Assessment of the coding accuracy of warfarin-related bleeding events

Predictive modeling of medication adherence in post myocardial infarction patients: a bayesian approach using beta-regression

AIMS

Predicting medication adherence in post myocardial infarction (MI) patients has the potential to improve patient outcomes. Most adherence prediction models dichotomize adherence metrics and status. This study aims to develop medication adherence prediction models that avoid dichotomizing adherence metrics and to test whether a simplified model including only 90-days adherence data would perform similarly to a full multivariable model.

METHODS

Post MI adult patients were followed for 1-year post the event. Data from pharmacy records were used to calculate proportion of days covered (PDC). We used Bayesian beta-regression to model PDC as a proportion, avoiding dichotomization. For each medication group, statins, P2Y12 inhibitors and aspirin, two prediction models were developed, a full and a simplified model.

RESULTS

3692 patients were included for model development. The median (Inter quartile range) PDC at 1-year for statins, P2Y12 inhibitors and aspirin was 0.8 (0.33, 1.00), 0.79 (0.23, 0.99) and 0.79 (0.23, 0.99), respectively. All models showed good fit to the data by visual predictive checks. Bayesian R2 for statins, P2Y12 inhibitors and aspirin models were 61.4%,71.2% and 55.2%, respectively. The simplified models showed similar performance compared with full complex models as evaluated by cross validation.

CONCLUSIONS

We developed Bayesian multilevel models for statins, P2Y12 inhibitors and aspirin in post MI patients that handled 1-year PDC as a proportion using the beta-distribution. In addition, simplified models, with 90-days adherence as single predictor, had similar performance compared with full complex models.

Predictors of Non-Variceal Hemorrhage in a National Cohort of Patients With Chronic Liver Disease

Background

Non-variceal hemorrhage in patients with chronic liver disease (CLD) increases morbidity, mortality, and healthcare costs. There are limited data on risk factors for non-variceal hemorrhage in the CLD population. The aim of this study was to assess the predictive value of various clinical and laboratory parameters for non-variceal hemorrhage in CLD patients.

Methods

We conducted a retrospective cohort study of US veterans diagnosed with CLD between 2002 and 2018 within the Veterans Health Administration database. We derived candidate variables from existing risk prediction models for hemorrhage, risk calculators for severity of liver disease, Charlson index of prognostic comorbidities, and prior literature. We used a competing risk analysis to study the relationship between putative risk factors and incidence of non-variceal hemorrhage in patients with CLD.

Results

Of 15,183 CLD patients with no history of cancer or anticoagulation use, 674 experienced non-variceal hemorrhage within 1 year of CLD diagnosis. In multivariable analysis, 11 of the 26 candidate variables independently predicted non-variceal hemorrhage: race, international normalized ratio (INR) > 1.5, bilirubin ≥ 2 mg/dL, albumin ≤ 3.5 g/dL, anemia, alcohol abuse, antiplatelet therapy, chronic kidney disease, dementia, proton pump inhibitor prescription, and recent infection.

Conclusions

In this study of almost 15,000 veterans, risk factors for non-variceal bleeding within the first year after diagnosis of CLD included non-Caucasian race, laboratory parameters indicating severe liver disease and recent infection in addition to the risk factors for bleeding observed in a general non-CLD population.

Clinical Outcomes of Direct Oral Anticoagulants vs Warfarin for Extended Treatment of Venous Thromboembolism

Importance

Extending the duration of oral anticoagulation for venous thromboembolism (VTE) beyond the initial 3 to 6 months of treatment is often recommended, but it is not clear whether clinical outcomes differ when using direct oral anticoagulants (DOACs) or warfarin.

Objective

To compare rates of recurrent VTE, hospitalizations for hemorrhage, and all-cause death among adults prescribed DOACs or warfarin whose anticoagulant treatment was extended beyond 6 months after acute VTE.

Design, Setting, and Participants

This cohort study was conducted in 2 integrated health care delivery systems in California with adults aged 18 years or older who received a diagnosis of incident VTE between 2010 and 2018 and completed at least 6 months of oral anticoagulant treatment with DOACs or warfarin. Patients were followed from the end of the initial 6-month treatment period until discontinuation of anticoagulation, occurrence of an outcome event, health plan disenrollment, or end of the study follow-up period (December 31, 2019). Data were obtained from the Kaiser Permanente Virtual Data Warehouse and electronic health records. Data analysis was conducted from March 2022 to January 2023.

Exposure

Dispensed prescriptions of DOACs or warfarin after a 6-month initial treatment for VTE.

Main Outcomes and Measures

The primary outcomes were rates per 100 person-years of recurrent VTE, hospitalizations for hemorrhage, and all-cause death. Comparison of DOAC and warfarin outcomes were performed using multivariable Cox proportional hazards regression.

Results

A total of 18 495 patients (5477 [29.6%] aged ≥75 years; 8973 women [48.5%]) with VTE who were treated with at least 6 months of anticoagulation were identified, of whom 2134 (11.5%) were receiving DOAC therapy and 16 361 (88.5%) were receiving warfarin therapy. Unadjusted event rates were lower for patients receiving DOAC therapy than warfarin therapy for recurrent VTE (event rate per 100 person-years, 2.92 [95% CI, 2.29-3.54] vs 4.14 [95% CI, 3.90-4.38]), hospitalizations for hemorrhage (event rate per 100 person-years, 1.02 [95% CI, 0.66-1.39] vs 1.81 [95% CI, 1.66-1.97]), and all-cause death (event rate per 100 person-years, 3.79 [95% CI, 3.09-4.49] vs 5.40 [95% CI, 5.13-5.66]). After multivariable adjustment, DOAC treatment was associated with a lower risk of recurrent VTE (adjusted hazard ratio [aHR], 0.66; 95% CI, 0.52-0.82). For patients prescribed DOAC treatment, the risks of hospitalization for hemorrhage (aHR, 0.79; 95% CI, 0.54-1.17) and all-cause death (aHR, 0.96; 95% CI, 0.78-1.19) were not significantly different than those for patients prescribed warfarin treatment.

