Composition Vendor Newgen KnowledgeWorks (P) Ltd. PLoS Med plos plosmed PLOS Medicine 1549-1277 1549-1676 Public Library of Science San Francisco, CA USA 10.1371/journal.pmed.1004532 PMEDICINE-D-24-02561 Research Article Medicine and health sciencesMental health and psychiatryMood disordersDepression Medicine and health sciencesHealth carePrimary care Medicine and health sciencesMental health and psychiatry Medicine and health sciencesHealth careHealth statisticsMorbidity Medicine and health sciencesOncologyCancers and neoplasms People and placesGeographical locationsEuropeEuropean UnionUnited KingdomWales People and placesGeographical locationsEuropeEuropean UnionUnited KingdomScotland Medicine and health sciencesPublic and occupational healthPhysical activity Depression and physical multimorbidity: A cohort study of physical health condition accrual in UK Biobank Depression and physical multimorbidity: A cohort study of physical health condition accrual https://orcid.org/0000-0002-1382-3095 Fleetwood Kelly J. Data curation Formal analysis Writing – original draft Writing – review & editing 1 * https://orcid.org/0000-0003-4191-4880 Guthrie Bruce Conceptualization Funding acquisition Supervision Writing – original draft Writing – review & editing 2 https://orcid.org/0000-0002-2067-2811 Jackson Caroline A. Conceptualization Funding acquisition Writing – review & editing 1 https://orcid.org/0000-0001-6997-9788 Kelly Paul A. T. Writing – review & editing 3 https://orcid.org/0000-0002-1703-3664 Mercer Stewart W. Conceptualization Funding acquisition Writing – review & editing 1 https://orcid.org/0000-0002-0063-8069 Morales Daniel R. Conceptualization Funding acquisition Writing – review & editing 4 https://orcid.org/0000-0001-9823-9252 Norrie John D. Conceptualization Funding acquisition Writing – review & editing 1 ¤ https://orcid.org/0000-0002-2267-1951 Smith Daniel J. Conceptualization Funding acquisition Writing – review & editing 5 https://orcid.org/0000-0002-7725-7520 Sudlow Cathie Conceptualization Funding acquisition Writing – review & editing 1 6 https://orcid.org/0000-0002-0489-684X Prigge Regina Data curation Formal analysis Writing – review & editing 1 Usher Institute, University of Edinburgh, Edinburgh, United Kingdom Advanced Care Research Centre, Usher Institute, University of Edinburgh, Edinburgh, United Kingdom Public Member of Study Advisory Board, Edinburgh, United Kingdom Division of Population Health and Genomics, University of Dundee, Dundee, United Kingdom Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, United Kingdom Health Data Research United Kingdom, London, United Kingdom Patel Vikram Academic Editor Harvard Medical School, UNITED STATES OF AMERICA

The authors have declared that no competing interests exist.

Current address: Queen’s University, Belfast, United Kingdom

 * E-mail: kelly.fleetwood@ed.ac.uk 1322025 22025 22 2 e1004532 782024 1312025 2025 Fleetwood et al 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. Background

Depression is associated with a range of adverse physical health outcomes. We aimed to quantify the association between depression and the subsequent rate of accrual of long-term physical health conditions in middle and older age.

 Methods and findings

We included 172,556 participants from the UK Biobank (UKB) cohort study, aged 40–71 years old at baseline assessment (2006–2010), who had linked primary care data available. Using self-report, primary care, hospital admission, cancer registry, and death records, we ascertained 69 long-term physical health conditions at both UKB baseline assessment and during a mean follow-up of 6.9 years. We used quasi-Poisson models to estimate associations between history of depression at baseline and subsequent rate of physical condition accrual. Within our cohort, 30,770 (17.8%) had a history of depression. Compared to those without depression, participants with depression had more physical conditions at baseline (mean 2.9 [SD 2.3] versus 2.1 [SD 1.9]) and accrued additional physical conditions at a faster rate (mean 0.20 versus 0.16 additional conditions/year during follow-up). After adjustment for age and sex, participants with depression accrued physical morbidities at a faster rate than those without depression (RR 1.32, 95% confidence interval [CI] [1.31, 1.34]). After adjustment for all sociodemographic characteristics, the rate of condition accrual remained higher in those with versus without depression (RR 1.30, 95% CI [1.28, 1.32]). This association attenuated but remained statistically significant after additional adjustment for baseline condition count and social/lifestyle factors (RR 1.10, 95% CI [1.09, 1.12]). The main limitation of this study is healthy volunteer selection bias, which may limit generalisability of findings to the wider population.

 Conclusions

Middle-aged and older adults with a history of depression have more long-term physical health conditions at baseline and accrue additional physical conditions at a faster rate than those without a history of depression. Our findings highlight the importance of integrated approaches to managing both mental and physical health outcomes.

 Author summary Why was this study done?

Mood disorders like depression are increasingly viewed as whole-body conditions affecting multiple systems across the brain and body.

People with depression are more likely to have long-term physical health conditions, such as diabetes and arthritis, than people without depression.

Previous studies have compared people with and without depression in terms of how many long-term physical health conditions they develop over time, however, most studies count 15 or fewer conditions, whereas recent research recommends counting more than 50 conditions.

 What did the researchers do and find?

We used data from over 170,000 people in middle and older age who participated in the UK Biobank study.

At the start of the study, 18% of the group had previously been diagnosed with depression.

We followed up the group for an average of 7 years after the start of the study, using general practitioner and hospital data to identify new diagnoses of 69 long-term physical health conditions.

At the start of the study, people without a previous diagnosis of depression had an average of 2 long-term physical health conditions, whilst people with a previous diagnosis of depression had an average of 3 such conditions.

On average, based on people who were the same age and sex, people with a previous diagnosis of depression gained long-term physical health conditions at a 30% faster rate than people without a previous diagnosis of depression.

 What do these findings mean?

A previous diagnosis of depression is a marker of risk for the subsequent development of long-term physical health conditions during middle and older age.

Existing healthcare systems are largely designed to treat individual conditions, instead of individual people with multiple conditions, and they especially struggle to treat people with both physical and mental health conditions.

We need healthcare services to take an integrated approach to caring for people who have both depression (or other mental health conditions) and long-term physical health conditions.

Using data from the UK Biobank, Kelly J Fleetwood and colleagues investigate the association between depression and the rate of accrual of long-term physical health conditions during middle and older age.

 http://dx.doi.org/10.13039/501100000265 Medical Research Council MC/S028013 https://orcid.org/0000-0003-4191-4880 GuthrieBruce This work was funded by the Medical Research Council (https://www.ukri.org/councils/mrc/)/National Institute for Health Research (https://www.nihr.ac.uk/) (MC/S028013) (BG [principal investigator]; CS, JN, SM, CJ, DM, DS [co-investigators]). The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. Data Availability This study was conducted using data from the UK Biobank (https://www.ukbiobank.ac.uk/). Researchers can apply to access the UK Biobank data for health research in the public interest. The code lists used in this study are available from https://github.com/rprigge-uoe/mltc-codelists. Introduction

Depression is the most common mental health condition. Over 300 million people worldwide, approximately 1 in 23 of the population, live with depression [1,2]. Prevalence of depression is higher in women than men and most common in adults aged 55–74 years [1]. Depression, irrespective of severity, is associated with increased mortality risk [3]. This is largely due to poorer physical health, particularly a higher risk of type 2 diabetes and cardiovascular disease [4,5].

