High-dimensional generalized median adaptive lasso with application to omics data.

Recently, there has been a growing interest in variable selection for causal inference within the context of high-dimensional data. However, when the outcome exhibits a skewed distribution, ensuring the accuracy of variable selection and causal effect estimation might be challenging. Here, we introduce the generalized median adaptive lasso (GMAL) for covariate selection to achieve an accurate estimation of causal effect even when the outcome follows skewed distributions. A distinctive feature of our proposed method is that we utilize a linear median regression model for constructing penalty weights, thereby maintaining the accuracy of variable selection and causal effect estimation even when the outcome presents extremely skewed distributions. Simulation results showed that our proposed method performs comparably to existing methods in variable selection when the outcome follows a symmetric distribution. Besides, the proposed method exhibited obvious superiority over the existing methods when the outcome follows a skewed distribution. Meanwhile, our proposed method consistently outperformed the existing methods in causal estimation, as indicated by smaller root-mean-square error. We also utilized the GMAL method on a deoxyribonucleic acid methylation dataset from the Alzheimer's disease (AD) neuroimaging initiative database to investigate the association between cerebrospinal fluid tau protein levels and the severity of AD.

Sensitivity analysis of individual treatment effects: A robust conformal inference approach.

We propose a model-free framework for sensitivity analysis of individual treatment effects (ITEs), building upon ideas from conformal inference. For any unit, our procedure reports the Γ-value, a number which quantifies the minimum strength of confounding needed to explain away the evidence for ITE. Our approach rests on the reliable predictive inference of counterfactuals and ITEs in situations where the training data are confounded. Under the marginal sensitivity model of [Z. Tan, J. Am. Stat. Assoc. 101, 1619-1637 (2006)], we characterize the shift between the distribution of the observations and that of the counterfactuals. We first develop a general method for predictive inference of test samples from a shifted distribution; we then leverage this to construct covariate-dependent prediction sets for counterfactuals. No matter the value of the shift, these prediction sets (resp. approximately) achieve marginal coverage if the propensity score is known exactly (resp. estimated). We describe a distinct procedure also attaining coverage, however, conditional on the training data. In the latter case, we prove a sharpness result showing that for certain classes of prediction problems, the prediction intervals cannot possibly be tightened. We verify the validity and performance of the methods via simulation studies and apply them to analyze real datasets.

Scalable kernel balancing weights in a nationwide observational study of hospital profit status and heart attack outcomes.

Weighting is a general and often-used method for statistical adjustment. Weighting has two objectives: first, to balance covariate distributions, and second, to ensure that the weights have minimal dispersion and thus produce a more stable estimator. A recent, increasingly common approach directly optimizes the weights toward these two objectives. However, this approach has not yet been feasible in large-scale datasets when investigators wish to flexibly balance general basis functions in an extended feature space. To address this practical problem, we describe a scalable and flexible approach to weighting that integrates a basis expansion in a reproducing kernel Hilbert space with state-of-the-art convex optimization techniques. Specifically, we use the rank-restricted Nyström method to efficiently compute a kernel basis for balancing in nearly linear time and space, and then use the specialized first-order alternating direction method of multipliers to rapidly find the optimal weights. In an extensive simulation study, we provide new insights into the performance of weighting estimators in large datasets, showing that the proposed approach substantially outperforms others in terms of accuracy and speed. Finally, we use this weighting approach to conduct a national study of the relationship between hospital profit status and heart attack outcomes in a comprehensive dataset of 1.27 million patients. We find that for-profit hospitals use interventional cardiology to treat heart attacks at similar rates as other hospitals but have higher mortality and readmission rates.

Validation of algorithms in studies based on routinely collected health data: general principles.

Clinicians, researchers, regulators, and other decision-makers increasingly rely on evidence from real-world data (RWD), including data routinely accumulating in health and administrative databases. RWD studies often rely on algorithms to operationalize variable definitions. An algorithm is a combination of codes or concepts used to identify persons with a specific health condition or characteristic. Establishing the validity of algorithms is a prerequisite for generating valid study findings that can ultimately inform evidence-based health care. This paper aims to systematize terminology, methods, and practical considerations relevant to the conduct of validation studies of RWD-based algorithms. We discuss measures of algorithm accuracy; gold/reference standard; study size; prioritizing accuracy measures; algorithm portability; and implication for interpretation. Information bias is common in epidemiologic studies, underscoring the importance of transparency in decisions regarding choice and prioritizing measures of algorithm validity. The validity of an algorithm should be judged in the context of a data source, and one size does not fit all. Prioritizing validity measures within a given data source depends on the role of a given variable in the analysis (eligibility criterion, exposure, outcome or covariate). Validation work should be part of routine maintenance of RWD sources.

