Why Most Published Research Findings Are False

Why there are so many contradicted or exaggerated findings in highly cited clinical research?

It is not uncommon that clinical studies of the same intervention contradicted with each other, e.g., one study produced positive results, while the other produced negative results. Ioanndis (2005a) found that among 49 highly-cited original clinical research studies, published in New England Journal of Medicine, Journal of the American Medical Association, Lancet or in a high-impact medical specialty journal, 32% of them were contradicted in subsequent large-scale studies, or were shown to have potentially overestimated the efficacy of the experimental intervention. This finding is disturbing and of serious concern given the widespread impact of these highly-cited studies and the rigorous standards used to design and conduct the studies. We perform Bayesian analysis of these highly-cited clinical studies based on Bayesian factor. We identified one cause of the issue: p values strongly overstated the experimental evidence. For the highly-cited studies, when the p value was 0.05, there was a 74.4% percentage chance that the null hypothesis was true. The use of a p value of 0.05 as the criterion for significance caused many researchers to mistakenly draw conclusions of positive findings, which were then contradicted by subsequent large-scale studies.

Patient Safety Incidents and Adverse Events in Ambulatory Dental Care: A Systematic Scoping Review

BACKGROUND

There have been efforts to understand the epidemiology of iatrogenic harm in hospitals and primary care and to improve the safety of care provision. There has in contrast been very limited progress in relation to the safety of ambulatory dental care.

OBJECTIVES

To provide a comprehensive overview of the range and frequencies of existing evidence on patient safety incidents and adverse events in ambulatory dentistry.

METHODS

We searched MEDLINE and EMBASE for articles reporting events that could have or did result in unnecessary harm in ambulatory dental care. We extracted and synthesized data on the types and frequencies of patient safety incidents and adverse events.

RESULTS

Forty articles were included. We found that the frequencies varied very widely between studies; this reflected differences in definitions, populations studied, and sampling strategies. The main 5 PSIs we identified were errors in diagnosis and examination, treatment planning, communication, procedural errors, and the accidental ingestion or inhalation of foreign objects. However, little attention was paid to wider organizational issues.

CONCLUSIONS

Patient safety research in dentistry is immature because current evidence cannot provide reliable estimates on the frequency of patient safety incidents in ambulatory dental care or the associated disease burden. Well-designed epidemiological investigations are needed that also investigate contributory factors.

Dissociable psychosocial profiles of adolescent substance users

OBJECTIVE

Alcohol, tobacco and cannabis use in adolescence is associated with adverse outcomes. Characterizing adolescent substance misusers, however, is difficult due to the wide range of risk and protective factors linked to substance use. The aim of the present study was to examine the role of the Individual, Family, School, Peer, and Social Environment on alcohol (lifetime and risky), tobacco (risky only), and cannabis use (lifetime and riskiness).

METHOD

Data were analyzed from a national sample of 5,680 adolescents, capturing substance use behavior alongside risk and protective factors across Individual, Family, School, Peer and Social domains. We applied a sophisticated machine learning classifier to develop models of alcohol, tobacco and cannabis initiation and misuse.

RESULTS

We found highly accurate (area under curve of receiver-operator-characteristic for out-of-sample performance was > .88) and replicable (over multiple iterations and in comparison with permuted outcomes) dissociable psychosocial profiles of alcohol, tobacco and cannabis use. Alongside common predictors (peer relations and externalizing behavior), dissociable risk and resilience factors were observed. Adolescent profiles of alcohol use were distinguished by the contribution of multiple domains. In contrast, tobacco use was characterized by a small number of individual variables, including female gender and poor perceived academic position. Cannabis use was differentiated by the distinct contribution of Individual risk factors, in particular male gender and feelings of anger. Differential associations were also evident, with the strength and direction of association differing substantially across substances.

CONCLUSION

This study indicates that the relationship between the environment and substance use is more complex than previously thought.

