Comorbidity and health-related quality of life in people with a chronic medical condition in randomised clinical trials: An individual participant data meta-analysis (original) (raw)

• Journal List
• PLoS Med
• v.20(1); 2023 Jan
• PMC9844862

PLoS Med. 2023 Jan; 20(1): e1004154.

Elaine W. Butterly, Data curation, Formal analysis, Project administration, Writing – original draft, Writing – review & editing,1 Peter Hanlon, Conceptualization, Writing – review & editing,1 Anoop S. V. Shah, Conceptualization, Writing – review & editing,2 Laurie J. Hannigan, Data curation, Writing – review & editing,3 ,4 ,5 Emma McIntosh, Conceptualization, Methodology, Writing – review & editing,1 Jim Lewsey, Conceptualization, Writing – review & editing,1 Sarah H. Wild, Conceptualization, Methodology, Writing – review & editing,6 Bruce Guthrie, Conceptualization, Writing – review & editing,6 Frances S. Mair, Methodology, Writing – review & editing,1 David M. Kent, Conceptualization, Writing – review & editing,7 Sofia Dias, Conceptualization, Methodology, Writing – review & editing,8 Nicky J. Welton, Conceptualization, Methodology, Writing – review & editing,4 and David A. McAllister, Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – review & editing1 ,*

Elaine W. Butterly

1School of Health and Wellbeing, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom

Peter Hanlon

1School of Health and Wellbeing, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom

Anoop S. V. Shah

2London School of Hygiene and Tropical Medicine, London, United Kingdom

Laurie J. Hannigan

3Nic Waals Institute, Lovisenberg Diaconal Hospital, Oslo, Norway

4Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, United Kingdom

5Department of Mental Disorders, Norwegian Institute of Public Health, Oslo, Norway

Emma McIntosh

1School of Health and Wellbeing, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom

Jim Lewsey

1School of Health and Wellbeing, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom

Sarah H. Wild

6Usher Institute, University of Edinburgh, Edinburgh, United Kingdom

Bruce Guthrie

6Usher Institute, University of Edinburgh, Edinburgh, United Kingdom

Frances S. Mair

1School of Health and Wellbeing, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom

David M. Kent

7Predictive Analytics and Comparative Effectiveness Center, Tufts Medical Center/Tufts University School of Medicine, Boston, Massachusetts, United States of America

Sofia Dias

8Centre for Reviews and Dissemination, University of York, York, United Kingdom

Nicky J. Welton

4Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, United Kingdom

David A. McAllister

1School of Health and Wellbeing, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom

1School of Health and Wellbeing, College of Medical, Veterinary and Life Sciences, University of Glasgow, Glasgow, United Kingdom

2London School of Hygiene and Tropical Medicine, London, United Kingdom

3Nic Waals Institute, Lovisenberg Diaconal Hospital, Oslo, Norway

4Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, United Kingdom

5Department of Mental Disorders, Norwegian Institute of Public Health, Oslo, Norway

6Usher Institute, University of Edinburgh, Edinburgh, United Kingdom

7Predictive Analytics and Comparative Effectiveness Center, Tufts Medical Center/Tufts University School of Medicine, Boston, Massachusetts, United States of America

8Centre for Reviews and Dissemination, University of York, York, United Kingdom

Corresponding author.

I have read the journal’s policy and the authors of this manuscript have the following competing interests. NJW has received honoraria for training and masterclasses from: Association of British Pharmaceutical Industries, Campbell Ireland, Centre for Global Development, NICE International and NICE Scientific Advice, All Wales Therapeutics and Toxicology Centre, University of Leuven. NJW has delivered training for Takeda, ICON plc, and University of Galway for which payment was made to her institution. DM has received funding for this work from the Wellcome Trust.

Received 2022 Jun 24; Accepted 2022 Dec 9.

This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Supplementary Materials

S1 PRISMA checklist: PRISMA checklist. This file is a PRISMA checklist for the manuscript.

(DOCX)

GUID: A48FBD7D-DB96-454C-BE1B-23A328A1C5E3

S1 Modelling description: Detailed description of modelling. This file includes a further detailed description of the statistical modelling performed. (DOCX)

GUID: B69516CF-6C20-4E81-AA80-BEB8C1CE14F6

S1 Additional figures tables: Supplementary figures and tables. This file contains supplementary figures and tables. (DOCX)

GUID: 2B5B2459-F095-4889-86D4-768BEFD04FE1

Attachment: Submitted filename: response to reviewers.docx

GUID: 0C04AB5E-FD07-4755-B93A-4E3240F16B8D

Attachment: Submitted filename: response_to_reviewers.docx

GUID: 8A420C66-5737-4217-9A70-79A627B66748

Data Availability Statement

The data that support the findings of this study are available from Clinical Study Data Request and the Yale Open Data Access repositories but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available by directly applying to these repositories via application process to their respective independent data access committees. All data released from the respective safe havens (Clinical Study Data Request and the Yale Open Data Access) has been made available at https://github.com/ChronicDiseaseEpi/como_qol_public.

Abstract

Background

Health-related quality of life metrics evaluate treatments in ways that matter to patients, so are often included in randomised clinical trials (hereafter trials). Multimorbidity, where individuals have 2 or more conditions, is negatively associated with quality of life. However, whether multimorbidity predicts change over time or modifies treatment effects for quality of life is unknown. Therefore, clinicians and guideline developers are uncertain about the applicability of trial findings to people with multimorbidity. We examined whether comorbidity count (higher counts indicating greater multimorbidity) (i) is associated with quality of life at baseline; (ii) predicts change in quality of life over time; and/or (iii) modifies treatment effects on quality of life.

Methods and findings

Included trials were registered on the United States trials registry for selected index medical conditions and drug classes, phase 2/3, 3 or 4, had ≥300 participants, a nonrestrictive upper age limit, and were available on 1 of 2 trial repositories on 21 November 2016 and 18 May 2018, respectively. Of 124 meeting these criteria, 56 trials (33,421 participants, 16 index conditions, and 23 drug classes) collected a generic quality of life outcome measure (35 EuroQol-5 dimension (EQ-5D), 31 36-item short form survey (SF-36) with 10 collecting both). Blinding and completeness of follow up were examined for each trial.

Using trials where individual participant data (IPD) was available from 2 repositories, a comorbidity count was calculated from medical history and/or prescriptions data. Linear regressions were fitted for the association between comorbidity count and (i) quality of life at baseline; (ii) change in quality of life during trial follow up; and (iii) treatment effects on quality of life. These results were then combined in Bayesian linear models. Posterior samples were summarised via the mean, 2.5th and 97.5th percentiles as credible intervals (95% CI) and via the proportion with values less than 0 as the probability (PBayes) of a negative association.

All results are in standardised units (obtained by dividing the EQ-5D/SF-36 estimates by published population standard deviations). Per additional comorbidity, adjusting for age and sex, across all index conditions and treatment comparisons, comorbidity count was associated with lower quality of life at baseline and with a decline in quality of life over time (EQ-5D −0.02 [95% CI −0.03 to −0.01], PBayes > 0.999). Associations were similar, but with wider 95% CIs crossing the null for SF-36-PCS and SF-36-MCS (−0.05 [−0.10 to 0.01], PBayes = 0.956 and −0.05 [−0.10 to 0.01], PBayes = 0.966, respectively). Importantly, there was no evidence of any interaction between comorbidity count and treatment efficacy for either EQ-5D or SF-36 (EQ-5D −0.0035 [95% CI −0.0153 to −0.0065], PBayes = 0.746; SF-36-MCS (−0.0111 [95% CI −0.0647 to 0.0416], PBayes = 0.70 and SF-36-PCS −0.0092 [95% CI −0.0758 to 0.0476], PBayes = 0.631.

Conclusions

Treatment effects on quality of life did not differ by multimorbidity (measured via a comorbidity count) at baseline—for the medical conditions studied, types and severity of comorbidities and level of quality of life at baseline, suggesting that evidence from clinical trials is likely to be applicable to settings with (at least modestly) higher levels of comorbidity.

Trial registration

A prespecified protocol was registered on PROSPERO (CRD42018048202).

Author summary

Why was this study done?

• ➢ The EuroQol-5 dimension (EQ-5D) and SF-36 are questionnaire-based tools that combine measures of physical and mental health into overall quality of life scores. These scores are used in randomised controlled trials of drug treatments to estimate how treatments affect quality of life. These estimates then inform decisions about which treatments should be offered to people with specific conditions.
• ➢ Multimorbidity, the presence of 2 or more conditions, makes diagnosis and treatment more complex and is associated with worse quality of life in some settings.
• ➢ People with multimorbidity are underrepresented in clinical trials, and little is known about whether and how multimorbidity changes quality of life in clinical trials. This makes it difficult for clinicians and clinical guideline developers to determine how results from clinical trials should be applied to people with multimorbidity.

What did the researchers do and find?

• ➢ To address this uncertainty, we re-analysed data from existing clinical trials. Among 33,421 participants in 56 trials of new treatments for 16 different medical conditions, we used data on medication usage and medical histories to produce a comorbidity count (higher counts indicating more multimorbidity) for each individual.
• ➢ We then used statistical models to examine associations for comorbidity counts finding that higher comorbidity counts were associated with worse quality of life at trial entry and predicted a more rapid decline in quality of life over the course of the trial. However, having a higher comorbidity count did not change the effect of treatment on quality of life.

