Individual Claims Data Research Articles

A Longitudinal Multivariable Analysis: State Policies and Opioid Dispensing in Medicare Beneficiaries Undergoing Surgery.

States have implemented policies to decrease clinically unnecessary opioid prescribing, but few studies have examined how state policies affect opioid dispensing rate trends for surgical patients. To examine trends in the perioperative opioid dispensing rates for fee-for-service Medicare beneficiaries and the effects of select state policies. A retrospective cohort study using 2006 to 2018 Medicare claims data for individuals undergoing surgical procedures for which opioid analgesic treatment is common. State policies mandating prescription drug monitoring program (PDMP; PDMP policies) use, initial opioid prescription duration limit (duration limit policies), and mandated continuing medical education (CME; CME pain policies) on pain management. Opioid dispensing rates, days' supply, and the daily morphine milligram equivalent dose (MMED). The percentage of Medicare beneficiaries dispensed opioids in the perioperative period increased from 2007 to 2018; MMED and days' supply decreased over the same period, with significant variation by age, sex, and race. None of the three state policies affected the likelihood of Medicare beneficiaries being dispensed perioperative opioids. However, CME pain policies and duration limit policies were associated with decreased days' supply and decreased MMED in the several years following implementation, respectively. While we observed a slight increase in the rate of Medicare beneficiaries dispensed opioids perioperatively and a substantial decrease in MMED and days' supply for those receiving opioids, state policies examined had relatively modest effects on the main measures. Our findings suggest that these state policies may have a limited impact on opioid dispensing for a patient population that is commonly dispensed opioid analgesics to help control surgical pain, and as a result may have little direct effect on clinical outcomes for this population. Changes in opioid dispensing for this population may be the result of broader societal trends than such state policies.

Development of a daily predictive model for the exacerbation of chronic obstructive pulmonary disease

Acute exacerbation (AE) of chronic obstructive pulmonary disease (COPD) compromises health status; it increases disease progression and the risk of future exacerbations. We aimed to develop a model to predict COPD exacerbation. We merged the Korean COPD subgroup study (KOCOSS) dataset with nationwide medical claims data, information regarding weather, air pollution, and epidemic respiratory virus data. The Korean National Health and Nutrition Examination Survey (KNHANES) dataset was used for validation. Several machine learning methods were employed to increase the predictive power. The development dataset consisted of 590 COPD patients enrolled in the KOCOSS cohort; these were randomly divided into training and internal validation subsets on the basis of the individual claims data. We selected demographic and spirometry data, medications for COPD and hospital visit for AE, air pollution data and meteorological data, and influenza virus data as contributing factors for the final model. Six machine learning and logistic regression tools were used to evaluate the performance of the model. A light gradient boosted machine (LGBM) afforded the best predictive power with an area under the curve (AUC) of 0.935 and an F1 score of 0.653. Similar favorable predictive performance was observed for the 2151 individuals in the external validation dataset. Daily prediction of the COPD exacerbation risk may help patients to rapidly assess their risk of exacerbation and will guide them to take appropriate intervention in advance. This might lead to reduction of the personal and socioeconomic burdens associated with exacerbation.

#5673 ESTIMATING AND VALIDATING MEASURES OF CKD PREVALENCE AND INCIDENCE USING OBSERVATIONAL AND CLAIMS DATA IN PERSONS AGED 70+