Using Standardized Test Scores to Include General Cognitive Ability in Education Research and Policy

In education research and education policy, much attention is paid to schools, curricula, and teachers, but little attention is paid to the characteristics of students. Differences in general cognitive ability (g) are often overlooked as a source of important variance among schools and in outcomes among students within schools. Standardized test scores such as the SAT and ACT are reasonably good proxies for g and are available for most incoming college students. Though the idea of g being important in education is quite old, we present contemporary evidence that colleges and universities in the United States vary considerably in the average cognitive ability of their students, which correlates strongly with other methods (including international methods) of ranking colleges. We also show that these g differences are reflected in the extent to which graduates of colleges are represented in various high-status and high-income occupations. Finally, we show how including individual-level measures of cognitive ability can substantially increase the statistical power of experiments designed to measure educational treatment effects. We conclude that education policy researchers should give more consideration to the concept of individual differences in cognitive ability as well as other factors.

The comparative evidence basis for the efficacy of second-generation antidepressants in the treatment of depression in the US: A Bayesian meta-analysis of Food and Drug Administration reviews

BACKGROUND

Studies have shown similar efficacy of different antidepressants in the treatment of depression.

METHOD

Data of phase-2 and -3 clinical-trials for 16 antidepressants (levomilnacipran, desvenlafaxine, duloxetine, venlafaxine, paroxetine, escitalopram, vortioxetine, mirtazapine, venlafaxine XR, sertraline, fluoxetine, citalopram, paroxetine CR, nefazodone, bupropion, vilazodone), approved by the FDA for the treatment of depression between 1987 and 2016, were extracted from the FDA reviews that were used to evaluate efficacy prior to marketing approval, which are less liable to reporting biases. Meta-analytic Bayes factors, which quantify the strength of evidence for efficacy, were calculated. In addition, posterior pooled effect-sizes were calculated and compared with classical estimations.

RESULTS

The resulted Bayes factors showed that the evidence load for efficacy varied strongly across antidepressants. However, all tested drugs except for bupropion and vilazodone showed strong evidence for their efficacy. The posterior effect-size distributions showed variation across antidepressants, with the highest pooled estimated effect size for venlafaxine followed by paroxetine, and the lowest for bupropion and vilazodone.

LIMITATIONS

Not all published trials were included in the study.

CONCLUSIONS

The results illustrate the importance of considering both the effect size and the evidence-load when judging the efficacy of a treatment. In doing so, the currently employed Bayesian approach provided clear insights on top of those gained with traditional approaches.

Do "Brain-Training" Programs Work?

In 2014, two groups of scientists published open letters on the efficacy of brain-training interventions, or "brain games," for improving cognition. The first letter, a consensus statement from an international group of more than 70 scientists, claimed that brain games do not provide a scientifically grounded way to improve cognitive functioning or to stave off cognitive decline. Several months later, an international group of 133 scientists and practitioners countered that the literature is replete with demonstrations of the benefits of brain training for a wide variety of cognitive and everyday activities. How could two teams of scientists examine the same literature and come to conflicting "consensus" views about the effectiveness of brain training?In part, the disagreement might result from different standards used when evaluating the evidence. To date, the field has lacked a comprehensive review of the brain-training literature, one that examines both the quantity and the quality of the evidence according to a well-defined set of best practices. This article provides such a review, focusing exclusively on the use of cognitive tasks or games as a means to enhance performance on other tasks. We specify and justify a set of best practices for such brain-training interventions and then use those standards to evaluate all of the published peer-reviewed intervention studies cited on the websites of leading brain-training companies listed on Cognitive Training Data (www.cognitivetrainingdata.org), the site hosting the open letter from brain-training proponents. These citations presumably represent the evidence that best supports the claims of effectiveness.Based on this examination, we find extensive evidence that brain-training interventions improve performance on the trained tasks, less evidence that such interventions improve performance on closely related tasks, and little evidence that training enhances performance on distantly related tasks or that training improves everyday cognitive performance. We also find that many of the published intervention studies had major shortcomings in design or analysis that preclude definitive conclusions about the efficacy of training, and that none of the cited studies conformed to all of the best practices we identify as essential to drawing clear conclusions about the benefits of brain training for everyday activities. We conclude with detailed recommendations for scientists, funding agencies, and policymakers that, if adopted, would lead to better evidence regarding the efficacy of brain-training interventions.

