Different analyses estimate different parameters of the effect of erythropoietin stimulating agents on survival in end stage renal disease: a comparison of payment policy analysis, instrumental variables, and multiple imputation of potential outcomes

Does Knowledge of Treatment Assignment Affect Patient Report of Symptoms, Function, and Health Status? An Evaluation Using Multiple Myeloma Trials

OBJECTIVES

Unblinded trials are common in oncology, but patient knowledge of treatment assignment may bias response to questionnaires. We sought to ascertain the extent of possible bias arising from patient knowledge of treatment assignment.

METHODS

This is a retrospective analysis of data from 2 randomized trials in multiple myeloma, 1 double-blind and 1 open label. We compared changes in patient reports of symptoms, function, and health status from prerandomization (screening) to baseline (pretreatment but postrandomization) across control and investigational arms in the 2 trials. Changes from prerandomization scores at ~2 and 6 months on treatment were evaluated only across control arms to avoid comparisons between 2 different experimental drugs. All scores were on 0- to 100-point scales. Inverse probability weighting, entropy balancing, and multiple imputation using propensity score splines were used to compare score changes across similar groups of patients.

RESULTS

Minimal changes from screening were seen at baseline in all arms. In the control arm, mean changes of <7 points were seen for all domains at 2 and 6 months. The effect of unblinding at 6 months in social function was a decline of less than 6 points (weighting: -3.09; 95% confidence interval -8.41 to 2.23; balancing: -4.55; 95% confidence interval -9.86 to 0.76; imputation: -5.34; 95% confidence interval -10.64 to -0.04).

CONCLUSION

In this analysis, we did not find evidence to suggest that there was a meaningful differential effect on how patients reported their symptoms, function or health status after knowing their treatment assignment.

Medical foster home is less costly than traditional nursing home care

OBJECTIVE

To compare the costs of Community Nursing Homes (CNHs) to Medical Foster Homes (MFHs) at Veteran Health Administration (VHA) Medical Centers that established MFH programs.

DATA SOURCES

Episode and costs data were derived from VA and Medicare files (inpatient, outpatient, emergency room, skilled nursing facility, dialysis, and hospice).

STUDY DESIGN

Propensity scores matched 354 MFH to 1693 CNH Veterans on demographics, clinical characteristics, health care utilization, and costs.

DATA EXTRACTION METHODS

Data were retrieved for years 2010-2011 from the VA Corporate Data Warehouse, VA Health Data Repository, and the VA MFH Program through the VA Informatics and Computing Infrastructure (VINCI).

PRINCIPAL FINDINGS

After matching on unique characteristics of MFH Veterans, costs were $71.28 less per day alive compared to CNH care. Home-based and mental health care costs increased with savings largely attributable to avoiding CNH residential care. When average out-of-pocket payments by Veterans of $74/day are considered, MFH is at least cost neutral. Mortality was 12 percent higher among matched Veterans in CNHs.

CONCLUSIONS

MFHs may serve as alternatives to traditional CNH care that do not increase total costs with mortality benefits. Future work should examine the differences for functional disability subgroups.

New Kidney Allocation System Associated With Increased Rates Of Transplants Among Black And Hispanic Patients

Before the 2014 implementation of a new kidney allocation system by the United Network for Organ Sharing, white patients were more likely than black or Hispanic patients to receive a kidney transplant. To determine the effect of the new allocation system on these disparities, we examined data for 179,071 transplant waiting list events in the period June 2013-September 2016, and we calculated monthly transplantation rates (34,133 patients actually received transplants). Implementation of the new system was associated with a narrowing of the disparities in the average monthly transplantation rates by 0.29 percentage point for blacks compared to whites and by 0.24 percentage point for Hispanics compared to whites, which resulted in both disparities becoming nonsignificant after implementation of the new system.

