Administrative Claims Data Research Articles

A refined ICD-10 diagnoses-based approach for retrospective analysis of potential palliative care need and coverage in claims data of deceased.

ICD-10-based approaches often provide the basis for retrospective estimation of potential palliative care need. Applying the ICD-10-based Murtagh et al. classification from 2014 (Murtagh classification), developed using mortality data, to administrative claims data leads to inconsistencies in estimating palliative care need. The aim of the study was to refine the classification for palliative care need estimation in deceased individuals with cancer and non-cancer diagnosis. A retrospective population-based study comparing Murtagh classification to a new ICD-10-based classification (revised by expert opinion) was conducted using outpatient and inpatient claims data, including billing codes for palliative care. Palliative care need was estimated for diagnoses groups and was contrasted with palliative care utilization rates in the last year of life. Our dataset included records of 417,405 individuals who deceased in 2016-2019. Out of individuals deceased in 2019 (n = 117,436), 81.4% had at least one diagnosis from the new classification, while 97.0% had at least one diagnosis from the Murtagh classification. Classification revision thus identified fewer individuals as potentially in need of palliative care. Among individuals with cancer, 70.7% (vs. 55.7% via Murtagh classification) received palliative care. In non-cancer subgroups, the utilization rate was considerably lower, with a maximum of 36.7% (vs. 33.7% via Murtagh classification) in 2019. Similar results were observed for the other years. Compared to the ICD10-based Murtagh classification, the revised ICD-10-based classification enables more realistic estimations if the cause of death is unavailable and reveals higher rates of palliative care coverage and differences especially in cancer versus non-cancer diseases. German Clinical Trials Register (DRKS00024133).

Treatment sequencing after failure to alectinib in patients with anaplastic lymphoma kinase-positive non-small-cell lung cancer.

Alectinib is the first-line therapy for anaplastic lymphoma kinase-positive non-small-cell lung cancer. Although some guidelines have recommended using other anaplastic lymphoma kinase inhibitors after alectinib failure, evidence for such regimens in patients who fail to respond to alectinib is limited. This study involved using administrative claims data from acute care hospitals in Japan. We extracted the data of 634 patients diagnosed with lung cancer between September 1, 2014, and January 31, 2023, who received alectinib treatment before treatment with another anaplastic lymphoma kinase inhibitor. We assessed distributions of patients according to their treatment sequencing and prognosis among three periods defined based on the initial marketing dates of lorlatinib and brigatinib. The type of anaplastic lymphoma kinase inhibitors after alectinib failure changed over time. In the most recent period, lorlatinib (58%) and brigatinib (40%) became predominant. Two-year overall survival improved over time (47%-84%), accompanied by an increased 2-year proportion of patients who continuously used anaplastic lymphoma kinase inhibitors after alectinib failure (13%-44%). The times to treatment discontinuation of the regimen between patients treated with lorlatinib and brigatinib were similar, with a hazard ratio of 1.02 (95% confidence interval, 0.64-1.64) in the period after marketing brigatinib. This study provides insights into the evolving treatment landscape for patients with anaplastic lymphoma kinase-positive non-small-cell lung cancer who experience failed alectinib treatment and highlights the need for further studies and data accumulation to determine the optimal treatment strategy.

Longitudinal medication adherence group-based trajectories of aging adults in the US: A retrospective analysis using monthly proportion of days covered calculations

