Predicting risk of suicidal behavior from insurance claims data vs. linked data from insurance claims and electronic health records.

PURPOSE: Observational studies assessing effects of medical products on suicidal behavior often rely on health record data to account for pre-existing risk. We assess whether high-dimensional models predicting suicide risk using data derived from insurance claims and electronic health records (EHRs) are superior to models using data from insurance claims alone. METHODS: Data were from seven large health systems identified outpatient mental health visits by patients aged 11 or older between 1/1/2009 and 9/30/2017. Data for the 5 years prior to each visit identified potential predictors of suicidal behavior typically available from insurance claims (e.g., mental health diagnoses, procedure codes, medication dispensings) and additional potential predictors available from EHRs (self-reported race and ethnicity, responses to Patient Health Questionnaire or PHQ-9 depression questionnaires). Nonfatal self-harm events following each visit were identified from insurance claims data and fatal self-harm events were identified by linkage to state mortality records. Random forest models predicting nonfatal or fatal self-harm over 90 days following each visit were developed in a 70% random sample of visits and validated in a held-out sample of 30%. Performance of models using linked claims and EHR data was compared to models using claims data only. RESULTS: Among 15 845 047 encounters by 1 574 612 patients, 99 098 (0.6%) were followed by a self-harm event within 90 days. Overall classification performance did not differ between the best-fitting model using all data (area under the receiver operating curve or AUC = 0.846, 95% CI 0.839-0.854) and the best-fitting model limited to data available from insurance claims (AUC = 0.846, 95% CI 0.838-0.853). Competing models showed similar classification performance across a range of cut-points and similar calibration performance across a range of risk strata. Results were similar when the sample was limited to health systems and time periods where PHQ-9 depression questionnaires were recorded more frequently. CONCLUSION: Investigators using health record data to account for pre-existing risk in observational studies of suicidal behavior need not limit that research to databases including linked EHR data.

Who gets treated for influenza: A surveillance study from the US Food and Drug Administration's Sentinel System.

OBJECTIVE: We describe the baseline characteristics and complications of individuals with influenza in the US FDA's Sentinel System by antiviral treatment timing. DESIGN: Retrospective cohort design. PATIENTS: Individuals aged ≥6 months with outpatient diagnoses of influenza in June 2014-July 2017, 3 influenza seasons. METHODS: We identified the comorbidities, vaccination history, influenza testing, and outpatient antiviral dispensings of individuals with influenza using administrative claims data from 13 data partners including the Centers for Medicare and Medicaid Services, integrated delivery systems, and commercial health plans. We assessed complications within 30 days: hospitalization, oxygen use, mechanical ventilation, critical care, ECMO, and death. RESULTS: There were 1,090,333 influenza diagnoses in 2014-2015; 1,005,240 in 2016-2017; and 578,548 in 2017-2018. Between 49% and 55% of patients were dispensed outpatient treatment within 5 days. In all periods >80% of treated individuals received treatment on the day of diagnosis. Those treated on days 1-5 after diagnosis had higher prevalences of diabetes, chronic obstructive pulmonary disease, asthma, and obesity compared to those treated on the day of diagnosis or not treated at all. They also had higher rates of hospitalization, oxygen use, and critical care. In 2014-2015, among those aged ≥65 years, the rates of hospitalization were 45 per 1,000 diagnoses among those treated on day 0; 74 per 1,000 among those treated on days 1-5; and 50 per 1,000 among those who were untreated. CONCLUSIONS: In a large, national analysis, approximately half of people diagnosed with influenza in the outpatient setting were treated with antiviral medications. Delays in outpatient dispensed treatment were associated with higher prevalence of comorbidities and higher rates of complication.

Real-world Use of and Spending on New Oral Targeted Cancer Drugs in the US, 2011-2018.

