729,372 (46.3%) continued their treatment, 1,306,338 (12.8%) switched to another medication, 3,586,156 (35.1%) discontinued their medication, and 599,279 (5.9%) augmented their treatment. CONCLUSIONS: We present a procedure for constructing a cohort using claims data. A surrogate measure for self-reported symptom remission based on the patterns of use of antidepressants has been proposed to address the absence of outcomes in claims. Future studies can use the procedures described here to organize studies of the comparative effectiveness of antidepressants.

Effectiveness of common antidepressants: a post market release study.

BACKGROUND: This study summarizes the experiences of patients, who have multiple comorbidities, with 15 mono-treated antidepressants. METHODS: This is a retrospective, observational, matched case control study. The cohort was organized using claims data available through OptumLabs for depressed patients treated with antidepressants between January 1, 2001 and December 31, 2018. The cohort included patients from all states within United States of America. The analysis focused on 3,678,082 patients with major depression who had 10,221,145 antidepressant treatments. Using the robust, and large predictors of remission, and propensity to prescribe an antidepressant, the study created 16,770 subgroups of patients. The study reports the remission rate for the antidepressants within the subgroups. The overall impact of antidepressant on remission was calculated as the common odds ratio across the strata. FINDINGS: The study accurately modelled clinicians' prescription patterns (cross-validated Area under the Receiver Operating Curve, AROC, of 82.0%, varied from 77% to 90%) and patients' remission (cross-validated AROC of 72.0%, varied from 69.5% to 78%). In different strata, contrary to published randomized studies, remission rates differed significantly and antidepressants were not equally effective. For example, in age and gender subgroups, the best antidepressant had an average remission rate of 50.78%, 1.5 times higher than the average antidepressant (30.30% remission rate) and 20 times higher than the worst antidepressant. The Breslow-Day chi-square test for homogeneity showed that across strata a homogenous common odds-ratio did not exist (alpha<0.0001). Therefore, the choice of the optimal antidepressant depended on the strata defined by the patient's medical history. INTERPRETATION: Study findings may not be appropriate for specific patients. To help clinicians assess the transferability of study findings to specific patient, the web site http://hi.gmu.edu/ad assesses the patient's medical history, finds similar cases in our data, and recommends an antidepressant based on the experience of remission in our data. Patients can share this site's recommendations with their clinicians, who can then assess the appropriateness of the recommendations. FUNDING: This project was funded by the Robert Wood Johnson foundation grant #76786. The development of related web site was supported by grant 247-02-20 from Virginia's Commonwealth Health Research Board.

Deep neural network models for identifying incident dementia using claims and EHR datasets.

This study investigates the use of deep learning methods to improve the accuracy of a predictive model for dementia, and compares the performance to a traditional machine learning model. With sufficient accuracy the model can be deployed as a first round screening tool for clinical follow-up including neurological examination, neuropsychological testing, imaging and recruitment to clinical trials. Seven cohorts with two years of data, three to eight years prior to index date, and an incident cohort were created. Four trained models for each cohort, boosted trees, feed forward network, recurrent neural network and recurrent neural network with pre-trained weights, were constructed and their performance compared using validation and test data. The incident model had an AUC of 94.4% and F1 score of 54.1%. Eight years removed from index date the AUC and F1 scores were 80.7% and 25.6%, respectively. The results for the remaining cohorts were between these ranges. Deep learning models can result in significant improvement in performance but come at a cost in terms of run times and hardware requirements. The results of the model at index date indicate that this modeling can be effective at stratifying patients at risk of dementia. At this time, the inability to sustain this quality at longer lead times is more an issue of data availability and quality rather than one of algorithm choices.

Predicting Onset of Dementia Using Clinical Notes and Machine Learning: Case-Control Study.