Prevalence and Characteristics of Adults with Difficult-to-Treat Rheumatoid Arthritis in a Large Patient Registry.

OBJECTIVES: An estimated 5-20% of patients with rheumatoid arthritis (RA) fail multiple treatments and are considered "difficult-to-treat" (D2T), posing a substantial clinical challenge for rheumatologists. A European Alliance of Associations for Rheumatology (EULAR) task force proposed a definition of D2T-RA in 2021. We applied EULAR's D2T definition in a cohort of patients with established RA to assess prevalence and we compared clinical characteristics of participants with D2T-RA with matched comparisons. METHODS: Data from the longitudinal Brigham and Women's Hospital Rheumatoid Arthritis Sequential Study (BRASS) registry was used. Participants were classified as D2T if they met EULAR's definition. A comparison group of non-D2T RA patients were matched 2:1 to every D2T patient, and differences in characteristics were evaluated in descriptive analyses. Prevalence rates of D2T were estimated using Poisson regression. RESULTS: We estimated the prevalence of D2T-RA to be 14.4 (95% CI: 12.8-16.3 per 100 persons) among 1,581 participants with RA, and 22.3 (95% CI: 19.9-25.0 per 100 persons) among 1,021 who were biologic/targeted synthetic DMARD experienced. We observed several differences in demographics, comorbidities, and RA disease activity between D2T-RA and non-D2T RA comparisons. Varying EULAR sub-criteria among all participants in BRASS resulted in a range of D2T-RA prevalence rates, from 0.6-17.5 per 100 persons. CONCLUSION: EULAR's proposed definition of D2T-RA identifies patients with RA who have not achieved treatment targets. Future research should explore heterogeneity in these patients and evaluate outcomes to inform the design of future studies aimed at developing more effective RA management protocols.

Ultrasound of the forefeet besides the hands in patients at risk for rheumatoid arthritis: is it worth the effort? A longitudinal cohort study.

OBJECTIVE: Ultrasound (US) can detect subclinical joint-inflammation in patients with clinically suspect arthralgia (CSA), which is valuable as predictor for rheumatoid arthritis (RA) development. In most research protocols both hands and forefeet are scanned, but it is unclear if US of the forefeet has additional value for predicting RA, especially since synovial hypertrophy in MTP-joints of healthy individuals is also common. To explore the possibility to omit scanning of the forefeet we determined if US of the forefeet is of additional predictive value for RA-development in CSA patients. METHODS: CSA patients of two independent cohorts underwent US of the hands and forefeet. We analyzed the association between RA-development and US-positivity for the full US-protocol, the full US-protocol with correction for Gray Scale(GS)-findings in the forefeet of healthy and the protocol without-forefeet. RESULTS: In total, 298 CSA patients were studied. In patients with a positive US, subclinical joint-inflammation was mostly present in the hands (90-86%). Only 10-14% of patients had subclinical joint-inflammation solely in the forefeet. US-positivity was associated with inflammatory arthritis development in both cohorts, with HRs 2.6(95%CI 0.9-7.5) and 3.1(95%CI 1.5-6.4) for the full protocol, 3.1(95%CI 1.3-7.7) and 2.7(95%CI 1.3-5.4) for the full US-protocol with correction, and 3.1(95%CI 1.4-6.9) and 2.8(95%CI 1.4-5.6) without the forefeet. AUROCs were equal across both cohorts. CONCLUSION: The forefeet can be omitted when US is used for the prediction of RA-development in CSA patients. This is due to the finding that subclinical joint-inflammation in the forefeet without concomitant inflammation in the hands is infrequent.

Racial and ethnic associations with interstitial lung disease and healthcare utilization in patients with systemic sclerosis.

