Search

Mika Kivimäki initiated the Individual Participant Data (IPD) Meta-Analysis Consortium, which currently has 50 members. The Consortium recently published several research reports on the relationship between job strain (high psychological demands and low decision latitude at work), on the one hand, and cardiovascular disease and its risk factors, on the other hand. Since IPD represents a novel way to conduct epidemiological research collaboration and as some of the findings from the IPD Consortium have been criticized, this commentary aims to address the rationales behind the approach and discuss some of the main criticisms of the Consortium. Researchers must tackle many problems when interpreting associations between a psychosocial work environment factor and a health outcome. First of all, work environment factors belong to a “distal” rather than a “proximal” group. In other words, the closer one gets to a biological mechanism relevant for disease development, the more likely it is that a relevant association will be strong. For instance, small samples are needed to establish an individual “brain” factor associated with depression or emotional exhaustion – simply because the brain factor is more or less depression. Factors related to work organization, on the other hand, are more “distal” since there are many factors that influence the relationship between the environment and the body’s organs. Accordingly, it is sometimes difficult to obtain sufficient statistical power for the establishment of an undisputable association. For instance, the long “distance” between job strain and the outcome, myocardial infarction (MI), explains why we should expect a weaker association than in the study of “proximal” factors, for instance myocardial metabolism in relation to MI. Nevertheless, on a societal level, job strain is very important since it affects many working people, with a prevalence in the working population (in the IPD Consortium study’s operational definition) of around 15%. Accordingly, if an unequivocal association is established, it is of major importance to those responsible for work organization and interventions designed to improve working conditions. However, large samples are needed to establish unequivocal proof. Since Karasek introduced his demand–control model (1), there have been many studies of the association between job strain and risk of MI. These studies have become increasingly sophisticated over the years. In addition, there is accumulated indirect evidence from longitudinal studies of the relationship between job strain, on the one hand, and blood pressure variations and endocrine, metabolic, and immunological parameters, on the other hand. The results from these studies give us a plausible physiological explanation of the assumed relationship between job strain and MI risk. A reason why this research field has attracted strong attention is that MI is an undisputable illness outcome. The study of MI risk, therefore, serves as a good scientific model for studying the relationship between job strain and adverse health outcomes in general. Establishment of and rationale behind the IPD Consortium There have been divided opinions about the relationship between job strain and risk of MI. The main reason for the controversy has been that, despite the relatively large size of several of the published cohort studies with number of observation years often in the range of 50 000, the statistical power has been too small for an unequivocal establishment of an association. As a result, Mika Kivimäki invited a number researchers, who had included psychosocial job factors in their study protocols and had or had not published results on the relationship between job strain and MI risk, to establish the IPD Consortium. Including unpublished cohorts was important as this provided a possibility to address the problem of publication bias – the tendency of researchers and journals to publish only positive findings that can lead to inflated associations. The IPD Consortium received financial support from research foundations in the UK, Sweden, Denmark, France, and Finland. The process started with the goal of testing the hypothesis that job strain is associated with MI risk. Research was divided into several stages, the first of which was to scrutinize questions regarding job strain in the participating cohorts. As expected, there were differences in formulations and response categories and also differences in the number of questions related to psychological demands and decision latitude. However, methodological work overseen by “judges” assessing the similarity between items in the different cohort questionnaires made it possible to establish which cohort questionnaires should be included as well as the minimum number of questions for each dimension in the total IPD study with acceptable precision for the two variables (ie, psychological demands and decision latitude). This was done and the results published (2) before the next phase started. Since the job strain variables in the different studies had been homogenized, it was possible to establish the median values for both demands and decision latitude within each cohort. According to the literature’s most common operationalization of job strain, the group with demand above median and, concomitantly decision latitude below median, was defined as the exposed group and the remaining population as the non-exposed. This enabled all individuals (almost 200 000) to be combined in a large cohort study. The average follow-up time was 7.5 years. The assessment of standard risk factors was treated in the same way – criteria were established before the analyses of the association between job strain and MI risk began. Two published articles (3, 4) described the scientific process in the IPD Consortium. Published findings The findings of the IPD Consortium have been published in The Lancet (5) and the Canadian Medical Association Journal (CMAJ) (6). They show a consistent and statistically significant age- and sex-adjusted relationship between job strain and MI risk, which remains even after further adjustments for country, socioeconomic position, and standard risk factors. The findings also show that exclusion of subjects with a short time lag between job strain assessment and onset of MI does not affect the statistical significance of the