Randomised controlled trials (RCTs) are often deemed the gold standard in comparative effectiveness research, since their synthesis through systematic reviews and meta-analyses is used to inform clinical care guidelines that guide evidence-informed practice.Reference Hariton and Locascio1 However, inconsistency and insufficiency in reporting of clinical trials, and in particular, their outcomes, is a long-standing issue in biomedical research, and challenges evidence-based care.Reference Glasziou, Altman, Bossuyt, Boutron, Clarke and Julious2–Reference Saldanha, Lindsley, Money, Kimmel, Smith and Dickersin8 Outcomes or end-points indicate intervention success or effectiveness, and are essential components of clinical trials.Reference Gorst, Gargon, Clarke, Blazeby, Altman and Williamson6,Reference Williamson, Altman, Blazeby, Clarke and Gargon9,Reference Dodd, Clarke, Becker, Mavergames, Fish and Williamson10 However, prior research has demonstrated that clinical trials insufficiently report the rationale for outcome selection, definition of the outcome, outcome measurement details and methodology for outcome analysis.Reference Dechartres, Trinquart, Atal, Moher, Dickersin and Boutron3,Reference Dwan, Gamble, Williamson and Kirkham5,Reference Azar, Riehm, McKay and Thombs11–Reference Chan and Altman13 Deficiencies in outcome reporting in trials (i.e. lack of sufficient details reported to ensure complete understanding of the end-point) impedes the reproducibility of trials and cross-study comparison of results, and further limits the uptake of research to clinical practice, thereby contributing to research waste.Reference Santor, Gregus and Welch14–Reference Macleod, Michie, Roberts, Dirnagl, Chalmers and Ioannidis17 Although prior research has examined primary outcome reporting in trials for adolescents with major depressive disorder (MDD),Reference Monsour, Mew, Szatmari, Patel, Saeed and Offringa18 reporting comprehensiveness of primary outcomes has not been assessed in RCTs for geriatric depression.
Outcome reporting in geriatric depression trials
Depression is one of the leading causes of disability for older adults worldwide, accounting for an estimated loss of 13.8 years of quality-adjusted life expectancy at 65 years of age.Reference Jia, Zack, Thompson, Crosby and Gottesman19 Adverse health outcomes for this clinical population often include a reduced quality of life,Reference Fassino, Leombruni, Daga, Brustolin, Rovera and Fabris20 disabilityReference Beekman, Penninx, Deeg, de Beurs, Geerling and van Tilburg21 and mortality.Reference Abas, Hotopf and Prince22 Geriatric MDD is often treated with one or a combination of interventions including, but not limited to, pharmacotherapy,Reference Kok and Reynolds23 psychotherapyReference Jayasekara, Procter, Harrison, Skelton, Hampel and Draper24 and exercise therapy.Reference Schuch, Vancampfort, Rosenbaum, Richards, Ward and Veronese25 However, there is still uncertainty regarding intervention effectiveness for this unique clinical population, given the prevalence of comorbid mental and physical illnesses that often accompany aging,Reference Kok and Reynolds23 and must be considered during selection of the treatment course because of potential drug–drug interactions between antidepressants and concomitant medications.Reference Kok and Reynolds23 The uncertainty in assessing intervention effectiveness may be partially attributed to variability in outcome reporting and subsequent challenges in interpretation and synthesis of trial findings, which impedes clinical decision-making for geriatric depression. Previous meta-analyses of pharmacologicalReference Kok, Nolen and Heeren26,Reference Mallery, MacLeod, Allen, McLean-Veysey, Rodney-Cail and Bezanson27 and psychosocialReference Wilson, Mottram and Vassilas28 interventions for older adults with depression have reported limitations in interpretability of findings as a result of the heterogeneity in the use of outcomes across trials.
