Describing diversity of real world data sources in pharmacoepidemiologic studies: The DIVERSE scoping review.

PURPOSE: Real-world evidence (RWE) is increasingly used for medical regulatory decisions, yet concerns persist regarding its reproducibility and hence validity. This study addresses reproducibility challenges associated with diversity across real-world data sources (RWDS) repurposed for secondary use in pharmacoepidemiologic studies. Our aims were to identify, describe and characterize practices, recommendations and tools for collecting and reporting diversity across RWDSs, and explore how leveraging diversity could improve the quality of evidence. METHODS: In a preliminary phase, keywords for a literature search and selection tool were designed using a set of documents considered to be key by the coauthors. Next, a systematic search was conducted up to December 2021. The resulting documents were screened based on titles and abstracts, then based on full texts using the selection tool. Selected documents were reviewed to extract information on topics related to collecting and reporting RWDS diversity. A content analysis of the topics identified explicit and latent themes. RESULTS: Across the 91 selected documents, 12 topics were identified: 9 dimensions used to describe RWDS (organization accessing the data source, data originator, prompt, inclusion of population, content, data dictionary, time span, healthcare system and culture, and data quality), tools to summarize such dimensions, challenges, and opportunities arising from diversity. Thirty-six themes were identified within the dimensions. Opportunities arising from data diversity included multiple imputation and standardization. CONCLUSIONS: The dimensions identified across a large number of publications lay the foundation for formal guidance on reporting diversity of data sources to facilitate interpretation and enhance replicability and validity of RWE.

Variation in incidence rates of outcomes relevant to vaccine safety monitoring in a US commercially insured population by case identification algorithm.

BACKGROUND: Large health insurance claims databases can be used to estimate rates of rare safety outcomes. We measured incidence rates of rare outcomes that could be used to contextualize adverse events among people receiving pneumococcal vaccines in clinical trials or clinical practice. However, algorithms used to identify outcomes in administrative databases are subject to error. Using two algorithms for each outcome, we assessed the influence of algorithm choice on the rates of the outcomes. METHODS: We used closed administrative medical and pharmacy claims in the Healthcare Integrated Research DatabaseSM (HIRD) to construct a broad cohort of individuals less than 100 years old (i.e., the target cohort) and a trial-similar cohort of individuals resembling those potentially eligible for a vaccine clinical trial (e.g., for a pneumococcal vaccine). We stratified by age and sex and used specific and sensitive algorithms to estimate rates of 39 outcomes including cardiac/cerebrovascular, metabolic, allergic/autoimmune, neurological, and hematologic outcomes. Specific algorithms intended to reduce false positive errors, while sensitive algorithms intended to reduce false negative errors, thereby providing lower and upper bounds for the "true" rates. RESULTS: We followed approximately 40 million individuals in the target cohort for an average of 3 years. Of 39 outcomes, 14 (36 %) had a rate from the specific algorithm that was less than half the rate from the sensitive algorithm. Rates of cardiac/cerebrovascular outcomes were most consistent (mean ratio of rates from specific algorithms compared to rates from sensitive algorithms = 0.76), while the rates of neurological and hematologic outcomes were the least consistent (mean ratio of rates = 0.33 and 0.36, respectively). CONCLUSIONS: For many cardiac/cerebrovascular outcomes, rates were similar regardless of the algorithm. For other outcomes, rates varied substantially by algorithm. Using multiple algorithms to ascertain outcomes in claims data can be informative about the extent of uncertainty due to outcome misclassification.

Risk of Anaphylaxis Among New Users of GLP-1 Receptor Agonists: A Cohort Study.

