Rolf H.H. Groenwold

Attributable mortality of ventilator-associated pneumonia: a meta-analysis of individual patient data from randomised prevention studies

BACKGROUND Estimating attributable mortality of ventilator-associated pneumonia has been hampered by confounding factors, small sample sizes, and the difficulty of doing relevant subgroup analyses. We estimated the attributable mortality using the individual original patient data of published randomised trials of ventilator-associated pneumonia prevention. METHODS We identified relevant studies through systematic review. We analysed individual patient data in a one-stage meta-analytical approach (in which we defined attributable mortality as the ratio between the relative risk reductions [RRR] of mortality and ventilator-associated pneumonia) and in competing risk analyses. Predefined subgroups included surgical, trauma, and medical patients, and patients with different categories of severity of illness scores. FINDINGS Individual patient data were available for 6284 patients from 24 trials. The overall attributable mortality was 13%, with higher mortality rates in surgical patients and patients with mid-range severity scores at admission (ie, acute physiology and chronic health evaluation score [APACHE] 20-29 and simplified acute physiology score [SAPS 2] 35-58). Attributable mortality was close to zero in trauma, medical patients, and patients with low or high severity of illness scores. Competing risk analyses could be done for 5162 patients from 19 studies, and the overall daily hazard for intensive care unit (ICU) mortality after ventilator-associated pneumonia was 1·13 (95% CI 0·98-1·31). The overall daily risk of discharge after ventilator-associated pneumonia was 0·74 (0·68-0·80), leading to an overall cumulative risk for dying in the ICU of 2·20 (1·91-2·54). Highest cumulative risks for dying from ventilator-associated pneumonia were noted for surgical patients (2·97, 95% CI 2·24-3·94) and patients with mid-range severity scores at admission (ie, cumulative risks of 2·49 [1·81-3·44] for patients with APACHE scores of 20-29 and 2·72 [1·95-3·78] for those with SAPS 2 scores of 35-58). INTERPRETATION The overall attributable mortality of ventilator-associated pneumonia is 13%, with higher rates for surgical patients and patients with a mid-range severity score at admission. Attributable mortality is mainly caused by prolonged exposure to the risk of dying due to increased length of ICU stay. FUNDING None.

Dealing With Missing Outcome Data in Randomized Trials and Observational Studies

Although missing outcome data are an important problem in randomized trials and observational studies, methods to address this issue can be difficult to apply. Using simulated data, the authors compared 3 methods to handle missing outcome data: 1) complete case analysis; 2) single imputation; and 3) multiple imputation (all 3 with and without covariate adjustment). Simulated scenarios focused on continuous or dichotomous missing outcome data from randomized trials or observational studies. When outcomes were missing at random, single and multiple imputations yielded unbiased estimates after covariate adjustment. Estimates obtained by complete case analysis with covariate adjustment were unbiased as well, with coverage close to 95%. When outcome data were missing not at random, all methods gave biased estimates, but handling missing outcome data by means of 1 of the 3 methods reduced bias compared with a complete case analysis without covariate adjustment. Complete case analysis with covariate adjustment and multiple imputation yield similar estimates in the event of missing outcome data, as long as the same predictors of missingness are included. Hence, complete case analysis with covariate adjustment can and should be used as the analysis of choice more often. Multiple imputation, in addition, can accommodate the missing-not-at-random scenario more flexibly, making it especially suited for sensitivity analyses.

Overestimation of risk ratios by odds ratios in trials and cohort studies: alternatives to logistic regression

Logistic regression analysis, which estimates odds ratios, is often used to adjust for covariables in cohort studies and randomized controlled trials (RCTs) that study a dichotomous outcome. In case–control studies, the odds ratio is the appropriate effect estimate, and the odds ratio can

Performance of the original EuroSCORE

The European system for cardiac operative risk evaluation (EuroSCORE) is a commonly used risk score for operative mortality following cardiac surgery. We aimed to conduct a systematic review of the performance of the additive and logistic EuroSCORE. A literature search resulted in 67 articles. Studies applying the EuroSCORE on patients undergoing cardiac surgery and which reported early mortality were included. Weighted meta-regression showed that the EuroSCORE overestimated mortality. However, this performance depended on the risk profile of patients: in high-risk patients, the additive model actually underestimated mortality. Discriminative performance was good. Given the poor predictive performance, the EuroSCORE may not be suitable as a tool for patient selection nor for benchmarking.

17α-hydroxyprogesterone caproate for the prevention of adverse neonatal outcome in multiple pregnancies: a randomized controlled trial.

OBJECTIVE: To estimate whether administration of 17&agr;-hydroxyprogesterone caproate can prevent neonatal morbidity in multiple pregnancies by reducing the preterm birth rate. METHODS: We conducted a multicenter, double-blind, placebo-controlled randomized trial in 55 obstetric clinics in the Netherlands. Women with a multiple pregnancy were randomized to weekly injections of either 250 mg 17&agr;-hydroxyprogesterone caproate or placebo, starting between 16 and 20 weeks of gestation and continuing until 36 weeks of gestation. The main outcome measure was adverse neonatal outcome. Secondary outcome measures were gestational age at delivery and delivery before 28, 32, and 37 weeks of gestation. RESULTS: We randomized 671 women. A composite measure of adverse neonatal outcome was present in 110 children (16%) born to mothers in the 17&agr;-hydroxyprogesterone caproate group, and in 80 children (12%) of mothers in the placebo group (relative risk [RR] 1.34; 95% confidence interval [CI] 0.95–1.89). The mean gestational age at delivery was 35.4 weeks for the 17&agr;-hydroxyprogesterone caproate group and 35.7 weeks for the placebo group (P=.32). Treatment with 17&agr;-hydroxyprogesterone caproate did not reduce the delivery rate before 28 weeks (6% in the 17&agr;-hydroxyprogesterone caproate group compared with 5% in the placebo group, RR 1.04; 95% CI 0.56–1.94), 32 weeks (14% compared with 10%, RR 1.37; 95% CI 0.91–2.05), or 37 weeks of gestation (55% compared with 50%, RR 1.11; 95% CI 0.97–1.28). CONCLUSION: 17&agr;-hydroxyprogesterone caproate does not prevent neonatal morbidity or preterm birth in multiple pregnancies. CLINICAL TRIAL REGISTRATION: ISRCTN Register, www.isrctn.org, ISRCTN40512715. LEVEL OF EVIDENCE: I