association – addressing the problem described above with possible vague pre-heart disease illness symptoms possibly influencing the job description. The findings also addressed a second question that had been raised in previous research (7): the risk associated with a combination of high psychological demands and low decision latitude was greater than the risks associated with each one of these two exposures. Finally, the IPD papers provided a response to the question regarding publication bias: Yes, there seemed to be some effect of publication bias since the odds ratio was lower in the unpublished versus published studies. However, even in the unpublished studies, there was a statistically significant relationship between job strain and subsequent MI risk. Criticism of the IPD findings Accordingly, the IPD Consortium has delivered clear responses to several of the questions that have been debated ever since Karasek introduced the model. The problem created by a “distal” relationship was solved by means of strict homogenization of the assessments in many cohorts so they could be used as one study. The findings were positive despite many conservative measures safeguarding against factors that could give rise to inflated relationships. Nevertheless, the IPD Consortium has been criticized for several reasons (8–10). The most critical voices have come from researchers involved in work-intervention research. The critical points can be divided into two groups. Standard risk factors’ role in the relationship between job strain and MI The first and most severe criticism relates to the way in which the role of the standard risk factors for heart disease (diabetes, high blood pressure, smoking, body mass index) has been presented in The Lancet and CMAJ. The latter article was based only on those cohorts that had full information about relevant lifestyle factors (102 000 participants), and the empirical conclusion has not been debated: job strain adds independently to illness risk even when standard risk factors have been taken into account. And, conversely, standard risk factors add independently to risk regardless of job strain. However, the independent association between job strain and MI risk is much weaker than the corresponding association between the standard risk factors and MI. No one could criticize the IPD Consortium for publishing that finding in itself. However, the practical interpretation formulated in The Lancet, “Our findings suggest that prevention of workplace stress might decrease disease incidence; however, this strategy would have a much smaller effect than would tackling of standard risk factors, such as smoking” (5, p1491) has not been endorsed by all members of the IPD. In my mind, the conclusion goes beyond the analyses of the IPD study. I tried to change these formulations but of course in a large group of researchers (46 authors for The Lancet article) one has to compromise, and it becomes practically difficult to discuss changes particularly in the final phase before publication. Critics of the IPD Consortium argue this statement could be used as a justification among clinicians (cardiologists, occupational physicians, and general practitioners) for disregarding the patient’s psychosocial working conditions entirely. So what is the role of standard risk factors in the relationship between job strain and MI risk? In its studies (11, 12), the Consortium has shown that job strain is significantly associated with diabetes, physical inactivity, and obesity as well as to a summarized Framingham risk factor score. According to a recent systematic review (13), job strain is associated with elevated blood pressure, one of the classical standard risk factors. There are divided opinions regarding this relationship with regard to blood pressure assessed in the classical way (at rest in the “doctor’s office”) but the relationship is more established for blood pressure that has been automatically monitored during normal daily activities (14). As has been pointed out by several authors, long lasting excessive stress may increase the risk of metabolic syndrome, which in itself increases heart disease risk. In addition, excessive intake of carbohydrates and cigarette smoking could be regarded as “external” efforts that an individual could make when the body’s own energy mobilization is failing. Accordingly, part of the relationships between job strain and lifestyle factors may mediate the relationship between job strain and MI risk. Thus, the relationships between standard risk factors and job strain make it very difficult to interpret analyses of job strain effects of risk after adjustment for such factors since some of the effect of job strain goes through them. Adjusting for lifestyle factors is valuable, but one should avoid far-reaching conclusions from the results due to this complexity. There is also a difference between the effect of job strain on different parts of the “ladder of lifestyle risk factors”. A closer look at the CMAJ table describing the relationship between number of lifestyle risk factors in groups with and without job strain reveals that there is indeed very little added risk going from “no job strain” to “job strain” (from 2.62 to 2.69) when participants (14 000 participants) have at least two of the lifestyle risk factors. But for those with no lifestyle risk factors (N=55 000) and those with only one such risk factor (N=33 000), job strain increases the risk from 1.00 (reference group) to 1.27 and from 1.47 to 1.87, respectively. From a population perspective, we should pay attention to these risk increases in the lower end of the lifestyle risk ladder. The prevalence of cigarette smoking varies vastly between countries. Therefore smoking is a risk factor, for example, with less potential payoff for cardiovascular prevention in Sweden than in many other countries since there are already few smokers in the risk ages in Sweden. Accordingly, it is difficult to make generalized statements regarding the possible pay-off of the potential effects of lifestyle versus work environment interventions. Working conditions and lifestyle The most important argument concerning working conditions and life style, however, is based upon clinical experience: employees with a poor work environment may be unwilling to follow lifestyle health promotion advice. In contrast, when subjects