Claims-based Measures Research Articles

Adherence and usage patterns of inhaled corticosteroids-long-acting beta-agonists by using inhaler-monitoring technology.

Background: Results of previous research indicate that adherence to prescribed inhaled corticosteroid-long-acting beta2-agonist (ICS-LABA) asthma controller medications is suboptimal, yet actual daily-use patterns are unclear and may be influenced by regimen complexity or dosing frequency. Objective: To investigate real-world use of asthma medications by using inhaler sensors for the ICS-LABA controllers: twice-daily fluticasone propionate (FP) plus salmeterol (SAL) and once-daily fluticasone furoate (FF) plus vilanterol (VI); and albuterol rescue medication. Methods: This longitudinal, two-phase, observational study included adults with asthma-prescribed FP-SAL (phase I) or FF-VI (phase II), and albuterol metered-dose inhalers. The participants completed baseline and follow-up surveys, and used clip-on inhaler sensors to monitor real-time inhaler use over the 6-month study period. Pharmacy claims data for the 6-month follow-up period were used to assess refills of ICS-LABA and albuterol inhalers. Results: Patients who used twice-daily FP-SAL received a sufficient dose (≥2 actuations/day) approximately one third of the time, those on once-daily FF-VI received a sufficient dose (≥1 actuation/day) ∼60% of the time. Patients who used once-daily FF-VI were more likely to take their medication as prescribed versus those who used twice-daily FP-SAL. There were no significant differences in the percentage of albuterol-free days (FP-SAL, 68.06% [n = 241]; FF-VI, 72.67% [n = 127]; p = 0.230). Exploratory outcomes are reported in this article's Online Supplemental Material. Claims-based measures of adherence were higher than sensor-based measures, hence claims data may have overestimated adherence, whereas sensors may have more accurately measured patients' medication use. Conclusion: These data supported the use of inhaler sensors as tools to directly and accurately measure ICS-LABA adherence and rescue medication use, and the adherence benefits of once-daily versus twice-daily ICS-LABA regimens.

Variation in the degree of concentration of prescription opioid utilization using different measures

BackgroundStudies of opioid prescribing patterns have shown that a small percentage of prescribers are responsible for a large portion of total prescriptions. There is some evidence that prescription opioid use may be similarly concentrated, but patterns may differ by population and choice of opioid utilization measure. The objective of this study is to determine what proportion of prescription opioid utilization was attributable to the top utilizers among Medicaid beneficiaries and how this proportion varies by the measure used. MethodsWe analyzed 2008–2009 Medicaid claims data from 14 states and created three claims-based measures of aggregate opioid utilization: total number of annual prescriptions, total annual milligrams morphine equivalent, and total days supply. We tested two versions of the total days supply measure: one assuming consecutive use and the other assuming concurrent use of prescriptions. ResultsThe top 20 % of prescription opioid users accounted for 66 % of prescriptions, 82–85 % of days supply depending on assumption, and 90 % of morphine milligram equivalents (MME). The degree to which prescription opioid utilization was concentrated among the top 20 % of users varied minimally across states. ConclusionsA small percentage of prescription opioid users account for a large share of prescription opioid use. Policy efforts should use metrics pertaining to days supply and total MME, not numbers of prescriptions, to more efficiently target heavy utilization. Policies targeting high-dose prescribing should consider the different ways that overlapping prescriptions may be taken.

Claims identification of patients with severe cancer-related symptoms.

The goal of this study was to establish a claims-based mechanism for identifying patients with metastatic non-small cell lung cancer (mNSCLC) and high levels of patient-reported cancer-related symptoms who could benefit from engagement with health care programs. A cross-sectional survey of patients with mNSCLC was conducted from July 2017 to May 2018. Surveys were mailed to patients who were within 3 months of cancer treatment and enrolled in a Medicare Advantage health plan. Pain, fatigue, and sleep disturbance were measured using the Patient-Reported Outcomes Measurement Information System. Depression was assessed using the Patient Health Questionnaire-2. Medical claims were linked to survey results to identify comorbidities and assess preindex health care resource utilization. Cluster analysis was used to differentiate patients based on patient-reported pain interference, pain intensity, depression, and sleep disturbance. Logistic regression was used to identify claims-based measures associated with more severe symptoms. For 698 respondents, 2 distinct symptom clusters were identified: a less severe (38.4%) cluster and a more severe (61.6%) cluster. Patients in the more severe cluster were younger, were more frequently dually eligible for Medicare and Medicaid, and more frequently had prescription fills for opioids. Claims-based factors associated with the more severe cluster included 2 or more 30-day fills for opioids in the prior 6 months, age younger than 75 years, depression diagnosis or antidepressants, bone metastases, and pain-related outpatient visits. The claims-based factors associated with the severe symptom cluster can enable identification of patients with mNSCLC who could benefit from clinical outreach programs to enhance the care and support provided to these patients.

