Propensity score model overfitting led to inflated variance of estimated odds ratios

Can supervised deep learning architecture outperform autoencoders in building propensity score models for matching?

PURPOSE

Propensity score matching is vital in epidemiological studies using observational data, yet its estimates relies on correct model-specification. This study assesses supervised deep learning models and unsupervised autoencoders for propensity score estimation, comparing them with traditional methods for bias and variance accuracy in treatment effect estimations.

METHODS

Utilizing a plasmode simulation based on the Right Heart Catheterization dataset, under a variety of settings, we evaluated (1) a supervised deep learning architecture and (2) an unsupervised autoencoder, alongside two traditional methods: logistic regression and a spline-based method in estimating propensity scores for matching. Performance metrics included bias, standard errors, and coverage probability. The analysis was also extended to real-world data, with estimates compared to those obtained via a double robust approach.

RESULTS

The analysis revealed that supervised deep learning models outperformed unsupervised autoencoders in variance estimation while maintaining comparable levels of bias. These results were supported by analyses of real-world data, where the supervised model's estimates closely matched those derived from conventional methods. Additionally, deep learning models performed well compared to traditional methods in settings where exposure was rare.

CONCLUSION

Supervised deep learning models hold promise in refining propensity score estimations in epidemiological research, offering nuanced confounder adjustment, especially in complex datasets. We endorse integrating supervised deep learning into epidemiological research and share reproducible codes for widespread use and methodological transparency.

A cross-sectional assessment of injection of "salts" and HIV transmission-related behaviours among a cohort of people who inject drugs in Kyrgyzstan

INTRODUCTION

Despite the increasing availability of new psychoactive substances (hereafter referred to as "salts") in Eastern Europe and Central Asia, there is a dearth of epidemiological data on the relationship between injecting "salts" and HIV risk behaviours. This is particularly relevant in settings where injection drug use accounts for a substantial proportion of the HIV burden, such as in Kyrgyzstan, a former Soviet Republic. This study assessed whether injecting "salts" is associated with sexual and injection-related HIV risk behaviours among people who inject drugs in Kyrgyzstan.

METHODS

The Kyrgyzstan InterSectional Stigma Study is a cohort of people who inject drugs in Kyrgyzstan's capital of Bishkek and the surrounding rural administrative division of Chuy Oblast. We conducted a cross-sectional analysis using survey data collected from cohort participants between July and November 2021, which included information on injection drug use (including "salts") and HIV risk behaviours. To minimize confounding by measured covariates, we used inverse-probability-weighted logistic and Poisson regression models to estimate associations between recent "salt" injection and HIV risk behaviours.

RESULTS

Of 181 participants included in the analysis (80.7% men, 19.3% women), the mean age was 40.1 years (standard deviation [SD] = 8.8), and 22% (n = 39) reported that they had injected "salts" in the past 6 months. Among people who injected "salts," 72% (n = 28) were men, and most were ethnically Russian 59% (n = 23), with a mean age of 34.6 (SD = 9.6). Injecting "salts" was significantly associated with a greater number of injections per day (adjusted relative risk [aRR] = 1.59, 95% confidence interval [CI] = 1.30-1.95) but lower odds of using syringe service programmes in the past 6 months (adjusted odds ratio [aOR] = 0.20, 95% CI = 0.12-0.32). Injecting "salts" was also significantly associated with lower odds of condomless sex in the past 6 months (aOR = 0.42, 95% CI = 0.24-0.76) and greater odds of having ever heard of pre-exposure prophylaxis (aOR = 4.80, 95% CI = 2.61-8.83).

CONCLUSIONS

(PWID) people who inject drugs who inject "salts" are a potentially emergent group with increased HIV acquisition risk in Kyrgyzstan. Targeted outreach bundled with comprehensive harm reduction and pre-exposure prophylaxis services are needed to prevent transmission of HIV and other blood-borne viruses.