Our recent review identified substantial variability in the outcomes reported by RCTs.Reference Rodrigues, Syed, Dufort, Sanger, Ghiassi and Sanger29 Additionally, up to 19 outcome measurement instruments (OMIs) were used to measure the single outcome, ‘depressive symptom severity’.Reference Rodrigues, Syed, Dufort, Sanger, Ghiassi and Sanger29 Although prior meta-analyses suggest variability in outcome measurement and descriptions,Reference Kok, Nolen and Heeren26–Reference Wilson, Mottram and Vassilas28 there has not been a systematic assessment of outcome reporting comprehensiveness for geriatric depression. A thorough assessment of the comprehensiveness of outcome reporting in trials is integral to understanding the presence and extent of the issue, and inform the need for standardising outcome reporting in trials assessing older adults with MDD. The objective of our study is to extend our previous work, and assess the comprehensiveness of primary outcome reporting in published geriatric depression trials.
Method
Study selection
This study is registered with the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD42021244753). Our study was conducted in conjunction with a systematic survey to identify eligible trials.Reference Rodrigues, Syed, Dufort, Sanger, Ghiassi and Sanger29 We included RCTs assessing any type of intervention for unipolar, non-psychotic MDD for adults aged 65 years and older, which were published in English between 1 January 2011 and 16 July 2021 inclusive. Trials evaluating people with comorbid mental disorders including depression, and those that presented a subgroup analysis containing adults aged 65 years and older, were also included. Pilot and feasibility trials, and follow-up studies and secondary analyses, were included when the primary RCT was published outside of our timeframe. The protocol for this study, which contains detailed search strategy and eligibility criteria, has been published.Reference Rodrigues, Sanger, Dufort, Sanger, Panesar and D'Elia30 In summary, we searched Medline, EMBASE, PsycINFO and the Cochrane Central Register of Controlled Trials databases to identify eligible trials. Title/abstract and full-text screening was conducted independently and in duplicate, using Covidence systematic review software.31 We supplemented our electronic search with a manual search for potentially eligible trials by reviewing the references of all included studies. Discrepancies regarding study inclusion resolved through discussion between reviewers, and a third reviewer, when necessary, to reach consensus during every stage of screening.
As our objective was to assess reporting comprehensiveness of primary outcomes, we restricted the sample to trials that specified a single, discernible primary outcome. Thus, for our present study, two reviewers applied additional eligibility criteria, independently and in duplicate. Specifically, these trials either (a) explicitly described these outcomes as ‘primary’ or using an appropriate synonym; (b) stated that the study aimed to examine the effect of an intervention on that specific outcome in the objectives or (c) used data from that outcome to power the sample size for the trial.Reference Hopewell, Dutton, Yu, Chan and Altman32 Studies with multiple primary outcomes, and/or those for which a primary outcome was not clearly stated, were therefore excluded from our present study as the primary outcome could not be inferred. For pilot and feasibility studies, which are conducted in preparation for full-scale RCTs and also include outcomes pertaining to feasibility,Reference Thabane, Ma, Chu, Cheng, Ismaila and Rios33 we solely considered effectiveness outcomes, in concordance with the objectives of our systematic survey.Reference Rodrigues, Syed, Dufort, Sanger, Ghiassi and Sanger29
Assessment of outcome reporting
We assessed the comprehensiveness of primary outcome reporting for trials included in our study by using a checklist of 70 outcome reporting items. These items were also used by a previous study to evaluate comprehensiveness of outcome reporting in adolescent depression trials,Reference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34 and in the development of the Consolidated Standards of Reporting Trials (CONSORT)-Outcomes checklist (an essential set of reporting items to be included for primary and secondary trial outcomes in published trials).Reference Butcher, Monsour, Mew, Chan, Moher and Mayo-Wilson35 The CONSORT-Outcomes checklist is an extension of the CONSORT 2010 statement (a minimum, recommended set of items to be reported by RCTs).Reference Schulz, Altman and Moher36 The 70-item checklist used in our study and the CONSORT-Outcomes checklist both contain outcome reporting items that spanned the following thematic categories: (a) who (source of information for the outcome), (b) what (outcome description), (c) where (location and setting of outcome assessment), (d) when (timing of outcome measurement), (e) why (rationale for outcome selection), (f) how (method of outcome measurement), (g) management and analysis of outcome data, (h) missing outcome data, (i) outcome interpretation and (j) any modifications made to the outcome.Reference Butcher, Monsour, Mew, Chan, Moher and Mayo-Wilson35,Reference Butcher37–Reference Butcher, Mew, Monsour, Chan, Moher and Offringa39
Of the 70 items, we found 12 items to be irrelevant or unable to be assessed in our study. These items are detailed with reasons for exclusion in Table 2. Thus, the outcome reporting assessment was conducted with the resulting 58-item checklist, similar to the assessment of primary outcome reporting across adolescent depression trials.Reference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34 Study team members (A.O., K.J.) were trained by a methodologist (M.R.) before conducting assessment of outcome reporting, using a sample of three randomly selected RCTs (see Supplementary File 1 available at https://doi.org/10.1192/bjo.2023.650 for the training guide). Once consensus was reached (≥80% agreement between reviewers) for each of the three trials, outcome reporting assessment was conducted for other studies independently and in duplicate, using predefined standardised data charting forms on Microsoft Excel (Microsoft Corporation; see https://office.microsoft.com/excel) from 31 January 2023 to 31 March 2023. Any disagreements were resolved through discussion, and by a third reviewer (M.R.) as needed to reach consensus. We used the same assessment process for every trial included in our study, in order to reach consensus on all appraised items.
