Individual meropenem epithelial lining fluid and plasma PK/PD target attainment

It is unclear whether plasma is a reliable surrogate for target attainment in the epithelial lining fluid (ELF). The objective of this study was to characterize meropenem target attainment in plasma and ELF using prospective samples. The first 24-hour T>MIC was evaluated vs 1xMIC and 4xMIC targets at the patient (i.e., fixed MIC of 2 mg/L) and population [i.e., cumulative fraction of response (CFR) according to EUCAST MIC distributions] levels for both plasma and ELF. Among 65 patients receiving ≥24 hours of treatment, 40% of patients failed to achieve >50% T>4xMIC in plasma and ELF, and 30% of patients who achieved >50% T>4xMIC in plasma had <50% T>4xMIC in ELF. At 1xMIC and 4xMIC targets, 3% and 25% of patients with >95% T>MIC in plasma had <50% T>MIC in ELF, respectively. Those with a CRCL >115 mL/min were less likely to achieve >50%T>4xMIC in ELF (P < 0.025). In the population, CFR for Escherichia coli at 1xMIC and 4xMIC was >97%. For Pseudomonas aeruginosa, CFR was ≥90% in plasma and ranged 80%-85% in ELF at 1xMIC when a loading dose was applied. CFR was reduced in plasma (range: 75%-81%) and ELF (range: 44%-60%) in the absence of a loading dose at 1xMIC. At 4xMIC, CFR for P. aeruginosa was 60%-86% with a loading dose and 18%-62% without a loading dose. We found that plasma overestimated ELF target attainment inup to 30% of meropenem-treated patients, CRCL >115 mL/min decreased target attainment in ELF, and loading doses increased CFR in the population.

Machine Learning To Stratify Methicillin-Resistant Staphylococcus aureus Risk among Hospitalized Patients with Community-Acquired Pneumonia

Methicillin-resistant Staphylococcus aureus (MRSA) is an uncommon but serious cause of community-acquired pneumonia (CAP). A lack of validated MRSA CAP risk factors can result in overuse of empirical broad-spectrum antibiotics. We sought to develop robust models predicting the risk of MRSA CAP using machine learning using a population-based sample of hospitalized patients with CAP admitted to either a tertiary academic center or a community teaching hospital. Data were evaluated using a machine learning approach. Cases were CAP patients with MRSA isolated from blood or respiratory cultures within 72 h of admission; controls did not have MRSA CAP. The Classification Tree Analysis algorithm was used for model development. Model predictions were evaluated in sensitivity analyses. A total of 21 of 1,823 patients (1.2%) developed MRSA within 72 h of admission. MRSA risk was higher among patients admitted to the intensive care unit (ICU) in the first 24 h who required mechanical ventilation than among ICU patients who did not require ventilatory support (odds ratio [OR], 8.3; 95% confidence interval [CI], 2.4 to 32). MRSA risk was lower among patients admitted to ward units than among those admitted to the ICU (OR, 0.21; 95% CI, 0.07 to 0.56) and lower among ICU patients without a history of antibiotic use in the last 90 days than among ICU patients with antibiotic use in the last 90 days (OR, 0.03; 95% CI, 0.002 to 0.59). The final machine learning model was highly accurate (receiver operating characteristic [ROC] area = 0.775) in training and jackknife validity analyses. We identified a relatively simple machine learning model that predicted MRSA risk in hospitalized patients with CAP within 72 h postadmission.

The First 2 Years of Biosimilar Epoetin for Cancer and Chemotherapy-Induced Anemia in the U.S.: A Review from the Southern Network on Adverse Reactions

Biosimilars are biologic drug products that are highly similar to reference products in analytic features, pharmacokinetics and pharmacodynamics, immunogenicity, safety, and efficacy. Biosimilar epoetin received Food and Drug Administration (FDA) approval in 2018. The manufacturer received an FDA nonapproval letter in 2017, despite receiving a favorable review by FDA's Oncologic Drugs Advisory Committee (ODAC) and an FDA nonapproval letter in 2015 for an earlier formulation. We discuss the 2018 FDA approval, the 2017 FDA ODAC Committee review, and the FDA complete response letters in 2015 and 2017; review concepts of litigation, naming, labeling, substitution, interchangeability, and pharmacovigilance; review European and U.S. oncology experiences with biosimilar epoetin; and review the safety of erythropoiesis-stimulating agents. In 2020, policy statements from AETNA, United Health Care, and Humana indicated that new epoetin oncology starts must be for biosimilar epoetin unless medical need for other epoetins is documented. Empirical studies report that as of 2012, reference epoetin use decreased from 40%-60% of all patients with cancer with chemotherapy-induced anemia to <5% of such patients because of safety concerns. Between 2018 and 2020, biosimilar epoetin use varied, increasing to 81% among one private insurer's patients covered by Medicare whose cancer care is administered with Oncology Analytics and to 41% with the same private insurer's patients with cancer covered by commercial health insurance and administered by the private insurer, to 0% in several Veterans Administration Hospitals, increasing to 100% in one large county hospital in California, and with yet-to-be-reported data from most oncology settings. We conclude that biosimilar epoetin appears to have overcome some barriers since 2015, although current uptake in the U.S. is variable. Pricing and safety considerations for all erythropoiesis-stimulating agents are primary determinants of biosimilar epoetin oncology uptake. IMPLICATIONS FOR PRACTICE: Few oncologists understand substitution and interchangeability of biosimilars with reference drugs. Epoetin biosimilar is new to the market, and physician and patient understanding is limited. The development of epoetin biosimilar is not familiar to oncologists.

Improving oncology biosimilar launches in the EU, the USA, and Japan: an updated Policy Review from the Southern Network on Adverse Reactions

The EU, the USA, and Japan account for the majority of biological pharmacotherapy use worldwide. Biosimilar regulatory approval pathways were authorised in the EU (2006), in Japan (2009), and in the USA (2015), to facilitate approval of biological drugs that are highly similar to reference products and to encourage market competition. Between 2007 and 2020, 33 biosimilars for oncology were approved by the European Medicines Agency (EMA), 16 by the US Food and Drug Administration (FDA), and ten by the Japan Pharmaceuticals and Medical Devices Agency (PMDA). Some of these approved applications were initially rejected because of manufacturing concerns (four of 36 [11%] with the EMA, seven of 16 [44%] with the FDA, none of ten for the PMDA). Median times from initial regulatory submission before approval of oncology biosimilars were 1·5 years (EMA), 1·3 years (FDA), and 0·9 years (PMDA). Pharmacists can substitute biosimilars for reference biologics in some EU countries, but not in the USA or Japan. US regulation prohibits substitution, unless the biosimilar has been approved as interchangeable, a designation not yet achieved for any biosimilar in the USA. Japan does not permit biosimilar substitution, as prescribers must include the product name on each prescription and that specific product must be given to the patient. Policy Reviews published in 2014 and 2016 in The Lancet Oncology focused on premarket and postmarket policies for oncology biosimilars before most of these drugs received regulatory approval. In this Policy Review from the Southern Network on Adverse Reactions, we identify factors preventing the effective launch of oncology biosimilars. Introduction to the market has been more challenging with therapeutic than for supportive care oncology biosimilars. Addressing region-specific competition barriers and educational needs would improve the regulatory approval process and market launches for these biologics, therefore expanding patient access to these products in the EU, the USA, and Japan.

Consequences to patients, clinicians, and manufacturers when very serious adverse drug reactions are identified (1997-2019): A qualitative analysis from the Southern Network on Adverse Reactions (SONAR)

BACKGROUND

Adverse drug/device reactions (ADRs) can result in severe patient harm. We define very serious ADRs as being associated with severe toxicity, as measured on the Common Toxicity Criteria Adverse Events (CTCAE)) scale, following use of drugs or devices with large sales, large financial settlements, and large numbers of injured persons. We report on impacts on patients, clinicians, and manufacturers following very serious ADR reporting.

METHODS

We reviewed clinician identified very serious ADRs published between 1997 and 2019. Drugs and devices associated with reports of very serious ADRs were identified. Included drugs or devices had market removal discussed at Food and Drug Advisory (FDA) Advisory Committee meetings, were published by clinicians, had sales > $1 billion, were associated with CTCAE Grade 4 or 5 toxicity effects, and had either >$1 billion in settlements or >1,000 injured patients. Data sources included journals, Congressional transcripts, and news reports. We reviewed data on: 1) timing of ADR reports, Boxed warnings, and product withdrawals, and 2) patient, clinician, and manufacturer impacts. Binomial analysis was used to compare sales pre- and post-FDA Advisory Committee meetings.

FINDINGS

Twenty very serious ADRs involved fifteen drugs and one device. Legal settlements totaled $38.4 billion for 753,900 injured persons. Eleven of 18 clinicians (61%) reported harms, including verbal threats from manufacturer (five) and loss of a faculty position (one). Annual sales decreased 94% from $29.1 billion pre-FDA meeting to $4.9 billion afterwards (p<0.0018). Manufacturers of four drugs paid $1.7 billion total in criminal fines for failing to inform the FDA and physicians about very serious ADRs. Following FDA approval, the median time to ADR reporting was 7.5 years (Interquartile range 3,13 years). Twelve drugs received Box warnings and one drug received a warning (median, 7.5 years following ADR reporting (IQR 5,11 years). Six drugs and 1 device were withdrawn from marketing (median, 5 years after ADR reporting (IQR 4,6 years)).