Long-term health conditions and UK labour market outcomes during the COVID-19 pandemic

BACKGROUND

Long-term health conditions can affect labour market outcomes. COVID-19 may have increased labour market inequalities, e.g. due to restricted opportunities for clinically vulnerable people. Evaluating COVID-19's impact could help target support.

AIM

To quantify the effect of several long-term conditions on UK labour market outcomes during the COVID-19 pandemic and compare them to pre-pandemic outcomes.

METHODS

The Understanding Society COVID-19 survey collected responses from around 20,000 UK residents in nine waves from April 2020-September 2021. Participants employed in January/February 2020 with a variety of long-term conditions were matched with people without the condition but with similar baseline characteristics. Models estimated probability of employment, hours worked and earnings. We compared these results with results from a two-year pre-pandemic period. We also modelled probability of furlough and home-working frequency during COVID-19.

RESULTS

Most conditions (asthma, arthritis, emotional/nervous/psychiatric problems, vascular/pulmonary/liver conditions, epilepsy) were associated with reduced employment probability and/or hours worked during COVID-19, but not pre-pandemic. Furlough was more likely for people with pulmonary conditions. People with arthritis and cancer were slower to return to in-person working. Few effects were seen for earnings.

CONCLUSION

COVID-19 had a disproportionate impact on people with long-term conditions' labour market outcomes.

Propensity score matching as an effective strategy for biomarker cohort design and omics data analysis

BACKGROUND

Analysis of omics data that contain multidimensional biological and clinical information can be complex and make it difficult to deduce significance of specific biomarker factors.

METHODS

We explored the utility of propensity score matching (PSM), a statistical technique for minimizing confounding factors and simplifying the examination of specific factors. We tested two datasets generated from cohorts of colorectal cancer (CRC) patients, one comprised of immunohistochemical analysis of 12 protein markers in 544 CRC tissues and another consisting of RNA-seq profiles of 163 CRC cases. We examined the efficiency of PSM by comparing pre- and post-PSM analytical results.

RESULTS

Unlike conventional analysis which typically compares randomized cohorts of cancer and normal tissues, PSM enabled direct comparison between patient characteristics uncovering new prognostic biomarkers. By creating optimally matched groups to minimize confounding effects, our study demonstrates that PSM enables robust extraction of significant biomarkers while requiring fewer cancer cases and smaller overall patient cohorts.

CONCLUSION

PSM may emerge as an efficient and cost-effective strategy for multiomic data analysis and clinical trial design for biomarker discovery.

Incidence of Esophageal Adenocarcinoma, Mortality, and Esophagectomy in Barrett's Esophagus Patients Undergoing Endoscopic Eradication Therapy

INTRODUCTION

Endoscopic eradication therapy (EET) is the preferred treatment for Barrett's esophagus (BE)-related neoplasia patients. However, the impact of EET on critical outcomes, outside of clinical trials and registry data, remains scarcely studied. We aimed to assess real-world practice patterns and clinical outcomes among BE patients undergoing EET.

METHODS

TriNetX is a large research network comprising linked inpatient and outpatient electronic-health record-derived data from over 80,000,000 patients. Patients with a diagnosis of BE from 2015 to 2020 were identified and included if they underwent EET during the study period. The primary outcome was the progression to EAC after index EET. Secondary outcomes included rate of esophagectomy, and all-cause mortality. All outcomes were stratified by baseline histology. The incidence of EAC and all-cause mortality were reported in person-years and adjusted for age and sex.

RESULTS

A total of 4114 patients were analyzed. Distribution of baseline histology was as follows: NDBE (11.8%), LGD (21.4%), HGD (26.4%), EAC (20.8%), and unspecified (19.6%). The total incidence of EAC after index EET was 6.01 per 1000 person-years (PY) for the entire cohort with the highest rate in HGD patients (12.9/1000 PY). The incidence of all-cause mortality was 13.23 per 1000 PY with the highest rates in EAC patients (25.1 per 1000 PY). Rates of esophagectomy were < 1% for all grades of dysplasia.