Importance: Launch prices of new cancer drugs in the US have substantially increased in recent years despite growing concerns about the quantity and quality of evidence supporting their approval by the US Food and Drug Administration (FDA). Objective: To assess the use of and spending on new oral targeted cancer drugs among US residents with employer-sponsored insurance between 2011 and 2018, stratified by the strength of available evidence of benefit. Design, Setting, and Participants: In this cross-sectional study, dispensing claims for oral targeted cancer drugs first approved by the FDA between January 1, 2011, and December 31, 2018, were analyzed. The number of patients with drugs dispensed and the total payment for all claims were aggregated by calendar year, and these outcomes were arrayed according to evidence underlying FDA approvals, including pivotal study design (availability of randomized clinical trials) and overall survival (OS) benefit, as documented in drug labels. This study was conducted from July 17, 2019, to July 23, 2021. Main Outcomes and Measures: Annual and cumulative numbers of patients who had dispensing events, and annual and cumulative sums of payment for eligible drugs. Results: Of 37 348 patients who had at least 1 of the 44 new oral targeted drugs dispensed between 2011 and 2018, 21 324 were men (57.1%); mean (SD) age was 64.1 (13.1) years. Most individuals (36 246 [97.0%]) received drugs for which evidence from randomized clinical trials existed; however, a growing share of patients received drugs without documented OS benefit during the study period: from 12.7% in 2011 to 58.8% in 2018. Cumulative spending on all sample drugs totaled $3.5 billion by the end of 2018, of which 96.8% was spent on drugs that were approved based on a pivotal randomized clinical trial. Cumulative spending on drugs without documented OS benefit ($1.8 billion [51.6%]) surpassed that on drugs with documented OS benefit ($1.7 billion [48.4%]) by the end of 2018. Conclusions and Relevance: The findings of this cross-sectional study suggest that drugs used for treatment of cancer without documented OS benefits are adopted in the health system and account for substantial spending.

Prevalence of prescription opioid use during pregnancy in eight US health plans during 2001-2014.

PURPOSE: To estimate prevalence of prescription opioid use during pregnancy in eight US health plans during 2001-2014. METHODS: We conducted a cohort study of singleton live birth deliveries. Maternal characteristics were ascertained from health plan and/or birth certificate data and opioids dispensed during pregnancy from health plan pharmacy records. Prevalence of prescription opioid use during pregnancy was calculated for any use, cumulative days of use, and number of dispensings. RESULTS: We examined prevalence of prescription opioid use during pregnancy in each health plan. Tennessee Medicaid had appreciably greater prevalence of use compared to the seven other health plans. Thus, results for the two groups were reported separately. In the seven health plans (n = 587 093 deliveries), prevalence of use during pregnancy was relatively stable at 9%-11% throughout 2001-2014. In Tennessee Medicaid (n = 256 724 deliveries), prevalence increased from 29% in 2001 to a peak of 36%-37% in 2004-2010, and then declined to 28% in 2014. Use for ≥30 days during pregnancy was stable at 1% in the seven health plans and increased from 2% to 7% in Tennessee Medicaid during 2001-2014. Receipt of ≥5 opioid dispensings during pregnancy increased in the seven health plans (0.3%-0.6%) and Tennessee Medicaid (3%-5%) during 2001-2014. CONCLUSION: During 2001-2014, prescription opioid use during pregnancy was more common in Tennessee Medicaid (peak prevalence in late 2000s) compared to the seven health plans (relatively stable prevalence). Although a small percentage of women had opioid use during pregnancy for ≥30 days or ≥ 5 dispensings, they represent thousands of women during 2001-2014.

A COVID-19-ready public health surveillance system: The Food and Drug Administration's Sentinel System.

The US Food and Drug Administration's Sentinel System was established in 2009 to use routinely collected electronic health data for improving the national capability to assess post-market medical product safety. Over more than a decade, Sentinel has become an integral part of FDA's surveillance capabilities and has been used to conduct analyses that have contributed to regulatory decisions. FDA's role in the COVID-19 pandemic response has necessitated an expansion and enhancement of Sentinel. Here we describe how the Sentinel System has supported FDA's response to the COVID-19 pandemic. We highlight new capabilities developed, key data generated to date, and lessons learned, particularly with respect to working with inpatient electronic health record data. Early in the pandemic, Sentinel developed a multi-pronged approach to support FDA's anticipated data and analytic needs. It incorporated new data sources, created a rapidly refreshed database, developed protocols to assess the natural history of COVID-19, validated a diagnosis-code based algorithm for identifying patients with COVID-19 in administrative claims data, and coordinated with other national and international initiatives. Sentinel is poised to answer important questions about the natural history of COVID-19 and is positioned to use this information to study the use, safety, and potentially the effectiveness of medical products used for COVID-19 prevention and treatment.

Translating claims-based CHA2 DS2 -VaSc and HAS-BLED to ICD-10-CM: Impacts of mapping strategies.

