Exclusion of older adults from randomized controlled trials in rheumatoid arthritis

Abstract

Objectives

To assess eligibility criteria that either explicitly or implicitly exclude older patients from randomized controlled trials (RCTs) in RA.

Methods

Our analysis included RCTs of pharmacological interventions registered with ClinicalTrials.gov and started between 2013 and 2022. Co-primary outcomes were proportions of trials with an upper age limit and the eligibility criteria indirectly increasing risk of the exclusion of older adults.

Results

A total of 143/290 (49%) trials had an upper age limit of 85 years or less. Multivariable analysis showed that the odds of an upper age limit were significantly lower in trials performed in the USA [adjusted odds ratio (aOR), 0.34; CI, 0.12–0.99; P = 0.04] and intercontinental trials (aOR, 0.4; CI, 0.18–0.87; P = 0.02). In total, 154/290 (53%) trials had at least one eligibility criterion implicitly excluding older adults. These included specific comorbidities (n = 114; 39%), compliance concerns (n = 67; 23%), and broad and vague exclusion criteria (n = 57; 20%); however, we found no significant associations between these criteria and trial characteristics. Overall, 217 (75%) trials either explicitly or implicitly excluded older patients; we also noted a trend towards increasing proportion of these trials over time. Only one trial (0.3%) enrolled solely patients aged 65 and older.

Conclusion

Older adults are commonly excluded from RCTs in RA based on both age limits and other eligibility criteria. This seriously limits the evidence base for the treatment of older patients in clinical practice. Given the growing prevalence of RA in older adults, relevant RCTs should be more inclusive to them.

Introduction

RA is a systemic inflammatory disease resulting in joint destruction and substantial patient disability [1]. In the past RA was considered a disease affecting primarily middle-aged patients [2]. However, several studies performed over the last three decades have consistently shown that the prevalence of RA has substantially increased in older persons. This trend was reported in publications originating from different countries including the USA, the UK, Finland, Sweden and Japan [2–6]. For instance, analysis of the data on 8826 RA patients from the Swedish National Patient Register revealed that the incidence of the disease peaked in the 70–79 years age group for both men and women [6].

The population of older adults with RA has been expanding rapidly largely as a result of increasing life expectancy in the general population [7, 8]. RA occuring in older individuals can be divided into two subsets: (i) RA with onset after 60 or 65 years of age (‘elderly-onset RA’; EORA); and (ii) RA that developed in younger patients who aged naturally (‘younger-onset RA’; YORA; [7–9]).

There are a number of significant differences between younger and older patients with RA. These include especially comorbidities and polypharmacy that are common in older adults as well as physiological changes affecting both the course of RA and the pharmacokinetics of antirheumatic drugs [10]. Therefore, to provide optimal care of older patients with RA, doctors need data on the safety and efficacy of anti-rheumatic treatments coming from randomized controlled trials (RCTs) enrolling adequate numbers of participants from the higher age groups. Unfortunately, two recent meta-analyses of RCTs in RA showed considerable underrepresentation of older adults in these trials, especially when compared with observational studies [11, 12]. However, the authors did not attempt to identify the factor(s) that are responsible for this.

In general, the most common barrier limiting the enrolment of older patients to clinical trials are the eligibility criteria [13]. A number of studies were already performed to assess the importance of these criteria in clinical trials in different diseases [14–18]. The purpose of this study was to evaluate the eligibility criteria in RCTs in RA, especially those that can limit the inclusion of older adults. To our knowledge the present study is the first to provide a comprehensive analysis of the eligibility criteria limiting the enrolment of older adults in RCTs in RA.

Method

Selection of eligible clinical trials

Our analysis included trials registered with ClinicalTrials.gov (CT.gov), the largest register of clinical studies being conducted around the world (https://www.clinicaltrials.gov/). Over the past two decades clinical trial registries, especially CT.gov, have emerged as a valuable source of data on clinical studies [19]. For instance, several studies on the enrolment criteria in clinical trials were performed on data from CT.gov [16, 18, 20].