BackgroundIt is thought that half of the patients with chronic conditions are not adherent to their medications, which contributes to significant health and economic burden. Many studies estimate medication non-adherence by implementing a threshold of ≥80% of Proportion of Days Covered (PDC), categorizing patients as either adherent or non-adherent. Healthcare quality metrics pertaining to medication use are based on this dichotomous approach of medication adherence, including the Medicare Part D Star Ratings. Among others, the Medicare Part D Star Ratings rewards part D plan sponsors with quality bonus payments based on this dichotomous categorization of beneficiaries’ medication adherence. ObjectivesDescribe the longitudinal adherence trajectories of adults ≥65 years of age covered by Medicare for 3 classes of drugs in the Part D Star Ratings: diabetes medications, statins, and select antihypertensives. MethodsThis study used Medicare healthcare administrative claims data linked to participants from the Health Retirement Study between 2008 and 2016. Group-based trajectory models (GBTM) elicited the number and shape of adherence trajectories from a sample of N = 11,068 participants for the three pharmacotherapeutic classes considered in this study. Medication adherence was estimated using monthly PDC. ResultsGBTM were estimated for the sample population taking antihypertensives (n = 7,272), statins (n = 8,221), and diabetes medications (n = 3,214). The hypertension model found three trajectories: high to very high adherence (47.55%), slow decline (32.99%), and rapid decline (19.47%) trajectories. The statins model found 5 trajectories: high to very high adherence (35.49%), slow decline (17.12%), low then increasing adherence (23.58%), moderate decline (12.62%), and rapid decline (11.20%). The diabetes medications model displayed 6 trajectories: high to very high adherence (24.15%), slow decline (16.84%), high then increasing adherence (25.56%), low then increasing (13.58%), moderate decline (10.60%), and rapid decline (9.27%). ConclusionsThis study showed the fluid nature of long-term medication adherence to the medications considered in the Medicare Part D Star Ratings and how it varies by pharmacotherapeutic class. These challenge previous assumptions about which patients were considered adherent to chronic medications. Policy and methodological implications about medication adherence are discussed.

Surgical prescription opioid trajectories among state Medicaid enrollees.

The objective of this study was to evaluate opioid use trajectories among a sample of 10,138 Medicaid patients receiving one of six index surgeries: lumbar spine, total knee arthroplasty, cholecystectomy, appendectomy, colon resection, and tonsillectomy. Retrospective cohort. Administrative claims data. Patients, aged 13 years and older, with 15-month continuous Medicaid eligibility surrounding index surgery, were selected from single-state Medicaid medical and pharmacy claims data for surgeries performed between 2014 and 2017. None. Baseline comorbidities and presurgery opioid use were assessed in the 6 months prior to admission, and patients' opioid use was followed for 9 months post-discharge. Generalized linear model with log link and Poisson distribution was used to determine risk of chronic opioid use for all risk factors. Group-based trajectory models identified groups of patients with similar opioid use trajectories over the 15-month study period. More than one in three (37.7 percent) patients were post-surgery chronic opioid users, defined as the dichotomous outcome of filling an opioid prescription 90 or more days after surgery. Key variables associated with chronic post-surgery opioid use include presurgery opioid use, 30-day post-surgery opioid use, and comorbidities. Latent trajectory modeling grouped patients into six distinct opioid use trajectories. Associates of trajectory group membership are reported. Findings support the importance of surgeons setting realistic patient expectations for post-surgical opioid use, as well as the importance of coordination of post-surgical care among patients failing to fully taper off opioids within 1-3 months of surgery.

Identifying neonatal intensive care (NICU) admissions using administrative claims data.

To define a method for identifying neonatal intensive care unit (NICU) admissions using administrative claims data. This was a retrospective cohort study using claims from Optum's de-identified Clinformatics® Data Mart Database (CDM) from 2016 -2020. We developed a definition to identify NICU admissions using a list of codes from the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM), Current Procedural Terminology (CPT), and revenue codes frequently associated with NICU admissions. We compared agreement between codes using Kappa statistics and calculated positive predictive values (PPV) and 95% confidence intervals (CI). On average, revenue codes (3.3%) alone identified more NICU hospitalizations compared to CPT codes alone (1.5%), whereas the use of CPT and revenue (8.9%) and CPT or revenue codes (13.7%) captured the most NICU hospitalizations, which aligns with rates of preterm birth. Gestational age alone (4.2%) and birthweight codes alone (2.0%) identified the least number of potential NICU hospitalizations. Setting CPT codes as the standard and revenue codes as the "test,", revenue codes resulted in identifying 86% of NICU admissions (sensitivity) and 97% of non-NICU admissions (specificity). Using administrative data, we developed a robust definition for identifying neonatal admissions. The identified definition of NICU codes is easily adaptable, repeatable, and flexible for use in other datasets.