feel that their working conditions improve they may feel more motivated to follow individual health promotion advice. Admittedly there is little scientific hard core evidence for this. However, of relevance is my research group’s observation in a controlled job-intervention study (a year-long program for improvement of psychosocial know-how among managers) showing that, after a year’s intervention, decision authority had developed statistically significantly more favorably among the intervention group’s employees (reporting to managers in the experimental group). At the same time, the morning plasma cortisol concentration (related to the metabolic syndrome) as well as the plasma concentration of the liver enzyme gamma glutamyl transferase and (among women only) the serum concentration of triglycerides had improved among employees in the intervention but not the control group. These findings indicate that an improved psychosocial work environment may contribute to an improved risk factor profile among the employees (15). Sorensen has discussed the effect of occupational class on smoking habits and its potential role for anti-smoking propaganda (16). It should also be pointed out that – with the use of terminology introduced above – lifestyle factors should be regarded as “more proximal” (thus likely to be associated with stronger direct associations with heart disease risk) in relation to the illness processes than psychosocial work environment factors. Studies showing that subjects with job strain who have suffered a MI are more likely than others during follow-up to experience a new cardiac event (17–19) illustrate that job strain is of importance to heart disease. The interrelationships between work environment and lifestyle risk factors should be studied preferably with more sophisticated statistical models. Karasek has more recently expanded his demand–control theory and discussed the complex nature of these relationships (20). He points out that the key factor in relationships that gives rise to long-term stress reactions at work is the control of storage and release of energy in the task of maintaining stable physiological self-regulation. He notes that in order to resolve current scientific dilemmas, one should ideally study multiple and linked levels of both proximal (even more proximal than we currently study) and more distal (for instance work organization and global economy) factors. Karasek and collaborators (21) have recently described the design of a possible large-scale study that would take these intermediary mechanisms into account. According to the IPD Consortium’s article in the CMAJ, “For many people, avoidance of stress at work is unrealistic. The absence of strong evidence for effective interventions to reduce job strain therefore raises the challenge of identifying additional approaches for dealing with the health impact of stress in the workplace” (6, p764). This has caused tension among the readership. In mass media and comments from cardiologists following the IPD publications, this seems to give the impression that organizational job interventions have no scientific basis (although this was emphatically not stated in the articles). While it is true that large-scale controlled intervention studies aimed at reducing job strain on a collective organizational level, with subsequent follow-up of health effects providing strong evidence for an effect on heart disease risk, have not been published, there are still publications indicating this could be worth a closer look. For a review, see La Montagne et al (22). Bourbonnais et al’s study (23) showed beneficial three-year effects of a work organization improvement program on the mental health of employees, with no similar change in the control group. Another study by Bond and Bunce (24) randomizing six offices to intervention and control groups, respectively, showed that mental health, sick leave, and self-rated productivity improved after one year in the intervention but not control group; this study also showed that increased perception of control was the mediating factor. In the study that my own group (15) performed on the effects of improved management on employee health, parameters related to the metabolic syndrome and decision authority reportedly improved for the employees in the intervention but not control group. Strength of the association between job strain and MI The second area of criticism leveled at the IPD findings relates to statements about the strength of the association between job strain and MI risk. It is true that in some of the IPD texts (for instance in the abstract of The Lancet article), the formulations regarding the studied associations were expanded to work stress in general. This is clearly a mistake since the analysis presented was on the effect of job strain – which is only one aspect of workplace stress. The IPD researchers are involved in research on other kinds of workplace stress, such as effort–reward imbalance (25), job insecurity (26) organizational justice (27), and leadership (28, 29), and the IPD Consortium plans to add exposures (effort–reward imbalance, job insecurity, and long working hours) to its study of job strain. This will enable the group to examine whether different kinds of psychosocial adverse factors compound risk. There are indications of this in previous literature (30, 31). Another question arising in the critique is whether adding health outcomes to one another would provide a more useful analysis of the total impact of the psychosocial work environment. The IPD Consortium has already published a study showing that job strain is associated with increased diabetes risk (11), while an ongoing study examines whether job strain is prospectively related to the risk of stroke. A study of job strain in relation to depression is also being planned. Therefore, in the future, we will see studies from the IPD group addressing the extent to which an adverse work environment (with several exposures) adds to the risk of developing any of these illnesses that are metabolically related to one another. Critical methodological questions have been made regarding the effect of the job strain assessment in the IPD study. A careful analysis of the questionnaires in the different cohort studies was performed and decisions were made regarding “sufficient homogeneity”. The statistical