Conclusions and Relevance

In this cohort study of patients with VTE who continued warfarin or DOAC anticoagulation beyond 6 months, DOAC treatment was associated with a lower risk of recurrent VTE, supporting the use of DOACs for the extended treatment of VTE in terms of clinical outcomes.

Arterial Thromboembolism in Patients With Atrial Fibrillation and CHA2DS2-VASc Score 0 to 2 With and Without Cancer

Background

It is unclear whether newly diagnosed cancer adds to the risk of arterial thromboembolism (ATE) in patients with atrial fibrillation/flutter (AF). This is especially relevant for AF patients with low to intermediate CHA₂DS₂-VASc scores in whom the risk-benefit ratios between ATE and bleeding are delicately balanced.

Objectives

The objectives were to evaluate the ATE risk in AF patients with a CHA₂DS₂-VASc score of 0 to 2 with and without cancer.

Methods

A population-based retrospective cohort study was performed. Patients with a CHA₂DS₂-VASc score of 0 to 2 not receiving anticoagulation at cancer diagnosis (or the matched index date) were included. Patients with embolic ATE or cancer before study index were excluded. AF patients were categorized into AF and cancer and AF and no cancer cohorts. Cohorts were matched for multinomial distribution of age, sex, index year, AF duration, CHA₂DS₂-VASc score, and low/high/undefined ATE risk cancer. Patients were followed from study index until the primary outcome or death. The primary outcome was acute ATE (ischemic stroke, transient ischemic attack, or systemic ATE) at 12 months using International Classification of Diseases-Ninth Revision codes from hospitalization. The Fine-Gray competing risk model was used to estimate the HR for ATE with death as a competing risk.

Results

The 12-month cumulative incidence of ATE was 2.13% (95% CI: 1.47-2.99) in 1,411 AF patients with cancer and 0.8% (95% CI: 0.56-1.10) in 4,233 AF patients without cancer (HR: 2.70; 95% CI: 1.65-4.41). The risk was highest in men with CHA₂DS₂-VASc = 1 and women with CHA₂DS₂-VASc = 2 (HR: 6.07; 95% CI: 2.45-15.01).

Conclusions

In AF patients with CHA₂DS₂-VASc scores of 0 to 2, newly diagnosed cancer is associated with an increased incidence of stroke, transient ischemic attack, or systemic ATE compared with matched controls without cancer.

COVID-RDNet: A novel coronavirus pneumonia classification model using the mixed dataset by CT and X-rays images

Corona virus disease 2019 (COVID-19) testing relies on traditional screening methods, which require a lot of manpower and material resources. Recently, to effectively reduce the damage caused by radiation and enhance effectiveness, deep learning of classifying COVID-19 negative and positive using the mixed dataset by CT and X-rays images have achieved remarkable research results. However, the details presented on CT and X-ray images have pathological diversity and similarity features, thus increasing the difficulty for physicians to judge specific cases. On this basis, this paper proposes a novel coronavirus pneumonia classification model using the mixed dataset by CT and X-rays images. To solve the problem of feature similarity between lung diseases and COVID-19, the extracted features are enhanced by an adaptive region enhancement algorithm. Besides, the depth network based on the residual blocks and the dense blocks is trained and tested. On the one hand, the residual blocks effectively improve the accuracy of the model and the non-linear COVID-19 features are obtained by cross-layer link. On the other hand, the dense blocks effectively improve the robustness of the model by connecting local and abstract information. On mixed X-ray and CT datasets, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), area under curve (AUC), and accuracy can all reach 0.99. On the basis of respecting patient privacy and ethics, the proposed algorithm using the mixed dataset from real cases can effectively assist doctors in performing the accurate COVID-19 negative and positive classification to determine the infection status of patients.

Venous thromboembolism prophylaxis with low molecular weight heparin versus unfractionated heparin for patients undergoing operative treatment of closed femoral shaft fractures

BACKGROUND

The objective of this study was to compare inpatient mortality rates for patients with operatively treated closed femoral shaft fractures (AO/OTA 32 A-C) who received venous thromboembolism (VTE) prophylaxis with either low molecular weight heparin (LMWH) or unfractionated heparin.

METHODS

This was a retrospective cohort study of a national database of patients presenting to Level I through IV trauma centers in the United States. All patients ≥18 years of age who sustained an operatively treated closed femoral shaft fracture were included. The primary outcome of inpatient mortality was compared between two groups: those who received LMWH or unfractionated heparin for VTE prophylaxis. Secondary outcomes were complications including VTE and bleeding events. Groups were compared using a multivariate regression model.

RESULTS

There were 2058 patients included in the study. Patients who received VTE prophylaxis with LMWH had lower odds of inpatient mortality compared to patients who received VTE prophylaxis with unfractionated heparin (OR 0.19; 95% CI 0.05 to 0.68, p = 0.011).

CONCLUSIONS

VTE prophylaxis with LMWH is associated with lower inpatient mortality compared to VTE prophylaxis with unfractionated heparin for patients undergoing operative treatment of closed femoral shaft fractures. To our knowledge this is the first study to report these associations for a specific subset of orthopedic trauma patients.

Different risks of hemorrhage in patients with elevated international normalized ratio from chronic liver disease versus warfarin therapy, a population-based retrospective cohort study

BACKGROUND

Patients with chronic liver disease (CLD) often present with an elevated international normalized ratio (INR). Although elevated INR reflects a higher risk of hemorrhage among warfarin users, its clinical significance in CLD patients is less clear.

OBJECTIVES

We used Veterans Health Administration data to quantify the association between INR and (non-variceal) hemorrhage in patients with CLD compared to warfarin users.

METHODS

We performed a multivariate competing risk analysis to study the association between INR and hemorrhage in the two cohorts. We used an interaction term between INR and cohort (CLD/warfarin users) to test if INR had different effects on hemorrhage in the two cohorts.