BACKGROUND: Clinical trials need efficient tools to assist in recruiting patients at risk of Alzheimer disease and related dementias (ADRD). Early detection can also assist patients with financial planning for long-term care. Clinical notes are an important, underutilized source of information in machine learning models because of the cost of collection and complexity of analysis. OBJECTIVE: This study aimed to investigate the use of deidentified clinical notes from multiple hospital systems collected over 10 years to augment retrospective machine learning models of the risk of developing ADRD. METHODS: We used 2 years of data to predict the future outcome of ADRD onset. Clinical notes are provided in a deidentified format with specific terms and sentiments. Terms in clinical notes are embedded into a 100-dimensional vector space to identify clusters of related terms and abbreviations that differ across hospital systems and individual clinicians. RESULTS: When using clinical notes, the area under the curve (AUC) improved from 0.85 to 0.94, and positive predictive value (PPV) increased from 45.07% (25,245/56,018) to 68.32% (14,153/20,717) in the model at disease onset. Models with clinical notes improved in both AUC and PPV in years 3-6 when notes' volume was largest; results are mixed in years 7 and 8 with the smallest cohorts. CONCLUSIONS: Although clinical notes helped in the short term, the presence of ADRD symptomatic terms years earlier than onset adds evidence to other studies that clinicians undercode diagnoses of ADRD. De-identified clinical notes increase the accuracy of risk models. Clinical notes collected across multiple hospital systems via natural language processing can be merged using postprocessing techniques to aid model accuracy.

Assessment of Potentially Inappropriate Prescribing of Opioid Analgesics Requiring Prior Opioid Tolerance.

Importance: Opioid-tolerant only (OTO) medications, such as transmucosal immediate-release fentanyl products and certain extended-release opioid analgesics, require prior opioid tolerance for safe use, as patients without tolerance may be at increased risk of overdose. Studies using insurance claims have found that many patients initiating these medications do not appear to be opioid tolerant. Objectives: To measure prevalence of opioid tolerance in patients initiating OTO medications and to determine whether linked electronic health record (EHR) data contribute evidence of opioid tolerance not found in insurance claims data. Design, Setting, and Participants: This retrospective cohort study used a national database of deidentified longitudinal health information, including medical and pharmacy claims, insurance enrollment, and EHR data, from January 1, 2007, to December 31, 2016. Data included 131 756 US residents with at least 183 days of continuous enrollment in commercial or Medicare Advantage insurance (including medical and pharmacy benefits) who had received an OTO medication and who had no inpatient stays in the 30 days prior to starting an OTO medication; of these, 20 044 individuals had linked EHR data within the prior 183 days. Data were analyzed from July 1, 2017, to August 31, 2018. Exposures: Initiating an OTO medication. Main Outcomes and Measures: Prior opioid tolerance demonstrated through pharmacy fills or EHR data on prescriptions written. Results: Among 153 385 OTO use episodes identified, 89 029 (58.0%) occurred among women, 62 900 (41.0%) occurred among patients with Medicare Advantage insurance, 39 394 (25.7%) occurred in the Midwest, 17 366 (11.3%) occurred in the Northeast, 73 316 (47.8%) occurred in the South, and 23 309 (15.2%) occurred in the West. Less than half of use episodes (73 117 episodes [47.7%]) involved patients with evidence in claims data of opioid tolerance prior to initiating therapy with an OTO medication, including 31 392 of 101 676 episodes (30.9%) involving transdermal fentanyl, 1561 of 2440 episodes (64.0%) involving transmucosal fentanyl, 36 596 of 43 559 episodes (84.0%) involving extended-release oxycodone, and 3568 of 5710 episodes (62.5%) involving extended-release hydromorphone. Among 20 044 OTO use episodes with linked EHR and claims data, less than 1% of OTO episodes identified in claims had evidence of opioid tolerance in structured EHR data that was not present in claims data (108 episodes [0.5%]). After limiting the sample to OTO episodes identified in claims with a matching OTO prescription within 14 days in the structured EHR data, only 40 of 939 episodes (4.0%) occurred among patients with evidence of tolerance that was not present in claims data. Conclusions and Relevance: This cohort study found that most patients initiating OTO medications did not have evidence of prior opioid tolerance, suggesting they were at increased risk of opioid-related harms, including fatal overdose. Data from EHRs did not contribute substantial additional evidence of opioid tolerance beyond the data found in prescription claims. Future research is needed to understand the clinical rationale behind these observed prescribing patterns and to quantify the risk of harm to patients associated with potentially inappropriate prescribing.