Scoring details
We assessed outcome reporting for each of the 58 checklist items as ‘fully reported’, ‘partially’ reported’ or ‘not reported’ for the primary outcome in every trial. A score of ‘fully reported’ was given to items where full details for the item were reported by included studies. This included instances where previously published supplementary materials (i.e. protocols, statistical analysis plans or other reports) were referenced by the authors regarding a particular reporting item. Conversely, items which were ‘partially reported’ by trials reported one or a few items of a multi-component item. This classification only applied to checklist items comprising multiple components (see Table 2 for list), i.e. item 23 (reliability of the OMI in a similar study setting). For instance, this item was scored as ‘partially reported’ when authors indicated that the OMI was reliable but did not specify whether reliability was established in a similar study setting. If no information was provided for the item, or the concept of the particular item was irrelevant to the particular trial based on the information provided in the study, items were classified as ‘not reported’ or ‘not applicable’, respectively. For instance, if the trial did not report having missing data, item 52 was scored as ‘not reported’, and item 55 (justification for methods used to handle missing data) was subsequently deemed ‘not applicable’.
Synthesis of findings
Study characteristics and results for reporting items were analysed descriptively with counts and frequencies. Outcome reporting comprehensiveness was calculated for each trial as a composite measure based on the percentage of items assessed as ‘fully reported, ‘partially reported’ and ‘not reported’.
Results
Search results
We identified 49 RCTs with the initial eligibility criteria, and excluded 18 trials for not having a single, discernible primary outcome. Our current study includes 31 RCTs; 22 studies (71%) explicitly deemed an outcome as ‘primary’, six (19%) aimed to assess the effect of an intervention on that particular end-point and three (10%) used data from the outcome to power the sample size for the trial (see Fig. 1 for the flow diagramReference Page, McKenzie, Bossuyt, Boutron, Hoffmann and Mulrow40). Our complete dataset may be found in Supplementary File 2, with references to all included trials in Supplementary File 3.
Fig. 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram for trials assessing treatment interventions for major depressive disorder in older adults.
Characteristics of included trials
The characteristics of the 31 RCTs included in our study are described in Table 1. Most included studies were conducted in Europe (number of studies k = 11, 36%) or North America (k = 8, 26%), with the majority being publicly funded (k = 16, 52%). Nearly half the trials assessed pharmacological interventions (k = 15, 48%), with the remainder of studies assessing psychosocial (k = 10, 32%), case management (k = 5, 16%) or acupressure (k = 1, 3%) interventions. The number of participants in included studies ranged from 13 to 1879, with a median sample size of 174. The most commonly reported primary outcome was ‘depressive symptom severity’, reported by 15 trials (48%), followed by ‘depression treatment response’ (k = 12, 39%; see Supplementary Table 1(a) for definitions and frameworks used to classify outcomes in our original study).