RESULTS

Data from 80 134 patients (14, 412 with CLD and 65, 722 taking warfarin) were analyzed. The effect of INR on the risk of hemorrhage differed between CLD patients and warfarin users (interaction P < .001). As INR increased above 1.5, the adjusted hazard ratio (aHR) for hemorrhage in CLD patients increased to 2.25 but remained fairly constant with further elevation of INR values. In contrast, the risk of hemorrhage in patients taking warfarin remained low with INR in the subtherapeutic (INR <2.0) and therapeutic ranges (INR 2.0-3.0), and increased exponentially with INR in the supratherapeutic range (aHR 1.64 with INR >3.0-3.5, and 4.70 with INR >3.5).

CONCLUSIONS

The relationship between INR and risk of hemorrhage in CLD patients is different from that in warfarin users. Caution should be exercised extrapolating data from warfarin users to make clinical decisions in CLD patients.

Comparison of the Efficacy and Safety of Direct Oral Anticoagulants and Vitamin K Antagonists in Patients with Atrial Fibrillation and Concomitant Liver Cirrhosis: A Systematic Review and Meta-Analysis

BACKGROUND

Patients with atrial fibrillation (AF) have a higher risk of developing thromboembolic events. Current guidelines recommend the use of oral anticoagulants for stroke prevention in these patients. Several clinical trials demonstrated that direct oral anticoagulants (DOACs) have similar efficacy and are safer alternatives to traditional oral anticoagulants. However, patients with concomitant liver cirrhosis were excluded from these trials.

OBJECTIVE

We aimed to systematically identify and review published clinical studies on the use of DOACs in patients with AF and liver cirrhosis and assess the efficacy and safety of DOACs in these patients.

METHODS

A systematic review of clinical trials and retrospective studies was conducted by searching the PubMed, Cochrane Library, Embase, SCOPUS, and Web of Science databases up to September 2020.

RESULTS

Three retrospective studies were included, involving 4011 patients with AF and liver cirrhosis. The use of DOACs was associated with a significant reduction in ischemic stroke (hazard ratio [HR] 0.62; 95% confidence interval [CI] 0.42-0.90; p = 0.01), major bleeding events (HR 0.64; 95% CI 0.57-0.72; p < 0.001), and intracranial hemorrhage (HR 0.49; 95% CI 0.40-0.59; p < 0.001).

CONCLUSIONS

Compared with warfarin in patients with AF and liver cirrhosis, DOACs appear to be associated with improved efficacy and safety outcomes. Randomized controlled trials are warranted to confirm these findings.

Risk of major bleeding associated with concomitant use of anticancer drugs and direct oral anticoagulant in patients with cancer and atrial fibrillation

This study evaluated the risk of major bleeding associated with concomitant use of direct oral anticoagulant (DOAC) and anticancer drugs (ACDs), which share metabolic pathways, in patients with atrial fibrillation (AF) and cancer. We performed a retrospective cohort study using Taiwan's National Health Insurance database and included patients with AF and cancer who received DOAC prescriptions from 1 to 2012 to 31 December 2017. The incidence of major bleeding in person-quarters with concomitant use of DOAC and any of 15 ACDs with inhibitory or competitive effects of CYP3A4 or P-gp activity (docetaxel, vinorelbine, methotrexate, irinotecan, etoposide, doxorubicin, cyclophosphamide, imatinib, nilotinib, abiraterone, bicalutamide, tamoxifen, anastrozole, cyclosporine, tacrolimus) was compared with that in person-quarters with DOAC alone. Adjusted incidence-rate differences between DOAC use with and without concurrent ACDs were estimated using Poisson regression models weighted by the inverse probability of treatment. In 13,158 patients with AF and cancer (76.9 ± 8.9 years; male 60%), 1545 major bleeding events occurred during 90,540 DOAC-exposed person-quarters. Concurrent use of DOAC and any of 15 ACDs occurred in only 18% of patients. Compared with use of DOAC alone, concomitant use of DOAC and these ACDs was not associated with an increased risk of major bleeding. Co-medication with DOAC and ACDs with inhibitory or competitive effects on CYP3A4 or P-gp activity was not associated with a higher risk of major bleeding than DOAC alone. Our findings may provide clinicians with confidence regarding the safety of concurrent use of DOAC and ACDs in patients with AF and cancer.

Novel bleeding prediction model in atrial fibrillation patients on new oral anticoagulants

OBJECTIVE

Clinical models such as the HAS-BLED (standing for Hypertension, Abnormal liver/renal function, Stroke history, Bleeding history or predisposition, Labile INR, Elderly, Drug/alcohol usage) were developed to predict risk of major bleeding on vitamin K antagonists/antiplatelet therapy. We aimed to develop a model that will improve the ability to predict major bleeding events in patients with non-valvular atrial fibrillation (AF) treated with new oral anticoagulants (NOACs).

METHODS

Clalit Health Services is the largest of four integrated healthcare organisations in Israel, which insures 4.7 million patients (53% of the population). We identified in Clalit Health Services all patients with AF, new users of an NOAC (2013-2017), and followed them until first occurrence of a major bleeding event, death, switch to another oral anticoagulant, 30 days after discontinuation of NOAC or end of follow-up (31 December 2019). Importance of the candidate model variables was estimated by inclusion frequencies across forward selection algorithm applied to 50 bootstrap samples. Then, backward selection algorithm using the modified Bayesian Information Criterion for competing risks was applied to select predictors for the final model.

RESULTS

47 623 patients with AF prescribed NOAC were studied. 28 055 patients with AF, initiators of apixaban (mean age 78.7, SD 9.0), were included in the first phase and had 662 major bleeding events. Nine variables were selected for inclusion in a final points-based risk-scoring system: male sex, anaemia, thrombocytopaenia (<99×10³/µL), concurrent antiplatelet therapy, hypertension, prior major bleeding, risk factors for a fall, low cholesterol level and low estimated glomerular filtration rate, with apparent area-under-curve (AUC) of 0.6546. Applicability of the model was then shown for 14 118 and 5450 patients with AF, initiators of dabigatran and rivaroxaban, where the score achieved c indices of 0.62 and 0.61, respectively.

CONCLUSIONS

We present a novel and simple risk score for prediction of major bleeding in patients with non-valvular AF treated with NOACs. Validation in additional cohorts is warranted.