Abstract Background and Aims Estimating prevalence and incidence of chronic kidney disease (CKD) based on claims data often underlies bias and limitations due to uncertainty of diagnostic validity. We investigated CKD prevalence and incidence over time using data from a community-dwelling cohort of individuals aged 70+ linked to individual claims data. We assessed the diagnostic validity of the claims-based compared with the eGFR-based CKD definition. Method We assessed CKD prevalence and incidence in two data sources: 1) eGFR values of participants aged 70+ (n = 2,069) of the Berlin Initiative Study (BIS), including five biennial study visits from 2009-2019, and 2) claims data of BIS participants matched on person-level. Using eGFR, we defined CKD prevalence as eGFR <60 ml/min/1.73m² calculated with the creatinine-based CKD-EPI (2009) equation. For each study visit, prevalent CKD cases were considered as incident if eGFR was ≥60 ml/min/1.73m² in the respective previous study visit. In claims data, in- and outpatient ICD-10 diagnoses (N18.3, N18.4, N18.5, N18.8, N18.9, N19) in the year preceding and/or following the study visit date were used to determine CKD prevalence. Incidence was defined as prevalent cases without a diagnosis in the year preceding the first CKD diagnosis. The denominator for incidence was defined as all persons at risk, i.e., who were not prevalent in a respective previous study visit and still under observation. We assessed the diagnostic validity of claims-based CKD prevalence and incidence using eGFR data as a reference. Analyses were stratified by sex. Results Between 2009 and 2019, CKD prevalence increased from 0.42 to 0.51 based on eGFR and from 0.30 to 0.43 based on claims data. Incidence varied between 0.16–0.22 using eGFR and 0.07–0.13 using BIS claims data. There was no clear time trend for incidence in any data source. Diagnostic validity of claims-based CKD prevalence and incidence detection is displayed in Table 1. Sensitivity for prevalence increased while other indicators remained stable over time. Detection of incident cases showed lower sensitivity (0.18–0.30 vs. 0.58–0.70) and PPVs (0.40–0.60 vs. 0.79–0.83) compared to prevalence, respectively. Prevalence and incidence based on claims data were slightly higher in males compared to females (Figure 1). Conclusion Using the eGFR-based or claims-based definition of CKD we found an increase in prevalence between 2009 and 2019. This might be due to the introduction of the KDIGO guidelines in 2012. During the same time, CKD incidence rates were stable. Comparing the claims-based with eGFR-based CKD definition, we found a sensitivity of 0.58-0.70 for prevalence and 0.18-0.30 for incidence, indicating that there is a risk of underdetection of CKD when using solely claims data, in particular for incident cases.

Association of Outpatient Behavioral Health Treatment With Medical and Pharmacy Costs in the First 27 Months Following a New Behavioral Health Diagnosis in the US

Outpatient behavioral health treatment (OPBHT) is an effective treatment for behavioral health conditions (BHCs) that may also be associated with improved medical health outcomes, but evidence regarding the cost-effectiveness of OPBHT across a large population has not been established. To investigate whether individuals newly diagnosed with a BHC who used OPBHT incurred lower medical and pharmacy costs over 15 and 27 months of follow-up compared with those not using OPBHT. This retrospective cohort study of commercially insured individuals in the US was conducted using administrative insurance claims data for individuals newly diagnosed with 1 or more BHCs between January 1, 2017, and December 31, 2018. Data were examined using a 12-month period before BHC diagnosis and 15- and 27-month follow-up periods. Participants included individuals aged 1 to 64 years who received any OPBHT with or without behavioral medication or who did not receive OPBHT or behavioral medication in the 15 months following diagnosis. Data were analyzed from May to October 2021. Receipt of OPBHT both as a dichotomous variable and categorized by number of OPBHT visits. The main outcome was the association between OPBHT treatment and 15- and 27-month medical and pharmacy costs, assessed using a generalized linear regression model with γ distribution, controlling for potential confounders. The study population included 203 401 individuals, of whom most were male (52%), White, non-Hispanic (75%), and 18 to 64 years of age (67%); 22% had at least 1 chronic medical condition in addition to a BHC. Having 1 or more OPBHT visits was associated with lower adjusted mean per-member, per-month medical and pharmacy costs across follow-up over 15 months (no OPBHT: $686 [95% CI, $619-$760]; ≥1 OPBHT: $571 [95% CI, $515-$632]; P < .001) and 27 months (no OPBHT: $464 [95% CI, $393-$549]; ≥1 OPBHT: $391 [95% CI, $331-$462]; P < .001). Furthermore, almost all doses of OPBHT across the 15 months following diagnosis were associated with lower costs compared with no OPBHT. In this cohort study, medical cost savings were associated with OPBHT among patients newly diagnosed with a BHC in a large, commercially insured population. The findings suggest that promoting and optimizing OPBHT may be associated with reduced overall medical spending among patients with BHCs.

The Impact of the COVID-19 Pandemic on Health Care Utilization Among Insured Individuals With Common Chronic Conditions.