Table 1 Characteristics and primary outcomes of included studies
a. As the mean or median ages for the included populations in the majority of studies were unclear, we have indicated the age cut-offs.
b. Funding sources categorised as follows: public: funded by a governmental organisation (e.g. National Institute of Mental Health, National Institute for Health Research); industry: for-profit corporation (e.g. Janssen Research & Development, AstraZeneca Pharmaceuticals); academic: university or other academic institution (e.g. Harvard Medical School, Tehran University of Medical Science); not for profit: not-for-profit foundation or organisation (e.g. The Health Foundation).
c. Study included both younger populations and older adults with major depressive disorder, but reported data stratified by age for those aged ≥65 years. Information has been extracted for this stratified population, which fulfilled our study inclusion criteria.
d. Feasibility trial.
e. Follow-up study.
Outcome reporting assessment
Overall, there was variation in the items scored as ‘fully reported’, ‘partially reported’ or ‘not reported’ across the thematic categories (Fig. 2). The category ‘Outcome data management and analyses’ had the highest percentage of fully reported items (73%), followed by ‘What: Description of the outcome’ (66%) and ‘Outcome interpretation’ (65%). The lowest percentage of fully reported items were observed for the categories ‘How: Method of outcome measurement’ (17%) and ‘Who: Source of information for the outcome’ (32%).
Fig. 2 Outcome reporting comprehensiveness across 31 geriatric major depressive disorder trials, by thematic item category.
The assessment of outcome comprehensiveness was variable for each of the included 31 RCTs. Overall, each study fully reported about half of the 58 checklist items (Fig. 3, Supplementary File 2). The percentage of items that were fully reported by each trial varied from 34 to 64%, with a median of 45%. The percentage of items that were fully reported remained relatively stable from 2011 through 2021, i.e. over a 10-year period (Fig. 3). We describe outcome reporting comprehensiveness for each thematic category in the following sections, with reporting frequencies for all 58 items presented in Table 2.
Fig. 3 Outcome reporting comprehensiveness across 31 geriatric major depressive disorder trials.
Table 2 Frequency of outcome reporting classifications for each reporting item for the primary outcome in included trials (n = 31)
This table has been adapted from an assessment of primary outcome reporting in adolescent depression trials.Reference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34
a. Not applicable refers to instances where ‘partially reported’ was not a valid assessment option. Items scored as ‘Not applicable’ were not included in the overall scoring, since they were deemed to be irrelevant to the assessment of outcome reporting by the research team (M.R., A.O., K.J., L.T., S.P., Z.S.), by consensus.
b. Outcome domain defined in accordance with core taxonomic framework proposed by Dodd et al.Reference Dodd, Clarke, Becker, Mavergames, Fish and Williamson10,Reference Rodrigues, Syed, Dufort, Sanger, Ghiassi and Sanger29 Given that domains are broad and not directly measurable, outcomes are selected to assess change within them. See Supplementary Table 1(a) for further details.
c. Outcome reporting items removed from the comprehensive item checklist, and subsequently excluded from reporting assessment.
d. Item was considered ‘fully reported’ only when all components for that item were reported in the trial, e.g. for item 13, if both scaling and scoring details were reported.
e. Several items do not add to a total denominator of N = 31 trials for the following reasons: items 19–26 (denominator: 30 trials) were not applied to a trial where the primary outcome was behavioural change, i.e. change in provider treatment adherence, which does not have gold standard measures of validity, reliability, etc.; item 27 (denominator: 27 trials) did not apply to trials that assessed only one outcome; items 44 and 45 (denominator: 18 trials) were not assessed for trials that did not include covariates/factors in their statistical models; item 51 (denominator: 15 trials) only applied to trials that conducted additional analyses; and items 53–56 and 58 (denominator: 26 trials) were only applied to trials that reported having missing data.
What: description of the outcome
Every included trial described the outcome domain, stated the outcome and specified the outcome as primary (k = 31, 100%; items 1–3, respectively). However, only 23% (k = 7/31) of included studies fully reported a rationale for classifying the outcome as primary (item 4). Although just over half (k = 16, 52%; item 5) of included RCTs defined clinical significance of the outcome, the criteria used to define meaningful change was infrequently reported by studies (k = 7, 23%; item 6).
