Propensity score matching versus coarsened exact matching in observational comparative effectiveness research

2021 ◽  
Author(s):  
David Guy ◽  
Igor Karp ◽  
Piotr Wilk ◽  
Joseph Chin ◽  
George Rodrigues
Keyword(s):  
Clinical Trials ◽  
Propensity Score ◽  
Propensity Score Matching ◽  
Observational Data ◽  
Randomized Clinical Trials ◽  
Original Data ◽  
Considerable Proportion ◽  
Exact Matching ◽  
Coarsened Exact Matching ◽  
Comparison Groups

Aim & methods: We compared propensity score matching (PSM) and coarsened exact matching (CEM) in balancing baseline characteristics between treatment groups using observational data obtained from a pan-Canadian prostate cancer radiotherapy database. Changes in effect estimates were evaluated as a function of improvements in balance, using results from randomized clinical trials to guide interpretation. Results: CEM and PSM improved balance between groups in both comparisons, while retaining the majority of original data. Improvements in balance were associated with effect estimates closer to those obtained in randomized clinical trials. Conclusion: CEM and PSM led to substantial improvements in balance between comparison groups, while retaining a considerable proportion of original data. This could lead to improved accuracy in effect estimates obtained using observational data in a variety of clinical situations.

Using Propensity Score Matching to Improve Validity in Public Administration Research

2022 ◽  
pp. 45-60
Author(s):  
Michael Howell-Moroney
Keyword(s):  
Clinical Trials ◽  
Propensity Score ◽  
Public Administration ◽  
Propensity Score Matching ◽  
Observational Data ◽  
Randomized Clinical Trials ◽  
Causal Effect ◽  
Social Science Research ◽  
Causal Effects ◽  
Government Employment

Randomized clinical trials have a longstanding status as the gold standard in detecting causal effects. In the social sciences, randomized clinical trials are rare because of their attendant logistical and cost burdens. Most social science research makes use of observational data. The empirical challenge posed by observational data is that treatment assignment is no longer random. This challenge continues to spur innovation across many disciplines toward more sophisticated techniques for estimating causal relationships. Scholars have developed a common theoretical framework for estimating causal effects, often called the potential outcomes or counterfactual framework. This chapter demonstrates the propensity score matching methodology as a way to estimate causal effects using observational data. Throughout, an example from public administration research, the effect of government employment on volunteerism, is used to illustrate the concepts. Empirical estimates of the treatment effects show that there may be a causal effect of government employment on volunteerism.

scholarly journals Drivers of green bond issuance and new evidence on the “greenium”

2021 ◽  
Vol 11 (1) ◽  
pp. 1-24
Author(s):  
Kristin Ulrike Löffler ◽  
Aleksandar Petreski ◽  
Andreas Stephan
Keyword(s):  
Propensity Score ◽  
Propensity Score Matching ◽  
Bond Markets ◽  
Exact Matching ◽  
Secondary Markets ◽  
Coarsened Exact Matching ◽  
Secondary Bond ◽  
New Evidence

AbstractThis paper examines whether a premium for green bonds, called “greenium”, found in previous studies, exists in primary and secondary bond markets. Using a universe of about 2000 green and 180,000 non-green bonds from 650 international issuers, we apply both propensity score matching and coarsened exact matching to determine a sample of conventional bonds that is most similar to the sample of green bonds. We find that green bonds have larger issue sizes and lower rated issuers, on average, compared to conventional bonds. The estimates show that the yield for green bonds is, on average, 15–20 basis points lower than that of conventional bonds, both on primary and secondary markets, thus a “greenium” exists.

scholarly journals Evaluating the Utility of Coarsened Exact Matching for Pharmacoepidemiology Using Real and Simulated Claims Data

2019 ◽  
Vol 189 (6) ◽  
pp. 613-622 ◽  
Author(s):  
John E Ripollone ◽  
Krista F Huybrechts ◽  
Kenneth J Rothman ◽  
Ryan E Ferguson ◽  
Jessica M Franklin
Keyword(s):  
Propensity Score ◽  
Propensity Score Matching ◽  
Mahalanobis Distance ◽  
Claims Data ◽  
The Elderly ◽  
Effect Estimate ◽  
Exact Matching ◽  
Coarsened Exact Matching ◽  
Covariate Balance ◽  
Insurance Claims Data