analyses showed that little information was lost with the applied criteria. However, it could still be that the standard Job Content Questionnaire which has been the international standard for job strain assessment is a better questionnaire than the methodological compromise that we achieved. Precision may have been lost due to this and, according to my understanding, it is more likely that this has given rise to random error than systematic overestimation of risk. Another important point is that there has only been one assessment of job strain (at start of follow-up) in the IPD study. In a previous study, Kivimäki et al (32) have shown that stable job strain lasting through at least two measurements with an interval of two years is a stronger predictor than one measurement alone. People change jobs and stop working. Accordingly, precision in predictions will improve when there are several longitudinal measurements. An additional problem is that subjects with the worst job conditions may not participate in this type of study. This is a methodological problem that is not only confined to the IPD Consortium studies. “Overstressed” subjects may have difficulties participating in our studies for practical reasons (irregular work exposure for instance). Also, those who have already developed a disease or illness caused by adverse working conditions will not be included. We know too little about the effects of this kind of non-participation. There is reason to believe that it may give rise to an underestimation of true risks, as discussed by Collins et al (33). Another limitation is that our study represents a Western European perspective. In the European countries participating in our study, work democracy has been an important topic in societal discussions for several decades. It is possible that the relationships would be different in Southern Asian, South American, and African countries for instance. The IPD Consortium also examines health outcomes other than cardiovascular disease. The first publication in this group (34) focused on colorectal, lung, breast, and prostate cancers. No significant associations were found between job strain and the incidence of these forms of cancer. Concluding remarks The IPD Consortium has great potential to provide solid knowledge in the field of psychosocial factors and health. IPD represents a new collaborative approach. Participating researchers have to search ways of homogenizing their exposure and outcome assessments in order make it possible to form very big cohorts in which every participant contributes equally. This is an improvement on the usual meta-analysis procedure. A possible drawback of the need to compromise is that the final number of questions that can be used is more limited than in the original study cohorts, possibly decreasing precision. Another issue is the complexity of managing collaboration between so many researchers. The IPD Consortium has presented good opportunities for everyone to provide input during the production of articles, but under time constraints (eg, the final phase of The Lancet article) it is difficult, if not impossible, to include opinions from 46 coauthors. Perhaps a formalized structure would have helped (eg, a small group could be democratically elected by all the participating researchers to act on their behalf in “critical situations”). The fact that there is an independent relationship between job strain and MI risk already provides an important rationale for employers to deal with psychosocial stress, regardless of the size of the association – simply because MI is a serious illness which gives rise to suffering, risk of death, and productivity loss both for the company and society as a whole. It should be pointed out that there are a number of other valid reasons for the employer to deal with employee stress, not the least financial ones (see, for instance, 35). Psychosocial stress at work may cause much more financial damage than employers know. The IPD Consortium will, in the near future, examine the total effect of job strain, job insecurity, and effort–reward imbalance on important health outcomes that to some extent have a shared etiology (depression, diabetes, stroke, and MI). One critique of the IPD Consortium is that our strategy may lead to over-interpretation of “bits and pieces” of findings. From that point of view, a joint analysis of all the exposures and outcomes already from the start would have been better. However, the IPD Consortium has worked very rapidly and effectively with its scientific process. Transparency has been key in its empirical analysis. A requirement that publications would have had to include all the possible exposures and outcomes in one analysis from start would have delayed the process considerably (possibly for several years) and may not have been feasible given the financing terms. However, the current intense debates put even more pressure on the scientific community to join forces to unravel the “proximal/distal” controversy. The IPD Consortium has shown an interesting example of the strong potential of scientific collaboration.

Background: Preterm birth is the most common cause of death and harm to newborn babies. Babies that are born early may have difficulties at birth and experience health problems during early childhood. Despite extensive study, there is still uncertainty about the effectiveness of progestogen (medications that are similar to the natural hormone progesterone) in preventing or delaying preterm birth, and in improving birth outcomes. The Evaluating Progestogen for Prevention of Preterm birth International Collaborative (EPPPIC) project aims to reduce uncertainty about the specific conditions in which progestogen may (or may not) be effective in preventing or delaying preterm birth and improving birth outcomes. Methods: The design of the study involves international collaborative individual participant data meta-analysis comprising systematic review, re-analysis, and synthesis of trial datasets. Inclusion criteria are as follows: randomized controlled trials comparing progestogen versus placebo or nonintervention, or comparing different types of progestogen, in asymptomatic women at risk of preterm birth. Main outcomes are as follows; fetal/infant death, preterm birth or fetal death (<=37 weeks, <=34 weeks, <= 28 weeks), serious neonatal complications