The Problem of Pain in Rheumatology: Variations in Case Definitions Derived From Chronic Pain Phenotyping Algorithms Using Electronic Health Records.

The aim of this study was to investigate and compare different case definitions for chronic pain to provide estimates of possible misclassification when researchers are limited by available electronic health record and administrative claims data, allowing for greater precision in case definitions. We compared the prevalence of different case definitions for chronic pain (N = 3042) in patients with autoimmune rheumatic diseases. We estimated the prevalence of chronic pain based on 15 unique combinations of pain scores, diagnostic codes, analgesic medications, and pain interventions. Chronic pain prevalence was lowest in unimodal pain phenotyping algorithms: 15% using analgesic medications, 18% using pain scores, 21% using pain diagnostic codes, and 22% using pain interventions. In comparison, the prevalence using a well-validated phenotyping algorithm was 37%. The prevalence of chronic pain also increased with the increasing number (bimodal to quadrimodal) of phenotyping algorithms that comprised the multimodal phenotyping algorithms. The highest estimated chronic pain prevalence (47%) was the multimodal phenotyping algorithm that combined pain scores, diagnostic codes, analgesic medications, and pain interventions. However, this quadrimodal phenotyping algorithm yielded a 10% overestimation of chronic pain compared to the well-validated algorithm. This is the first empirical study to our knowledge that shows that established common modes of phenotyping chronic pain can lead to substantially varying estimates of the number of patients with chronic pain. These findings can be a reference for biases in case definitions for chronic pain and could be used to estimate the extent of possible misclassifications or corrections in using datasets that cannot include specific data elements.

Weathering the Storm: Syringe Services Program Laws and Human Immunodeficiency Virus During the COVID-19 Pandemic.

Syringe services programs (SSPs) are community-based prevention programs that provide a range of harm reduction services to persons who inject drugs. Despite their benefits, SSP laws vary across the United States. Little is known regarding how legislation surrounding SSPs may have influenced HIV transmission over the COVID-19 pandemic, a period in which drug use increased. This study examined associations between state SSP laws and HIV transmission among the Medicaid population before and after the COVID-19 pandemic. State-by-month counts of new HIV diagnoses among the Medicaid population were produced using administrative claims data from the Transformed Medicaid Statistical Information System from 2019 to 2020. Data on SSP laws were collected from the Prescription Drug Abuse Policy System. Associations between state SSP laws and HIV transmission before and after the start of the COVID-19 pandemic were evaluated using an event study design, controlling for the implementation of COVID-19 nonpharmaceutical interventions and state and time fixed effects. State laws allowing the operation of SSPs were associated with 0.54 (P = 0.044) to 1.18 (P = 0.001) fewer new monthly HIV diagnoses per 100,000 Medicaid enrollees relative to states without such laws in place during the 9 months after the start of the COVID-19 pandemic. The largest effects manifested for population subgroups disproportionately affected by HIV, such as male and non-Hispanic Black Medicaid enrollees. Less restrictive laws on SSPs may have helped mitigate HIV transmission among the Medicaid population throughout the COVID-19 pandemic. Policymakers can consider implementing less restrictive SSP laws to mitigate HIV transmission resulting from future increases in injection drug use. The findings and conclusions in this report are those of the author(s) and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Treatment Patterns of Long-Acting Somatostatin Analogs for Neuroendocrine Tumors