Four reasons to prefer Bayesian analyses over significance testing

Inference using significance testing and Bayes factors is compared and contrasted in five case studies based on real research. The first study illustrates that the methods will often agree, both in motivating researchers to conclude that H1 is supported better than H0, and the other way round, that H0 is better supported than H1. The next four, however, show that the methods will also often disagree. In these cases, the aim of the paper will be to motivate the sensible evidential conclusion, and then see which approach matches those intuitions. Specifically, it is shown that a high-powered non-significant result is consistent with no evidence for H0 over H1 worth mentioning, which a Bayes factor can show, and, conversely, that a low-powered non-significant result is consistent with substantial evidence for H0 over H1, again indicated by Bayesian analyses. The fourth study illustrates that a high-powered significant result may not amount to any evidence for H1 over H0, matching the Bayesian conclusion. Finally, the fifth study illustrates that different theories can be evidentially supported to different degrees by the same data; a fact that P-values cannot reflect but Bayes factors can. It is argued that appropriate conclusions match the Bayesian inferences, but not those based on significance testing, where they disagree.

Wo sind die wissenschaftlichen Standards für hochwertige Replikationsforschung?

Zusammenfassung. Es gibt einen breiten Konsens, dass Replikation ein wichtiges Instrument ist, um valide Befunde und solide Forschung zu erkennen. Wenn sie aber wissenschaftlich bedeutsam ist, dann muss auch die Replikationsforschung an strengen methodischen Regeln und an klar artikulierten wissenschaftlichen Zielen gemessen werden. Eine kritische Beschäftigung mit der aktuellen Replikationsforschung – etwa im jüngst veröffentlichten Bericht der Open Science Collaboration – zeigt jedoch, dass eine strenge und forschungslogisch begründete Methodologie für Replikationsstudien bislang weder angewandt noch entwickelt wurde. Infolgedessen bleibt die Validität der Schlüsse, die aus Replikationsstudien gezogen werden dürfen, oftmals unklar. Dieses grundlegende Problem wird hier unter vier Gesichtspunkten diskutiert: Unklarheit des Gegenstandes der Replikation (Replicandum), Vernachlässigung einschlägiger methodischer Probleme (Regressivität; Reliabilität der Veränderungsmessung), einseitige Vermeidung von angeblich kostenträchtigen „Falsch-Positiven“ ohne Versuch einer systematischen Kosten-Nutzen-Messung sowie das vernachlässigte Ziel, Replikationsforschung so zu implementieren, dass sie echte Erkenntnisfortschritte bringt und als exzellente Forschung anerkannt werden kann.

Can infants learn phonology in the lab? A meta-analytic answer

Two of the key tasks facing the language-learning infant lie at the level of phonology: establishing which sounds are contrastive in the native inventory, and determining what their possible syllabic positions and permissible combinations (phonotactics) are. In 2002-2003, two theoretical proposals, one bearing on how infants can learn sounds (Maye, Werker, & Gerken, 2002) and the other on phonotactics (Chambers, Onishi, & Fisher, 2003), were put forward on the pages of Cognition, each supported by two laboratory experiments, wherein a group of infants was briefly exposed to a set of pseudo-words, and plausible phonological generalizations were tested subsequently. These two papers have received considerable attention from the general scientific community, and inspired a flurry of follow-up work. In the context of questions regarding the replicability of psychological science, the present work uses a meta-analytic approach to appraise extant empirical evidence for infant phonological learning in the laboratory. It is found that neither seminal finding (on learning sounds and learning phonotactics) holds up when close methodological replications are integrated, although less close methodological replications do provide some evidence in favor of the sound learning strand of work. Implications for authors and readers of this literature are drawn out. It would be desirable that additional mechanisms for phonological learning be explored, and that future infant laboratory work employ paradigms that rely on constrained and unambiguous links between experimental exposure and measured infant behavior.