or fetal/infant death, neurosensory disability (measured at 18 months or later) or infant/child death, important maternal morbidity, or maternal death. In statistical methods, IPD will be synthesized across trials using meta-analysis. Both 'two-stage' models (where effect estimates are calculated for each trial and subsequently pooled in a meta-analysis) and 'one-stage' models (where all IPD from all trials are analyzed in one step, while accounting for the clustering of participants within trials) will be used. If sufficient suitable data are available, a network meta-analysis will compare all types of progesterone and routes of administration extending the one-stage models to include multiple treatment arms. Discussion: EPPPIC is an international collaborative project being conducted by the forming EPPPIC group, which includes trial investigators, an international secretariat, and the research project team. Results, which are intended to contribute to improvements in maternal and child health, are expected to be publicly available in mid 2018. [ABSTRACT FROM AUTHOR]

Background: Demands are increasingly being made for clinical trialists to actively share individual participant data (IPD) collected from clinical trials using responsible methods that protect the confidentiality and privacy of clinical trial participants. Clinical trialists, particularly those receiving public funding, are often concerned about the additional time and money that data-sharing activities will require, but few published empirical data are available to help inform these decisions. We sought to evaluate the activity and resources required to prepare anonymised IPD from a clinical trial in anticipation of a future data-sharing request.Methods: Data from two UK publicly funded clinical trials were used for this exercise: 2437 participants with epilepsy recruited from 90 hospital outpatient clinics in the SANAD trial and 146 children with neuro-developmental problems recruited from 18 hospitals in the MENDS trial. We calculated the time and resources required to prepare each anonymised dataset and assemble a data pack ready for sharing.Results: The older SANAD trial (published 2007) required 50 hours of staff time with a total estimated associated cost of £3185 whilst the more recently completed MENDS trial (published 2012) required 39.5 hours of staff time with total estimated associated cost of £2540.Conclusions: Clinical trial researchers, funders and sponsors should consider appropriate resourcing and allow reasonable time for preparing IPD ready for subsequent sharing. This process would be most efficient if prospectively built into the standard operational design and conduct of a clinical trial. Further empirical examples exploring the resource requirements in other settings is recommended.Trial Registration: SANAD: International Standard Randomised Controlled Trials Registry: ISRCTN38354748 . Registered on 25 April 2003.Mends: EU Clinical Trials Register Eudract 2006-004025-28 . Registered on 16 May 2007. International Standard Randomised Controlled Trials Registry: ISRCTN05534585 /MREC 07/MRE08/43. Registered on 26 January 2007. [ABSTRACT FROM AUTHOR]

Aims Epidemiological evidence indicates a protective effect of light to moderate alcohol consumption compared to non-drinking and heavy drinking. Although several mechanisms have been suggested, the effect of alcohol on atherosclerotic changes in vessel walls is unclear. Therefore, we explored the relationship between alcohol consumption and common carotid intima media thickness, a marker of early atherosclerosis in the general population. Methods Individual participant data from eight cohorts, involving 37,494 individuals from the USE-IMT collaboration were used. Multilevel age and sex adjusted linear regression models were applied to estimate mean differences in common carotid intima-media thickness (CIMT) with alcohol consumption. Results The mean age was 57.9 years (SD 8.6) and the mean CIMT was 0.75 mm (SD 0.177). About, 40.5% reported no alcohol consumed, and among those who drank, mean consumption was 13.3 g per day (SD 16.4). Those consuming no alcohol or a very small amount (<5 g per day) had significantly lower common CIMT values than those consuming >10 g per day, after adjusting for a range of confounding factors. Conclusion In this large CIMT consortium, we did not find evidence to support a protective effect of alcohol on CIMT. [ABSTRACT FROM AUTHOR]

Background: Cigarette smoking is associated with a lower risk of Parkinson's disease. It is unclear what constituent of tobacco smoke may lower the risk. Use of Swedish moist smokeless tobacco (snus) can serve as a model to disentangle what constituent of tobacco smoke may lower the risk. The aim of this study was to determine whether snus use was associated with a lower risk of Parkinson's disease.Methods: Individual participant data were collected from seven prospective cohort studies, including 348 601 men. We used survival analysis with multivariable Cox regression to estimate study-specific relative risk of Parkinson's disease due to snus use, and random-effects models to pool estimates in a meta-analysis. The primary analyses were restricted to never-smokers to eliminate the potential confounding effect of tobacco smoking.Results: During a mean follow-up time of 16.1 years, 1199 incident Parkinson's disease cases were identified. Among men who never smoked, ever-snus users had about 60% lower Parkinson's disease risk compared with never-snus users [pooled hazard ratio (HR) 0.41, 95% confidence interval (CI) 0.28-0.61]. The inverse association between snus use and Parkinson's disease risk was more pronounced in current (pooled HR 0.38, 95% CI 0.23-0.63), moderate-heavy amount (pooled HR 0.41, 95% CI 0.19-0.90) and long-term snus users (pooled HR 0.44, 95% CI 0.24-0.83).Conclusions: Non-smoking men who used snus had a substantially lower risk of Parkinson's disease. Results also indicated an inverse dose-response relationship between snus use and Parkinson's disease risk. Our findings suggest that nicotine or other components of tobacco leaves may influence the development of Parkinson's disease. [ABSTRACT FROM AUTHOR]

