In the context of competing risks data, the subdistribution hazard ratio has limited clinical interpretability to measure treatment effects. An alternative is the difference in restricted mean times ...lost (RMTL), which gives the mean time lost to a specific cause of failure between treatment groups. In non-randomized studies, the average causal effect is conventionally used for decision-making about treatment and public health policies. We show how the difference in RMTL can be estimated by contrasting the integrated cumulative incidence functions from a Fine-Gray model. We also show how the difference in RMTL can be estimated by using inverse probability of treatment weighting and contrasts between weighted non-parametric estimators of the area below the cumulative incidence. We use pseudo-observation approaches to estimate both component models and we integrate them into a doubly-robust estimator. We demonstrate that this estimator is consistent when either component is correctly specified. We conduct simulation studies to assess its finite-sample performance and demonstrate its inherited consistency property from its component models. We also examine the performance of this estimator under varying degrees of covariate overlap and under a model misspecification of nonlinearity. We apply the proposed method to assess biomarker-treatment interaction in subpopulations of the POPLAR and OAK randomized controlled trials of second-line therapy for advanced non-small-cell lung cancer.
The growth in scientific production may threaten the capacity for the scientific community to handle the ever-increasing demand for peer review of scientific publications. There is little evidence ...regarding the sustainability of the peer-review system and how the scientific community copes with the burden it poses. We used mathematical modeling to estimate the overall quantitative annual demand for peer review and the supply in biomedical research. The modeling was informed by empirical data from various sources in the biomedical domain, including all articles indexed at MEDLINE. We found that for 2015, across a range of scenarios, the supply exceeded by 15% to 249% the demand for reviewers and reviews. However, 20% of the researchers performed 69% to 94% of the reviews. Among researchers actually contributing to peer review, 70% dedicated 1% or less of their research work-time to peer review while 5% dedicated 13% or more of it. An estimated 63.4 million hours were devoted to peer review in 2015, among which 18.9 million hours were provided by the top 5% contributing reviewers. Our results support that the system is sustainable in terms of volume but emphasizes a considerable imbalance in the distribution of the peer-review effort across the scientific community. Finally, various individual interactions between authors, editors and reviewers may reduce to some extent the number of reviewers who are available to editors at any point.
We aimed to compare empirically the treatment effects measured by the hazard ratio (HR) and by the difference (and ratio) of restricted mean survival times (RMST) in oncology randomized trials.
We ...selected oncology randomized controlled trials from five leading journals during the last 6 months of 2014. We reconstructed individual patient data for one time-to-event outcome from each trial, preferably the primary outcome. We reanalyzed each trial and compared the treatment effect estimated by the HR with that by the difference (and ratio) of RMST. We estimated an average ratio of the HR to the ratio of RMST; an average ratio less than one indicates more optimistic assessments with HRs.
We analyzed 54 randomized controlled trials totaling 33,212 patients. The selected outcome was overall survival in 21 (39%) trials. There was evidence of nonproportionality of hazards in 13 (24%) trials. The HR and RMST-based measures were in agreement regarding the statistical significance of the effect, except in one case. The median HR was 0.84 (Q1 to Q3 range, 0.67 to 0.97) and the median difference in RMST was 1.12 months (range, 0.22 to 2.75 months). The average ratio of the HR to the ratio of RMST was 1.11 (95% CI, 1.07 to 1.15), with substantial between-trial variability (I(2) = 86%). Results were consistent by outcome type (overall survival v other outcomes) and whether the proportional hazard assumption held or not.
On average, the HR provided significantly larger treatment effect estimates than the ratio of RMST. The HR may seem large when the absolute effect is small. RMST-based measures should be routinely reported in randomized trials with time-to-event outcomes.
A recent study suggested that results of single-center trials are frequently contradicted when similar trials are performed in multicenter settings.
To perform a meta-epidemiologic study to evaluate ...whether estimates of treatment effect differ between single-center and multicenter randomized, controlled trials (RCTs).