PURPOSE: The CHA2 DS2 -VaSc and HAS-BLED risk scores are commonly used in the studies of oral anticoagulants (OACs). The best ways to map these scores to the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes is unclear, as is how they perform in various types of OAC users. We aimed to assess the distributions of CHA2 DS2 -VaSc and HAS-BLED scores and C-statistics for outcome prediction in the ICD-10-CM era using different mapping strategies. METHODS: We compared the distributions of CHA2 DS2 -VaSc and HAS-BLED scores from various mapping strategies in atrial fibrillation patients before, during, and after ICD-10-CM transition. We estimated the C-statistics predicting the 90-day risk of hospitalized stroke (for CHA2 DS2 -VaSc) or hospitalized bleeding (for HAS-BLED) in patients identified at least 6 months after the ICD-10-CM transition, overall and by anticoagulant type. RESULTS: Forward-backward mapping produced higher CHA2 DS2 -VaSc and HAS-BLED scores in the ICD-10-CM era compared to the ICD-9-CM era: the mean difference was 0.074 (95% confidence interval 0.064-0.085) for CHA2 DS2 -VaSc and 0.055 (0.048-0.062) for HAS-BLED. Both scores had higher C-statistics in patients taking no OACs (0.697 [0.677-0.717] for CHA2 DS2 -VaSc; 0.719 [0.702-0.737] for HAS-BLED) or direct OACs (0.695 [0.654-0.735] for CHA2 DS2 -VaSc; 0.700 [0.673-0.728] for HAS-BLED) than those taking warfarin (0.655 [0.613-0.697] for CHA2 DS2 -VaSc; 0.663 [0.6320.695] for HAS-BLED). CONCLUSIONS: Existing mapping strategies generally preserved the distributions of CHA2 DS2 -VaSc and HAS-BLED scores after ICD-10-CM transition. Both scores performed better in patients on no OACs or direct OACs than patients on warfarin.

Use of Antidiabetic drugs during pregnancy among U.S. women with Livebirth deliveries in the Mini-Sentinel system.

BACKGROUND: As the prevalence of diabetes mellitus increases in the population, the exposure to antidiabetic drugs (ADDs) during pregnancies is expected to grow, as has been seen over the last decade. The objective of this study was to estimate the prevalence of ADD use during pregnancy among women in the Mini-Sentinel Distributed Database (MSDD) who delivered a liveborn infant. METHODS: We identified qualifying livebirth pregnancies among women aged 10 to 54 years in the MSDD from 2001 to 2013. ADD use was estimated using outpatient pharmacy dispensing claims and days-supplied among three cohorts: all livebirth pregnancies, pregnancies among women with pre-existing diabetes, and pregnancies among women without prior ADD use. RESULTS: Among the 1.9 million pregnancies in the MSDD that resulted in a livebirth from 2001 to 2013, 4.4% were exposed to an ADD. Of the 15,606 pregnancies (0.8%) with pre-existing diabetes, 92.8% were also exposed during the pregnancy period. The most commonly used product in these pregnancies was insulin (75.6% of pregnancies). In contrast, in pregnancies of women without prior ADD use, the most commonly used products were glyburide and insulin, and most of these users were diagnosed with gestational diabetes. CONCLUSIONS: Patterns of ADD use during pregnancy described here, along with changes in disease incidence and management, highlight the importance of continuing surveillance of ADD utilization patterns and examining the safety and effectiveness of these products in pregnancy.

Identification of potential drug name confusion errors in the Sentinel System.

PURPOSE: In July 2015, the US Food and Drug Administration (FDA) published a drug safety communication regarding errors in prescribing and dispensing of the antidepressant Brintellix (vortioxetine) and the antiplatelet Brilinta (ticagrelor) that arose due to proprietary drug name confusion. Brintellix is indicated for major depressive disorder; Brilinta is indicated to reduce cardiovascular death, myocardial infarction, and stroke in patients with acute coronary syndrome or history of myocardial infarction. Brintellix was renamed to Trintellix in May 2016. Using Brilinta and Brintellix as a proof-of-concept feasibility use case, we assessed whether drug name confusion errors between the pair could be identified in electronic health care data via the combination of a claims-based algorithm and limited manual claims data review. METHODS: Using data from the Sentinel System, we defined potential errors as Brintellix users without an on- or off-label indication for Brintellix, without a dispensing for a drug with the same indications as Brintellix, and with an on- or off-label indication for Brilinta between -365 and +30 days after index Brintellix dispensing; the reverse was done for Brilinta. We manually reviewed claims profiles of potential cases. RESULTS: We identified 27 (0.1%) potential errors among 21 208 Brintellix users; 16 appeared to be likely errors based on claims profile review. Fifty-one (0.3%) of the 16 779 Brilinta users were identified as potential errors, and four appeared to be likely errors. CONCLUSIONS: A claims-based algorithm combined with manual review of claims profiles could identify potential drug name confusion errors, and narrow down likely errors that warrant further investigation.