OBJECTIVE: Racial and ethnic differences in presentation and outcomes have been reported in systemic sclerosis (SSc) and SSc-interstitial lung disease (ILD). However, prior studies have limited diversity. We aim to evaluate if there are racial/ethnic differences associated with ILD, time intervals between SSc and ILD and with emergency department (ED) visit or hospitalization rates. METHODS: Clinical and sociodemographic variables were extracted for 756 patients with SSc from longitudinal health records in an integrated health-system. Logistic regression models analyzed the association of covariates with ILD and age at SSc-ILD. Healthcare outcomes were analyzed with complementary log-log regression models. RESULTS: Overall, 33.7% of patients in the cohort had an ILD code, with increased odds for Asian (odds ratio [OR], 2.60; 95% confidence interval [CI], 1.29-5.28; p=0.008) compared with White patients. The predicted age in years of SSc-ILD was younger for Hispanic (estimate, -6.5; 95% CI, -13--0.21; p = 0.04) and Black/African American patients (-10; 95% CI -16--4.9; p < 0.001) compared with White patients. Black/African American patients were more likely to have an ILD code before an SSc code (59% compared with 20.6% of White patients), and the shortest interval from SSc to ILD (3 months). Black/African American (HR, 2.59; 95% CI 1.47-4.49; p = 0.001) and Hispanic patients (HR 2.29; 95% CI 1.37- 3.82; p = 0.002) had higher rates of an ED visit. CONCLUSION: We found that odds of SSc-ILD differed by racial/ethnic group, minoritized patients had earlier age of presentation, and greater rates of an ED visit.

Which advanced treatment should be used following the failure of a first-line anti-TNF in patients with rheumatoid arthritis? 15 years of evidence from the Quebec registry RHUMADATA.

BACKGROUND: Since 2000, advanced therapies (AT) have revolutionized the treatment of moderate to severe rheumatoid arthritis (RA). Randomized control trials as well as observational studies together with medication availability often determine second-line choices after the failure of first Tumor Necrosis Factor inhibitors (TNFi). This led to the observation that specific sequences provide better long-term effectiveness. We investigated which alternative medication offers the best long-term sustainability following the first TNFi failure in RA. METHODS: Data were extracted from RHUMADATA from January2007. Patients were followed until treatment discontinuation, loss to follow-up, or November 25, 2022. Kaplan-Meier and Cox regression models were used to compare discontinuation between groups. Missing data were imputed, and propensity scores were computed to reduce potential attribution bias. Complete, unadjusted, and propensity score-adjusted imputed data analyses were produced. RESULTS: 611 patients (320 treated with a TNFi and 291 treated with molecules having another mechanism of action (OMA)) were included. The mean age at diagnosis was 44.5 and 43.9 years, respectively. The median retention was 2.84 and 4.48 years for TNFi and OMAs groups. Using multivariable analysis, the discontinuation rate of the OMA group was significantly lower than TNFi (adjHR: 0.65; 95% CI: 0.44-0.94). This remained true for the PS-adjusted MI Cox models. In a stratified analysis, rituximab (adjHR: 0.39; 95% CI: 0.18-0.84) had better retention than TNFi after adjusting for patient characteristics. CONCLUSION: Switching to an OMA, especially rituximab, in patients with failure to a first TNFi appears to be the best strategy as a second line of therapy.

Effectiveness and safety of the combination of sodium-glucose transport protein 2 inhibitors and glucagon-like peptide-1 receptor agonists in patients with type 2 diabetes mellitus: a systematic review and meta-analysis of observational studies.