Multimorbidity is defined as the coexistence of two or more long-term health conditions, with physical multimorbidity specifically referring to the coexistence of two or more long-term physical health conditions. Although estimates of multimorbidity prevalence differ due to large variations in the number of conditions counted [6], multimorbidity increases rapidly with age and is the norm in older people. Health and social care systems are challenged by the increasing prevalence of multimorbidity, driven by population aging, improved survival from acute conditions, and increasing incidence of chronic conditions such as type 2 diabetes [7]. However, the role of depression in multimorbidity is understudied, with only half of multimorbidity studies including depression in their condition count [6]. Whilst depression has been shown to be associated with increased risk of various individual physical health conditions, less is known about its role as a risk factor for physical multimorbidity.

Most previous research on depression and physical multimorbidity has been cross-sectional, with studies indicating that approximately one-fifth of people with any physical condition also have depression, and that the prevalence of depression increases with the number of physical conditions [8]. However, cross-sectional studies are difficult to interpret given a plausible bidirectional relationship between depression and physical health [9].

Previous cohort studies have generally reported an association between baseline depression and accrual of subsequent physical conditions [9–16]. However, these studies have a number of limitations. Most measured physical condition accrual using intermittent surveys where loss to follow-up may be informative because participants who are unwell are more likely to die or drop out [9,10,13,15,17]. Furthermore, many studies measured physical multimorbidity using between three and 15 conditions [9–15,17], whereas a recent Delphi study recommended over 50 conditions for inclusion in multimorbidity measures [18]. Further limitations of existing studies are their relatively short follow-up, with some studies having only three or four years follow-up [9,13,17], and limited adjustment for potential confounders.

We therefore aimed to quantify the association between history of depression and the subsequent accrual of 69 long-term physical health conditions in middle and older age, using electronic health records to identify the incidence of physical conditions during an average of seven years follow-up.

 Methods

We have reported this study in accordance with the STROBE checklist [19] (S1 Checklist). This study is part of a wider project using UK Biobank (UKB) data to investigate relationships between depression and multimorbidity. Our grant proposal and our application to UKB for the wider project included a broad overview of the plans for this study, but we did not publish a prospective analysis plan.

Study design and participants

The UKB is a cohort study of half a million middle-aged and older adults with information on a wide range of health conditions, sociodemographic, lifestyle, and social factors [20]. People registered with a general practitioner in England, Scotland, or Wales were invited to participate, with baseline assessment conducted between 2006 and 2010 [21]. Baseline assessment included a touch-screen questionnaire, verbal interview, and physical measurements [20]. Participants provided written informed consent for follow-up through linkage to national datasets, including primary care, hospital, cancer registry, and death records. UKB has ethical approval from the NHS North West Research Ethics Committee (reference: 21/NW/0157).

In order to appropriately evaluate a broad range of long-term physical health conditions, our study population was derived from UKB participants with linked primary care data (approximately 45% of the UKB cohort) [22]. Linked primary care data were available from Scotland, Wales and practices in England that used either the TPP or Vision practice management systems. We included participants with a continuous primary care record (no gaps of >90 days between practice registrations) from at least a year before their baseline assessment to at least one day after baseline assessment [23]. We excluded the small proportion of primary care records from the UKB extract of the Vision practice management system in England because this linked dataset is missing records from people who died before data extraction. We also excluded participants who withdrew from the study.

 Linked electronic health records

In addition to the primary care records, all UKB participants are linked to hospital, cancer registry, and death records from England, Scotland, and Wales. Records are available for different dates in each country, as described in Table A in S1 Text.

We defined conditions at baseline using all primary care records up to and including the date of the participant’s baseline assessment. Primary care records transfer between practices in the UK when a patient moves, and so should capture an individual’s entire medical history. Cancer registry and hospital records were available from different dates for England, Wales, and Scotland, with a minimum of eight years of records prior to the baseline assessments (Table A in S1 Text). To ensure consistency of look-back period across the secondary care data sources, we defined conditions at baseline for each participant using cancer registry and hospital records from the eight years up to and including their baseline assessment date.

During the follow-up period, both primary care and cancer registry records are available up to at least 2016, with hospital and death records available for longer (Table A in S1 Text). Participants were therefore followed up to the earliest of death, end of continuous primary care record, or end of cancer registry follow-up.

 Outcome

We selected 69 long-term physical health conditions relevant to middle-aged and older adults and recommended for multimorbidity research (Table B in S1 Text) [18,24]. We use the term condition broadly, with some conditions including more than one disease. For example, the condition ‘solid organ malignancies’ includes both primary and secondary malignancies. This approach reduces double counting of diagnoses that evolve or are related (e.g., for people whose cancer has metastasised).

For each participant, we identified long-term physical health conditions using information collected from the participant at baseline assessment and from primary care, hospital, cancer registry, and death records. We identified conditions from primary care records using Read V2 and Clinical Terms V3 (CTV3) codes, from hospital records using ICD-10 codes and OPCS-4 procedure codes, and from cancer registry and death records using ICD-10 codes. All code lists are available in our GitHub repository (https://github.com/rprigge-uoe/mltc-codelists), with a more detailed description of our approach available in the accompanying manuscript [22].

Conditions were defined as prevalent at each participant’s date of UKB baseline assessment if the condition was self-reported at the baseline assessment or the first linked record was on or before the date of the baseline assessment, with the exception of five conditions present from birth (Table B in S1 Text) which were always considered prevalent regardless of the timing of the first record. We defined incident conditions as those where the first record was after the date of baseline assessment. The outcome was the count of incident long-term physical health conditions during follow-up. Since we focused on long-term health conditions, our analyses did not allow for recovery or remission from conditions. Furthermore, such information is not well recorded in the electronic health records.

 Depression

Our exposure is history of depression at baseline: defined as a diagnosis of depression at any time before or on the participant’s UKB baseline assessment date. In the United Kingdom, depression is most often diagnosed in primary care, and primary care records should capture a participant’s entire medical history, however, in order to capture as many depression cases as possible, we defined history of depression based on multiple sources. We identified diagnoses of depression from linked primary care or hospital records, or self-reported depression in response to the baseline assessment question “Has a doctor ever told you that you have had any other serious medical conditions or disabilities?”. Code lists for depression (Read V2 and CTV3 codes for primary care records, and ICD-10 codes for hospital records) are available in our GitHub repository (https://github.com/rprigge-uoe/mltc-codelists).

 Covariates

Age and sex were ascertained from recruitment data and optionally updated by participants at the baseline assessment. We categorised self-reported ethnicity into five groups (Black, mixed, South Asian, White, and any other ethnic group) [25]. Country of residence (England, Wales, or Scotland) and area-based deprivation, measured by the Townsend Deprivation Index (in deciles of the whole UKB cohort) [26], were derived from participants’ home addresses at baseline.

For each participant, we calculated a total (prevalent) condition count at their UKB baseline assessment date. This count included the 69 physical health conditions used in the outcome and additionally 10 non-depression mental health conditions (Table C in S1 Text). We ascertained these mental health conditions from baseline assessment, primary care and hospital data using the same approach that we used to ascertain the physical health conditions.