What do these findings mean?

• ➢ These findings suggest that where treatments improve quality of life for participants overall, they are similarly likely to do so for people with multimorbidity.
• ➢ Whether this is true for individuals with higher numbers of comorbidities or with conditions and treatments not included in our analyses remains uncertain.
• ➢ Nonetheless, our findings help inform clinical decision-making by reassuring clinicians, health economists, and guideline developers that overall trial results can be used when considering how best to manage many people with multimorbidity.

Introduction

Measures of health-related quality of life evaluate the impact of treatments in ways that matter to patients [1], and, by informing economic models, are important underpinnings of clinical guideline recommendations and therefore clinical practice. Generic quality of life measures, such as the EuroQol-5 dimension (EQ-5D) [2] and the 36-item short form survey (SF-36) [3] are valid across all conditions. This means they can be used to prioritise treatments even across different conditions. Consequently, quality of life measures are often included in randomised clinical trials (hereafter trials).

Multimorbidity, where individuals have 2 or more health conditions, is common and important as it is strongly associated with higher mortality and hospitalisation rates [4,5], and treatments that have been shown in clinical trials to improve quality of life are prescribed to people with multimorbidity less frequently than those without any comorbidities [6,7]. Cross-sectional associations between multimorbidity and quality of life have been reported [8–22]. Health-related quality of life has been identified as an essential core outcome in multimorbidity research [23]. Despite this, few studies have examined whether multimorbidity predicts change in quality of life over time [24–29], and whether the effect of treatments on quality of life differ according to the presence and extent of multimorbidity is unknown. Consequently, clinicians, regulatory agencies, and guideline developers are uncertain as to the applicability of trial quality of life findings for people with multimorbidity.

In previous individual participant data (IPD) analyses from a set of 124 clinical trials ranging across a number of index conditions and treatment comparisons, we have previously shown that multimorbidity, although underrepresented is present among trial participants and predicts increased rates of both serious adverse events and trial attrition [5,30,31]. Using the 56 trials from this set for which measures of quality of life are available, we now aim to determine whether comorbidity count, which is increased in multimorbidity, predicts change in quality life, and whether treatment effects on quality of life differ by comorbidity count at baseline.

Methods

This study is reported as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline (S1 PRISMA Checklist).

Design

We performed a meta-analysis of trial IPD to determine the associations between comorbidity count and (i) quality of life at baseline; (ii) change in quality of life at trial follow-up; and (iii) the effect of treatment on change in quality of life at trial follow-up. Each analysis was done in 2 stages to account for the fact that trial data were stored securely on separate platforms and could not be analysed together in a single model. In the first stage, associations for individual trials were modelled within each secure trial analysis platform. In the second stage, the coefficients (and standard errors) from the first stage were meta-analysed using Bayesian linear models. We used a method that allowed partial pooling across index conditions and drug treatment comparisons in order to obtain overall drug treatment comparison specific and index-condition specific estimates of these associations.

Trial inclusion

Industry sponsored randomised trials with available IPD were identified using a prespecified protocol (Prospero CRD42018048202) [32]. The full selection process and analysis plan has been documented previously [30].

In brief, the United States clinical trial registry at clinicaltrials.gov [33] was searched for eligible studies meeting the following criteria; randomised phase 2/3, 3 or 4 trials studying preselected drug classes used to treat or prevent 23 selected long-term medical index conditions, registered from January 1990 until the dates of initial IPD access on 21 November 2016 and 18 May 2018, for Clinical Study Data Request (CSDR) and Yale University Open Data Access (YODA) repositories, respectively. Trials were excluded where they examined treatment for neoplastic, infectious, affective, psychotic, or developmental disorders. Trials were included if they were registered to clinicaltrials.gov [33] and had IPD availability within CSDR or YODA IPD repositories at the time of the original data request to these platforms. To make efficient use of analyst time (both ours and that of the trial sponsors who anonymised the IPD), included trials were limited to those with ≥300 participants. Given our focus on comorbidity, trials were excluded if they limited participants to those aged less than 60 years old. For the current analysis, this set of trials was restricted to those with at least 1 generic health-related quality of life measure. As our aim in the current analysis was to examine the impact across a range of index conditions, we did not include trials which only included condition-specific measures of health-related quality of life (e.g., St. George’s Respiratory Questionnaire). Included trials were categorised based on index medical condition and the World Health Organization Anatomic Therapeutic Chemical (ATC) drug class of the intervention drug [34].

Comorbidity data

A comorbidity count was calculated for each trial participant. Twenty-one conditions were included in the count: cardiovascular disease, chronic pain, arthritis, affective disorders, acid-related disorders, asthma/chronic obstructive pulmonary disease, diabetes mellitus, osteoporosis, thyroid disease, thromboembolic disease, inflammatory conditions, benign prostatic hyperplasia, gout, glaucoma, urinary incontinence, erectile dysfunction, psychotic disorders, epilepsy, migraine, parkinsonism, and dementia.

The methods used to derive this count have been described previously [30,35]. Briefly, where only data on concomitant medication was available, specific ATC codes were used to define conditions (e.g., use inhaled of corticosteroids indicated the presence of asthma or chronic obstructive pulmonary disease). Where medical history data was also available, specific medical dictionary for regulatory activities (MedDRA) codes were additionally used to identify individuals with any of the 21 conditions. The comorbidity counts were summed for each participant. The full list of ATC and MedDRA codes and the analysis code used to derive the comorbidity count are available at the project’s GitHub repository (https://github.com/ChronicDiseaseEpi/como_qol_public).

Outcome measures

Trials reported EQ-5D, SF-36, or both. EQ-5D is a preference-based generic health-related quality of life questionnaire with 5 domains (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) and a global measure of current health (measured via a visual analogue scale). Given EQ-5D is preference based, incorporating values for health outcomes, it can be used in cost-effectiveness analyses [36]. The domains are scored on either a 3 or 5 level scale (EQ-5D-3L or EQ-5D-5L). Since trials in our analysis included either type, but not both, we mapped each to a single EQ-5D index value (using United Kingdom population-based value sets for EQ-5D-3L and EQ-5D-5L, respectively, to obtain a single value that is comparable across all the EQ-5D trials [37]). EQ-5D index can range from negative values (reflecting a state felt to be “less preferable than death”), through 0 being “a state as bad as death” to 1 being “perfect health” [2].

The SF-36 questionnaire is also a commonly used generic quality of life instrument and includes 36 questions over 8 domains. Each domain score provides a weighted summary value of the questions within that domain. These domain scores are then summarised using a physical component score (PCS) and a mental component score (MCS). As is standard, we calculated these by first standardising the scales to z scores, then aggregating the physical and mental domains and transforming these into summary t scores ranging from 0 to 100 with higher scores denoting higher quality of life [3]. SF-36-PCS and SF-36-MCS scores are known to be correlated and therefore findings for each should not interpreted as if they provided independent verification [38].

Treatment arms

Treatment arm comparisons were specified prior to undertaking the quality of life analyses. For multi-arm trials, the most extreme arms were selected for comparison (e.g., if different dosages were used, the highest dose was compared to placebo or usual care—e.g., canagliflozin 300 mg, rather than 100 mg, versus placebo). Where placebo or usual care was included as a treatment arm, this was selected as the comparator. Otherwise, we chose the arm with the oldest treatment as the comparator arm.

Statistical analysis

Summary statistics were calculated for each index condition for the available EQ-5D and SF-36 trials including age (mean and standard deviation; SD), sex (number and %), and comorbidity count (mean and SD) and proportion with 0, 1 or 2, or more comorbidities.

Full descriptions of the modelling are provided in the supplementary (S1 Modelling description) and are described briefly below.

• (i) Association between comorbidity count and quality of life at baseline

In linear regression models, for each trial and each measure, baseline quality of life was modelled adjusting for age (per 15-year increment, which was close to the SD for most trials), sex (male versus referent group of females), and comorbidity count (per additional comorbidity). The effect measure estimates and associated standard errors for each model were then exported from the YODA and CSDR secure analysis platforms. In order to convert the measures onto a similar scale, we standardised each; we did so by dividing the estimates and standard errors by published estimates of the standard deviation (EQ-5D-index– 0.23 [39], SF-36-PCS—9.08, and SF-36-MCS– 10.16 [40]).

The effect measure estimates for comorbidity count (adjusted for age and sex) terms were then separately meta-analysed in Bayesian linear regression models. We used Bayesian models since these allowed partial pooling across index conditions and treatment comparisons and because they allowed us to obtain credible intervals for estimates at the level of index condition and treatment comparison directly from the posterior without a need for post hoc calculations. We fitted a range of meta-analyses, from the simplest where all trial-level estimates were pooled (ignoring treatment comparison and index condition), through models where there was partial pooling between either index conditions or treatment comparisons, to the most complex model where the estimates were partially pooled across both treatment comparisons and index condition. The Bayesian models were fitted using the brms package [41] in R statistical software. Samples from the posterior distribution for each association was summarised as the mean and the 2.5th and 97.5th percentiles (credible intervals) of the posterior distribution. The proportion of the distribution less than 0 (probability of negative association, i.e., Bayesian P) was also reported.

• (ii) Association between comorbidity count and change in quality of life at trial follow-up

We obtained estimates similarly for the association between comorbidity count and change in health-related quality of life from baseline to trial follow-up by fitting the same models but with final quality of life score as the outcome and a term for the baseline score in the trial-level linear regression models.