OBJECTIVE: To assess risk of anaphylaxis among patients with type 2 diabetes mellitus who are initiating therapy with a glucagon-like peptide 1 receptor agonist (GLP-1 RA), with a focus on those starting lixisenatide therapy. RESEARCH DESIGN AND METHODS: A cohort study was conducted in three large, U.S. claims databases (2017-2021). Adult (aged ≥18 years) new users of a GLP-1 RA who had type 2 diabetes mellitus and ≥6 months enrollment in the database before GLP-1 RA initiation (start of follow-up) were included. GLP-1 RAs evaluated were lixisenatide, an insulin glargine/lixisenatide fixed-ratio combination (FRC), exenatide, liraglutide or insulin degludec/liraglutide FRC, dulaglutide, and semaglutide (injectable and oral). The first anaphylaxis event during follow-up was identified using a validated algorithm. Incidence rates (IRs) and 95% CIs were calculated within each medication cohort. The unadjusted IR ratio (IRR) comparing anaphylaxis rates in the lixisenatide cohort with all other GLP-1 RAs combined was analyzed post hoc. RESULTS: There were 696,089 new users with 456,612 person-years of exposure to GLP-1 RAs. Baseline demographics, comorbidities, and use of other prescription medications in the 6 months before the index date were similar across medication cohorts. IRs (95% CIs) per 10,000 person-years were 1.0 (0.0-5.6) for lixisenatide, 6.0 (3.6-9.4) for exenatide, 5.1 (3.7-7.0) for liraglutide, 3.9 (3.1-4.8) for dulaglutide, and 3.6 (2.6-4.9) for semaglutide. The IRR (95% CI) for the anaphylaxis rate for the lixisenatide cohort compared with the pooled other GLP-1 RA cohort was 0.24 (0.01-1.35). CONCLUSIONS: Anaphylaxis is rare with GLP-1 RAs. Lixisenatide is unlikely to confer higher risk of anaphylaxis than other GLP-1 RAs.

Risk of Death and Cardiovascular Events Following an Exacerbation of COPD: The EXACOS-CV US Study.

Purpose: This study estimated the magnitude and duration of risk of cardiovascular events and mortality following acute exacerbations of chronic obstructive pulmonary disease (AECOPD), and whether risks varied by number and severity of exacerbation in a commercially insured population in the United States. Methods: This was a retrospective cohort study of newly diagnosed COPD patients ≥40 years old in the Healthcare Integrated Research Database from 2012 to 2019. Patients experiencing exacerbations comprised the "exacerbation cohort". Moderate exacerbations were outpatient visits with contemporaneous antibiotic or glucocorticoid administration; severe exacerbations were emergency department visits or hospitalizations for AECOPD. Follow-up started on the exacerbation date. Distribution of time between diagnosis and first exacerbation was used to assign index dates to the "unexposed" cohort. Cox proportional hazards models estimated risks of a cardiovascular event or death following an exacerbation adjusted for medical and prescription history and stratified by follow-up time, type of cardiovascular event, exacerbation severity, and rank of exacerbation (first, second, or third). Results: Among 435,925 patients, 170,236 experienced ≥1 exacerbation. Risk of death was increased for 2 years following an exacerbation and was highest during the first 30 days (any exacerbation hazard ratio (HR)=1.79, 95% CI=1.58-2.04; moderate HR=1.22, 95% CI=1.04-1.43; severe HR=5.09, 95% CI=4.30-6.03). Risks of cardiovascular events were increased for 1 year following an AECOPD and highest in the first 30-days (any exacerbation HR=1.34, 95% CI=1.23-1.46; moderate HR=1.23 (95% CI 1.12-1.35); severe HR=1.93 (95% CI=1.67-2.22)). Each subsequent AECOPD was associated with incrementally higher rates of both death and cardiovascular events. Conclusion: Risk of death and cardiovascular events was greatest in the first 30 days and rose with subsequent exacerbations. Risks were elevated for 1-2 years following moderate and severe exacerbations, highlighting a sustained increased cardiopulmonary risk associated with exacerbations.

Comparative safety of conjugated estrogens/bazedoxifene versus estrogen/progestin combination hormone therapy among women in the United States: a multidatabase cohort study.