Why: rationale for selecting the outcome
There was variation in the descriptions of the rationale for outcome selection by trials included in our study. Outcome items that were most frequently reported included explanations of how the outcome addresses the research question (k = 31, 100%; item 8) and described how the outcome relates to the hypothesis of the study (k = 27, 87%; item 7). In this category, less frequently reported items described why the primary outcome was relevant to stakeholders (k = 2, 6%; item 9), and which stakeholders were actively involved in selection of the outcome (k = 2, 6%; item 10).
How: the way the outcome is measured
Overall, items pertaining to the way the outcome was measured were reported poorly by geriatric depression trials. Although all trials (k = 31, 100%) described the OMI used, less than half (k = 15, 48%; item 13) included details regarding instrument scaling and scoring. No trial (k = 0, 0%; item 15) specified a recall period for outcome assessment. Thirty of the 31 included trials could be assessed for reporting on measurement properties, as the primary outcome for one RCT was provider treatment adherence, which does not have measures of validity, reliability, etc. Only nine studies (30%; item 19) described the validity of the OMI in individuals similar to the study sample, with 17% of trials (k = 5; item #0) justifying validity of the OMI in the study setting. Four RCTs (13%; item 22) fully reported reliability of the OMI in a relevant study sample, with even fewer studies (k = 2, 7%; item 23) describing reliability of the OMI in the specified study setting. Only a paucity of trials explicitly described responsiveness of the OMI used in the study (k = 3, 10%; item 24) or the feasibility (k = 2, 7%; item 25), acceptability and/or burden of the OMI in the study sample (k = 1, 3%; item 26).
Who: source of information of the outcome
Descriptions related to the identity and number of outcome assessors were fully reported by just over half of the included trials (k = 16, 52%; item 29). However, justification regarding the choice of outcome assessors (k = 3, 10%; item 30) and trial-specific training required for outcome assessors (k = 8, 26%; item 32) were less frequently reported.
Where: assessment location and setting of the outcome
The location of outcome assessment was reported by 97% of included studies (k = 30; item 34). However, descriptions of the setting of outcome assessment (i.e. clinic, home, other) were reported by 61% of RCTs (k = 19, 61%; item 35), with no trial justifying why the outcome setting was suitable for the study sample (k = 0, 0%; item 36).
When: timing of measurement of the outcome
Every included study described the timing and frequency of outcome assessment (k = 31, 100%; item 37); however, only 32% of studies (k = 10; item 38) provided justification for timing of outcome measurement.
Outcome data management and analyses
Overall, geriatric depression trials demonstrated good reporting of items pertaining to outcome data management and analyses. All trials (k = 31, 100%) described the outcome analysis population (item 39), the unit of analysis of the outcome (item 40), the outcome analysis metric (item 41), the method of aggregation for outcome data (item 42), the statistical methods/significance tests used in analysis (item 43) and the time period for outcome analysis (item 47).
There was variability in the description of items pertaining to outcome management, with between 3 and 29% of items being fully reported by RCTs (items 48–50). Less than a third of studies (k = 9, 29%; item 48) described the outcome data, assessment process and analysis for participants who discontinued or deviated from the assigned interventional protocol.
Missing outcome data
Of the RCTs, 46% or more described how much data was missing, described reasons for missingness in each study arm and explained the statistical methods used to handle missing outcome data (items 52–54). However, only 15% of studies (k = 4; item 55) provided justification for the methods used to handle missing data, which was the least frequently reported item in this category.
Outcome interpretation
Although every study reported an interpretation of outcome data in relation to clinical outcomes (k = 31, 100%; item 57), only a paucity of RCTs (k = 6, 23%; item 58) discussed the impact of missing outcome data on the interpretation of findings.
Discussion
Our study found that comprehensiveness of primary outcome reporting in geriatric depression trials published between 2011 and 2021 was variable and mostly insufficient. Notably, the level of detail and descriptions of primary end-points were inconsistent, which impedes full comprehension of markers used to indicate intervention effectiveness. Overall, less than half of the reporting items from the checklist of 58 items were fully reported by trials. Furthermore, outcome reporting was relatively stable and did not improve over the 10-year period. Items that described analysis of the primary outcome were generally fully reported, whereas those that detailed how the end-point was measured were only fully reported in 17% of included trials.