• (iii) Association between comorbidity count and the effect of treatment on change in quality of life at trial follow-up

Lastly, we obtained estimates for comorbidity count–treatment interactions for change in health-related quality of life from baseline to trial follow-up by fitting the same model as (ii) with additional terms for treatment arm and an arm–comorbidity count interaction. We repeated this model in a sensitivity analysis after excluding trials that did not demonstrate a benefit in quality of life.

In order to allow other researchers to use the treatment–covariate interaction results to inform subsequent analyses (e.g., as an informative prior), we obtained samples from the posterior. We obtained samples for index conditions and treatment comparisons included in our model, as well as for a notional new index condition and new treatment comparison not included in our model. We summarised these samples as student t-distributions. As with the main analysis, these models were fit using the brms package (S1 Modelling description for additional details).

Ethics approval and consent to participate

Ethical approval was obtained from the University of Glasgow, College of Medicine, Veterinary and Life Sciences ethics committee (200160070).

Results

Included trials

Of the 124 trials meeting our criteria [30,32], 56 provided 1 or more quality of life measures. Twenty-five trials (19,070 participants) provided EQ-5D only, 21 trials (8,595 participants) provided SF-36 only, and 10 trials provided both measures (5,756 participants) (Fig 1). One further trial that had collected EQ-5D and SF-36 was excluded as these results had been redacted by the study sponsors as part of the anonymisation process [42].

PRISMA diagram for included trials in quality of life analyses.

Flow diagram of included trials with IPD and reporting QoL measures; EQ-5D or the 36-item short form survey (SF-36). IPD, individual participant data; QoL, quality of life.

The characteristics of the 56 included trials were summarised by index condition (Table 1). The clinicaltrials.gov national clinical trial (NCT) identifiers of the included trials can be found in the Supporting information (Table A in S1 Additional figures tables). Index conditions were axial spondyloarthritis, chronic idiopathic urticaria, dementia, type 2 diabetes mellitus, pulmonary hypertension, inflammatory bowel disease, migraine, osteoporosis, Parkinson’s disease, psoriasis, psoriatic arthropathy, pulmonary fibrosis, restless leg syndrome, rheumatoid arthritis, systemic lupus erythematosus (SLE), and thromboembolism. Not all arms were analysed from multi-arm trials; therefore, although included trials were limited to those with ≥300 participants, the number of analysed participants per trial ranged from 102 to 2,568 participants. Approximately 82.1% of trials had a placebo or usual care arm as the comparator. The follow-up ranged from 6 to 104 weeks. Mean follow-up was 34.5 weeks. Thirty-seven trials and 19 trials were analysed on the CSDR and YODA platforms, respectively. The mean age of trial participants ranged from 38 years in SLE trials to 73 years in dementia trials and the percentage of male participants ranged from 6% in SLE trials to 79% in pulmonary fibrosis trials.

Table 1

Participant characteristics by condition and drug class.

As we had access to IPD, we were able to make the following assessments of trial robustness to possible biases: 94.6% were at least double blinded (double blind n = 26, triple blind n = 8, quadruple blind n = 19), there being only 3 open label trials. Participant completion ranged from 38.4% to 100%, with a mean (SD) completion of 81.1% (13.2%) (median 83.3% and IQR 71.4% to 90.6%). Given this is an IPD analysis there was no reporting bias.

Comorbidity counts

As published previously, participants with comorbidities were present in trials for all index conditions [30]. Trials of selective immunosuppressants in SLE had the lowest proportion of participants with no comorbidities (6%) and the highest proportion of participants with ≥2 comorbidities (66%). Trials of tumour necrosis factor α (TNF-α) inhibitors in rheumatoid arthritis had the lowest proportion with ≥2 comorbidities (12%). Over half of participants in trials with SLE, pulmonary fibrosis, osteoporosis, pulmonary hypertension, and chronic idiopathic urticaria as the index condition, had 2 or more comorbidities.

Comorbidity count and quality of life at baseline

The mean baseline EQ-5D index scores and SF-36-PCS and SF-36-MCS scores varied across conditions and intervention drug classes (Table 1). On average, participants in trials of TNF-α drugs in rheumatoid arthritis had the lowest mean baseline EQ-5D index value at 0.74 (0.08). The highest mean baseline EQ-5D index value was found in participants from trials in type 2 diabetes mellitus of glucagon-like peptide-1 receptor agonists and sodium glucose co-transporter 2 inhibitors 0.92 (0.09).

The lowest SF-36 component scores were seen in SLE trials of selective immunosuppressants with a mean SF-36-PCS 27.9 (2.5) and SF-36-MCS 20.68 (2.09). The highest SF-36 component scores, SF-36-PCS 66.42 (13.37) and SF-36-MCS 63.67 (13.89), were seen in psoriasis trials of interleukin inhibitors.

Across all index conditions and treatment comparisons, there was a negative association between comorbidity count and quality of life at baseline, when adjusted for age and sex, for all 3 measures: EQ-5D effect estimate −0.04 standardised units (95% CI −0.05, −0.02; Bayesian P > 0.999) per additional comorbidity; SF-36-PCS effect estimate −0.13 standardised units (95% CI −0.2, −0.06; Bayesian P > 0.999) per additional comorbidity; SF-36-MCS effect estimate −0.1 standardised units (95% CI −0.15, −0.04; Bayesian P = 0.999) per additional comorbidity.

Comorbidity count at baseline and subsequent change in quality of life

For all measures, relative to individuals without comorbidity, there was a decrease in quality of life over the course of the trial among participants with higher comorbidity counts. Table 2 shows these associations per one-unit increment in comorbidity count. The associations were similar in the simplest model where all trials were simply pooled and in the more complex models where trial was nested within index condition and/or drug class (Table 2). For EQ-5D, no model included the null (i.e., no association). For the SF-36 models, the results varied slightly by model. The 95% credible interval including the null for the model including drug class and index condition but did not include the null for the other models. However, even for the SF-36-PCS model, which had the widest 95% CI, the probability (Bayesian P value) that comorbidity count was negatively associated with quality of life (i.e., there was a smaller improvement/greater fall in quality of life in those trials where quality of life improved/worsened, respectively) was 95.6%.

Table 2

Change in quality of life measures from baseline to end of follow-up by comorbidity count.

There was no evidence of departure from linearity for the association between comorbidity count and quality of life for any of the measures (Table B in S1 Additional figures tables). This means that for a participant with a comorbidity count of 1 compared to a participant with a comorbidity count of 0 (or indeed for any given one-unit increment in comorbidity count), the increase in quality of life during the trial was 0.02 standardised units lower for EQ-5D (difference −0.02; 95% CI −0.03 to −0.01; Bayesian P value > 0.999), 0.05 units lower for SF-36-PCS, and 0.05 units lower for SF-36-MCS (Table 2). Back transforming these values to the original scales, this equates to 0.01 for the EQ-5D index, 0.48 for the SF-36-PCS, and 0.44 for the SF-36-MCS. These associations between comorbidity count and change in quality of life are approximately similar in magnitude (per 1 additional comorbidity) to the treatment effects on change in quality of life for the 56 trials (Figure A in S1 Additional figures tables). Therefore, the presence of 2 or more comorbidities has a larger effect than treatment on quality of life. The associations were similar for different index conditions (Fig 2) and treatment comparisons (Fig 3).

EQ-5D, SF-36-PCS, and SF-36-MCS: change in quality of life and treatment interaction by condition.

The association of comorbidity and change in quality of life measures with treatment interaction (orange) and without (black). Standardised effect estimates with 50%, 80%, and 95% credibility intervals per trial index condition. EQ-5D, EuroQol-5 dimension; MCS, mental component score; PCS, physical component score; SF-36, the 36-item short form survey.

EQ-5D, SF-36-PCS and SF-36-MCS: change in quality of life and treatment interaction by treatment comparison.

The association of comorbidity and change in quality of life measures with treatment interaction (orange) and without (black). Standardised effect estimates with 50%, 80%, and 95% credibility intervals per trial intervention ATC drug treatment comparisons. ATC, anatomic therapeutic chemical; EQ-5D, EuroQol-5 dimension; MCS, mental component score; PCS; physical component score; SF-36, the 36-item short form survey.

Comorbidity count at baseline and variation in the effect of treatment on quality of life during trial follow-up

There was no evidence of an interaction between comorbidity count and treatment effect on EQ-5D, SF-36-PCS, or SF-36-MCS. The probability that treatment was less effective in improving quality of life in those with a greater number of comorbidities (Bayesian P values) ranged from 0.556 to 0.749 (Table 3). The point estimates for these interactions were an order of magnitude smaller than those for comorbidity count and change in quality of life (see previous section). This was true across index conditions (Fig 2) and treatment comparisons (Fig 3), and the associations were null across all models from the simplest where all trials were pooled to the most complex where trial was nested within treatment comparison and index condition. Summarised posterior predictions (as student t-distributions) can be found in Table C in S1 Additional figures tables, for use by other researchers as informative priors for subsequent analyses. We also repeated the interaction analysis after excluding the trials (23 EQ-5D, 20 SF-36-PCS, and 23 SF-36-MCS) that did not demonstrate a benefit in quality of life; the findings were similarly null (Table D in S1 Additional figures tables).

Table 3

Variation in the effect of treatment on quality of life by comorbidity count.