Abstract Coarsened exact matching (CEM) is a matching method proposed as an alternative to other techniques commonly used to control confounding. We compared CEM with 3 techniques that have been used in pharmacoepidemiology: propensity score matching, Mahalanobis distance matching, and fine stratification by propensity score (FS). We evaluated confounding control and effect-estimate precision using insurance claims data from the Pharmaceutical Assistance Contract for the Elderly (1999–2002) and Medicaid Analytic eXtract (2000–2007) databases (United States) and from simulated claims-based cohorts. CEM generally achieved the best covariate balance. However, it often led to high bias and low precision of the risk ratio due to extreme losses in study size and numbers of outcomes (i.e., sparse data bias)—especially with larger covariate sets. FS usually was optimal with respect to bias and precision and always created good covariate balance. Propensity score matching usually performed almost as well as FS, especially with higher index exposure prevalence. The performance of Mahalanobis distance matching was relatively poor. These findings suggest that CEM, although it achieves good covariate balance, might not be optimal for large claims-database studies with rich covariate information; it might be ideal if only a few (<10) strong confounders must be controlled.

scholarly journals Can the Ceftriaxone Breakpoints Be Increased Without Compromising Patient Outcomes?

2018 ◽  
Vol 5 (6) ◽  
Author(s):  
Pranita D Tamma ◽  
Virginia M Pierce ◽  
Sara E Cosgrove ◽  
Ebbing Lautenbach ◽  
Anthony Harris ◽  
...  
Keyword(s):  
Propensity Score ◽  
Propensity Score Matching ◽  
Patient Outcomes ◽  
Nearest Neighbor ◽  
Clinical Laboratory ◽  
Severity Of Illness ◽  
Bloodstream Infections ◽  
Source Control ◽  
Exact Matching ◽  
Extended Spectrum

Abstract Background In 2010, the Clinical Laboratory and Standards Institute recommended a 3-fold lowering of ceftriaxone breakpoints to 1 mcg/mL for Enterobacteriaceae. Supportive clinical data at the time were from fewer than 50 patients. We compared the clinical outcomes of adults with Enterobacteriaceae bloodstream infections treated with ceftriaxone compared with matched patients (with exact matching on ceftriaxone minimum inhibitory concentrations [MICs]) treated with extended-spectrum agents to determine if ceftriaxone breakpoints could be increased without negatively impacting patient outcomes. Methods A retrospective cohort study was conducted at 3 large academic medical centers and included patients with Enterobacteriaceae bacteremia with ceftriaxone MICs of 2 mcg/mL treated with ceftriaxone or extended-spectrum β-lactams (ie, cefepime, piperacillin/tazobactam, meropenem, or imipenem/cilastatin) between 2008 and 2014; 1:2 nearest neighbor propensity score matching was performed to estimate the odds of recurrent bacteremia and mortality within 30 days. Results Propensity score matching yielded 108 patients in the ceftriaxone group and 216 patients in the extended-spectrum β-lactam group, with both groups well-balanced on demographics, preexisting medical conditions, severity of illness, source of bacteremia, and source control interventions. No difference in recurrent bacteremia (odds ratio [OR], 1.16; 95% confidence interval [CI], 0.49–2.73) or mortality (OR, 1.27; 95% CI, 0.56–2.91) between the treatment groups was observed for patients with isolates with ceftriaxone MICs of 2 mcg/mL. Only 6 isolates (1.6%) with ceftriaxone MICs of 2 mcg/mL were extended-spectrum β-lactamase (ESBL)–producing. Conclusions Our findings suggest that patient outcomes are similar when receiving ceftriaxone vs extended-spectrum agents for the treatment of Enterobacteriaceae bloodstream infections with ceftriaxone MICs of 2 mcg/mL. This warrants consideration of adjusting the ceftriaxone susceptibility breakpoint from 1 to 2 mcg/mL, as a relatively small increase in the antibiotic breakpoint could have the potential to limit the use of large numbers of extended-spectrum antibiotic agents.

scholarly journals Why Propensity Scores Should Not Be Used for Matching

2019 ◽  
Vol 27 (4) ◽  
pp. 435-454 ◽  
Author(s):  
Gary King ◽  
Richard Nielsen
Keyword(s):  
Propensity Score ◽  
Causal Inference ◽  
Propensity Score Matching ◽  
Propensity Scores ◽  
Original Data ◽  
The Other ◽  
Randomized Experiment ◽  
Random Matching ◽  
Popular Method ◽  
Matching Methods

We show that propensity score matching (PSM), an enormously popular method of preprocessing data for causal inference, often accomplishes the opposite of its intended goal—thus increasing imbalance, inefficiency, model dependence, and bias. The weakness of PSM comes from its attempts to approximate a completely randomized experiment, rather than, as with other matching methods, a more efficient fully blocked randomized experiment. PSM is thus uniquely blind to the often large portion of imbalance that can be eliminated by approximating full blocking with other matching methods. Moreover, in data balanced enough to approximate complete randomization, either to begin with or after pruning some observations, PSM approximates random matching which, we show, increases imbalance even relative to the original data. Although these results suggest researchers replace PSM with one of the other available matching methods, propensity scores have other productive uses.