Eligible trials were searched for in CT.gov using the search term ‘rheumatoid arthritis’ (field ‘Condition or disease’). The search was narrowed down by using the following inclusion criteria: (i) interventional studies; (ii) randomized; (iii) primary purpose ‘treatment’; (iv) study start date on 01/01/2013 or later (advanced search function). Studies with the recruitment status ‘withdrawn’, ‘suspended’, and ‘unknown’, paediatric clinical trials, trials enrolling healthy volunteers as well as trials of non-pharmacological interventions were excluded. The record of each of the potentially eligible trials was manually reviewed to ensure that it met all the inclusion criteria.

Data extraction and analysis

We focused on two types of data: (i) general characteristics of the trials; and (ii) the eligibility criteria. The general characteristics of the trials included the data on the sponsor, status, trial start date and registration date, CT.gov identifier, title, condition/disease, intervention name and type, trial phase, sample size, timeframe for primary end point assessment, and centre location. The sponsors were divided into two main categories: (i) industrial (for trials in which the pharmaceutical industry was either the primary sponsor or a ‘collaborator’ according to the terminology used by CT.gov); and (ii) other (with no involvement of the pharmaceutical industry). Our preliminary analysis of the potentially eligible trials showed that the National Institutes of Health (NIH) were involved in sponsoring only three trials so these were included in the category ‘Other’.

When designing analysis of the eligibility criteria, we took into consideration the results of previous studies that aimed to identify the main protocol barriers limiting the enrolment of older adults in clinical trials [14–16, 18]. These studies showed that there are two main types of the eligibility criteria that limit the inclusion of older patients: (i) the upper age limits; and (ii) the criteria indirectly increasing risk of the exclusion of older persons. The latter include especially the criteria related to specific comorbidities, broad and vague exclusion criteria as well as the criteria concerning compliance concerns.

The analysis of the exclusion criteria concerning comorbidities was based on the classification developed by Lewis et al. [21]. The system was also used in other studies on the exclusion of older adults from clinical trials including one study from our group [14, 22]. It distinguishes two main types of the exclusion criteria: moderate and strict. Strict exclusion criteria require normal or almost normal organ/system function, while moderate criteria allow for some mild abnormalities. The details of the classification are available at https://theoncologist.onlinelibrary.wiley.com/doi/suppl/10.1634/theoncologist.2014-0093.

Broad and vague criteria generally involve unspecified conditions and are open to wide investigators’ interpretation. These included, for instance, a criterion referring to ‘subject considered by the investigator, for any reason, to be an unsuitable candidate for the study’. Compliance concerns explicitly referred to a participant’s ability to comply with the trial’s requirements including an ability to participate in control visits throughout the trial period.

The data were extracted from records of individual trials, inserted into an Excel spreadsheet, and analysed in Excel. Data extraction and analysis were performed independently by two investigators (A.M. and J.B.). All discrepancies were resolved through consensus. Co-primary outcomes were proportions of trials with an upper age limit of 85 years or lower, at least one strict exclusion criterion related to a specific comorbidity, a broad and vague exclusion criterion, or an exclusion criterion concerning compliance concerns. The study was not registered.

Statistical analysis

Descriptive statistics was used to characterize included trials. Continuous variables were presented as medians with interquartile ranges and discrete variables as absolute numbers and percentages. In order to determine the associations between certain variables and the co-primary outcomes, univariate and multivariable logistic regression was used. A range of pre-defined characteristics of the trials were used as co-variates. Each of the co-variates was included in both univariate and multivariable models. Temporal trends regarding the frequency of different types of the eligibility criteria were assessed with univariate logistic regression. Statistical significance of variables included in logistic regression was determined with the Wald test. All statistical calculations were performed using R programming language (packages finalfit and dplyr). P < 0.05 was considered a statistically significant threshold.

Results

Selection and characteristics of randomized controlled trials in RA

Our study focused on RCTs of pharmacological interventions used in the treatment of RA. Eligible trials were searched for in CT.gov. The register was queried on 14 April 2022. Trial selection is shown in Fig. 1. The inclusion criteria were met by 290 trials (Supplementary Table S1, available at Rheumatology online).