INTERPRETATION

Because very serious ADRs impacts are so large, policy makers should consider developing independently funded pharmacovigilance centers of excellence to assist with clinician investigations.

FUNDING

This work received support from the National Cancer Institute (1R01 CA102713 (CLB), https://www.nih.gov/about-nih/what-we-do/nih-almanac/national-cancer-institute-nci; and two Pilot Project grants from the American Cancer Society's Institutional Grant Award to the University of South Carolina (IRG-13-043-01) https://www.cancer.org/ (SH; BS).

803. Risk factors associated with Clostridioides difficile infection in hospitalized patients with community-acquired pneumonia

Background

Patients with community-acquired pneumonia (CAP) who are hospitalized and treated with antibiotics may carry an increased risk for developing Clostridioides difficile infection (CDI). Accurate risk estimation tools are needed to guide monitoring and CDI mitigation efforts. We aimed to identify patient-specific risk factors associated with CDI among hospitalized patients with CAP.

Methods

Design: retrospective case-control study of hospitalized patients who received CAP-directed antibiotic therapy between 1/1/2014 and 5/29/2018. Cases were hospitalized CAP patients who developed CDI post-admission. Control patients did not develop CDI and were selected at random from CAP patients hospitalized during this period. Variables: comorbidities, laboratory results, vital signs, severity of illness, prior hospitalization, and past antibiotic use. Propensity-score weights: identified via structural decomposition analysis of pre-treatment variables. Analysis: weighted classification tree models that predicted any CDI, hospital-onset CDI, and any healthcare-associated CDI according to CAP antibiotic treatment. Performance: percent accuracy in classification (PAC) and weighted positive (PPV) and negative predictive values (NPV). Modeling: completed using the ODA package (v1.0.1.3) for R (v3.5.1).

Results

A total of 32 cases and 232 controls were identified. Sixty pre-treatment variables were screened. Structural decomposition analysis, completed in two stages, identified prior hospitalization (OR 6.56, 95% CI: 3.01-14.31; PAC: 80.3%) and BUN greater than 29 mg/dL (OR 11.67, 95% CI: 2.41-56.5; PAC: 80.8%) as propensity-score weights. With respect to CDI, receipt of broad-spectrum anti-pseudomonal antibiotics was significantly (all P’s< 0.05) associated with any CDI (NPV: 90.29%, PPV: 27.94%), hospital-onset CDI (NPV: 97.53%, PPV: 26.86%), and healthcare-associated CDI (NPV: 92.89%, PPV: 27.94%).

Conclusion

We identified risk factors available at hospital admission and empiric use of broad-spectrum Gram-negative antibiotics as being associated with the development of CDI. Model PPVs were over two-fold greater than our sample base rate. Increased monitoring and avoidance of overly broad antibiotic use in high-risk patients appears warranted.

Disclosures

All Authors: No reported disclosures

238. Novel Way to Evaluate Antibiotic Use Appropriateness: Moving Towards the “Never Event” Classification by Electronic Algorithm

Background

Antibiotic use is commonly tracked electronically by antimicrobial stewardship programs (ASPs). Traditionally, evaluating the appropriateness of antibiotic use requires time- and labor-intensive manual review of each drug order. A drug-specific “appropriateness” algorithm applied electronically would improve the efficiency of ASPs. We thus created an antibiotic “never event” (NE) algorithm to evaluate vancomycin use, and sought to determine the performance characteristics of the electronic data capture strategy.

Methods

An antibiotic NE algorithm was developed to characterize vancomycin use (Figure) at a large academic institution (1/2016–8/2019). Patients were electronically classified according to the NE algorithm using data abstracted from their electronic health record. Type 1 NEs, defined as continued use of vancomycin after a vancomycin non-susceptible pathogen was identified, were the focus of this analysis. Type 1 NEs identified by automated data capture were reviewed manually for accuracy by either an infectious diseases (ID) physician or an ID pharmacist. The positive predictive value (PPV) of the electronic data capture was determined.

Antibiotic Never Event (NE) Algorithm to Characterize Vancomycin Use

Results

A total of 38,774 unique cases of vancomycin use were available for screening. Of these, 0.6% (n=225) had a vancomycin non-susceptible pathogen identified, and 12.4% (28/225) were classified as a Type 1 NE by automated data capture. All 28 cases included vancomycin-resistant Enterococcus spp (VRE). Upon manual review, 11 cases were determined to be true positives resulting in a PPV of 39.3%. Reasons for the 17 false positives are given in Table 1. Asymptomatic bacteriuria (ASB) due to VRE in scenarios where vancomycin was being appropriately used to treat a concomitant vancomycin-susceptible infection was the most common reason for false positivity, accounting for 64.7% of false positive cases. After removing urine culture source (n=15) from the algorithm, PPV improved to 53.8%.

Conclusion

An automated vancomycin NE algorithm identified 28 Type 1 NEs with a PPV of 39%. ASB was the most common cause of false positivity and removing urine culture as a source from the algorithm improved PPV. Future directions include evaluating Type 2 NEs (Figure) and prospective, real-time application of the algorithm.

Disclosures

Marc H. Scheetz, PharmD, MSc, Merck and Co. (Grant/Research Support)

Type A behaviour and psychological androgyny among Greek college students

Research conducted primarily in the United States suggests that the combination of Type A behaviour, high instrumentality, and low expressiveness may place individuals at relatively high risk for coronary‐artery and heart disease. The present research investigates the trans‐societal generalizability of the structure, reliability, relationships among, and distributional characteristics of these measures for 117 college students in Athens, Greece. As hypothesized, the Greek students were significantly less instrumental and less expressive than the American students, and scored significantly higher on the measure of Type A. Although there were significant mean differences between these samples, the results suggest that many of the interscale relationships are comparable, supporting an etic (universal) interpretation of the findings. Nevertheless, the results suggested several emic (unique) characteristics of these measures and their interrelationships.

End of an era of administering erythropoiesis stimulating agents among Veterans Administration cancer patients with chemotherapy-induced anemia

Erythropoisis stimulating agent (ESA) use was addressed in Food and Drug Administration (FDA) Oncology Drug Advisory Committee (ODAC) meetings between 2004 and 2008. FDA safety-focused regulatory actions occurred in 2007 and 2008. In 2007, black box warnings advised of early death and venous thromboembolism (VTE) risks with ESAs in oncology. In 2010, a Risk Evaluation Strategies (REMS) was initiated, with cancer patient consent that mortality and VTE risks were noted with ESAs. We report warnings and REMS impacts on ESA utilization among Veterans Administration (VA) cancer patients with chemotherapy-induced anemia (CIA). Data were from Veterans Affairs database (2003–2012). Epoetin and darbepoetin use were primary outcomes. Segmented linear regression was used to estimate changes in ESA use levels and trends, clinical appropriateness, and adverse events (VTEs) among chemotherapy-treated cancer patients. To estimate changes in level of drug prescription rate after policy actions, model-specific indicator variables as covariates based on specific actions were included. ESA use fell by 95% and 90% from 2005, for epoetin and darbepoetin, from 22% and 11%, respectively, to 1% and 1%, respectively, among cancer patients with CIA, respectively (p<0.01). Following REMS in 2010, mean hematocrit levels at ESA initiation decreased from 30% to 21% (p<0.01). Black box warnings preceded decreased ESA use among VA cancer patients with CIA. REMS was followed by reduced hematocrit levels at ESA initiation. Our findings contrast with privately- insured and Medicaid insured cancer patient data on chemotherapy-induced anemia where ESA use decreased to 3% to 7% by 2010–2012. By 2012, the era of ESA administration to VA to cancer patients had ended but the warnings remain relevant and significant. In 2019, oncology/hematology national guidelines (ASCO/ASH) recommend that cancer patients with chemotherapy-induced anemia should receive ESAs or red blood cell transfusions after risk-benefit evaluation.

End of an era for erythropoiesis-stimulating agents in oncology

Erythropoiesis-stimulating agents (ESAs) are available to treat chemotherapy-induced anemia (CIA). In 2007-2008, regulatory notifications advised of venous thromboembolism and mortality risks while the Center for Medicare and Medicaid Services' restricted ESA initiation to patients with hemoglobin <10 g/dl. In 2010, a Risk Evaluation and Mitigation Strategies required consent prior to administration. We evaluated ESA utilization from 2003 to 2012 and obtained private health insurer claims data for persons with lung, colorectal, or breast cancer from 2001 to 2012. ESA use for CIA was determined by an ESA claim after chemotherapy, up to 6 months after treatment. We identified 839,948 commercially insured patients, including 24,785 patients with ESA-treated CIA (3.2%). Darbepoetin use increased 3.9-fold from 2003 to 2007 (12.3% to 48.7%) and then decreased 95% to 2.6% by 2012. Epoetin use decreased 90% from 2003 to 2012 (30.3% to 3.1%). Between 2003 and 2012, mean epoetin dosing decreased 0.8-fold (244,979 in 2003 vs. 196,216 units in 2012), but increased 1.8-fold for darbepoetin-treated CIA (262 in 2003 to 467 μg in 2012). Among CIA patients, transfusions were low (4.5%) in 2002-2007, then increased 2.2-fold between 2008 and 2012. Safety initiatives between 2007 and 2010 facilitated reductions in ESA use combined with changes in coverage. These data show the efficacy of regulatory efforts, publication of adverse events and changes in reimbursement in reducing use of ESAs. Future studies are warranted to optimize deimplementation strategies to improve patient safety.