On the Evaluation of Treatments for Narcotics Addiction

1979 ◽  
Vol 9 (2) ◽  
pp. 205-211 ◽  
Author(s):  
Vincent P. Dole ◽  
Burton Singer
Keyword(s):  
Clinical Trials ◽  
Randomized Trial ◽  
Observational Data ◽  
Chronic Diseases ◽  
Drug Addiction ◽  
Randomized Trials ◽  
Randomized Clinical Trials ◽  
The Other ◽  
Domain Of Applicability

This study is concerned with the domain of applicability of randomized clinical trials. For evaluation of well-defined treatments of acute diseases over limited periods of time, the randomized trial technique is unquestionably the best. However, in the field of chronic diseases (as illustrated by drug addiction) the physician's responsibility extends over periods of years, and his judgements involve consideration of many contingent factors which vary in the course of the disease. In this domain, randomized clinical trials, however ambitious in design, give only partial guidance. Observational data therefore must be used if treatment is to be optimized for individual patients. Two randomized trials in the treatment of narcotics addiction — one testing methadone and the other naltrexone — are reviewed, with comments on their conclusions and limitations.

Comparative effectiveness of treatment options in elderly patients with colon cancer receiving adjuvant chemotherapy.

2012 ◽  
Vol 30 (4_suppl) ◽  
pp. 470-470
Author(s):  
Sacha Satram-Hoang ◽  
Devi Ramanan ◽  
Luen F. Lee ◽  
Shui Yu ◽  
Carolina M. Reyes ◽  
...  
Keyword(s):  
Colon Cancer ◽  
Clinical Trials ◽  
Propensity Score ◽  
Elderly Patients ◽  
Older Patients ◽  
Randomized Clinical Trials ◽  
Cohort Analysis ◽  
Risk Of Death ◽  
Morbidity Score ◽  
Co Morbidity

470 Background: While colon cancer (CC) is predominantly a disease of the elderly, older patients are underrepresented in clinical trials. We sought to evaluate whether the treatment patterns and benefits realized by trial participants pertain to older patients in the real-world setting. Methods: Using the linked Surveillance, Epidemiology, and End Results (SEER)-Medicare database, we conducted a retrospective cohort analysis of 3390 stage II and III CC patients diagnosed between 1/1/ 2004 to 12/31/2007, who were >66 years, enrolled in Medicare Parts A and B, and received adjuvant treatment with 5FU/LV (n=1368), FOLFOX (n=1398), CAP (capecitabine; n=507), and CAPOX (CAP + oxaliplatin; n=117) within 3 months after surgery. Date of last follow-up was 12/31/2007. Chi-square test and ANOVA or t-test assessed differences in patient and disease characteristics by treatment. Propensity score weighted Cox regression assessed the relative risk of death by treatment. Results: Patients treated with CAP were older (mean age 77 years; p<.0001), more likely female (61%; p<.05), more likely non-white (19%; p<.05) and had higher co-morbidity score (p<.0001) compared to the other treatment groups. The mean time to chemo initiation after surgery were similar between the groups (mean 46-49 days) while mean duration of treatment were longer for 5FU/LV (149 days) and FOLFOX (144 days), compared to CAP (121 days) and CAPOX (111 days); p<.0001. The incidence of adverse events (AEs) within 180 days after initiation of treatment were higher in patients treated with FOLFOX (82%) and 5FU/LV (78%) compared to CAP (74%) and CAPOX (71%); p=0.0002. Propensity score adjusted multivariate analysis demonstrated comparable survival for CAP-based regimens vs. 5-FU/LV- based regimens ( table ). Conclusions: Treatment outcomes for elderly patients observed in routine clinical practice were comparable between CAP-based and 5FU/LV-based regimens and consistent with results reported in randomized clinical trials. AEs associated with medical resource utilization were less frequent with CAP-based regimens. [Table: see text]

Sign in / Sign up

Export Citation Format

Share Document