CONCLUSION

The results of this study provide "real-world" data on critical outcomes for BE patients undergoing EET, demonstrating a low risk of incident EAC, all-cause mortality, and need for esophagectomy.

First-line opioid agonist treatment as prevention against assisting others in initiating injection drug use: A longitudinal cohort study of people who inject drugs in Vancouver, Canada

Background

Among people who inject drugs, frequent injecting and experiencing withdrawal are associated with facilitating others' first injections. As these factors may reflect an underlying substance use disorder, we investigated whether first-line oral opioid agonist treatment (OAT; methadone or buprenorphine/naloxone) reduces the likelihood that people who inject drugs help others initiate injecting.

Methods

We used questionnaire data from semi-annual visits between December 2014-May 2018 on 334 people who inject drugs with frequent non-medical opioid use in Vancouver, Canada. We estimated the effect of current first-line OAT on subsequent injection initiation assistance provision (i.e., helped someone initiate injecting in the following six months) using inverse-probability-weighted estimation of repeated measures marginal structural models to reduce confounding and informative censoring by time-fixed and time-varying covariates.

Results

By follow-up visit, 54-64% of participants reported current first-line OAT whereas 3.4-6.9% provided subsequent injection initiation assistance. Per the primary weighted estimate (n = 1114 person-visits), participants currently on first-line OAT (versus no OAT) were 50% less likely, on average, to subsequently help someone initiate injecting (relative risk [RR]=0.50, 95% CI=0.23-1.11). First-line OAT was associated with reduced risk of subsequent injection initiation assistance provision in participants who, at baseline, injected opioids less than daily (RR=0.15, 95% CI=0.05-0.44) but not in those who injected opioids daily (RR=0.86, 95% CI=0.35-2.11).

Conclusions

First-line OAT seemingly reduces the short-term likelihood that people who inject drugs facilitate first injections. However, the extent of this potential effect remains uncertain due to imprecise estimation and observed heterogeneity by baseline opioid injecting frequency.

Oolong Tea Consumption and the Risk of Oral Squamous Cell Carcinoma: A Propensity Score-Based Analysis in Southeast China

Oolong tea is one of the world's most popular non-alcoholic beverages, particularly in coastal Southeast China. Hitherto, epidemiological studies on the association between oolong tea consumption and the risk of patients with oral squamous cell carcinoma (OSCC) are very limited. This study aimed to evaluate the potential effect of oolong tea consumption on OSCC risk in Southeast China. From January 2010 to October 2020, face-to-face interviews were conducted for 744 newly diagnosed OSCC patients and 1,029 healthy controls to collect information on demographics, oolong tea consumption behaviors, and other lifestyle factors. Propensity score matching (PSM), inverse probability of treatment weight (IPTW), and stabilized inverse probability of treatment weight (SIPTW) were utilized to minimize confounding effects. Multivariate, conditional, and weighted logistic regression was used to evaluate the associations of oolong tea consumption behaviors with OSCC risk. Participants who drank oolong tea showed a lower risk of OSCC when compared to their non-drink counterparts [PSM population, OR (95%CI): 0.69 (0.49–0.97); SIPTW population, OR (95%CI): 0.74 (0.58–0.94)]. Moreover, the reduced risk was found to be significantly associated with certain tea-drinking habits (consumed amount over 500 mL per day, a duration of <20 years, age at initiation older than 30 years, and warm and moderately concentrated tea). Similar results were yielded in the sensitivity analyses (Multivariate adjustment and the IPTW analysis). Furthermore, subgroup analysis revealed that the negative association of oolong tea drinking with OSCC risk was more evident among those with poor oral hygiene. This study provides supportive evidence that oolong tea consumption may have a potentially beneficial effect in preventing OSCC, especially for those with poor oral hygiene.

