Misclassification Bias Research Articles

Harmful Effects of Ambient Nitrogen Dioxide on Atopic Dermatitis: Comparison of Exposure Assessment Based on Monitored Concentrations and Modeled Estimates

Precise exposure assessment of air pollutants is crucial in epidemiologic studies to ensure valid estimates of health effects. We conducted a longitudinal study to evaluate the role of air quality monitoring (AQM) measurements and high-resolution modeling outcomes focusing on nitrogen dioxide (NO2) exposure and atopic dermatitis (AD). A total of 128 young children with AD in Seoul Metropolitan Area, Korea, were recruited as a panel. We estimated the participants’ exposure to NO2 for four months, from 1 April through 31 July 2014 based on (1) monitored levels from 60 AQM stations located at varying distances from residential areas (AQM station-based NO2, AQM-NO2) and (2) estimates from a community multi-scale air quality (CMAQ) modeling system with a high-resolution (1 × 1 km) (CMAQ-NO2). We then compared the effect of AQM-NO2 on AD symptoms with that of CMAQ-NO2. The average distance between the participants’ residences and the nearest AQM station was 2.03 ± 1.06 km, ranging from 0.28 km to 5.73 km. Based on AQM-NO2, the AD symptoms increased by 10.28% (95% confidence interval (CI): 3.24, 17.79) with an increase of 10 ppb of NO2. The effect estimates of CMAQ-NO2 were similar to those of AQM-NO2 when assessed in patients living within 3 km from the nearest AQM station. Even within 1 km, the CI estimate obtained from the CMAQ was much narrower than from AQM (44.18–49.54 vs. 7.02–64.75). However, the association of AQM-NO2 with AD symptoms of patients living beyond 3 km was not positive, whereas that of CMAQ-NO2 maintained positive. In conclusion, exposure to ambient NO2 is significantly associated with aggravation of AD symptoms in young children. In addition, our study suggests that exposure assessment of NO2 using measurement data obtained from monitoring stations far from residential locations can lead to misclassification bias.

Acute Cardiovascular Events Associated With Influenza in Hospitalized Adults : A Cross-sectional Study.

Influenza may contribute to the burden of acute cardiovascular events during annual influenza epidemics. To examine acute cardiovascular events and determine risk factors for acute heart failure (aHF) and acute ischemic heart disease (aIHD) in adults with a hospitalization associated with laboratory-confirmed influenza. Cross-sectional study. U.S. Influenza Hospitalization Surveillance Network during the 2010-to-2011 through 2017-to-2018 influenza seasons. Adults hospitalized with laboratory-confirmed influenza and identified through influenza testing ordered by a practitioner. Acute cardiovascular events were ascertained using discharge codes from the International Classification of Diseases (ICD), Ninth Revision, Clinical Modification, and ICD, 10th Revision. Age, sex, race/ethnicity, tobacco use, chronic conditions, influenza vaccination, influenza antiviral medication, and influenza type or subtype were included as exposures in logistic regression models, and marginal adjusted risk ratios and 95% CIs were estimated to describe factors associated with aHF or aIHD. Among 89999 adults with laboratory-confirmed influenza, 80261 had complete medical record abstractions and available ICD codes (median age, 69 years [interquartile range, 54 to 81 years]) and 11.7% had an acute cardiovascular event. The most common such events (non-mutually exclusive) were aHF (6.2%) and aIHD (5.7%). Older age, tobacco use, underlying cardiovascular disease, diabetes, and renal disease were significantly associated with higher risk for aHF and aIHD in adults hospitalized with laboratory-confirmed influenza. Underdetection of cases was likely because influenza testing was based on practitioner orders. Acute cardiovascular events were identified by ICD discharge codes and may be subject to misclassification bias. In this population-based study of adults hospitalized with influenza, almost 12% of patients had an acute cardiovascular event. Clinicians should ensure high rates of influenza vaccination, especially in those with underlying chronic conditions, to protect against acute cardiovascular events associated with influenza. Centers for Disease Control and Prevention.