BACKGROUND: Randomized controlled trials and real-world studies suggest that combination therapy with sodium-glucose transport protein 2 inhibitors (SGLT2is) and glucagon-like peptide-1 receptor agonists (GLP-1RAs) is associated with improvement in fasting plasma glucose (FPG), glycated hemoglobin (HbA1c), systolic blood pressure (SBP), body mass index (BMI), and total cholesterol levels. However, a systematic review of available real-world evidence may facilitate clinical decision-making in the real-world scenario. This meta-analysis assessed the safety and effectiveness of combinations of SGLT2is + GLP-1RAs with a focus on their cardioprotective effects along with glucose-lowering ability in patients with type 2 diabetes mellitus (T2DM) in a real-world setting. METHODS: Electronic searches were performed in the PubMed/MEDLINE, PROQuest, Scopus, CINAHL, and Google Scholar databases. Qualitative analyses and meta-analyses were performed using the Joanna Briggs Institute SUMARI software package and Review Manager v5.4, respectively. RESULTS: The initial database search yielded 1445 articles; of these, 13 were included in this study. The analyses indicated that SGLT2is + GLP-1RAs combinations were associated with significantly lower all-cause mortality when compared with individual therapies (odds ratio [95% confidence interval [CI] 0.49 [0.41, 0.60]; p < 0.00001). Significant reductions in BMI (- 1.71 [- 2.74, - 0.67]; p = 0.001), SBP (- 6.35 [- 10.17, - 2.53]; p = 0.001), HbA1c levels (- 1.48 [- 1.75, - 1.21]; p < 0.00001), and FPG (- 2.27 [- 2.78, - 1.76]; p < 0.00001) were associated with the simultaneous administration of the combination. Changes in total cholesterol levels and differences between simultaneous and sequential combination therapies for this outcome were not significant. CONCLUSION: This systematic review and meta-analysis based on real-world data suggests that the combination of SGLT2is + GLP-1RAs is associated with lower all-cause mortality and favorable improvements in cardiovascular, renal, and glycemic measurements. The findings drive a call-to-action to incorporate this combination early and simultaneously in managing T2DM patients and achieve potential cardiovascular benefits and renal protection.

Effectiveness of cladribine compared to fingolimod, natalizumab, ocrelizumab and alemtuzumab in relapsing-remitting multiple sclerosis.

BACKGROUND: Comparisons between cladribine and other potent immunotherapies for multiple sclerosis (MS) are lacking. OBJECTIVES: To compare the effectiveness of cladribine against fingolimod, natalizumab, ocrelizumab and alemtuzumab in relapsing-remitting MS. METHODS: Patients with relapsing-remitting MS treated with cladribine, fingolimod, natalizumab, ocrelizumab or alemtuzumab were identified in the global MSBase cohort and two additional UK centres. Patients were followed for ⩾6/12 and had ⩾3 in-person disability assessments. Patients were matched using propensity score. Four pairwise analyses compared annualised relapse rates (ARRs) and disability outcomes. RESULTS: The eligible cohorts consisted of 853 (fingolimod), 464 (natalizumab), 1131 (ocrelizumab), 123 (alemtuzumab) or 493 (cladribine) patients. Cladribine was associated with a lower ARR than fingolimod (0.07 vs. 0.12, p = 0.006) and a higher ARR than natalizumab (0.10 vs. 0.06, p = 0.03), ocrelizumab (0.09 vs. 0.05, p = 0.008) and alemtuzumab (0.17 vs. 0.04, p < 0.001). Compared to cladribine, the risk of disability worsening did not differ in patients treated with fingolimod (hazard ratio (HR) 1.08, 95% confidence interval (CI) 0.47-2.47) or alemtuzumab (HR 0.73, 95% CI 0.26-2.07), but was lower for patients treated with natalizumab (HR 0.35, 95% CI 0.13-0.94) and ocrelizumab (HR 0.45, 95% CI 0.26-0.78). There was no evidence for a difference in disability improvement. CONCLUSION: Cladribine is an effective therapy that can be viewed as a step up in effectiveness from fingolimod, but is less effective than the most potent intravenous MS therapies.

The National Institutes of Health COVID-19 Neuro Databank/Biobank: Creation and Evolution.

INTRODUCTION: Diverse neurological conditions are reported associated with the SARS-CoV-2 virus; neurological symptoms are the most common conditions to persist after the resolution of acute infection, affecting 20% of patients 6 months after acute illness. The COVID-19 Neuro Databank (NeuroCOVID) was created to overcome the limitations of siloed small local cohorts to collect detailed, curated, and harmonized de-identified data from a large diverse cohort of adults with new or worsened neurological conditions associated with COVID-19 illness, as a scientific resource. METHODS: A Steering Committee including US and international experts meets quarterly to provide guidance. Initial study sites were recruited to include a wide US geographic distribution; academic and non-academic sites; urban and non-urban locations; and patients of different ages, disease severity, and comorbidities seen by a variety of clinical specialists. The NeuroCOVID REDCap database was developed, incorporating input from professional guidelines, existing common data elements, and subject matter experts. A cohort of eligible adults is identified at each site; inclusion criteria are: a new or worsened neurological condition associated with a COVID-19 infection confirmed by testing. De-identified data are abstracted from patients' medical records, using standardized common data elements and five case report forms. The database was carefully enhanced in response to feedback from site investigators and evolving scientific interest in post-acute conditions and their timing. Additional US and international sites were added, focusing on diversity and populations not already described in published literature. By early 2024, NeuroCOVID included over 2,700 patient records, including data from 16 US and 5 international sites. Data are being shared with the scientific community in compliance with NIH requirements. The program has been invited to share case report forms with the National Library of Medicine as an ongoing resource for the scientific community. CONCLUSION: The NeuroCOVID database is a unique and valuable source of comprehensive de-identified data on a wide variety of neurological conditions associated with COVID-19 illness, including a diverse patient population. Initiated early in the pandemic, data collection has been responsive to evolving scientific interests. NeuroCOVID will continue to contribute to scientific efforts to characterize and treat this challenging illness and its consequences.