Evaluation of serious bleeding signals during concomitant use of clopidogrel and hypnotic drugs

BACKGROUND

In a previous drug-drug interaction (DDI) screening study intended to generate hypotheses, clopidogrel + either eszopiclone or zolpidem (vs. clopidogrel alone) were associated with serious bleeding.

OBJECTIVES

To confirm or refute these DDI signals and examine associations with other hypnotics in an independent population of United States Medicaid beneficiaries METHODS: We employed a bi-directional self-controlled case series design in eligible individuals concomitantly exposed to one of 12 hypnotics (precipitants, exposures of interest) plus either clopidogrel (the object drug) or pravastatin (the negative control object drug). The outcome was hospital presentation with serious bleeding. Using conditional Poisson regression, we calculated confounder-adjusted rate ratios (RRs) and 95% confidence intervals for serious bleeding during clopidogrel + precipitant use (vs. clopidogrel alone). To distinguish a DDI from a precipitant's inherent effect on bleeding, we divided effect measures by the adjusted RR for the corresponding pravastatin + precipitant pair to obtain ratios of RR (RRRs).

RESULTS

Among 23,194 users of clopidogrel and 3824 of pravastatin who experienced serious bleeding during an active prescription for one of these agents, confounder-adjusted RRRs for serious bleeding were 6.63 (0.39-113.01) and 0.77 (0.53-1.11) with eszopiclone and zolpidem, respectively, whereas confounder-adjusted RRRs for other hypnotics ranged from 0.18 (0.04-0.85) for triazolam to 1.79 (0.16-20.44) for zaleplon. Statistical imprecision therefore precluded us from confirming or refuting these prior signals with eszopiclone and zolpidem.

CONCLUSIONS

While we could not confirm or refute previously identified DDI signals, numerically elevated RRRs for serious bleeding with several clopidogrel + hypnotic pairs warrant further examination.

Clinical outcomes and costs associated with procalcitonin utilization in hospitalized patients with pneumonia, heart failure, viral respiratory infection, or chronic obstructive pulmonary disease

Lower respiratory tract infections (LRTIs) due to bacterial pneumonia are common among hospitalized patients and are frequently treated with antibiotics. Viral illnesses and exacerbations of heart failure or COPD may present with symptoms mimicking a LRTI, resulting in unnecessary antibiotic utilization. Procalcitonin testing may be useful in these clinical scenarios. We attempted to assess the utility of procalcitonin testing versus not testing, and positive versus negative results among hospitalized patients with suspected LRTI. We performed a retrospective cohort study using multivariable analysis comparing clinical outcomes of patients with and without procalcitonin testing. Patients were 18 years or older, hospitalized for pneumonia, heart failure, COPD, or a viral respiratory illness between October 2014 and October 2015 (n = 2353). All patients received at least one dose of antibiotics. Major outcomes were duration of antibiotic therapy, length of hospital stay, C. difficile testing and infections, and normalized total direct costs. Procalcitonin testing occurred in 14.0% of patients and pneumonia (70.6%) was the most common diagnosis. After covariate adjustments, mean length of stay (5.61 vs. 6.67 days, p < 0.001) and duration of antibiotics (3.95 vs. 4.47 days, p < 0.001) were shorter among tested patients. Fewer 30-day readmissions (OR 0.62, 95% CI 0.40-0.95) were observed, and total direct healthcare costs were 34% lower (0.66, 95% CI 0.58-0.74) among tested patients. Negative procalcitonin results were associated with further reductions in some outcomes. In conclusion, procalcitonin testing among hospitalized patients with suspected LRTI is associated with reductions in antibiotic duration, length of stay, 30-day readmission, and healthcare costs.

Accuracy of identifying hospital acquired venous thromboembolism by administrative coding: implications for big data and machine learning research

Big data analytics research using heterogeneous electronic health record (EHR) data requires accurate identification of disease phenotype cases and controls. Overreliance on ground truth determination based on administrative data can lead to biased and inaccurate findings. Hospital-acquired venous thromboembolism (HA-VTE) is challenging to identify due to its temporal evolution and variable EHR documentation. To establish ground truth for machine learning modeling, we compared accuracy of HA-VTE diagnoses made by administrative coding to manual review of gold standard diagnostic test results. We performed retrospective analysis of EHR data on 3680 adult stepdown unit patients identifying HA-VTE. International Classification of Diseases, Ninth Revision (ICD-9-CM) codes for VTE were identified. 4544 radiology reports associated with VTE diagnostic tests were screened using terminology extraction and then manually reviewed by a clinical expert to confirm diagnosis. Of 415 cases with ICD-9-CM codes for VTE, 219 were identified with acute onset type codes. Test report review identified 158 new-onset HA-VTE cases. Only 40% of ICD-9-CM coded cases (n = 87) were confirmed by a positive diagnostic test report, leaving the majority of administratively coded cases unsubstantiated by confirmatory diagnostic test. Additionally, 45% of diagnostic test confirmed HA-VTE cases lacked corresponding ICD codes. ICD-9-CM coding missed diagnostic test-confirmed HA-VTE cases and inaccurately assigned cases without confirmed VTE, suggesting dependence on administrative coding leads to inaccurate HA-VTE phenotyping. Alternative methods to develop more sensitive and specific VTE phenotype solutions portable across EHR vendor data are needed to support case-finding in big-data analytics.

Association Between Post-Dural Puncture Headache After Neuraxial Anesthesia in Childbirth and Intracranial Subdural Hematoma

Importance

Women giving birth have high rates of dural puncture secondary to neuraxial anesthesia and are at high risk for a resulting headache. It appears to be unknown whether there is a significant association between post-dural puncture headache and subsequent intracranial subdural hematoma.

Objective

To determine the association of post-dural puncture headache with postpartum intracranial subdural hematoma.

Design, Setting, and Participants

This cohort study of patients used hospital discharges recorded in the US Agency for Healthcare Research and Quality National Readmission Database for women who experienced childbirth from January 2010 to December 2016. Patients were included if they had been admitted for childbirth, had 2 months of follow-up data, and did not receive a diagnostic lumbar puncture. Only the first delivery for a calendar year was studied. Data were analyzed from January 2018 to June 2019.