Evaluating the use of bootstrapping in cohort studies conducted with 1:1 propensity score matching-A plasmode simulation study.

PURPOSE: Bootstrapping can account for uncertainty in propensity score (PS) estimation and matching processes in 1:1 PS-matched cohort studies. While theory suggests that the classical bootstrap can fail to produce proper coverage, practical impact of this theoretical limitation in settings typical to pharmacoepidemiology is not well studied. METHODS: In a plasmode-based simulation study, we compared performance of the standard parametric approach, which ignores uncertainty in PS estimation and matching, with two bootstrapping methods. The first method only accounted for uncertainty introduced during the matching process (the observation resampling approach). The second method accounted for uncertainty introduced during both PS estimation and matching processes (the PS reestimation approach). Variance was estimated based on percentile and empirical standard errors, and treatment effect estimation was based on median and mean of the estimated treatment effects across 1000 bootstrap resamples. Two treatment prevalence scenarios (5% and 29%) across two treatment effect scenarios (hazard ratio of 1.0 and 2.0) were evaluated in 500 simulated cohorts of 10 000 patients each. RESULTS: We observed that 95% confidence intervals from the bootstrapping approaches but not the standard approach, resulted in inaccurate coverage rates (98%-100% for the observation resampling approach, 99%-100% for the PS reestimation approach, and 95%-96% for standard approach). Treatment effect estimation based on bootstrapping approaches resulted in lower bias than the standard approach (less than 1.4% vs 4.1%) at 5% treatment prevalence; however, the performance was equivalent at 29% treatment prevalence. CONCLUSION: Use of bootstrapping led to variance overestimation and inconsistent coverage, while coverage remained more consistent with parametric estimation.

Misclassification in Assessment of First Trimester In-utero Exposure to Drugs Used Proximally to Conception: the Example of Letrozole Utilization for Infertility Treatment.

Letrozole is an aromatase inhibitor that has an unapproved use for ovulation induction with infertility. Because of the proximity of this use to conception, we selected letrozole to study the effect of 3 different methods for identifying the pregnancy start date and their impact on exposure misclassification. Using electronic health data from the US Sentinel database (2001-2015), we identified live-birth pregnancies conceived through in-vitro fertilization or intrauterine insemination. The pregnancy start was calculated using 1) a validated algorithm to estimate the last menstrual period (LMP), 2) LMP + 14 days (i.e., conception estimate), and 3) the fertility-procedure date. We identified 47,628 live-births after intrauterine insemination (n = 24,962) and in-vitro fertilization (n = 22,666), in which 2,458 (5.3%) mothers received letrozole. The algorithm-based conception estimate occurred within 14 days of the fertility procedure for 78.3% of pregnancies. Defining pregnancy start as LMP (45.7/1,000 pregnancies) or LMP + 14 days (12.7/1,000 pregnancies) overestimated letrozole exposure during pregnancy by 8.4-fold and 2.3-fold, respectively, compared with defining it at the date of the fertility procedure (5.5/1,000 pregnancies). While most studies of drug utilization in pregnancy use LMP as the conventional pregnancy start, this introduced substantial exposure misclassification in the example of letrozole. LMP + 14 days was less biased. Researchers should carefully consider the impact of the method for identifying the pregnancy start date on the potential for exposure misclassification.

The impact of FDA regulatory activities on incident dispensing of LABA-containing medication: 2005-2011.