Objective:The COVID pandemic has had a significant impact on the US health care system. Our primary objective was to understand the impact of the COVID pandemic on non–COVID-related health care utilization among insured individuals with chronic conditions. Our secondary objective was to examine the differential impact by individual characteristics.Main Data Source:Medical and pharmacy claims data for individuals enrolled in a large insurer across the United States.Research Design:A retrospective and repeated cross-sectional study. Overall and condition-specific health care utilization and cost metrics in (1) March 1 to June 15 and (2) June 16 to September 30, 2020 were compared with the same months during 2016–2019.Subjects:Members of all ages with a diagnosis of diabetes, cardiovascular disease, or chronic kidney disease with commercial or Medicare Advantage insurance.Results:Most non–COVID-related health care utilization decreased drastically on March 1 to June 15, 2020 [odds ratio (OR) range across condition-specific tests: 0.55–0.69; incidence rate ratio (IRR) range for hospitalization/emergency department (ED) visit/outpatient visit: 0.65–0.77] but returned to closer to pre-COVID levels by June 16 to September 30, 2020 [OR range across condition-specific tests: 0.93–1.08; IRR range for hospitalization/ED visit/outpatient visit: 0.77–0.97]. Our study found an enormous increase in telehealth use on March 1 to June 15, 2020 (90–170 times prepandemic levels). A differential impact was observed by age, sex, region of residence, and insurance type.Implications:Further investigation is needed to assess the impact of these changes in health care utilization on long-term health outcomes.

Dementia in a Hearing-impaired Population According to Hearing Aid Use: A Nationwide Population-based Study in Korea.

Background:Hearing loss is considered a potentially modifiable risk factor for dementia. However, the effect of use of a hearing aid on the development of dementia has not been clearly established. We aimed to assess the incidence of dementia in hearing-impaired individuals according to hearing aid use in a nationwide population-based cohort study with matched controls.Methods:This was a retrospective, longitudinal, cohort study of South Korean national claims data for individuals newly registered with hearing disabilities (better ear ≥40 dBHL with worse ear ≥80 dBHL, or better ear ≥60 dBHL) between 2004 and 2008. The hearing aid cohort comprised individuals who received hearing aid subsidies from the National Health Insurance within a year from disability registration. The comparison cohort comprised individuals without a record of a hearing aid claim during the study period after 1:1 matching for audiologic and sociodemographic factors with the hearing aid cohort. The occurrence of dementia was followed up until 2018.Results:Each cohort comprised 8780 individuals. Overall incidence of dementia in the hearing aid and comparison cohorts were 156.0 and 184.5 per 10,000 person-years, respectively (incidence rate ratio = 0.85, 95% confidence interval 0.79–0.91). In a multivariable analysis of the whole study populations, hearing aid use (hazard ratio = 0.75, 95% confidence interval 0.70–0.81) attenuated the risk of dementia.Conclusions:Dementia incidence in individuals with hearing disabilities was lower in hearing aid users than that in nonusers. Hearing rehabilitation with hearing aids should be encouraged for individuals with hearing loss.

Collective reserving using individual claims data

The aim of this paper is to operationalize claims reserving based on individual claims data. We design a modeling architecture that is based on six different neural networks. Each network is a separate module that serves a certain modeling purpose. We apply our architecture to individual claims data and predict their settlement processes on a monthly time grid. A proof of concept is provided by benchmarking the resulting claims reserves with the ones received from the classical chain-ladder method which uses much coarser (aggregated) data.

Lung Function Monitoring in Allogenic Hematopoietic Cell Transplant (A-HCT) Patients