Risk Factors Associated with Blood Exposure for Sporadic Hepatitis E in Dhaka, Bangladesh

Fecal contamination of drinking water is associated with large hepatitis E virus (HEV) outbreaks of genotypes 1 and 2 in endemic areas. Sporadic transmission of HEV genotypes 3 and 4 in high-income countries has been associated with exposure to blood and animal contact. The objective of the study was to identify the risk factors for hepatitis E and the genotype(s) causing sporadic hepatitis E in Dhaka, Bangladesh. We selected, from a diagnostic center in Dhaka between November 2008 and November 2009, cases presenting with jaundice and anti-HEV IgM antibodies and age-matched controls were defined as those with no history of jaundice and normal blood test results. Serum samples were tested for HEV RNA using real-time reverse transcriptase polymerase chain reaction followed by a sequencing and phylogenetic analysis. A total of 109 cases and 109 controls were enrolled. The cases were more likely to be male (adjusted matched odds ratios [mOR] 2.2, 95% CI: 1.2-3.9; P = 0.01), or have reported contact with another person's blood or blood product, or contact with blood-contaminated sharp instruments (adjusted mOR 2.1, 95% CI: 1.1-4.1; P = 0.03) than controls. There were no significant differences between the cases and the controls in terms of reported high-risk sexual intercourse, consumption of undercooked meat, or contact or drinking fecally-contaminated water. The sera from three cases carried HEV RNA, all belonging to genotype 1. Findings from this study suggest that contact with human blood and sharing sharp instruments may transmit sporadic hepatitis E in Dhaka, Bangladesh. Efforts to prevent the transmission of blood-borne pathogens may also prevent sporadic HEV transmission in this endemic setting.

Gene Set Enrichment Analyses: lessons learned from the heart failure phenotype

BACKGROUND

Genetic studies for complex diseases have predominantly discovered main effects at individual loci, but have not focused on genomic and environmental contexts important for a phenotype. Gene Set Enrichment Analysis (GSEA) aims to address this by identifying sets of genes or biological pathways contributing to a phenotype, through gene-gene interactions or other mechanisms, which are not the focus of conventional association methods.

RESULTS

Approaches that utilize GSEA can now take input from array chips, either gene-centric or genome-wide, but are highly sensitive to study design, SNP selection and pruning strategies, SNP-to-gene mapping, and pathway definitions. Here, we present lessons learned from our experience with GSEA of heart failure, a particularly challenging phenotype due to its underlying heterogeneous etiology.

CONCLUSIONS

This case study shows that proper data handling is essential to avoid false-positive results. Well-defined pipelines for quality control are needed to avoid reporting spurious results using GSEA.

Research Review: The effectiveness of multidimensional family therapy in treating adolescents with multiple behavior problems - a meta-analysis

BACKGROUND

Multidimensional family therapy (MDFT) is a well-established treatment for adolescents showing both substance abuse and/or antisocial behavior.

METHOD

The effectiveness of MDFT in reducing adolescents' substance abuse, delinquency, externalizing and internalizing psychopathology, and family malfunctioning was examined by means of a (three-level) meta-analysis, summarizing 61 effect sizes from 19 manuscripts (N = 1,488 participants).

RESULTS

Compared with other therapies, the overall effect size of MDFT was significant, albeit small in magnitude (d = 0.24, p < .001), and similar across intervention outcome categories. Moderator analysis revealed that adolescents with high severity problems, including severe substance abuse and disruptive behavior disorder, benefited more from MDFT than adolescents with less severe conditions.

CONCLUSIONS

It can be concluded that MDFT is effective for adolescents with substance abuse, delinquency, and comorbid behavior problems. Subsequently, it is important to match specific characteristics of the adolescents, such as extent of impairment, with MDFT.