Caveat Medicus: Clinician experiences in publishing reports of serious oncology-associated adverse drug reactions

Oncology-associated adverse drug/device reactions can be fatal. Some clinicians who treat single patients with severe oncology-associated toxicities have researched case series and published this information. We investigated motivations and experiences of select individuals leading such efforts. Clinicians treating individual patients who developed oncology-associated serious adverse drug events were asked to participate. Inclusion criteria included having index patient information, reporting case series, and being collaborative with investigators from two National Institutes of Health funded pharmacovigilance networks. Thirty-minute interviews addressed investigational motivation, feedback from pharmaceutical manufacturers, FDA personnel, and academic leadership, and recommendations for improving pharmacovigilance. Responses were analyzed using constant comparative methods of qualitative analysis. Overall, 18 clinicians met inclusion criteria and 14 interviewees are included. Primary motivations were scientific curiosity, expressed by six clinicians. A less common theme was public health related (three clinicians). Six clinicians received feedback characterized as supportive from academic leaders, while four clinicians received feedback characterized as negative. Three clinicians reported that following the case series publication they were invited to speak at academic institutions worldwide. Responses from pharmaceutical manufacturers were characterized as negative by 12 clinicians. One clinician’s wife called the post-reporting time the “Maalox month,” while another clinician reported that the manufacturer collaboratively offered to identify additional cases of the toxicity. Responses from FDA employees were characterized as collaborative for two clinicians, neutral for five clinicians, unresponsive for negative by six clinicians. Three clinicians endorsed developing improved reporting mechanisms for individual physicians, while 11 clinicians endorsed safety activities that should be undertaken by persons other than a motivated clinician who personally treats a patient with a severe adverse drug/device reaction. Our study provides some of the first reports of clinician motivations and experiences with reporting serious or potentially fatal oncology-associated adverse drug or device reactions. Overall, it appears that negative feedback from pharmaceutical manufacturers and mixed feedback from the academic community and/or the FDA were reported. Big data, registries, Data Safety Monitoring Boards, and pharmacogenetic studies may facilitate improved pharmacovigilance efforts for oncology-associated adverse drug reactions. These initiatives overcome concerns related to complacency, indifference, ignorance, and system-level problems as barriers to documenting and reporting adverse drug events- barriers that have been previously reported for clinician reporting of serious adverse drug reactions.

Regulatory and Clinical Experiences with Biosimilar Filgrastim in the U.S., the European Union, Japan, and Canada

Biosimilar filgrastims are primarily indicated for chemotherapy-induced neutropenia prevention. They are less expensive formulations of branded filgrastim, and biosimilar filgrastim was the first biosimilar oncology drug administered in European Union (EU) countries, Japan, and the U.S. Fourteen biosimilar filgrastims have been marketed in EU countries, Japan, the U.S., and Canada since 2008, 2012, 2015, and 2016, respectively. We reviewed experiences and policies for biosimilar filgrastim markets in EU countries and Japan, where uptake has been rapid, and in the U.S. and Canada, where experience is rapidly emerging. U.S. regulations for designating biosimilar interchangeability are under development, and such regulations have not been developed in most other countries. Pharmaceutical substitution is allowed for new filgrastim starts in some EU countries and in Canada, but not Japan and the U.S. In EU countries, biosimilar adoption is facilitated with favorable hospital tender offers. U.S. adoption is reportedly 24%, while the second filgrastim biosimilar is priced 30% lower than branded filgrastim and 20% lower than the first biosimilar filgrastim approved by the U.S. Food and Drug Administration. Utilization is about 60% in EU countries, where biosimilar filgrastim is marketed at a 30%-40% discount. In Japan, biosimilar filgrastim utilization is 45%, primarily because of 35% discounts negotiated by Central Insurance and hospital-only markets. Overall, biosimilar filgrastim adoption barriers are small in many EU countries and Japan and are diminishing in Canada in the U.S. Policies facilitating improved U.S. adoption of biosimilar filgrastim, based on positive experiences in EU countries and Japan, including favorable insurance coverage; larger price discount relative to reference filgrastim pricing; closing of the "rebate trap" with transparent pricing information; formal educational efforts of patients, physicians, caregivers, and providers; and allowance of pharmaceutical substitution of biosimilar versus reference filgrastim, should be considered. IMPLICATIONS FOR PRACTICE: We reviewed experiences and policies for biosimilar filgrastims in Europe, Japan, Canada, and the U.S. Postmarketing harmonization of regulatory policies for biosimilar filgrastims has not occurred. Acceptance of biosimilar filgrastims for branded filgrastim, increasing in the U.S. and in Canada, is commonplace in Japan and Europe. In the U.S., some factors, accepted in Europe or Japan, could improve uptake, including acceptance of biosimilars as safe and effective; larger cost savings, decreasing "rebate traps" where pharmaceutical benefit managers support branded filgrastim, decreased use of patent litigation/challenges, and allowing pharmacists to routinely substitute biosimilar for branded filgrastim.

Using machine learning to evaluate treatment effects in multiple-group interrupted time series analysis

RATIONALE, AIMS, AND OBJECTIVES

Interrupted time series analysis (ITSA) is a popular evaluation methodology in which a single treatment unit's outcome is studied over time, and the intervention is expected to "interrupt" the level and/or trend of the outcome, subsequent to its introduction. The internal validity of this analysis is strengthened considerably if the treated unit is contrasted with a comparable control group. In this paper, we introduce a novel machine learning approach using optimal discriminant analysis (ODA) to evaluate treatment effects in multiple-group ITSA.

METHOD

We evaluate the effect of California's Proposition 99 (passed in 1988) for reducing cigarette sales, by comparing California (CA) to Montana (MT)-the best matching control state not exposed to any smoking reduction initiatives. We contrast results from ODA to those of ITSA regression (ITSAREG)-a commonly used approach for evaluating treatment effects in ITSA studies.

RESULTS

Both approaches found CA and MT to be comparable on their preintervention time series, and both approaches equally found CA to have statistically lower cigarette sales in the post-intervention period (P < 0.0001). The ODA model achieved very high effect strength of sensitivity (a measure of classification accuracy) of 91.67%, which remained high (75.00%) after conducting leave-one-out analysis to assess generalizability.

CONCLUSIONS

The ODA framework achieved results comparable to ITSAREG, bolstering confidence in the intervention effect. In addition, ODA confers several advantages over conventional approaches that may make it a better approach to use in multiple group ITSA studies: insensitivity to skewed data, model-free permutation tests to derive P values, identification of the threshold value which best discriminates intervention and control groups, a chance- and maximum-corrected index of classification accuracy, and cross-validation to assess generalizability.

Identifying causal mechanisms in health care interventions using classification tree analysis

RATIONALE, AIMS, AND OBJECTIVES

Mediation analysis identifies causal pathways by testing the relationships between the treatment, the outcome, and an intermediate variable that mediates the relationship between the treatment and outcome. This paper introduces classification tree analysis (CTA), a machine-learning procedure, as an alternative to conventional methods for analysing mediation effects.

METHOD

Using data from the JOBS II study, we compare CTA to structural equation models (SEMs) by assessing their consistency in revealing mediation effects on 2 outcomes; reemployment (a binary variable) and depressive symptoms (a continuous variable). Because study participants were not randomized sequentially to both treatment and mediator, an additional model was generated including baseline covariates to strengthen the validity of some key identifying assumptions required of all mediation analyses.

RESULTS

Using SEM, no statistically significant treatment or mediated effects were found for either outcome. In contrast, CTA found a significant treatment effect for reemployment (P = .047) and a mediated pathway for individuals in the treatment group (P = .014). No CTA model could be generated for the reemployment outcome. When covariates were added to the model, CTA found numerous interactions, while SEM found no effects.

CONCLUSIONS

CTA may uncover mediation effects where conventional approaches do not, because CTA does not require any assumptions about the distribution of variables nor of the functional form of the model, and CTA will systematically identify all statistically viable interactions. The versatility of CTA enables the investigator to explore the theorized underlying causal mechanism of an intervention in a much more comprehensive manner than conventional mediation analytic approaches.

Estimating causal effects for survival (time-to-event) outcomes by combining classification tree analysis and propensity score weighting

RATIONALE, AIMS AND OBJECTIVES

A common approach to assessing treatment effects in nonrandomized studies with time-to-event outcomes is to estimate propensity scores and compute weights using logistic regression, test for covariate balance, and then estimate treatment effects using Cox regression. A machine-learning alternative-classification tree analysis (CTA)-used to generate propensity scores and to estimate treatment effects in time-to-event data may identify complex relationships between covariates not found using conventional regression-based approaches.

METHOD

Using empirical data, we identify all statistically valid CTA propensity score models and then use them to compute strata-specific, observation-level propensity score weights that are subsequently applied in outcomes analyses. We compare findings obtained using this framework to the conventional method, by evaluating covariate balance and treatment effect estimates obtained using Cox regression and a weighted CTA outcomes model.