Background Bronchiolitis obliterans syndrome (BOS) is an obstructive airway disease of the lungs that affects 5.5 to 14.8% of allogeneic hematopoietic stem cell transplant (allo-HSCT) patients. One of its defining clinical manifestations is a decline in Forced expiratory volume in 1 second (FEV1) in the presence of airflow obstruction, as quantified by spirometry and other lung function tests. Prompt diagnosis through lung function testing may improve outcomes, but lung function testing rates following allo-HSCT have not been examined. This study analyzed lung function testing rates before and after allo-HSCT in the US. Methods Data sources for this longitudinal retrospective study were the IQVIA PharMetrics Plus commercial database and the Medicare Limited Dataset, both with enrollment, demographic and medical claims data for individuals in the US. Study patients had at least one claim with a Common Procedure Terminology (CPT) or International Classification of Diseases (ICD), 9th and 10th revision, procedure code, or an International Code of Disease (ICD-9 or ICD-10) for allo-HSCT, following a 6-month period with no evidence of transplantation. Commercially insured patients were limited to age 0y to &amp;lt;65y because those ≥65y are likely to have primary coverage through Medicare. The study period was 1/1/06 - 9/30/18 for commercially insured, and 1/1/10 -12/31/18 for Medicare patients. Lung function tests were identified using CPT and ICD procedure codes. Outcome measures were the percent of patients receiving testing each year, and the mean annual number of tests per patient. All measures were calculated for patients receiving at least one lung function test of any kind, and for specific tests: spirometry, lung diffusion capacity, and plethysmography/lung function volume. McNemar tests were conducted to assess whether the percent of patients with any lung function testing was significantly different in year 3, compared with years 1 and 2. ANOVA tests assessed whether mean testing rates were significantly different in the three post-transplant years. All tests of significance were conducted at an alpha level of .05. Results Among 2,187 commercially insured and 1,864 Medicare patients, a minority (41%) received at least one lung function test in the first year following transplantation. Among those who survived more than one year, the percentage with any lung function test declined over time, with 42% (Commercial) and 40% (Medicare) receiving any tests in year 2 and 31% (Commercial) and 26% (Medicare) in year 3 (Commercial p=.61; Medicare p&amp;lt;.05). The proportions of patients receiving specific tests further declined over the study period, including spirometry, with 41% (Commercial) and 34% (Medicare) of all patients receiving spirometry in year 1 and 38% (Commercial) and 36% (Medicare) in year 2, but only 28% (Commercial) and 24% (Medicare) receiving spirometry tests in year 3 (Commercial p&amp;lt;.05; Medicare p=.12). The mean annual number of tests administered per patient also declined. Rates of spirometry, which were 0.73±1.1 (Commercial) and 0.55±0.9 (Medicare) in year 1 and 0.66±1.2 (Commercial) and 0.55±0.9 (Medicare) in year 2 declined to 0.46±1.1 (Commercial) and 0.34±0.7 (Medicare) in year 3 (Commercial p&amp;lt;.05; Medicare p&amp;lt;.05). Rates of lung diffusion capacity and plethysmography/lung function volume testing also declined, with significantly fewer tests per patient in year 3, compared with years 1 and 2 (all p&amp;lt;0.05) (Table). Conclusion Morbidity and mortality from BOS remain high in allo-HSCT patients, but only a minority of patients receive lung function testing in the first year, and even fewer receive testing in year 3 following allo-HSCT. The mean number of tests per patient is significantly lower for all patients in year 3, compared with years 1 and 2, suggesting that annual testing frequency also decreases over time. This decrease is evident in commercially insured patients in the second year post-transplant, and is greatest for all patients in the third year post transplant, when patients remain at risk of BOS. Declines in testing may lead to a delayed or missed diagnosis of BOS. Increased and sustained monitoring of allo-HSCT patients could lead to earlier detection of BOS and earlier intervention and treatment. Disclosures No relevant conflicts of interest to declare.

Conducting a national burden of disease study in South Korea: from past to present

Abstract South Korea has been measuring its own burden of disease since the early 2000s and has developed a methodology for its calculation. Since 2012, South Korea has been carrying out disease burden research through the Research and Development Project, supported by the Ministry of Health and Welfare. Because the Korean health insurance system covers the entire population of the country, the Korean National Burden of Disease (KNBD) study is based on individual claims data. Disease prevalence, incidence, and mortality rates, which are the epidemiologic indicators used to calculate burden of disease, are based on actual data, and the incidence-based approach is used for Years Lived with disability (YLD), Years of Life Lost (YLL), and Disability Adjusted Life Year (DALY). The KNBD study is conducted for 260 diseases and injuries and subdivided into sub-national levels. Disability weights (DW) are calculated to reflect the sociocultural characteristics of a disease. For cause-specific DWs, we conducted a self-administered web-based survey for medical doctors and students. Using a ranking method, we calculated cause-specific DWs. For this reason, KNBD study results cannot be directly compared with GBD results. The major findings of the KNBD study are as follows. In Korea, the DALY rate (per 100,000 people) steadily increased for all income levels. The higher the income, the lower the burden of disease; low-income people had a higher burden of disease than high-income people. However, thyroid cancer, refraction and accommodation disorders had a high burden of disease in the high-income group. Growth in the YLD rate was faster than in the YLL rate. In 2016 the leading causes of disease burden were diabetes mellitus, low back pain, and chronic obstructive pulmonary disease. We will present South Korea's disease burden calculation methodology and its results and share the efforts to calculate disease burden in other countries.