1131-P: Clinical, Demographic, Provider, and Formulary Characteristics Associated with Initiation of a SGLT2i or GLP-1RA among Medicare Beneficiaries with Type 2 Diabetes on Prior Oral Agents: A Cohort Study

Using 2014-2016 claims data from a random 5% sample of Medicare beneficiaries, we sought to describe characteristics associated with initiation of a SLT2i or GLP-1RA, focusing on measures of plan generosity (e.g., the number of drugs covered with no or minimal copays). Our cohort included beneficiaries with type 2 diabetes who had at least 1 prescription claim for an oral diabetes medication in the 180 days before the start of follow-up (1/1/2015). To identify new initiators of these agents, we excluded beneficiaries who filled any insulin, SGLT2i or GLP-1RA prior to the index date or who died before 12/31/16. We developed multivariable logistic regression models controlling a priori for age, race, region, prior medication use, claims-based measures of diabetes severity and clinical comorbidities, provider specialty and measures of formulary generosity. Of 103,112 eligible beneficiaries (mean age 73, 57% female), 5,076 (4.9%) initiated a SGLT2i or GLP-1RA between 1/1/15-12/31/16. After adjusting for all baseline characteristics, beneficiaries in plans covering 2 or more target drugs in preferred tiers (i.e., 1-3) were more likely (adjusted odds ratio [aOR] 1.18 [95% CI, 1.05 to 1.32]) to initiate one of these medications than patients enrolled in plans covering 0 drugs. Other characteristics associated with higher odds included: age<65 (aOR 1.78 [95% CI, 1.64 to 1.92]), diabetic retinopathy (aOR 1.19 [95% CI, 1.08 to 1.31], obesity (aOR 1.49 [95% CI, 1.38 to 1.62) and having baseline medications prescribed by an endocrinologist (aOR 1.68 [95% CI, 1.49 to 1.89]. Black race (aOR 0.66 [95% CI, 0.60 to 0.73]) and CKD (aOR 0.84 [95% CI, 0.76 to 0.92]) were associated with lower risk of initiation. After controlling for a large number of prespecified covariates, Medicare beneficiaries with type 2 diabetes with more generous prescription drug coverage were more likely to initiate a SGLT2i or GLP-1RA. Disclosure J. Luo: Consultant; Self; Alosa Health. I. Hernandez: None. N. Gabriel: None. W.F. Gellad: None. Funding National Center for Advancing Translational Sciences (KL2TR001856)

Reliability and correlations among quality measures for lung, breast, and colorectal cancer.

2073 Background: Alternative payment models for oncology seek to improve quality and reduce spending. Yet the ability to measure high-quality care across oncology practices remains uncertain. We characterized quality of care for oncology practices using registry and claims-based measures of processes, utilization, end-of-life care, and survival and assessed correlations of practice-level performance across measure type and cancers. Methods: Using SEER-Medicare data, we studied individuals with newly diagnosed lung (N = 95,635), breast (N = 78,736), or colorectal (CRC, N = 51,385) cancers in 2010-2015 treated in oncology practices with ≥20 patients (502, 492, and 347 practices, respectively). We measured receipt of guideline-recommended treatment and surveillance (processes), hospitalizations or emergency department visits during 6-month chemotherapy episodes (utilization), care intensity in the last month of life (EOL), and 12-month survival (lung and CRC only). We calculated summary process, utilization, and EOL measures for each patient (number of measures met divided by the number for which the patient was eligible). We used hierarchical linear models with practice-level random effects to estimate summary measures and survival for each practice. We calculated practice-level reliability (a measurement’s reproducibility) for each measure based on the between-measure variance, within-measure variance, and sample size. Results: Few practices had ≥20 patients eligible for most measures (38%, 37%, and 31% of practices had ≥20 patients for any lung, breast, and CRC measures, respectively). Measure reliability was low. Only 13%, 7%, and 20% of measures for lung, breast, and CRC, respectively, had a median reliability across practices ≥0.7. Among practices with ≥20 patients with summary measures of each type within cancer, correlations across measure types were low (all correlation coefficients (r)≤0.21 except a weak correlation of the CRC process summary measure with 1-year CRC survival, r = 0.38, p < 0.001). Summary process measures were minimally or not correlated across cancer type (lung, breast, CRC; all correlation coefficients ≤0.16). Conclusions: Claims-based measures of care processes, utilization, EOL care, and survival are limited by small numbers of fee-for-service Medicare patients across practices, even after pooling 6 years of data. Measures have poor reliability and are poorly correlated across measure or cancer type. Additional research is needed to identify reliable quality measures for practice-level alternate payment models.