Randomized controlled trials: not always the "gold standard" for evidence in obstetrics and gynecology.

Randomized controlled trials are considered the "gold standard" for therapeutic interventions, and it is not uncommon for sweeping changes in medical practice to follow positive results from such trials. However, randomized controlled trials are not without their limitations. Physicians frequently view randomized controlled trials as infallible, whereas they tend to dismiss evidence derived from sources other than randomized controlled trials as less credible or reliable. In several situations in obstetrics and gynecology, there are no randomized controlled trials to help guide the clinician. In these circumstances, it is important to evaluate the entire body of evidence including observational studies, rather than dismiss interventions altogether simply because no randomized controlled trials exist. Randomized controlled trials and observational studies should be viewed as complementary rather than at odds with each other. Some reversals in widely adopted clinical practice have recently been implemented following subsequent studies that contradicted the outcomes of major randomized controlled trials. The most notable of these was the withdrawal from the market of 17-hydroxyprogesterone caproate for preterm birth prevention. Such reversals could potentially have been averted if the inherent limitations of randomized controlled trials were carefully considered before implementing these universal practice changes. This Clinical Opinion underscores the limitations of an exclusive reliance on randomized controlled trials while disregarding other evidence in determining how best to care for patients. Solutions are proposed that advocate that clinicians adopt a more balanced perspective that considers the entirety of the available medical evidence and the individual patient characteristics, needs, and wishes.

'Spin' in urology non-randomised studies comparing therapeutic interventions: a temporal analysis.

OBJECTIVE: To determine the prevalence of 'spin' (i.e., reporting practices that distort the interpretation of results by positively reflecting negative findings or downplaying potential harms) strategies and level of spin in urological observational studies and whether the use of spin has changed over time. MATERIALS AND METHODS: MEDLINE and Embase were searched to identify observational studies comparing therapeutic interventions in the top five urology journals and major urological subspecialty journals, published between 2000 and 2001, 2010 and 2011, and 2020 and 2021. RESULTS: A total of 235 studies were included. Spin was identified in 81% of studies, with a median of two strategies per study. The most commonly used strategies were inadequate implication for clinical practice (30%), causal language or causal claim (29%), and use of linguistic spin (29%). Moderate to high levels of spin were found in 55% of conclusions. From 2000 to 2020, the average number of strategies used has significantly decreased each decade (H = 27.459, P < 0.001), and the median level of spin in conclusions was significantly lower in studies published in the 2020s and 2010s than in the 2000s (H = 11.649, P = 0.003). CONCLUSIONS: Our results suggest that 81% of urological observational studies comparing therapeutic interventions contained spin. Over the past two decades, the use of spin has significantly declined, but this remains an area for improvement, with 70% of included studies published in the 2020s employing spin. Medical writing should scrupulously avoid words or phrases that are not supported by data in the manuscript.

Optimal refinement of strata to balance covariates.