Machine learning models to predict onset of dementia: A label learning approach.

INTRODUCTION: The study objective was to build a machine learning model to predict incident mild cognitive impairment, Alzheimer's Disease, and related dementias from structured data using administrative and electronic health record sources. METHODS: A cohort of patients (n = 121,907) and controls (n = 5,307,045) was created for modeling using data within 2 years of patient's incident diagnosis date. Additional cohorts 3-8 years removed from index data are used for prediction. Training cohorts were matched on age, gender, index year, and utilization, and fit with a gradient boosting machine, lightGBM. RESULTS: Incident 2-year model quality on a held-out test set had a sensitivity of 47% and area-under-the-curve of 87%. In the 3-year model, the learned labels achieved 24% (71%), which dropped to 15% (72%) in year 8. DISCUSSION: The ability of the model to discriminate incident cases of dementia implies that it can be a worthwhile tool to screen patients for trial recruitment and patient management.

Identifying incident dementia by applying machine learning to a very large administrative claims dataset.

Alzheimer's disease and related dementias (ADRD) are highly prevalent conditions, and prior efforts to develop predictive models have relied on demographic and clinical risk factors using traditional logistical regression methods. We hypothesized that machine-learning algorithms using administrative claims data may represent a novel approach to predicting ADRD. Using a national de-identified dataset of more than 125 million patients including over 10,000 clinical, pharmaceutical, and demographic variables, we developed a cohort to train a machine learning model to predict ADRD 4-5 years in advance. The Lasso algorithm selected a 50-variable model with an area under the curve (AUC) of 0.693. Top diagnosis codes in the model were memory loss (780.93), Parkinson's disease (332.0), mild cognitive impairment (331.83) and bipolar disorder (296.80), and top pharmacy codes were psychoactive drugs. Machine learning algorithms can rapidly develop predictive models for ADRD with massive datasets, without requiring hypothesis-driven feature engineering.

Development and Validation of HealthImpact: An Incident Diabetes Prediction Model Based on Administrative Data.

OBJECTIVE: To develop and validate a model of incident type 2 diabetes based solely on administrative data. DATA SOURCES/STUDY SETTING: Optum Labs Data Warehouse (OLDW), a national commercial administrative dataset. STUDY DESIGN: HealthImpact model was developed and internally validated using nested case-control study design; n = 473,049 in training cohort and n = 303,025 in internal validation cohort. HealthImpact was externally validated in 2,000,000 adults followed prospectively for 3 years. Only adults ≥18 years were included. DATA COLLECTION/EXTRACTION METHODS: Patients with incident diabetes were identified using HEDIS rules. Control subjects were sampled from patients without diabetes. Medical and pharmacy claims data collected over 3 years prior to index date were used to build the model variables. PRINCIPAL FINDINGS: HealthImpact, scored 0-100, has 48 variables with c-statistic 0.80815. We identified HealthImpact threshold of 90 as identifying patients at high risk of incident diabetes. HealthImpact had excellent discrimination in external validation cohort (c-statistic 0.8171). The sensitivity, specificity, positive predictive value, and negative predictive value of HealthImpact >90 for new diagnosis of diabetes within 3 years were 32.35, 94.92, 22.25, and 96.90 percent, respectively. CONCLUSIONS: HealthImpact is an efficient and effective method of risk stratification for incident diabetes that is not predicated on patient-provided information or laboratory tests.

All-cause and incremental per patient per year cost associated with chronic hepatitis C virus and associated liver complications in the United States: a managed care perspective.