Quantifying misclassification and bias errors due to hierarchical sepsis scores in real-time sepsis diagnosis

Severe sepsis and septic shock are common causes of intensive care unit (ICU) admission, length of stay, mortality and cost. Sepsis is very difficult to diagnose in real-time and early treatment can show positive benefits. Current sepsis scores are hierarchically calculated, creating significant case-control and/or misclassification biases in developing real-time diagnostic classifiers based on real-time clinical data. This study examines the impact on kernel density classifier performance of a continuous sepsis score derived from a hierarchical score and based on current well-accepted metrics.Clinical data and model-based insulin sensitivity from 36 patients in the Christchurch Hospital ICU (6071 h) is used to segregate severe sepsis and septic shock from mild sepsis and SIRS using a modified continuous sepsis score using a kernel density estimator (KDE) method. Best case resubstitution, worst case bootstrap, and expected case .632 bootstrap estimates are calculated. Diagnostic quality is assessed using likelihood ratios (LHRs), ROC curves, diagnostic odds-ratios (DOR), sensitivity and specificity. Results were compared to those using the hierarchical sepsis score and the same KDE.The continuous sepsis score yielded smoother, more realistic time-varying trajectories for sepsis evolution. The classifier provided very good results, potentially able to change clinical decisions, but did not reach the best diagnostic accuracy for all performance metrics. Using several metrics avoided assessment bias, and likelihood ratios provided the best overall assessment relative to clinical decision making processes. Finally, quantifying infection presence is a source of case-control and misclassification bias in classifier development, showing clear need for improved sepsis scores.

Non-steroidal Anti-inflammatory Drugs and the Risk of Pneumonia Complications: A Systematic Review.

There have been concerns regarding the safety of nonsteroidal antiinflammatory drugs (NSAIDs) in patients with respiratory infections. However, to date, the quality of the evidence has not been systematically assessed. The purpose of this systematic review was to evaluate the role of NSAIDs on pneumonia complications. OVID MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, Database of Abstracts of Reviews of Effects, and Google Scholar were searched. Studies that examined pneumonia complications in patients who had taken NSAIDs before onset of symptoms were identified. Quality assessment was conducted using the Risk of Bias in Non-randomized Studies - of Interventions (ROBINS-I) assessment tool, which was adapted to include biases that were pertinent to this question. The search strategy identified 1721 potential studies through the 5 primary databases and searching reference lists. Of these, 10 studies met the inclusion criteria, including 5 nested case-control studies, 2 population-based case-control studies, and 3 cohort studies. In total, 59,724 adults were included from 4 of the studies (range = 57-59,250) and 1217 children from 5 studies (range = 148-540). All studies demonstrated a positive association; in adults (odds ratio/risk ratio range = 1.8-8.1) and children (odds ratio/risk ratio range = 1.9-6.8). Studies were limited by moderate or serious risk of confounding bias, exposure misclassification, and protopathic biases and sparse data bias. The results of this review demonstrate that published studies on the effect of NSAIDs use and risk of pneumonia complications are subject to a number of biases. These results should not be extrapolated as evidence of harm for NSAIDs, including ibuprofen, in respiratory ailments but highlight the need for more methodologically robust studies to evaluate this potential relationship.

Epidemiological Studies on Fetal Loss - Better Data and Research Methods are Needed.

In perinatal epidemiology, fetal deaths occur over the first timeline, from conception to birth or fetal death. Majority of other epidemiological research on human diseases focus on the second timeline, from birth to death. The transition from the first to the second timeline is not a fixed duration of time and it depends on when birth occurs. We discussed the complications when switching from the first to the second timeline in epidemiological studies of early life exposures, pregnancy events, and future health outcomes. Population-based studies often lack accurate data on the date of conception for most pregnancies and the complete count of fetal death, leading to chances for selection and misclassification biases. To address these problems, better research data and methodological advancement in study designs and biases evaluations are needed.

Assessing the validity of health administrative data compared to population health survey data for the measurement of low back pain.