We obtained information on stressful life events, loneliness, multisite pain, sleep, smoking, alcohol intake frequency, and frailty (weight loss, slow walking speed, weak grip strength, and low physical activity) from the baseline assessment touchscreen questionnaire (Table D in S1 Text). We obtained body mass index (BMI), cholesterol:HDL ratio, HbA1c, and systolic blood pressure (SBP) from laboratory tests or measurements taken during the baseline assessment.

 Statistical analysis

Our aim was to describe to what extent the rate of accrual of long-term physical health conditions differs between people with and without depression.

To provide a simple visualisation of the accrual of long-term physical health conditions, we plotted the cumulative mean number of conditions per participant from baseline to end of follow-up by age, sex and history of depression at baseline. The cumulative mean at each time point was based on participants followed up until at least that time point.

We then compared the rate of long-term physical health condition accrual during follow-up between participants with and without a history of depression at baseline using quasi-Poisson models, which allow for over-dispersion in the outcome. We accounted for each participant’s follow-up time by including the log follow-up time as an offset term in the models. We estimated the unadjusted association between history of depression at baseline and the count of incident physical conditions, then adjusted for age at baseline and sex, followed by additional sociodemographic characteristics (ethnicity, country of residence at baseline, and area-based deprivation). Finally, we fitted a fully adjusted model which additionally accounted for clinical, social, and lifestyle factors (baseline condition count, count of stressful life events, loneliness, multisite pain, sleep, smoking, alcohol intake frequency, BMI, cholesterol:HDL ratio, HbA1c, SBP, and four markers of frailty: weight loss, slow walking speed, weak grip strength, and low physical activity). Age, count of physical morbidities at baseline, cholesterol:HDL ratio, and SBP were included in the models as continuous variables, each with a linear term and additionally a quadratic term where this improved the fit. Age was scaled by subtracting the mean age and dividing by 10. Count of morbidities at baseline was centred by subtracting the mean. We log transformed cholesterol:HDL ratio and SBP to improve the normality of their distributions (in preparation for the multiple imputation, see below), then centred the logged values by subtracting their mean. All other variables were categorical.

We used multiple imputation by chained equations (MICE) to account for missing data in 17 covariates, assuming that data was missing at random. Most covariates were missing for less than 2% of participants; however, the cholesterol:HDL ratio and HbA1c were missing for 14.2% and 6.1%, respectively (mostly due to assay problems). The imputation models included the outcome and all covariates (including the quadratic terms for the continuous covariates). They did not include any auxiliary variables because the covariates already covered a broad set of sociodemographic, clinical, social and lifestyle factors. Overall, 23.3% of our cohort had at least one missing covariate. Hence, we conducted 25 imputations [27]. Each imputed dataset was analysed and we used Rubin’s rules to pool the results [28]. We also conducted a complete case analysis and compared the results to those of the multiple imputation.

All analysis was conducted in R version 4.0.0 or above [29]. We used the mice package version 3.16.0 to perform multiple imputation [30].

 Results

We included 172,556 UKB participants in our cohort (Fig 1). Of these, 30,770 (17.8%) had a history of depression at baseline. History of depression at baseline was predominately identified from primary care records; 90.7% of cases had a primary care diagnosis on or before baseline, 4.2% had a hospital diagnosis on or before baseline, and 31.5% of cases self-reported a previous diagnosis of depression (Fig A in S1 Text). Mean follow-up was 6.9 (SD 1.9) years (Table 1). The mean age at baseline was 57 (SD 8) years. Two-thirds of participants with a history of depression were women, compared to half of participants without a history of depression. The majority of participants were White and from England. Participants with a history of depression more commonly lived in more deprived areas, and on average had 1.3 more long-term health conditions (0.8 more physical health conditions; 0.4 more mental health conditions) at baseline than those without a history of depression. The three most common physical health conditions present at UKB baseline assessment were hypertension (33.4% of people with a history of depression versus 29.6% of people without a history of depression), allergic and chronic rhinitis (32.4% versus 26.7%), and osteoarthritis (27.0% versus 19.1%) (Tables E and F in S1 Text). The most common comorbid mental health conditions present at baseline were anxiety (41.8% versus 6.3%) and alcohol misuse (4.1% versus 1.7%) (Table G in S1 Text). At baseline, chronic multisite pain, sleeplessness, stressful life events, loneliness, smoking and obesity were all more common amongst participants with a history of depression versus those without. Participants with a history of depression were more likely to abstain from alcohol, or only drink alcohol occasionally. Cholesterol:HDL ratio and HbA1c were marginally higher in participants with a history of depression, and SBP was marginally lower, although absolute differences were small. Participants with a history of depression also had more markers of frailty, including weight loss, slow walking speed, weak grip strength, and low physical activity.

10.1371/journal.pmed.1004532.g001Flow diagram for the study cohort.

*Excluding participants who withdrew permission for their data to be included in research before 13 October 2023. UK, United Kingdom.