Collective Reserving using Individual Claims Data

The aim of this paper is to operationalize claims reserving based on general insurance individual claims data. We design a modeling architecture that is based on six different neural networks. Each network is a separate module that serves a certain modeling purpose. We apply our architecture to individual claims data and predict their settlement processes on a monthly time grid. A proof of concept is provided by benchmarking the resulting claims reserves with the ones received from the classical chain-ladder method which uses much coarser (aggregated) data.

A machine learning approach for individual claims reserving in insurance

AbstractAccurate loss reserves are an important item in the financial statement of an insurance company and are mostly evaluated by macrolevel models with aggregate data in run‐off triangles. In recent years, a new set of literature has considered individual claims data and proposed parametric reserving models based on claim history profiles. In this paper, we present a nonparametric and flexible approach for estimating outstanding liabilities using all the covariates associated to the policy, its policyholder, and all the information received by the insurance company on the individual claims since its reporting date. We develop a machine learning–based method and explain how to build specific subsets of data for the machine learning algorithms to be trained and assessed on. The choice for a nonparametric model leads to new issues since the target variables (claim occurrence and claim severity) are right‐censored most of the time. The performance of our approach is evaluated by comparing the predictive values of the reserve estimates with their true values on simulated data. We compare our individual approach with the most used aggregate data method, namely, chain ladder, with respect to the bias and the variance of the estimates. We also provide a short real case study based on a Dutch loan insurance portfolio.

03:09 PM Abstract No. 142 Market shifts in transcatheter dialysis conduit procedures in the Medicare population: a 15-year national and state-level analysis

To evaluate trends in transcatheter hemodialysis conduit procedures in the Medicare population over a recent 15-year interval. Aggregate national claims data were extracted from CMS PSPS Master Files from 2001-2015 for hemodialysis conduit angiography and thrombectomy procedures. Utilization was stratified by billing specialty and site of service. Additionally, individual claims data from 2004-2015 CMS 5% Research Identifiable Files were used to assess state-level utilization. Utilization was normalized per 100,000 Medicare fee-for-service beneficiaries. From 2001-2015, hemodialysis conduit angiography utilization rates increased from 385 to 1,045 per 100,000 beneficiaries [compound annual growth rate (CAGR) +7.4%)], and thrombectomy rates increased from 114 to 168 (CAGR +2.8%). The CAGR for angiography was, by specialty, +1.5% for radiologists, +18.4% for surgeons, and +24.0% for nephrologists, and by site, +29.1% for office and +0.8% for hospital settings. Radiologists’ overall market share of angiography decreased from 81.5% in 37.0%. By combination of specialty and site of service, angiography utilization growth was greatest for nephrologists in the office (from 5 to 265) and surgeons in the office (0 to 128). The greatest decline was for radiologists in the hospital (299 to 205). At the state level, there was marked heterogeneity in dialysis angiography utilization in 2015 [0 (Wyoming) to 1,1,73 (Georgia)], temporal change in angiography utilization from 2004-2015 [CAGR -100.0% (Wyoming) to +19.9% (Nevada)], and radiologists’ 2015 market share [4.8% (Washington DC) to 100.0% (North Dakota)]. Nonetheless, radiologists’ market share decreased in 49 states, and in some states dramatically (e.g., in Nevada, from 100.0% in 2004 to 6.7% in 2015). Transcatheter dialysis conduit angiography utilization has grown substantially, and more so than thrombectomy. This growth has been accompanied by a drastic market shift from radiologists in the hospital to nephrologists and surgeons in the office. Despite wide geographic heterogeneity across the U.S., decreasing radiologist market share has been observed in nearly every state.

Claims data analyses unable to properly characterize the value of neurologists in epilepsy care.