Discussion

In an IPD meta-analysis of 56 clinical trials across 16 index conditions and 14 treatment comparisons, we found that a higher comorbidity count at baseline was associated with poorer quality of life at baseline (as measured by EQ-5D, SF-36-PCS, and SF-36-MCS). The higher comorbidity count at baseline also predicted less improvement in EQ-5D over time. The findings were similar, but with wider 95% CIs which in some models just included the null, for SF-36-PCS and SF-36-MCS. However, baseline comorbidity count was not associated with differences in the estimated effect of treatment on quality of life for any of these measures.

Ours is the largest study, to our knowledge, to examine whether comorbidity predicts change in quality of life over time or in response to treatment in trial participants [24–29]. However, there are several limitations. First although comorbidity must, by definition increase with increases in multimorbidity, a simple count such as we used does not capture the full complexity of multimorbidity, including issues such as interactions between concordant and discordant conditions [43]. It is plausible that more nuanced measures of multimorbidity, feasible in trials designed to collect multimorbidity information prospectively, would lead to more nuanced results as to, for example, the effect of specific combinations of conditions on quality of life and heterogeneity of treatment effects. Moreover, while there was no evidence of any departure from linearity with the range of comorbidity counts we observed, this may not be true for higher comorbidity counts. Similarly, care should be taken when applying our findings to individuals with lower quality of life pretreatment, since the quality of life scores in the trial participants were generally fairly high at baseline. Secondly, the trials analysed were only those available via the CSDR and YODA trial repositories. Not all sponsors share data via these repositories, and among sponsors who do, not all trials are shared. Therefore, our dataset is not representative of all clinical trials, even for those index conditions and treatment comparisons included. As trials were identified from the US clinical trials register (clinicaltrials.gov) rather than a database of published papers (e.g., PubMed), there was no risk of publication bias. However, this does not mean that the trials were a random or complete sample of all registered trials. As we have noted previously [44], this set of trials were broadly similar to trials where IPD was not available with respect to the set indications, phases, number of participants, start dates, and exclusion criteria. However, inflammatory bowel disease and arthritis trials and trials of immunosuppressants were overrepresented, while phase 4 trials and those with especially large enrolment sizes were underrepresented.

Thirdly, while our definitions were prespecified, based on high-quality data, and have previously been used to demonstrate associations between comorbidity and both trial withdrawal and trial serious adverse events [5,31], the trials were not designed to measure comorbidity. It is possible that stronger associations may have been found if bespoke measures had been available. Thirdly, few of the trial participants had more than 2 comorbidities. As such, care should be taken in extrapolating our findings to individuals with 3 or more conditions as well as to people with more severe comorbidities who are likely to be excluded from trials, since it is possible that these types of comorbidities do modify treatment responses. Finally, our analysis was designed to examine variation in treatment effects by comorbidity count not overall treatment effects. In our modelling, we shared information (partial pooling) across multiple index conditions and treatment comparisons, pooling more extensively than would have been appropriate had our focus been on specific treatment comparisons. Consequently, our analysis of comorbidity treatment interactions should not be used to make inferences about the overall efficacy of specific treatment comparisons, but rather about plausible variation in treatment effects by comorbidity count.

A number of previous studies have examined cross-sectional associations between multimorbidity (or the presence of multiple comorbidities and quality of life, consistently showing negative associations [8–22]). Two systematic reviews with meta-analyses demonstrated a negative association; one looked at studies in any setting (mean reduction in quality of life for each additional comorbid condition −1.55% (95% CI −2.97% to −0.13%) and −4.37% (95% CI −7.13% to −1.61%) for mental and physical quality of life, respectively) [21], and one examined studies in non-hospital settings (reduction in quality of life 3.8% to 4.1% per additional comorbid condition) [22]. Multimorbidity has also been found to be associated with lower quality of life in the USA Medical Expenditure panel survey [19], a population-based survey of cancer patients in the Netherlands [20], a cross-sectional study of 1,649 people in primary care in India [8], and among 3,256 people with spinal osteoarthritis identified from Korean national health survey data [14].

In contrast, we found only a small number of studies that examined whether comorbidity or multimorbidity at baseline predicted longitudinal change in quality of life. In a cohort study of 1,211 adults in Japan, multimorbidity at baseline predicted more rapid decline in SF-36 over 12 months [27]. Similarly, among 1,582 people undergoing total hip arthroplasty in Denmark comorbidity at baseline was associated with a smaller increase in quality of life (high comorbidity versus no comorbidity EQ-5D change 0.09; 95% CI 0.02 to 0.16) [28]. In a study comprising 351 individuals attending an Australian clinic for people with complex chronic diseases, comorbidity count predicted more rapid decline in SF-36 (−0.11 per comorbidity; −0.96 to 0.76) [29]. However, all 3 studies had substantial loss to follow-up. In a secondary analysis of a clinical trial including 379 participants with early rheumatoid arthritis, quality of life measures were regressed on multimorbidity (measured via the rheumatoid arthritis comorbidity index) using linear mixed models. SF-36-PCS was found to decrease more quickly among participants with multimorbidity but there was no association for SF-36-MCS [25]. Two studies reported multimorbidity at baseline and reported longitudinal measures of quality of life in people with head and neck cancer and prostate cancer, respectively, but neither presented effect estimates or 95% confidence intervals making it difficult to judge whether this was due to the low sample sizes [24,26]. To this literature, we add the observation that in 33,421 participants in 56 trials across 16 index conditions, baseline multimorbidity is associated with lower quality of life at baseline and—at least for the EQ-5D-index—less improvement in quality of life associated with trial participation.

While we did not undertake a formal systematic review, in a wide-ranging search including a number of terms for comorbidity, multimorbidity, and quality of life (Table E in S1 Additional figures tables), we found only 1 study that reported findings on variation of treatment effects on quality of life by comorbidity or multimorbidity. In a convenience sample of 3 trials (1 asthma, 1 heartburn, and 1 gastric ulcer disease) [45], the reported associations differed across the included trials, measures of quality of life and measures of multimorbidity. To this sparse literature, we add findings, from a large number of participants, that there was no evidence of heterogeneity of treatment effect by multimorbidity for EQ-5D, SF-36-PCS, or SF-36-MCS.

The observation that treatment effects on quality of life did not differ by multimorbidity (measured via a comorbidity count) at baseline has implications for the interpretation of clinical trials. Unless there are strong a priori reasons to believe that an effect of a treatment on quality of life will be differential by comorbidity (e.g., a trial of a diabetes drug given by injection in a device that might be more difficult for people with arthritis to use may lead to differential efficacy in those people), reports of heterogeneity from individual trials and small meta-analyses should most likely be considered chance findings. Similarly, treatments that involve complex regimes may also differentially impact quality of life according to the participant’s comorbidity count. For example, individuals with more comorbidities (and hence greater polypharmacy) may find additional treatments either more or less burdensome than individuals with fewer comorbidities. Our findings imply that treatment efficacy estimates for EQ-5D, SF-36-PCS, and SF-36-MCS from clinical trials are also likely to be applicable to settings with (at least modestly) higher levels of comorbidity, at least for the kinds of conditions and drug classes covered in our analysis.

Based on our models, for each outcome, we produced posterior predictions for comorbidity–treatment interactions for an unobserved drug class and index condition (i.e., one not included in our models). These distributions represent an estimate of the likely variation in treatment effect according to comorbidity count before seeing the trial data for any given specific treatment comparison. We had originally intended to repeat the modelling after log-transforming quality of life measures. However, since there were no interactions on the linear scale (which is standard for these measures), we did not examine the effect of changing scale. As such, these distributions can be used as informative priors for subsequent meta-analyses examining treatment effects in people with comorbidity, improving the reliability and precision of the resultant estimates. The prior distributions we produced can also be used to inform choices in probabilistic health economic models that are used to apply trial findings to “real-world” populations, such as in Health Technology Assessments [46]. For example, if a trial was conducted among participants with fewer comorbidities than patients in the target population, our findings can be used to model the likely effect of such treatments on quality of life in real-world settings. We observed a decline in quality of life in participants with comorbidities, relative to other participants. This finding provides new evidence to suggest that the association between comorbidity and quality of life is causal. Most previous studies reporting associations between multimorbidity or comorbidity and quality of life have been cross-sectional [8–22], and the few longitudinal studies have been limited by small sizes [24–26], limited numbers of baseline conditions, or substantial loss to follow-up [27–29]. Our findings therefore strengthen the evidence for causation between comorbidity and baseline and subsequent change in quality of life.

There are a number of possible mechanisms for the relative decline in quality of life among people with multiple comorbidities. First, the decline may be a direct result of the underlying conditions (alone or in combination); symptoms such as pain, impaired sleep, limited mobility, and impaired function may increasingly adversely impact quality of life over time. Alternatively, the (non-treatment related) benefits of trial participation such as improved access to clinical care may be different in people with and without comorbidity. Similarly, attending visits, undergoing procedures, and following treatment regimens may also impose a greater treatment burden among people with comorbidity. This concept of trial participation burden is analogous to the treatment burden (visits, drug regimens, etc.) described in routine clinical practice [47], which is known to be more challenging for people with multimorbidity.

In these clinical trials, higher comorbidity count is associated with lower quality of life at baseline and predicts subsequent relative decline in EQ-5D (and most likely SF-36) over time. However, the effect of treatment on quality of life does not differ by comorbidity count at baseline. Trial-derived estimates for treatment effects for quality of life are likely to be applicable to people with moderate numbers of comorbidities.

