A mega-analysis of genome-wide association studies for major depressive disorder.

Prior genome-wide association studies (GWAS) of major depressive disorder (MDD) have met with limited success. We sought to increase statistical power to detect disease loci by conducting a GWAS mega-analysis for MDD. In the MDD discovery phase, we analyzed more than 1.2 million autosomal and X chromosome single-nucleotide polymorphisms (SNPs) in 18 759 independent and unrelated subjects of recent European ancestry (9240 MDD cases and 9519 controls). In the MDD replication phase, we evaluated 554 SNPs in independent samples (6783 MDD cases and 50 695 controls). We also conducted a cross-disorder meta-analysis using 819 autosomal SNPs with P<0.0001 for either MDD or the Psychiatric GWAS Consortium bipolar disorder (BIP) mega-analysis (9238 MDD cases/8039 controls and 6998 BIP cases/7775 controls). No SNPs achieved genome-wide significance in the MDD discovery phase, the MDD replication phase or in pre-planned secondary analyses (by sex, recurrent MDD, recurrent early-onset MDD, age of onset, pre-pubertal onset MDD or typical-like MDD from a latent class analyses of the MDD criteria). In the MDD-bipolar cross-disorder analysis, 15 SNPs exceeded genome-wide significance (P<5 × 10(-8)), and all were in a 248 kb interval of high LD on 3p21.1 (chr3:52 425 083-53 822 102, minimum P=5.9 × 10(-9) at rs2535629). Although this is the largest genome-wide analysis of MDD yet conducted, its high prevalence means that the sample is still underpowered to detect genetic effects typical for complex traits. Therefore, we were unable to identify robust and replicable findings. We discuss what this means for genetic research for MDD. The 3p21.1 MDD-BIP finding should be interpreted with caution as the most significant SNP did not replicate in MDD samples, and genotyping in independent samples will be needed to resolve its status.

An efficient landmark model for prediction of suicide attempts in multiple clinical settings.

Growing evidence has shown that applying machine learning models to large clinical data sources may exceed clinician performance in suicide risk stratification. However, many existing prediction models either suffer from "temporal bias" (a bias that stems from using case-control sampling) or require training on all available patient visit data. Here, we adopt a "landmark model" framework that aligns with clinical practice for prediction of suicide-related behaviors (SRBs) using a large electronic health record database. Using the landmark approach, we developed models for SRB prediction (regularized Cox regression and random survival forest) that establish a time-point (e.g., clinical visit) from which predictions are made over user-specified prediction windows using historical information up to that point. We applied this approach to cohorts from three clinical settings: general outpatient, psychiatric emergency department, and psychiatric inpatients, for varying prediction windows and lengths of historical data. Models achieved high discriminative performance (area under the Receiver Operating Characteristic curve 0.74-0.93 for the Cox model) across different prediction windows and settings, even with relatively short periods of historical data. In short, we developed accurate, dynamic SRB risk prediction models with the landmark approach that reduce bias and enhance the reliability and portability of suicide risk prediction models.

ATLAS: an automated association test using probabilistically linked health records with application to genetic studies.

OBJECTIVE: Large amounts of health data are becoming available for biomedical research. Synthesizing information across databases may capture more comprehensive pictures of patient health and enable novel research studies. When no gold standard mappings between patient records are available, researchers may probabilistically link records from separate databases and analyze the linked data. However, previous linked data inference methods are constrained to certain linkage settings and exhibit low power. Here, we present ATLAS, an automated, flexible, and robust association testing algorithm for probabilistically linked data. MATERIALS AND METHODS: Missing variables are imputed at various thresholds using a weighted average method that propagates uncertainty from probabilistic linkage. Next, estimated effect sizes are obtained using a generalized linear model. ATLAS then conducts the threshold combination test by optimally combining P values obtained from data imputed at varying thresholds using Fisher's method and perturbation resampling. RESULTS: In simulations, ATLAS controls for type I error and exhibits high power compared to previous methods. In a real-world genetic association study, meta-analysis of ATLAS-enabled analyses on a linked cohort with analyses using an existing cohort yielded additional significant associations between rheumatoid arthritis genetic risk score and laboratory biomarkers. DISCUSSION: Weighted average imputation weathers false matches and increases contribution of true matches to mitigate linkage error-induced bias. The threshold combination test avoids arbitrarily choosing a threshold to rule a match, thus automating linked data-enabled analyses and preserving power. CONCLUSION: ATLAS promises to enable novel and powerful research studies using linked data to capitalize on all available data sources.