 10.1371/journal.pmed.1004532.t001Baseline characteristics of included participants by history of depression at baseline.
History of depression at baseline
Depression (N = 30,770) No depression (N = 141,786)
Follow-up (years) 6.7 (2.0) 6.9 (1.9)
Age (years) 56.3 (7.9) 56.8 (8.0)
Sex: Female 20,592 (66.9%) 73,431 (51.8%)
Ethnicity
 Black 190 (0.6%) 1,355 (1·0%)
 Mixed 179 (0.6%) 610 (0.4%)
 South Asian 363 (1.2%) 2,440 (1.7%)
 White 29,620 (96.3%) 135,086 (95.3%)
 Other ethnic group 277 (0.9%) 1,693 (1.2%)
 Missing 141 (0.5%) 602 (0.4%)
Country of residence
 England 23,709 (77.1%) 107,911 (76.1%)
 Scotland 3,617 (11.8%) 18,497 (13.0%)
 Wales 3,429 (11.1%) 15,335 (10.8%)
 Missing 15 (0.0%) 43 (0.0%)
Area-based deprivation (Townsend Deprivation Index quintile*)
 1 (least deprived) 5,356 (17.4%) 29,876 (21.1%)
 2 5,581 (18.1%) 29,263 (20.6%)
 3 6,135 (19.9%) 29,869 (21.1%)
 4 6,431 (20.9%) 28,162 (19.9%)
 5 (most deprived) 7,220 (23.5%) 24,457 (17.2%)
 Missing 47 (0.2%) 159 (0.1%)
Number of prevalent morbidities 3.5 (2.4) 2.2 (1.9)
 Physical 2.9 (2.3) 2.1 (1.9)
 Mental 0.5 (0.6) 0.1 (0.3)
Chronic multisite pain
 No 20,729 (67.4%) 115,487 (81.5%)
 Yes 9,928 (32.3%) 25,755 (18.2%)
 Missing 113 (0.4%) 544 (0.4%)
Sleeplessness/insomnia
 Never/rarely 4,747 (15.4%) 36,011 (25.4%)
 Sometimes 13,820 (44.9%) 68,121 (48.0%)
 Usually 12,122 (39.4%) 37,337 (26.3%)
 Missing 81 (0.3%) 317 (0.2%)
Stressful life events (count, excluding own illness/injury/assault)
 0 15,910 (51.7%) 88,353 (62.3%)
 1 10,439 (33.9%) 41,529 (29.3%)
 2 or more 4,097 (13.3%) 10,794 (7.6%)
 Missing 324 (1.1%) 1,110 (0.8%)
Loneliness
 No 19,298 (62.7%) 119,240 (84.1%)
 Yes 10,711 (34.8%) 20,169 (14.2%)
 Missing 761 (2.5%) 2,377 (1.7%)
Smoking status
 Never 15,116 (49.1%) 79,952 (56.4%)
 Previous 10,972 (35.7%) 48,005 (33.9%)
 Current 4,524 (14.7%) 13,128 (9.3%)
 Missing 158 (0.5%) 701 (0.5%)
Alcohol intake
 Daily or almost daily 5,411 (17.6%) 28,329 (20.0%)
 Three or four times a week 5,832 (19.0%) 34,203 (24.1%)
 Once or twice a week 7,629 (24.8%) 38,156 (26.9%)
 One to three times a month 3,850 (12.5%) 15,543 (11.0%)
 Special occasions only 4,566 (14.8%) 14,777 (10.4%)
 Never 3,394 (11.0%) 10,493 (7.4%)
 Missing 88 (0.3%) 285 (0.2%)
BMI (kg/m2)
  < 25 8,873 (28.8%) 46,121 (32.5%)
 25–29.9 12,434 (40.4%) 60,998 (43.0%)
 30–34.9 6,143 (20.0%) 24,737 (17.4%)
  ≥35 3,121 (10.1%) 9,132 (6.4%)
 Missing 199 (0.6%) 798 (0.6%)
Cholesterol:HDL ratio
 Mean (SD) 4.2 (1.1) 4.1 (1.1)
 Missing 4,530 (14.7%) 19,913 (14.0%)
HbA1c (mmol/mol)
  < 32 4,949 (16.1%) 24,394 (17.2%)
 32–34 4,909 (16.0%) 23,667 (16.7%)
 34–36 5,934 (19.3%) 28,770 (20.3%)
 36–38 5,237 (17.0%) 23,703 (16.7%)
  ≥38 7,820 (25.4%) 32,603 (23.0%)
 Missing 1,921 (6.2%) 8,649 (6.1%)
Systolic blood pressure (mmHg)
 Mean (SD) 136.2 (18.4) 139.2 (18.7)
 Missing 160 (0.5%) 451 (0.3%)
Weight loss
 No 24,752 (80.4%) 118,619 (83.7%)
 Yes 5,433 (17.7%) 20,545 (14.5%)
 Missing 585 (1.9%) 2,622 (1.8%)
Slow walking speed
 No 26,191 (85.1%) 130,868 (92.3%)
 Yes 4,163 (13.5%) 9,878 (7.0%)
 Missing 416 (1.4%) 1,040 (0.7%)
Weak grip strength
 No 21,875 (71.1%) 109,169 (77.0%)
 Yes 8,641 (28.1%) 31,647 (22.3%)
 Missing 254 (0.8%) 970 (0.7%)
Low physical activity
 No 25,934 (84.3%) 126,933 (89.5%)
 Yes 4,441 (14.4%) 13,654 (9.6%)
 Missing 395 (1.3%) 1,199 (0.8%)

Data are mean (SD) or n (%).

*The statistical models adjust for Townsend Deprivation Index decile, however, quintiles are shown here for conciseness.

BMI, Body Mass Index; HbA1c, glycated haemoglobin; HDL, high density lipoprotein.

During follow-up, participants with a history of depression accrued an average of 0.20 additional physical health conditions per year compared to 0.16 conditions per year for participants without a history of depression (Fig 2). The three most common incident physical health conditions during follow-up in both participants with and without a history of depression were osteoarthritis (15.7% of people with depression, but without osteoarthritis at baseline versus 12.5% of people without depression or osteoarthritis at baseline), hypertension (12.9% versus 12.0%) and gastro-oesophageal reflux disease (and similar) (13.8% versus 9.6%), (Table E in S1 Text).

10.1371/journal.pmed.1004532.g002Cumulative mean number of long-term physical health conditions at baseline and during follow-up*, stratified by history of depression at baseline, age at baseline and sex (n = 172,556).

*The cumulative mean at each time point is based on participants followed up until at least that time point.

Based on all participants in our cohort, in the unadjusted model, people with a history of depression accrued new long-term physical health conditions at a faster rate than people without a history of depression (rate ratio [RR] 1.25, 95% confidence interval [CI] [1.23, 1.27]). After adjustment for age and sex, the rate of condition accrual remained higher in people with versus without a history of depression (RR 1.32, 95% CI [1.31, 1.34]) (Table 2). Additional adjustment for sociodemographic factors only slightly attenuated the effect estimate (RR 1.30, 95% CI [1.28, 1.32]). In the fully adjusted model, which included potential mediators, a history of depression was associated with a marginally increased rate of morbidity accrual (RR 1.10, 95% CI [1.09, 1.12]).

10.1371/journal.pmed.1004532.t002Rate ratios for the association of history of depression at baseline, sociodemographic, social, lifestyle and clinical factors with physical health condition accrual during follow-up.
Rate ratio (95% CI)* 
Adjusted for age and sex Adjusted for sociodemographic characteristics Fully adjusted
History of depression at baseline 1.32 (1.31, 1.34) 1.30 (1.28, 1.32) 1.10 (1.09, 1.12)
Age† Age (RR for each additional 10 years of age) 1.58 (1.57, 1.59) 1.59 (1.58, 1.61) 1.43 (1.41, 1.44)
Sex (ref: Male) Female 0.80 (0.79, 0.81) 0.81 (0.80, 0.82) 0.82 (0.81, 0.83)
Socioeconomic statusTownsend deprivation decile(ref: 1, least deprived) 2 0.99 (0.96, 1.01) 0.98 (0.96, 1.01)
3 1.01 (0.98, 1.03) 0.99 (0.97, 1.01)
4 1.04 (1.01, 1.07) 1.02 (0.99, 1.04)
5 1.06 (1.03, 1.08) 1.02 (0.99, 1.04)
6 1.06 (1.03, 1.08) 1.00 (0.98, 1.03)
7 1.10 (1.07, 1.12) 1.01 (0.99, 1.04)
8 1.13 (1.11, 1.16) 1.02 (1.00, 1.05)
9 1.21 (1.18, 1.24) 1.04 (1.02, 1.07)
10 (most deprived) 1.35 (1.32, 1.39) 1.07 (1.04, 1.10)
Ethnicity (ref: White) Black 1.20 (1.13, 1.28) 1.08 (1.02, 1.14)
Mixed 1.11 (1.02, 1.21) 1.06 (0.97, 1.15)
South Asian 1.33 (1.28, 1.39) 1.16 (1.11, 1.20)
Other ethnic group 1.10 (1.04, 1.16) 1.03 (0.98, 1.09)
Country of residence at baseline (ref: England) Scotland 0.81 (0.80, 0.82) 0.82 (0.81, 0.84)
Wales 0.99 (0.98, 1.01) 0.95 (0.93, 0.97)
Morbidities at baseline† Count 1.06 (1.06, 1.07)
Count-squared 0.997 (0.996, 0.997)
Smoking(ref: never) Previous 1.09 (1.08, 1.11)
Current 1.25 (1.23, 1.28)
Alcohol intake(ref: daily or almost daily) 3–4 times a week 0.97 (0.95, 0.99)
1–2 times a week 1.00 (0.98, 1.02)
1–3 times a month 1.01 (0.99, 1.03)
Special occasions 1.05 (1.03, 1.07)
Never 1.08 (1.06, 1.10)
BMI (kg/m2)(ref: < 25) 25–29.9 1.10 (1.08, 1.11)
30–34.9 1.20 (1.18, 1.22)
 ≥35 1.33 (1.30, 1.36)
SBP (mmHg)† log(SBP) 1.42 (1.35, 1.48)
(log(SBP))-squared 3.20 (2.59, 3.95)
Cholesterol:HDL ratio† log(ratio) 1.06 (1.03, 1.08)
(log(ratio))-squared 1.24 (1.17, 1.32)
HbA1c (mmol/mol)(ref: < 32) 32–34 0.99 (0.97, 1.01)
34–36 1.02 (1.00, 1.04)
36–38 1.04 (1.02, 1.06)
 ≥38 1.21 (1.19, 1.23)
Sleeplessness/insomnia(ref: never/rarely) Sometimes 1.04 (1.02, 1.05)
Usually 1.09 (1.07, 1.11)
Count of stressful life events (ref: 0) 1 1.03 (1.02, 1.04)
2 or more 1.06 (1.04, 1.08)
Chronic multisite pain (ref: no) Yes 1.15 (1.14, 1.17)
Loneliness (ref: no) Yes 1.06 (1.05, 1.08)
Weight loss (ref: no) Yes 1.04 (1.03, 1.06)
Slow walking speed (ref: no) Yes 1.12 (1.10, 1.14)
Weak grip strength (ref: no) Yes 1.05 (1.03, 1.06)
Low physical activity (ref: no) Yes 1.08 (1.06, 1.10)