Exposures

Women with post-dural puncture headache associated with neuraxial anesthesia in the 2-month postpartum period were identified using International Classification of Disease (Ninth Edition and Tenth Edition) codes and were compared with those without post-dural puncture headaches.

Main Outcome and Measures

The primary outcome was intracranial subdural hematoma in the 2-month postpartum period. Secondary outcomes included in-hospital mortality and occurrence of neurosurgery.

Results

A total of 26 469 771 patients with 26 498 194 deliveries were included. Exclusion of repeated deliveries (n = 28 423), deliveries without 2 months of follow-up data (n = 4 329 621), and deliveries with diagnostic lumbar puncture (n = 9334) resulted in a final cohort of 22 130 815 patients and deliveries. For the cohort, the mean (SD) age was 28.1 (6.0) years, and there were 68 374 post-dural puncture headaches, for an overall rate of 309 (95% CI, 302-316) per 100 000. There were 342 cases of subdural hematoma identified, indicating a rate of 1.5 (95% CI, 1.3-1.8) per 100 000 women. Of these, 100 cases were in women with post-dural puncture headache, indicating a rate of 147 (95% CI, 111-194) hematoma cases per 100 000 deliveries in this subgroup. Post-dural puncture headache had an unadjusted absolute risk increase of 145 (95% CI, 117-174) subdural hematoma cases per 100 000 deliveries. After adjusting for confounders, post-dural puncture headache had an odds ratio for subdural hematoma of 199 (95% CI, 126-317; P < .001) and an adjusted absolute risk increase of 130 (95% CI, 90-169; P < .001) per 100 000 deliveries.

Conclusions and Relevance

The presence of presumed post-dural puncture headache after neuraxial anesthesia in childbirth, compared with no headache, was associated with a small but statistically significant absolute increase in the risk of being diagnosed with intracranial subdural hematoma. Further research is needed to establish if this association is causal for this rare outcome.

Bleeding risk in patients with venous thromboembolic events treated with new oral anticoagulants

New oral anticoagulants (NOACs) is the preferred treatment in secondary prophylaxis of venous thromboembolic events (VTE). The aim of this study was to investigate possible risk factors associated with major bleeding in VTE-patients treated with NOACs. In this retrospective register-based study we screened the Swedish anticoagulation register Auricula (during 2012.01.01–2017.12.31) to find patients and used other national registers for outcomes. Primary endpoint was major bleeding defined as bleeding leading to hospital care. Multivariate Cox-regression analysis was used to reveal risk factors. 18 219 patients with NOAC due to VTE were included. 85.6% had their first VTE, mean age was 69.4 years and median follow-up time was 183 days. The most common NOAC was rivaroxaban (54.8%), followed by apixaban (42.0%), dabigatran (3.2%) and edoxaban (0.1%). The rate of major bleeding was 6.62 (95% CI 6.19–7.06) per 100 treatment years in all patients and 11.27 (CI 9.96–12.57) in patients above 80 years of age. Statistically independent risk factors associated with major bleeding were age (normalized HR 1.38, CI 1.27–1.50), earlier major bleeding (HR 1.58, Cl 1.09–2.30), COPD (HR 1.28, CI 1.04–1.60) and previous stroke (HR 1.28, Cl 1.03–1.58) or transient ischemic attack (TIA) (HR 1.33, Cl 1.01–1.76). Prior warfarin treatment was protective (HR 0.67, CI 0.58–0.78). This real world cohort shows a high bleeding rate especially among the elderly and in patients with previous major bleeding, COPD and previous stroke or TIA. This should be considered when deciding on treatment duration and NOAC dose in these patients.

Vitamin K versus warfarin interruption alone in patients without bleeding and an international normalized ratio > 10

BACKGROUND

Reversal of an international normalized ratio (INR) > 10 with vitamin K is recommended in patients experiencing bleeding; however, information on outcomes with reversal using vitamin K in non-bleeding patients is lacking.

OBJECTIVE

To compare clinical and safety outcomes between non-bleeding patients receiving warfarin with an INR > 10 who did and did not receive a prescription for vitamin K.

PATIENTS/METHODS

This was a retrospective cohort study conducted in an integrated health-care delivery system. Adult patients receiving warfarin therapy who experienced an INR > 10 without bleeding between 01/01/2006 and 06/30/2018 were included. Patients were assessed for an outpatient dispensing or in-office administration of vitamin K on the day of or the day after an INR > 10 and then clinically relevant bleeding, thromboembolism, all-cause mortality, and time to INR < 4 within the next 30 days.

RESULTS

A total of 809 patients was included with 332 and 477 who were and were not dispensed vitamin K, respectively. Overall, mean patient age was 71.7 years, 60.1% were female and the mean INR was 10.4 at presentation. There were no differences between groups in 30-day rates of bleeding or thromboembolism (both P > .05). Patients dispensed vitamin K had a higher likelihood of mortality (15.1% versus 10.1%, P = .032, adjusted odds ratio = 1.63, 95% confidence interval 1.03 to 2.57). Overall, time to an INR < 4 was similar between groups.

CONCLUSION

Vitamin K administration was not associated with improved clinical outcomes in asymptomatic patients with an INR > 10.

Coronary Artery Calcium for Personalized Allocation of Aspirin in Primary Prevention of Cardiovascular Disease in 2019: The MESA Study (Multi-Ethnic Study of Atherosclerosis)

BACKGROUND

Recent American College of Cardiology/American Heart Association Primary Prevention Guidelines recommended considering low-dose aspirin therapy only among adults 40 to 70 years of age who are at higher atherosclerotic cardiovascular disease (ASCVD) risk but not at high risk of bleeding. However, it remains unclear how these patients are best identified. The present study aimed to assess the value of coronary artery calcium (CAC) for guiding aspirin allocation for primary prevention by using 2019 aspirin meta-analysis data on cardiovascular disease relative risk reduction and bleeding risk.