Association between proton pump inhibitor use and gastric cancer: a population-based cohort study using two different types of nationwide databases in Korea

OBJECTIVE

The association between proton pump inhibitor (PPI) use and gastric cancer related to Helicobacter pylori eradication has not been fully investigated in geographical regions with high risk of gastric cancer. We aimed to evaluate the association between PPIs and gastric cancer in Korea.

DESIGN

This study analysed the original and common data model versions of the Korean National Health Insurance Service database from 2002 to 2013. We compared the incidence rates of gastric cancer after 1-year drug exposure, between new users of PPIs and other drugs excluding PPIs, by Cox proportional hazards model. We also analysed the incidence of gastric cancer among PPI users after H. pylori eradication.

RESULTS

The analysis included 11 741 patients in matched PPI and non-PPI cohorts after large-scale propensity score matching. During a median follow-up of 4.3 years, PPI use was associated with a 2.37-fold increased incidence of gastric cancer (PPI≥30 days vs non-PPI; 118/51 813 person-years vs 40/49 729 person-years; HR 2.37, 95% CI 1.56 to 3.68, p=0.001). The incidence rates of gastric cancer showed an increasing trend parallel to the duration of PPI use. In H. pylori-eradicated subjects, the incidence of gastric cancer was significantly associated with PPI use over 180 days compared with the non-PPI group (PPI≥180 days vs non-PPI; 30/12 470 person-years vs 9/7814 person-years; HR 2.22, 95% CI 1.05 to 4.67, p=0.036).

CONCLUSION

PPI use was associated with gastric cancer, regardless of H. pylori eradication status. Long-term PPIs should be used with caution in high-risk regions for gastric cancer.

Simulation-based design of pragmatic trials in psoriatic arthritis using propensity scores

BACKGROUND/AIMS

Design of clinical trials requires careful decision-making across several dimensions, including endpoints, eligibility criteria, and subgroup enrichment. Clinical trial simulation can be an informative tool in trial design, providing empirical evidence by which to evaluate and compare the results of hypothetical trials with varying designs. We introduce a novel simulation-based approach using observational data to inform the design of a future pragmatic trial.

METHODS

We utilize propensity score-adjusted models to simulate hypothetical trials under alternative endpoints and enrollment criteria. We apply our approach to the design of pragmatic trials in psoriatic arthritis, using observational data embedded within the Tight Control of Inflammation in Early Psoriatic Arthritis study to simulate hypothetical open-label trials comparing treatment with tumor necrosis factor-α inhibitors to methotrexate. We first validate our simulations of a trial with traditional enrollment criteria and endpoints against a recently published trial. Next, we compare simulated treatment effects in patient populations defined by traditional and broadened enrollment criteria, where the latter is consistent with a future pragmatic trial. In each trial, we also consider five candidate primary endpoints.

RESULTS

Our results highlight how changes in the enrolled population and primary endpoints may qualitatively alter study findings and the ability to detect heterogeneous treatment effects between clinical subgroups. For treatments of interest in the study of psoriatic arthritis, broadened enrollment criteria led to diluted estimated treatment effects. Endpoints with greater responsiveness to treatment compared with a traditionally used endpoint were identified. These considerations, among others, are important for designing a future pragmatic trial aimed at having high external validity with relevance for real-world clinical practice.

CONCLUSION

Observational data may be leveraged to inform design decisions in pragmatic trials. Our approach may be generalized to the study of other conditions where existing trial data are limited or do not generalize well to real-world clinical practice, but where observational data are available.

Circulating ANGPTL8 levels and risk of kidney function decline: Results from the 4C Study

Background

ANGPTL8, an important regulator of lipid metabolism, was recently proven to have additional intracellular and receptor-mediated functions. This study aimed to investigate circulating levels of ANGPTL8 and its potential association with the risk of kidney function decline in a cohort study.