*Results of Poisson models, multiple imputation was used to account for missing covariates in the model adjusted for sociodemographic characteristics and the fully adjusted model.

† For each quantitative covariate, we scaled the values by subtracting their mean.

BMI, body mass index; HbA1c, glycated haemoglobin; HDL, high-density lipoprotein; SBP, systolic blood pressure.

For all models, results based on complete case analysis were similar to those based on multiple imputation (Table H in S1 Text).

 Discussion

Compared to people without a history of depression, people with a history of depression had more long-term physical health conditions at baseline, and accrued physical conditions at a faster rate even after accounting for sociodemographic differences. The association attenuated after further adjustment for baseline clinical, social, and lifestyle factors.

A number of other studies have examined the association between depression or depressive symptoms and the accrual of long-term physical health conditions [9–16]. These studies varied in the number of physical health conditions studied, how conditions were followed-up and the length of follow-up, but most also found that depression or depressive symptoms were associated with a higher rate of condition accrual. Using UKB data, Qiao and colleagues (2022) found that baseline depression was associated with higher rates of cardiometabolic multimorbidity (narrowly defined as two or more of type 2 diabetes, stroke, and coronary heart disease) [14]. A large study from Canada of adults without physical morbidity at baseline observed associations between a major depressive episode in the previous 12 months and the development of one or more of 15 conditions in the following 10 years [12]. Whilst a Swedish study of older adults defined physical multimorbidity based on 54 conditions and had a follow-up of 15 years, this study was much smaller than our own with approximately 3,000 participants in total and fewer than 300 with depression [16]. Nevertheless, it also found an association between depression and rate of morbidity accumulation. In contrast, a study from the United States found that adults with baseline bipolar disorder, but not depression, had a higher rate of morbidity accrual compared to adults without a mood disorder [17]. However, the study population was younger than in the present study, with almost half the participants younger than 45 years of age at baseline, and follow-up was only three years. Using data from residents of one US county, Bobo and colleagues (2011) found that women with depression experienced higher rates of accrual of 15 physical conditions across the life span, although associations in men varied by age and by the presence of comorbid anxiety, with a higher rate of condition accrual in some groups of men, but no evidence of association in others [11].

Strengths of this study are the use of a very large cohort with multiple linked electronic health records including primary care data, which is essential for detecting many conditions that are treated mainly or exclusively in primary care, including depression [22]. Our definition of physical multimorbidity included a wide range of long-term physical health conditions based on recent recommendations for conditions to include in multimorbidity research [18]. However, we acknowledge that there is ongoing debate about what conditions should be included in multimorbidity measures, and whether some conditions, such as hypertension, should instead be treated as risk factors. Average follow-up in this study was seven years and used linked electronic health records, including death records, meaning that outcome ascertainment is likely to be better than previous studies which relied on repeat survey data, and are therefore subject to recall and reporting bias, and ascertainment bias due to loss to follow-up [31].

The study also has a number of limitations. Only 5.5% of people invited to join the UK Biobank participated in the baseline assessments, with participants less likely on average to live in deprived areas, and more likely to have better health, compared with the general population [32]. Baseline condition counts and rates of accrual of new conditions are therefore like to underestimate values in the general population. Furthermore, our estimates of the association between history of depression at baseline and physical health condition accrual may not be generalisable to the wider population [33,34]; however, it is reassuring that other multimorbidity associations estimated from the UK Biobank cohort were generally similar to associations estimated from a nationally representative sample [35]. Whilst our partially adjusted models included information on covariates that likely preceded both our exposure and outcome and thus acted as confounding factors, some covariates included in our final model may lie on the causal pathway or have a conceptual overlap with depression. Thus, our final model may have underestimated the strength of the association between depression and physical condition accrual. Our cohort had relatively high rates of missingness for cholesterol:HDL ratio (14.2%) and HbA1c (6.1%), although data for other covariates was near complete. We accounted for missing data using multiple imputation under the assumption that data was missing at random which is plausible since missing laboratory data was primarily due to assay problems after sample collection. Whilst we have examined a broad range of long-term physical health conditions, examining the relationship between depression and specific physical conditions was outside the scope of this study. Finally, we were unable to examine the role of depression remission and relapse and depression severity because such information was not reliably captured by the data.

Middle-aged and older adults with a history of depression have higher prevalence of physical health conditions at baseline, and have an increased rate of physical condition accrual subsequently. The higher rate of accrual is partly driven by differences in modifiable risk factors like smoking, high BMI and low physical activity, meaning that there are potential opportunities for preventive care to improve future health. Multimorbidity challenges existing healthcare because individual needs do not always sit neatly with organisational boundaries, and that is often most true when people have both physical and mental health problems. Better identification and management of depression in physical healthcare is needed, but mental health services also need to involve themselves in supporting their patients to maintain or improve their physical health, in relation to smoking, diet, obesity, and exercise, for example. Relatively intensive collaborative care approaches have been shown to be effective in improving mental and physical outcomes in people with depression and common conditions like diabetes and heart disease but there is a need to develop and evaluate more consistent preventive approaches across all services [36]. Such approaches are likely to be most needed in less affluent areas, with the least affluent developing any multimorbidity 10–15 years earlier than the most affluent, and even larger differences in physical-mental multimorbidity [8,37]. In this study, people living in more deprived areas were more likely to have depression at baseline. This highlights the additional need for better integrated physical-mental healthcare in less affluent areas, which is not reflected in current resource allocation [38].