Enabling phenotypic big data with PheNorm.

Objective: Electronic health record (EHR)-based phenotyping infers whether a patient has a disease based on the information in his or her EHR. A human-annotated training set with gold-standard disease status labels is usually required to build an algorithm for phenotyping based on a set of predictive features. The time intensiveness of annotation and feature curation severely limits the ability to achieve high-throughput phenotyping. While previous studies have successfully automated feature curation, annotation remains a major bottleneck. In this paper, we present PheNorm, a phenotyping algorithm that does not require expert-labeled samples for training. Methods: The most predictive features, such as the number of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes or mentions of the target phenotype, are normalized to resemble a normal mixture distribution with high area under the receiver operating curve (AUC) for prediction. The transformed features are then denoised and combined into a score for accurate disease classification. Results: We validated the accuracy of PheNorm with 4 phenotypes: coronary artery disease, rheumatoid arthritis, Crohn's disease, and ulcerative colitis. The AUCs of the PheNorm score reached 0.90, 0.94, 0.95, and 0.94 for the 4 phenotypes, respectively, which were comparable to the accuracy of supervised algorithms trained with sample sizes of 100-300, with no statistically significant difference. Conclusion: The accuracy of the PheNorm algorithms is on par with algorithms trained with annotated samples. PheNorm fully automates the generation of accurate phenotyping algorithms and demonstrates the capacity for EHR-driven annotations to scale to the next level - phenotypic big data.

Surrogate-assisted feature extraction for high-throughput phenotyping.

OBJECTIVE: Phenotyping algorithms are capable of accurately identifying patients with specific phenotypes from within electronic medical records systems. However, developing phenotyping algorithms in a scalable way remains a challenge due to the extensive human resources required. This paper introduces a high-throughput unsupervised feature selection method, which improves the robustness and scalability of electronic medical record phenotyping without compromising its accuracy. METHODS: The proposed Surrogate-Assisted Feature Extraction (SAFE) method selects candidate features from a pool of comprehensive medical concepts found in publicly available knowledge sources. The target phenotype's International Classification of Diseases, Ninth Revision and natural language processing counts, acting as noisy surrogates to the gold-standard labels, are used to create silver-standard labels. Candidate features highly predictive of the silver-standard labels are selected as the final features. RESULTS: Algorithms were trained to identify patients with coronary artery disease, rheumatoid arthritis, Crohn's disease, and ulcerative colitis using various numbers of labels to compare the performance of features selected by SAFE, a previously published automated feature extraction for phenotyping procedure, and domain experts. The out-of-sample area under the receiver operating characteristic curve and F -score from SAFE algorithms were remarkably higher than those from the other two, especially at small label sizes. CONCLUSION: SAFE advances high-throughput phenotyping methods by automatically selecting a succinct set of informative features for algorithm training, which in turn reduces overfitting and the needed number of gold-standard labels. SAFE also potentially identifies important features missed by automated feature extraction for phenotyping or experts.

Toward high-throughput phenotyping: unbiased automated feature extraction and selection from knowledge sources.

OBJECTIVE: Analysis of narrative (text) data from electronic health records (EHRs) can improve population-scale phenotyping for clinical and genetic research. Currently, selection of text features for phenotyping algorithms is slow and laborious, requiring extensive and iterative involvement by domain experts. This paper introduces a method to develop phenotyping algorithms in an unbiased manner by automatically extracting and selecting informative features, which can be comparable to expert-curated ones in classification accuracy. MATERIALS AND METHODS: Comprehensive medical concepts were collected from publicly available knowledge sources in an automated, unbiased fashion. Natural language processing (NLP) revealed the occurrence patterns of these concepts in EHR narrative notes, which enabled selection of informative features for phenotype classification. When combined with additional codified features, a penalized logistic regression model was trained to classify the target phenotype. RESULTS: The authors applied our method to develop algorithms to identify patients with rheumatoid arthritis and coronary artery disease cases among those with rheumatoid arthritis from a large multi-institutional EHR. The area under the receiver operating characteristic curves (AUC) for classifying RA and CAD using models trained with automated features were 0.951 and 0.929, respectively, compared to the AUCs of 0.938 and 0.929 by models trained with expert-curated features. DISCUSSION: Models trained with NLP text features selected through an unbiased, automated procedure achieved comparable or slightly higher accuracy than those trained with expert-curated features. The majority of the selected model features were interpretable. CONCLUSION: The proposed automated feature extraction method, generating highly accurate phenotyping algorithms with improved efficiency, is a significant step toward high-throughput phenotyping.