Development and validation of a claims-based measure of abortion services.

Data on abortion services are critical for monitoring trends in access and utilization, evaluating policies, and examining a wide range of research questions. Accurate and timely data, however, can be difficult to obtain for abortion services. Oregon is one of several states that use state funds to finance abortion services in their Medicaid programmes. Oregon's Medicaid programme contracts with managed care plans that receive global budgets to provide care. Abortion services, however, must be billed directly to the state through fee-for-service (FFS) billing to ensure that federal funds are not used. In this study, we identify possible abortions using Medicaid insurance claims data from Oregon and categorize identified abortions as high, medium, or low confidence according to convergent validity analysis of FFS billing. We used individually linked Medicaid eligibility and claims data from women ages 15 to 44 enrolled in Oregon's Medicaid programme from 2008 to 2013. Abortion-related Medicaid claims were identified and categorized based on diagnosis, procedure, and drug codes. These categories were assessed for convergent validity by examining FFS billing for possible abortions to women enrolled in managed care plans. In total, 23 763 possible abortions obtained by 18 518 women were classified with high (n = 21 450), medium (n = 562), and low (n = 1751) confidence. Among managed care abortions, more than 99% of high confidence abortions were billed on an FFS basis compared with 72% of medium confidence and <1% of low confidence abortions. The majority of high confidence abortions were to urban-residing (89%) white (73%) women. Research on abortion services using insurance claims has important implications for women's health care and public health policy. A high-quality claims-based measure can facilitate monitoring the provision of abortion services within health systems and evaluation of initiatives to increase equitable abortion access.

Tobacco use by pregnant Medicaid beneficiaries: Validating a claims-based measure in Oregon.

In Oregon, more than 4 in 5 pregnant women who smoke are covered by Medicaid. Although birth certificate data for smoking during pregnancy are not accessible in a timely manner, Medicaid claims data are available monthly and provide person-level data. This study utilized an individually linked database of Medicaid claims and birth certificate data to compare the prevalence of tobacco use diagnosis codes in Medicaid claims data to self-reported smoking during pregnancy reported on birth certificates. We computed the sensitivity and specificity of Medicaid claims data to ascertain tobacco use during pregnancy compared to self-report on linked birth certificates. Using logistic regression models, we also examined demographic, prenatal care, and behavioral health factors that predicted agreement between claims and birth certificates. From 2008 to 2013, 17.9% of women with Medicaid births reported smoking during pregnancy on birth certificates compared to 3.8% of non-Medicaid births. Tobacco-related claims during pregnancy were present for 12.6% of Medicaid births. Overall agreement between claims and birth certificates rose from 87.0% in 2008 to 90.2% in 2013; sensitivity rose from 43.0% to 62.2%. Sensitivity was lowest for Hispanic women and highest for White women, and declined as maternal education increased. Sensitivity was 33.9 percentage points higher for women with any mental illness diagnosis and 27.3 percentage points higher for women with any substance use disorder diagnosis. Specificity was greater than 95% in all years. Medicaid claims data may help in surveillance of maternal smoking rates and assessment of smoking cessation programs for female Medicaid beneficiaries of reproductive age.

Development and Validation of an Administrative Claims-based Measure for All-cause 30-day Risk-standardized Readmissions After Discharge From Inpatient Psychiatric Facilities

The objective of this study was to develop and test a measure that estimates unplanned, 30-day, all-cause risk-standardized readmission rates (RSRRs) after inpatient psychiatric facility (IPF) discharge. We established a retrospective cohort of adults with a principal diagnosis of psychiatric illness or dementia discharged from IPFs to nonacute care settings, using 2012-2013 Medicare fee-for-service claims data. All-cause unplanned readmissions within 3-30 days post-IPF discharge were assessed by constructing then validating a parsimonious logistic regression model of 56 risk factors (selected via empirical data, systematic literature review, clinical expert opinion) for readmission using bootstrapping. RSRRs were calculated from the ratio of predicted versus expected readmission rates for each IPF using hierarchical regression. Measure reliability and validity were assessed via multiple strategies. The measure development cohort included 716,174 admissions to 1679 IPFs and 149,475 (20.9%) readmissions. Most readmissions (>80%) had principal diagnoses of mood, schizoaffective or substance use disorders, delirium/dementia, infections or drug/substance poisoning. Facility RSRRs ranged from 11.0% to 35.4%. The risk adjustment model showed good calibration and moderate discrimination similar to other readmission risk models (c statistic 0.66). Sensitivity analyses solidified the risk modeling approach. The intraclass correlation coefficient of estimated IPF RSRRs was 0.78, indicating good reliability. The measure identified 8.3% of hospitals as having better and 13.4% as having worse RSRRs than the national readmission rate. The measure provides an assessment of facility-level quality and insight into risk factors useful for informing preventive interventions. The measure will be included in the Centers for Medicare and Medicaid Services (CMS) Inpatient Psychiatric Quality Reporting program in 2019.