MEDLINE was searched via PubMed for meta-analyses of RCTs with binary outcomes that were published between August 2008 and January 2009 and in the first 6 months of 2010 in the 10 leading journals of each medical specialty. One issue of the Cochrane Database of Systematic Reviews was also searched.
All individual RCTs included in the meta-analyses were selected.
Data were extracted and their quality was assessed by use of the risk of bias tool of the Cochrane Collaboration.
The primary outcome was the ratio of odds ratios (ROR), used to quantify the difference in estimated intervention effect between single-center and multicenter RCTs. An ROR less than 1 would indicate larger estimates of the intervention effect in single-center trials. Sensitivity analyses were performed with adjustment for sample size, risk of bias within RCTs, and variance of the log odds ratio to take publication bias into account. Forty-eight meta-analyses were selected, including 421 RCTs (223 were single-center and 198 were multicenter). Single-center RCTs showed a larger intervention effect than did multicenter RCTs (combined ROR, 0.73 95% CI, 0.64 to 0.83), with low heterogeneity across individual meta-analyses (I(2) = 12.0%; P = 0.24). Adjustment for sample size yielded consistent results (ROR, 0.85 CI, 0.74 to 0.97), as did adjustment for risk of bias within RCTs, such as allocation concealment (ROR, 0.76 CI, 0.67 to 0.86), and variance of log odds ratio (ROR, 0.83 CI, 0.72 to 0.96).
Despite sensitivity analyses, meta-confounding cannot be fully excluded.
Single-center RCTs showed larger treatment effects than did multicenter RCTs, a finding that was consistent in all sensitivity analyses. These results suggest that this item should be considered when the results of RCTs and meta-analyses are interpreted.
Academic grant Recherche sur la Recherche from the Délégation Interrégionale à la Recherche Clinique (DIRC), Ile de France, Assistance Publique-Hôpitaux de Paris (APHP).
Although several public health organizations have recommended population-wide reduction in salt intake, the evidence on the population benefits remains unclear. We conducted a metaknowledge analysis ...of the literature on salt intake and health outcomes.
We identified reports--primary studies, systematic reviews, guidelines and comments, letters or reviews--addressing the effect of sodium intake on cerebro-cardiovascular disease or mortality. We classified reports as supportive or contradictory of the hypothesis that salt reduction leads to population benefits, and constructed a network of citations connecting these reports. We tested for citation bias using an exponential random graph model. We also assessed the inclusion of primary studies in systematic reviews on the topic.
We identified 269 reports (25% primary studies, 5% systematic reviews, 4% guidelines and 66% comments, letters, or reviews) from between 1978 and 2014. Of these, 54% were supportive of the hypothesis, 33% were contradictory and 13% were inconclusive. Reports were 1.51 95% confidence interval (CI) 1.38 to 1.65 times more likely to cite reports that drew a similar conclusion, than to cite reports drawing a different conclusion. In all, 48 primary studies were selected for inclusion across 10 systematic reviews. If any given primary study was selected by a review, the probability that a further review would also have selected it was 27.0% (95% CI 20.3% to 33.7%).
We documented a strong polarization of scientific reports on the link between sodium intake and health outcomes, and a pattern of uncertainty in systematic reviews about what should count as evidence.
AbstractObjectiveTo examine the association between risk factor burdens—categorized as optimal, borderline, or elevated—and the lifetime risk of atrial fibrillation.DesignCommunity based cohort ...study.SettingLongitudinal data from the Framingham Heart Study.ParticipantsIndividuals free of atrial fibrillation at index ages 55, 65, and 75 years were assessed. Smoking, alcohol consumption, body mass index, blood pressure, diabetes, and history of heart failure or myocardial infarction were assessed as being optimal (that is, all risk factors were optimal), borderline (presence of borderline risk factors and absence of any elevated risk factor), or elevated (presence of at least one elevated risk factor) at index age.Main outcome measureLifetime risk of atrial fibrillation at index age up to 95 years, accounting for the competing risk of death.ResultsAt index age 55 years, the study sample comprised 5338 participants (2531 (47.4%) men). In this group, 247 (4.6%) had an optimal risk profile, 1415 (26.5%) had a borderline risk profile, and 3676 (68.9%) an elevated risk profile. The prevalence of elevated risk factors increased gradually when the index ages rose. For index age of 55 years, the lifetime risk of atrial fibrillation was 37.0% (95% confidence interval 34.3% to 39.6%). The lifetime risk of atrial fibrillation was 23.4% (12.8% to 34.5%) with an optimal risk profile, 33.4% (27.9% to 38.9%) with a borderline risk profile, and 38.4% (35.5% to 41.4%) with an elevated risk profile. Overall, participants with at least one elevated risk factor were associated with at least 37.8% lifetime risk of atrial fibrillation. The gradient in lifetime risk across risk factor burden was similar at index ages 65 and 75 years.ConclusionsRegardless of index ages at 55, 65, or 75 years, an optimal risk factor profile was associated with a lifetime risk of atrial fibrillation of about one in five; this risk rose to more than one in three a third in individuals with at least one elevated risk factor.