Low back pain (LBP) is a high-burden condition that lacks routine surveillance data. Health administrative data may be used for surveillance, but their validity for measuring LBP in the general population has not been established. We aimed to (1) determine the validity of health administrative data to measure LBP compared to self-reported LBP in a population-based sample of Ontario adults; and (2) describe the differences in characteristics of LBP cases based on data sources. Adult respondents (≥18 years) of the Canadian Community Health Survey (CCHS) from 2003 to 2012 were included (N = 150,695). Canadian Community Health Survey data were individually linked to health administrative data, including Ontario Health Insurance Plan and hospitalization data. The reference standard was defined as self-reported back problem diagnosed by a health professional in the CCHS. Measurement of LBP from billing records was defined as ≥1 physician billing or procedural code for LBP during the year preceding CCHS interview date. We measured concurrent validity by comparing prevalence, agreement (kappa), and accuracy (sensitivity, specificity, and positive and negative predictive values [PV]) of administrative data to measure LBP. Prevalence of LBP was higher using self-reported (21.2%) than administrative data (10.2%), and agreement was low (kappa = 0.21). Administrative data had sensitivity 23.9% (95% CI 23.1-24.6), specificity 93.4% (95% CI 93.2-93.7), positive PV 50.4% (95% CI 49.1-51.7), and negative PV 82.0% (95% CI 81.7-82.3). Characteristics of LBP cases based on data sources differed in sex, health/behaviour characteristics, and allied health care utilization. Using health administrative data significantly underestimates the prevalence of LBP. This can lead to misclassification bias that is likely nondifferential in epidemiological studies.

Validating the optimal classification approach using International Classification of Diseases, 10th Revision codes to identify dermatology patients with acne

Validating the optimal classification approach using International Classification of Diseases, 10th Revision codes to identify dermatology patients with acne

External validation demonstrated the Ottawa SAH prediction models can identify pSAH using health administrative data

ObjectivesThe objective of the study is to externally validate three primary subarachnoid hemorrhage (pSAH) identification models. Study Design and SettingWe evaluated three models that identify pSAH using recursive partitioning (A), logistic regression (B), and a prevalence-adjusted logistic regression(C), respectively. Blinded chart review and/or linkage to existing registries determined pSAH status. We included all patients aged ≥18 in four participating center registries or whose discharge abstracts contained ≥1 administrative codes of interest between January 1, 2012 and December 31, 2013. ResultsA total of 3,262 of 193,190 admissions underwent chart review (n = 2,493) or registry linkage (n = 769). A total of 657 had pSAH confirmed (20·1% sample, 0·34% admissions). The sensitivity, specificity, and positive predictive value (PPV) were as follows: i) model A: 98·3% (97·0–99·2), 53·5% (51·5–55·4), and 34·8% (32·6–37·0); ii) model B (score ≥6): 98·0% (96·6–98·9), 47·4% (45·5–49·4), and 32·0% (30·0–34·1); and iii) model C (score ≥2): 95·7% (93·9–97·2), 85·5% (84·0–86·8), and 62·3 (59·3–65·3), respectively. Model C scores of 0, 1, 2, 3, or 4 had probabilities of 0·5% (0·2–1·5), 1·5% (1·0–2·2), 24·8% (21·0–29·0), 90·0% (86·8–92·0), and 97·8% (88·7–99·6), without significant difference between centers (P = 0·86). The PPV of the International Classification of Diseases code (I60) was 63·0% (95% confidence interval: 60·0–66·0). ConclusionsAll three models were highly sensitive for pSAH. Model C could be used to adjust for misclassification bias.

Predictive Value of International Classification of Diseases Codes for Idiopathic Intracranial Hypertension in a University Health System.