Accuracy of Prescription Claims Data in Identifying Truly Nonadherent Patients.

Administrative claims data are increasingly used to identify nonadherent patients. This necessitates a comprehensive review and assessment of their accuracy in identifying nonadherent patients. To (a) compare administrative claims-based measures of adherence with nonadherence verified by patient interview; (b) determine if and to what extent patients classified as nonadherent based on prescription claims differ from patients classified as nonadherent based on interventions designed to gather multiple types of medication lists to compare against the prescription fill history; and (c) assess the various patient-reported reasons for nonadherence. A cross-sectional study was used to identify patients from the Southern Piedmont Community Care Network of North Carolina who were enrolled with Medicaid between January 1, 2012, and May 31, 2013, and were using prescription medications for 1 or more chronic conditions. Patients with more than a 30-day gap in refill history were identified using prescription claims and were interviewed by pharmacists to assess the reasons for nonadherence. Based on the patient-reported reasons for a gap in refill, patients were classified as interview-verified nonadherent patients or interview-verified adherent patients. The positive predictive value of prescription claims in identifying nonadherent patients was calculated, and descriptive statistics were reported. Characteristics of interview-verified nonadherent patients were compared with adherent patients using t-tests and chi-square statistics. 1,425 patients representing 2,936 patient-class of medication combinations were included in the final analysis. 824 (28.07%) of the 2,936 records that were flagged as nonadherent using claims analysis were confirmed as adherent during patient interviews. The positive predictive value of claims records in identifying nonadherent patients was 0.72. The 2 most common reasons for patients to be misclassified as nonadherent in claims data following self-report were discontinuation of medication on prescribers' directions (21.93%) and having an alternate channel for receiving the medication (6.13%). Among interview-verified nonadherent patients, side effects, patient beliefs, education, and socioeconomic barriers were the most common patient-reported reasons for gaps in refill. Prescription claims may underestimate adherence in patients. When interviewed directly by a pharmacist, most patients reported discontinuation of medication as per prescribers' directions. To determine the overall validity of prescription claims data, further analysis is required to assess its accuracy in identifying truly nonadherent patients among those who are identified as nonadherent by claims data. No outside funding supported this study. Glassberg and Wei were employees at Community Care of North Carolina when this research was conducted. Trygstad is an employee of Community Care of North Carolina; Robinson is an employee of Community Care of Southern Piedmont, a subsidiary of Community Care of North Carolina. The geographies, health care professionals, and subjects involved in the study were related to the care coordination work that Community Care of North Carolina was charged with implementing through its informatics and subject matter expertise assistance provided to these local entities to augment primary care activities. Farley has received funding from the Agency for Healthcare Research and Quality, Centers for Disease Control and Prevention, American College of Clinical Pharmacy, the National Institutes of Health, and Community Care of North Carolina and has also received consulting funds from UCB. The other authors have nothing additional to report.