Background: Long-acting somatostatin analog therapy (LA-SSA) is recommended as first-line therapy for treatment of unresectable or metastatic neuroendocrine tumors (NETs). Understanding treatment sequencing and dosing patterns of LA-SSA is essential for clinical decision-making to provide value-based management of NETs. Objective: To describe treatment patterns of LA-SSA among patients with NETs and subgroups with carcinoid syndrome (CS) in the United States. Methods: This retrospective study utilized claims data from MarketScan® databases to identify patients with NETs and newly treated with LA-SSA between January 1, 2015, and October 31, 2020. Patients were stratified by index LA-SSA (lanreotide and octreotide long-acting release [LAR]). Reported 28-day doses were based on claim fields for days’ supply/drug quantity or units of service. Dose escalation was defined as increases in quantity or frequency. Continuous variables, categorical variables, and Kaplan-Meier estimated treatment durations were compared using t-tests, chi-square/Fisher’s tests, and log-rank tests, respectively. Results: The study included 241 lanreotide and 521 octreotide LAR patients. Compared with octreotide LAR patients, treatment duration was longer for lanreotide patients (median, 41.3 vs 26.8 months; log-rank p=.004). Fewer lanreotide patients received rescue treatment with short-acting octreotide (7.9% vs 14.4%; p=.011), and a first (6.2% vs 27.3%) and second dose escalation (0.8% vs 5.2%; both p<.05). Among patients with doses reported, fewer lanreotide patients received above-label doses (2.5% [5/202] vs 14.4% [60/416]; p<.001). Among patients who ended treatment during follow-up, fewer lanreotide patients transitioned to another LA-SSA (18.9% [17/90] vs 33.6% [92/274]; p=.008). Similar treatment patterns were observed in CS subgroups. Results for switched treatment patterns were limited due to insufficient sample sizes. Discussion: Real-world treatment patterns of LA-SSA were assessed using more recent administrative claims data. Compared with octreotide LAR patients, lanreotide patients were more likely to remain longer on initial treatment and starting dose without dose escalations and less likely to use rescue treatment and transition to another LA-SSA after discontinuation of the index treatment. Conclusions: Findings from this claims study suggest a potential clinical benefit of lanreotide in NET management.

Treatment Patterns of Long-Acting Somatostatin Analogs for Neuroendocrine Tumors.

Background: Long-acting somatostatin analog therapy (LA-SSA) is recommended as first-line therapy for treatment of unresectable or metastatic neuroendocrine tumors (NETs). Understanding treatment sequencing and dosing patterns of LA-SSA is essential for clinical decision-making to provide value-based management of NETs. Objective: To describe treatment patterns of LA-SSA among patients with NETs and subgroups with carcinoid syndrome (CS) in the United States. Methods: This retrospective study utilized claims data from MarketScan® databases to identify patients with NETs and newly treated with LA-SSA between January 1, 2015, and October 31, 2020. Patients were stratified by index LA-SSA (lanreotide and octreotide long-acting release [LAR]). Reported 28-day doses were based on claim fields for days' supply/drug quantity or units of service. Dose escalation was defined as increases in quantity or frequency. Continuous variables, categorical variables, and Kaplan-Meier estimated treatment durations were compared using t-tests, chi-square/Fisher's tests, and log-rank tests, respectively. Results: The study included 241 lanreotide and 521 octreotide LAR patients. Compared with octreotide LAR patients, treatment duration was longer for lanreotide patients (median, 41.3 vs 26.8 months; log-rank p=.004). Fewer lanreotide patients received rescue treatment with short-acting octreotide (7.9% vs 14.4%; p=.011), and a first (6.2% vs 27.3%) and second dose escalation (0.8% vs 5.2%; both p<.05). Among patients with doses reported, fewer lanreotide patients received above-label doses (2.5% [5/202] vs 14.4% [60/416]; p<.001). Among patients who ended treatment during follow-up, fewer lanreotide patients transitioned to another LA-SSA (18.9% [17/90] vs 33.6% [92/274]; p=.008). Similar treatment patterns were observed in CS subgroups. Results for switched treatment patterns were limited due to insufficient sample sizes. Discussion: Real-world treatment patterns of LA-SSA were assessed using more recent administrative claims data. Compared with octreotide LAR patients, lanreotide patients were more likely to remain longer on initial treatment and starting dose without dose escalations and less likely to use rescue treatment and transition to another LA-SSA after discontinuation of the index treatment. Conclusions: Findings from this claims study suggest a potential clinical benefit of lanreotide in NET management.