There are a number of implications for future research. Although people with a history of depression accrued physical conditions at a higher rate than people without, they also already had more physical health conditions at baseline. Life course approaches would be useful to better understand the interplay of depression and physical health at younger ages, and to unpick whether the observed associations are driven by a subset of physical conditions or whether associations are similar for all conditions. Investigating the ordering of subsequent long-term physical health conditions [39] would also be worthwhile, although it is often difficult to establish the order of conditions from electronic health records, especially where multiple diagnoses are recorded on the same day. Similarly, it would be useful to explore whether depression is also associated with subsequent physical condition severity and disease-specific outcomes, as well as generic outcomes like quality of life. We estimated the association between depression and physical condition accrual in a broad cohort, including both men and women, ranging in age from 40 to 71 years at baseline. However, it would also be valuable to explore this association in analyses stratified by key sociodemographic characteristics such as age and sex. Further research should also more carefully explore the causal relationship between depression and the subsequent accrual of long-term physical health conditions in order to identify potential interventions, including whether effective management of depression and/or more assertive management of risk factors for physical disease in people with depression leads to lower rates of physical disease accrual. Further research which stratifies depression by severity and chronicity would be useful to examine if there are particular groups of patients where such interventions might be most effectively targeted. Research is also needed to better understand potential mechanisms, for example, the role of inflammation in the development of both depression and physical health conditions like coronary heart disease.

In the most comprehensive study to date on this topic, to our knowledge, we identified a higher rate of accrual for comorbid physical health problems in people with a history of depression compared to those without. Our findings highlight that depression should be viewed as a ‘whole body’ condition, as well as the importance of integrated approaches to managing both mental and physical health outcomes.

 Supporting information STROBE checklist.

(DOCX)

 Supplementary tables and figures.

(DOCX)

 Abbreviations BMI

body mass index

CI

confidence interval

CTV3

Clinical Terms V3

HDL

high-density lipoprotein

RR

rate ratio

SBP

systolic blood pressure

MICE

multiple imputation by chained equations

UKB

UK Biobank

The study was conducted using the UK Biobank Resource under application number 57213. The authors would like to thank the UK Biobank participants and the UK Biobank staff for their contributions to this study. The authors would also like to thank Pat Watson, a public member of our advisory board, for providing thoughtful feedback throughout our project, Dr Emma Davidson for providing us with unpublished depression code lists and Dr Kristiina Rannikmäe for providing us with stroke, B12 deficiency anaemia, folate deficiency anaemia and iron deficiency anaemia code lists which were unpublished at the time.

 References World Health Organization. Depression and other common mental disorders: global health estimates. World Health Organization. 2017 [cited 2025 Jan 8]. Available from: https://apps.who.int/iris/bitstream/handle/10665/254610/W?sequence=1. Vos T, Lim SS, Abbafati C, Abbas KM, Abbasi M, Abbasifard M, et al. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet. 2020;396(10258):1204–22. doi: 10.1016/S0140-6736(20)30925-9 33069326 Cuijpers P, Vogelzangs N, Twisk J, Kleiboer A, Li J, Penninx BW. Differential mortality rates in major and subthreshold depression: meta-analysis of studies that measured both. Br J Psychiatry. 2013;202(1):22–7. doi: 10.1192/bjp.bp.112.112169 23284149 Jackson CA, Kerssens J, Fleetwood K, Smith DJ, Mercer SW, Wild SH. Incidence of ischaemic heart disease and stroke among people with psychiatric disorders: retrospective cohort study. Br J Psychiatry. 2020;217(2):442–9. doi: 10.1192/bjp.2019.250 31753047 Jackson CA, Fleetwood K, Kerssens J, Smith DJ, Mercer S, Wild SH. Incidence of type 2 diabetes in people with a history of hospitalization for major mental illness in Scotland, 2001–2015: a retrospective cohort study. Diabetes Care. 2019;42(10):1879–85. doi: 10.2337/dc18-2152 31471379 Ho IS-S, Azcoaga-Lorenzo A, Akbari A, Black C, Davies J, Hodgins P, et al. Examining variation in the measurement of multimorbidity in research: a systematic review of 566 studies. Lancet Public Health. 2021;6(8):e587–97. doi: 10.1016/S2468-2667(21)00107-9 34166630 The Academy of Medical Sciences. Multimorbidity: a priority for global health research. 2018 [cited 2025 Jan 8]. Available from: https://acmedsci.ac.uk/file-download/82222577 Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study. Lancet. 2012;380(9836):37–43. doi: 10.1016/S0140-6736(12)60240-2 22579043 Qiao Y, Liu S, Zhang Y, Wu Y, Shen Y, Ke C. Bidirectional association between depression and multimorbidity in middle-aged and elderly Chinese adults: a longitudinal cohort study. Aging Ment Health. 2022;26(4):784–90. doi: 10.1080/13607863.2021.1877609 33512250 Arias-de la Torre J, Ronaldson A, Prina M, Matcham F, Pinto Pereira SM, Hatch SL, et al. Depressive symptoms during early adulthood and the development of physical multimorbidity in the UK: an observational cohort study. Lancet Healthy Longev. 2021;2(12):e801–10. doi: 10.1016/S2666-7568(21)00259-2 34901908 Bobo WV, Grossardt BR, Virani S, St Sauver JL, Boyd CM, Rocca WA. Association of depression and anxiety with the accumulation of chronic conditions. JAMA Netw Open. 2022;5(5):e229817. doi: 10.1001/jamanetworkopen.2022.9817 35499825 Li A, Rosella LC, Kurdyak P, Wodchis WP. Depression as a risk factor for physical illness and multimorbidity in a cohort with no prior comorbidity. Can J Psychiatry. 2021;66(8):726–36. doi: 10.1177/0706743720974832 33242993 Niles AN, O’Donovan A. Comparing anxiety and depression to obesity and smoking as predictors of major medical illnesses and somatic symptoms. Health Psychol. 2019;38(2):172–81. doi: 10.1037/hea0000707 30556708 Qiao Y, Ding Y, Li G, Lu Y, Li S, Ke C. Role of depression in the development of cardiometabolic multimorbidity: findings from the UK Biobank study. J Affect Disord. 2022;319:260–6. Xu X, Mishra GD, Jones M. Depressive symptoms and the development and progression of physical multimorbidity in a national cohort of Australian women. Health Psychol. 2019;38(9):812–21. doi: 10.1037/hea0000738 31144827 Triolo F, Sjöberg L, Calderón-Larrañaga A, Belvederi Murri M, Vetrano DL, Fratiglioni L, et al. Late-life depression and multimorbidity trajectories: the role of symptom complexity and severity. Age Ageing. 2023;52(2):afac315. doi: 10.1093/ageing/afac315 36735844 Stapp EK, Williams SC, Kalb LG, Holingue CB, Van Eck K, Ballard ED, et al. Mood disorders, childhood maltreatment, and medical morbidity in US adults: an observational study. J Psychosom Res. 2020;137:110207. Ho ISS, Azcoaga-Lorenzo A, Akbari A, Davies J, Khunti K, Kadam UT, et al. Measuring multimorbidity in research: Delphi consensus study. BMJ Med. 2022;1(1):e000247. doi: 10.1136/bmjmed-2022-000247 36936594 von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med. 2007;4(10):e296. doi: 10.1371/journal.pmed.0040296 17941714 Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12(3):e1001779. doi: 10.1371/journal.pmed.1001779 25826379 UK Biobank. UK Biobank: protocol for a large-scale prospective epidemiological resource. 2007 [cited 2025 Jan 8]. Available from: https://www.ukbiobank.ac.uk/media/gnkeyh2q/study-rationale.pdf Prigge R, Fleetwood KJ, Jackson CA, Mercer S, Kelly PA, Sudlow C, et al. Robustly measuring multiple long-term health conditions using disparate linked datasets in UK biobank. Lancet. 2024. Available from: doi: http://doi.org/10.2139/ssrn.4863974 [preprint] Lewis JD, Bilker WB, Weinstein RB, Strom BL. The relationship between time since registration and measured incidence rates in the General Practice Research Database. Pharmacoepidemiol Drug Saf. 2005;14(7):443–51. doi: 10.1002/pds.1115 15898131 Kuan V, Denaxas S, Gonzalez-Izquierdo A, Direk K, Bhatti O, Husain S, et al. A chronological map of 308 physical and mental health conditions from 4 million individuals in the English National Health Service. Lancet Digit Health. 2019;1(2):e63–77. doi: 10.1016/S2589-7500(19)30012-3 31650125 Khunti K, Routen A, Banerjee A, Pareek M. The need for improved collection and coding of ethnicity in health research. J Public Health (Oxf). 2021;43(2):e270–2. doi: 10.1093/pubmed/fdaa198 33283239 Townsend P. Deprivation. J Soc Policy. 1987;16:125–46. White IR, Royston P, Wood AM. Multiple imputation using chained equations: Issues and guidance for practice. Stat Med. 2011;30(4):377–99. doi: 10.1002/sim.4067 21225900 Rubin DB. Multiple imputation for nonresponse in surveys. New York: Wiley; 1987. R Core Team. R: a language and environment for statistical computing. R version 4.0.0. Vienna, Austria: R Foundation for Statistical Computing; 2020. van Buuren S, Groothuis-Oudshoorn K. Mice: multivariate imputation by chained equations in R. J Stat Softw. 2011;45(3):1–67. Plana-Ripoll O, Chen D, Laustsen LM, Momen NC. The challenges at the core of multimorbidity research. Lancet Psychiatry. 2024;11(6):399–400. doi: 10.1016/S2215-0366(24)00129-9 38642561 Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population. Am J Epidemiol. 2017;186(9):1026–34. doi: 10.1093/aje/kwx246 28641372 Keyes KM, Westreich D. UK Biobank, big data, and the consequences of non-representativeness. Lancet. 2019;393(10178):1297. doi: 10.1016/S0140-6736(18)33067-8 30938313 Stamatakis E, Owen KB, Shepherd L, Drayton B, Hamer M, Bauman AE. Is cohort representativeness passé? Poststratified associations of lifestyle risk factors with mortality in the UK Biobank. Epidemiology. 2021;32(2):179–88. doi: 10.1097/EDE.0000000000001316 33492009 Hanlon P, Jani BD, Nicholl B, Lewsey J, McAllister DA, Mair FS. Associations between multimorbidity and adverse health outcomes in UK Biobank and the SAIL Databank: a comparison of longitudinal cohort studies. PLoS Med. 2022;19(3):e1003931. doi: 10.1371/journal.pmed.1003931 35255092 Firth J, Siddiqi N, Koyanagi A, Siskind D, Rosenbaum S, Galletly C, et al. The Lancet Psychiatry Commission: a blueprint for protecting physical health in people with mental illness. Lancet Psychiatry. 2019;6(8):675–712. doi: 10.1016/S2215-0366(19)30132-4 31324560 MacRae C, Mercer SW, Henderson D, McMinn M, Morales DR, Jefferson E, et al. Age, sex, and socioeconomic differences in multimorbidity measured in four ways: UK primary care cross-sectional analysis. Br J Gen Pract. 2023;73(729):e249–56. doi: 10.3399/BJGP.2022.0405 36997222 Mercer SW, Patterson J, Robson JP, Smith SM, Walton E, Watt G. The inverse care law and the potential of primary care in deprived areas. Lancet. 2021;397(10276):775–6. doi: 10.1016/S0140-6736(21)00317-2 33640047 Hayward CJ, Batty JA, Westhead DR, Johnson O, Gale CP, Wu J, et al. Disease trajectories following myocardial infarction: insights from process mining of 145 million hospitalisation episodes. EBioMedicine. 2023;96:104792. doi: 10.1016/j.ebiom.2023.104792 37741008 10.1371/journal.pmed.1004532.r001 Author response to Decision Letter 0 Submission Version 0