COMPARATIVE PERFORMANCE OF FOUR CLAIMS-BASED FRAILTY MEASURES IN THE UNITED STATES MEDICARE DATA

Background: There has been increasing effort to measure frailty in the United States Medicare data. The performance of claims-based frailty measures has not been compared. Methods: This retrospective cohort study included 2,326 community-dwelling Medicare beneficiaries who participated in the 2008 assessment of the Health and Retirement Study. The claims-based frailty measures developed by Davidoff, Faurot, Segal, and Kim were compared against clinical measures of frailty (gait speed, grip strength) using correlation coefficients and health outcomes (e.g., mortality, hospitalization, activities-of-daily-living disabilities) over 2 years using C-statistics. Results: The Davidoff, Faurot, Segal, and Kim indices were negatively correlated with gait speed (-0.19, -0.33, -0.37, and -0.37, respectively), but age and sex adjustment variably attenuated the correlation to -0.17, -0.22, -0.18, and -0.33, respectively. The corresponding correlation coefficients with grip strength were -0.17, -0.27, -0.35, and -0.24, which attenuated to -0.09, -0.14, -0.05, and -0.23 after age and sex adjustment, respectively. The models that included age, sex, and each of Davidoff, Faurot, Segal, and Kim indices showed C-statistics of 0.67, 0.71, 0.71, 0.75 for mortality (versus C-statistic for age and sex: 0.66); 0.59, 0.64, 0.63, 0.70 for hospitalization (versus C-statistic for age and sex: 0.58); and 0.64, 0.63, 0.63, 0.70 for activities-of-daily-living disabilities (versus C-statistic for age and sex: 0.61), respectively. Conclusions: The choice of a claims-based frailty measure results in a meaningful variation in the identification of frail older adults at high risk for adverse health outcomes. Claims-based frailty measures that included demographic variables offer limited risk adjustment beyond age and sex.

Validating Reports of Chronic Conditions in the Medicare CAHPS Survey.

The Medicare Consumer Assessment of Healthcare Providers and Systems (CAHPS) survey includes items about chronic conditions, health history, and self-rated health. Questions remain about the concordance between patient reports and administrative sources on questions related to health history. To validate CAHPS measures of chronic conditions against claims-based measures from the Medicare Chronic Conditions Warehouse (CCW). We linked CAHPS responses from 301,050 fee-for-service Medicare beneficiaries in 2010-2012 with summaries of their claims in the CCW and identified nearest equivalent measures of conditions across sources. We calculated sensitivities and specificities for conditions and estimated regression models to assess the effects of patient characteristics on the sensitivity. The sensitivity of CAHPS measures differed across conditions, ranging from 0.513 for history of stroke to 0.773 for history of cancer. Sensitivity was generally lower for older beneficiaries, those reporting good self-rated health, and those with fair or poor mental health. Specificity was 0.904 or greater for all conditions, up to 0.961 for stroke. Despite difference in timeframes and definitions of conditions, the measured sensitivities demonstrated reasonable validity. Variation in sensitivity is consistent with covariates that either directly measure health severity within a diagnosis or can be construed as a proxy for severity of illness.

A Comprehensive Review of Methods to Measure Oral Oncolytic Dose Intensity Using Retrospective Data.

Understanding the real-world use of oral oncolytics is essential to assess drug effectiveness. Retrospective analyses using medical and pharmacy claims data allow observation of drug use patterns and health outcomes. However, studies of medication adherence to oral oncolytics may not be sufficient in characterizing exposure because they typically measure refill frequency, not the administered dose or dose changes. Patients who appear fully adherent by traditional measures may be receiving different doses and experiencing differing effectiveness. Relative dose intensity (RDI) is a measure that has been used for intravenous drugs to capture the amount of a particular chemotherapeutic agent administered per unit of time (dose intensity), expressed as the fraction of the amount recommended in evidence-based guidelines. Such a measure would be useful for real-world studies of comparative effectiveness to characterize patient exposure to oral oncolytics. To identify studies that used administrative claims data to measure real-world oral oncolytic dose intensity, RDI, or similar constructs. Two health sciences librarians conducted a literature search (PubMed, January 1, 1809-February 6, 2018) including terms in each of the following concept areas: oncology drugs, dosage, and retrospective data sources. At least 2 reviewers scanned each title and abstract of publications retrieved from PubMed. Abstracts that indicated the study reported dose or related concepts and oral oncolytics using retrospective data sources were marked for full-text review. During full-text review, papers were excluded if they did not study oral oncolytics (i.e., only described intravenous chemotherapy); if they did not report drug dosage; or if the study was not retrospective. Resulting studies were included for full-text data extraction. Of the 1,640 publications returned from the search, 41 were marked for full-text review. Full-text review established that 17 studies addressed a concept related to dose of oral oncolytics using retrospective data. Twenty-four studies were excluded: 11 did not measure dose; 9 did not study oral oncolytics; and 4 were not retrospective studies. Among the 17 articles marked for extraction, 5 articles reported dose intensity or RDI using medical records or electronic health record (EHR) data. This study reveals not only the need for a claims-based measure of dose intensity for oral oncolytics, but also provides a basis for the development of such a measure based on previous EHR-based studies. While several claims data studies have characterized oral oncolytic dosing and duration, we found that no studies combined these dimensions into a single measure such as dose intensity. Methods using EHR data may be translatable to a claims data study. Future research is needed to develop and validate such measures. Novartis Pharmaceuticals provided funding for this study and is a manufacturer of oral onalytics, which is under study in this article. Arcona and Zacker are employees of Novartis. Slejko reports grants from PhRMA, PhRMA Foundation, and Takeda Pharmaceuticals and consulting fees from Pfizer, outside the submitted work. Stuart reports consulting fees from the University of Maryland during the study. The other authors have nothing to disclose. The preliminary findings of this study were presented in a poster at AMCP Nexus 2018, October 22-25, 2018, in Orlando, FL.