METHODS

The study included 6470 participants from the MESA Study (Multi-Ethnic Study of Atherosclerosis). ASCVD risk was estimated using the pooled cohort equations, and 3 strata were defined: <5%, 5% to 20%, and >20%. All participants underwent CAC scoring at baseline, and CAC scores were stratified as =0, 1 to 99, ≥100, and ≥400. A 12% relative risk reduction in cardiovascular disease events was used for the 5-year number needed to treat (NNT₅) calculations, and a 42% relative risk increase in major bleeding events was used for the 5-year number needed to harm (NNH₅) estimations.

RESULTS

Only 5% of MESA participants would qualify for aspirin consideration for primary prevention according to the American College of Cardiology/American Heart Association guidelines and using >20% estimated ASCVD risk to define higher risk. Benefit/harm calculations were restricted to aspirin-naive participants <70 years of age not at high risk of bleeding (n=3540). The overall NNT₅ with aspirin to prevent 1 cardiovascular disease event was 476 and the NNH₅ was 355. The NNT₅ was also greater than or similar to the NNH₅ among estimated ASCVD risk strata. Conversely, CAC≥100 and CAC≥400 identified subgroups in which NNT₅ was lower than NNH₅. This was true both overall (for CAC≥100, NNT₅=140 versus NNH₅=518) and within ASCVD risk strata. Also, CAC=0 identified subgroups in which the NNT₅ was much higher than the NNH₅ (overall, NNT₅=1190 versus NNH₅=567).

CONCLUSIONS

CAC may be superior to the pooled cohort equations to inform the allocation of aspirin in primary prevention. Implementation of current 2019 American College of Cardiology/American Heart Association guideline recommendations together with the use of CAC for further risk assessment may result in a more personalized, safer allocation of aspirin in primary prevention. Confirmation of these findings in experimental settings is needed.

Reflection on modern methods: five myths about measurement error in epidemiological research

Epidemiologists are often confronted with datasets to analyse which contain measurement error due to, for instance, mistaken data entries, inaccurate recordings and measurement instrument or procedural errors. If the effect of measurement error is misjudged, the data analyses are hampered and the validity of the study's inferences may be affected. In this paper, we describe five myths that contribute to misjudgments about measurement error, regarding expected structure, impact and solutions to mitigate the problems resulting from mismeasurements. The aim is to clarify these measurement error misconceptions. We show that the influence of measurement error in an epidemiological data analysis can play out in ways that go beyond simple heuristics, such as heuristics about whether or not to expect attenuation of the effect estimates. Whereas we encourage epidemiologists to deliberate about the structure and potential impact of measurement error in their analyses, we also recommend exercising restraint when making claims about the magnitude or even direction of effect of measurement error if not accompanied by statistical measurement error corrections or quantitative bias analysis. Suggestions for alleviating the problems or investigating the structure and magnitude of measurement error are given.

Splanchnic vein thrombosis predicts worse survival in patients with advanced pancreatic cancer

INTRODUCTION

Pancreatic cancer is a thrombogenic malignancy with nearly half of venous thrombotic events occurring in the splanchnic circulation. The effect of splanchnic vein thrombosis on mortality in pancreatic cancer is unknown. We studied the effect of splanchnic vein thrombosis on mortality in veterans with advanced pancreatic adenocarcinoma, and explored the association of anticoagulant therapy on mortality and hemorrhage.

METHODS

Using International Classification of Diseases (ICD) codes, we identified eligible patients and outcomes in the Veterans Health Administration database. Using Cox proportional hazards regression, we analyzed the association between splanchnic vein thrombosis and mortality among patients with advanced pancreatic cancer. We used propensity score inverse probability-of-treatment weighting to balance the groups who did and did not receive anticoagulation. To understand the role of anticoagulant therapy, we used Cox proportional hazards regression to analyze mortality and competing risk analysis to assess the risk of hemorrhage.

RESULTS

Of the patients with advanced pancreatic cancer (N = 6164), 122 developed splanchnic vein thrombosis. Splanchnic vein thrombosis was associated with a two-fold increase in mortality, aHR 2.02, 95% CI 1.65-2.47. The finding held true after restricting the analysis to patients undergoing treatment for pancreatic cancer, and after adjusting for immortal time bias by a 30-day landmark analysis. Anticoagulant therapy did not affect mortality (aHR 0.99, 95% CI 0.65-1.51), and increased the risk of hemorrhage (aHR 2.7, 95% CI 1.02-7.07).

CONCLUSION

Splanchnic vein thrombosis predicts worse survival in patients with advanced pancreatic adenocarcinoma. Anticoagulant therapy may not mitigate this increased mortality, and increases the risk of hemorrhage.

Sensitivity and specificity of an algorithm based on medico-administrative data to identify hospitalized patients with major bleeding presenting to an emergency department

Background

Validation studies on an ICD-10-based algorithm to identify major bleeding events are scarce, and mostly focused on positive predictive values.

Objective

To evaluate the sensitivity and specificity of an ICD-10-based algorithm in adult patients referred to hospital.

Methods

This was a cross-sectional, retrospective analysis. Among all hospital stays of adult patients referred to Rennes University Hospital, France, through the emergency ward in 2014, we identified major bleeding events according to an index test based on a list of ICD-10 diagnoses. As a reference, a two-step process was applied: firstly, a computerized request for electronic health records from the emergency ward, using several hemorrhage-related diagnostic codes and specific emergency therapies so as to discard stays with a very low probability of bleeding; secondly, a chart review of selected records was conducted by a medical expert blinded to the index test results and each hospital stay was classified into one of two exclusive categories: major bleeding or no major bleeding, according to pre-specified criteria.

Results

Out of 16,012 hospital stays, the reference identified 736 major bleeding events and left 15,276 stays considered as without the target condition. The index test identified 637 bleeding events: 293 intracranial hemorrhages, 197 gastrointestinal hemorrhages and 147 other bleeding events. Overall, sensitivity was 65% (95%CI, 62 to 69), and specificity was 99.0%. We observed differential sensitivity and specificity across bleeding types, with the highest values for intracranial hemorrhage. Positive predictive values ranged from 59% for “other” bleeding events, to 71% (95%CI, 65 to 78) for gastrointestinal hemorrhage, and 96% for intracranial hemorrhage.

Conclusions

Low sensitivity and differential measures of accuracy across bleeding types support the need for specific data collection and medical validation rather than using an ICD-10-based algorithm for assessing the incidence of major bleeding.