We construct a framework for meta-analysis and other multi-level data structures that codifies the sources of heterogeneity between studies or settings in treatment effects and examines their implications for analyses. The key idea is to consider, for each of the treatments under investigation, the subject's potential outcome in each study or setting were he to receive that treatment. We consider four sources of heterogeneity: (1) response inconsistency, whereby a subject's response to a given treatment would vary across different studies or settings, (2) the grouping of nonequivalent treatments, where two or more treatments are grouped and treated as a single treatment under the incorrect assumption that a subject's responses to the different treatments would be identical, (3) nonignorable treatment assignment, and (4) response-related variability in the composition of subjects in different studies or settings. We then examine how these sources affect heterogeneity/homogeneity of conditional and unconditional treatment effects. To illustrate the utility of our approach, we re-analyze individual participant data from 29 randomized placebo-controlled studies on the cardiovascular risk of Vioxx, a Cox-2 selective nonsteroidal anti-inflammatory drug approved by the FDA in 1999 for the management of pain and withdrawn from the market in 2004. [ABSTRACT FROM AUTHOR]

Meta-analysis using individual participant data (IPD) obtains and synthesises the raw, participant-level data from a set of relevant studies. The IPD approach is becoming an increasingly popular tool as an alternative to traditional aggregate data meta-analysis, especially as it avoids reliance on published results and provides an opportunity to investigate individual-level interactions, such as treatment-effect modifiers. There are two statistical approaches for conducting an IPD meta-analysis: one-stage and two-stage. The one-stage approach analyses the IPD from all studies simultaneously, for example, in a hierarchical regression model with random effects. The two-stage approach derives aggregate data (such as effect estimates) in each study separately and then combines these in a traditional meta-analysis model. There have been numerous comparisons of the one-stage and two-stage approaches via theoretical consideration, simulation and empirical examples, yet there remains confusion regarding when each approach should be adopted, and indeed why they may differ. In this tutorial paper, we outline the key statistical methods for one-stage and two-stage IPD meta-analyses, and provide 10 key reasons why they may produce different summary results. We explain that most differences arise because of different modelling assumptions, rather than the choice of one-stage or two-stage itself. We illustrate the concepts with recently published IPD meta-analyses, summarise key statistical software and provide recommendations for future IPD meta-analyses. © 2016 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd. [ABSTRACT FROM AUTHOR]

Background: It is widely accepted and acknowledged that data harmonization is crucial: in its absence, the co-analysis of major tranches of high quality extant data is liable to inefficiency or error. However, despite its widespread practice, no formalized/systematic guidelines exist to ensure high quality retrospective data harmonization.Methods: To better understand real-world harmonization practices and facilitate development of formal guidelines, three interrelated initiatives were undertaken between 2006 and 2015. They included a phone survey with 34 major international research initiatives, a series of workshops with experts, and case studies applying the proposed guidelines.Results: A wide range of projects use retrospective harmonization to support their research activities but even when appropriate approaches are used, the terminologies, procedures, technologies and methods adopted vary markedly. The generic guidelines outlined in this article delineate the essentials required and describe an interdependent step-by-step approach to harmonization: 0) define the research question, objectives and protocol; 1) assemble pre-existing knowledge and select studies; 2) define targeted variables and evaluate harmonization potential; 3) process data; 4) estimate quality of the harmonized dataset(s) generated; and 5) disseminate and preserve final harmonization products.Conclusions: This manuscript provides guidelines aiming to encourage rigorous and effective approaches to harmonization which are comprehensively and transparently documented and straightforward to interpret and implement. This can be seen as a key step towards implementing guiding principles analogous to those that are well recognised as being essential in securing the foundational underpinning of systematic reviews and the meta-analysis of clinical trials. [ABSTRACT FROM AUTHOR]

Prefrontal transcranial direct current stimulation (tDCS) shows promise as an effective treatment for depression. However, factors influencing treatment and the time-course of symptom improvements remain to be elucidated. Individual participant data was collected from ten randomised controlled trials of tDCS in depression. Depressive symptom scores were converted to a common scale, and a linear mixed effects individual growth curve model was fit to the data using k-fold cross-validation to prevent overfitting. Data from 576 participants were analysed (tDCS: n = 311; sham: n = 265), of which 468 were unipolar and 108 had bipolar disorder. tDCS effect sizes reached a peak at approximately 6 weeks, and continued to diverge from sham up to 10 weeks. Significant predictors associated with worse response included higher baseline depression severity, treatment resistance, and those associated with better response included bipolar disorder and anxiety disorder. Our findings suggest that longer treatment courses, lasting at least 6 weeks in duration, may be indicated. Further, our results show that tDCS is effective for depressive symptoms in bipolar disorder. Compared to unipolar depression, participants with bipolar disorder may require additional maintenance sessions to prevent rapid relapse. • tDCS antidepressant effects peak at approximately 6 weeks. • Anxiety, benzodiazepine use, and baseline severity predicted mood outcomes. • Bipolar participants show a different response trajectory to unipolar participants. [ABSTRACT FROM AUTHOR]