To determine the association of a neurologist visit with health care use and cost outcomes for patients with incident epilepsy. Using health care claims data for individuals insured by United Healthcare from 2001 to 2016, we identified patients with incident epilepsy. The population was defined by an epilepsy/convulsion diagnosis code (ICD codes 345.xx/780.3x, G40.xx/R56.xx), an antiepileptic prescription filled within the succeeding 2 years, and neither criterion met in the 2 preceding years. Cases were defined as patients who had a neurologist encounter for epilepsy within 1 year after an incident diagnosis; a control cohort was constructed with propensity score matching. Primary outcomes were emergency room (ER) visits and hospitalizations for epilepsy. Secondary outcomes included measures of cost (epilepsy related, not epilepsy related, and antiepileptic drugs) and care escalation (including EEG evaluation and epilepsy surgery). After participant identification and propensity score matching, there were 3,400 cases and 3,400 controls. Epilepsy-related ER visits were more likely for cases than controls (year 1: 5.9% vs 2.3%, p < 0.001), as were hospitalizations (year 1: 2.1% vs 0.7%, p < 0.001). Total medical costs for epilepsy care, nonepilepsy care, and antiepileptic drugs were greater for cases (p ≤ 0.001). EEG evaluation and epilepsy surgery occurred more commonly for cases (p ≤ 0.001). Patients with epilepsy who visited a neurologist had greater subsequent health care use, medical costs, and care escalation than controls. This comparison using administrative claims is plausibly confounded by case disease severity, as suggested by higher nonepilepsy care costs. Linking patient-centered outcomes to claims data may provide the clinical resolution to assess care value within a heterogeneous population.

Cost of herpes zoster and herpes zoster-related complications among immunocompromised individuals

Herpes zoster (HZ) is a painful dermatomal rash caused by reactivation of latent varicella-zoster virus. The incidence of HZ is increased for immunocompromised (IC) individuals. The objective of this study is to assess the healthcare costs incurred by IC individuals who develop HZ with or without associated complications. We conducted a retrospective case-control study across the US over a 5-year period, based on health insurance claims data for individuals aged ≥50 years identified as IC by disease or immunosuppressive treatment. A cohort of 30,107 IC individuals who experienced HZ was matched to a cohort of 113,875 IC individuals without HZ. Average all-cause healthcare costs over 18 months were calculated and compared between IC individuals with and without HZ. In addition, the costs of HZ in IC individuals with HZ-related complications were compared to the costs of those with uncomplicated HZ. During the year following HZ onset, IC individuals with HZ had on average total unadjusted costs that were US$3879 higher than the controls. After adjusting costs, controlling for comorbidities and healthcare costs before the onset of HZ, the average annual costs for HZ cases and controls without HZ were similar. HZ-related complications led to increases in average adjusted annual costs compared to uncomplicated HZ ranging from US$612 for eye complications to US$4535 for neurologic complications. In conclusion, in IC individuals, episodes of HZ lead to substantially increased unadjusted annual healthcare costs. HZ-related complications add considerably to adjusted annual healthcare costs compared to uncomplicated HZ.

PHS113 - Geographic Variation in Utilization of Physician Services Among Older Adults Diagnosed with Non-Hodgkin Lymphoma

Geographic characteristics such as county-level socioeconomic status (SES) or accessibility of health services may play a role in behavioral decisions to visit a physician following a diagnosis of Non-Hodgkin Lymphoma (NHL). There is limited research to identify which geographic characteristics factor into the behavioral decisions. We analyzed linked registry and Medicare claims data for individuals diagnosed with NHL between 2000 and 2013. Outcome measures were the proportion and counts of: 1) all outpatient physician (OP) visits; 2) medical oncologist (MO) visits. The primary predictors were county-level SES and health services supply factors (HSSF). We utilized hierarchical generalized linear models (HGLM) with an offset for differential follow-up. We reported intraclass correlation coefficients (ICC) for the binary models. For both binary and count models, we reported median odds ratios (MORs) for the county-level random effect and interval odds ratios (IORs) for the county-level fixed effects. The sample included 72,973 individuals living in 590 counties. The average age was 78 years and the sample was 87% White, non-Hispanic. The median number of OP visits and MO visits per person was 19 and 14, respectively, over a median follow-up of 883 days. For the OP logistic HGLM: 1) the ICC was 3%; 2) the MOR was 1.3; 3) the IORs were [0.65; 1.21] for the SES index and [1.04; 1.93] for the HSSF index. For the MO logistic HGLM: 1) the ICC was 8%; 2) the MOR was 1.6; 3) the IORs were [0.4; 2.5] for the SES index and [0.35; 2.21] for the HSSF index. There was a larger role for county-level HSSF compared to the county-level SES in the HGLM count models. Among individuals diagnosed with NHL, there is a larger role for the county-level measure of health services accessibility compared to the county-level measure of socioeconomic status.