The Need for Formal Evidence Synthesis in Food Policy: A Case Study of Willingness-to-Pay

Meta-analysis is increasingly utilised in the understanding of consumer behaviour, including in relation to farm animal welfare. However, the issue of publication bias has received little attention. As willingness-to-pay (WTP) is widely used in policy, it is important to explore publication bias. This research aimed to evaluate publication bias in WTP, specifically public WTP for farm animal welfare. A systematic review of four databases yielded 54 studies for random effects meta-analysis. Publication bias was assessed by the Egger test, rank test, contour-enhanced funnel plots, and the Vevea and Hedges weight-function model. Results consistently indicated the presence of publication bias, highlighting an overestimation of WTP for farm animal welfare. Stakeholders should be wary of WTP estimates that have not been critically evaluated for publication bias.

Beyond the Genetic Information Nondiscrimination Act: ethical and economic implications of the exclusion of disability, long-term care and life insurance

The Genetic Information Nondiscrimination Act (GINA) was intended to protect individuals in the USA from discrimination based on their genetic data, but does not apply to life, long-term care or disability insurance. Patient advocates and ethicists have argued that GINA does not go far enough. Others express concerns for the viability of insurance companies if millions of potential customers know more than professional actuaries. Here we discuss the exclusion of certain insurance types from GINA. We explore the ethical and economic implications of this distinction, and potential paths forward. We suggest that because long-term care and disability insurance can be essential for well-being, there is no good reason to place them in a class with life insurance and therefore beyond GINA's reach.

What Constitutes Strong Psychological Science? The (Neglected) Role of Diagnosticity and A Priori Theorizing

A Bayesian perspective on Ioannidis’s (2005) memorable statement that “Most Published Research Findings Are False” suggests a seemingly inescapable trade-off: It appears as if research hypotheses are based either on safe ground (high prior odds), yielding valid but unsurprising results, or on unexpected and novel ideas (low prior odds), inspiring risky and surprising findings that are inevitably often wrong. Indeed, research of two prominent types, sexy hypothesis testing and model testing, is often characterized by low priors (due to astounding hypotheses and conjunctive models) as well as low-likelihood ratios (due to nondiagnostic predictions of the yin-or-yang type). However, the trade-off is not inescapable: An alternative research approach, theory-driven cumulative science, aims at maximizing both prior odds and diagnostic hypothesis testing. The final discussion emphasizes the value of pluralistic science, within which exploratory phenomenon-driven research can play a similarly strong part as strict theory-testing science.

Toward evidence-based medical statistics: a Bayesian analysis of double-blind placebo-controlled antidepressant trials in the treatment of anxiety disorders

The Food and Drug Administration (FDA) uses a p < 0.05 null-hypothesis significance testing framework to evaluate "substantial evidence" for drug efficacy. This framework only allows dichotomous conclusions and does not quantify the strength of evidence supporting efficacy. The efficacy of FDA-approved antidepressants for the treatment of anxiety disorders was re-evaluated in a Bayesian framework that quantifies the strength of the evidence. Data from 58 double-blind placebo-controlled trials were retrieved from the FDA for the second-generation antidepressants for the treatment of anxiety disorders. Bayes factors (BFs) were calculated for all treatment arms compared to placebo and were compared with the corresponding p-values and the FDA conclusion categories. BFs ranged from 0.07 to 131,400, indicating a range of no support of evidence to strong evidence for the efficacy. Results also indicate a varying strength of evidence between the trials with p < 0.05. In sum, there were large differences in BFs across trials. Among trials providing "substantial evidence" according to the FDA, only 27 out of 59 dose groups obtained strong support for efficacy according to the typically used cutoff of BF ≥ 20. The Bayesian framework can provide valuable information on the strength of the evidence for drug efficacy. Copyright © 2016 John Wiley & Sons, Ltd.