Validation of electronic health record phenotyping of bipolar disorder cases and controls.

OBJECTIVE: The study was designed to validate use of electronic health records (EHRs) for diagnosing bipolar disorder and classifying control subjects. METHOD: EHR data were obtained from a health care system of more than 4.6 million patients spanning more than 20 years. Experienced clinicians reviewed charts to identify text features and coded data consistent or inconsistent with a diagnosis of bipolar disorder. Natural language processing was used to train a diagnostic algorithm with 95% specificity for classifying bipolar disorder. Filtered coded data were used to derive three additional classification rules for case subjects and one for control subjects. The positive predictive value (PPV) of EHR-based bipolar disorder and subphenotype diagnoses was calculated against diagnoses from direct semistructured interviews of 190 patients by trained clinicians blind to EHR diagnosis. RESULTS: The PPV of bipolar disorder defined by natural language processing was 0.85. Coded classification based on strict filtering achieved a value of 0.79, but classifications based on less stringent criteria performed less well. No EHR-classified control subject received a diagnosis of bipolar disorder on the basis of direct interview (PPV=1.0). For most subphenotypes, values exceeded 0.80. The EHR-based classifications were used to accrue 4,500 bipolar disorder cases and 5,000 controls for genetic analyses. CONCLUSIONS: Semiautomated mining of EHRs can be used to ascertain bipolar disorder patients and control subjects with high specificity and predictive value compared with diagnostic interviews. EHRs provide a powerful resource for high-throughput phenotyping for genetic and clinical research.

Stratification of risk for hospital admissions for injury related to fall: cohort study.

OBJECTIVE: To determine whether the ability to stratify an individual patient's hazard for falling could facilitate development of focused interventions aimed at reducing these adverse outcomes. DESIGN: Clinical and sociodemographic data from electronic health records were utilized to derive multiple logistic regression models of hospital readmissions for injuries related to falls. Drugs used at admission were summarized based on reported adverse effect frequencies in published drug labeling. SETTING: Two large academic medical centers in New England, United States. PARTICIPANTS: The model was developed with 25,924 individuals age ≥ 40 with an initial hospital discharge. The resulting model was then tested in an independent set of 13,032 inpatients drawn from the same hospital and 36,588 individuals discharged from a second large hospital during the same period. MAIN OUTCOME MEASURE: Hospital readmissions for injury related to falls. RESULTS: Among 25,924 discharged individuals, 680 (2.6%) were evaluated in the emergency department or admitted to hospital for a fall within 30 days of discharge, 1635 (6.3%) within 180 days of discharge, 2360 (9.1%) within one year, and 3465 (13.4%) within two years. Older age, female sex, white or African-American race, public insurance, greater number of drugs taken on discharge, and score for burden of adverse effects were each independently associated with hazard for fall. For drug burden, presence of a drug with a frequency of adverse effects related to fall of 10% was associated with 3.5% increase in odds of falling over the next two years (odds ratio 1.04, 95% confidence interval 1.02 to 1.05). In an independent testing set, the area under the receiver operating characteristics curve was 0.65 for a fall within two years based on cross sectional data and 0.72 with the addition of prior utilization data including age adjusted Charlson comorbidity index. Portability was promising, with area under the curve of 0.71 for the longitudinal model in a second hospital system. CONCLUSIONS: It is potentially useful to stratify risk of falls based on clinical features available as artifacts of routine clinical care. A web based tool can be used to calculate and visualize risk associated with drug treatment to facilitate further investigation and application.

Limbic system white matter microstructure and long-term treatment outcome in major depressive disorder: a diffusion tensor imaging study using legacy data.