Misclassification bias is introduced into medical claims-based research because of reliance on diagnostic coding rather than full medical record review. We sought to characterize this bias for idiopathic intracranial hypertension (IIH) and evaluate strategies to reduce it. A retrospective review of medical records was conducted using a clinical data warehouse containing medical records and administrative data from an academic medical center. Patients with 1 or more instances of International Classification of Diseases (ICD)-9 or -10 codes for IIH (348.2 or G93.2) between 1989 and 2017 and original results of neuroimaging (head CT or MRI), lumbar puncture, and optic nerve examination were included in the study. Diagnosis of IIH was classified as definite, probable, possible, or inaccurate based on review of medical records. The positive predictive value (PPV) for IIH ICD codes was calculated for all subjects, subjects with an IIH code after all testing was completed, subjects with high numbers of IIH ICD codes and codes spanning longer periods, subjects with IIH ICD codes associated with expert encounters (ophthalmology, neurology, or neurosurgery), and subjects with acetazolamide treatment. Of 1,005 patients with ICD codes for IIH, 103 patients had complete testing results and were included in the study. PPV of ICD-9/-10 codes for IIH was 0.63. PPV in restricted samples was 0.82 (code by an ophthalmologist n = 57), 0.70 (acetazolamide treatment n = 87), and 0.72 (code after all testing, n = 78). High numbers of code instances and longer duration between the first and last code instance also increased the PPV. An ICD-9 or -10 code for IIH had a PPV of 63% for probable or definite IIH in patients with necessary diagnostic testing performed at a single institution. Coding accuracy was improved in patients with an IIH ICD code assigned by an ophthalmologist. Use of coding algorithms considering treatment providers, number of codes, and treatment is a potential strategy to reduce misclassification bias in medical claims-based research on IIH. However, these are associated with a reduced sample size.

Definition and diagnosis of cerebral palsy in genetic studies: a systematic review.

To conduct a systematic review of phenotypic definition and case ascertainment in published genetic studies of cerebral palsy (CP) to inform guidelines for the reporting of such studies. Inclusion criteria comprised genetic studies of candidate genes, with CP as the outcome, published between 1990 and 2019 in the PubMed, Embase, and BIOSIS Citation Index databases. Fifty-seven studies met the inclusion criteria. We appraised how CP was defined, the quality of information on case ascertainment, and compliance with international consensus guidelines. Seven studies (12%) were poorly described, 33 studies (58%) gave incomplete information, and 17 studies (30%) were well described. Missing key information precluded determining how many studies complied with the definition by Rosenbaum et al. Only 18 out of 57 studies (32%) were compliant with the Surveillance of Cerebral Palsy in Europe (SCPE) international guidelines on defining CP. Limited compliance with international consensus guidelines on phenotypic definition and mediocre reporting of CP case ascertainment hinders the comparison of results among genetic studies of CP (including meta-analyses), thereby limiting the quality, interpretability, and generalizability of study findings. Compliance with the SCPE guidelines is important for ongoing gene discovery efforts in CP, given the potential for misclassification of unrelated neurological conditions as CP.

A pseudo-likelihood method for estimating misclassification probabilities in competing-risks settings when true-event data are partially observed.

Outcome misclassification occurs frequently in binary-outcome studies and can result in biased estimation of quantities such as the incidence, prevalence, cause-specific hazards, cumulative incidence functions, and so forth. A number of remedies have been proposed to address the potential misclassification of the outcomes in such data. The majority of these remedies lie in the estimation of misclassification probabilities, which are in turn used to adjust analyses for outcome misclassification. A number of authors advocate using a gold-standard procedure on a sample internal to the study to learn about the extent of the misclassification. With this type of internal validation, the problem of quantifying the misclassification also becomes a missing data problem as, by design, the true outcomes are only ascertained on a subset of the entire study sample. Although, the process of estimating misclassification probabilities appears simple conceptually, the estimation methods proposed so far have several methodological and practical shortcomings. Most methods rely on missing outcome data to be missing completely at random (MCAR), a rather stringent assumption which is unlikely to hold in practice. Some of the existing methods also tend to be computationally-intensive. To address these issues, we propose a computationally-efficient, easy-to-implement, pseudo-likelihood estimator of the misclassification probabilities under a missing at random (MAR) assumption, in studies with an available internal-validation sample. We present the estimator through the lens of studies with competing-risks outcomes, though the estimator extends beyond this setting. We describe the consistency and asymptotic distributional properties of the resulting estimator, and derive a closed-form estimator of its variance. The finite-sample performance of this estimator is evaluated via simulations. Using data from a real-world study with competing-risks outcomes, we illustrate how the proposed method can be used to estimate misclassification probabilities. We also show how the estimated misclassification probabilities can be used in an external study to adjust for possible misclassification bias when modeling cumulative incidencefunctions.