Promises and pitfalls of Web-based experimentation in the advance of replicable psychological science: A reply to Plant (2015)

In a recent letter, Plant (2015) reminded us that proper calibration of our laboratory experiments is important for the progress of psychological science. Therefore, carefully controlled laboratory studies are argued to be preferred over Web-based experimentation, in which timing is usually more imprecise. Here we argue that there are many situations in which the timing of Web-based experimentation is acceptable and that online experimentation provides a very useful and promising complementary toolbox to available lab-based approaches. We discuss examples in which stimulus calibration or calibration against response criteria is necessary and situations in which this is not critical. We also discuss how online labor markets, such as Amazon's Mechanical Turk, allow researchers to acquire data in more diverse populations and to test theories along more psychological dimensions. Recent methodological advances that have produced more accurate browser-based stimulus presentation are also discussed. In our view, online experimentation is one of the most promising avenues to advance replicable psychological science in the near future.

Weighing Evidence "Steampunk" Style via the Meta-Analyser

The funnel plot is a graphical visualization of summary data estimates from a meta-analysis, and is a useful tool for detecting departures from the standard modeling assumptions. Although perhaps not widely appreciated, a simple extension of the funnel plot can help to facilitate an intuitive interpretation of the mathematics underlying a meta-analysis at a more fundamental level, by equating it to determining the center of mass of a physical system. We used this analogy to explain the concepts of weighing evidence and of biased evidence to a young audience at the Cambridge Science Festival, without recourse to precise definitions or statistical formulas and with a little help from Sherlock Holmes! Following on from the science fair, we have developed an interactive web-application (named the Meta-Analyser) to bring these ideas to a wider audience. We envisage that our application will be a useful tool for researchers when interpreting their data. First, to facilitate a simple understanding of fixed and random effects modeling approaches; second, to assess the importance of outliers; and third, to show the impact of adjusting for small study bias. This final aim is realized by introducing a novel graphical interpretation of the well-known method of Egger regression.

How Modeling Standards, Software, and Initiatives Support Reproducibility in Systems Biology and Systems Medicine

OBJECTIVE

Only reproducible results are of significance to science. The lack of suitable standards and appropriate support of standards in software tools has led to numerous publications with irreproducible results. Our objectives are to identify the key challenges of reproducible research and to highlight existing solutions.

RESULTS

In this paper, we summarize problems concerning reproducibility in systems biology and systems medicine. We focus on initiatives, standards, and software tools that aim to improve the reproducibility of simulation studies.

CONCLUSIONS

The long-term success of systems biology and systems medicine depends on trustworthy models and simulations. This requires openness to ensure reusability and transparency to enable reproducibility of results in these fields.

Will big data yield new mathematics? An evolving synergy with neuroscience

New mathematics has often been inspired by new insights into the natural world. Here we describe some ongoing and possible future interactions among the massive data sets being collected in neuroscience, methods for their analysis and mathematical models of the underlying, still largely uncharted neural substrates that generate these data. We start by recalling events that occurred in turbulence modelling when substantial space-time velocity field measurements and numerical simulations allowed a new perspective on the governing equations of fluid mechanics. While no analogous global mathematical model of neural processes exists, we argue that big data may enable validation or at least rejection of models at cellular to brain area scales and may illuminate connections among models. We give examples of such models and survey some relatively new experimental technologies, including optogenetics and functional imaging, that can report neural activity in live animals performing complex tasks. The search for analytical techniques for these data is already yielding new mathematics, and we believe their multi-scale nature may help relate well-established models, such as the Hodgkin-Huxley equations for single neurons, to more abstract models of neural circuits, brain areas and larger networks within the brain. In brief, we envisage a closer liaison, if not a marriage, between neuroscience and mathematics.

Precision-scored parental report questions and HL-scaled tympanometry as informative measures of hearing in otitis media 1: Large-sample evidence on determinants and complementarity to pure-tone audiometry

INTRODUCTION

In otitis media with effusion (OME), hearing loss is a core sign/symptom and basis of concern, with absolute pure-tone threshold sensitivity (in dB HL) by air-conduction providing the default measure of hearing. However several fundamental problems limiting the value of HL measures in otitis media are insufficiently appreciated. To appraise the joint value and implications of multiple hearing measures towards more comprehensive hearing assessment in OM, we examine in two related articles the interrelations and common or diverging determinants of three measures, two of them objective: binaural HL, and ACET (the published quasi-continuous scaling of binaural tympanometry to HL). The third measure is partly subjective: parentally reported hearing difficulties (RHD-4); this is the precision-scored total of the 4 items selected for the OM8-30 general purpose questionnaire for parents in OM.