OBJECTIVE: Evidence of safety issues associated with long-acting beta2-agonist (LABA) treatment has led to multiple regulatory activities by the U.S. Food and Drug Administration (FDA) on this class of medications. This study describes the impact of the regulatory activities on incident LABA-containing medication dispensing. METHODS: A monthly rolling cohort of asthma patients who were eligible to initiate a LABA-containing product was created in the Mini-Sentinel Distributed Database between January 2005 and June 2011. Cohorts of individuals who initiated LABA were examined for the changes in the proportions of single-ingredient to fixed-dose inhaled corticosteroid (ICS)-LABA initiators, appropriate initiation of LABA-containing products, and use of controller medications. The impact of the 2005 and 2010 FDA regulatory activities associated with LABA-containing products was measured using interrupted time series with segmented regression. RESULTS: LABA-containing product initiation was declining prior to the 2005 regulatory activities and continued to decline over the study period, accompanied by increased initiation of fixed dose ICS-LABA among LABA initiators. While the 2010 regulatory activities had no immediate impact on the proportion of LABA initiation in patients with prior controller medication dispensing and/or poor asthma control, there was an increasing positive trend toward LABA initiation in the appropriate patient population after the regulatory activities. CONCLUSION: The 2005 and 2010 FDA regulatory activities likely had an impact on communicating the safety concerns of LABA products. However, the impact cannot be viewed independent of scientific publications, guidelines for asthma treatment and other regulatory activities.

A proposed approach to accelerate evidence generation for genomic-based technologies in the context of a learning health system.

Genomic technologies should demonstrate analytical and clinical validity and clinical utility prior to wider adoption in clinical practice. However, the question of clinical utility remains unanswered for many genomic technologies. In this paper, we propose three building blocks for rapid generation of evidence on clinical utility of promising genomic technologies that underpin clinical and policy decisions. We define promising genomic tests as those that have proven analytical and clinical validity. First, risk-sharing agreements could be implemented between payers and manufacturers to enable temporary coverage that would help incorporate promising technologies into routine clinical care. Second, existing data networks, such as the Sentinel Initiative and the National Patient-Centered Clinical Research Network (PCORnet) could be leveraged, augmented with genomic information to track the use of genomic technologies and monitor clinical outcomes in millions of people. Third, endorsement and engagement from key stakeholders will be needed to establish this collaborative model for rapid evidence generation; all stakeholders will benefit from better information regarding the clinical utility of these technologies. This collaborative model can create a multipurpose and reusable national resource that generates knowledge from data gathered as part of routine care to drive evidence-based clinical practice and health system changes.

Safety assessment of niacin in the US Food and Drug Administration's mini-sentinel system.

PURPOSE: The Heart Protection Study 2-Treatment of HDL to Reduce the Incidence of Vascular Events (HPS2-THRIVE) trial found higher incidence rates of adverse reactions, including bleeding, in patients receiving the combination of extended-release niacin and laropiprant versus placebo. It is not known whether these adverse events are attributable to laropiprant, not approved in the USA, or to extended-release niacin. We compared rates of major gastrointestinal bleeding and intracranial hemorrhage among initiators of extended-release niacin and initiators of fenofibrate. METHODS: We used Mini-Sentinel (now Sentinel) to conduct an observational, new user cohort analysis. We included data from 5 Data Partners covering the period from January 1, 2007 to August 31, 2013. Individuals who initiated extended-release niacin were propensity score-matched to individuals who initiated fenofibrate. Within the matched cohorts, we used Cox proportional hazards models to compare rates of hospitalization for major gastrointestinal bleeding events and intracranial hemorrhage assessed using validated claims-based algorithms. RESULTS: A total of 234 242 eligible extended-release niacin initiators were identified, of whom 210 389 (90%) were 1:1 propensity score-matched to eligible fenofibrate initiators. In propensity score-matched analyses, no differences were observed between exposure groups in rates of major gastrointestinal bleeding (hazard ratio [HR], 0.98; 95% confidence interval [CI], 0.82 to 1.18) or intracranial hemorrhage (HR, 1.21; 95% CI, 0.66 to 2.22). Results were similar in pre-specified sensitivity and subgroup analyses. CONCLUSIONS: We did not observe evidence for an association between extended-release niacin versus fenofibrate and rates of major gastrointestinal bleeding or intracranial hemorrhage.

Trends in 30-day readmission rates after COPD hospitalization, 2006-2012.