Detailed characteristics of included RCTs are presented in Table 1. Most of the trials (n = 159; 55%) involved the use of a biological DMARD followed by a targeted synthetic DMARD (n = 73; 25%). The median number of participants was 160 [interquartile range (IR), 60–381]. Most of the trials were phase 2 trials (n = 105; 36%), followed by phase 3 trials (n = 98; 34%). The pharmaceutical industry was involved in sponsoring 220 trials (76%). The majority of the trials were intercontinental (106; 37%); many trials were also performed in Europe (n = 65; 22%) and Asia (n = 63; 22%). The median timeframe for primary end point assessment was 23 weeks (IR range 12–24 weeks).

Age limits

Detailed data on the age limits used in the included RCTs are presented in Table 2. We found that 154 trials (53%) had an upper age limit. These mostly fell within the range between 71 and 75 (n = 57; 20%). Overall, the number of RCTs with an upper age limit of 85 years of age or less (co-primary outcome) was 143 (49%). Only one trial (0.3%) enrolled solely participants aged 65 or older. The objective of this trial was to assess the effectiveness, safety and cost-effectiveness of low-dose glucocorticoid treatment in older patients with RA (GLORIA study; NCT02585258). In addition, there were two trials only enrolling participants aged 50 and older. The purpose of the first trial (NCT02982083) was to evaluate the efficacy of raloxifene in postmenopausal women with RA, while the aim of the other (NCT02092467) was to assess the safety of tofacitinib and some TNF alpha inhibitors in patients with RA. None of the three trials had an upper age limit.

Using logistic regression, we also identified the factors affecting the odds of a trial having an upper age limit of 85 years of age or less. We performed both univariate and multivariable analyses. The following covariates were included in both analyses: centre location, trial phase, sample size, timeframe for primary end point assessment, the sponsor and the intervention type.

Detailed results of logistic regression analysis regarding the upper age limits are shown in Supplementary Table S2, available at Rheumatology online. In univariate analysis, significantly higher odds of a trial excluding older adults were associated with industry funding [odds ratio (OR), 1.92; CI, 1.11–3.37; P = 0.02] and lower in trials with a higher number of participants (OR, 0.99; CI, 0.99–0.99; P < 0.001), longer timeframes for primary end point assessment (OR, 0.98; CI, 0.96–0.99; P = 0.002), and in intercontinental trials (OR, 0.44; CI, 0.23–0.84; P = 0.01). However, in multivariable analysis the only covariate significantly associated with the odds of a trial having an upper age limit of 85 years of age or less was centre location [adjusted OR (aOR), 0.34; CI, 0.12–099; P = 0.04 for the USA and aOR, 0.4; CI, 0.18–0.87; P = 0.02 for intercontinental trials].

Moreover, we assessed whether there is any temporal trend regarding the frequency of trials with an upper age limit. Proportions of RCTs with an upper age limit in individual years are shown in Fig. 2 (by trial start date) and Fig. 3 (by trial registration date). Univariate logistic regression showed a trend towards increasing the proportion of such trials over time (OR 1.07, 95% CI 0.98, 1.17, P = 0.09 in analysis by both the start date and registration date).

Criteria indirectly increasing risk of the exclusion of older adults

We also evaluated the eligibility criteria that implicitly exclude older adults (co-primary outcome). These included the criteria related to specific comorbidities, broad and vague exclusion criteria as well as the criteria concerning compliance concerns.

Detailed data on the exclusion criteria involving comorbidities are presented in Table 2. Overall, 114 trials (39%) had at least one strict exclusion criterion related to a specific comorbidity. These included mostly prior malignancies (n = 44; 15%), followed by the liver function disorders (n = 39; 13%), and the criteria concerning cardiac diseases (n = 32; 11%). We also found that 78 out of 143 (55%) trials with an upper limit of 85 years of age or less did not have any strict exclusion criterion concerning comorbidity.

Broad and vague exclusion criteria were listed in 57 (20%) trials. Moreover, 67 (23%) trials had the exclusion criteria related to compliance concerns. Overall, the number of trials with at least one criterion indirectly increasing risk of the exclusion of older adults was 154 (53%; some trials may have had more than one such criterion). Univariate analysis showed that the odds of a trial having at least one criterion implicitly excluding older patients were lower in intercontinental trials (OR, 0.47; CI, 0.24–0.89; P = 0.02). However, this association was not confirmed by multivariable analysis (P >0.05; Supplementary Table S3, available at Rheumatology online).