BACKGROUND:The long-term probability of developing atrial fibrillation (AF) considering genetic predisposition and clinical risk factor burden is unknown.
METHODS:We estimated the lifetime risk of AF ...in individuals from the community-based Framingham Heart Study. Polygenic risk for AF was derived using a score of ≈1000 AF-associated single-nucleotide polymorphisms. Clinical risk factor burden was calculated for each individual using a validated risk score for incident AF comprised of height, weight, systolic and diastolic blood pressure, current smoking status, antihypertensive medication use, diabetes mellitus, history of myocardial infarction, and history of heart failure. We estimated the lifetime risk of AF within tertiles of polygenic and clinical risk.
RESULTS:Among 4606 participants without AF at 55 years of age, 580 developed incident AF (median follow-up, 9.4 years; 25th–75th percentile, 4.4–14.3 years). The lifetime risk of AF >55 years of age was 37.1% and was substantially influenced by both polygenic and clinical risk factor burden. Among individuals free of AF at 55 years of age, those in low-polygenic and clinical risk tertiles had a lifetime risk of AF of 22.3% (95% confidence interval, 15.4−9.1), whereas those in high-risk tertiles had a risk of 48.2% (95% confidence interval, 41.3−55.1). A lower clinical risk factor burden was associated with later AF onset after adjusting for genetic predisposition (P<0.001).
CONCLUSIONS:In our community-based cohort, the lifetime risk of AF was 37%. Estimation of polygenic AF risk is feasible and together with clinical risk factor burden explains a substantial gradient in long-term AF risk.
BACKGROUND:The frequency of cardiac rhythm abnormalities and their risk factors in community-dwelling adults are not well characterized.
METHODS:We determined the frequency of rhythm abnormalities in ...the UK Biobank, a national prospective cohort. We tested associations between risk factors and incident rhythm abnormalities using multivariable proportional hazards regression.
RESULTS:Of 502 627 adults (median age, 58 years interquartile range, 13; 54.4% women), 2.35% had a baseline rhythm abnormality. The prevalence increased with age with 4.84% of individuals aged 65 to 73 years affected. During 3 368 332 person-years of follow-up, 15 906 new rhythm abnormalities were detected (4.72 per 1000 person-years; 95% confidence interval CI4.65–4.80). Atrial fibrillation (3.11 per 1000 person-years; 95% CI3.05–3.17), bradyarrhythmias (0.89 per 1000 person-years; 95% CI0.86–0.92), and conduction system diseases (1.06 per 1000 person-years; 95% CI1.02–1.09) were more common than supraventricular (0.51 per 1000 person-years; 95% CI0.48–0.53) and ventricular arrhythmias (0.57 per 1000 person-years; 95% CI0.55–0.60). Older age (hazard ratio HR2.35 per 10-year increase; 95% CI2.29–2.41; P<0.01), male sex (HR1.83; 95% CI1.76–1.89; P<0.01), hypertension (HR1.49; 95% CI1.44–1.54; P<0.01), chronic kidney disease (HR1.95; 95% CI1.67–2.27; P<0.01), and heart failure (HR1.99; 95% CI1.76–2.26; P<0.01) were associated with new rhythm abnormalities.