Expanded Demographic Norms for Version 3 of the Alzheimer Disease Centers' Neuropsychological Test Battery in the Uniform Data Set.

Norms for the Uniform Data Set Version 3 Neuropsychological Battery are available for cognitively normal individuals based on age, education, and sex; however, these norms do not include race. We provide expanded norms for African Americans and whites. Data from 32 Alzheimer's Disease Centers (ADCs) and ADC affiliated cohorts with global Clinical Dementia Rating Scale (CDR) Dementia Staging Instrument scores of 0 were included. Descriptive statistics for each test were calculated by age, sex, race, and education. Multiple linear regressions were conducted to estimate the effect of each demographic variable; squared semipartial correlation coefficients measured the relative importance of variables. There were 8313 participants (16% African American) with complete demographic information, ranging from 6600 to 7885 depending on the test. Lower scores were found for older and less educated groups, and African Americans versus whites. Education was the strongest predictor for most tests, followed in order by age, race, and sex. Quadratic terms were significant for age and education, indicating some nonlinearity, but did not substantially increase R. Although race-based norms represent incomplete proxies for other sociocultural variables, the appropriate application of these norms is important given the potential to improve diagnostic accuracy and to reduce misclassification bias in cognitive disorders of aging such as Alzheimer disease.

Effect of Smoking on Breast Cancer by Adjusting for Smoking Misclassification Bias and Confounders Using a Probabilistic Bias Analysis Method.

PurposeThe aim of this study was to determine the association between smoking and breast cancer after adjusting for smoking misclassification bias and confounders.MethodsIn this case–control study, 1000 women with breast cancer and 1000 healthy controls were selected. Using a probabilistic bias analysis method, the association between smoking and breast cancer was adjusted for the bias resulting from misclassification of smoking secondary to self-reporting as well as a minimally sufficient adjustment set of confounders derived from a causal directed acyclic graph (cDAG). Population attributable fraction (PAF) for smoking was calculated using Miettinen’s formula.ResultsWhile the odds ratio (OR) from the conventional logistic regression model between smoking and breast cancer was 0.64 (95% CI: 0.36–1.13), the adjusted ORs from the probabilistic bias analysis were in the ranges of 2.63–2.69 and 1.73–2.83 for non-differential and differential misclassification, respectively. PAF ranges obtained were 1.36–1.72% and 0.62–2.01% using the non-differential bias analysis and differential bias analysis, respectively.ConclusionAfter misclassification correction for smoking, the non-significant negative-adjusted association between smoking and breast cancer changed to a significant positive-adjusted association.

Exome-Wide Association Study Reveals Several Susceptibility Genes and Pathways Associated With Acute Coronary Syndromes in Han Chinese.

Genome-wide association studies have identified more than 150 susceptibility loci for coronary artery disease (CAD); however, there is still a large proportion of missing heritability remaining to be investigated. This study sought to identify population-based genetic variation associated with acute coronary syndromes (ACS) in individuals of Chinese Han descent. We proposed a novel strategy integrating a well-developed risk prediction model into control selection in order to lower the potential misclassification bias and increase the statistical power. An exome-wide association analysis was performed for 1,669 ACS patients and 1,935 healthy controls. Promising variants were further replicated using the existing in silico dataset. Additionally, we performed gene- and pathway-based analyses to investigate the aggregate effect of multiple variants within the same genes or pathways. Although none of the association signals were consistent across studies after Bonferroni correction, one promising variant, rs10409124 at STRN4, showed potential impact on ACS in both European and East Asian populations. Gene-based analysis explored four genes (ANXA7, ZNF655, ZNF347, and ZNF750) that showed evidence for association with ACS after multiple test correction, and identification of ZNF655 was successfully replicated by another dataset. Pathway-based analysis revealed that 32 potential pathways might be involved in the pathogenesis of ACS. Our study identified several candidate genes and pathways associated with ACS. Future studies are needed to further validate these findings and explore these genes and pathways as potential therapeutic targets in ACS.