Poor Quality of Reporting Confounding Bias in Observational Intervention Studies : A Systematic Review

PURPOSE To systematically review observational studies on medical interventions to determine the quality of reporting of confounding. METHODS Articles on observational studies on medical interventions in five general medical journals and five epidemiological journals published between January 2004 and April 2007 were systematically reviewed. All relevant items pertaining to confounding bias were scored for each article. The overall quality of reporting was determined with an 8-point score. RESULTS The MEDLINE search resulted in 2993 publications, and 174 (5.8%) articles were included in the analysis. In the majority of studies (>98%), the potential for confounding bias was reported. Details on the selection and inclusion of observed confounders were reported in 10% and 51%, respectively. The potential for unobserved confounding was reported in 60%, and 9% commented on the potential effect of such remaining confounding. The quality of reporting of confounding score was mediocre (a median score of 4 points; interquartile range 3 to 5), and scores were similar in all years. CONCLUSION The quality of reporting of confounding in articles on observational medical intervention studies was poor. However, the STROBE statement for reporting of observational studies may considerably impact the reporting of such studies.

Sensitivity analyses to estimate the potential impact of unmeasured confounding in causal research

BACKGROUND The impact of unmeasured confounders on causal associations can be studied by means of sensitivity analyses. Although several sensitivity analyses are available, these are used infrequently. This article is intended as a tutorial on sensitivity analyses, in which we discuss three methods to conduct sensitivity analysis. METHODS Each method is based on assumed associations between confounder and exposure, confounder and outcome and the prevalence of the confounder in the population at large. In the first method an unmeasured confounder is simulated and subsequently adjusted. The other two methods are analytical methods, in which either the (adjusted) effect estimate is multiplied with a factor based on assumed confounder characteristics, or the (adjusted) risks for the outcome among exposed and unexposed subjects are adjusted by such a factor. These methods are illustrated with a clinical example on influenza vaccine effectiveness. RESULTS When applied to a dataset constructed to assess the effect of influenza vaccination on mortality, the three reviewed methods provided similar results. After adjustment for observed confounders, influenza vaccination reduced mortality by 42% [odds ratio (OR) 0.58, 95% confidence interval (CI) 0.46-0.73]. To arrive at a 95% CI including one requires a very common confounder (40% prevalence) with strong associations with both vaccination status and mortality, respectively OR < or =0.3 and OR > or =3.0 (OR 0.79, 95% CI 0.62-1.00). CONCLUSIONS In every non-randomized study on causal associations the robustness of the results with respect to unmeasured confounding can, and should, be assessed using sensitivity analyses.

Quantitative assessment of unobserved confounding is mandatory in nonrandomized intervention studies

OBJECTIVE In nonrandomized intervention studies unequal distribution of patient characteristics in the groups under study may hinder comparability of prognosis and therefore lead to confounding bias. Our objective was to review methods to control for observed confounding, as well as unobserved confounding STUDY DESIGN AND SETTING We reviewed epidemiologic literature on methods to control for observed and unobserved confounding. RESULTS Various methods are available to control for observed (i.e., measured) confounders, either in the design of data collection (i.e., matching, restriction), or in data analysis (i.e., multivariate analysis, propensity score analysis). Methods to quantify unobserved confounding can be categorized in methods with and without prior knowledge of the effect estimate. Without prior knowledge of the effect estimate, unobserved confounding can be quantified using different types of sensitivity analysis. When prior knowledge is available, the size of unobserved confounding can be estimated directly by comparison with prior knowledge. CONCLUSION Unobserved confounding should be addressed in a quantitative way to value the inferences of nonrandomized intervention studies.

Measuring balance and model selection in propensity score methods.

Propensity score (PS) methods focus on balancing confounders between groups to estimate an unbiased treatment or exposure effect. However, there is lack of attention in actually measuring, reporting and using the information on balance, for instance for model selection. We propose to use a measure for balance in PS methods and describe several of such measures: the overlapping coefficient, the Kolmogorov‐Smirnov distance, and the Lévy distance.

Get Real in Individual Participant Data (IPD) Meta-Analysis: A Review of the Methodology.

Individual participant data (IPD) meta‐analysis is an increasingly used approach for synthesizing and investigating treatment effect estimates. Over the past few years, numerous methods for conducting an IPD meta‐analysis (IPD‐MA) have been proposed, often making different assumptions and modeling choices while addressing a similar research question. We conducted a literature review to provide an overview of methods for performing an IPD‐MA using evidence from clinical trials or non‐randomized studies when investigating treatment efficacy. With this review, we aim to assist researchers in choosing the appropriate methods and provide recommendations on their implementation when planning and conducting an IPD‐MA.