ICD-10 Coding Will Challenge Researchers: Caution and Collaboration may Reduce Measurement Error and Improve Comparability Over Time.

The October 1, 2015 US health care diagnosis and procedure codes update, from the 9th to 10th version of the International Classification of Diseases (ICD), abruptly changed the structure, number, and diversity of codes in health care administrative data. Translation from ICD-9 to ICD-10 risks introducing artificial changes in claims-based measures of health and health services. Using published ICD-9 and ICD-10 definitions and translation software, we explored discontinuity in common diagnoses to quantify measurement changes introduced by the upgrade. Using 100% Medicare inpatient data, 2012-2015, we calculated the quarterly frequency of condition-specific diagnoses on hospital discharge records. Years 2012-2014 provided baseline frequencies and historic, annual fourth-quarter changes. We compared these to fourth quarter of 2015, the first months after ICD-10 adoption, using Centers for Medicare and Medicaid Services Chronic Conditions Data Warehouse (CCW) ICD-9 and ICD-10 definitions and other commonly used definitions sets. Discontinuities of recorded CCW-defined conditions in fourth quarter of 2015 varied widely. For example, compared with diagnosis appearance in 2014 fourth quarter, in 2015 we saw a sudden 3.2% increase in chronic lung disease and a 1.8% decrease in depression; frequency of acute myocardial infarction was stable. Using published software to translate Charlson-Deyo and Elixhauser conditions yielded discontinuities ranging from -8.9% to +10.9%. ICD-9 to ICD-10 translations do not always align, producing discontinuity over time. This may compromise ICD-based measurements and risk-adjustment. To address the challenge, we propose a public resource for researchers to share discovered discontinuities introduced by ICD-10 adoption and the solutions they develop.

Effectiveness of Medicare cost-sharing elimination for Cancer screening on utilization

BackgroundThe Patient Protection and Affordable Care Act (ACA) eliminated the cost-sharing requirement for several preventive cancer screenings. This study examined the cancer screening utilization of mammogram, Pap smear and colonoscopy in Medicare fee-for-service (FFS) under the ACA.MethodsThe primary data were the 2007–2013 Medicare Current Beneficiary Survey linked to FFS claims. The effect of the cost-sharing removal on the probability of receiving a preventive cancer screening test was estimated using a logistic regression, separately for each screening test, adjusting for the complex survey design. The model was also separately estimated for different socioeconomic and race/ethnic groups.The study sample included beneficiaries with Part B coverage for the entire calendar year, excluding beneficiaries in Medicaid or Medicare Advantage plans. Beneficiaries with a claims-documented or self-reported history of targeted cancers, who were likely to have diagnostic tests or have surveillance screenings were excluded.The screening measures were constructed separately following Medicare coverage and U.S. Preventive Services Task Force (USPSTF) recommendations. We measured the screening utilization outcome drawing from claims data, as well as using the self-reported survey data.ResultsAfter the cost-sharing removal policy, we found no statistically significant difference in a beneficiary’s probability of receiving a colonoscopy (transition period: OR = 1.08, 95% CI = 0.90–1.29; post-policy period: OR = 1.08, 95% CI = 0.83–1.42), a mammogram (transition period: OR = 1.03, 95% CI = 0.91–1.17; post-policy period: OR = 1.07, 95% CI = 0.88–1.30), or a biennial Pap smear (transition period: OR = 0.87, 95% CI = 0.69–1.09; post-policy period: OR = 0.72, 95% CI = 0.51–1.03) in claims-based measures following Medicare coverage. Similarly, we found null effects of the policy change on utilization of colonoscopy among enrollees 50–75 years old, biennial mammograms by women 50–74, and triennial Pap smear tests among women 21–65 in claims-based measures according to USPSTF. The findings from survey-based measures were consistent with the estimates from claims-based measures, except that the use of Pap smear declined since 2011. Further, the policy change did not increase utilization in patients with disadvantaged socioeconomic characteristics. Yet the disparate patterns in adjusted screening rates by socioeconomic status and race/ethnicity persisted over time.ConclusionsRemoving out-of-pocket costs for screenings did not provide enough incentives to increase the screening rates among Medicare beneficiaries.