The reporting of outcomes must be conducted in a comprehensive manner, i.e. with sufficient detail to permit full understanding of an end-point, to facilitate transparency of information about the trial from design stage, through to conduct and outcome assessment.Reference Chan, Song, Vickers, Jefferson, Dickersin and Gøtzsche41 Conversely, variability in outcome reporting, including reporting of insufficient details to permit full understanding of any aspect of a trial's end-point measures, impedes the comparison and synthesis of findings. In particular, this creates difficulty in translating research findings into evidence synthesis products, such as systematic reviews and meta-analyses, consequently reducing their ability to be utilised in clinical decision-making.Reference Mayo-Wilson, Fusco, Li, Hong, Canner and Dickersin16 Below, we discuss potential reasons for our findings, and implications for pertinent stakeholders, which should be considered in the interpretation, replicability and synthesis of geriatric depression trials.
Overview of outcome reporting
Although we observed variability in primary outcome reporting across geriatric depression RCTs, it should be noted that several items on our checklist were well-reported across trials. Reporting elements which were well-reported described the timing and frequency of outcome assessment and analyses. Specifically, all trials in our study described the outcome analysis population, unit of analysis, outcome analysis metric, method of aggregation, statistical methods for analysis and the time period for outcome analysis. Our findings also echo those of a recently conducted study on primary outcome reporting in adolescent depression trials.Reference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34 These results may be attributed to the CONSORT reporting guidelines.Reference Schulz, Altman and Moher36 In particular, timing of outcome assessment and outcome analysis represent iterations of items present in the CONSORT reporting guideline, which has been widely used, and is considered the current gold standard for reporting findings from clinical trials.Reference Schulz, Altman and Moher36 Although prior research has demonstrated that the CONSORT guideline facilitates comprehensiveness in reporting practices for RCTs,Reference Devereaux, Manns, Ghali, Quan and Guyatt42–Reference Plint, Moher, Morrison, Schulz, Altman and Hill44 our study highlights that deficiencies in outcome reporting still remain. The general guidance in outcome reporting provided by the original CONSORT statementReference Schulz, Altman and Moher36 may be insufficient to fully ensure full comprehensiveness of reporting practices. Consequently, the recently developed CONSORT-Outcomes checklistReference Butcher, Monsour, Mew, Chan, Moher and Mayo-Wilson35 may facilitate standardisation of reporting outcome-specific information in future geriatric depression trials, among other fields.
Our findings indicated that only a paucity of included trials detailed measurement properties of the OMI (i.e. validity, reliability, feasibility), which varied from 3 to 30%. Evaluation of depression symptom severity and/or depression treatment response are subjective health outcomes directly reported by patients, and considered latent constructs that are unable to be measured directly. Thus, psychometric scales are used as OMIs in geriatric depression and psychiatry research at large.Reference Balsamo, Cataldi, Carlucci, Padulo and Fairfield45 Despite the extensive use of these scales, however, one cannot assume that different OMIs are equally valid in assessing an outcome. A content analysis by FriedReference Fried46 demonstrated only a mean moderate overlap (Jaccard index: 0.41 (average coefficient of overlap across all scales); range: 0 (no overlap) to 1 (complete overlap)) between common OMIs used in depression research.Reference Fried46,Reference Fried47 Thus, it is particularly important to report validity, reliability and other measurement properties, to evaluate whether particular OMIs are able to assess such constructs in a valid and reliable manner for the target population in clinical trials. Similarly, a recent review has demonstrated the necessity of including details on measurement properties of OMIs, to communicate the validity of results obtained from using a particular measurement tool, thereby further facilitating understanding of trial outcomes.Reference Butcher, Mew, Monsour, Chan, Moher and Offringa39
Strengths and limitations
Our assessment was conducted in a systematic manner, and employed methodology used in another study that examined reporting in adolescent depression trials.Reference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34 Specifically, two trained reviewers performed reporting assessments independently and in duplicate, using a consensus-based approach to resolve discrepancies.