Characteristics and End-of-Life Care Pathways of Decedents From a National Cohort of Assisted Living Residents.

Assisted living (AL) is an increasingly common, place of care for dying persons. However, it remains unclear to what extent residents are able to age in place or if AL represents an additional transition before death. Examine the sociodemographic characteristics, comorbidities, health care utilization, and end-of-life care pathways of AL residents before death. A national cohort study of fee-for-service Medicare beneficiaries residing in large AL communities (25+ beds) during the month of January 2017 with 3 years of follow-up, using administrative claims data. 268,812 AL residents. Sociodemographic characteristics, comorbidities, and health care utilization at the end of life. Between 2017 and 2019, 35.1% of the study cohort died. Decedents were more likely than the overall AL population to be 85 years old or older (76.5% vs. 59.5%), and diagnosed with Alzheimer's disease and related dementia (70.3% vs. 51.6%). Most decedents (96.2%) had some presence in AL during the last year of life, but over 1 in 5 left AL before the last month of life. Among those in AL on day 30 before death, nearly half (46.4%) died in place without any health care transition, while 13.2% had 3 or more transfers before dying. AL is an important place of care for dying persons, especially for those with dementia. These findings indicate a need to assess existing policies and processes guiding the care of the frail and vulnerable population of dying AL residents.

Predicting seizure recurrence after an initial seizure-like episode from routine clinical notes using large language models: a retrospective cohort study.

The evaluation and management of first-time seizure-like events in children can be difficult because these episodes are not always directly observed and might be epileptic seizures or other conditions (seizure mimics). We aimed to evaluate whether machine learning models using real-world data could predict seizure recurrence after an initial seizure-like event. This retrospective cohort study compared models trained and evaluated on two separate datasets between Jan 1, 2010, and Jan 1, 2020: electronic medical records (EMRs) at Boston Children's Hospital and de-identified, patient-level, administrative claims data from the IBM MarketScan research database. The study population comprised patients with an initial diagnosis of either epilepsy or convulsions before the age of 21 years, based on International Classification of Diseases, Clinical Modification (ICD-CM) codes. We compared machine learning-based predictive modelling using structured data (logistic regression and XGBoost) with emerging techniques in natural language processing by use of large language models. The primary cohort comprised 14 021 patients at Boston Children's Hospital matching inclusion criteria with an initial seizure-like event and the comparison cohort comprised 15 062 patients within the IBM MarketScan research database. Seizure recurrence based on a composite expert-derived definition occurred in 57% of patients at Boston Children's Hospital and 63% of patients within IBM MarketScan. Large language models with additional domain-specific and location-specific pre-training on patients excluded from the study (F1-score 0·826 [95% CI 0·817-0·835], AUC 0·897 [95% CI 0·875-0·913]) performed best. All large language models, including the base model without additional pre-training (F1-score 0·739 [95% CI 0·738-0·741], AUROC 0·846 [95% CI 0·826-0·861]) outperformed models trained with structured data. With structured data only, XGBoost outperformed logistic regression and XGBoost models trained with the Boston Children's Hospital EMR (logistic regression: F1-score 0·650 [95% CI 0·643-0·657], AUC 0·694 [95% CI 0·685-0·705], XGBoost: F1-score 0·679 [0·676-0·683], AUC 0·725 [0·717-0·734]) performed similarly to models trained on the IBM MarketScan database (logistic regression: F1-score 0·596 [0·590-0·601], AUC 0·670 [0·664-0·675], XGBoost: F1-score 0·678 [0·668-0·687], AUC 0·710 [0·703-0·714]). Physician's clinical notes about an initial seizure-like event include substantial signals for prediction of seizure recurrence, and additional domain-specific and location-specific pre-training can significantly improve the performance of clinical large language models, even for specialised cohorts. UCB, National Institute of Neurological Disorders and Stroke (US National Institutes of Health).