The effect of the look-back period for estimating incidence using administrative data

BackgroundThe look-back period is needed to define baseline population for estimating incidence. However, short look-back period is known to overestimate incidence of diseases misclassifying prevalent cases to incident cases. The purpose of this study is to evaluate the impact of the various length of look-back period on the observed incidences of uterine leiomyoma, endometriosis and adenomyosis, and to estimate true incidences considering the misclassification errors in the longitudinal administrative data in Korea.MethodsA total of 319,608 women between 15 to 54 years of age in 2002 were selected from Korea National Health Insurance Services (KNHIS) cohort database. In order to minimize misclassification bias incurred when applying various length of look-back period, we used 11 years of claim data to estimate the incidence by equally setting the look-back period to 11 years for each year using prediction model. The association between the year of diagnosis and the number of prevalent cases with the misclassification rates by each look-back period was investigated. Based on the findings, prediction models on the proportion of misclassified incident cases were developed using multiple linear regression.ResultsThe proportion of misclassified incident cases of uterine leiomyoma, endometriosis and adenomyosis were 32.8, 10.4 and 13.6% respectively for the one-year look-back period in 2003. These numbers decreased to 6.3% in uterine leiomyoma and − 0.8% in both endometriosis and adenomyosis using all available look-back periods (11 years) in 2013.ConclusionThis study demonstrates approaches for estimating incidences considering the different proportion of misclassified cases for various length of look-back period. Although the prediction model used for estimation showed strong R-squared values, follow-up studies are required for validation of the study results.

Definition of Clinical Outcomes in Pediatric Anesthesia Research: It Is Like the Tower of Babel!

Definition of Clinical Outcomes in Pediatric Anesthesia Research: It Is Like the Tower of Babel!

Identification of Pediatric Autism Spectrum Disorder Cases Using Health Administrative Data.

Administrative data are frequently used to identify Autism Spectrum Disorder (ASD) cases in epidemiological studies. However, validation studies on this mode of case ascertainment have lacked access to high-quality clinical diagnostic data and have not followed published reporting guidelines. We report on the diagnostic accuracy of using readily available health administrative data for pediatric ASD case ascertainment. The validation cohort included almost all the ASD-positive children born in British Columbia, Canada from April 1, 2000 to December 31, 2009 and consisted of 8,670 children in total. 4,079 ASD-positive and 2,787 ASD-negative children were identified using Autism Diagnostic Observation Schedule (ADOS) and Autism Diagnostic Interview-Revised (ADI-R) assessments done through the British Columbia Autism Assessment Network (BCAAN). An additional 1,804 ADOS/ADI-R assessed ASD-positive children were identified using Ministry of Education records. This prospectively collected clinical data (the diagnostic gold standard) was then linked to each child's physician billing and hospital discharge data. The diagnostic accuracy of 11 algorithms that used the administrative data to assign ASD case status was assessed. For all algorithms, high positive predictive values (PPVs) were observed alongside low values for other measures of diagnostic accuracy illustrating that PPVs alone are not an adequate measure of diagnostic accuracy. We show that British Columbia's health administrative data cannot reliably be used to discriminate between children with ASD and children with other developmental disorders. Utilizing these data may result in misclassification bias. Methodologically sound, region-specific validation studies are needed to support the use of administrative data for ASD case ascertainment. Autism Res 2020, 13: 456-463. © 2019 International Society for Autism Research, Wiley Periodicals, Inc. LAY SUMMARY: Health administrative data are frequently used to identify Autism Spectrum Disorder (ASD) cases for research purposes. However, previous validation studies on this sort of case identification have lacked access to high-quality clinical diagnostic data and have not followed published reporting guidelines. We show that British Columbia's health administrative data cannot reliably be used to discriminate between children with ASD and children with other developmental disorders.

Recognizing misclassification bias in research and medical practice.

Recognizing misclassification bias in research and medical practice.