METHODS

The Eurotitis-2 study (Total N=2886) internationally standardises OM8-30 and its OMQ-14 short form. The clinical and parent-response variables acquired cover many issues in diagnosis, symptomatology and impact of OM. Data acquisition was built upon routine clinic practice, enabling us also to document some properties of that practice, such as patterns of missing HL data. To address possible confounding or loss of representativeness from this, we investigated the implications of substituting tympanometry-based ACET for missing HL to give an HL/ACET hybrid. ACET is the mapping of categorical tympanometry to continuous HL. We simulated degrees of artificial missingness of HL up to 35% on the 1430 complete-data cases, using random deletion, with 1000-version bootstrapping. Correlations of this HL/ACET hybrid with pure (100%) HL then documented the degree of correlation retained under dilution of HL by an admixture of ACET; we also documented distribution shapes. For RHD-4, we then probed the determining influences on severity of score as an auditory disability measure, both background ones (from centre, age, sex, socio-economic status, length of history, diagnosis and season) and the two underlying objective hearing measures (HL, ACET). We ran these multiple regressions (GLMs), for representativeness and generality, both on 1430 complete-data cases (i.e. all 3 hearing variables present) and also on supplemented samples according to data required only for particular analyses (N increased by +56% to +68%). A further method of sample supplementation (by up to +96%) used the HL/ACET hybrid.

RESULTS

Sex made negligible difference in any analysis. The particular collaborating centre, age, season and diagnosis collectively influenced presence/absence of HL data very strongly. (Area under ROC 0.944). Socio-economic status did not influence HL presence; surprisingly, nor did RHD, ACET or length of history, after control for centre, age, diagnosis and season. Of the inter-correlations between hearing measures, only the one between ACET and RHD was influenced (slightly reduced) by the inclusion of cases without HL data. In the simulated substitutions, Pearson correlation of hybrid HL/ACET with true HL remained above 0.90 for substitution by ACET of up to 30% rate of artificially 'missing' HL. Centre differences were adequately summarised by simple absolute additive differences in mean local case severity. In the determinant models for RHD on the 1430 complete-data cases, HL and the set of background determinants collectively explained broadly similar proportions of RHD's variability, totalling 36.8% explained. On the larger maximum case samples, slightly less absolute variability was explicable than on complete-case data, but relative magnitudes of contribution from individual determinants, both background and hearing measures, remained similar. The expected mean differences in RHD between diagnoses (RAOM, OME, and combined) were found, but the patterns of background and objective measure influences determining RHD did not differ significantly between the diagnoses.

CONCLUSIONS

(1) In the Eurotitis-2 database, descriptive differences in various background demographic and clinical measures between cases on whom HL data were obtained versus not, were only of material magnitude for length of history and reported hearing difficulties. Such descriptive differences are not necessarily bases of confounding, so using our framework of 6 background adjuster variables, (particular collaborating centre, age, season, diagnosis, socioeconomic status and length of history) we isolated the determinants of HL data presence. The first four listed strongly predicted HL data presence/absence so are sufficient to control analyses well for any bias or confounding by HL data presence. (2) Diagnoses as OME and combined (OME+RAOM) had higher probability of HL data being present relative to RAOM, indicating that HL acquisition is chiefly seen as confirming and quantifying hearing loss in (suspect) OME, not as ruling it out (e.g. in suspected RAOM). Given this, also using RHD and or ACET as pre-triage to efficiently target capacity and/or reduce costs and opportunity costs of acquiring HL would be rational, but there was no evidence of such precise use of initial hearing-related information to decide on HL acquisition. (3) The full six background variables explained comparable variance in Reported Hearing Difficulties (RHD) to what was explained by ACET, but not quite as much as by HL. Achieving a high percentage explained (32-37% from good models) required both classes of determinant to be entered as predictors. The pattern of background determining influences for RHD was largely stable, with or without objective measures as additional predictors, and on maximum or complete-data cases. Length of history strongly determines RHD for a given concurrent HL. (4) Accepting ACET as substitute where HL was missing in OM cases gave a sample-size enhancement of 17% in Eurotitis-2, with negligible difference in the pattern of determinants. This hybrid measure can be recommended as reasonable next-best when moderate percentages of HL data are missing. (5) The stable pattern of prediction of RHD suggests that our six background determinants provide a very promising low-cost yet comprehensive framework for determination. It hence offers pluripotent statistical adjustment against confounding, applicable to RAOM, OME and combined diagnoses in any analysis using this database. Claims that it thereby offers a sufficient framework for full European standardisation of all the scores from the OM8-30 questionnaire measures await parallel demonstrations for symptom areas other than RHD. As 25% of the variance in RHD severity can be explained by the six adjusters in our framework, none of the six variables should be omitted from acquisition and analytic use in future OM research.