However, our study is not without its limitations. First, we focused on RCTs that specified a single, discernible primary outcome, thereby excluding trials with multiple primary outcomes or those unclear in specifying a primary outcome. Additionally, we included pilot and feasibility studies, whose outcomes are not powered to detect effectiveness.Reference Thabane, Ma, Chu, Cheng, Ismaila and Rios33 Thus, our study findings may not reflect the true state of primary outcome reporting in full-scale geriatric MDD trials, particularly in the case of selective reporting in trials with multiple primary outcomes, as evidenced by prior research.Reference Tyler, Normand and Horton48 Second, we assessed published trials from 2011 to 2021, thereby excluding unpublished studies or RCTs published outside this period. Nonetheless, given that the trials included in our study spanned a 10-year timeframe, we believe that this is sufficient to assess primary outcome reporting in geriatric depression trials.Reference Williamson, Altman, Bagley, Barnes, Blazeby and Brookes49 Third, the distinction between the categories ‘fully reported’ and ‘partially reported’ may be susceptible to subjectivity in assessment. However, this risk was mitigated by conducting assessment by two reviewers independently and in duplicate (A.O., K.J.), who used a training guide with descriptions and examples of scoring categories, which was developed by methodological experts (M.R., L.T.). Fourth, our 58-item checklist has not been validated, as our study was conducted before publication of the CONSORT-Outcomes guideline,Reference Butcher, Monsour, Mew, Chan, Moher and Mayo-Wilson35 and all items may not be equally relevant in reporting assessment. However, this checklist has been used in a prior study to assess outcome reportingReference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34 and in the development of the eventual CONSORT-Outcomes guideline,Reference Butcher, Monsour, Mew, Chan, Moher and Mayo-Wilson35 with overlap between items in both checklists.
Implications for patients, caregivers and clinicians
Two of our findings in particular pose implications for stakeholders of geriatric depression trials: notably, the rationale for primary outcome selection and consideration of meaningful change.
First, the rationale for classifying an outcome as primary was reported by only 23% of trials, suggesting limited consideration of why a particular outcome is used to indicate treatment success. Given that there is an overall lack of consensus about which outcomes are important to measure in a clinical trial for geriatric depression,Reference Rodrigues, Syed, Dufort, Sanger, Ghiassi and Sanger29 it is unsurprising that the rationale for selecting an outcome as a primary indicator of effectiveness is likewise poorly reported. This finding has important implications for patients, caregivers and clinicians. Knowledge of the trial's primary aims and, consequently, clarity in the rationale for outcome selection, would facilitate patient and caregiver understanding of the relevance of the outcome as a marker of treatment success, particularly when the outcome assessed is meaningful to them.Reference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34 Requirements for reporting the rationale for outcome selection (i.e. through the CONSORT-Outcomes checklist)Reference Butcher, Monsour, Mew, Chan, Moher and Mayo-Wilson35 may potentially encourage trialists to include primary outcomes that reflect intervention effectiveness in accordance with patient and caregiver perspectives, such as improvements in social functioning, as identified through prior research.Reference Kan, Jörg, Buskens, Schoevers and Alma50,Reference McIntyre, Ismail, Watling, Weiss, Meehan and Musingarimi51 Furthermore, an explanation as to why a particular end-point was selected for assessment in a trial would increase its selection in other trials, thereby facilitating the comprehension and comparison of findings between trials through aggregation of results in meta-analyses, and consequently improve evidence-based decision-making.
