Active and passive smoking and risk of breast cancer: a meta-analysis.

Studies on active and passive tobacco smoking and breast cancer have found inconsistent results. A meta-analysis of observational studies on tobacco smoking and breast cancer occurrence was conducted based on systematic searches for studies with retrospective (case-control) and prospective (cohort) designs. Eligible studies were identified, and relative risk measurements were extracted for active and passive tobacco exposures. Random-effects meta-analyses were used to compute summary relative risks (SRR). Heterogeneity of results between studies was evaluated using the (I (2)) statistics. For ever active smoking, in 27 prospective studies, the SRR for breast cancer was 1.10 (95 % CI [1.09-1.12]) with no heterogeneity (I (2) = 0 %). In 44 retrospective studies, the SRR was 1.08 (95 % CI [1.02-1.14]) with high heterogeneity (I (2) = 59 %). SRRs for current active smoking were 1.13 (95 % CI [1.09-1.17]) in 27 prospective studies and 1.08 (95 % CI [0.97-1.20]) in 22 retrospective studies. The results were stable across different subgroup analyses, notably pre/post-menopause, alcohol consumption adjustments, including/excluding passive smokers from the referent group. For ever passive smoking, in 11 prospective studies, the SRR for breast cancer was 1.07 (95 % CI [1.02-1.13]) with no heterogeneity (I (2) = 1 %). In 20 retrospective studies, the SRR was 1.30 (95 % CI [1.10-1.54]) with high heterogeneity (I (2) = 74 %). Too few prospective studies were available for meaningful subgroup analyses. There is consistent evidence for a moderate increase in the risk of breast cancer in women who smoke tobacco. The evidence for a moderate increase in risk with passive smoking is more substantial than a few years ago.

Dose-response analysis of toxicological and pharmacological mixtures with the model deviation ratio method: Problems and solutions.

Risk assessment for mixtures of chemicals requires to investigate the magnitude of their potential adverse effects on living organisms. This is usually done by assessing how experimental toxicological mixture data depart from the model of Loewe additivity. Several recent scientific studies propose to perform this task using an ad hoc method known as model deviation ratio (MDR) method. Moreover, the first official European regulatory document for the study of combined exposures explicitly recommends the use of the MDR method (EFSA Scientific Committee et al. Guidance on harmonised methodologies for human health, animal health and ecological risk assessment of combined exposure to multiple chemicals. EFSA Journal, 2019). We show here that the MDR method is not rooted in statistical principles and can lead to erroneous claims. We show however that the distribution of the MDR can be evaluated by simulations and show how this allows us to devise and carry out a bona fide statistical test. The proposed method accounts for uncertainty in the estimation of ED/EC50 and does not require a minimum sample size. The computer code developped in this study is made available as an R package called MDR.

Effect of the epidemiological heterogeneity on the outbreak outcomes.

Multi-host pathogens infect and are transmitted by different kinds of hosts and, therefore, the host heterogeneity may have a great impact on the outbreak outcome of the system. This paper deals with the following problem: consider the system of interacting and mixed populations of hosts epidemiologically different, what would be the outbreak outcome for each host population composing the system as a result of mixing in comparison to the situation with zero mixing? To address this issue we have characterized the epidemic response function for a single-host population and defined a heterogeneity index measuring how host systems are epidemiologically different in terms of generation time, basic reproduction number R0 and, therefore, epidemic response function. Based on the individual epidemiological characteristics of populations, with heterogeneities and mixing affinities, the response of subpopulations in a multi-host system is compared to that of a single-host system. The case of a two-host system, in which the infection transmission depends solely on the infection susceptibility of the receiver, is analyzed in detail. Three types of responses are observed: dilution, amplification or no effect, corresponding to lower, higher or equal attack rates, respectively, for a host population in an interacting multi-host system compared to the zero-mixing situation. We find that no effect is generally observed for zero heterogeneity. A dilution effect is always observed for all the host populations when their individual R0,i < 1. Whereas, when at least one of the individual R0,i > 1, then the hosts ''i'' with R0,i > R0,j undergo a dilution effect while the hosts ''j'' undergo an amplification effect.

Effect of vitamin D supplementation on non-skeletal disorders: a systematic review of meta-analyses and randomised trials.

Randomised trials reported up to Dec 31, 2012, did not confirm that vitamin D supplementation could protect from non-skeletal health conditions affecting adults, as was expected on the basis of data from observational studies. To examine whether the more recently published meta-analyses and trials would change past conclusions, we systematically reviewed meta-analyses of vitamin D supplementation and non-skeletal disorders published between Jan 1, 2013, and May 31, 2017, that included study participants of all ages, including pregnant women. We also searched for randomised trials not included in meta-analyses. We identified 87 meta-analyses, of which 52 were excluded because they contained less recent literature or were of suboptimal quality. We retrieved 202 articles on trials that were not included in meta-analyses. Recent meta-analyses reinforce the finding that 10-20 µg per day of vitamin D can reduce all-cause mortality and cancer mortality in middle-aged and older people. Although vitamin D doses were greater than those assessed in the past, we found no new evidence that supplementation could have an effect on most non-skeletal conditions, including cardiovascular disease, adiposity, glucose metabolism, mood disorders, muscular function, tuberculosis, and colorectal adenomas, or on maternal and perinatal conditions. New data on cancer outcomes were scarce. The compilation of results from 83 trials showed that vitamin D supplementation had no significant effect on biomarkers of systemic inflammation. The main new finding highlighted by this systematic review is that vitamin D supplementation might help to prevent common upper respiratory tract infections and asthma exacerbations. There remains little evidence to suggest that vitamin D supplementation has an effect on most conditions, including chronic inflammation, despite use of increased doses of vitamin D, strengthening the hypothesis that low vitamin D status is a consequence of ill health, rather than its cause. We further hypothesise that vitamin D supplementation could exert immunomodulatory effects that strengthen resistance to acute infections, which would reduce the risk of death in debilitated individuals. We identified many meta-analyses of suboptimal quality, which is of concern. Future systematic reviews on vitamin D should be based on data sharing so that data for participants with the same outcomes measured in the same way can be pooled to generate stronger evidence.