7 Aug 2024

 10.1371/journal.pmed.1004532.r002 Decision Letter 0 Dodd Philippa C. Senior Editor 2025 Philippa C. Dodd 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. Submission Version 0

9 Aug 2024

Dear Dr Fleetwood,

Thank you for submitting your manuscript entitled "Depression and multimorbidity: a cohort study of physical health condition accrual in UK Biobank" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff 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 Aug 13 2024 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 plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Pippa

Philippa C. Dodd, MBBS MRCP PhD

Senior Editor

PLOS Medicine

pdodd@plos.org

 10.1371/journal.pmed.1004532.r003 Author response to Decision Letter 1 Submission Version 1

12 Aug 2024

 10.1371/journal.pmed.1004532.r004 Decision Letter 1 Dodd Philippa C. Senior Editor 2025 Philippa C. Dodd 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. Submission Version 1

22 Oct 2024

Dear Dr Fleetwood,

Many thanks for submitting your manuscript "Depression and multimorbidity: a cohort study of physical health condition accrual in UK Biobank" (PMEDICINE-D-24-02561R1) to PLOS Medicine. The paper has been reviewed by subject experts and a statistician; their comments are included below and can also be accessed here: [LINK]

As you will see, the reviewers were in agreement that the paper dealt with an important area of research. However, they also raised a number of points that will need addressing. After discussing the paper with the editorial team and an academic editor with relevant expertise, I'm pleased to invite you to revise the paper in response to the reviewers' comments. We plan to send the revised paper to some or all of the original reviewers, and we cannot provide any guarantees at this stage regarding publication.

When you upload your revision, please include a point-by-point response that addresses all of the reviewer and the editorial points, indicating the changes made in the manuscript and either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please also be sure to check the general editorial comments at the end of this letter and include these in your point-by-point response. When you resubmit your paper, please include a clean version of the paper as the main article file and a version with changes tracked as a marked-up manuscript. It may also be helpful to check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper.

We ask that you submit your revision by Nov 12 2024 11:59PM. However, if this deadline is not feasible, please contact me by email, and we can discuss a suitable alternative.

Don't hesitate to contact me directly with any questions (ssunny@plos.org).

Best regards,

Syba

Syba Sunny MBBS, MRes, FRCPath

Associate Editor

PLOS Medicine

ssunny@plos.org

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

Comments from the academic editor:

The academic editor was very much in favour of your paper and stated the following: ‘this study is one of the most robust in all respects (sample size, measures, longitudinal design etc) to disentangle the complicated relationship between depression and chronic conditions and has important findings for policy and practice’. Nevertheless, he agreed with the reviewer comments and, thus, asks that the authors revise their manuscript accordingly.