RESULTS

All models had some imbalanced covariates. Nevertheless, treatment effect estimates were generally consistent across all weighted models.

CONCLUSIONS

In the study sample, given that all approaches elicited similar results, using CTA increased confidence that bias could not be reduced any further. Because the CTA algorithm identifies all statistically valid propensity score models for a sample, it is most likely to identify a correctly specified propensity score model-and therefore should be used either to confirm results using traditional methods, or to reveal biases that may be missed by traditional methods. Moreover, given that the true treatment effect is never known in observational data, CTA should be considered for estimating outcomes because no statistical assumptions are required.

Minimizing imbalances on patient characteristics between treatment groups in randomized trials using classification tree analysis

RATIONALE, AIMS, AND OBJECTIVES

Randomization ensures that treatment groups do not differ systematically in their characteristics, thereby reducing threats to validity that may otherwise explain differences in outcomes. Large observed imbalances in patient characteristics may indicate that selection bias is being introduced into the treatment allocation process. We introduce classification tree analysis (CTA) as a novel algorithmic approach for identifying potential imbalances in characteristics and their interactions when provisionally assigning each new participant to one or the other treatment group. The participant is then permanently assigned to the treatment group that elicits either no or less imbalance than if assigned to the alternate group.

METHOD

Using data on participant characteristics from a clinical trial, we compare 3 different treatment allocation approaches: permuted block randomization (the original allocation method), minimization, and CTA. Treatment allocation performance is assessed by examining balance of all 17 patient characteristics between study groups for each of the allocation techniques.

RESULTS

While all 3 treatment allocation techniques achieved excellent balance on main effect variables, Classification tree analysis further identified imbalances on interactions and in the distributions of some of the continuous variables.

CONCLUSIONS

Classification tree analysis offers an algorithmic procedure that may be used with any randomization methodology to identify and then minimize linear, nonlinear, and interactive effects that induce covariate imbalance between groups. Investigators should consider using the CTA approach as a real-time complement to randomization for any clinical trial to safeguard the treatment allocation process against bias.

Modeling time-to-event (survival) data using classification tree analysis

RATIONALE, AIMS, AND OBJECTIVES

Time to the occurrence of an event is often studied in health research. Survival analysis differs from other designs in that follow-up times for individuals who do not experience the event by the end of the study (called censored) are accounted for in the analysis. Cox regression is the standard method for analysing censored data, but the assumptions required of these models are easily violated. In this paper, we introduce classification tree analysis (CTA) as a flexible alternative for modelling censored data. Classification tree analysis is a "decision-tree"-like classification model that provides parsimonious, transparent (ie, easy to visually display and interpret) decision rules that maximize predictive accuracy, derives exact P values via permutation tests, and evaluates model cross-generalizability.

METHOD

Using empirical data, we identify all statistically valid, reproducible, longitudinally consistent, and cross-generalizable CTA survival models and then compare their predictive accuracy to estimates derived via Cox regression and an unadjusted naïve model. Model performance is assessed using integrated Brier scores and a comparison between estimated survival curves.

RESULTS

The Cox regression model best predicts average incidence of the outcome over time, whereas CTA survival models best predict either relatively high, or low, incidence of the outcome over time.

CONCLUSIONS

Classification tree analysis survival models offer many advantages over Cox regression, such as explicit maximization of predictive accuracy, parsimony, statistical robustness, and transparency. Therefore, researchers interested in accurate prognoses and clear decision rules should consider developing models using the CTA-survival framework.

Using classification tree analysis to generate propensity score weights

RATIONALE, AIMS AND OBJECTIVES

In evaluating non-randomized interventions, propensity scores (PS) estimate the probability of assignment to the treatment group given observed characteristics. Machine learning algorithms have been proposed as an alternative to conventional logistic regression for modelling PS in order to avoid limitations of linear methods. We introduce classification tree analysis (CTA) to generate PS which is a "decision-tree"-like classification model that provides accurate, parsimonious decision rules that are easy to display and interpret, reports P values derived via permutation tests, and evaluates cross-generalizability.

METHOD

Using empirical data, we identify all statistically valid CTA PS models and then use them to compute strata-specific, observation-level PS weights that are subsequently applied in outcomes analyses. We compare findings obtained using this framework to logistic regression and boosted regression, by evaluating covariate balance using standardized differences, model predictive accuracy, and treatment effect estimates obtained using median regression and a weighted CTA outcomes model.

RESULTS

While all models had some imbalanced covariates, main-effects logistic regression yielded the lowest average standardized difference, whereas CTA yielded the greatest predictive accuracy. Nevertheless, treatment effect estimates were generally consistent across all models.

CONCLUSIONS

Assessing standardized differences in means as a test of covariate balance is inappropriate for machine learning algorithms that segment the sample into two or more strata. Because the CTA algorithm identifies all statistically valid PS models for a sample, it is most likely to identify a correctly specified PS model, and should be considered as an alternative approach to modeling the PS.

Melanoma complicating treatment with natalizumab for multiple sclerosis: A report from the Southern Network on Adverse Reactions (SONAR)

A 43-year-old female with multiple sclerosis developed urethral melanoma. The only potential risk factor was treatment with natalizumab, a humanized monoclonal antibody against α4 integrins. To investigate the risk-exposure relationship, we reviewed this case, all other published cases, and cases of natalizumab-associated melanoma reported to regulatory agencies. Data sources included the Food and Drug Administration's (FDA) Adverse Event Reporting System (FAERS) (2004-2014), a FDA Advisory Committee Meeting Report, and peer-reviewed publications. In the United States, the manufacturer maintains an FDA-mandated Tysabri Safety Surveillance Program (part of the Tysabri Outcomes Unified Commitment to Health (TOUCH)) of natalizumab-treated patients. We statistically compared reporting completeness for natalizumab-associated melanoma cases in FAERs for which information was obtained entirely from the TOUCH program versus cases where FAERS information was supplemented by TOUCH program information. FAERS included 137 natalizumab-associated melanoma reports in patients with multiple sclerosis. Median age at melanoma diagnosis was 45 years (range: 21-74 years). Changes in preexisting nevi occurred in 16%, history of cutaneous nevi occurred in 22%, diagnosis within 2 years of beginning natalizumab occurred in 34%, and 74% had primary surgical treatment. Among seven natalizumab-treated MS patients who developed biopsy-confirmed melanoma on treatment and reported in the literature, median age at diagnosis was 41 years (range: 38-48 years); and the melanoma diagnosis occurred following a median of 12 natalizumab doses (range: 1-77 doses). A history of mole or nevi was noted in four patients and a history of prior melanoma was noted in one patient. Completeness scores for reports were significantly lower for FAERS cases reported from the TOUCH program versus FAERS cases supplemented by TOUCH information (median score of 2 vs. 4 items out of 8-possible items, P < 0.0007). Clinicians should monitor existing nevi and maintain suspicion for melanoma developing in natalizumab-treated patients. The TOUCH Safety Surveillance Program, currently focused on progressive multifocal leukoencephalopathy, should be expanded to include information on other serious complications including malignancies, particularly if they are immunologic in nature.

Tale of Two Erythropoiesis-Stimulating Agents: Utilization, Dosing, Litigation, and Costs of Darbepoetin and Epoetin Among South Carolina Medicaid-Covered Patients With Cancer and Chemotherapy-Induced Anemia

PURPOSE

The US Food and Drug Administration (FDA) has approved epoetin and darbepoetin for chemotherapy-induced anemia (CIA). Approved epoetin and darbepoetin dosing schedules were three times per week and weekly, respectively, although off-label, less frequent scheduling was common. In 2004, 2007, and 2008, a US Food and Drug Administration Advisory Committees warned of risks associated with erythropoiesis-stimulating agents. During this period, lawsuits alleging illegal darbepoetin marketing practices have concluded, resulting in $1.1 billion in fines and settlements and one criminal conviction. No prior study, to our knowledge, has reported on the use of darbepoetin versus epoetin for CIA.

METHODS

We evaluated the dosing, utilization, and costs of erythropoiesis-stimulating agents among 3,761 South Carolina Medicaid patients with CIA.

RESULTS

Epoetin and darbepoetin utilization rates were 22% and 28% in 2003, 10% and 33% in 2007, and 3% and 7% in 2010, respectively. Mean per-patient per-administration epoetin and darbepoetin doses were 40,983 IU and 191 µg, respectively, in 2003 and 47,753 IU and 369 µg, respectively, in 2010. Mean monthly patient costs for epoetin and darbepoetin were $1,030 and $981, respectively, in 2003 and $932 and $1,352, respectively, in 2010. Epoetin use decreased steadily between 2002 and 2010; darbepoetin use increased steadily between 2003 and 2007 and then decreased steadily thereafter. Per-patient dosing of darbepoetin, but not epoetin, increased steadily between 2003 and 2010, and monthly per-patient epoetin costs decreased 3% while the per-patients costs of darbepoetin increased 30% between 2003 and 2010.

CONCLUSION

To our knowledge, our findings are the first data reporting on epoetin versus darbepoetin use for CIA and support recently concluded lawsuits involving allegations of illegal marketing practices of the manufacturer of darbepoetin.