Proportions of RCTs with the criteria indirectly increasing risk of the exclusion of older adults in individual years are shown in Fig. 2 (by trial start date) and Fig. 3 (by trial registration date). Univariate logistic regression showed a trend towards increasing the proportion of these trials over time (OR 1.03, 95% CI 0.95, 1.12, P = 0.43 in analysis by the start date and OR 1.01, 95% CI 0.92, 1.10, P = 0.81 in analysis by the registration date).

Overall, 217 trials (75%) had either an upper age limit of 85 years of age or less or at least one criterion indirectly increasing risk of the exclusion of older patients (some studies may have had both types of criteria).

Exploratory analysis of trials of non-pharmacological interventions

Our study focused on RCTs of pharmacological interventions; as shown in Fig. 1, trials of non-pharmacological interventions were excluded from the primary analysis. However, we also performed a separate exploratory analysis to evaluate the extent of the exclusion of older adults from RCTs of non-pharmacological interventions (n = 68). These included 65 RCTs of non-invasive interventions and three RCTs of invasive interventions (detailed data not shown). The analysis showed that 41 (60%) trials had an upper age limit of 85 years of age or less. In addition, 22 (32%) trials had at least one criterion indirectly increasing risk of the exclusion of older adults. Overall, 50 (74%) RCTs of non-pharmacological interventions had either an upper age limit or at least one criterion implicitly excluding older patients. Thus, the results of analysis of RCTs of the pharmacological interventions are generalizable also to trials of other types of interventions.

Discussion

Our study has shown a remarkable scale of the exclusion of older adults from RCTs in RA based on different types of the eligibility criteria. First of all, 49% of the trials started between 2013 and 2022 had an upper age limit of 85 years of age or less. Moreover, 53% of the trials had a criterion indirectly increasing risk of the exclusion of older patients. Overall, 75% of the trials either explicitly excluded older adults or had high risk of such exclusion. Furthermore, we noted a trend towards increasing the proportion of trials with upper age limits and those implicitly excluding older patients over time.

The odds of the presence of an upper age limit were significantly lower in RCTs performed in the USA. This likely results from long-standing efforts by some key US institutions, especially the FDA and the NIH, aiming to improve the enrolment of older adults in clinical trials [23–26].Over the past decade a number of studies have been performed to assess the generalizability of the results of RCTs in RA [11, 12, 27–29]. A consistent finding relevant to our considerations is underrepresentation of older adults in RCTs, especially when compared with observational studies concerning RA [11, 12]. However, only two studies provided some data on age limits used in clinical trials in RA [27, 29]. Furthermore, none of them assessed the eligibility criteria indirectly increasing risk of the exclusion of older adults. To our knowledge, our study is the first to provide comprehensive analysis of the eligibility criteria limiting the enrolment of older patients in RCTs of antirheumatic treatments.

In general, the eligibility criteria aim to ensure the homogeneity of the sample of participants and their safety. Many of these are justified and essential to meet the trial objective. However, restrictive criteria may substantially reduce the generalizability of trial results [25, 30]. This problem (especially the exclusion of older adults) is relevant to RA because, as consistently shown by the epidemiological data, the number of older patients with RA has been growing rapidly. Thus, the substantial extent of the exclusion of older persons from RCTs in RA seriously limits the evidence base for the treatment of patients from higher age groups in clinical practice. Therefore, investigators involved in the design of the eligibility criteria in trials of antirheumatic drugs should carefully consider whether each of the planned criteria should be introduced. Some authors expressed concerns that certain exclusion criteria used in newer trials may not be entirely scientifically justified, but rather replicated from the protocols of earlier trials [30]. It is noteworthy here that over the last decade several research groups reported inadequate generalizability of trial results associated with restrictive eligibility criteria in different diseases [15, 16, 20, 31].