OBJECTIVE: To assess the risk of select safety outcomes including endometrial cancer, endometrial hyperplasia, and breast cancer among women using conjugated estrogens/bazedoxifene (CE/BZA) as compared with estrogen/progestin combination hormone therapy (EP). METHODS: We conducted a new-user cohort study in five US healthcare claims databases representing more than 92 million women. We included CE/BZA or EP new users from May 1, 2014, to August 30, 2019. EP users were propensity score (PS) matched to users of CE/BZA. Incidence of endometrial cancer, endometrial hyperplasia, breast cancer, and eight additional cancer and cardiovascular outcomes were ascertained using claims-based algorithms. Rate ratios (RR) and differences pooled across databases were estimated using random-effects models. RESULTS: The study population included 10,596 CE/BZA and 33,818 PS-matched EP new users. Rates of endometrial cancer and endometrial hyperplasia were slightly higher among CE/BZA users (1.6 and 0.4 additional cases per 10,000 person-years), although precision was limited because of small numbers of cases (endometrial cancer: RR, 1.50 [95% confidence interval {CI}, 0.79-2.88]; endometrial hyperplasia: RR, 1.69 [95% CI, 0.51-5.61]). Breast cancer incidence was lower in CE/BZA users (9.1 fewer cases per 10,000 person-years; RR, 0.79; 95% CI, 0.58-1.05). Rates of other outcomes were slightly higher among CE/BZA users, but with confidence intervals compatible with a wider range of possible associations. CONCLUSIONS: CE/BZA users might experience slightly higher rates of endometrial cancer and endometrial hyperplasia, and a lower rate of breast cancer, than EP users in the first years of use.

Real-world treatment patterns, healthcare resource utilization, and cost among adults with pulmonary arterial hypertension in the United States.

Treatment for pulmonary arterial hypertension (PAH) has evolved over the past decade, including approval of new medications and growing evidence to support earlier use of combination therapy. Despite these changes, few studies have assessed real-world treatment patterns, healthcare resource utilization (HCRU), and costs among people with PAH using recent data. We conducted a retrospective cohort study using administrative claims from the HealthCore Integrated Research Database®. Adult members with claims for a PAH diagnosis, right heart catheterization, and who initiated PAH treatment (index date) between October 1, 2015 and November 30, 2020 were identified. Members had to be continuously enrolled in the health plan for 6 months before the index date (baseline) and ≥30 days after. Treatment patterns, HCRU, and costs were described. A total of 843 members with PAH (mean age 62.3 years, 64.2% female) were included. Only 21.0% of members received combination therapy as their first-line treatment, while most members (54.6%) received combination therapy as second-line treatment. All-cause HCRU remained high after treatment initiation with 58.0% of members having ≥1 hospitalization and 41.3% with ≥1 emergency room visit. Total all-cause costs declined from $15,117 per patient per month at baseline to $14,201 after treatment initiation, with decreased medical costs ($14,208 vs. $6,349) more than offsetting increased pharmacy costs ($909 vs. $7,852). In summary, despite growing evidence supporting combination therapy, most members with PAH initiated treatment with monotherapy. Total costs decreased following treatment, driven by a reduction in medical costs even with increases in pharmacy costs.

Development and validation of a predictive model algorithm to identify anaphylaxis in adults with type 2 diabetes in U.S. administrative claims data.