CO84 Characterization of Long-COVID Phenotypes and Associated Clinical Phenotypes in Administrative Claims Data

CO84 Characterization of Long-COVID Phenotypes and Associated Clinical Phenotypes in Administrative Claims Data

Real-World Study of Single-Inhaler Triple Therapy with Fluticasone Furoate/Umeclidinium/Vilanterol on Asthma Control in the US.

Real-world asthma control data among patients initiating fluticasone furoate/umeclidinium/vilanterol (FF/UMEC/VI) are limited. This study assessed rescue medication use and asthma-related exacerbations in patients with asthma before and after initiating single-inhaler FF/UMEC/VI using administrative claims data. This retrospective, pre-post cohort study analyzed data from the IQVIA PharMetrics Plus database (September 18, 2016‒March 31, 2020). Patients aged ≥18 years that had ≥1 dispensing of single-inhaler FF/UMEC/VI 100/62.5/25 mcg (first dispensing = index date), ≥12 months of continuous health insurance enrollment prior to (pre-treatment) and following (post-treatment) FF/UMEC/VI initiation and ≥1 diagnosis of asthma during the pre-treatment period or on the index date were included. The primary endpoint was the number of oral corticosteroid (OCS) dispensings per patient per year during pre- and post-treatment periods. Secondary endpoints included asthma-related exacerbation rates and short-acting β2-agonist (SABA) use. Comparisons between pre- and post-treatment periods were made using risk and rate ratios. Overall, 890 patients with asthma initiating treatment with FF/UMEC/VI were included. The most recently dispensed controller medications prior to FF/UMEC/VI initiation were inhaled corticosteroids/long-acting β2-agonists (33.5%) and leukotriene modifiers (33.0%). Patients had a 29% reduction in the number of OCS dispensings (rate ratio [95% confidence interval (CI)]: 0.71 [0.65, 0.77], P < 0.001) during post-treatment versus pre-treatment, with a 23% reduction in the proportion of patients with ≥1 OCS dispensing post-treatment (risk ratio [95% CI]: 0.77 [0.73, 0.82], P < 0.001). Significant reductions in rates (rate ratio [95% CI]) of asthma-related exacerbations (0.59 [0.52, 0.67], P < 0.001) and SABA use (0.80 [0.74, 0.86], P < 0.001) were also observed. In this real-world study, patients with asthma had significantly lower OCS use, asthma-related exacerbations, and SABA use following treatment initiation with FF/UMEC/VI compared with their pre-treatment period. These results suggest better asthma control following initiation of FF/UMEC/VI in a routine clinical practice setting.

CO142 A Retrospective Study Characterizing Age at Key Clinical Disease Indicators Among Patients With Friedreich Ataxia Using Health Administrative Claims Data in the United States

CO142 A Retrospective Study Characterizing Age at Key Clinical Disease Indicators Among Patients With Friedreich Ataxia Using Health Administrative Claims Data in the United States

Impact of Dementia on Days at Home after Traumatic Brain Injury among Older Medicare Beneficiaries

AbstractBackgroundTraumatic brain injury (TBI) is a highly disabling event among older adults, contributing to disability in activities of daily living and elevated risk for nursing home admission. Older adults with Alzheimer’s disease and related dementias (ADRD) are more susceptible to TBI given increased fall incidence in this population, but how ADRD influences long‐term disability outcomes after TBI has been challenging to ascertain because of small sample sizes in prior work. Home time, a claims‐based measure of time spent at home and not hospitalized following a health‐related event, may aid in determining the association of ADRD and post‐TBI outcomes among older adults. Thus, the objective of this study was to examine the impact of TBI on home time among older adults with ADRD.MethodsWe conducted a retrospective cohort study using a random sample of Medicare administrative claims data for years 2010‐2018. Community‐dwelling Medicare beneficiaries aged ≥65 years hospitalized with a primary diagnosis of TBI were included and ADRD was defined at baseline (i.e. before TBI) using diagnostic codes. Home time was calculated by subtracting the number of days spent in a care environment or deceased from the total follow‐up available for each of the twelve follow‐up months post‐TBI. We modelled monthly days at home as a function of ADRD and covariates using generalized estimating equations.ResultsThere were 20,350 participants in the study, of whom 6,683 (33.8%) had a pre‐existing diagnosis of ADRD. In regression models that accounted for pre‐TBI days at home, demographic and clinical characteristics, and probability of survival, beneficiaries with ADRD spent fewer days at home following TBI compared to beneficiaries without ADRD (rate ratio 0.90; 95% confidence interval 0.89, 0.92). This disparity was equivalent to almost 50 fewer days spent at home over the year following TBI.ConclusionTBI resulted in a significant loss of home time over the year following injury, a loss markedly worse among older adults with ADRD.