BACKGROUND: Approximately 3.2-3.9 million U.S. residents are infected with the hepatitis C virus (HCV). Total annual costs (direct and indirect) in the United States for HCV were estimated to be $5.46 billion in 1997, and direct medical costs have been predicted to increase to $10.7 billion for the 10-year period from 2010 through 2019, due in part to the increasing number of HCV patients developing advanced liver disease (AdvLD). OBJECTIVE: To quantify in a sample of commercially insured enrollees (a) total per patient per year (PPPY) all-cause costs to the payer, overall and by the stage of liver disease, for patients diagnosed with HCV; and (b) incremental all-cause costs for patients diagnosed with HCV relative to a matched non-HCV cohort. METHODS: This retrospective, matched cohort study included patients aged at least 18 years and with at least 6 months of continuous enrollment in a large managed care organization (MCO) claims database from July 1, 2001, through March 31, 2010. Patients with a diagnosis of HCV (ICD-9-CM codes 070.54, 070.70) were identified and stratified into those with and without AdvLD, defined as decompensated cirrhosis (ICD-9-CM codes 070.44, 070.71, 348.3x, 456.0, 456.1, 456.2x, 572.2, 572.3, 572.4, 782.4, 789.59); hepatocellular carcinoma (HCC, ICD-9-CM code 155); or liver transplant (ICD-9-CM codes V42.7, 50.5 or CPT codes 47135, 47136). For patients without AdvLD, the index date was the first HCV diagnosis date observed at least 6 months after the first enrollment date, and at least 6 months of continuous enrollment after the index date were required. HCV patients without AdvLD were stratified into those with and without compensated cirrhosis (ICD-9-CM codes 571.2, 571.5, 571.6). For patients with AdvLD, the index date was the date of the first AdvLD diagnosis observed at least 6 months after the first enrollment date, and at least 1 day of enrollment after the index date was required. Cases were matched in an approximate 1:10 ratio to comparison patients without an HCV diagnosis or AdvLD diagnosis who met all other inclusion criteria based on gender, age, hospital referral region state, pre-index health care costs, alcoholism, human immunodeficiency virus/acquired immune deficiency syndrome (HIV/AIDS), and a modified Charlson Comorbidity Index. For the HCV and comparison patient cohorts, PPPY all-cause costs to the payer were calculated as total allowed charges summed across all patients divided by total patient-days of follow-up for the cohort, multiplied by 365, inflation-normalized to 2009 dollars. Because the calculation of PPPY cost generated a single value for each cohort, bootstrapping was used to generate descriptive statistics. Incremental PPPY costs for HCV patients relative to non-HCV patients were calculated as between-group differences in PPPY costs. T-tests for independent samples were used to compare costs between case and comparison cohorts. RESULTS: A total of 34,597 patients diagnosed with HCV, 78.0% with HCV without AdvLD, 4.4% with compensated cirrhosis, 12.3% with decompensated cirrhosis, 2.8% with HCC, and 2.6% with liver transplant, were matched to 330,435 comparison patients. Mean (SD) age of all HCV cases was 49.9 (8.5) years; 61.7% were male. Incremental mean (SD) PPPY costs in 2009 dollars for all HCV patients relative to comparison patients were $ 9,681 ($176) PPPY. Incremental PPPY costs were $5,870 ($157) and $5,330 ($491) for HCV patients without liver disease and with compensated cirrhosis, respectively. Incremental PPPY costs for patients with AdvLD were $27,845 ($ 965) for decompensated cirrhosis, $43,671 ($2,588) for HCC, and $ 93,609 ($4,482) for transplant. Incremental prescription drug costs, including the cost of antiviral drugs, were $2,739 ($37) for HCV patients overall, $2,659 ($41) for HCV without liver involvement, and $3,102 ($157) for HCV with compensated cirrhosis. These between-group differences were statistically significant at P<0.001. CONCLUSIONS: Based on a retrospective analysis of data from a large, MCO claims database, patients diagnosed with HCV had annual all-cause medical costs that were almost twice as high as those of enrollees without a diagnosis of HCV. Health care costs increased dramatically with AdvLD. Data from this study may help MCOs project future HCV costs and facilitate planning for HCV patient management efforts.