OBJECTIVES: Treatment-resistant depression is a common clinical occurrence among patients with major depressive disorder (MDD), but its neurobiology is poorly understood. We used data collected as part of routine clinical care to study white matter integrity of the brain's limbic system and its association to treatment response. METHODS: Electronic medical records of multiple large New England hospitals were screened for patients with an MDD billing diagnosis, and natural language processing was subsequently applied to find those with concurrent diffusion-weighted images, but without any diagnosed brain pathology. Treatment outcome was determined by review of clinical charts. MDD patients (n = 29 non-remitters, n = 26 partial-remitters, and n = 37 full-remitters), and healthy control subjects (n = 58) were analyzed for fractional anisotropy (FA) of the fornix and cingulum bundle. RESULTS: Failure to achieve remission was associated with lower FA among MDD patients, statistically significant for the medial body of the fornix. Moreover, global and regional-selective age-related FA decline was most pronounced in patients with treatment-refractory, non-remitted depression. CONCLUSIONS: These findings suggest that specific brain microstructural white matter abnormalities underlie persistent, treatment-resistant depression. They also demonstrate the feasibility of investigating white matter integrity in psychiatric populations using legacy data.

An electronic health records study of long-term weight gain following antidepressant use.

IMPORTANCE: Short-term studies suggest antidepressants are associated with modest weight gain but little is known about longer-term effects and differences between individual medications in general clinical populations. OBJECTIVE: To estimate weight gain associated with specific antidepressants over the 12 months following initial prescription in a large and diverse clinical population. DESIGN, SETTING, AND PARTICIPANTS: We identified 22,610 adult patients who began receiving a medication of interest with available weight data in a large New England health care system, including 2 academic medical centers and affiliated outpatient primary and specialty care clinics. We used electronic health records to extract prescribing data and recorded weights for any patient with an index antidepressant prescription including amitriptyline hydrochloride, bupropion hydrochloride, citalopram hydrobromide, duloxetine hydrochloride, escitalopram oxalate, fluoxetine hydrochloride, mirtazapine, nortriptyline hydrochloride, paroxetine hydrochloride, venlafaxine hydrochloride, and sertraline hydrochloride. As measures of assay sensitivity, additional index prescriptions examined included the antiasthma medication albuterol sulfate and the antiobesity medications orlistat, phentermine hydrochloride, and sibutramine hydrochloride. Mixed-effects models were used to estimate rate of weight change over 12 months in comparison with the reference antidepressant, citalopram. MAIN OUTCOME AND MEASURE: Clinician-recorded weight at 3-month intervals up to 12 months. RESULTS: Compared with citalopram, in models adjusted for sociodemographic and clinical features, significantly decreased rate of weight gain was observed among individuals treated with bupropion (ß [SE]: -0.063 [0.027]; P = .02), amitriptyline (ß [SE]: -0.081 [0.025]; P = .001), and nortriptyline (ß [SE]: -0.147 [0.034]; P < .001). As anticipated, differences were less pronounced among individuals discontinuing treatment prior to 12 months. CONCLUSIONS AND RELEVANCE: Antidepressants differ modestly in their propensity to contribute to weight gain. Short-term investigations may be insufficient to characterize and differentiate this risk.

Rare copy number variation in treatment-resistant major depressive disorder.

BACKGROUND: While antidepressant treatment response appears to be partially heritable, no consistent genetic associations have been identified. Large, rare copy number variants (CNVs) play a role in other neuropsychiatric diseases, so we assessed their association with treatment-resistant depression (TRD). METHODS: We analyzed data from two genome-wide association studies comprising 1263 Caucasian patients with major depressive disorder. One was drawn from a large health system by applying natural language processing to electronic health records (i2b2 cohort). The second consisted of a multicenter study of sequential antidepressant treatments, Sequenced Treatment Alternatives to Relieve Depression. The Birdsuite package was used to identify rare deletions and duplications. Individuals without symptomatic remission, despite two antidepressant treatment trials, were contrasted with those who remitted with a first treatment trial. RESULTS: CNV data were derived for 778 subjects in the i2b2 cohort, including 300 subjects (37%) with TRD, and 485 subjects in Sequenced Treatment Alternatives to Relieve Depression cohort, including 152 (31%) with TRD. CNV burden analyses identified modest enrichment of duplications in cases (empirical p = .04 for duplications of 100-200 kilobase) and a particular deletion region spanning gene PABPC4L (empirical p = .02, 6 cases: 0 controls). Pathway analysis suggested enrichment of CNVs intersecting genes regulating actin cytoskeleton. However, none of these associations survived genome-wide correction. CONCLUSIONS: Contribution of rare CNVs to TRD appears to be modest, individually or in aggregate. The electronic health record-based methodology demonstrated here should facilitate collection of larger TRD cohorts necessary to further characterize these effects.