Persistent erectile dysfunction in men exposed to the 5α-reductase inhibitors, finasteride, or dutasteride

Importance

Case reports describe persistent erectile dysfunction (PED) associated with exposure to 5α-reductase inhibitors (5α-RIs). Clinical trial reports and the manufacturers’ full prescribing information (FPI) for finasteride and dutasteride state that risk of sexual adverse effects is not increased by longer duration of 5α-RI exposure and that sexual adverse effects of 5α-RIs resolve in men who discontinue exposure.

Objective

Our chief objective was to assess whether longer duration of 5α-RI exposure increases risk of PED, independent of age and other known risk factors. Men with shorter 5α-RI exposure served as a comparison control group for those with longer exposure.

Design

We used a single-group study design and classification tree analysis (CTA) to model PED (lasting ≥90 days after stopping 5α-RI). Covariates included subject attributes, diseases, and drug exposures associated with sexual dysfunction.

Setting

Our data source was the electronic medical record data repository for Northwestern Medicine.

Subjects

The analysis cohorts comprised all men exposed to finasteride or dutasteride or combination products containing one of these drugs, and the subgroup of men 16–42 years old and exposed to finasteride ≤1.25 mg/day.

Main outcome and measures

Our main outcome measure was diagnosis of PED beginning after first 5α-RI exposure, continuing for at least 90 days after stopping 5α-RI, and with contemporaneous treatment with a phosphodiesterase-5 inhibitor (PDE₅I). Other outcome measures were erectile dysfunction (ED) and low libido. PED was determined by manual review of medical narratives for all subjects with ED. Risk of an adverse effect was expressed as number needed to harm (NNH).

Results

Among men with 5α-RI exposure, 167 of 11,909 (1.4%) developed PED (persistence median 1,348 days after stopping 5α-RI, interquartile range (IQR) 631.5–2320.5 days); the multivariable model predicting PED had four variables: prostate disease, duration of 5α-RI exposure, age, and nonsteroidal anti-inflammatory drug (NSAID) use. Of 530 men with new ED, 167 (31.5%) had new PED. Men without prostate disease who combined NSAID use with >208.5 days of 5α-RI exposure had 4.8-fold higher risk of PED than men with shorter exposure (NNH 59.8, all p < 0.002). Among men 16–42 years old and exposed to finasteride ≤1.25 mg/day, 34 of 4,284 (0.8%) developed PED (persistence median 1,534 days, IQR 651–2,351 days); the multivariable model predicting PED had one variable: duration of 5α-RI exposure. Of 103 young men with new ED, 34 (33%) had new PED. Young men with >205 days of finasteride exposure had 4.9-fold higher risk of PED (NNH 108.2, p < 0.004) than men with shorter exposure.

Conclusion and relevance

Risk of PED was higher in men with longer exposure to 5α-RIs. Among young men, longer exposure to finasteride posed a greater risk of PED than all other assessed risk factors.

Using machine learning to identify structural breaks in single-group interrupted time series designs

RATIONALE, AIMS AND OBJECTIVES

Single-group interrupted time series analysis (ITSA) is a popular evaluation methodology in which a single unit of observation is being studied, the outcome variable is serially ordered as a time series and the intervention is expected to 'interrupt' the level and/or trend of the time series, subsequent to its introduction. Given that the internal validity of the design rests on the premise that the interruption in the time series is associated with the introduction of the treatment, treatment effects may seem less plausible if a parallel trend already exists in the time series prior to the actual intervention. Thus, sensitivity analyses should focus on detecting structural breaks in the time series before the intervention.

METHOD

In this paper, we introduce a machine-learning algorithm called optimal discriminant analysis (ODA) as an approach to determine if structural breaks can be identified in years prior to the initiation of the intervention, using data from California's 1988 voter-initiated Proposition 99 to reduce smoking rates.

RESULTS

The ODA analysis indicates that numerous structural breaks occurred prior to the actual initiation of Proposition 99 in 1989, including perfect structural breaks in 1983 and 1985, thereby casting doubt on the validity of treatment effects estimated for the actual intervention when using a single-group ITSA design.

CONCLUSIONS

Given the widespread use of ITSA for evaluating observational data and the increasing use of machine-learning techniques in traditional research, we recommend that structural break sensitivity analysis is routinely incorporated in all research using the single-group ITSA design.

Using machine learning to model dose-response relationships

RATIONALE, AIMS AND OBJECTIVES

Establishing the relationship between various doses of an exposure and a response variable is integral to many studies in health care. Linear parametric models, widely used for estimating dose-response relationships, have several limitations. This paper employs the optimal discriminant analysis (ODA) machine-learning algorithm to determine the degree to which exposure dose can be distinguished based on the distribution of the response variable. By framing the dose-response relationship as a classification problem, machine learning can provide the same functionality as conventional models, but can additionally make individual-level predictions, which may be helpful in practical applications like establishing responsiveness to prescribed drug regimens.

METHOD

Using data from a study measuring the responses of blood flow in the forearm to the intra-arterial administration of isoproterenol (separately for 9 black and 13 white men, and pooled), we compare the results estimated from a generalized estimating equations (GEE) model with those estimated using ODA.

RESULTS

Generalized estimating equations and ODA both identified many statistically significant dose-response relationships, separately by race and for pooled data. Post hoc comparisons between doses indicated ODA (based on exact P values) was consistently more conservative than GEE (based on estimated P values). Compared with ODA, GEE produced twice as many instances of paradoxical confounding (findings from analysis of pooled data that are inconsistent with findings from analyses stratified by race).

CONCLUSIONS

Given its unique advantages and greater analytic flexibility, maximum-accuracy machine-learning methods like ODA should be considered as the primary analytic approach in dose-response applications.

Using machine learning to assess covariate balance in matching studies

In order to assess the effectiveness of matching approaches in observational studies, investigators typically present summary statistics for each observed pre-intervention covariate, with the objective of showing that matching reduces the difference in means (or proportions) between groups to as close to zero as possible. In this paper, we introduce a new approach to distinguish between study groups based on their distributions of the covariates using a machine-learning algorithm called optimal discriminant analysis (ODA). Assessing covariate balance using ODA as compared with the conventional method has several key advantages: the ability to ascertain how individuals self-select based on optimal (maximum-accuracy) cut-points on the covariates; the application to any variable metric and number of groups; its insensitivity to skewed data or outliers; and the use of accuracy measures that can be widely applied to all analyses. Moreover, ODA accepts analytic weights, thereby extending the assessment of covariate balance to any study design where weights are used for covariate adjustment. By comparing the two approaches using empirical data, we are able to demonstrate that using measures of classification accuracy as balance diagnostics produces highly consistent results to those obtained via the conventional approach (in our matched-pairs example, ODA revealed a weak statistically significant relationship not detected by the conventional approach). Thus, investigators should consider ODA as a robust complement, or perhaps alternative, to the conventional approach for assessing covariate balance in matching studies.

Combining machine learning and propensity score weighting to estimate causal effects in multivalued treatments

RATIONALE, AIMS AND OBJECTIVES

Interventions with multivalued treatments are common in medical and health research; examples include comparing the efficacy of competing interventions and contrasting various doses of a drug. In recent years, there has been growing interest in the development of methods that estimate multivalued treatment effects using observational data. This paper extends a previously described analytic framework for evaluating binary treatments to studies involving multivalued treatments utilizing a machine learning algorithm called optimal discriminant analysis (ODA).

METHOD

We describe the differences between regression-based treatment effect estimators and effects estimated using the ODA framework. We then present an empirical example using data from an intervention including three study groups to compare corresponding effects.

RESULTS

The regression-based estimators produced statistically significant mean differences between the two intervention groups, and between one of the treatment groups and controls. In contrast, ODA was unable to discriminate between distributions of any of the three study groups.

CONCLUSIONS

Optimal discriminant analysis offers an appealing alternative to conventional regression-based models for estimating effects in multivalued treatment studies because of its insensitivity to skewed data and use of accuracy measures applicable to all prognostic analyses. If these analytic approaches produce consistent treatment effect P values, this bolsters confidence in the validity of the results. If the approaches produce conflicting treatment effect P values, as they do in our empirical example, the investigator should consider the ODA-derived estimates to be most robust, given that ODA uses permutation P values that require no distributional assumptions and are thus, always valid.

Combining machine learning and matching techniques to improve causal inference in program evaluation

RATIONALE, AIMS AND OBJECTIVES

Program evaluations often utilize various matching approaches to emulate the randomization process for group assignment in experimental studies. Typically, the matching strategy is implemented, and then covariate balance is assessed before estimating treatment effects. This paper introduces a novel analytic framework utilizing a machine learning algorithm called optimal discriminant analysis (ODA) for assessing covariate balance and estimating treatment effects, once the matching strategy has been implemented. This framework holds several key advantages over the conventional approach: application to any variable metric and number of groups; insensitivity to skewed data or outliers; and use of accuracy measures applicable to all prognostic analyses. Moreover, ODA accepts analytic weights, thereby extending the methodology to any study design where weights are used for covariate adjustment or more precise (differential) outcome measurement.