Methods

We analysed 2,311 participants aged 40 years old and older from the China Cardiometabolic Disease and Cancer Cohort (4C) Study. Kidney function decline was defined as an estimated glomerular filtration rate (eGFR) less than 60 mL per minute per 1.73 m² of body surface area, a decrease in eGFR of ≥ 30% from baseline, chronic kidney disease (CKD)-related hospitalization or death, or end-stage renal disease. The association between baseline ANGPTL8 levels and kidney function decline was assessed using multivariable-adjusted Cox proportional hazards models, and inverse possibility of treatment weight (IPTW) was utilized to prevent overfitting.

Results

There were 136 (5.9%) cases of kidney function decline over a median of 3.8 years of follow-up. We found that serum ANGPTL8 levels at baseline were elevated in individuals with kidney function decline compared to those without kidney function decline during follow-up (718.42 ± 378.17 vs. 522.04 ± 283.07 pg/mL, p < 0.001). Compared with the first quartile, multivariable-adjusted hazard ratio (95% confidence intervals [CIs]) for kidney function decline was 2.59 (95% CI, 1.41–4.77) for the fourth ANGPTL8 quartile. Furthermore, compared with patients in the first ANGPTL8 quartile, those in the fourth ANGPTL8 quartile were more likely to report a higher stage of CKD (relative risk: 1.33; 95% CI, 1.01–1.74). The conclusions of the regression analyses were not altered in the IPTW models. Multivariable-adjusted restricted cubic spline analyses suggested a linear relationship of ANGPTL8 with kidney function decline (p for nonlinear trend = 0.66, p for linear trend < 0.001).

Conclusions

Participants with higher circulating ANGPTL8 levels were at increased risk for kidney function decline, highlighting the importance of future studies addressing the pathophysiological role of ANGPTL8 in CKD.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-021-01317-3.

Enabling Technologies and the Role of Private Firms: A Machine Learning Matching Analysis

Investments in enabling technologies—including the fifth-generation technology standard for broadband cellular networks (5G), artificial intelligence (AI), and light detection and ranging (LIDAR) technology—are important strategic decisions for firms. This paper asks how inventions that private firms developed with (versus without) public-sector partners differ in their enabling technology trajectory. Using a novel method of machine learning matching, we compare patented technologies generated from more than 30,000 public–private relationships with comparable technologies invented by private firms alone during a 21-year period. To measure the enabling potential of a technology, we introduce a new enabling technology index. The findings show that private-firm relationships with the public sector—in particular cooperative agreements and grants with mission agencies (National Aeronautics and Space Administration and Department of Defense)—are likely starting points for enabling technology trajectories. We thus put a spotlight on organizational arrangements that combine the breadth of exploration (agreements, grants) with deep exploitation in a particular domain (mission agency). A key contribution is a better understanding of the types of private-firm efforts that are associated with enabling technologies. We also challenge the common assumption that enabling technologies have their origins only in public-sector projects and show how private firms are involved. Our significant contribution is to show how private firms can change evolution of ecosystems through technology development.

Using propensity scores to estimate effects of treatment initiation decisions: State of the science

Confounding can cause substantial bias in nonexperimental studies that aim to estimate causal effects. Propensity score methods allow researchers to reduce bias from measured confounding by summarizing the distributions of many measured confounders in a single score based on the probability of receiving treatment. This score can then be used to mitigate imbalances in the distributions of these measured confounders between those who received the treatment of interest and those in the comparator population, resulting in less biased treatment effect estimates. This methodology was formalized by Rosenbaum and Rubin in 1983 and, since then, has been used increasingly often across a wide variety of scientific disciplines. In this review article, we provide an overview of propensity scores in the context of real-world evidence generation with a focus on their use in the setting of single treatment decisions, that is, choosing between two therapeutic options. We describe five aspects of propensity score analysis: alignment with the potential outcomes framework, implications for study design, estimation procedures, implementation options, and reporting. We add context to these concepts by highlighting how the types of comparator used, the implementation method, and balance assessment techniques have changed over time. Finally, we discuss evolving applications of propensity scores.