Persistent pain after motor vehicle collision: comparative effectiveness of opioids vs nonsteroidal antiinflammatory drugs prescribed from the emergency department-a propensity matched analysis

Each year millions of Americans present to the emergency department (ED) for care after a motor vehicle collision (MVC); the majority (>90%) are discharged to home after evaluation. Acute musculoskeletal pain is the norm in this population, and such patients are typically discharged to home with prescriptions for oral opioid analgesics or nonsteroidal antiinflammatory drugs (NSAIDs). The influence of acute pain management on subsequent pain outcomes in this common ED population is unknown. We evaluated the effect of opioid analgesics vs NSAIDs initiated from the ED on the presence of moderate to severe musculoskeletal pain and ongoing opioid use at 6 weeks in a large cohort of adult ED patients presenting to the ED after MVC (n = 948). The effect of opioids vs NSAIDs was evaluated using an innovative quasi-experimental design method using propensity scores to account for covariate imbalances between the 2 treatment groups. No difference in risk for moderate to severe musculoskeletal pain at 6 weeks was observed between those discharged with opioid analgesics vs NSAIDs (risk difference = 7.2% [95% confidence interval: -5.2% to 19.5%]). However, at follow-up participants prescribed opioids were more likely than those prescribed NSAIDs to report use of prescription opioids medications at week 6 (risk difference = 17.5% [95% confidence interval: 5.8%-29.3%]). These results suggest that analgesic choice at ED discharge does not influence the development of persistent moderate to severe musculoskeletal pain 6 weeks after an MVC, but may result in continued use of prescription opioids. Supported by NIAMS R01AR056328 and AHRQ 5K12HS022998.

Comparative outcomes of immunochemotherapy regimens in Waldenström macroglobulinaemia

Comparative data on immunochemotherapy regimens for Waldenström macroglobulinaemia/lymphoplasmacytic lymphoma (WM/LPL) are lacking. We analysed overall survival (OS), risk of hospitalizations, transfusions and plasmapheresis in a population-based cohort of patients ≥65 years old initiating WM/LPL therapy in 1999-2013. To minimize bias, we applied a propensity score-based causal inference method. We conducted three analyses of: patients treated with or without rituximab, patients treated with rituximab monotherapy or with combination immunochemotherapy, and regimens based on classic purine analogues or alkylators. Among 1310 patients, 78·5% received rituximab. Patients who received rituximab had significantly better OS [hazard ratio (HR) 0·62, 95% confidence interval (CI) 0·55-0·71] and lower risk of transfusions (risk difference -3·3%, 95% CI -6·3 to -0·3) than those who did not, without a significant difference in hospitalizations or plasmapheresis. We observed no significant difference in OS (HR 0·91, 95% CI 0·79-1·04) between rituximab monotherapy and combination immunochemotherapy, but toxicity outcomes were lower with rituximab alone. Neither survival (HR 1·10, 95%CI 0·92-1·32) nor toxicity outcomes differed significantly between regimens based on purine analogues or alkylators. The survival advantage strongly supports rituximab as part of upfront therapy for WM/LPL, whereas regimens with either purine analogues or alkylating agents result in similar outcomes.

Identification of predicted individual treatment effects in randomized clinical trials

In most medical research, treatment effectiveness is assessed using the average treatment effect or some version of subgroup analysis. The practice of individualized or precision medicine, however, requires new approaches that predict how an individual will respond to treatment, rather than relying on aggregate measures of effect. In this study, we present a conceptual framework for estimating individual treatment effects, referred to as predicted individual treatment effects. We first apply the predicted individual treatment effect approach to a randomized controlled trial designed to improve behavioral and physical symptoms. Despite trivial average effects of the intervention, we show substantial heterogeneity in predicted individual treatment response using the predicted individual treatment effect approach. The predicted individual treatment effects can be used to predict individuals for whom the intervention may be most effective (or harmful). Next, we conduct a Monte Carlo simulation study to evaluate the accuracy of predicted individual treatment effects. We compare the performance of two methods used to obtain predictions: multiple imputation and non-parametric random decision trees. Results showed that, on average, both predictive methods produced accurate estimates at the individual level; however, the random decision trees tended to underestimate the predicted individual treatment effect for people at the extreme and showed more variability in predictions across repetitions compared to the imputation approach. Limitations and future directions are discussed.