CONCLUSIONS:The frequency of rhythm abnormalities in middle-aged to older community-dwelling adults is substantial. Atrial fibrillation, bradyarrhythmias, and conduction system diseases account for most rhythm conditions.
Mapping the international landscape of clinical trials may inform global health research governance, but no large-scale data are available. Industry or non-industry sponsorship may have a major ...influence in this mapping. We aimed to map the global landscape of industry- and non-industry-sponsored clinical trials and its evolution over time.
We analyzed clinical trials initiated between 2006 and 2013 and registered in the WHO International Clinical Trials Registry Platform (ICTRP). We mapped single-country and international trials by World Bank's income groups and by sponsorship (industry- vs. non- industry), including its evolution over time from 2006 to 2012. We identified clusters of countries that collaborated significantly more than expected in industry- and non-industry-sponsored international trials.
119,679 clinical trials conducted in 177 countries were analysed. The median number of trials per million inhabitants in high-income countries was 100 times that in low-income countries (116.0 vs. 1.1). Industry sponsors were involved in three times more trials per million inhabitants than non-industry sponsors in high-income countries (75.0 vs. 24.5) and in ten times fewer trials in low- income countries (0.08 vs. 1.08). Among industry- and non-industry-sponsored trials, 30.3% and 3.2% were international, respectively. In the industry-sponsored network of collaboration, Eastern European and South American countries collaborated more than expected; in the non-industry-sponsored network, collaboration among Scandinavian countries was overrepresented. Industry-sponsored international trials became more inter-continental with time between 2006 and 2012 (from 54.8% to 67.3%) as compared with non-industry-sponsored trials (from 42.4% to 37.2%).
Based on trials registered in the WHO ICTRP we documented a substantial gap between the globalization of industry- and non-industry-sponsored clinical research. Only 3% of academic trials but 30% of industry trials are international. The latter appeared to be conducted in preferentially selected countries.
Background/aims
Non-inferiority trials with time-to-event outcomes are becoming increasingly common. Designing non-inferiority trials is challenging, in particular, they require very large sample ...sizes. We hypothesized that the difference in restricted mean survival time, an alternative to the hazard ratio, could lead to smaller required sample sizes.
Methods
We show how to convert a margin for the hazard ratio into a margin for the difference in restricted mean survival time and how to calculate the required sample size under a Weibull survival distribution. We systematically selected non-inferiority trials published between 2013 and 2016 in seven major journals. Based on the protocol and article of each trial, we determined the clinically relevant time horizon of interest. We reconstructed individual patient data for the primary outcome and fit a Weibull distribution to the comparator arm. We converted the margin for the hazard ratio into the margin for the difference in restricted mean survival time. We tested for non-inferiority using the difference in restricted mean survival time and hazard ratio. We determined the required sample size based on both measures, using the type I error risk and power from the original trial design.
Results
We included 35 trials. We found evidence of non-proportional hazards in five (14%) trials. The hazard ratio and the difference in restricted mean survival time were consistent regarding non-inferiority testing, except in one trial where the difference in restricted mean survival time led to evidence of non-inferiority while the hazard ratio did not. The median hazard ratio margin was 1.43 (Q1–Q3, 1.29–1.75). The median of the corresponding margins for the difference in restricted mean survival time was −21 days (Q1–Q3, −36 to −8) for a median time horizon of 2.0 years (Q1–Q3, 1–3 years). The required sample size according to the difference in restricted mean survival time was smaller in 71% of trials, with a median relative decrease of 8.5% (Q1–Q3, 0.4%–38.0%). Across all 35 trials, about 25,000 participants would have been spared from enrollment using the difference in restricted mean survival time compared to hazard ratio for trial design.
Conclusion
The margins for the hazard ratio may seem large but translate to relatively small differences in restricted mean survival time. The difference in restricted mean survival time offers meaningful interpretation and can result in considerable reductions in sample size. Restricted mean survival time-based measures should be considered more widely in the design and analysis of non-inferiority trials with time-to-event outcomes.
PDF