Second, only 23% of trials in our study justified the criteria for meaningful change, i.e. the minimal important change (MIC) or the minimally clinically important difference. The MIC is a respondent-centred indicator of treatment success that highlights the smallest change on an OMI between two time points that may be considered clinically meaningful.Reference Jaeschke, Singer and Guyatt52 When fully reported, the MIC has the potential to provide meaningful context and guidance for clinical decision-making, as it constitutes a good or poor outcome, which may therefore be used to infer intervention effectiveness in a clinical trial.Reference Krause, Hetrick, Courtney, Cost, Butcher and Offringa53 Our finding that only a few trials reported the MIC suggests that determining what constitutes a good or poor outcome is currently based on statistical significance (i.e. mean differences between intervention groups on OMIs), with little regard for what meaningful change would represent to patients, caregivers and clinicians.Reference De Vet, Terwee, Mokkink and Knol54 Notably, the MIC may be determined with an anchor-based approach, which provides an opportunity for engagement of older adults with depression and their caregivers, and is reflective of the increased movement toward inclusion of patients in health research.Reference Manafo, Petermann, Mason-Lai and Vandall-Walker55 An anchor-based MIC is considered ‘a threshold for a minimal within-person change over time above which patients perceive themselves importantly changed’.Reference Terwee, Peipert, Chapman, Lai, Terluin and Cella56 The MIC may be calculated for different respondent groups (patients, caregivers, clinicians) and, when reported in the published report of a clinical trial, serve as binary indicator(s) demonstrating intervention effectiveness. Furthermore, clinicians may utilise established MIC thresholds in their practice when discussing interventions and expected outcomes with patients. Our study therefore highlights the need for determination and reporting of MIC thresholds for OMIs in geriatric depression trials, to extend our understanding of intervention effectiveness beyond mere statistical significance into critical evaluations of whether clinically meaningful change has been achieved for older adults with depression.
Suggestions for journals
Our study revealed a consistent lack of comprehensive outcome reporting over a 10-year period, which echoes findings from the review on reporting in adolescent depression trials.Reference Monsour, Mew, Patel, Chee-a-tow, Saeed and Santos34 Prior research has demonstrated that journal endorsement of CONSORT guidelines are beneficial in improving reporting of RCTs.Reference Devereaux, Manns, Ghali, Quan and Guyatt42–Reference Plint, Moher, Morrison, Schulz, Altman and Hill44 Given that our study has revealed deficiencies in outcome reporting, in particular, the rationale for outcome selection and definition of clinically meaningful change, journals are recommended to incorporate the CONSORT-Outcomes guideline for reporting outcomes in published trials in the editorial process. Specifically, journals may endorse the CONSORT-Outcomes statement, recommend authors and peer reviewers to follow these guidelines when preparing materials or reviewing manuscripts for publication and/or require submission of the CONSORT-Outcomes checklist by authors.Reference Turner, Shamseer, Altman, Schulz and Moher43
In conclusion, we found substantial variability in the reporting of primary outcomes across published geriatric depression RCTs. Omission of key details regarding trial outcomes may impede interpretation, replicability and eventual aggregation of trials through knowledge synthesis products that inform clinical guidelines and guide evidence-based decision-making. There is a need for trialists to understand patient perspectives on clinically meaningful outcomes in geriatric depression, and to adhere to outcome reporting guidelines such as the CONSORT-Outcomes statement, when reporting findings from geriatric MDD RCTs.
Supplementary material
Supplementary material is available online at https://doi.org/10.1192/bjo.2023.650
Data availability
The authors confirm that the data supporting the findings of this study are available within the article and/or its supplementary materials.
Acknowledgements
The authors wish to acknowledge Zuhayr Syed, who assisted with screening articles for the study.
Author contributions
M.R. contributed to the conception and design of the study and study protocol, screening of the articles for inclusion, data extraction and synthesis, writing of the first and subsequent drafts of the manuscript, resolved discrepancies in reporting assessments, interpreted the results and assisted in development of the search strategy and data collection tool. A.O. and K.J. conducted reporting quality assessments and contributed to the interpretation of results. P.G. contributed to data extraction and synthesis, and provided critical revision and review of the final manuscript. S.S. contributed critically to the development of the search strategy and final review of the manuscript. N.S., A. Dufort, B.P., A. D'Elia and S.P. screened articles for inclusion and provided critical revision and review of the final manuscript. Z.S. and L.T. contributed to the conception and design of the study, and provided critical revision and approval of the final manuscript. All authors read and approved the final manuscript.
Funding
This work is not funded by a specific grant. M.R. is supported by the Ontario Graduate Scholarship (OGS; 2021–2023) and the Research Institute of St. Joseph's Studentship Award (2023–2024). Z.S. received funding from Alternate Funding Plan (grant number 20-10178-480925-75153) and Canadian Institutes for Health Research (award number PJT-156306).
Declaration of interest
None.
Open access
eLetters
No eLetters have been published for this article.