Aggregating Claims Data Across Payers: Approaches, Challenges, and Lessons Learned From the Comprehensive Primary Care Initiative.

The Comprehensive Primary Care (CPC) initiative fueled the emergence of new organizational alliances and financial commitments among payers and primary care practices to use data for performance improvement. In most regions of the country, practices received separate confidential feedback reports of claims-based measures from multiple payers, which varied in content and provided an incomplete picture of a practice's patient panel. Over CPC's last few years, participating payers in several regions resisted the tendency to guard data as a proprietary asset, instead working collaboratively to produce aggregated performance feedback for practices. Aggregating claims data across payers is a potential game changer in improving practice performance because doing so potentially makes the data more accessible, comprehensive, and useful. Understanding lessons learned and key challenges can help other initiatives that are aggregating claims or clinical data across payers for primary care practices or other types of providers.

The Effect of Frailty on Discharge Location for Medicare Beneficiaries After Acute Stroke

ObjectiveTo examine the effect of frailty on poststroke discharge location with respect to stroke severity and create a risk-adjusted model for understanding the effects of frailty on discharge to an inpatient rehabilitation facility. DesignRetrospective cohort. SettingA 2014 5% Medicare sample. ParticipantsPatients hospitalized for a first-time acute ischemic stroke (N=7258). InterventionsNot applicable. Main Outcome MeasuresA prehospitalization 6-month baseline was used to calculate a frailty score. Logistic regression to predict odds of discharge to inpatient rehabilitation was used to calculate for 3 levels of baseline frailty, controlling for patient demographics, stroke severity, and comorbidities. ResultsAbout 1603 patients were discharged to inpatient rehabilitation. Patients who were nonfrail (odds ratio [OR] 1.716; 95% confidence interval [95% CI], 1.463-2.013) or prefrail (OR 1.519; 95% CI, 1.296-1.779) were more likely to be discharged to inpatient rehabilitation. The final logistic regression model had a C-statistic of 0.63. Most of the patients discharged to inpatient rehabilitation were nonfrail (44.2%) and had moderate strokes (38.9%). Individuals who were frail and suffered a moderate (OR 0.78; 95% CI, 0.558-1.091) or severe stroke (OR 0.509; 95% CI, 0.358-0.721) were less likely to be discharged to an inpatient rehabilitation facility. ConclusionsA lack of a claims-based measure for prestroke functional ability makes it difficult to understand discharge decision-making patterns for individuals’ poststroke. Prestroke frailty was found to have a significant effect on predicating inpatient rehabilitation discharge after an acute stroke when controlling for stroke severity, comorbidities, and age. Further investigation is warranted to examine differences in rehabilitation utilization based on frailty and to quantify the effect of rehabilitation on frailty status in individuals poststroke.

Mortality by a proxy performance status as defined by a claims-based measure for disability status in older patients with newly diagnosed multiple myeloma in the United States.

We applied a claims-based definition of disability status as a proxy for performance status (PS) and examined associations between PS and mortality in a population-based cohort of older US adults with multiple myeloma (MM). We identified older (≥66 years) Medicare beneficiaries diagnosed with MM January 1, 2008-December 31, 2011, who began first-line therapy in the study period (through December 31, 2012). We estimated predicted probability of poor PS for each patient at initiation of each line up to fourth-line therapy, classified as poor (predicted probability ≥0.11) or good (<0.11) PS, and examined mortality. Crude overall survival was estimated using the Kaplan-Meier method with log-rank test for survival comparison between PS groups. Cox proportional hazards models evaluated the association between poor PS and mortality risk, adjusted for baseline characteristics by lines of therapy. Of 12,547 patients, 5841, 2372, and 819 initiated second-, third-, and fourth-line in the study period. Poor PS proportions were 16.6%, 21.8%, 18.4%, and 18.2% at each line. Crude overall survival was worse for poor PS patients across lines (P < 0.01 for each). Adjusted hazards ratios (95% CI) of mortality for patients with poor versus good PS were 1.28 (1.18-1.40), first-line; 1.55 (1.36-1.77), second-line; 1.35 (1.10-1.65), third-line; 1.22 (0.84-1.76), fourth-line. The claims-based prediction model for disability status performed as expected as a proxy for PS in older Medicare patients with MM. PS was an independent risk factor for mortality. Further studies assessing the effect of PS on mortality by therapies are warranted.

New approaches to measuring the comprehensiveness of primary care physicians.