CO105 A Retrospective Study Characterizing Age at Loss of Ambulation Among Patients With Friedreich Ataxia Using Health Administrative Claims Data in the United States

CO105 A Retrospective Study Characterizing Age at Loss of Ambulation Among Patients With Friedreich Ataxia Using Health Administrative Claims Data in the United States

An Economic Evaluation of the Relationship Between Glycemic Control and Total Healthcare Costs for Adults with Type2 Diabetes: Retrospective Cohort Study.

Glycemic control is associated with better outcomes among individuals with type2 diabetes (T2D). This research examines total US all-cause medical costs for adults with T2D with recommended glycemic control (HbA1c < 7%) compared to poor glycemic control (HbA1c ≥ 7%). The study used administrative claims data linked to HbA1c laboratory test results from January1, 2015 through June30, 2021 to identify adults with T2D with a recorded HbA1c test. Patients with recommended glycemic control at index date were propensity score matched to patients with poor glycemic control. General linear models and two-part models were used to compare all-cause outpatient, drug, acute care and total costs for 1year post index date. The study included 59,830 propensity-matched individuals. Results indicate that recommended glycemic control, compared to poor glycemic control, was associated with statistically significantly lower all-cause acute care ($23,868 ± $21,776 vs. $24,352 ± $22,223), drug ($10,277 ± $14,671 vs. $10,540 ± $14,928), and total medical costs ($41,381 ± $42,757 vs. $42,054 ± $43,422) but significantly higher outpatient costs ($7290 ± $12,028 vs. $7026 ± $11,587) (all p < 0.0001). Sensitivity analyses examined results based upon alternative HbA1c thresholds of ≤ 6.5% and < 8%. Results were generally robust to alternative HbA1c thresholds, with higher HbA1c thresholds associated with higher all-cause total costs as well as increased savings for having HbA1c below threshold. Glycemic control was associated with significantly lower all-cause total, drug, and acute care medical costs. Given the high prevalence of T2D in the USA, our results suggest potential economic benefits associated with glycemic control for healthcare providers.

A Geospatial Analysis of Social and Structural Determinants of Health and High HIV Prevalence in Alabama, USA.

This study utilizes geospatial analytic techniques to examine HIV hotspots in Alabama leveraging Medicaid utilization data. This cross-sectional study leveraged Medicaid utilization data from Alabama's 67 counties, averaging 9,861 Medicaid recipients aged > 18 years old per county. We used Alabama Medicaid administrative claims data from January 1, 2016, to December 31, 2020, to identify individuals with HIV. Using Microsoft SQL Server, we obtained the average annual count of HIV Medicaid claims in each of the 67 Alabama counties (numerator) and the number of adult Medicaid recipients in each county (denominator), and standardized with a multiplier of 100,000. We also examined several other area-level summary variables (e.g., non-high school completion, income greater than four times the federal poverty level, social associations, urbanicity/rurality) as social and structural determinants of health. County-boundary choropleth maps were created representing the geographic distribution of HIV rates per 100,000 adult Medicaid recipients in Alabama. Leveraging ESRI ArcGIS and local indicators of spatial association (LISA), results were examined using local Moran's I to identify geographic hotspots. Eleven counties had HIV rates higher than 100 per 100,000. Three were hotspots. Being an HIV hotspot was significantly associated with relatively low educational attainment and less severe poverty than other areas in the state. Findings suggesting that the HIV clusters in Alabama were categorized by significantly less severe poverty and lower educational attainment can aid ongoing efforts to strategically target resources and end the HIV epidemic in U.S.' Deep South.