METHOD

One-to-one matching on the propensity score was used as the matching strategy. Covariate balance was assessed using standardized difference in means (conventional approach) and measures of classification accuracy (ODA). Treatment effects were estimated using ordinary least squares regression and ODA.

RESULTS

Using empirical data, ODA produced results highly consistent with those obtained via the conventional methodology for assessing covariate balance and estimating treatment effects.

CONCLUSIONS

When ODA is combined with matching techniques within a treatment effects framework, the results are consistent with conventional approaches. However, given that it provides additional dimensions and robustness to the analysis versus what can currently be achieved using conventional approaches, ODA offers an appealing alternative.

Using data mining techniques to characterize participation in observational studies

Data mining techniques are gaining in popularity among health researchers for an array of purposes, such as improving diagnostic accuracy, identifying high-risk patients and extracting concepts from unstructured data. In this paper, we describe how these techniques can be applied to another area in the health research domain: identifying characteristics of individuals who do and do not choose to participate in observational studies. In contrast to randomized studies where individuals have no control over their treatment assignment, participants in observational studies self-select into the treatment arm and therefore have the potential to differ in their characteristics from those who elect not to participate. These differences may explain part, or all, of the difference in the observed outcome, making it crucial to assess whether there is differential participation based on observed characteristics. As compared to traditional approaches to this assessment, data mining offers a more precise understanding of these differences. To describe and illustrate the application of data mining in this domain, we use data from a primary care-based medical home pilot programme and compare the performance of commonly used classification approaches - logistic regression, support vector machines, random forests and classification tree analysis (CTA) - in correctly classifying participants and non-participants. We find that CTA is substantially more accurate than the other models. Moreover, unlike the other models, CTA offers transparency in its computational approach, ease of interpretation via the decision rules produced and provides statistical results familiar to health researchers. Beyond their application to research, data mining techniques could help administrators to identify new candidates for participation who may most benefit from the intervention.

Note on the Multidisciplinary Scope of Psychological Androgyny Theory

Psychological androgyny theory posits two independent behavioral domains: instrumental behavior reflects a problem-solving orientation and expressive behavior reflects an affective concern for the welfare of others. The hypothesis that these dimensions resemble others studied in literatures on leadership and conflict resolution was tested. A factor analysis of six scores, two from each of the three literatures, for a sample of 47 undergraduate men offered empirical support for the similarity hypothesis. Theoretical implications of the finding are discussed in terms of directions for further research.

Social Conformity and the Type a Behavior Pattern

The relation between the Type A behavior pattern and social conformity was examined on a perceptual skill task. 23 extreme Type A and B persons participated in an experiment based on the Asch conformity paradigm. Type B subjects conformed twice as often as Type A persons ( p < .02). The same task administered to a subset of the sample and a larger normative sample indicated that social pressure to conform was responsible for differences in judgmental accuracy of stimuli (the conformity measure). The results suggest that Type A persons' resistance to subtle social coercion to conform may reflect their increased sensitivity to threats to their personal control. The findings are discussed in the context of reactance and control theories of Type A behavior.

Risk Factors and Outcomes Associated With Non–Enterococcus faecalis, Non–Enterococcus faeciumEnterococcal Bacteremia

Objectives.

To compare risk factors, clinical features, and outcomes in patients withEnterococcus avium,Enterococcus casseliflavus,Enterococcus durans,Enterococcus gallinarum, andEnterococcus mundtiibacteremia (cases) with those in patients withEnterococcus faecalisbacteremia (controls).

Design.

A retrospective case-control study.

Setting.

A 725-bed, university-affiliated, academic medical center.

Patients.

The clinical microbiology database at Northwestern Memorial Hospital from January 1994 to May 2003 was searched to identify cases; each case was matched to one control on the basis of date of admission.

Results.

Thirty-three cases were identified and matched with 33 controls. The mean duration of hospital stay was longer (29.7 vs 17.2 days;P= .03) and the mean time to acquisition of bacteremia was greater (16.5 vs 6.3 days;P= .003) for cases than controls. Cases were more likely to have underlying hematologic malignancies (P< .001), to have been treated with corticosteroids (P= .02), and to be neutropenic (P= .003). Controls were more likely to have an indwelling bladder catheter (P= .01), and cases were more likely to have the gastrointestinal tract as a source of infection (P= .007) and to have concurrent cholangitis (P= .002). There were no differences in severity of illness or in mortality rates.

Conclusions.

Compared with patients withE. faecalisbacteremia, patients with non-E. faecalis, non-Enterococcus faeciumenterococcal bacteremia were more likely to have a hematologic malignancy, prior treatment with corticosteroids, neutropenia, and cholangitis; longer duration of hospital stay was also identified as a clinical feature. However, non-E. faecalis, non-E. faeciumspecies are not associated with any differences in mortality.

Adverse Event Reporting in Clinical Trials of Finasteride for Androgenic Alopecia: A Meta-analysis

IMPORTANCE

Two meta-analyses conclude that finasteride treatment of androgenic alopecia (AGA) is safe but do not assess quality of safety reporting.

OBJECTIVE

To assess safety reporting for clinical trial reports of finasteride for AGA.

DATA SOURCES

MEDLINE, ClinicalTrials.gov, and a clinical data repository for an academic medical center.

STUDY SELECTION

Published clinical trial reports for finasteride treatment of AGA.

DATA EXTRACTION AND SYNTHESIS

For each trial, we assessed quality of adverse event reporting, extracted the number and type of adverse events in treatment and placebo groups, and assessed duration of safety evaluation and adequacy of blinding. Two observers independently extracted the data; differences were resolved by consensus. We assessed generalizability in a large cohort of men prescribed finasteride, 1.25 mg/d or less, by assessing for eligibility in the finasteride-AGA pivotal trials.

MAIN OUTCOMES AND MEASURES

Quality was assessed as adequate, partially adequate, inadequate, or no events reported. We used funnel plots of the hazard ratio to assess bias.

RESULTS

Of 34 clinical trials, none had adequate safety reporting, 19 were partially adequate, 12 were inadequate, and 3 reported no adverse events. Funnel plots were asymmetric with a bias toward lower odds ratio for sexual adverse effects, suggesting systematic underdetection. No reports assessed adequacy of blinding, 18 (53%) disclosed conflicts of interest, and 19 (56%) received funding from the manufacturer. Duration of drug safety evaluation was 1 year or less for 26 of 34 trials (76%). Of 5704 men in the clinical data repository who were treated for AGA with finasteride, 1.25 mg/d or less, for AGA, only 31% met inclusion criteria for the pivotal trials referenced in the manufacturer's full prescribing information and 33% took finasteride for more than 1 year.

CONCLUSIONS AND RELEVANCE

Available toxicity information from clinical trials of finasteride in men with AGA is very limited, is of poor quality, and seems to be systematically biased. In a cohort of men prescribed finasteride for routine treatment of AGA, most would have been excluded from the pivotal studies that supported US Food and Drug Administration approval for AGA. Published reports of clinical trials provide insufficient information to establish the safety profile for finasteride in the treatment of AGA.

Cisplatin: process and future

One of the most important anticancer agents is cisplatin (CDDP). Numerous studies with a CDDP-based combination have been reported over the last 30 years. The use of CDDP in the 1980s and 1990s showed responses in advanced stage non-small-cell lung cancer (NSCLC). Over the years it was found that the side effects of this agent (particularly nephrotoxicity) were a common problem. Agents such as carboplatin, taxanes, gemcitabine, irinotecan and pemetrexed proved to be effective in NSCLC with reduced or no nephrotoxicity. The administration of these newer agents improved several side effects, but without improving efficacy. When prophylactic (adjuvant) treatment for NSCLC was introduced, CDDP was the agent selected, which indicated the value of the drug. Recently, a novel formulation of CDDP, liposomal cisplatin, which has shown very low toxicity, no nephrotoxicity and equal effectiveness was produced; its importance is its higher effectiveness than standard CDDP in lung adenocarcinoma.