Feasibility of studying brain morphology in major depressive disorder with structural magnetic resonance imaging and clinical data from the electronic medical record: a pilot study.

For certain research questions related to long-term outcomes or to rare disorders, designing prospective studies is impractical or prohibitively expensive. Such studies could instead utilize clinical and magnetic resonance imaging data (MRI) collected as part of routine clinical care, stored in the electronic medical record (EMR). Using major depressive disorder (MDD) as a disease model, we examined the feasibility of studying brain morphology and associations with remission using clinical and MRI data exclusively drawn from the EMR. Advanced automated tools were used to select MDD patients and controls from the EMR who had brain MRI data, but no diagnosed brain pathology. MDD patients were further assessed for remission status by review of clinical charts. Twenty MDD patients (eight full-remitters, six partial-remitters, and six non-remitters), and 15 healthy control subjects met all study criteria for advanced morphometric analyses. Compared to controls, MDD patients had significantly smaller right rostral-anterior cingulate volume, and level of non-remission was associated with smaller left hippocampus and left rostral-middle frontal gyrus volume. The use of EMR data for psychiatric research may provide a timely and cost-effective approach with the potential to generate large study samples reflective of the real population with the illness studied.

QT interval and antidepressant use: a cross sectional study of electronic health records.

OBJECTIVE: To quantify the impact of citalopram and other selective serotonin reuptake inhibitors on corrected QT interval (QTc), a marker of risk for ventricular arrhythmia, in a large and diverse clinical population. DESIGN: A cross sectional study using electrocardiographic, prescribing, and clinical data from electronic health records to explore the relation between antidepressant dose and QTc. Methadone, an opioid known to prolong QT, was included to demonstrate assay sensitivity. SETTING: A large New England healthcare system comprising two academic medical centres and outpatient clinics. PARTICIPANTS: 38,397 adult patients with an electrocardiogram recorded after prescription of antidepressant or methadone between February 1990 and August 2011. MAIN OUTCOME MEASURES: Relation between antidepressant dose and QTc interval in linear regression, adjusting for potential clinical and demographic confounding variables. For a subset of patients, change in QTc after drug dose was also examined. RESULTS: Dose-response association with QTc prolongation was identified for citalopram (adjusted beta 0.10 (SE 0.04), P<0.01), escitalopram (adjusted beta 0.58 (0.15), P<0.001), and amitriptyline (adjusted beta 0.11 (0.03), P<0.001), but not for other antidepressants examined. An association with QTc shortening was identified for bupropion (adjusted beta 0.02 (0.01) P<0.05). Within-subject paired observations supported the QTc prolonging effect of citalopram (10 mg to 20 mg, mean QTc increase 7.8 (SE 3.6) ms, adjusted P<0.05; and 20 mg to 40 mg, mean QTc increase 10.3 (4.0) ms, adjusted P<0.01). CONCLUSIONS: This study confirmed a modest prolongation of QT interval with citalopram, and identified additional antidepressants with similar observed risk. Pharmacovigilance studies using electronic health record data may be a useful method of identifying potential risk associated with treatments.

Modeling disease severity in multiple sclerosis using electronic health records.

OBJECTIVE: To optimally leverage the scalability and unique features of the electronic health records (EHR) for research that would ultimately improve patient care, we need to accurately identify patients and extract clinically meaningful measures. Using multiple sclerosis (MS) as a proof of principle, we showcased how to leverage routinely collected EHR data to identify patients with a complex neurological disorder and derive an important surrogate measure of disease severity heretofore only available in research settings. METHODS: In a cross-sectional observational study, 5,495 MS patients were identified from the EHR systems of two major referral hospitals using an algorithm that includes codified and narrative information extracted using natural language processing. In the subset of patients who receive neurological care at a MS Center where disease measures have been collected, we used routinely collected EHR data to extract two aggregate indicators of MS severity of clinical relevance multiple sclerosis severity score (MSSS) and brain parenchymal fraction (BPF, a measure of whole brain volume). RESULTS: The EHR algorithm that identifies MS patients has an area under the curve of 0.958, 83% sensitivity, 92% positive predictive value, and 89% negative predictive value when a 95% specificity threshold is used. The correlation between EHR-derived and true MSSS has a mean R(2) = 0.38±0.05, and that between EHR-derived and true BPF has a mean R(2) = 0.22±0.08. To illustrate its clinical relevance, derived MSSS captures the expected difference in disease severity between relapsing-remitting and progressive MS patients after adjusting for sex, age of symptom onset and disease duration (p = 1.56×10(-12)). CONCLUSION: Incorporation of sophisticated codified and narrative EHR data accurately identifies MS patients and provides estimation of a well-accepted indicator of MS severity that is widely used in research settings but not part of the routine medical records. Similar approaches could be applied to other complex neurological disorders.

