The causes and effects of socio-demographic exclusions from clinical trials
To investigate the exclusion from trials of women, older people and minority ethnic groups, focusing on two drug exemplars, statins and non-steroidal anti-inflammatory drugs (NSAIDs). Medical and ethical databases. Workshops with stakeholders. Literature was reviewed on exclusions in healthcare research and three workshops were held with stakeholders. Twenty-seven randomised controlled trials (RCTs) of statins use for secondary prevention of coronary heart disease (CHD) and 25 NSAIDs trials for pain in osteoarthritis (OA) were analysed. Using a Scottish cohort with record-linkage, profiling was carried out for 3188 people needing secondary prevention for CHD (1993-1996), ascertaining the independent effects of statins, and 131,410 people dispensed NSAIDs (1989-1996), examining adverse effects. Routine data sources were accessed to profile the need for secondary prevention of CHD in England and usage was estimated by consulting published surveys. The Somerset and Avon Survey of Health (SASH) 1996-97 and published data were accessed for information on potential need and usage of NSAIDs in OA. For both drugs, the socio-demographic profiles of trial samples, the population in potential need and those on treatment were compared. An evidence synthesis was produced to clarify the effects of statins on women and older people and the relationship of absolute effectiveness outcomes with underlying risk levels of disease events was modelled, examining the likely effects of trial exclusions. The average age of statins trial participants was 58.5 years; only 16.3% were women. Statins reduced cardiovascular disease (CVD) incidence by about 25% in both men and women. Older people up to 75 years of age also benefited. Meta-analysis and two landmark trials confirmed these results. The average age of NSAIDs trial participants was 61.9 years and women were well represented (68.5%). Gastrointestinal (GI) adverse events were commonly reported, but renal side-effects were not. Outcomes were seldom reported according to socio-demographic group. For both drugs, USA trials were more inclusive than UK/European trials. Ethnicity was not well reported for either drug. Some 23% of the cohort were treated with statins. Users were younger than non-statins users (but no more likely to be male) and had superior outcomes. High current exposure to NSAIDs elevated the risk of GI side-effects by about 50% versus no current exposure and renal impairment risk by nearly 140%. Side-effect risk increased with age; being female diminished risk. Approximately 537,000 incident cases of CVD would qualify for statins use in England each year. Women constitute 45% of this population with need, two-thirds of whom are aged 65 years or over. Need varies by ethnic group. No sex bias in prescribing statins was detected, but use was commoner in younger people. For NSAIDs, 6.3% of adults aged 35+ years reported hip and/or knee pain associated with OA; 3.9% of adults used prescribed analgesics for this and they were more likely to be women and to be >65 years old. For statins, women formed almost half of the 'with need' and 'on treatment' populations, but were markedly under-represented in trials. Those aged 65+ years formed nearly two-thirds of the 'with need' population, but only one-fifth of trial samples, and were less likely to be treated than younger subjects. For NSAIDs, women formed similar proportions. Associations of side-effects with socio-demographic factors was revealed in cohort data but not in trials. The issue of exclusion from trials of women, older people and ethnic minorities has been relatively neglected in the UK research community, and there is confusion about diversity issues. Under-representation occurs, but in drug trials at least this may not always affect the external validity of relative effect estimates. However, measures of absolute effectiveness, absolute harm and cost-effectiveness are associated with underlying risk levels in different socio-demographic groups. Under-representation will therefore bias absolute effect estimates. The following areas are suggested for future research: multi-disciplinary assessment of realistic options for trialists to address the issue of exclusions; clarification of the use of ethnic categories in health research and of the implications of the different dimensions of ageing and sex/gender; identification of barriers and facilitators to the involvement of different population groups in research, further investigation of the susceptibility of older men to NSAID adverse events, and the development of a 'register of registries and databases' and exploration of how linked health information systems in the UK could be improved.
[Show abstract] [Hide abstract] ABSTRACT: Background: Type 2 diabetes is common, on the rise, and disproportionately affects ethnic minority groups. Telehealth interventions may mitigate diabetes-related complications, but might under-recruit or even exclude ethnic minorities, in part because of English language requirements. The under-representation of minority patients in trials could threaten the generalizability of the findings, whereby the patients who might stand to benefit most from such interventions are not being included in their evaluation. Objective: The aims of this systematic review are twofold: (1) to assess the reporting and prevalence of ethnic minorities in published telehealth trials for type 2 diabetes, including identifying trial features associated with successful patient recruitment; and (2) to determine the proportion of such trials that report English language proficiency as an inclusion/exclusion criterion, including how and why they do so. Methods: Randomized controlled trials (RCTs) of adults with type 2 diabetes in Western, English-speaking countries that included telehealth interventions targeting diabetes as a primary condition, and those that did not specifically recruit minority groups will be included. Search strategies were devised for indexed and keyword terms capturing type 2 diabetes, telehealth/health technology, and RCTs in English language publications from 2000 to July 2015 in MEDLINE, PsycINFO, EMBASE, CINAHL, and CENTRAL. Reference lists of included studies will also be searched. Two reviewers will independently screen abstracts and full-text articles against inclusion criteria, mediated by a third reviewer if consensus cannot be reached. Data extracted from included studies will be checked by a second reviewer and will be summarized using narrative synthesis. Results: This research is in progress, with findings expected by Spring 2016. Conclusions: This review will address research reporting and recruitment practices of ethnic minorities in telehealth RCTs for type 2 diabetes. Prevalence estimates will elucidate generalizability of existing research, with implications for researchers, health professionals, and policy makers. Identifying trial or intervention features that appear to facilitate ethnic minority recruitment, as well as language barriers that impede it might suggest ways to improve recruitment in future trials. Trial registration: PROSPERO International Prospective Register of Systematic Reviews: CRD42015024899; http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015024899 (Archived by WebCite at http://www.webcitation.org/6fUMqbJ0f).