PURPOSE: To use medical record adjudication and predictive modeling methods to develop and validate an algorithm to identify anaphylaxis among adults with type 2 diabetes (T2D) in administrative claims. METHODS: A conventional screening algorithm that prioritized sensitivity to identify potential anaphylaxis cases was developed and consisted of diagnosis codes for anaphylaxis or relevant signs and symptoms. This algorithm was applied to adults with T2D in the HealthCore Integrated Research Database (HIRD) from 2016 to 2018. Clinical experts adjudicated anaphylaxis case status from redacted medical records. We used confirmed case status as an outcome for predictive models developed using lasso regression with 10-fold cross-validation to identify predictors and estimate the probability of confirmed anaphylaxis. RESULTS: Clinical adjudicators reviewed medical records with sufficient information from 272 adults identified by the anaphylaxis screening algorithm, which had an estimated Positive Predictive Value (PPV) of 65% (95% confidence interval [CI]: 60%-71%). The predictive model algorithm had a c-statistic of 0.95. The model's probability threshold of 0.60 excluded 89% (84/94) of false positives identified by the screening algorithm, with a PPV of 94% (95% CI: 91%-98%). The model excluded very few true positives (15 of 178), and identified 92% (95% CI: 87%-96%) of the cases selected by the screening algorithm. CONCLUSIONS: Predictive modeling techniques yielded an accurate algorithm with high PPV and sensitivity for identifying anaphylaxis in administrative claims. This algorithm could be considered in future safety studies using similar claims data to reduce potential outcome misclassification.

Patient Knowledge of Safe Use of ER/LA Opioid Analgesics Following Implementation of the Class-Wide REMS: A Survey Study.

BACKGROUND/RATIONALE: The US Food and Drug Administration (FDA) approved a Risk Evaluation and Mitigation Strategy (REMS) for extended release/long-acting (ER/LA) opioids in 2012. The purpose of this study was to assess patient knowledge of the safe use of these products following implementation of the REMS and to determine possible effects of the REMS, including impact on medication access. OBJECTIVE: To assess patient knowledge of safe use of ER/LA opioids and use of REMS patient education tools such as the Medication Guide (MG) and Patient Counseling Document (PCD). METHODS: This was a cross-sectional survey of commercially insured (Commercial) and Medicare Advantage-insured (Medicare) adults with ≥1 pharmacy claim for an ER/LA opioid (10/01/2015 - 02/28/2017) in the HealthCore Integrated Research Database and Medicaid-insured (Medicaid) adult members of a research panel, about their knowledge of safe use of ER/LA opioids and receipt/comprehension of the MG and PCD. RESULTS: Survey respondents consisted of 382 Commercial, 43 Medicare and 40 Medicaid adults. While ≥95% of respondents received and read the MG, fewer were aware of the PCD (Commercial: 47%, Medicare: 65%, Medicaid: 53%). Almost 75% of the knowledge questions were answered correctly by ≥80% of all respondents; fewer respondents recognized that use of opioids as directed can lead to death (Commercial: 73%, Medicare: 56%, Medicaid: 63%), the MG should be read at each dispensing (Commercial: 78%, Medicare: 53%, Medicaid: 75%), opioids should not be stored in the medicine cabinet (Commercial: 77%, Medicare: 79%, Medicaid: 58%), missed doses should not be taken as soon as possible (Commercial: 56%, Medicare: 51%, Medicaid: 50%), and pills should not be crushed (Commercial: 85%, Medicare: 67%, Medicaid: 52%). CONCLUSION: Although most respondents reported reading and understanding the MG and exhibited knowledge of safe use of ER/LA opioids, providers' use of the PCD and increased understanding of safe use core messages need reinforcement.

Real-world safety of palbociclib in breast cancer patients in the United States: a new user cohort study.