Bleeding risk of ticagrelor compared to clopidogrel in intensive care unit patients with acute coronary syndrome: A propensity-score matching analysis

BACKGROUND

Intensive care unit (ICU) patients with the most severe forms of acute coronary syndrome (ACS) require invasive therapies such as extracorporeal life support. The risk of bleeding in ICU patients with ACS treated with a dual antiplatelet therapy of aspirin and ticagrelor is unknown. The primary objective of this study was to compare the bleeding risk of ticagrelor and clopidogrel in ICU patients with ACS.

METHODS AND FINDINGS

We conducted a retrospective study based on a propensity score and a proportional hazards model. All patients with ACS hospitalized in the ICU of a French university hospital between January 2013 and January 2017 were included in the study. Bleeding during ICU stay was defined as all Thrombolysis in myocardial infarction (TIMI) major or minor events. A total of 155 patients were included in the study. According to propensity score matching, 57 patients treated with aspirin and ticagrelor were matched with 57 patients treated with aspirin and clopidogrel. Median (first-third quartile) Simplified Acute Physiology Score II was 61.5 (41.0-85.0). Bleeding during ICU stay occurred in 12 patients (21.1%) treated with clopidogrel and in 35 patients (61.4%) treated with ticagrelor (p<0.0001). This significant association was found for both TIMI major bleeding (12.3% vs. 35.1%, p = 0.004) and TIMI minor bleeding (8.8% vs. 26.3%, p = 0.01). The relative risk of bleeding occurrence during ICU stay was 2.60 (confidence interval 95%: 1.55-4.35) for ticagrelor compared to clopidogrel. No significant difference in ICU mortality was found between the two groups (45.6% in the clopidogrel group vs. 29.8% in the ticagrelor group, p = 0.08).

CONCLUSIONS

Bleeding complications are frequent and serious in ICU patients with ACS. A dual antiplatelet therapy of aspirin and ticagrelor is associated with a higher risk of bleeding compared to a dual antiplatelet therapy of aspirin and clopidogrel.

CF290 for pancolonic chromoendoscopy improved sessile serrated polyp detection and procedure time: a propensity score-matching study

Background and study aims  The benefits of the new-generation CF290 (Olympus, Tokyo, Japan) for pancolonic chromoendoscopy (PCC) for colorectal polyp detection and its procedure time remain questionable. We compared the CF290 with the previous CF260 for PCC. Methods  We performed a propensity score-matching study using baseline characteristics such as age, sex, indications, endoscopist, and bowel preparation. We compared the detection of adenomas and sessile serrated polyps (SSPs) and procedure times of two expert endoscopists who performed PCC using the CF290 series (high-quality system with flushing pump) and the CF260 series (high-definition system). Results  We matched 374 patients who underwent PCC using the CF290 and 187 patients who underwent PCC using the CF260. The adenoma detection rate of the 290 series was higher than that of the 260 series, but not significantly. The SSP detection rate for the 290 series was higher than that for the 260 series ( P  = 0.01). Insertion time required for the 290 series was shorter than that required for the 260 series ( P  < 0.0001). Withdrawal time of the 290 series was shorter than that of the 260 series ( P  < 0.0001). Conclusion  Advanced technology can provide accuracy and help save time, and therefore, should be applied whenever possible.

Should they stay or should they go? Who benefits from interfacility transfer to a higher-level trauma center following initial presentation at a lower-level trauma center

BACKGROUND

Interfacility transfer of patients from Level III/IV to Level I/II (tertiary) trauma centers has been associated with improved outcomes. However, little data are available classifying the specific subsets of patients that derive maximal benefit from transfer to a tertiary trauma center. Drawbacks to transfer include increased secondary overtriage. Here, we ask which injury patterns are associated with improved survival following interfacility transfer.

METHODS

Data from the National Trauma Data Bank was utilized. Inclusion criteria were adults (≥16 years). Patients with Injury Severity Score of 10 or less or those who arrived with no signs of life were excluded. Patients were divided into two cohorts: those admitted to a Level III/IV trauma center versus those transferred into a tertiary trauma center. Multiple imputation was performed for missing values, and propensity scores were generated based on demographics, injury patterns, and disease severity. Using propensity score-stratified Cox proportional hazards regression, the hazard ratio for time to death was estimated.