Background: The association between cannabis use and positive symptoms in schizophrenia spectrum disorders is well documented, especially via meta-analyses. Yet, findings are inconsistent regarding negative symptoms, while other dimensions such as disorganization, depression, and excitement, have not been investigated. In addition, meta-analyses use aggregated data discarding important confounding variables which is a source of bias., Methods: PubMed, ScienceDirect and PsycINFO were used to search for publications from inception to September 27, 2022. We contacted the authors of relevant studies to extract raw datasets and perform an Individual Participant Data meta-analysis (IPDMA). Inclusion criteria were: psychopathology of individuals with schizophrenia spectrum disorders assessed by the Positive and Negative Syndrome Scale (PANSS); cannabis-users had to either have a diagnosis of cannabis use disorder or use cannabis at least twice a week. The main outcomes were the PANSS subscores extracted via the 3-factor (positive, negative and general) and 5-factor (positive, negative, disorganization, depression, excitement) structures. Preregistration is accessible via Prospero: ID CRD42022329172., Findings: Among the 1149 identified studies, 65 were eligible and 21 datasets were shared, totaling 3677 IPD and 3053 complete cases. The adjusted multivariate analysis revealed that relative to non-use, cannabis use was associated with higher severity of positive dimension (3-factor: Adjusted Mean Difference, aMD = 0.34, 95% Confidence Interval, CI = [0.03; 0.66]; 5-factor: aMD = 0.38, 95% CI = [0.08; 0.63]), lower severity of negative dimension (3-factor: aMD = -0.49, 95% CI [-0.90; -0.09]; 5-factor: aMD = -0.50, 95% CI = [-0.91; -0.08]), higher severity of excitement dimension (aMD = 0.16, 95% CI = [0.03; 0.28]). No association was found between cannabis use and disorganization (aMD = -0.13, 95% CI = [-0.42; 0.17]) or depression (aMD = -0.14, 95% CI = [-0.34; 0.06])., Interpretation: No causal relationship can be inferred from the current results. The findings could be in favor of both a detrimental and beneficial effect of cannabis on positive and negative symptoms, respectively. Longitudinal designs are needed to understand the role of cannabis is this association. The reported effect sizes are small and CIs are wide, the interpretation of findings should be taken with caution., Funding: This research did not receive any specific grant or funding. Primary financial support for authors was provided by Le Vinatier Psychiatric Hospital., Competing Interests: OA received consulting fees from Cortechs. ai, and payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing or educational events from Janssen, Sunovion, Lundbeck (all to him personally). EF received consulting fees; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing or educational events; support for attending meetings and/or travel; participation on a Data Safety Monitoring Board or Advisory Board from Boehringer-Ingelheim, Janssen, Lundbeck, Otsuka. TW received honoraria for speakership by ROVI, Janssen-Cilag, Recordati; support for participation in APA congress in 2023 by ROVI; participation on a Data Safety Monitoring Board or Advisory Board for DFG sponsored studies, NicStim, Target Flame, Early member in DSMB, member in advisory board of Otsuka/Lundbeck, Janssen-Cilag and ROVI. JA declared support for the present manuscript through payments made to the University of Calgary from NIMH, from 2008 to 2014. OA declared support for the present manuscript through payments made to institution from KG Jebsen Stiftelsen, Research Council Norway, EU H2020 and Horiz Europe, NIMH, Regional Health Authorities, and NordForsk. MS declared support for the present manuscript through initial data collection supported by Grant No. 02.001033 from the Swiss Federal Office of Public Health. RR declared support for the present manuscript from Canada First Research Excellence Fund, Grant doe Healthy Brains for Healthy Lives; Fonds de recherche du Québec – Santé. No other disclosures were reported. All other authors declare no competing interests., (© 2023 The Authors.)

Individual participant-data meta-analysis (IPDMA) uses data at the participant level to synthesize evidence on clinical decision-making. We discuss the differences from aggregate-data meta-analysis and the advantages and challenges of IPDMA. IPDMA provides an opportunity to study effect modifiers at the participant level and can reduce bias in comparison to aggregate-data meta-analysis. A challenge with IPDMA is that it often requires considerable effort to obtain all the data, and an inability to obtain data for all the studies can lead to availability bias. PATIENT SUMMARY: Systematic reviews of the literature are performed to summarize available evidence regarding clinical decisions. Here we explain the differences between reviews that use aggregated data from published studies and reviews that use the participant-level data from those studies., (Copyright © 2023 The Author(s). Published by Elsevier B.V. All rights reserved.)

Observational data provide invaluable real-world information in medicine, but certain methodological considerations are required to derive causal estimates. In this systematic review, we evaluated the methodology and reporting quality of individual-level patient data meta-analyses (IPD-MAs) published in 2009, 2014, and 2019 that sought to estimate a causal relationship in medicine. We screened over 16,000 titles and abstracts, reviewed 45 full-text articles out of the 167 deemed potentially eligible, and included 29 into the analysis. Unfortunately, we found that causal methodologies were rarely implemented, and reporting was generally poor across studies. Specifically, only three of the 29 articles used quasi-experimental methods, and no study used G-methods to adjust for time-varying confounding. To address these issues, we propose stronger collaborations between physicians and methodologists to ensure that causal methodologies are properly implemented in IPD-MAs. In addition, we put forward a suggested checklist of reporting guidelines for IPD-MAs that utilize causal methods. This checklist could improve reporting thereby potentially enhancing the quality and trustworthiness of IPD-MAs, which can be considered one of the most valuable sources of evidence for health policy., Competing Interests: Competing interests Authors declare that they have no competing interests.