BACKGROUND: There is limited real-world safety information on palbociclib for treatment of advanced stage HR+/HER2- breast cancer. METHODS: We conducted a cohort study of breast cancer patients initiating palbociclib and fulvestrant from February 2015 to September 2017 using the HealthCore Integrated Research Database (HIRD), a longitudinal claims database of commercial health plan members in the United States. The historical comparator cohort comprised patients initiating fulvestrant monotherapy from January 2011 to January 2015. Propensity score matching and Cox regression were used to estimate hazard ratios for various safety events. For acute liver injury (ALI), additional analyses and medical record validation were conducted. RESULTS: There were 2445 patients who initiated palbociclib including 566 new users of palbociclib-fulvestrant, and 2316 historical new users of fulvestrant monotherapy. Compared to these historical new users of fulvestrant monotherapy, new users of palbociclib-fulvestrant had a greater than 2-fold elevated risk for neutropenia, leukopenia, thrombocytopenia, stomatitis and mucositis, and ALI. Incidence of anemia and QT prolongation were more weakly associated, and incidences of serious infections and pulmonary embolism were similar between groups after propensity score matching. After adjustment for additional ALI risk factors, the elevated risk of ALI in new users of palbociclib-fulvestrant persisted (e.g. primary ALI algorithm hazard ratio (HR) = 3.0, 95% confidence interval (CI) = 1.1-8.4). CONCLUSIONS: This real-world study found increased risks of several adverse events identified in clinical trials, including neutropenia, leukopenia, and thrombocytopenia, but no increased risk of serious infections or pulmonary embolism when comparing new users of palbociclib-fulvestrant to fulvestrant monotherapy. We observed an increased risk of ALI, extending clinical trial findings of significant imbalances in grade 3/4 elevations of alanine aminotransferase (ALT).

Prior presumed coronavirus infection reduces COVID-19 risk: A cohort study.

BACKGROUND: Immunological cross-reactivity between common cold coronaviruses (CCC) and SARS-CoV-2 might account for the reduced incidence of COVID-19 in children. Evidence to support speculation includes in vitro evidence for humoral and cellular cross-reactivity with SARS-CoV-2 in specimens obtained before the pandemic started. METHOD: We used retrospective health insurance enrollment records, claims, and laboratory results to assemble a cohort of 869,236 insured individuals who had a PCR test for SARS-CoV-2. We estimated the effects of having clinical encounters for various diagnostic categories in the year preceding the study period on the risk of a positive test result. FINDINGS: After adjusting for age, gender and care seeking behavior, we identified that individuals with diagnoses for common cold symptoms, including acute sinusitis, bronchitis, or pharyngitis in the preceding year had a lower risk of testing positive for SARS-CoV-2 (OR=0.76, 95%CI=0.75, 0.77). No reduction in the odds of a positive test for SARS-CoV-2 was seen in individuals under 18 years. The reduction in odds in adults remained stable for four years but was strongest in those with recent common cold symptoms. INTERPRETATION: While this study cannot attribute this association to cross-immunity resulting from a prior CCC infection, it is one potential explanation. Regardless of the cause, the reduction in the odds of being infected by SARS-CoV-2 among those with a recent diagnosis of common cold symptoms may have a role in shifting future COVD-19 infection patterns from endemic to episodic.

A real-world study on characteristics, treatments and outcomes in US patients with advanced stage ovarian cancer.

BACKGROUND: Detailed epidemiologic descriptions of large populations of advanced stage ovarian cancer patients have been lacking to date. This study aimed to describe the patient characteristics, treatment patterns, survival, and incidence rates of health outcomes of interest (HOI) in a large cohort of advanced stage ovarian cancer patients in the United States (US). METHODS: This cohort study identified incident advanced stage (III/IV) ovarian cancer patients in the US diagnosed from 2010 to 2018 in the HealthCore Integrated Research Database (HIRD) using a validated predictive model algorithm. Descriptive characteristics were presented overall and by treatment line. The incidence rates and 95% confidence intervals for pre-specified HOIs were evaluated after advanced stage diagnosis. Overall survival, time to treatment discontinuation or death (TTD), and time to next treatment or death (TTNT) were defined using treatment information in claims and linkage with the National Death Index. RESULTS: We identified 12,659 patients with incident advanced stage ovarian cancer during the study period. Most patients undergoing treatment received platinum agents (75%) and/or taxanes (70%). The most common HOIs (> 24 per 100 person-years) included abdominal pain, nausea and vomiting, anemia, and serious infections. The median overall survival from diagnosis was 4.5 years, while approximately half of the treated cohort had a first-line time to treatment discontinuation or death (TTD) within the first 4 months, and a time to next treatment or death (TTNT) from first to second-line of about 6 months. CONCLUSIONS: This study describes commercially insured US patients with advanced stage ovarian cancer from 2010 to 2018, and observed diverse treatment patterns, incidence of numerous HOIs, and limited survival in this population.