Association of breast cancer in men with exposure to 5-α reductase inhibitors: A RADAR report

2532

Background: Breast cancers in men (BCM) account for <1% of all breast cancers. Dihydrotestosterone (DHT) inhibits proliferation of normal and neoplastic mammary tissue and constrains the effect of estrogens. Finasteride (F) and dutasteride (D) are 5-α reductase inhibitors (5-αRIs) that reduce systemic and local dihydrotestosterone and cause gynecomastia in 1–3% of men. The package inserts for F and D state, “the relationship between long-term use of (finasteride/dutasteride) and male breast neoplasia is currently unknown.” F and D are marketed for treatment of symptomatic benign-prostatic hyperplasia. F is marketed for treatment of androgenetic alopecia. Methods: To detect disproportionality in the FDA MedWatch dataset, we calculated the empiric Bayes geometric mean (EBGM) for association of BCM with F or D. We also calculated the attributable risk of BCM exposed to F or D among men at an urban academic hospital (Northwestern Memorial Hospital) and at a rural healthcare system (Marshfield Clinic). Results: In the MedWatch dataset, we identified 33 reports of F-associated BCM and 5 reports of D-associated BCM. For F–associated BCM, the EBGM was 58.95 (95% CI 24.47-81.76; p=0.0001). For D-associated BCM, the EBGM was 15.79 (95% CI 4.57-35.49; p=0.0001). The mean age for BCM after 5-αRI exposure was 70±11 years; 11/38 (29%) had gynecomastia. There were 38 cases of BCM associated with 5-αRI in the combined Northwestern and Marshfield cohort (see table below). Conclusions: We found a highly significant association between BCM and 5-αRI exposure in each of 6 separate analyses (3 sources X 2 drugs), with an estimated 1 extra BCM per 564 men exposed to 5-αRIs. We now plan to assess BRCA status and other risk factors. Given that 5-αRIs are marketed for control of lower urinary tract symptoms or for cosmetic purposes, it is not immediately obvious that use of finasteride or dutasteride for their labeled indications would provide any net benefit. [Table: see text]

Quality of methods for assessing and reporting serious adverse events in clinical trials of cancer drugs

The validity of information regarding drug toxicity in humans depends on the quality of the methods and instruments used to assess adverse drug events (ADEs). This study evaluates the quality of instruments used to assess and report ADEs to institutional review boards (IRBs) at US cancer centers. Forms from all 49 National Cancer Institute (NCI)-designated centers were assessed for utility in abstracting event type, severity, and causality; patient demographics; safety monitoring; and consequent changes in the conduct of the relevant study. Of the 55 items considered essential for ADE reporting, one item (event description) was present on all the forms. Seventy-eight percent of the instruments prompted for global introspection of the investigator, a method known to be unreliable. Of the 34 items that our panel of experts considered essential for event description, the median number of items present was four (domain = 1-11). The use of a validated tool to describe and assess event type, severity, and causality may lead to more timely, accurate identification of safety signals in cancer treatment.

Evaluation of a potential clinical interaction between ceftriaxone and calcium

In April 2009, the FDA retracted a warning asserting that ceftriaxone and intravenous calcium products should not be coadministered to any patient to prevent precipitation events leading to end-organ damage. Following that announcement, we sought to evaluate if the retraction was justified. A search of the FDA Adverse Event Reporting System was conducted to identify any ceftriaxone-calcium interactions that resulted in serious adverse drug events. Ceftazidime-calcium was used as a comparator agent. One hundred four events with ceftriaxone-calcium and 99 events with ceftazidime-calcium were identified. Adverse drug events were recorded according to the listed description of drug involvement (primary or secondary suspect) and were interpreted as probable, possible, unlikely, or unrelated. For ceftriaxone-calcium-related adverse events, 7.7% and 20.2% of the events were classified as probable and possible for embolism, respectively. Ceftazidime-calcium resulted in fewer probable embolic events (4%) but more possible embolic events (30.3%). Among cases that considered ceftriaxone or ceftazidime and calcium as the primary or secondary drug, one case was classified as a probable embolic event. That patient received ceftriaxone-calcium and died, although an attribution of causality was not possible. Our analysis suggests a lack of support for the occurrence of ceftriaxone-calcium precipitation events in adults. The results of the current analysis reinforce the revised FDA recommendations suggesting that patients >28 days old may receive ceftriaxone and calcium sequentially and provide a transparent and reproducible methodology for such evaluations.

Racial variations in care and outcomes for inpatient HIV-related pneumocystis pneumonia

Racial disparities in HIV-care include the disproportionate impact of HIV/AIDS on African Americans. We conducted a retrospective review of 1,855 cases at 78 hospitals in nine cities to evaluate racial variations in inpatient care for AIDS-related Pneumocystis pneumonia (PCP) shortly after the introduction of highly active anti-retroviral therapies. While inpatient HIV-related PCP mortality was comparable between Whites and Hispanics (p=0.94), African Americans were less likely than Whites to die in-hospital (AOR 0.69, 95% CI 0.48, 0.99) and more likely to receive timely anti-PCP medications (AOR 1.67, 95% CI 1.21, 2.30) and timely corticosteroids (AOR 1.46, 95% CI 1.17, 1.82). Findings were compared with those from our study involving 1,547 patients at 82 hospitals in five cities over the first decade of the AIDS epidemic. In contrast to the first study, in the second decade African Americans were more likely to receive timely and appropriate therapy for HIV-related PCP, and resultantly were more likely to survive the hospitalization.

Ticlopidine- and clopidogrel-associated thrombotic thrombocytopenic purpura (TTP): review of clinical, laboratory, epidemiological, and pharmacovigilance findings (1989-2008)

Thrombotic thrombocytopenic purpura (TTP) is a fulminant disease characterized by platelet aggregates, thrombocytopenia, renal insufficiency, neurologic changes, and mechanical injury to erythrocytes. Most idiopathic cases of TTP are characterized by a deficiency of ADAMTS13 (a disintegrin and metalloprotease, with thrombospondin-1-like domains) metalloprotease activity. Ironically, use of anti-platelet agents, the thienopyridine derivates clopidogrel and ticlopidine, is associated with drug induced TTP. Data were abstracted from a systematic review of English-language literature for thienopyridine-associated TTP identified in MEDLINE, EMBASE, the public website of the Food and Drug Administration, and abstracts from national scientific conferences from 1991 to April 2008. Ticlopidine and clopidogrel are the two most common drugs associated with TTP in FDA safety databases. Epidemiological studies identify recent initiation of anti-platelet agents as the most common risk factor associated with risks of developing TTP. Laboratory studies indicate that most cases of thienopyridine-associated TTP involve an antibody to ADAMTS13 metalloprotease, present with severe thrombocytopenia, and respond to therapeutic plasma exchange (TPE); a minority of thienopyridine-associated TTP presents with severe renal insufficiency, involves direct endothelial cell damage, and is less responsive to TPE. The evaluation of this potentially fatal drug toxicity can serve as a template for future efforts to comprehensively characterize other severe adverse drug reactions.

Hepatotoxicity associated with long- versus short-course HIV-prophylactic nevirapine use: a systematic review and meta-analysis from the Research on Adverse Drug events And Reports (RADAR) project

BACKGROUND AND OBJECTIVE

The antiretroviral nevirapine can cause severe hepatotoxicity when used 'off-label' for preventing mother-to-child HIV transmission (PMTCT), newborn post-exposure prophylaxis and for pre- and post-exposure prophylaxis among non-HIV-infected individuals. We describe the incidence of hepatotoxicity with short- versus long-course nevirapine-containing regimens in these groups.

METHODS

We reviewed hepatotoxicity cases among non-HIV-infected individuals and HIV-infected pregnant women and their offspring receiving short- (or=5 days) nevirapine prophylaxis. Sources included adverse event reports from pharmaceutical manufacturers and the US FDA, reports from peer-reviewed journals/scientific meetings and the Research on Adverse Drug events And Reports (RADAR) project. Hepatotoxicity was scored using the AIDS Clinical Trial Group criteria.

RESULTS

Toxicity data for 8216 patients treated with nevirapine-containing regimens were reviewed. Among 402 non-HIV-infected individuals receiving short- (n=251) or long-course (n=151) nevirapine, rates of grade 1-2 hepatotoxicity were 1.99% versus 5.30%, respectively, and rates of grade 3-4 hepatotoxicity were 0.00% versus 13.25%, respectively (p<0.001 for both comparisons). Among 4740 HIV-infected pregnant women receiving short- (n=3031) versus long-course (n=1709) nevirapine, rates of grade 1-2 hepatotoxicity were 0.62% and 7.04%, respectively, and rates of grade 3-4 hepatotoxicity were 0.23% versus 4.39%, respectively (p<0.001 for both comparisons). The rates of grade 3-4 hepatotoxicity among 3074 neonates of nevirapine-exposed HIV-infected pregnant women were 0.8% for those receiving short-course (n=2801) versus 1.1% for those receiving long-course (n=273) therapy (p<0.72).

CONCLUSIONS

Therapy duration appears to significantly predict nevirapine hepatotoxicity. Short-course nevirapine for HIV prophylaxis is associated with fewer hepatotoxic reactions for non-HIV-infected individuals or pregnant HIV-infected women and their offspring, but administration of prophylactic nevirapine for >or=2 weeks appears to be associated with high rates of hepatotoxicity among non-HIV-infected individuals and HIV-infected pregnant mothers. When full highly active antiretroviral therapy (HAART) regimens are not available, single-dose nevirapine plus short-course nucleoside reverse transcriptase inhibitors to decrease the development of HIV viral resistance is an essential therapeutic option for PMTCT and these data support the safety of single-dose nevirapine in this setting.

Predicting outcome after emergent cerclage using classification tree analysis

We sought to develop a predictive model for gestational age at delivery after placement of an emergent cerclage in the second trimester. Data were obtained for women undergoing emergent cerclage in response to documented cervical change on physical examination at a university hospital between 1980 and 2000. Hierarchically optimal classification tree analysis (CTA) was used to predict delivery prior to 24 weeks, between 24 and 27 6/7 weeks, or after 27 6/7 weeks. One hundred sixteen women were available for analysis. Delivery prior to 24 weeks was best predicted by presence of prolapsed membranes and gestational age at cerclage placement; delivery between 24 and 27 6/7 weeks was best predicted by parity alone; delivery of at least 28 weeks was best predicted by cervical dilation and length, presence of prolapsed membranes, and parity. When choosing a single model to predict delivery at the three different gestational age periods, the last model yielded the most accurate results. CTA can be used to construct a predictive model for outcome after emergent cerclage that may be informative for both patients and physicians.