Renin-angiotensin-aldosterone system inhibitors are associated with improved paclitaxel-induced peripheral neuropathy in lung cancer: a study using administrative claims data.

Chemotherapy-induced peripheral neuropathy (CIPN) has been reported to reduce patients' quality of life and impair cancer treatment by causing anticancer drug withdrawal or interruption. However, there are currently no effective methods for the prevention of CIPN. Renin-angiotensin-aldosterone system (RAAS) inhibitors may be associated with a reduced risk of developing oxaliplatin-induced peripheral neuropathy, and it would be valuable to examine whether they have the same effect on CIPN caused by other anticancer drugs. Our study explored the potential preventive effects of RAAS inhibitors on preventing paclitaxel-induced peripheral neuropathy (PIPN). An exploratory cohort study was conducted using commercially available administrative claims data on lung cancer patients treated with paclitaxel-based chemotherapy. Cumulative paclitaxel doses, RAAS inhibitor prescriptions, and incidences of PIPN were identified using patient medical records. Fine-Gray analyses with death as a competing risk were performed. A propensity score approach was applied to address the problem of confounding. Patients with lung cancer who received paclitaxel-based chemotherapy were classified into users of RAAS inhibitor (n = 1320) and non-users of RAAS inhibitor (n = 4566). The doses of RAAS inhibitors in our study were similar to those commonly used to treat hypertension. The PIPN incidence was significantly lower in users of RAAS inhibitor than in the non-users of RAAS inhibitor (sub-distribution hazard ratio, 0.842; 95% confidence interval, 0.762-0.929). The result was consistent in various sensitivity analyses and important subgroup analyses. RAAS inhibitors at doses commonly used for hypertension were associated with a reduced incidence of PIPN in patients with lung cancer.

ATRAcTR (Authentic Transparent Relevant Accurate Track-Record): a screening tool to assess the potential for real-world data sources to support creation of credible real-world evidence for regulatory decision-making

Adoption and use of real-world data (RWD) for decision-making has been complicated by concerns regarding whether RWD was fit-for-purpose or was of sufficient validity to support the creation of credible RWE. This has greater urgency as regulatory agencies begin to use real world evidence (RWE) to inform decisions about treatment effectiveness. Researchers need an efficient and systematic method to screen the quality of RWD sources considered for use in studies of effectiveness and safety. Based on a literature review we developed a listing of screening criteria that have been previously proposed to assess the quality of RWD sources. We also developed an additional criterion based on Modern Validity Theory. While there has occurred some convergence of conceptual frameworks to assess data quality (DQ) and there is much agreement on specific assessment criteria, consensus has yet to emerge on how to assess whether a specific RWD source is reliable and fit-for-purpose. To create a user-friendly tool to assess whether RWD sources may have sufficient quality to support a well-designed RWE study for submission to a regulatory authority, we grouped the quality criteria with a view to harmonize published frameworks and to be consistent with how researchers generally evaluate existing RWD sources for research that they intend to submit to regulatory agencies. Screening data quality criteria were grouped into five dimensions after a comprehensive literature review via PubMed: authenticity, transparency, relevance, accuracy, and track record. The resultant tool was tested for its response burden using a hypothetical administrative claims data source. Providing responses to the screening criteria required only few hours effort by an experienced data source manager. Thus, the tool should not be an onerous burden on data source providers if asked by prospective researchers to provide the required information. Assessing whether a particular data source is fit-for-purpose will be facilitated by the use of this tool, but it will not be sufficient by itself. Fit-for-purpose judgements will still require further careful consideration based on the context and the specific scientific question of interest. Unlike prior DQ frameworks (DQF), the track record dimension of the tool adds the consideration of experience with RWD sources consistent with Modern Validity Theory. However, the tool does not address issues of study design and analysis that are critical to regulatory agencies in evaluating the robustness and credibility of the real-world evidence generated.