Outcome misclassification: Impact, usual practice in pharmacoepidemiology database studies and an online aid to correct biased estimates of risk ratio or cumulative incidence.

PURPOSE: It is well documented that outcome misclassification can bias a point estimate. We aimed to understand current practice in addressing this bias in pharmacoepidemiology database studies and to develop an open source application (app) from existing methodology to demonstrate the impact and mechanism of this bias on results. METHODS: Studies of an exposure and a clinical outcome were selected from all Pharmacoepidemiology and Drug Safety publications during 2017 and any reference to outcome misclassification described. An app to correct risk ratio (RR) and cumulative incidence for outcome misclassification was developed from a published methodology and used to demonstrate the impact of correction on point estimates. RESULTS: Eight (19%) of 43 papers selected reported estimates of outcome ascertainment accuracy with positive predictive value (PPV) the most commonly reported measure (7 of 8 studies). Three studies (7%) corrected for the bias, 1 by exposure strata, and 5 (12%) restricted analyses to confirmed cases. The app (app http://apps.p-95.com/ISPE/) uses values of PPV and sensitivity (or a range of possible values) in each exposure strata and returns corrected point estimates and confidence intervals. The app demonstrates that small differences between comparison groups in PPV or sensitivity can introduce bias even when accuracy estimates are high. CONCLUSIONS: Outcome misclassification is not usually corrected in pharmacoepidemiology database studies although correction methods using routinely measured indices are available. Error indices are needed for each comparison group to correct RR estimates for these errors. The app should encourage understanding of this bias and increase adjustment.

Incidence of Opioid Overdose Among Patients Using ER/LA Opioid Analgesics Before and After Implementation of the Class-Wide Opioid Risk Evaluation and Mitigation Strategy.

INTRODUCTION: The United States (US) Food and Drug Administration (FDA) required a Risk Evaluation and Mitigation Strategy (REMS) for extended-release and long-acting (ER/LA) opioid analgesics on 09 July 2012. METHODS: This study compared the incidence of opioid overdose before (July 2010-June 2012) and after (July 2013-September 2016) the initiation of the Risk Evaluation and Mitigation Strategy (REMS) for extended-release and long-acting (ER/LA) opioid analgesics. We identified patients with ≥1 ER/LA opioid dispensing in either time period in national data from the HealthCore Integrated Research DatabaseSM (HIRD) and in United States (US) Medicaid claims data from four states. We described each population, calculated the incidence rate (IR) of opioid overdose, and assessed crude and propensity score adjusted incidence rate ratios (IRR) comparing the overdose rate after vs before implementation of the REMS. RESULTS: A total of 121,229 commercially insured and 11,488 Medicaid patients were included in the analysis. Rates of overdose were substantially higher in Medicaid patients than in the commercially insured patients (IR 192.0, 95% confidence interval [CI] 162.60-225.18 versus 102.60, 95% CI 93.0-112.93 in the active period). The IRRs for opioid overdose were 1.01 (95% CI 0.87-1.17) in the commercially insured population and 0.70 (95% CI 0.52-0.93) in Medicaid. CONCLUSION: This leveling off of overdose rates among commercially insured patients and decline among Medicaid patients is encouraging, but it is difficult to disentangle the specific impact of the REMS from many other ongoing initiatives with similar goals.

Testosterone Therapy in Relation to Prostate Cancer in a U.S. Commercial Insurance Claims Database.