Objective: Identify independent and novel risk factors for late-preterm (28-36 weeks) and term (≥37 weeks) stillbirth and explore development of a risk-prediction model., Design: Secondary analysis of an Individual Participant Data (IPD) meta-analysis investigating modifiable stillbirth risk factors., Setting: An IPD database from five case-control studies in New Zealand, Australia, the UK and an international online study., Population: Women with late-stillbirth (cases, n = 851), and ongoing singleton pregnancies from 28 weeks' gestation (controls, n = 2257)., Methods: Established and novel risk factors for late-preterm and term stillbirth underwent univariable and multivariable logistic regression modelling with multiple sensitivity analyses. Variables included maternal age, body mass index (BMI), parity, mental health, cigarette smoking, second-hand smoking, antenatal-care utilisation, and detailed fetal movement and sleep variables., Main Outcome Measures: Independent risk factors with adjusted odds ratios (aOR) for late-preterm and term stillbirth., Results: After model building, 575 late-stillbirth cases and 1541 controls from three contributing case-control studies were included. Risk factor estimates from separate multivariable models of late-preterm and term stillbirth were compared. As these were similar, the final model combined all late-stillbirths. The single multivariable model confirmed established demographic risk factors, but additionally showed that fetal movement changes had both increased (decreased frequency) and reduced (hiccoughs, increasing strength, frequency or vigorous fetal movements) aOR of stillbirth. Poor antenatal-care utilisation increased risk while more-than-adequate care was protective. The area-under-the-curve was 0.84 (95% CI 0.82-0.86)., Conclusions: Similarities in risk factors for late-preterm and term stillbirth suggest the same approach for risk-assessment can be applied. Detailed fetal movement assessment and inclusion of antenatal-care utilisation could be valuable in late-stillbirth risk assessment., (© 2023 The Authors. BJOG: An International Journal of Obstetrics and Gynaecology published by John Wiley & Sons Ltd.)

Background: Brief motivational interventions (BMIs) are one of the most effective individually focused alcohol intervention strategies for college students. Despite the central theoretical role of motivation for change in BMIs, it is unclear whether BMIs increase motivation to change drinking behavior. We conducted a two-step meta-analysis of individual participant data (IPD) to examine whether BMIs increase motivation for change. N = 5903;59% women, 72% White) from Project INTEGRATE. The BMIs included individually delivered motivational interviewing with personalized feedback (MI + PF), stand-alone personalized feedback (PF), and group-based motivational interviewing (GMI)., Methods: We included 15 trials of BMI (N = 5903;59% women, 72% White) from Project INTEGRATE. The BMIs included individually-delivered motivational interviewing with personalized feedback (MI + PF), stand-alone personalized feedback (PF), and group-based motivational interviewing (GMI). Different measures and responses used in the original trials were harmonized. Effect size estimates were derived from a model that adjusted for baseline motivation and demographic variables for each trial (step 1) and subsequently combined in a random-effects meta-analysis (step 2)., Results: The overall intervention effect of BMIs on motivation for change was not statistically significant (standard mean difference [SMD]: 0.026, 95% CI: [-0.001, 0.053], p = 0.06, k = 19 comparisons). Of the three subtypes of BMIs, GMI, which tended to provide motivation-targeted content, had a statistically significant intervention effect on motivation, compared with controls (SMD: 0.055, 95% CI: [0.007, 0.103], p = 0.025, k = 5). By contrast, there was no evidence that MI + PF (SMD = 0.04, 95% CI: [-0.02, 0.10], k = 6, p = 0.20) nor PF increased motivation (SMD = 0.005, 95% CI: [-0.028, 0.039], k = 8, p = 0.75), compared with controls. Post hoc meta-regression analysis suggested that motivation sharply decreased each month within the first 3 months postintervention (b = -0.050, z = -2.80, p = 0.005 for k = 14)., Conclusions: Although BMIs provide motivational content and normative feedback and are assumed to motivate behavior change, the results do not wholly support the hypothesis that BMIs improve motivation for change. Changing motivation is difficult to assess during and following interventions, but it is still a theoretically important clinical endpoint. Further, the evidence cautiously suggests that changing motivation may be achievable, especially if motivation-targeted content components are provided., (© 2023 The Authors. Alcohol: Clinical and Experimental Research published by Wiley Periodicals LLC on behalf of Research Society on Alcohol.)