What is the best way to split one stratum into two to maximally reduce the within-stratum imbalance in many covariates? We formulate this as an integer program and approximate the solution by randomized rounding of a linear program. A linear program may assign a fraction of a person to each refined stratum. Randomized rounding views fractional people as probabilities, assigning intact people to strata using biased coins. Randomized rounding is a well-studied theoretical technique for approximating the optimal solution of certain insoluble integer programs. When the number of people in a stratum is large relative to the number of covariates, we prove the following new results: (i) randomized rounding to split a stratum does very little randomizing, so it closely resembles the linear programming relaxation without splitting intact people; (ii) the linear relaxation and the randomly rounded solution place lower and upper bounds on the unattainable integer programming solution; and because of (i), these bounds are often close, thereby ratifying the usable randomly rounded solution. We illustrate using an observational study that balanced many covariates by forming matched pairs composed of 2016 patients selected from 5735 using a propensity score. Instead, we form 5 propensity score strata and refine them into 10 strata, obtaining excellent covariate balance while retaining all patients. An R package optrefine at CRAN implements the method. Supplementary materials are available online.

G-formula for observational studies under stratified interference, with application to bed net use on malaria.

Assessing population-level effects of vaccines and other infectious disease prevention measures is important to the field of public health. In infectious disease studies, one person's treatment may affect another individual's outcome, that is, there may be interference between units. For example, the use of bed nets to prevent malaria by one individual may have an indirect effect on other individuals living in close proximity. In some settings, individuals may form groups or clusters where interference only occurs within groups, that is, there is partial interference. Inverse probability weighted estimators have previously been developed for observational studies with partial interference. Unfortunately, these estimators are not well suited for studies with large clusters. Therefore, in this paper, the parametric g-formula is extended to allow for partial interference. G-formula estimators are proposed for overall effects, effects when treated, and effects when untreated. The proposed estimators can accommodate large clusters and do not suffer from the g-null paradox that may occur in the absence of interference. The large sample properties of the proposed estimators are derived assuming no unmeasured confounders and that the partial interference takes a particular form (referred to as 'weak stratified interference'). Simulation studies are presented demonstrating the finite-sample performance of the proposed estimators. The Demographic and Health Survey from the Democratic Republic of the Congo is then analyzed using the proposed g-formula estimators to assess the effects of bed net use on malaria.

crossnma: An R package to synthesize cross-design evidence and cross-format data using network meta-analysis and network meta-regression.

BACKGROUND: Although aggregate data (AD) from randomised clinical trials (RCTs) are used in the majority of network meta-analyses (NMAs), other study designs (e.g., cohort studies and other non-randomised studies, NRS) can be informative about relative treatment effects. The individual participant data (IPD) of the study, when available, are preferred to AD for adjusting for important participant characteristics and to better handle heterogeneity and inconsistency in the network. RESULTS: We developed the R package crossnma to perform cross-format (IPD and AD) and cross-design (RCT and NRS) NMA and network meta-regression (NMR). The models are implemented as Bayesian three-level hierarchical models using Just Another Gibbs Sampler (JAGS) software within the R environment. The R package crossnma includes functions to automatically create the JAGS model, reformat the data (based on user input), assess convergence and summarize the results. We demonstrate the workflow within crossnma by using a network of six trials comparing four treatments. CONCLUSIONS: The R package crossnma enables the user to perform NMA and NMR with different data types in a Bayesian framework and facilitates the inclusion of all types of evidence recognising differences in risk of bias.

Designing and analyzing studies of coronavirus disease 2019 and post-acute sequelae of severe acute respiratory syndrome coronavirus 2 among immunocompromised individuals.

Observational studies of coronavirus disease 2019 (COVID-19) among transplant candidates and recipients remain important as immunocompromised patients formed a very small proportion of patients included in COVID-19 trials and large database analyses. We discuss methods that have been used in such analyses to evaluate the impact of vaccination on the risk of symptomatic COVID-19 in such patients and on the probability of developing post-acute sequelae of severe acute respiratory syndrome coronavirus 2 after the onset of infection. We also propose future directions for research and discuss the methods that will be useful to conduct such investigations. The study design and analytical issues that we consider have the potential to be helpful not only for COVID-19 research but also for other infections as well.

The association between circulating 25-hydroxyvitamin D and pancreatic cancer: a systematic review and meta-analysis of observational studies.