BACKGROUND: Reduction in 30-day readmission rate after chronic obstructive pulmonary disease (COPD)-related hospitalization is a national objective. However, little is known about trends in readmission rates in recent years, particularly in priority populations defined by the Agency for Healthcare Research and Quality (AHRQ)(e.g., the elderly, women, racial/ethnic minorities, low-income and rural populations, and populations with chronic illnesses). METHODS: We conducted a retrospective cohort study using data from the State Inpatient Database of eight geographically-dispersed US states (Arkansas, California, Florida, Iowa, Nebraska, New York, Utah, and Washington) from 2006 through 2012. We identified all COPD-related hospitalizations by patients ?40 years old. The primary outcome was any-cause readmission within 30 days of discharge from the index hospitalization for COPD. RESULTS: From 2006 to 2012, a total of 845,465 hospitalizations at risk for 30-day readmissions were identified. Overall, 30-day readmission rate for COPD-related hospitalization decreased modestly from 20.0% in 2006 to 19.2% in 2012, an 0.8% absolute decrease (OR 0.991, 95%CI 0.989-0.995, Ptrend<0.001). This modest decline remained statistically significant after adjusting for patient demographics and comorbidities (adjusted OR 0.981, 95%CI 0.977-0.984, Ptrend<0.001). Similar to the overall population, the readmission rate over the 7-year period remained persistently high in most of AHRQ-defined priority populations. CONCLUSIONS: Our observations provide a benchmark for future investigation of the impact of Hospital Readmissions Reduction Program on readmissions after COPD hospitalization. Our findings encourage researchers and policymakers to develop effective strategies aimed at reducing readmissions among patients with COPD in an already-stressed healthcare system.

Stakeholders' views on data sharing in multicenter studies.

AIM: To understand stakeholders' views on data sharing in multicenter comparative effectiveness research studies and the value of privacy-protecting methods. MATERIALS & METHODS: Semistructured interviews with five US stakeholder groups. RESULTS: We completed 11 interviews, involving patients (n = 15), researchers (n = 10), Institutional Review Board and regulatory staff (n = 3), multicenter research governance experts (n = 2) and healthcare system leaders (n = 4). Perceptions of the benefits and value of research were the strongest influences toward data sharing; cost and security risks were primary influences against sharing. Privacy-protecting methods that share summary-level data were acknowledged as being appealing, but there were concerns about increased cost and potential loss of research validity. CONCLUSION: Stakeholders were open to data sharing in multicenter studies that offer value and minimize security risks.

Trends in elective labor induction for six United States health plans, 2001-2007.

BACKGROUND: To describe trends in labor induction, including elective induction, from 2001 to 2007 for six U.S. health plans and to examine the validity of induction measures derived from birth certificate and health plan data. METHODS: This retrospective cohort study included 339,123 deliveries at 35 weeks' gestation or greater. Linked health plan and birth certificate data provided information about induction, maternal medical conditions, and pregnancy complications. Induction was defined from diagnosis and procedure codes and birth certificate data and considered elective if no accepted indication was coded. We calculated induction prevalence across health plans and years. At four health plans, we reviewed medical records to validate induction measures. RESULTS: Based on electronic data, induction prevalence rose from 28% in 2001 to 32% in 2005, then declined to 29% in 2007. The trend was driven by changes in the prevalence of apparent elective induction, which rose from 11% in 2001 to 14% in 2005 and then declined to 11% in 2007. The trend was similar for subgroups by parity and gestational age. Elective induction prevalence varied considerably across plans. On review of 86 records, 36% of apparent elective inductions identified from electronic data were confirmed as valid. CONCLUSIONS: Elective induction appeared to peak in 2005 and then decline. The decrease may reflect quality improvement initiatives or changes in policies, patient or provider attitudes, or coding practices. The low validation rate for measures of elective induction defined from electronic data has important implications for existing quality measures and for research studies examining induction's outcomes.

Data quality assessment for comparative effectiveness research in distributed data networks.

BACKGROUND: Electronic health information routinely collected during health care delivery and reimbursement can help address the need for evidence about the real-world effectiveness, safety, and quality of medical care. Often, distributed networks that combine information from multiple sources are needed to generate this real-world evidence. OBJECTIVE: We provide a set of field-tested best practices and a set of recommendations for data quality checking for comparative effectiveness research (CER) in distributed data networks. METHODS: Explore the requirements for data quality checking and describe data quality approaches undertaken by several existing multi-site networks. RESULTS: There are no established standards regarding how to evaluate the quality of electronic health data for CER within distributed networks. Data checks of increasing complexity are often used, ranging from consistency with syntactic rules to evaluation of semantics and consistency within and across sites. Temporal trends within and across sites are widely used, as are checks of each data refresh or update. Rates of specific events and exposures by age group, sex, and month are also common. DISCUSSION: Secondary use of electronic health data for CER holds promise but is complex, especially in distributed data networks that incorporate periodic data refreshes. The viability of a learning health system is dependent on a robust understanding of the quality, validity, and optimal secondary uses of routinely collected electronic health data within distributed health data networks. Robust data quality checking can strengthen confidence in findings based on distributed data network.