BACKGROUND: We conducted a study to assess whether testosterone therapy (TT) alters prostate cancer risk using a large U.S. commercial insurance research database. METHODS: From the HealthCore Integrated Research Database (HIRD), we selected men ages 30 years or greater who were new users of TT during 2007 to 2015. We selected two comparison groups: (i) unexposed (matched 10:1) and (ii) new users of phosphodiesterase type 5 inhibitor (PDE5i). Incident prostate cancer was defined as diagnosis of prostate cancer within 4 weeks following prostate biopsy. Propensity scores and inverse probability of treatment weights were used in Poisson regression models to estimate adjusted incidence rates, incidence rate ratios (IRR), and 95% confidence intervals (CI). Subgroup analyses included stratification by prostate cancer screening, hypogonadism, and follow-up time. RESULTS: The adjusted prostate cancer IRR was 0.77 (95% CI, 0.68-0.86) when comparing TT with the unexposed group and 0.85 (95% CI, 0.79-0.91) in comparison with the PDE5i group. Inverse associations between TT and prostate cancer were observed in a majority of subgroup analyses, although in both comparisons estimates generally attenuated with increasing time following initial exposure. Among TT users, duration of exposure was not associated with prostate cancer. CONCLUSIONS: Men who received TT did not have a higher rate of prostate cancer compared with the unexposed or PDE5i comparison groups. The inverse association between TT and prostate cancer could be the result of residual confounding, contraindication bias, or undefined biological effect. IMPACT: This study suggests that limited TT exposure does not increase risk of prostate cancer in the short term.

Medical record-based ascertainment of behaviors suggestive of opioid misuse, diversion, abuse, and/or addiction among individuals showing evidence of doctor/pharmacy shopping.

OBJECTIVES: Doctor/pharmacy shopping, the practice of seeking prescriptions from multiple healthcare sources without their coordination, may be a measure of prescription medicine abuse. This cross-sectional study examined the relationship between a claims-based doctor/pharmacy shopping definition and medical record documented behaviors suggestive of misuse, diversion, abuse and/or addiction. METHODS: Patients with ≥2 opioid dispensings starting in 2012 in a US administrative claims database were grouped into doctor/pharmacy shopping categories by number of providers and pharmacies used over 18 months: no shopping, minimal shopping, moderate shopping and severe shopping. Medical charts of opioid prescribers were reviewed to identify behaviors suggestive of misuse, diversion, abuse and/or addiction. RESULTS: Among 581,940 opioid users, 78% were classified as no shopping, 11% minimal shopping, 8% moderate shopping and 3% severe shopping. Almost 40% of severe shopping patients had no medical record documented behaviors (positive predictive value: 24.3%). Compared with no shopping, the odds ratio [OR] of ≥3 behaviors was 1.70 (95% confidence interval [CI] 0.50-5.78) for minimal shopping, 1.81 (95% CI 0.54-6.03) for moderate shopping, and 8.93 (95% CI 3.12-25.54) for severe shopping. CONCLUSIONS: Claims-identified severe doctor/pharmacy shopping was strongly associated with behaviors suggestive of misuse, diversion, abuse and/or addiction, but the proportion of medical records documenting these was low.

Development of predictive models to identify advanced-stage cancer patients in a US healthcare claims database.

BACKGROUND: Although healthcare databases are a valuable source for real-world oncology data, cancer stage is often lacking. We developed predictive models using claims data to identify metastatic/advanced-stage patients with ovarian cancer, urothelial carcinoma, gastric adenocarcinoma, Merkel cell carcinoma (MCC), and non-small cell lung cancer (NSCLC). METHODS: Patients with ≥1 diagnosis of a cancer of interest were identified in the HealthCore Integrated Research Database (HIRD), a United States (US) healthcare database (2010-2016). Data were linked to three US state cancer registries and the HealthCore Integrated Research Environment Oncology database to identify cancer stage. Predictive models were constructed to estimate the probability of metastatic/advanced stage. Predictors available in the HIRD were identified and coefficients estimated by Least Absolute Shrinkage and Selection Operator (LASSO) regression with cross-validation to control overfitting. Classification error rates and receiver operating characteristic curves were used to select probability thresholds for classifying patients as cases of metastatic/advanced cancer. RESULTS: We used 2723 ovarian cancer, 6522 urothelial carcinoma, 1441 gastric adenocarcinoma, 109 MCC, and 12,373 NSCLC cases of early and metastatic/advanced cancer to develop predictive models. All models had high discrimination (C > 0.85). At thresholds selected for each model, PPVs were all >0.75: ovarian cancer = 0.95 (95% confidence interval [95% CI]: 0.94-0.96), urothelial carcinoma = 0.78 (95% CI: 0.70-0.86), gastric adenocarcinoma = 0.86 (95% CI: 0.83-0.88), MCC = 0.77 (95% CI 0.68-0.89), and NSCLC = 0.91 (95% CI 0.90 - 0.92). CONCLUSION: Predictive modeling was used to identify five types of metastatic/advanced cancer in a healthcare claims database with greater accuracy than previous methods.