RESULTS

Twelve thousand five hundred thirty-four (5.2%) were admitted to Level III/IV trauma centers, and 227,315 (94.8%) were transferred to a tertiary trauma center. Patients transferred to a tertiary trauma center had reduced mortality (hazard ratio, 0.69; p < 0.001). We identified that patients with traumatic brain injury with Glasgow Coma Scale score less than 13, pelvic fracture, penetrating mechanism, solid organ injury, great vessel injury, respiratory distress, and tachycardia benefited from interfacility transfer to a tertiary trauma center. In this sample, 56.8% of the patients benefitted from transfer. Among those not transferred, 49.5% would have benefited from being transferred.

CONCLUSION

Interfacility transfer is associated with a survival benefit for specific patients. These data support implementation of minimum evidence-based criteria for interfacility transfer.

LEVEL OF EVIDENCE

Therapeutic/Care Management, Level IV.

Impact of levosimendan on weaning from peripheral venoarterial extracorporeal membrane oxygenation in intensive care unit

BACKGROUND

Few data are available on the impact of levosimendan in refractory cardiogenic shock patients undergoing peripheral venoarterial extracorporeal membrane oxygenation (VA-ECMO). The aim of this study was to evaluate the impact of levosimendan on VA-ECMO weaning in patients hospitalized in intensive care unit (ICU).

METHODS

This retrospective cohort study was conducted in a French university hospital from 2010 to 2017. All patients hospitalized in ICU undergoing VA-ECMO were consecutively evaluated.

RESULTS

A total of 150 patients undergoing VA-ECMO were eligible for the study. Thirty-eight propensity-matched patients were evaluated in the levosimendan group and 65 in the non-levosimendan group. In patients treated with levosimendan, left ventricular ejection fraction had increased from 21.5 ± 9.1% to 30.7 ± 13.5% (P < 0.0001) and aortic velocity-time integral from 8.9 ± 4 cm to 12.5 ± 3.8 cm (P = 0.002) 24 h after drug infusion. After propensity score matching, levosimendan was the only factor associated with a significant reduction in VA-ECMO weaning failure rates (hazard ratio = 0.16; 95% confidence interval 0.04-0.7; P = 0.01). Kaplan-Meier survival curves showed that survival rates at 30 days were 78.4% for the levosimendan group and 49.5% for the non-levosimendan group (P = 0.02). After propensity score matching analysis, the difference in 30-day mortality between the two groups was not significant (hazard ratio = 0.55; 95% confidence interval 0.27-1.10; P = 0.09).

CONCLUSIONS

Our results suggest that levosimendan was associated with a beneficial effect on VA-ECMO weaning in ICU patients.

Impact of an invasive strategy in the elderly hospitalized with acute coronary syndrome with emphasis on the nonagenarians

BACKGROUND

Published data about nonagenarians with acute coronary syndrome (ACS) were mainly descriptive and limited by small sample sizes and unadjusted outcomes. We aim to describe the characteristics, management, and the impact of an invasive strategy on major adverse events in elderly patients hospitalized with ACS with focus on the nonagerians.

METHODS AND RESULTS

We analyzed data collected as part of the AMI-OPTIMA study, a cluster-randomized study of knowledge translation intervention versus usual care on optimal discharge medications in patients admitted with ACS at 24 Canadian hospitals. To determine whether an invasive strategy improved outcomes in the elderly, we used inverse probability weighting to adjust for confounders between patients who underwent invasive versus conservative strategies. Of 4,569 consecutive patients: 2,395 (52%) were <70 years old, 1,031 (23%) were septuagenarians, 941 (21%) were octogenarians, and 202 (4.4%) were nonagenarians. An invasive strategy was associated with reduced in-hospital all-cause mortality in all age groups: 1.1% versus 3.8% in patients <70 years old (P < 0.001), 2.9% versus 7.4% in septuagenarians (P < 0.001), 5.1% versus 14.7% in octogenarians (P < 0.001), and 12.0% versus 25.1% in nonagenarians (P = 0.001). An invasive strategy was also associated with higher thrombolysis in myocardial infarction major bleeds in the nonagenarians (9.0% vs. 2.0%; P = 0.003).