To develop claims-based measures of comprehensiveness of primary care physicians (PCPs) and summarize their associations with health care utilization and cost. A total of 5359 PCPs caring for over 1 million Medicare fee-for-service beneficiaries from 1404 practices. We developed Medicare claims-based measures of physician comprehensiveness (involvement in patient conditions and new problem management) and used a previously developed range of services measure. We analyzed the association of PCPs' comprehensiveness in 2013 with their beneficiaries' emergency department, hospitalizations rates, and ambulatory care-sensitive condition (ACSC) admissions (each per 1000 beneficiaries per year), and Medicare expenditures (per beneficiary per month) in 2014, adjusting for beneficiary, physician, practice, and market characteristics, and clustering. Each measure varied across PCPs and had low correlation with the other measures-as intended, they capture different aspects of comprehensiveness. For patients whose PCPs' comprehensiveness score was at the 75th vs 25th percentile (more vs less comprehensive), patients had lower service use (P<0.05) in one or more measures: involvement with patient conditions: total Medicare expenditures, -$17.4 (-2.2 percent); hospitalizations, -5.5 (-1.9 percent); emergency department (ED) visits, -16.3 (-2.4 percent); new problem management: total Medicare expenditures, -$13.3 (-1.7 percent); hospitalizations, -7.0 (-2.4 percent); ED visits, -19.7 (-2.9 percent); range of services: ED visits, -17.1 (-2.5 percent). There were no significant associations between the comprehensiveness measures and ACSC admission rates. These measures demonstrate strong content and predictive validity and reliability. Medicare beneficiaries of PCPs providing more comprehensive care had lower hospitalization rates, ED visits, and total Medicare expenditures.

Claims-based measures of continuity of care have non-linear associations with health: data linkage study.

BackgroundContinuity of care (CoC) is considered a central element of good primary care and is often measured using medical claims data. Possible values of CoC depend on the number of claims which is related to health status. This study investigated the relationships between CoC and health status and risk of emergency hospitalisation.MethodsHealth insurance claims for consultations with general practitioners (GPs) in the 24 months following entry to the 45 and Up Study were used to calculate usual provider continuity (UPC) and the Continuity of Care Index (CoC Index). Relationships of CoC with number of claims, self-rated health and emergency hospitalisation were investigated using descriptive statistics and logistic regression models.ResultsBoth measures of CoC were strongly related to number of claims and to measures of health status, which were in turn highly associated. Multivariable logistic regression models showed a weak positive relationship between CoC and odds of emergency hospitalisation for those with CoC less than 1, while individuals with perfect CoC had significantly lower odds of hospitalisation compared to all other categories of CoC. However, analyses stratified by, or adjusting for, number of claims showed no clear associations between CoC and risk of hospitalisation.ConclusionsThe pattern of association between CoC categories and emergency hospitalisation was non-linear and was confounded by the effect of number of claims. Future studies should apply caution in using claims-based measures of CoC as a continuous variable or employing an arbitrary cut-point, and should adjust for number of claims.

An Approach to Identify Delivery of Palliative Radiation Therapy Using Health Care Claims Data: A Proof-of-Concept Application of a Visual Analytics Tool.

There is limited information on the use of data visualization tools for health services research applications. We provide a proof-of-concept application that focuses on claims-based measures of palliative radiation therapy. We investigate whether a guided, data-driven investigation contributes information for subsequent statistical analysis and algorithm development. This retrospective cohort study used linked registry and claims data on men who were diagnosed with stage IV M0 or stage IV M1b prostate cancer between 2005 and 2009, with associated claims from 2005 through 2010, and receiving radiation therapy. Preprocessing of data was accomplished by using EventFlow software to investigate longitudinal patterns in claims for radiation therapy in the 13 months after cancer diagnosis. Guided by results from EventFlow, we developed descriptive statistics to investigate the length of radiation therapy, use of bone metastasis coding, and mortality between M1b and M0 patients. A total of 1,151 patients met the inclusion criteria. Taking advantage of the novel aggregation capability of EventFlow, we observed differences in the length of radiation therapy and the use of bone metastasis coding between men with (M1b) and without (M0) a diagnosis of bone metastasis. Seventy-nine percent of M1b patients received radiation for a duration ≤ 4 weeks, which suggested palliative radiation (to the bone). Seventy-six percent of M0 patients received radiation for ≥ 6 weeks, which suggested radiation to the prostate. Mortality was higher among those who received a shorter duration of radiation therapy compared with those who received a longer duration of therapy. Use of EventFlow, followed by statistical analysis of the linked registry and claims data, identified useful components of a claims-based measure of radiation to the bone.