Validation of an Algorithm for Claims-based Incidence of Prostate Cancer.

BACKGROUND: Prostate cancer is a commonly studied outcome in administrative claims studies, but there is a dearth of validated case identifying algorithms. The long-term development of the disease increases the difficulty in separating prevalent from incident prostate cancer. The purpose of this validation study was to assess the accuracy of a claims algorithm to identify incident prostate cancer among men in commercial and Medicare Advantage US health plans. METHODS: We identified prostate cancer in claims as a prostate cancer diagnosis within 28 days after a prostate biopsy and compared case ascertainment in the claims with the gold standard results from the Georgia Comprehensive Cancer Registry (GCCR). RESULTS: We identified 74,008 men from a large health plan claims database for possible linkage with GCCR. Among the 382 prostate cancer cases identified in claims, 312 were also identified in the GCCR (positive predictive value [PPV] = 82%). Of the registry cases, 91% (95% confidence interval = 88, 94) were correctly identified in claims. Claims and registry diagnosis dates of prostate cancer matched exactly in 254/312 (81%) cases. Nearly half of the false-positive cases also had claims for prostate cancer treatment. Thirteen (43%) false-negative cases were classified as noncases by virtue of having a biopsy and diagnosis >28 days apart as required by the algorithm. Compared to matches, false-negative cases were older men with less aggressive prostate cancer. CONCLUSIONS: Our algorithm demonstrated a PPV of 82% with 92% sensitivity in ascertaining incident PC. Administrative health plan claims can be a valuable and accurate source to identify incident prostate cancer cases.

Predictive model algorithms identifying early and advanced stage ER+/HER2- breast cancer in claims data.

PURPOSE: Claims databases offer large populations for research, but lack clinical details. We aimed to develop predictive models to identify estrogen receptor positive (ER+) and human epidermal growth factor negative (HER2-) early breast cancer (ESBC) and advanced stage breast cancer (ASBC) in a claims database. METHODS: Female breast cancer cases in Anthem's Cancer Care Quality Program served as the gold standard validation sample. Predictive models were developed from clinical knowledge and empirically from claims data using logistic and lasso regression. Model performance was assessed by classification rates and c-statistics. Models were applied to the HealthCore Integrated Research Database (claims) to identify cohorts of women with ER+/HER2- ESBC and ASBC. RESULTS: The validation sample included 3184 women with ER+/HER2- ESBC and 1436 with ER+/HER2- ASBC. Predictive models for ER+/HER2- ESBC and ASBC included 25 and 20 factors, respectively. Models had robust discrimination in identifying cases (c-stat = 0.92 for ESBC and 0.95 for ASBC). Compared with a traditional a priori algorithm developed with clinical insight alone, the ER+/HER2- ASBC-predictive model had better positive predictive value (PPV) (0.91, 95% CI, 0.90-0.93, vs 0.69, 95% CI, 0.66-0.73) and sensitivity (0.54 vs 0.35). Models were applied to the claims database to identify cohorts of 33 001 and 3198 women with ER+/HER2- ESBC and ASBC. CONCLUSION: We conducted a validation study and developed predictive models to identify in a claims database cohorts of women with ER+/HER2- ESBC and ASBC. The models identified large cohorts in the claims data that can be used to characterize indications in the evaluation of targeted therapies.