Supporting information

S1 PRISMA checklist

PRISMA checklist.

This file is a PRISMA checklist for the manuscript.

(DOCX)

S1 Modelling description

Detailed description of modelling. This file includes a further detailed description of the statistical modelling performed.

(DOCX)

S1 Additional figures tables

Supplementary figures and tables. This file contains supplementary figures and tables.

(DOCX)

Acknowledgments

This study, carried out under YODA Project # 2017–1746, used data obtained from the Yale University Open Data Access Project, which has an agreement with JANSSEN RESEARCH & DEVELOPMENT, L.L.C. The interpretation and reporting of research using this data are solely the responsibility of the authors and does not necessarily represent the official views of the Yale University Open Data Access Project or JANSSEN RESEARCH & DEVELOPMENT, L.L.C. This study was also carried out under ClinicalStudyDataRequest.com project number 1732, used data from the ClinicalStudyDataRequest.com repository, who provided data from Boehringer-Ingelheim, GSK, Lilly, Roche, Takeda, and Sanofi. The interpretation and reporting of research using these data are solely the responsibility of the authors and does not necessarily represent the official views of ClinicalStudyDataRequest.com or Boehringer-Ingelheim, GSK, Lilly, Roche, Takeda, or Sanofi.

Abbreviations

Funding Statement

DM is funded via an Intermediate Clinical Fellowship and Beit Fellowship from the Wellcome Trust, who also supported other costs related to this project such as data access costs and database licenses (“Treatment effectiveness in multimorbidity: Combining efficacy estimates from clinical trials with the natural history obtained from large routine healthcare databases to determine net overall treatment Benefits.” - 201492/Z/16/Z. SD is supported by the Medical Research Council [grant no. MR/R025223/1]. LH is supported by the South-Eastern Norway Regional Health Authority (#2020023, #2019097). PH is funded through a Clinical Research Training Fellowship from the Medical Research Council (Grant reference: MR/S021949/1). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability

The data that support the findings of this study are available from Clinical Study Data Request and the Yale Open Data Access repositories but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available by directly applying to these repositories via application process to their respective independent data access committees. All data released from the respective safe havens (Clinical Study Data Request and the Yale Open Data Access) has been made available at https://github.com/ChronicDiseaseEpi/como_qol_public.

References

1. Sherbourne CD, Sturm R, Wells KB. What Outcomes Matter to Patients?. J Gen Intern Med. 1999;14:357–363. doi: 10.1046/j.1525-1497.1999.00354.x[PMC free article] [PubMed] [CrossRef] [Google Scholar]

3. Ware JEJ. SF-36 Health Survey Update. Spine. 2000;25:3130–3139. doi: 10.1097/00007632-200012150-00008 [PubMed] [CrossRef] [Google Scholar]

4. Wen Y-C, Chen L-K, Hsiao F-Y. Predicting mortality and hospitalization of older adults by the multimorbidity frailty index. PLoS ONE. 2017;12:e0187825. doi: 10.1371/journal.pone.0187825[PMC free article] [PubMed] [CrossRef] [Google Scholar]

5. Hanlon P, Butterly E, Shah AS, Hannigan LJ, Wild SH, Guthrie B, et al. Assessing trial representativeness using Serious Adverse Events: An observational analysis using aggregate and individual-level data from clinical trials and routine healthcare data. BMC Med. 2022. doi: 10.1186/s12916-022-02594-9[PMC free article] [PubMed] [CrossRef] [Google Scholar]

6. Smith DJ, Martin D, McLean G, Langan J, Guthrie B, Mercer SW. Multimorbidity in bipolar disorder and undertreatment of cardiovascular disease: a cross sectional study. BMC Med. 2013;11:263. doi: 10.1186/1741-7015-11-263[PMC free article] [PubMed] [CrossRef] [Google Scholar]

7. Wang PS, Avorn J, Brookhart MA, Mogun H, Schneeweiss S, Fischer MA, et al. Effects of noncardiovascular comorbidities on antihypertensive use in elderly hypertensives. Hypertens Dallas Tex. 1979;2005(46):273–279. doi: 10.1161/01.HYP.0000172753.96583.e1 [PubMed] [CrossRef] [Google Scholar]

8. Pati S, Swain S, Knottnerus JA, Metsemakers JFM, van den Akker M. Health related quality of life in multimorbidity: a primary-care based study from Odisha. India Health Qual Life Outcomes. 2019;17:116. doi: 10.1186/s12955-019-1180-3[PMC free article] [PubMed] [CrossRef] [Google Scholar]

9. Fortin M, Dubois M-F, Hudon C, Soubhi H, Almirall J. Multimorbidity and quality of life: a closer look. Health Qual Life Outcomes. 2007;5:52. doi: 10.1186/1477-7525-5-52[PMC free article] [PubMed] [CrossRef] [Google Scholar]

10. Fortin M, Lapointe L, Hudon C, Vanasse A, Ntetu AL, Maltais D. Multimorbidity and quality of life in primary care: a systematic review. Health Qual Life Outcomes. 2004;2:51. doi: 10.1186/1477-7525-2-51[PMC free article] [PubMed] [CrossRef] [Google Scholar]

11. Brettschneider C, Leicht H, Bickel H, Dahlhaus A, Fuchs A, Gensichen J, et al. Relative Impact of Multimorbid Chronic Conditions on Health-Related Quality of Life–Results from the MultiCare Cohort Study. PLoS ONE. 2013;8:e66742. doi: 10.1371/journal.pone.0066742[PMC free article] [PubMed] [CrossRef] [Google Scholar]

12. Chen W, Lynd LD, FitzGerald JM, Marra CA, Rousseau R, Sadatsafavi M. The added effect of comorbidity on health-related quality of life in patients with asthma. Qual Life Res Int J Qual Life Asp Treat Care Rehabil. 2015;24:2507–2517. doi: 10.1007/s11136-015-0995-6 [PubMed] [CrossRef] [Google Scholar]

13. Fraser SD, Barker J, Roderick PJ, Yuen HM, Shardlow A, Morris JE, et al. Health-related quality of life, functional impairment and comorbidity in people with mild-to-moderate chronic kidney disease: a cross-sectional study. BMJ Open. 2020;10:e040286. doi: 10.1136/bmjopen-2020-040286[PMC free article] [PubMed] [CrossRef] [Google Scholar]

14. Kim S-K, Choe J-Y. Comorbidities and Health-Related Quality of Life in Subjects with Spine Osteoarthritis at 50 Years of Age or Older: Data from the Korea National Health and Nutrition Examination Survey. Medicina (Kaunas).2022;58:126. doi: 10.3390/medicina58010126[PMC free article] [PubMed] [CrossRef] [Google Scholar]

15. Adams RJ, Wilson DH, Taylor AW, Daly A, Tursan d’Espaignet E, Dal Grande E, et al. Coexistent chronic conditions and asthma quality of life: a population-based study. Chest. 2006;129:285–291. doi: 10.1378/chest.129.2.285 [PubMed] [CrossRef] [Google Scholar]

16. Al-Asadi AM, Klein B, Meyer D. Multiple comorbidities of 21 psychological disorders and relationships with psychosocial variables: a study of the online assessment and diagnostic system within a web-based population. J Med Internet Res. 2015;17:e55. doi: 10.2196/jmir.4143[PMC free article] [PubMed] [CrossRef] [Google Scholar]

17. Millá-Perseguer M, Guadalajara-Olmeda N, Vivas-Consuelo D, Usó-Talamantes R. Measurement of health-related quality by multimorbidity groups in primary health care. Health Qual Life Outcomes. 2019;17:8. doi: 10.1186/s12955-018-1063-z[PMC free article] [PubMed] [CrossRef] [Google Scholar]

18. Nelis SM, Wu Y-T, Matthews FE, Martyr A, Quinn C, Rippon I, et al. The impact of co-morbidity on the quality of life of people with dementia: findings from the IDEAL study. Age Ageing. 2019;48:361–367. doi: 10.1093/ageing/afy155[PMC free article] [PubMed] [CrossRef] [Google Scholar]

19. Sullivan PW, Ghushchyan VH, Bayliss EA. The impact of co-morbidity burden on preference-based health-related quality of life in the United States. Pharmacoeconomics. 2012;30:431–442. doi: 10.2165/11586840-000000000-00000 [PubMed] [CrossRef] [Google Scholar]

20. Vissers PAJ, Thong MSY, Pouwer F, Zanders MMJ, Coebergh JWW, van de Poll-Franse LV. The impact of comorbidity on Health-Related Quality of Life among cancer survivors: analyses of data from the PROFILES registry. J Cancer Surviv Res Pract. 2013;7:602–613. doi: 10.1007/s11764-013-0299-1 [PubMed] [CrossRef] [Google Scholar]

21. Makovski TT, Schmitz S, Zeegers MP, Stranges S, van den Akker M. Multimorbidity and quality of life: Systematic literature review and meta-analysis. Ageing Res Rev. 2019;53:100903. doi: 10.1016/j.arr.2019.04.005 [PubMed] [CrossRef] [Google Scholar]