Usefulness of Coronary Artery Calcium to Identify Adults of Sufficiently High Risk for Atherothrombotic Cardiovascular Events to Consider Low-Dose Rivaroxaban Thromboprophylaxis (from MESA)

Low-dose rivaroxaban was effective in secondary prevention of atherosclerotic cardiovascular disease (ASCVD) in the COMPASS trial. There is no established role, however, for oral anticoagulants in primary prevention. We evaluated whether coronary artery calcium (CAC) scoring identifies a high-risk primary prevention adult population who may benefit from low-dose rivaroxaban to prevent ASCVD events. We modeled expected outcomes of low-dose rivaroxaban in 5,196 Multiethnic Study of Atherosclerosis (MESA) cohort participants not already on antiplatelet or anticoagulant therapy. We applied relative risk ratios from COMPASS to absolute MESA event rates in order to estimate number needed to treat (NNT) to avoid a composite of cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke, as well as number needed to harm (NNH) to cause 1 hospitalized bleed; with both NNT and NNH stratified by calculated ASCVD risk and by baseline CAC. MESA participants with CAC ≥300 had crude ASCVD event rate of 20 per 1000 patient-years, which is comparable to that observed in the COMPASS control-arm. CAC was independently associated with the composite ASCVD outcome (p <0.001 for trend). However, CAC was not independently associated with adjusted hazard ratio for hospitalized major bleeding. Predicted 5-year NNT (modeled from COMPASS) was 75 in persons with CAC 100-299 and 45 with CAC ≥300 despite NNH values of 252 and 98, respectively. In conclusion, CAC helps to distinguish estimated ASCVD benefit from estimated bleeding harm, thereby identifying very high-risk primary prevention adults without established cardiovascular disease who may derive net-benefit from low-dose rivaroxaban.

Assessing major bleeding risk in atrial fibrillation patients concurrently taking non-vitamin K antagonist oral anticoagulants and antiepileptic drugs

Aims

This study compared the risk of major bleeding between atrial fibrillation (AF) patients who took non-vitamin K antagonist oral anticoagulants (NOACs) and antiepileptic drugs (AEDs) concurrently and those who took only NOACs.

Methods and results

We performed a retrospective cohort study using Taiwan National Health Insurance database and included AF patients who received NOAC prescriptions from 1 June 2012 to 31 December 2017. The major bleeding risks of person-quarters exposed to NOAC and 11 concurrent AEDs (carbamazepine, gabapentin, lamotrigine, levetiracetam, oxcarbazepine, phenobarbital, phenytoin, pregabalin, topiramate, valproic acid, and zonisamide) were compared with person-quarters exposed to NOAC alone. Adjusted incidence rate differences between NOAC with or without concurrent AEDs were estimated using Poisson regression models weighted by the inverse probability of treatment. Among 104 319 patients (age 75.0 ± 10.3 years; men, 56.2%), 8546 major bleeding events occurred during 731 723 person-quarters with NOAC prescriptions. Concurrent AED use was found in 15.3% of NOAC-treated patients. Concurrent use of NOAC with valproic acid, phenytoin, or levetiracetam increased adjusted incidence rates per 1000 person-years of major bleeding more significantly than NOAC alone: 153.49 for NOAC plus valproic acid vs. 55.06 for NOAC alone [difference 98.43, 95% confidence interval (CI) 82.37–114.49]; 135.83 for NOAC plus phenytoin vs. 54.43 for NOAC alone (difference 81.4, 95% CI 60.14–102.66); and 132.96 for NOAC plus levetiracetam vs. 53.08 for NOAC alone (difference 79.88, 95% CI 64.47–95.30).

Conclusion

For AF patients, the concurrent use of NOACs and valproic acid, phenytoin, or levetiracetam was associated with a higher risk of major bleeding.

Effect of data validation audit on hospital mortality ranking and pay for performance

Background

Quality improvement and epidemiology studies often rely on database codes to measure performance or impact of adjusted risk factors, but how validity issues can bias those estimates is seldom quantified.

Objectives

To evaluate whether and how much interhospital administrative coding variations influence a typical performance measure (adjusted mortality) and potential incentives based on it.

Design

National cross-sectional study comparing hospital mortality ranking and simulated pay-for-performance incentives before/after recoding discharge abstracts using medical records.

Setting

Twenty-four public and private hospitals located in France

Participants

All inpatient stays from the 78 deadliest diagnosis-related groups over 1 year.

Interventions

Elixhauser and Charlson comorbidities were derived, and mortality ratios were computed for each hospital. Thirty random stays per hospital were then recoded by two central reviewers and used in a Bayesian hierarchical model to estimate hospital-specific and comorbidity-specific predictive values. Simulations then estimated shifts in adjusted mortality and proportion of incentives that would be unfairly distributed by a typical pay-for-performance programme in this situation.

Main outcome measures

Positive and negative predictive values of routine coding of comorbidities in hospital databases, variations in hospitals’ mortality league table and proportion of unfair incentives.

Results

A total of 70 402 hospital discharge abstracts were analysed, of which 715 were recoded from full medical records. Hospital comorbidity-level positive predictive values ranged from 64.4% to 96.4% and negative ones from 88.0% to 99.9%. Using Elixhauser comorbidities for adjustment, 70.3% of hospitals changed position in the mortality league table after correction, which added up to a mean 6.5% (SD 3.6) of a total pay-for-performance budget being allocated to the wrong hospitals. Using Charlson, 61.5% of hospitals changed position, with 7.3% (SD 4.0) budget misallocation.

Conclusions

Variations in administrative data coding can bias mortality comparisons and budget allocation across hospitals. Such heterogeneity in data validity may be corrected using a centralised coding strategy from a random sample of observations.

Comparison of 2 Natural Language Processing Methods for Identification of Bleeding Among Critically Ill Patients

IMPORTANCE

To improve patient safety, health care systems need reliable methods to detect adverse events in large patient populations. Events are often described in clinical notes, rather than structured data, which make them difficult to identify on a large scale.