One of the main types of the eligibility criteria that are particularly controversial involves the upper age limits. Each of such criteria is arbitrary. The use of upper age limits in clinical trials was criticized [15, 16, 18] largely due to heterogeneity of the ageing process, which implies that an older person’s chronological age does not always correlate with his/her state of health [32]. Unfortunately, as shown by the results of our study, the exclusion of older adults based on age limits is prevalent also in RCTs in RA. The most common limits fell within the range between 76 and 80 years of age. This finding is in line with the results of a recent cross-sectional study which showed that the group of older adults that are most underrepresented in pivotal RCTs in different diseases other than RA are those older than 75 or 80 years of age [33]. We also noted that 55% of the trials with an upper age limit did not have any strict exclusion criterion concerning comorbidity. Thus, these trials excluded older patients solely based on their age, but were potentially open to younger individuals with a poorer state of health.

RA is a disease associated with a high risk of some comorbidities including but not limited to diseases of the cardiovascular system [34], the respiratory system [35] and malignancies [36]. Because the ageing process itself is associated with higher prevalence of at least some of these diseases, they are also an important clinical problem in older patients with RA [37]. For instance, Mochizuki et al. found that in 129 RA patients aged 65 and older, common comorbidities included hypertension (37.2%), cardiovascular diseases (14.7%) and cancer (10.1%) [38]. Thus, strict exclusion criteria related to comorbidities frequently occurring in patients with RA also substantially limit generalizability of trial results.

Another problem relevant to the participation of older adults in clinical trials in RA is frailty. RA increases risk of frailty [39]. Recent meta-analyses revealed that pooled prevalence of frailty in RA patients was 24% and 33.5% [40, 41]. While some studies showed that in patients with RA age is an independent risk factor of frailty [42], neither of the meta-analyses confirmed this [40, 41]. Another study performed on 210 older patients with RA showed that the prevalence of frailty is 37.6% [43]. Thus, while a substantial proportion of patients with RA suffer from frailty, most of them, including many older individuals, are not frail. Furthermore, if the investigator suspects that an older patient can be frail, this may be verified by performing comprehensive geriatric assessment (CGA [44]). The use of CGA in older patients with RA seems to be an option more reasonable than the introduction of an arbitrary upper age limit. Importantly, some RA-specific tools to identify frail patients were already developed and validated [45].

The investigators might also have concerns over risk of increased rate of premature dropouts of older trial participants. To assess this problem, Palmowski et al. performed mixed effects meta-regression analysis of 243 RCTs involving over 48 000 patients with RA or osteoarthritis. The authors showed that the proportion of older adults among trial participants does not affect the general participant retention [46]. Furthermore, some methods were shown to improve the retention of older participants in clinical trials [13]; these could also be employed in clinical trials in RA.

Our study focused on the eligibility criteria as the barrier to the inclusion of older adults to clinical trials. While this barrier is the most common, there are also other important barriers including the reluctance of physicians, lack of interest of the patients, and transportation problems [13]. Therefore, multifaceted activities are necessary to improve the inclusion of older adults to clinical trials [47]. The most apparent aspect of this problem is to design trials with the eligibility criteria more inclusive to older individuals. Other activities might include training of rheumatologists who may be involved in recruitment of RA patients to clinical trials. Eventually, these efforts should lead to the development of treatment guidelines taking into account the distinctiveness of older patients. Unfortunately, the current major guidelines by the American College of Rheumatology and the European Alliance of Associations for Rheumatology (formerly EULAR) do not provide any specific recommendations regarding the treatment of older adults [48, 49].

The main strength of this study is that it is the first to assess different types of eligibility criteria that either explicitly or implicitly exclude older adults from RCTs in RA. Unlike the previous studies which focused on the age limits, we also evaluated other criteria that indirectly increase risk of the exclusion of older patients. The main limitation is that our study included only trials registered with CT.gov, which is not the sole registry of clinical trials. Thus, some trials may have been missing from our analysis. However, CT.gov is the largest registry of clinical trials and previous studies on the exclusion of older adults, as well as patients with comorbidities, from clinical trials were performed on the data from CT.gov [14, 18, 20]. Another limitation is that we relied only on the data from CT.gov and did not have access to trial protocols.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: The authors have declared no conflicts of interest.

Acknowledgements

The authors are grateful to Dr Magdalena Zielenkiewicz, Faculty of Mathematics, Informatics, and Mechanics, University of Warsaw, for performing expert statistical analysis of the results.

References