Enrichment of Claims Data with Official Causes of Death Using a Record Linkage with the Epidemiological Cancer Registry of North Rhine-Westphalia: Feasibility Study and Comparison of Procedures

Claims data of the statutory health insurance (SHI) are an important data source for the evaluation of cancer prevention programs. However, this source does not contain relevant information on cause of death. This study examined whether individual claims data can be enriched with data on the required cause of death using record linkage procedures with suitable external data sources. In the German pharmacoepidemiologic research database (GePaRD) we identified a sample of 25,528 deceased female residents of North Rhine Westphalia (NRW) who, according to GePaRD information, died between 2006 and 2013. Date and cause of all deaths among inhabitants of NRW since 2005 were available in the epidemiological cancer registry of NRW. In cooperation with 2 SHI companies, we tried to match each individual of the sample with a case of death in NRW and the corresponding cause of death using a probabilistic and, alternatively, a deterministic linkage procedure. Of the study sample, 94.72% were successfully matched by the probabilistic and 93.36% by the deterministic method. The probabilistic and the deterministic record linkage approach produced comparably high matching rates. Cases without matches are probably due to errors occurring at the stage of personal data entry. Given the lower technical efforts, the deterministic approach appears to be the method of choice for the enrichment of claims data with cause of death information from suitable external data sources in Germany.

Machine learning in individual claims reserving

Machine learning techniques make it feasible to calculate claims reserves on individual claims data. This paper illustrates how these techniques can be used by providing an explicit example in individual claims reserving.

Understanding Reporting Delay in General Insurance

The aim of this paper is to understand and to model claims arrival and reporting delay in general insurance. We calibrate two real individual claims data sets to the statistical model of Jewell and Norberg. One data set considers property insurance and the other one casualty insurance. For our analysis we slightly relax the model assumptions of Jewell allowing for non-stationarity so that the model is able to cope with trends and with seasonal patterns. The performance of our individual claims data prediction is compared to the prediction based on aggregate data using the Poisson chain-ladder method.

Healthcare resource use and costs associated with renal impairment in US patients with bone metastases from solid tumors.

Introduction During cancer progression, more than half of patients develop renal insufficiency, including chronic kidney disease. The primary and secondary objectives of this study were to estimate healthcare resource use and costs, respectively, associated with renal impairment in patients with bone metastases from solid tumors in the United States. Methods and materials This was a retrospective, observational cohort study conducted using administrative claims data for individuals with solid tumors and bone metastases. Control patients were matched to renal impairment patients using propensity scores (ratio up to 3:1) based on demographics, clinical characteristics, and baseline costs. Average per-patient per-year healthcare resource utilization and costs (total costs and cost components; 2013 dollars) were reported. Results In total, 2616 renal impairment patients were matched to 7211 control patients. Renal impairment patients had greater healthcare resource use compared with controls, including a greater mean number of hospital admissions (4.4 versus 2.1), longer average stay per hospital admission (7.4 versus 6.5 days), as well as greater mean number of physician office visits (22.9 versus 18.8), emergency department visits (3.1 versus 2.0), and hospital-based outpatient visits (18.8 versus 16.0) compared with control patients. Total costs were > $50,000 higher among renal impairment patients ($142,267 versus $88,839; P < 0.001), with hospital costs accounting for $72,557 for renal impairment patients, and $27,858 for control patients ( P < 0.001). Conclusion The healthcare resource use and costs associated with renal impairment in patients with bone metastases from solid tumors is high; efforts to reduce renal impairment in this population, including the potential avoidance of nephrotoxic agents, are warranted.

Direct costs in impaired glucose regulation: results from the population-based Heinz Nixdorf Recall study

ObjectiveFor the first time, this population-based study sought to analyze healthcare utilization and associated costs in people with normal fasting glycemia (NFG), impaired fasting glycemia (IFG), as well as previously...