OBJECTIVE

To develop and compare 2 natural language processing methods, a rules-based approach and a machine learning (ML) approach, for identifying bleeding events in clinical notes.

DESIGN, SETTING, AND PARTICIPANTS

This diagnostic study used deidentified notes from the Medical Information Mart for Intensive Care, which spans 2001 to 2012. A training set of 990 notes and a test set of 660 notes were randomly selected. Physicians classified each note as present or absent for a clinically relevant bleeding event during the hospitalization. A bleeding dictionary was developed for the rules-based approach; bleeding mentions were then aggregated to arrive at a classification for each note. Three ML models (support vector machine, extra trees, and convolutional neural network) were developed and trained using the 990-note training set. Another instance of each ML model was also trained on a sample of 450 notes, with equal numbers of bleeding-present and bleeding-absent notes. The notes were represented using term frequency-inverse document frequency vectors and global vectors for word representation.

MAIN OUTCOMES AND MEASURES

The main outcomes were accuracy, sensitivity, specificity, positive predictive value, and negative predictive value for each model. Following training, the models were tested on the test set and sensitivities were compared using a McNemar test.

RESULTS

The 990-note training set represented 769 patients (296 [38.5%] female; mean [SD] age, 67.42 [14.7] years). The 660-note test set represented 527 patients (211 [40.0%] female; mean [SD] age, 67.86 [14.7] years). Bleeding was present in 146 notes (22.1%). The extra trees down-sampled model and rules-based approaches were similarly sensitive (93.8% vs 91.1%; difference, 2.7%; 95% CI, -3.8% to 7.9%; P = .44). The positive predictive value for the extra trees model, however, was 48.6%. The rules-based model had the best performance overall, with 84.6% specificity, 62.7% positive predictive value, and 97.1% negative predictive value.

CONCLUSIONS AND RELEVANCE

Bleeding is a common complication in health care, and these results demonstrate an automated and scalable detection method. The rules-based natural language processing approach, compared with ML, had the best performance in identifying bleeding, with high sensitivity and negative predictive value.

Disease burden, complication rates, and health-care costs of heparin-induced thrombocytopenia in the USA: a population-based study

BACKGROUND

Heparin-induced thrombocytopenia can be a life-threatening and limb-threatening complication of heparin therapy. Incidence and complication rates of this condition have been extrapolated from studies with modest sample sizes, and despite the availability of therapeutic interventions the outcomes of heparin-induced thrombocytopenia are not well understood. We aimed to estimate disease burden, complication rates, and costs of heparin-induced thrombocytopenia in the USA.

METHODS

In this population-based study we analysed data from 2009 to 2013 from the Nationwide (National) Inpatient Sample (NIS), a large, all-payer inpatient health-care database in the USA. To validate the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code for heparin-induced thrombocytopenia (289.84), we defined the sensitivity and specificity of this code using patient data from 2013 from a local hospital (Froedtert Memorial Lutheran Hospital, Milwaukee, WI, USA). The primary outcomes assessed were the incidence of hospital discharges with codes for heparin-induced thrombocytopenia and of discharges for heparin-induced thrombocytopenia associated with cardiopulmonary bypass, haemodialysis, hip or knee arthroplasty, trauma or injury (or both), and gingival or periodontal disease (or both). We also assessed the incidence of thrombosis, bleeding, limb or digit amputation, mortality, length of hospital stay, and associated hospital charges.

FINDINGS

Between 2009 and 2013, 97 566 discharges from the NIS assigned the ICD-9-CM code for heparin-induced thrombocytopenia, and 149 911 247 discharges coded for non-heparin-induced thrombocytopenia, were analysed. Overall, heparin-induced thrombocytopenia was identified in 97 566 (0·065%; SE 0·0012) of 150 008 813 discharges, corresponding to approximately one in 1500 hospital admissions. Patients undergoing cardiopulmonary bypass had the highest rates of heparin-induced thrombocytopenia (7702 [0·63%; SE 0·03] of 1 230 362), followed by those undergoing haemodialysis (23 012 [0·47%; 0·01] of 4 908 100), those with gingival or periodontal disease, or both (106 [0·12%; 0·03] of 88 621), and those with trauma or injury, or both (541 [0·09%; 0·01] of 602 944); patients with hip (845 [0·04%; 0·004] of 1 943 353) and knee (676 [0·02%; 0·002] of 3 022 602) arthroplasty had the lowest rates of heparin-induced thrombocytopenia. Thrombosis (29 079 [29·8%; SE 0·4] of 97 566) and bleeding (6044 [6·2%; 0·2] of 97 566) were common complications in heparin-induced thrombocytopenia, and 1446 (23·9%; 1·2) of 6044 patients with heparin-induced thrombocytopenia who had haemorrhage died. 742 (0·76%; SE 0·06) of 97 566 patients with heparin-induced thrombocytopenia discharges underwent amputations compared with 173 043 (0·12%; 0·001) of 149 911 247 with non-heparin-induced thrombocytopenia discharges (adjusted odds ratio 5·095 [95% CI 4·309-6·023]; p<0·0001). Overall, in-hospital mortality was 9842 (10·1%; SE 0·2) of 97 508 in discharge summaries coded for heparin-induced thrombocytopenia compared with 3 206 700 (2·1%; 0·01) of 149 811 891 in discharges for non-heparin-induced thrombocytopenia (adjusted odds ratio 4·075 [95% CI 3·846-4·317]; p<0·0001). The median length of stay among live discharges was 8·9 days (IQR 4·6-17·1) and total hospital charges were US$83 072 (IQR 37 240-188 419) for heparin-induced thrombocytopenia discharges compared with 2·6 days (1·4-4·8) and $21 360 (11 426-41 917) for non-heparin-induced thrombocytopenia discharges (p<0·0001 for both). 333 discharges from a local hospital were analysed to assess the diagnostic sensitivity and specificity of the heparin-induced thrombocytopenia ICD-9-CM code; sensitivity was 90·9% (95% CI 57·1-99·5) and specificity was 94·4% (91·1-96·6).

INTERPRETATION

Complication rates for heparin-induced thrombocytopenia remain high and more effective preventive and treatment interventions are needed.

FUNDING

None.