PURPOSE: The relationship between circulating 25-hydroxyvitamin D [25(OH)D] and pancreatic cancer has been well studied but remains unclear. The purpose of this study was to elucidate the association between circulating 25(OH)D and pancreatic cancer by using a meta-analytic approach. METHODS: PubMed, Embase, and Wed of Science databases were searched through October 15, 2022. A random or fixed-effects model was used to estimate the pooled odds ratio (OR), risk ratio (RR), hazard ratio (HR) and their 95% confidence intervals (CIs). RESULTS: A total of 16 studies including 529,917 participants met the inclusion criteria, of which 10 reported incidence and 6 reported mortality. For the highest versus lowest categories of circulating 25(OH)D, the pooled OR of pancreatic cancer incidence in case-control studies was 0.98 (95% CI 0.69-1.27), and the pooled HRs of pancreatic cancer mortality in cohort and case-control studies were 0.64 (95% CI 0.45-0.82) and 0.78 (95% CI 0.62-0.95), respectively. The leave-one-out sensitivity analyses found no outliers and Galbraith plots indicated no substantial heterogeneity. CONCLUSION: Evidence from this meta-analysis suggested that high circulating 25(OH)D levels may be associated with decreased mortality but not incidence of pancreatic cancer. Our findings may provide some clues for the treatment of pancreatic cancer and remind us to be cautious about widespread vitamin D supplementation for the prevention of pancreatic cancer.

Use of clinical classifications software to address ICD coding transition in large healthcare databases analyzed via high-dimensional propensity scores.

PURPOSE: The EUPAS26595 study characterized the rate of acute renal failure (ARF) in patients exposed to levetiracetam versus other antiepileptic drugs using healthcare claims data and a high-dimensional propensity score (hd-PS) for confounding adjustment. The data contained several coding systems by design and an update in International Classification of Diseases (ICD) coding dictionary. Such coding heterogeneity can affect the performance of hd-PS, and manually coding harmonization is not feasible. Our objective was to explore the impact of code aggregation via Clinical Classifications Software (CCS) on the analysis of a large claims-based database using hd-PS. METHODS: Patients with epilepsy, who were new-users of an antiepileptic drug, were identified from the IBM® MarketScan® Research Databases. We used CCS categories to harmonize coding and compared the results with other alternatives. Incidence rate ratios (IRRs) were computed using modified Poisson regression model with a robust variance estimator. RESULTS: For January 2008-October 2015 (before ICD update), 34 833 eligible patients initiated levetiracetam and 52 649 initiated a comparator drug; IRR (95% CI) for ARF for the hd-PS analysis was 1.34 (0.72-2.50) without CCS categories and 1.30 (0.71-2.39) with CCS categories. For January 2008-December 2017 (including ICD coding change), 45 672 eligible patients initiated levetiracetam and 64 664 initiated a comparator drug; IRR (95% CI) for the hd-PS analysis was 1.34 (0.78-2.29) without CCS categories and 1.37 (0.80-2.34) with CCS categories. CONCLUSIONS: Using single-level CCS categories to overcome differences in coding provides consistent results and can be used in studies that use large claims data and hd-PS for adjustment.

Metadata for Data dIscoverability aNd Study rEplicability in obseRVAtional Studies (MINERVA): Development and Pilot of a Metadata List and Catalogue in Europe.

PURPOSE: Metadata for data dIscoverability aNd study rEplicability in obseRVAtional studies (MINERVA), a European Medicines Agency-funded project (EUPAS39322), defined a set of metadata to describe real-world data sources (RWDSs) and piloted metadata collection in a prototype catalogue to assist investigators from data source discoverability through study conduct. METHODS: A list of metadata was created from a review of existing metadata catalogues and recommendations, structured interviews, a stakeholder survey, and a technical workshop. The prototype was designed to comply with the FAIR principles (findable, accessible, interoperable, reusable), using MOLGENIS software. Metadata collection was piloted by 15 data access partners (DAPs) from across Europe. RESULTS: A total of 442 metadata variables were defined in six domains: institutions (organizations connected to a data source); data banks (data collections sustained by an organization); data sources (collections of linkable data banks covering a common underlying population); studies; networks (of institutions); and common data models (CDMs). A total of 26 institutions were recorded in the prototype. Each DAP populated the metadata of one data source and its selected data banks. The number of data banks varied by data source; the most common data banks were hospital administrative records and pharmacy dispensation records (10 data sources each). Quantitative metadata were successfully extracted from three data sources conforming to different CDMs and entered into the prototype. CONCLUSIONS: A metadata list was finalized, a prototype was successfully populated, and a good practice guide was developed. Setting up and maintaining a metadata catalogue on RWDSs will require substantial effort to support discoverability of data sources and reproducibility of studies in Europe.