22. Schmitz S, Makovski TT, Adams R, van den Akker M, Stranges S, Zeegers MP. Bayesian Hierarchical Models for Meta-Analysis of Quality-of-Life Outcomes: An Application in Multimorbidity. Pharmacoeconomics. 2020;38:85–95. doi: 10.1007/s40273-019-00843-z [PubMed] [CrossRef] [Google Scholar]

23. Smith SM, Wallace E, Salisbury C, Sasseville M, Bayliss E, Fortin M. A Core Outcome Set for Multimorbidity Research (COSmm). Ann Fam Med. 2018;16:132–138. doi: 10.1370/afm.2178[PMC free article] [PubMed] [CrossRef] [Google Scholar]

24. Daskivich TJ, van de Poll-Franse LV, Kwan L, Sadetsky N, Stein DM, Litwin MS. From bad to worse: comorbidity severity and quality of life after treatment for early-stage prostate cancer. Prostate Cancer Prostatic Dis. 2010;13:320–327. doi: 10.1038/pcan.2010.33 [PubMed] [CrossRef] [Google Scholar]

25. Stouten V, Westhovens R, De Cock D, Van der Elst K, Pazmino S, Bertrand D, et al. Having a co-morbidity predicts worse outcome in early rheumatoid arthritis despite intensive treatment: a post hoc evaluation of the pragmatic randomized controlled CareRA trial. Rheumatology (Oxford).2021;60:3699–3708. doi: 10.1093/rheumatology/keaa841 [PubMed] [CrossRef] [Google Scholar]

26. Oozeer NB, Benbow J, Downs C, Kelly C, Welch A, Paleri V. The effect of comorbidity on quality of life during radiotherapy in head and neck cancer. Otolaryngol Head Neck Surg. 2008;139:268–272. doi: 10.1016/j.otohns.2008.05.027 [PubMed] [CrossRef] [Google Scholar]

27. Aoki T, Fukuhara S, Fujinuma Y, Yamamoto Y. Effect of multimorbidity patterns on the decline in health-related quality of life: a nationwide prospective cohort study in Japan. BMJ Open. 2021;11:e047812. doi: 10.1136/bmjopen-2020-047812[PMC free article] [PubMed] [CrossRef] [Google Scholar]

28. Glassou EN, Pedersen AB, Aalund PK, Mosegaard SB, Hansen TB. Is gain in health-related quality of life after a total hip arthroplasty depended on the comorbidity burden? Acta Orthop. 2018;89:374–379. doi: 10.1080/17453674.2018.1457885[PMC free article] [PubMed] [CrossRef] [Google Scholar]

29. Tyack Z, Frakes K, Barnett A, Cornwell P, Kuys S, McPhail S. Predictors of health-related quality of life in people with a complex chronic disease including multimorbidity: a longitudinal cohort study. Qual Life Res. 2016;25:2579–2592. doi: 10.1007/s11136-016-1282-x [PubMed] [CrossRef] [Google Scholar]

30. Hanlon P, Hannigan L, Rodriguez-Perez J, Fischbacher C, Welton NJ, Dias S, et al. Representation of people with comorbidity and multimorbidity in clinical trials of novel drug therapies: an individual-level participant data analysis. BMC Med. 2019;17:201. doi: 10.1186/s12916-019-1427-1[PMC free article] [PubMed] [CrossRef] [Google Scholar]

31. Lees JS, Hanlon P, Butterly EW, Wild SH, Mair FS, Taylor RS, et al. Effect of age, sex, and morbidity count on trial attrition: meta-analysis of individual participant level data from phase 3/4 industry funded clinical trials. BMJ Med. 2022:1. doi: 10.1136/bmjmed-2022-000217 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

35. Hanlon P, Jani BD, Butterly E, Nicholl B, Lewsey J, McAllister DA, et al. An analysis of frailty and multimorbidity in 20,566 UK Biobank participants with type 2 diabetes. Commun Med. 2021;1:1–9. doi: 10.1038/s43856-021-00029-9[PMC free article] [PubMed] [CrossRef] [Google Scholar]

36. Neumann PJ, Goldie SJ, Weinstein MC. Preference-based measures in economic evaluation in health care. Annu Rev Public Health. 2000;21:587–611. doi: 10.1146/annurev.publhealth.21.1.587 [PubMed] [CrossRef] [Google Scholar]

38. Taft C, Karlsson J, Sullivan M. Do SF-36 Summary Component Scores Accurately Summarize Subscale Scores? Qual Life Res. 2001;10:395–404. doi: 10.1023/a:1012552211996 [PubMed] [CrossRef] [Google Scholar]

40. Garratt AM, Stavem K. Measurement properties and normative data for the Norwegian SF-36: results from a general population survey. Health Qual Life Outcomes. 2017;15:51. doi: 10.1186/s12955-017-0625-9[PMC free article] [PubMed] [CrossRef] [Google Scholar]

43. Overview | Multimorbidity: clinical assessment and management | Guidance | NICE. [cited 2022 Oct 14]. Available from: https://www.nice.org.uk/guidance/ng56.

45. Xuan J, Kirchdoerfer LJ, Boyer JG, Norwood GJ. Effects of comorbidity on health-related quality-of-life scores: an analysis of clinical trial data. Clin Ther. 1999;21:383–403. doi: 10.1016/S0149-2918(00)88295-8 [PubMed] [CrossRef] [Google Scholar]

46. Dias S, Welton NJ, Sutton AJ, Ades AE. Evidence Synthesis for Decision Making 1: Introduction. Med Decis Making. 2013;33:597–606. doi: 10.1177/0272989X13487604[PMC free article] [PubMed] [CrossRef] [Google Scholar]

47. Mair FS, May CR. Thinking about the burden of treatment. BMJ. 2014;349:g6680. doi: 10.1136/bmj.g6680 [PubMed] [CrossRef] [Google Scholar]

2023 Jan; 20(1): e1004154.

Decision Letter 0

This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

29 Jun 2022

Dear Dr McAllister,

Thank you for submitting your manuscript entitled "Comorbidity and health related quality of life in randomised clinical trials: an individual participant data meta-analysis" for consideration at PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff as well as by an academic editor with relevant expertise and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by Jul 01 2022 11:59PM.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at gro.solp@enicidemsolp if you have any queries relating to your submission.

Kind regards,

Dr Philippa Claire Dodd, MBBS MRCP PhD

Senior Editor

PLOS Medicine

2023 Jan; 20(1): e1004154.

Decision Letter 1

This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

26 Sep 2022

Dear Dr. McAllister,

Thank you very much for submitting your manuscript "Comorbidity and health related quality of life in randomised clinical trials: an individual participant data meta-analysis" (PMEDICINE-D-22-02156R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at gro.solp@enicideMSOLP.

We expect to receive your revised manuscript by Oct 17 2022 11:59PM. Please email us (gro.solp@enicidemsolp) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Philippa Dodd, MBBS MRCP PhD

PLOS Medicine

gro.solp@ddodp

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

GENERAL

As per our email correspondence, please include the code used to determine mortality scores. See also reviewer comments below.

Please report your SR/MA according to the PRISMA guidelines provided at the EQUATOR site. http://www.equator-network.org/reporting-guidelines/prisma/

Please provide the completed PRISMA checklist. When completing the checklist, please use section and paragraph numbers, rather than page numbers.

Please add the following statement, or similar, to the Methods: "This study is reported as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline (S1 Checklist)."

Your search end dates are not clear (until one reaches the suppl files) please include in the abstract & main manuscript (see below)

Please update your search to the present time (within the last 6 months, if not already).

Please embed abbreviations within the main manuscript text at the time the abbreviation is first referred to and remove from the end of the manuscript. Please ensure inclusion of abbreviations where relevant in all tables/figures

Please remove consent/data availability statement/competing interests statement from the end of the manuscript and include in the manuscript submission form instead

ABSTRACT

Please report your abstract according to PRISMA for abstracts, following the PLOS Medicine abstract structure (Background, Methods and Findings, Conclusions) http://www.plosmedicine.org/article/info:doi/10.1371/journal.pmed.1001419 .

Please combine the Methods and Findings sections into one section, “Methods and findings”.

Abstract Methods and Findings:

Please ensure that all numbers presented in the abstract are present and identical to numbers presented in the main manuscript text.

Please provide the dates of search, data sources, types of study designs included (specify RCT results line 1)

Please quantify the main results p values, as well as with 95% CIs.

Please include the important dependent variables that are adjusted for in the analyses.

In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

AUTHOR SUMMARY

At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

INTRODUCTION

Line 69: Please replace “background” with “Introduction”

METHODS and RESULTS

Please quantify the main results p values, as well as with 95% CIs, where relevant

Where p-values are reported, please also provide the statistical tests used to determine them

Please report the end date of your search - line 108: “…from January 1990 onwards…”

Please evaluate study quality and risk of bias.

Please evaluate evidence of publication bias.

Please clearly report how you searched for and selected published studies including the search platforms

Please ensure that your ethics statement is included in the manuscript methods section, as stated in the submission form, and remove from the end of the manuscript

Line 106: “…(Prospero CRD42018048202).(30) The full selection process and analysis plan has been documented previously. (31)” please include the analysis plan with the manuscript, see also reviewer comments below

DISCUSSION

Thank you for organizing the discussion largely as follows: a short, clear summary of the article's findings; what the study adds to existing research and where and why the results may differ from previous research; strengths and limitations of the study; implications and next steps for research, clinical practice, and/or public policy. Please remove the sub-headings from the discussion and place the conclusion as the final paragraph (removing the heading “conclusion”).