- "One frequently cited challenge in reference to minority group participation in RCTs relates to patients' language proficiency and literacy [19,20]. Among those living in the United States, around 25 million people are unable to speak English fluently , while it is estimated that almost 300,000 adults in England and Wales from the 4 most common ethnic minority groups speak little or no English . "
[Show abstract] [Hide abstract] ABSTRACT: To compare characteristics of individuals participating in randomized control trials (RCTs) of treatments of substance use disorder (SUD) with individuals receiving treatment in usual care settings, and to provide a summary quantitative measure of differences between characteristics of these two groups of individuals using propensity score methods. Analyses using data from RCT samples from the National Institute of Drug Abuse Clinical Trials Network (CTN) and target populations of patients drawn from the Treatment Episodes Data Set-Admissions (TEDS-A). Multiple clinical trial sites and nationwide usual SUD treatment settings in the United States. A total of 3,592 individuals from 10 CTN samples and 1,602,226 individuals selected from TEDS-A between 2001 and 2009. The propensity scores for enrolling in the RCTs were computed based on the following nine observable characteristics: sex, race/ethnicity, age, education, employment status, marital status, admission to treatment through criminal justice, intravenous drug use, and the number of prior treatments. The proportion of those with ≥12 years of education and the proportion of those who had full-time jobs were significantly higher among RCT samples than among target populations (in seven and nine trials, respectively, at p<.001). The pooled difference in the mean propensity scores between the RCTs and the target population was 1.54 standard deviations and was statistically significant at p<.001. In the USA, individuals recruited into randomized control trials (RCT) of substance use disorder treatments appear to be very different from individuals receiving treatment in usual care settings. Notably, RCT participants tend to have more years of education and a greater likelihood of full-time work compared with people receiving care in usual care settings.
- "Lack of external validity is a concern when the RCT participants are different from the target population. Findings from recent studies have heightened these concerns by showing that RCT samples might not represent the types of patients encountered in usual clinical practice settings [2,3]. In the context of substance use disorders (SUD), recent studies have shown that tight exclusion criteria commonly employed in RCTs might have resulted in RCT samples that are different with regard to sex and race distribution from the treatmentseeking populations in usual care settings4567 . "
[Show abstract] [Hide abstract] ABSTRACT: Direct comparisons of the effect of a glycated haemoglobin measurement or an oral glucose tolerance test on the uptake and yield of screening in people of South Asian origin have not been made. We evaluated this in 18 to 60-year-old South Asian Surinamese. We invited 3173 South Asian Surinamese for an oral glucose tolerance test between June 18th 2009- December 31st 2009 and 2012 for a glycated hemoglobin measurement between April 19th 2010-November 11th, 2010. Participants were selected from 48 general practices in The Hague, The Netherlands. We used mixed models regression to analyse differences in response and participation between the groups. We described differences in characteristics of participants and calculated the yield as the percentage of all cases identified, if all invitees had been offered screening with the specified method. The response and participation in the glycated hemoglobin group was higher than in the group offered an oral glucose tolerance test (participation 23.9 vs. 19.3; OR: 1.30, 95%-confidence interval1.01-1.69). After adjustment for age and sex, characteristics of participants were similar for both groups. Overall, glycated hemoglobin identified a similar percentage of type 2 diabetes cases but a higher percentage of prediabetes cases, in the population than the oral glucose tolerance test. We found that glycated hemoglobin and the oral glucose tolerance test may be equally efficient for identification of type 2 diabetes in populations of South Asian origin. However, for programs aimed at identifying people at high risk of type 2 diabetes (i.e. with prediabetes), the oral glucose tolerance test may be a less efficient choice than glycated hemoglobin.
- "As compared to screening studies among European populations, the uptake is only slightly lower in our study [4,6789101112. This lower uptake was expected as a lower participation is often observed in studies among migrant populations in industrialised countries [4,282930. Differences in uptake in HbA1c vs. OGTT group. The higher uptake for HbA1c is in line with the assumption that a more burdensome test is associated with a lower response [4, 20]. "