A translational engine at the national scale: informatics for integrating biology and the bedside.

Informatics for integrating biology and the bedside (i2b2) seeks to provide the instrumentation for using the informational by-products of health care and the biological materials accumulated through the delivery of health care to conduct discovery research and to study the healthcare system in vivo. This complements existing efforts such as prospective cohort studies or trials outside the delivery of routine health care. i2b2 has been used to generate genome-wide studies at less than one tenth the cost and one tenth the time of conventionally performed studies as well as to identify important risk from commonly used medications. i2b2 has been adopted by over 60 academic health centers internationally.

Antidepressant response in patients with major depression exposed to NSAIDs: a pharmacovigilance study.

OBJECTIVE It has been suggested that there is a mechanism by which nonsteroidal anti-inflammatory drugs (NSAIDs) may interfere with antidepressant response, and poorer outcomes among NSAID-treated patients were reported in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study. To attempt to confirm this association in an independent population-based treatment cohort and explore potential confounding variables, the authors examined use of NSAIDs and related medications among 1,528 outpatients in a New England health care system. METHOD Treatment outcomes were classified using a validated machine learning tool applied to electronic medical records. Logistic regression was used to examine the association between medication exposure and treatment outcomes, adjusted for potential confounding variables. To further elucidate confounding and treatment specificity of the observed effects, data from the STAR*D study were reanalyzed. RESULTS NSAID exposure was associated with a greater likelihood of depression classified as treatment resistant compared with depression classified as responsive to selective serotonin reuptake inhibitors (odds ratio=1.55, 95% CI=1.21-2.00). This association was apparent in the NSAIDs-only group but not in those using other agents with NSAID-like mechanisms (cyclooxygenase-2 inhibitors and salicylates). Inclusion of age, sex, ethnicity, and measures of comorbidity and health care utilization in regression models indicated confounding; association with outcome was no longer significant in fully adjusted models. Reanalysis of STAR*D results likewise identified an association in NSAIDs but not NSAID-like drugs, with more modest effects persisting after adjustment for potential confounding variables. CONCLUSIONS These results support an association between NSAID use and poorer antidepressant outcomes in major depressive disorder but indicate that some of the observed effect may be a result of confounding.

Incident user cohort study of risk for gastrointestinal bleed and stroke in individuals with major depressive disorder treated with antidepressants.

OBJECTIVE: To examine the association between exposure to newer antidepressants and risk of gastrointestinal (GI) and other bleeding complications among individuals with major depressive disorder (MDD). DESIGN: This study uses an incident user cohort design to compare associations between incidence of vascular/bleeding events and the relative affinity (low, moderate or high) of the antidepressant for the serotonin transporter during an exposure risk period for each patient. SETTING: New England healthcare system electronic medical record database. PARTICIPANTS: 36 389 individuals with a diagnosis of MDD and monotherapy with a selective serotonin reuptake inhibitor, serotonin-norepinephrine reuptake inhibitor or other new-generation antidepressant were identified from among 3.1 million patients in a New England healthcare system. PRIMARY AND SECONDARY OUTCOME MEASURES: Rates of bleeding or other vascular complications, including acute liver failure, acute renal failure, asthma, breast cancer and hip fractures. RESULTS: 601 GI bleeds were observed in the 21 462 subjects in the high-affinity group versus 333 among the 14 927 subjects in the lower affinity group (adjusted RR: 1.17, 95% CI 1.02 to 1.34). Similarly, 776 strokes were observed in the high-affinity group versus 434 in the lower affinity treatment group (adjusted RR: 1.18, 95% CI 1.06 to 1.32). No significant association with risk for a priori negative control outcomes, including acute liver failure, acute renal failure, asthma, breast cancer and hip fractures, was identified. CONCLUSIONS: Use of antidepressants with high affinity for the serotonin transporter may confer modestly elevated risk for GI and other bleeding complications. While multiple methodologic limitations must be considered, these results suggest that antidepressants with lower serotonin receptor affinity may be preferred in patients at greater risk for such complications.