CONCLUSIONS

The reduction in in-hospital mortality associated with an invasive strategy in elderly and nonagenarians presented with ACS is generating hypothesis and merits further studies to confirm these benefits and to guide clinicians in the management of these high-risk patients.

Preoperative oral care and effect on postoperative complications after major cancer surgery

Background

Improving patients' oral hygiene is an option for preventing postoperative pneumonia that may be caused by aspiration of oral and pharyngeal secretions. Whether preoperative oral care by a dentist can decrease postoperative complications remains controversial. A retrospective cohort study was undertaken to assess the association between preoperative oral care and postoperative complications among patients who underwent major cancer surgery.

Methods

The nationwide administrative claims database in Japan was analysed. Patients were identified who underwent resection of head and neck, oesophageal, gastric, colorectal, lung or liver cancer between May 2012 and December 2015. The primary outcomes were postoperative pneumonia and all-cause mortality within 30 days of surgery. Patient background was adjusted for with inverse probability of treatment weighting using propensity scoring.

Results

Of 509 179 patients studied, 81 632 (16·0 per cent) received preoperative oral care from a dentist. A total of 15 724 patients (3·09 per cent) had postoperative pneumonia and 1734 (0·34 per cent) died within 30 days of surgery. After adjustment for potential confounding factors, preoperative oral care by a dentist was significantly associated with a decrease in postoperative pneumonia (3·28 versus 3·76 per cent; risk difference − 0·48 (95 per cent c.i. −0·64 to−0·32) per cent) and all-cause mortality within 30 days of surgery (0·30 versus 0·42 per cent; risk difference − 0·12 (−0·17 to −0·07) per cent).

Conclusion

Preoperative oral care by a dentist significantly reduced postoperative complications in patients who underwent cancer surgery.

An evaluation of exact matching and propensity score methods as applied in a comparative effectiveness study of inhaled corticosteroids in asthma

Background

Cohort matching and regression modeling are used in observational studies to control for confounding factors when estimating treatment effects. Our objective was to evaluate exact matching and propensity score methods by applying them in a 1-year pre–post historical database study to investigate asthma-related outcomes by treatment.

Methods

We drew on longitudinal medical record data in the PHARMO database for asthma patients prescribed the treatments to be compared (ciclesonide and fine-particle inhaled corticosteroid [ICS]). Propensity score methods that we evaluated were propensity score matching (PSM) using two different algorithms, the inverse probability of treatment weighting (IPTW), covariate adjustment using the propensity score, and propensity score stratification. We defined balance, using standardized differences, as differences of <10% between cohorts.

Results

Of 4064 eligible patients, 1382 (34%) were prescribed ciclesonide and 2682 (66%) fine-particle ICS. The IPTW and propensity score-based methods retained more patients (96%–100%) than exact matching (90%); exact matching selected less severe patients. Standardized differences were >10% for four variables in the exact-matched dataset and <10% for both PSM algorithms and the weighted pseudo-dataset used in the IPTW method. With all methods, ciclesonide was associated with better 1-year asthma-related outcomes, at one-third the prescribed dose, than fine-particle ICS; results varied slightly by method, but direction and statistical significance remained the same.

Conclusion

We found that each method has its particular strengths, and we recommend at least two methods be applied for each matched cohort study to evaluate the robustness of the findings. Balance diagnostics should be applied with all methods to check the balance of confounders between treatment cohorts. If exact matching is used, the calculation of a propensity score could be useful to identify variables that require balancing, thereby informing the choice of matching criteria together with clinical considerations.