From big data to deep insight in developmental science

The use of the term 'big data' has grown substantially over the past several decades and is now widespread. In this review, I ask what makes data 'big' and what implications the size, density, or complexity of datasets have for the science of human development. A survey of existing datasets illustrates how existing large, complex, multilevel, and multimeasure data can reveal the complexities of developmental processes. At the same time, significant technical, policy, ethics, transparency, cultural, and conceptual issues associated with the use of big data must be addressed. Most big developmental science data are currently hard to find and cumbersome to access, the field lacks a culture of data sharing, and there is no consensus about who owns or should control research data. But, these barriers are dissolving. Developmental researchers are finding new ways to collect, manage, store, share, and enable others to reuse data. This promises a future in which big data can lead to deeper insights about some of the most profound questions in behavioral science.

Driving with pets and motor vehicle collision involvement among older drivers: A prospective population-based study

OBJECTIVE

Distracted driving is a major cause of motor vehicle collision (MVC) involvement. Pets have been identified as potential distraction to drivers, particularly in the front. This type of distraction could be worse for those with impairment in the cognitive aspects of visual processing. The purpose of this study is to evaluate the association between driving with pets and rates of motor vehicle collision involvement in a cohort of older drivers.

METHODS

A three-year prospective study was conducted in a population-based sample of 2000 licensed drivers aged 70 years and older. At the baseline visit, a trained interviewer asked participants about pet ownership, whether they drive with pets, how frequently, and where the pet sits in the vehicle. Motor vehicle collision (MVC) involvement during the three-year study period was obtained from the Alabama Department of Public Safety. At-fault status was determined by the police officer who arrived on the scene. Participants were followed until the earliest of death, driving cessation, or end of the study period. Poisson regression was used to calculate crude and adjusted rate ratios (RR) examining the association between pet ownership, presence of a pet in a vehicle, frequency of driving with a pet, and location of the pet inside with vehicle with any and at-fault MVC involvement. We examined whether the associations differed by higher order visual processing impairment status, as measured by Useful Field of View, Trails B, and Motor-free Visual Perception Test.

RESULTS

Rates of crash involvement were similar for older adults who have ever driven with a pet compared to those who never drove with their pet (RR=1.15, 95% CI 0.76-1.75). Drivers who reported always or sometimes driving with their pet had higher MVC rates compared to pet owners who never drive with a pet, but this association was not statistically significant (RR=1.39, 95% CI 0.86-2.24). In terms of location, those reporting having a pet frequently ride in the front of the vehicle had similar rates of MVC involvement compared to those who did not drive with a pet in the front. A similar pattern of results was observed for at-fault MVCs. This association was not modified by visual processing impairment status.

CONCLUSION

The current study demonstrates a positive but non-significant association between frequently driving with pets and MVC involvement. More research is needed, particularly on restraint use and whether the pet was in the car at the time of the crash, to help characterize the public safety benefit of regulations on driving with pets.