Hispanic and Black Patients with Adult B-Cell Acute Lymphoblastic Leukemia Have a Significantly Worse Survival in the Modern Era - a SEER 2010-2016 Analysis

Prior studies have evaluated incidence and survival trends for B-Acute Lymphoblastic Leukemia (B-ALL) in children and adults. However, there have been no recent studies evaluating the difference in survival between different race and ethnicities in the era of tyrosine kinase inhibitors and novel combination therapy. We wanted to determine 5-year observed survival for adult patients with B-ALL diagnosed in recent years and assess for any difference in survival by race-ethnicity. Methods: We used Surveillance Epidemiology End Results (SEER) 18 registry to identify B-ALL patients using ICD-O-3 codes 9811-9818 and 9836. SEER 18 covers ~28% of US population. The year of diagnosis was limited to 2010-2016 in order to capture a patient population most likely treated with modern therapies. We limited our study to adults aged 20 years (yrs) or more, which were then divided into the following age groups: 20-29, 30-39, 40-49, 50-59, 60-69,70-79,80+ yrs of age. Gender, race-ethnicity, median family income and observed overall survival (OS) were obtained from SEER. For multivariate survival analysis, Cox proportional hazard model was used to adjust for clinically important and other relevant variables (age, gender, race-ethnicity, median income). We included median family income, a county level characteristic in our analysis as a surrogate for access to care. We divided these counties into quintiles based on median family income and included that variable in the multivariate model. We did not adjust for genetic risk or patient insurance status, as the information provided in SEER was inadequate and would likely lead to misclassification bias. SEER Stat 8.3.5 and SAS student edition were used for analysis. Results: 4244 cases of B-ALL were identified with an age adjusted incidence rate of 0.92 per 100,000. B-ALL occurred at all ages, but incidence was higher in young adults (<30 yrs) and adults older than 50 yrs of age, with peak incidence noted in the oldest age group (80+yrs). Of note, our cohort included adults older than or equal to 20 yrs of age, hence did not capture the early peak of childhood B-ALL. Females had a statistically significant lower incidence of B-ALL as compared to males (Incidence rate ratio (IRR) 0.77, p-value<0.05). Table 1 lists counts and IRR by different patient characteristics. Age adjusted incidence was the highest among Hispanics (1.61{1.53-1.71}), followed by Non-Hispanic Whites (NHW)(0.77{0.73-0.8}, Non-Hispanic Asian Pacific Islander (NHAPI)(0.7{0.63-0.78}) and Non-Hispanic Blacks (NHB)(0.54{0.47-0.61}); this difference was statistically significant (p-value <0.05). When we evaluated the age adjusted incidence rate in each age group by different race-ethnicities, Hispanics had a statistically significant higher incidence in each age group except in 80+ yrs age group. About 52% of population died during the study period from any cause. We limited our survival analysis to patients without second malignancy to avoid the confounding effect of another cancer associated mortality. We evaluated OS differences between race-ethnicity in a multivariate model that adjusted for age, sex and income. We found that when compared to NHW, Hispanics (Hazard Ratio (HR) 1.3{1.16-1.46}; p<0.01) and NHB (HR 1.24{1.03-1.5};p 0.02) had worse overall survival. We also showed increased mortality in older than younger adults with B-ALL (Table 2), in line with prior knowledge. There was slightly decreased OS seen in males compared to females, and the difference barely reached statistical significance (HR 0.91{0.83-1.0}; p 0.05). Patients from the poorest counties had worse survival than those in counties with higher median income (Table 2). Conclusion: Our study showed a significant survival disparity in adult B-ALL by race and ethnicity in the modern era. This can partly be attributed to differences in access to care as shown in our study. Interestingly, Hispanic and NHB have a significantly worse overall survival compared to NHW and NHAPI even after accounting for income differences, as a surrogate for access to care. This could be due to other unaccounted measure of health disparity, availability of allogeneic transplantation and/or difference in disease biology. Further studies are needed to evaluate such differences, identify barriers to care in minority populations and characterize potential differences in the genetic make-up of B-ALL in the various ethnic/racial groups. Disclosures Sica: Physician's Education Resources (PER): Honoraria. Verma:Stelexis: Equity Ownership, Honoraria; Acceleron: Honoraria; Celgene: Honoraria; BMS: Research Funding; Janssen: Research Funding.