FIGURES

Thank you for including the figure captions and the figures. Please also place these within the manuscript results section

I agree with the reviewer comments below that the PRISMA flowchart would be a helpful addition to the main manuscript.

Please consider avoiding the use of red (and green) to make the figures accessible to those with colour blindness

TABLES

Table 2 – please also provide unadjusted analyses

REFERENCES

Please select the PLOS Medicine reference style in your citation manager. In-text reference call outs Citations should be in square brackets, and preceding punctuation, note the absence of spaces within the square brackets, “…symptomatic [2,8].”

In the bibliography please ensure no more than 6 authors are listed before et al where more than n6 authors contribute

Journal name abbreviations should be those found in the National Center for Biotechnology Information (NCBI) databases.

Comments from the reviewers:

Reviewer #1: Thanks for the opportunity to review your manuscript. My role is as a statistical reviewer, so my review concentrates on the study design, data, and analysis that are presented. I have put general questions first, followed by queries relevant to a specific section of the manuscript (with a page/line reference).

The manuscript presents and IPD meta-analysis, of data drawn from pharma sponsored RCTs that collected both a HRQOL measure and a comorbidity count. The data is used to assess two questions, 1) whether higher comorbidity count is associated with greater declines in HRQOL over time (from baseline of study to a follow-up point), and 2) whether there is an interaction between treatment effects (on HRQOL) and baseline comorbidity (i.e. a subgroup analysis for HRQOL outcome). A two-stage approach (analysis on each trial, then synthesis of the estimates from each trial) was used. This approach was used over a one-stage (multilevel model) approach because of the availability of trial data stored across different platforms. The relevant coefficients and SEs were estimated with linear regression, then in the second synthesis stage, these were pooled using a Bayesian LM. The 'ANCOVA' approach to estimating change from baseline is used for the interaction between treatment effects and baseline comorbidity. Clearly a lot of work went into doing the first stage of the analysis.

The study is registered on PROSPERO, and the approach detailed in this manuscript matches the registration. One thing I was curious about was whether you kept any data about how comorbidities were recorded across all of the trials. My own experience suggests a high degree of heterogeneity when you go from one therapeutic area to another (e.g. renal trialists seem to try to capture everything), and wondered if this might account for some of the differences in comorbidity counts by type of disease.

Is it possible to see a copy of the code that derives the comorbidity count? Usually during the review process names are concealed so granting a view request might not work, but is it too unwieldy to attach the syntax.

I would consider some adjustment to the title if there is space, in particular to reflect that the study is focused on drug trials of chronic disease patients.

With only a limited number of data sets available for IPD, do you consider the results to be generalizable to multi-morbidity in all drug RCTs?

P6, L108. One important detail I would add here from the PROSPERO protocol was the disease selection criteria i.e. that ID, cancer etc. are excluded

P9, L172. To clarify, this means that age was rescaled to be in units of SD?

P10, L174. Were any checks of residuals from this stage done? (I can understand that this would be a pain with so many trials to check). What criteria was used to decide that the polynomial (squared) term of # comorbidities was consistent with linearity? A second order polynomial is a decent way to account for non-linearity but it won't always work in my experience.

P12, L252. I think this is an accurate statement about the main results, one thing I struggled with was translating what the coefficient (i.e. change in HRQOL with one extra comorbidity) translates to. I didn't have a sense of whether this was a very dramatic change or a minor one. I am not sure if this can be changed easily - and might just reflect that I don't usually work with HRQOL outcomes more than any problem with the manuscript.

P13, L280. This is the posterior probability of a negative interaction beta coefficient?

Supp Appendix. I would describe the prior for the overall intercept to be 'weakly informative' (SD=10), but it looks like the trial/cond/comp priors are more strongly informative (SD=1). What was the rationale of this choice (or have I misunderstood this part?)?

Reviewer #2: Thanks for the opportunity to review this IPD that aims to improve understanding about the impact of multimorbidity in HRQoL (and whether multimorbidity affects treatment effects on HRQoL).

It is a policy relevant piece of research that has potential to contribute to understanding about the causal relationship between multimorbidity and QoL.

I don't have any major reservations about the core methods deployed and these are mainly well described.

However I was struck by the conceptual inconsistency that perhaps undermines the thrust of the analyses and the arguments off the back of the findings.

1. In the Introduction the study is contextualised in relation to multimorbidity, characterised as two or more conditions. But in the methods the authors pivot to talk about comorbidity and this seems to represent significant conceptual slippage that is not adequately addressed. Their approach is fundamentally about co-morbidity where there is an index condition with an intention to then add up numbers of comorbidities, which essentially assumes comorbidity is additive. But this approach (as discussed by Ng et al doi: 10.1093/ije/dyy134 [and others]) does not account for multimorbidity that occurs by chance and is reliant on the range of illnesses in the dataset. It does not take into account the way certain illness might cluster and have syntergistic affects on health outcomes and as such I wonder if the authors can address this and refer to literature on multimorbidity clusters, to more adequately defend their approach. See for example: Sinnige et al. doi:10.1371/journal.pone.0079641 and Busija https://doi.org/10.1007/s10654-019-00568-5; Prados-Torros http://dx.doi.org/10.1016/j.jclinepi.2013.09.021

2. The number of index conditions (21) seems quite small compared to previous epidemiological studies of multimorbidity e.g. Barnett et al. https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(12)60240-2/fulltext

3. Excluding trials with populations younger than 60 risks excluding populations with multimorbidity from socio-economically disadvantaged backgrounds where multimorbidity is known to affect people 10 years younger (see for example https://www.thelancet.com/journals/lanhl/article/PIIS2666-7568(21)00146-X/fulltext)

4. Additionally, the list of index conditions excludes mental health problems, both common and serious (if we exclude dementia as not belonging to that family of clinical health problems). This seems to be a major omission given the prevalence of mental and physical multimorbidity and this should be explained and made clear in the title or at least the main description of the study.

Reviewer #3: Thank you for asking me to review this paper that read with considerable interest. I think there is a very important message here that has the potential to impact on important aspects of trial design and interpretation. That is that factors that predict outcome do not necessarily predict response to treatment. This means we have some reassurance that, at least in the context of the trials studied here, that effects observed in people with fewer co-morbidities are likely to be generalisable and can be applied clinically to those with more co-morbidities.

I look forward to the team repeating these analyses, subject to data availability, for other baseline factors measured in clinical trials such as sex/gender, ethnicity, age, specific co-morbidities (e.g. depression). For the record I would still have liked the paper even if the results were different or inconclusive.

I have reviewed this paper from the perspective of a clinical trialist with experience of IPD meta-analysis for sub-group effects. Whilst I think a Bayesian approach is appropriate for work I am not competent to comment on the actual methods used. This would require specialist statistical review.

I have some, largely minor comments

1. My understanding is that the Euroqol group do not consider EQ-5D to be an abbreviation and that this is the full name of the measure and so no need to define this as Euroqol-5 dimension.

2. EQ-5D-3L, EQ-5D-3L, and SF-36 are usually hyphenated in this style

3. This is one of several papers published by this group on this dataset. I think it would be helpful to briefly document these, and their headline conclusions in the background to make it clear where this paper sits in the overall body of the work

4. I think there is too much extraneous data in the abstract. Detailing the baseline relationship between co-morbidity & EQ-5D / SF-36 distracts from the main points of interest - that people who are less well do less well overall but that they have just as much to gain from effective treatments. It is the absence of any interaction between co-morbidity and that treatment efficacy that is the golden nugget from this paper. This could be drawn out more in the abstract. Also, for the more general reader I think some explanation is needed of 'standardised units'

5. We are directed to the authors previous paper documenting that co-morbidity in clinical trials is less common than in the general population as a source for the full study selection process and the analysis plan. This paper tells us more about the selection process but provides no information on the statistical analysis plan for this new paper. Further, the entry in Prospero is too limited to indicate the nature of the analysis plan. Here it states;

'For each trial, for each outcome, estimates of covariate treatment-interactions will be obtained. The resultant trial-level results will be combined in Bayesian hierarchical generalized linear model to estimate interactions at the level of drug-classes and wider groupings of related drug classes.'

This indicates that original focus was on effects within drug classes rather than the overall effects reported here. I think the full, a priori, statistical analysis plan should be available with this paper.

6. In lines 142 to 145 the authors allude to mapping between EQ-5D-5L and EQ-5D-3L and refer here to a standard look-up table. I assume here that hey are using the Van Hout algorithm. Thus I assume they are standardising to a UK value set for EQ-5D-3L. If so this should be clear. There should be a link to where this can be found directly rather than the generic Euroqol website. There is a link to this in the EQ-5D-5L manual - but the link is broken and so I was unable to quickly find the relevant details.

7. For this current paper this sentence on lines 155 - 156 is probably superfluous 'SF36 can be mapped to preference based measures such as EQ5D for use in economic evaluations.'