Venous thromboembolism and mortality associated with recombinant erythropoietin and darbepoetin administration for the treatment of cancer-associated anemia

CONTEXT

The erythropoiesis-stimulating agents (ESAs) erythropoietin and darbepoetin are licensed to treat chemotherapy-associated anemia in patients with nonmyeloid malignancies. Although systematic overviews of trials have identified venous thromboembolism (VTE) risks, none have identified mortality risks with ESAs.

OBJECTIVE

To evaluate VTE and mortality rates associated with ESA administration for the treatment of anemia among patients with cancer.

DATA SOURCES

A published overview from the Cochrane Collaboration (search dates: January 1, 1985-April 1, 2005) and MEDLINE and EMBASE databases (key words: clinical trial, erythropoietin, darbepoetin, and oncology), the public Web site of the US Food and Drug Administration and ESA manufacturers, and safety advisories (search dates: April 1, 2005-January 17, 2008).

STUDY SELECTION

Phase 3 trials comparing ESAs with placebo or standard of care for the treatment of anemia among patients with cancer.

DATA EXTRACTION

Mortality rates, VTE rates, and 95% confidence intervals (CIs) were extracted by 3 reviewers from 51 clinical trials with 13 611 patients that included survival information and 38 clinical trials with 8172 patients that included information on VTE.

DATA SYNTHESIS

Patients with cancer who received ESAs had increased VTE risks (334 VTE events among 4610 patients treated with ESA vs 173 VTE events among 3562 control patients; 7.5% vs 4.9%; relative risk, 1.57; 95% CI, 1.31-1.87) and increased mortality risks (hazard ratio, 1.10; 95% CI, 1.01-1.20).

CONCLUSIONS

Erythropoiesis-stimulating agent administration to patients with cancer is associated with increased risks of VTE and mortality. Our findings, in conjunction with basic science studies on erythropoietin and erythropoietin receptors in solid cancers, raise concern about the safety of ESA administration to patients with cancer.

Derivation of a triage algorithm for chest radiography of community-acquired pneumonia patients in the emergency department

BACKGROUND

Community-acquired pneumonia (CAP) accounts for 1.5 million emergency department (ED) patient visits in the United States each year.

OBJECTIVES

To derive an algorithm for the ED triage setting that facilitates rapid and accurate ordering of chest radiography (CXR) for CAP.

METHODS

The authors conducted an ED-based retrospective matched case-control study using 100 radiographic confirmed CAP cases and 100 radiographic confirmed influenzalike illness (ILI) controls. Sensitivities and specificities of characteristics assessed in the triage setting were measured to discriminate CAP from ILI. The authors then used classification tree analysis to derive an algorithm that maximizes sensitivity and specificity for detecting patients with CAP in the ED triage setting.

RESULTS

Temperature greater than 100.4 degrees F (likelihood ratio = 4.39, 95% confidence interval [CI] = 2.04 to 9.45), heart rate greater than 110 beats/minute (likelihood ratio = 3.59, 95% CI = 1.82 to 7.10), and pulse oximetry less than 96% (likelihood ratio = 2.36, 95% CI = 1.32 to 4.20) were the strongest predictors of CAP. However, no single characteristic was adequately sensitive and specific to accurately discriminate CAP from ILI. A three-step algorithm (using optimum cut points for elevated temperature, tachycardia, and hypoxemia on room air pulse oximetry) was derived that is 70.8% sensitive (95% CI = 60.7% to 79.7%) and 79.1% specific (95% CI = 69.3% to 86.9%).

CONCLUSIONS

No single characteristic adequately discriminates CAP from ILI, but a derived clinical algorithm may detect most radiographic confirmed CAP patients in the triage setting. Prospective assessment of this algorithm will be needed to determine its effects on the care of ED patients with suspected pneumonia.

Cancer-associated neutropenic fever: clinical outcome and economic costs of emergency department care

Purpose. Febrile neutropenia (FN) is a common, costly, and potentially fatal complication in oncology. While FN in the inpatient setting has been extensively studied, only one study has evaluated emergency department (ED) care for FN cancer patients. That study found that 96% of patients survived the complication. We evaluated clinical and economic outcomes for cancer patients with chemotherapy-associated FN treated in an ED. Methods. ED records for consecutive oncology patients with FN were reviewed for information on death, intensive care unit (ICU) use, blood cultures, and costs. Results. Forty-eight patients (n = 57 visits) were evaluated. Six patients died from FN (12%) and four received ICU care within 2 weeks and survived (8%). Blood cultures were positive for 37% of the ED visits. The median ED time was 3.3 hours. In 91% of visits, i.v. antibiotics were administered in the ED, ordered at a median of 1.7 hours from triage (interquartile range [IQR], 1.2-2.8 hours). All patients with death or ICU in 2 weeks and all but one patient with positive blood cultures received antibiotics. The median per patient ED costs were $1,455 (IQR, $1,300-$1,579)-42.4% for hospital/nursing, 23.5% for radiology, 20.8% for physician services, 10.9% for diagnostic tests, and 2.4% for antibiotics. Conclusions. Cancer patients with FN in this sample presenting to the ED frequently had no identified source of infection. One third of the patients had positive ED blood cultures and one fifth died or required ICU care within 2 weeks. Costs of ED care were similar to the cost of a single day of inpatient care. Disclosure of potential conflicts of interest is found at the end of this article.

Developing Indicators of Inpatient Adverse Drug Events Through Nonlinear Analysis Using Administrative Data

BACKGROUND

Because of uniform availability, hospital administrative data are appealing for surveillance of adverse drug events (ADEs). Expert-generated surveillance rules that rely on the presence of International Classification of Diseases, 9th Revision Clinical Modification (ICD-9-CM) codes have limited accuracy. Rules based on nonlinear associations among all types of available administrative data may be more accurate.

OBJECTIVES

By applying hierarchically optimal classification tree analysis (HOCTA) to administrative data, derive and validate surveillance rules for bleeding/anticoagulation problems and delirium/psychosis.

RESEARCH DESIGN

Retrospective cohort design.

SUBJECTS

A random sample of 3987 admissions drawn from all 41 Utah acute-care hospitals in 2001 and 2003.

MEASURES

Professional nurse reviewers identified ADEs using implicit chart review. Pharmacists assigned Medical Dictionary for Regulatory Activities codes to ADE descriptions for identification of clinical groups of events. Hospitals provided patient demographic, admission, and ICD9-CM data.

RESULTS

Incidence proportions were 0.8% for drug-induced bleeding/anticoagulation problems and 1.0% for drug-induced delirium/psychosis. The model for bleeding had very good discrimination and sensitivity at 0.87 and 86% and fair positive predictive value (PPV) at 12%. The model for delirium had excellent sensitivity at 94%, good discrimination at 0.83, but low PPV at 3%. Poisoning and adverse event codes designed for the targeted ADEs had low sensitivities and, when forced in, degraded model accuracy.

CONCLUSIONS

Hierarchically optimal classification tree analysis is a promising method for rapidly developing clinically meaningful surveillance rules for administrative data. The resultant model for drug-induced bleeding and anticoagulation problems may be useful for retrospective ADE screening and rate estimation.

Two mechanistic pathways for thienopyridine-associated thrombotic thrombocytopenic purpura: a report from the SERF-TTP Research Group and the RADAR Project

OBJECTIVES

We sought to describe clinical and laboratory findings for a large cohort of patients with thienopyridine-associated thrombotic thrombocytopenic purpura (TTP).

BACKGROUND

The thienopyridine derivatives, ticlopidine and clopidogrel, are the 2 most common drugs associated with TTP in databases maintained by the U.S. Food and Drug Administration (FDA).

METHODS

Clinical reports of TTP associated with clopidogrel and ticlopidine were identified from medical records, published case reports, and FDA case reports (n = 128). Duration of thienopyridine exposure, clinical and laboratory findings, and survival were recorded. ADAMTS13 activity (n = 39) and inhibitor (n = 30) were measured for a subset of individuals.

RESULTS

Compared with clopidogrel-associated TTP cases (n = 35), ticlopidine-associated TTP cases (n = 93) were more likely to have received more than 2 weeks of drug (90% vs. 26%), to be severely thrombocytopenic (84% vs. 60%), and to have normal renal function (72% vs. 45%) (p < 0.01 for each). Compared with TTP patients with ADAMTS13 activity >15% (n = 13), TTP patients with severely deficient ADAMTS13 activity (n = 26) were more likely to have received ticlopidine (92.3% vs. 46.2%, p < 0.003). Among patients who developed TTP >2 weeks after thienopyridine, therapeutic plasma exchange (TPE) increased likelihood of survival (84% vs. 38%, p < 0.05). Among patients who developed TTP within 2 weeks of starting thienopyridines, survival was 77% with TPE and 78% without.

CONCLUSIONS

Thrombotic thrombocytopenic purpura is a rare complication of thienopyridine treatment. This drug toxicity appears to occur by 2 different mechanistic pathways, characterized primarily by time of onset before versus after 2 weeks of thienopyridine administration. If TTP occurs after 2 weeks of ticlopidine or clopidogrel therapy, therapeutic plasma exchange must be promptly instituted to enhance likelihood of survival.