Estimating the Risk of Myocardial Infarction in Persons with Hemophilia A Using a Machine-Learning Approach with US Claims Data

Introduction Epidemiological studies in the literature provide conflicting conclusions regarding the risk of myocardial infarction (MI) in people diagnosed with congenital hemophilia A (HA). As a result, the question of protection from MI conferred by HA remains debated. Using insurance claims data, we conducted a study to explore this potential relationship. Methods Initially, a traditional pharmaco-epidemiological approach was conducted using the Truven MarketScan Commercial Database and/or Medicare Supplemental Database. A cohort of persons with HA (PwHA) was identified based on the following criteria: having had a confirmed diagnosis of congenital HA between Jan 1, 2000 and Sept 30, 2017; at least three claims for HA within 365 consecutive days; and continuous enrollment with insurance coverage for the 6 months after first diagnosis of HA. In addition, all individuals were required to be aged ≥18 years, male, and with no evidence of a diagnosis of von Willebrand disease (VWD), hemophilia B, acquired HA, or MI prior to their first HA diagnosis. Based on the results of the first analysis, a second, novel approach was undertaken, using a 2-step method incorporating machine learning and drug utilization for cohort identification. For this approach, the inclusion criteria for the study were further refined to include persons with at least one medical or pharmacy claim for factor VIII (FVIII) therapy, activated prothrombin complex concentrate, or activated factor VIIa therapy; or at least one medical or pharmacy claim for FVIII/VWD therapy; or at least one medical or pharmacy claim for desmopressin and at least one medically-attended visit with a diagnosis of HA in the same claim line; or at least one medically-attended visit with a HA diagnosis. The earliest date for fulfilling any of these inclusion criteria was deemed the individual's index date. Participants also had to have 183 days of continuous insurance enrollment prior to study entry in order to participate. Secondly, a HA classification algorithm, first developed and validated by Lyons et al (Value in Health 2018), was adapted and applied to the above refined cohort. A cohort of individuals with no evidence of HA in the study period was then randomly selected from the MarketScan database and frequency matched to the HA cohort by age, sex, insurance type, region, enrollment length, diabetes status, and hypertensive status at a 1:3 ratio, yielding a control cohort of 18,817 individuals. A Poisson regression model was then fitted to estimate the adjusted incidence rate ratio (IRR). The model was adjusted for all baseline covariates as well as HIV and hepatitis C status, with age as a time-varying covariate. Results Based on the chosen criteria, an initial cohort of 2148 PwHA was identified. The crude incidence rate of MI in this cohort was estimated to be 1.75 (95% confidence interval [CI] 1.43-2.11) per 100 person-years. Relative to the matched cohort of individuals with no evidence of HA (N=10,661), this yielded an IRR of 1.68. Such a high relative risk was investigated further by examining concomitant medications. This revealed evidence of misclassification bias with 16% of participants having been prescribed anticoagulants and a low frequency of hemophilia drug utilization. These results prompted the revision of methods underlying cohort identification to the machine learning/drug utilization approach. The revised cohort yielded a 98.5% specificity and 77.8% sensitivity for selection of PwHA, identifying 3154 individuals with a ≥90% probability of being true PwHA. Ten were excluded as they had a previous history of MI, leaving a final cohort of 3144 PwHA. The crude incidence rate of MI in this HA cohort was calculated to be 0.25 (95% CI 0.15-0.34) per 100 person-years and 0.22 (95% CI 0.18-0.27) in the non-HA population, yielding an unadjusted IRR of 1.13. The adjusted IRR was estimated to be 1.31 (95% CI 0.85-2.00, p=0.22), indicating no statistically significant difference in the risk of MI in the HA population versus a matched non-HA control. Conclusions No evidence of a different risk of MI in PwHA relative to non-HA counterparts was observed in this analysis. To our knowledge, this represents the largest real-world data study investigating MI in the HA population. While every effort was taken to mitigate for the effects of confounders and bias, the results should be interpreted in the context of the limitations of a secondary data use study. Disclosures Faghmous: F. Hoffmann-La Roche Ltd: Employment. Sarouei:Genentech, Inc.: Employment. Chang:Genentech, Inc.: Employment; Genentech/Roche: Equity Ownership. Patel:Genentech: Employment; Roche/Genentech: Equity Ownership. Sima:Roche/Genentech: Employment; Roche: Equity Ownership. Kuebler:Genentech, Inc.: Employment, Equity Ownership.