8. In table 1 do we really mean age down to two places of decimals? Might whole numbers be easier on the eye and just as informative? Similarly for percentage of males whole percentages are likely to just as informative as presenting to one decimal place. If authors disagree on this point they need to edit the 28% figure from R03DX study

9. In the review copy figures 1&2 were rather blurred. A clearer version will be needed for publication

10. I think the PRISMA flow chart in the supplementary files might be better included in the main paper

Any attachments provided with reviews can be seen via the following link:

[LINK]

2023 Jan; 20(1): e1004154.

Author response to Decision Letter 1

17 Oct 2022

Attachment

Submitted filename: response to reviewers.docx

2023 Jan; 20(1): e1004154.

Decision Letter 2

This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

14 Nov 2022

Dear Dr. McAllister,

Thank you very much for re-submitting your manuscript "Comorbidity and health related quality of life in people with a chronic medical disease in randomised clinical trials: an individual participant data meta-analysis" (PMEDICINE-D-22-02156R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by 3 reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (gro.solp@enicidemsolp) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at gro.solp@enicidemsolp.

If you have any questions in the meantime, please contact me or the journal staff on gro.solp@enicidemsolp.

We look forward to receiving the revised manuscript by Nov 21 2022 11:59PM.

Sincerely,

Pippa

Philippa Dodd, MBBS MRCP PhD

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

Thank you for your considered and detailed responses to previous reviewer and editor comments (including those that may have appeared irrelevant) which are vital to maximizing the transparency of data. Please see below for further minor comments and suggested revisions.

COMMENTS FROM THE ACADEMIC EDITOR

A couple points in the article that were not entirely clear to me were:

1. how the evidence that they have in the article/appendix about linear relationship between predictor and outcome is consistent with the conclusion that non-linear relationship is not present (as opposed to it is unclear whether the relationship is linear or not) -- the linear assumption may be important if there are some people with outlier # of comorbidities which can have a lot of leverage in a linear model's slope estimates; also trial exclusions may systematically take out some types of comorbidities which could be more influential and hence generalization from trial efficacy and efficacy modification by comorbidity to population effectiveness and effect modification may be more challenging

2. the baseline QoL in the trials is generally relatively high (almost all 0.80 or above for EQ4D QoL weight) so it is probably important to not overly confidently extrapolate too far out of sample (to patients with baseline QoL that is substantially below) in terms of the impact of comorbidities on change in QoL (e.g., just because comorbidities for relatively healthy people [high QoL at baseline] may not have a very strong impact on the size of a treatment effect does not necessarily imply that this will be the case for people with much lower QoL at baseline [whose QoL is substantially outside the range of the study populations in the trials considered])

GENERAL

Please address all reviewer and editor comments in full

Please check the in-text reference call-outs throughout the manuscript there is occasionally a space missing between the text and the opening parentheses e.g. line 103 “patients[1]”, line 112 “reported[8–22]” and so on. Please ensure that punctuation follows closing parentheses e.g. line 282 “…conditions. [30] Trials….” should read "…conditions [30]. Trials….” Please check and amend throughout.

TITLE

suggest “conditions” in place of “diseases” or something similar

ABSTRACT

Thank you for updating the abstract and reporting according to PRISMA guidance. We note the comments from reviewer 3 regarding inclusion of information in the abstract and acknowledge that PRISMA reporting necessitates this for purposes of transparent data reporting. In context of the reporting guidance all findings should be reported in the abstract.

To me, not dissimilar to reviewer 3, the finding of an absence of interaction between co-morbidity and treatment efficacy is more interesting and more novel than other findings and I suspect may be to others reading your manuscript also. I agree with reviewer 3 that with respect to this particular outcome, as written, the importance of this point is somewhat lost, particularly on the general reader. I would suggest re-wording/elaborating the final sentences of the methods and findings section (lines 57-59) as well as the abstract conclusion such that the meaning/implications of the treatment efficacy data are clearer especially to the more general reader – i.e. what does the negative association mean in real terms at the patient and/or policy level? (see also author summary below)

Line 59: “EQ5D or SF36” – should hyphenate here also, please check throughout and amend where necessary.

AUTHOR SUMMARY

Line 74: perhaps “physical and mental health” instead of “mental function and wellbeing”, or something similar

Line 78: beginning “Multimorbidity…” suggest “Multimorbidity, the presence of two or more conditions, makes diagnosis and treatment more complex and is associated with worse quality of life in some settings.” The first sentence, as written, is perhaps redundant?

Line 80: beginning “Moreover, people with multimorbidity…” suggest beginning as a separate bulleted point as “People with multimorbidity are….”

What do these findings mean?

In general, I think you undersell your study, perhaps in an attempt not to oversell it (which is always important). I would suggest you revise this section with the below comments in mind:

Line 96: “…which helps inform clinical decision making” – how? Please justify the statement

Line 93: “a higher comorbidity count did not change the effect of treatment on quality of life.” and line 98: “…suggest that the effect of treatments on quality of life scores…” It may help to explicitly/simply state what the effect of treatment is. Throughout this is somewhat left to assumption which may not be helpful to the general reader.

Line 97: “These findings also provide some reassurance for clinicians and guideline developers…” – again how? I would revise this point in line with the above, what is the effect of treatment on QoL scores? I think this could be structured/re-worded to pack a bigger punch – which I think it deserves – it’s important not to overstate findings but I think you’re somewhat understating/underselling the importance of your study outcomes here.

Line 100: “…and for the types of treatments and conditions included in this analysis.” Rather paradoxically the statement is both broad and vague. I appreciate it would be impossible to list these but is there a better way to umbrella the included conditions (or refer to categories of those excluded)? Perhaps not given the nature of the study, in which case would it be better removed altogether? I'll leave it to your discretion, but I would have a think about how this is written such that you include as much necessary information as possible.

METHODS and RESULTS

Line 278: “…mean (sd) completion” should this read (SD) if representing ?standard deviation

FIGURE 1

Thank you for including this. Line 261: please hyphenate QoL surveys EQ-5D and SF-36 in the caption

FIGURE 2

Thank you for kindly altering the colour schemes of your figures to improve accessibility to the reader with colour blindness. Your caption(s) still refer to your previous colour scheme “...(blue) and without (red)…”. Please amend according to the new scheme and throughout, including supplementary files, where relevant.

DISCUSSION

The discussion (and preceding results section) achieves much better clarity regarding the implications of your study outcomes than the abstract/author summary. I would encourage you to cross reference any specific changes you make to the abstract/author summary with these sections to ensure consistency of language and terms, thus interpretation, throughout the manuscript.

Line 369: “Ours is the largest study to examine whether…” please temper assertations of primacy – suggest “to our knowledge” or something similar

LANGUAGE

Please ensure the term disease is replaced with condition or something similar in the title and please check throughout for the same

SOCIAL MEDIA

Please include any twitter handles for your institution/funding body etc such that we can help to extend the reach of your research

Comments from Reviewers:

Reviewer #1: Thanks for the revised manuscript and responses to my review. Compiling the information into the GitHub repository made completing the re-review much easier.

The sensitivity check with penalised splines is consistent with the polynomial terms, I think this looks fine to me.

I don't think there are any changes needed with the coefficient (my question about interpretability of the beta coefficient), I think you are right that there aren't any benchmarks because it's almost always normally an input in a CE analysis (which has a more obvious interpretation).

The explanation about the priors was helpful thank you - with the additional information available via the appendix I don't think there's any changes that need to be made here.

I happily recommend the manuscript be accepted.

Reviewer #2: Thank you for further reflecting on how to position this paper in the context of research on multimorbidity and comorbidity. I think the additional contributions clarifies the approach taken to conceptualising multimorbidity and justifying the selection of comorbidities.

Reviewer #3: I was pleased to see this revised article.

I must first apply some peer review to the editorial requirements with regard to how systematic reviews should be presented. To simplistically apply quality standards used for a conventional review to an IPD meta-analysis is simply wrong. This is an analysis on data are available in an IPD resources. It is simply not feasible to include trials published within six months. There is quite clearly no need to update the search to being within the previous six months. Further I suggest that conventional risk of bias assessment is also inappropriate. The authors are not here trying to suggest any main effects of an intervention. Rather they are looking at within trials data to help us better understand effects of co-morbidity. Above a very low bar for quality, that is met by having data of suitable quality to be included, and evidence of robust randomisation all these data will be of good quality. Essentially the data are what the data are

I remain very surprised that the authors think it a novel finding that people with multi-morbidity are less well, and have poorer outcomes is a novel finding. There is a vast amount of observational data, and indeed clinical experience, on this matter. I would also suggest that the highly selective nature of recruitment to the included trials means that these data are far inferior to the existing observational data on this point. Whilst these data need reporting the really novel finding from this work is that multi-morbidity does not moderate outcome, within this dataset. Thus I remain surprised that the authors report effect sizes for their first two analyses in the abstract when not including the (non) effect size for treatment moderation.

Any attachments provided with reviews can be seen via the following link:

[LINK]

2023 Jan; 20(1): e1004154.

Author response to Decision Letter 2

16 Nov 2022

Attachment

Submitted filename: response_to_reviewers.docx

2023 Jan; 20(1): e1004154.

Decision Letter 3

This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

9 Dec 2022

Dear Dr McAllister,

On behalf of my colleagues and the Academic Editor, Professor Jeremy Goldhaber-Fiebert, I am pleased to inform you that we have agreed to publish your manuscript "Comorbidity and health related quality of life in people with a chronic medical condition in randomised clinical trials: an individual participant data meta-analysis" (PMEDICINE-D-22-02156R3) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with gro.solp@sserpenicidem. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Best wishes,

Pippa

Philippa Dodd, MBBS MRCP PhD

PLOS Medicine

Articles from PLOS Medicine are provided here courtesy of PLOS