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

Comments from the reviewers:

Reviewer #1: See attachment

Michael Dewey

Reviewer #2: Review - PMEDICINE-D-24-02561R1

Authors present a study examining the association between history of depression and the accumulation of physical diseases using data from UKB. Using a large study population with a wide age range at baseline (40-71 years), they report a faster rate of disease accumulation in people with a history of depression. These associations are preserved after accounting for ethnicity, area-level deprivation, and country of residence. Subsequent adjustment for clinical, behavioural, and functional indicators, as well as prevalent diseases, reduces the effect of depression history, although it remains. The paper is well-written and the analysis is sound. I have some concerns about conceptualizing depression, lack of age-group stratification given the wide age-span of the sample, and overall innovativeness of the study, given the many studies examining depression-multimorbidity connection out there.

1. Authors are not consistent in their terminology/conceptualization of depression. In some sections, they speak about prevalent depression (eg., people with/without depression), whereas in most other places, they write "history of depression". In my reading, their operationalization of depression is more consistent with the notion of "depression history", at least when it comes to the self-report item. For the electronic records this is less clear, and more detail is required from authors here. Does depression assessment from the records only refer to the last eight years, or is it possible to gauge the history of depression from these data? Depending on the answer here, the conceptualization of depression diagnosis will vary. Consequently, if one source of depression diagnosis used here refers to history (self-repot), while the other to the last eight years, then combining them into a single measure may not be suitable. Please provide more detail about conceptualizing depression from the different sources and consider sensitivity analysis with alternative operationalizations of depression (using just one source of diagnosis?) to limit misclassification and ensure consistency between operational definition and its conceptualization.

2. What is the overlap between the self-report of depression and the diagnosis from the electronic records?

3. Did the authors consider episodic nature of depression in their operationalization? Is it handled in any way, particularly for data from electronic records? At the very least, this should be discussed in the limitations.

4. Authors are working with a very wide age range (40-71 years at baseline). We know that both multimorbidity and depression are age-sensitive syndromes, increasing in prevalence with age (multimorbidity), or exhibiting a more chronic course and poor prognosis, with an altered clinical presentation (depression). Indeed, some have argued that old-age depression may be conceptually distinct from middle-aged depression, with an outsized role played by somatic and cognitive symptoms (as opposed to affective/low mood ones). Given such age-dependency in both exposure and outcome, I think it is imperative to not lump all ages together, but rather conduct stratified analysis and discuss age-specific findings at length.

5. To a large extent, this also applies to sex. In fact, authors themselves mention sex differences in the introduction, therefore it would only be natural to revisit them later on in the paper.

6. For the operationalization of chronic conditions, I have several points. 1) Did the authors consider dropping hypertension from disease count and re-calculating accrual rates without it? Some in MM literature have argued that hypertension represents a risk factor rather than an overt disease, and given its high prevalence, the inclusion of hypertension could be misleading. 2) I was unclear about the mental conditions other than depression... Were they considered as part of the total count in the accrual analysis? If they were, authors should do a sensitivity analysis without them, considering the comorbidity of mental disorders, particularly depression and anxiety (especially in older adults). 3) please consider providing baseline prevalences of diseases in the Supplementary Table 1.

7. The inclusion of stressful life events, loneliness, pain, and sleep could be problematic, considering their possible conceptual overlap with depression (indeed sleeping difficultly is one of depression symptoms). This is further complicated by the fact they are assessed concurrently with depression. I suggest their inclusion is reconsidered.

8. Indeed, the analysis in which factors on the causal pathway between depression and disease count are considered as a way of explaining away depression's effect is not well justified in the introduction. As I read, the aim seems to be about predictive effects of depression on multimorbidity change, whereby depression and MM are kept temporally and conceptually distinct to strengthen inferences. This analysis, blurs this conceptual separation, and requires better justification.

9. How much of the drop in the effect of depression is attributable to controlling for prevalent conditions?

10. I have doubts about the usefulness of the results in Table 2 according to prevalent/incident individual diseases. To me, this goes back to lacking age-group perspective. Ignoring huge age differences runs the risk of misrepresenting these cross-tabs. Also, in the limitations, authors write that examining the relationship between depression and specific physical conditions was outside the scope of this study. Still, they devoted a lengthy table 2 to individual diseases. I think the paper would benefit more from an explicit age/sex perspective, considering the size of authors' dataset.

11. In the face of lacking age/sex/disease type/depression severity/course perspective, I am left wondering about the place of this study in the field where depression-multimorbidity associations have been extensively researched before (and synthesized in previous reviews, including scoping and systematic ones). Undoubtedly, authors address some previous gaps (primarily study size and depth of multimorbidity assessment), although they are confronted with others such as depression conceptualization issues and non-representativeness of the study population. And given the lack of an "angle" that helps triangulate these findings a specific and well-defined at-risk subpopulation, I am left wondering exactly what new this study brings to the table (or, to rephrase it, whether should it be published in a high-impact journal like PLOS Medicine).

Reviewer #3: This paper tackles a highly relevant topic, investigating the relationship between depression and the accumulation of physical health conditions (multimorbidity) using the UK Biobank cohort. The study is well-structured, leverages a robust dataset, and contributes valuable insights to the growing body of literature on mental health and its intersection with chronic disease. The paper demonstrates methodological rigor and offers important implications for public health practice and clinical care. However, there are several areas where the manuscript can be improved to enhance clarity, rigor, and interpretability of results.

One question relates to the choice of covariates:

The authors include a lot of covariates in their analysis, and the justification for selecting manyof these is not clear. Were they chosen because of theoretical importance?

Reviewer #4: This study addresses an important research topic on the impact of depression on the subsequent accrual of physical health conditions. The strengths of the work include its large cohort size (172,556) participants and the comprehensive inclusion of conditions in the definition of multimorbidity - which set this paper aside from many of the existing studies in the literature at present. Whilst at face value the analyses is well conducted and presented, in particular, a high quality imputation approach as been implemented - I do have several major concerns regarding the overall analyses and modelling approach which are critical to the overall reliability and quality of the results.

1. Introduction - paragraph on multimorbidity and depression studies being cross-sectional references many of the key papers, however it is incomplete and could be more balanced therefore. e.g. There are studies which look at disease accrual over time/in sequence (e.g. Hayward et al. Lancet eBioMedicine, 2023 Volume 96, 104792 - disease trajectories) which shows data on the time sequenced accrual of mental health related hospital admissions following cardiovascular diseases and its association with increased risk of death. A more comprehensive review of the literature to highlight the novel contribution of this work therefore is recommended.

2. Whilst UK Biobank is a fantastic resource - the disadvantages of its use for multimorbidity research must be highlighted up front. The authors use strong language in the introduction to suggest they "robustly" identify physical conditions, and whilst I appreciate they have gone over and above existing studies in their attempts to be much more inclusive and comprehensive in accrual of conditions than others - the level of selection bias and healthy user bias inherent in UK Biobank does limit the robustness of this assessment. Further linked primary care data is available for fewer than 50% of the cohort. Further data on the level of linkage to other data sources do not feature strongly (or at all?) - and these all affect the level of confidence in ascertainment and the equality of ascertainment of both depression and the physical conditions for individuals in the study. Moreover over half the cohort are excluded given the lack of primary care linkage - I realise this is a necessary choice given the data but this does introduce further queries over the representativeness of these data and its conclusions which need to be fully acknowledged as limitations. The language in the intro and discussion need to be appropriately moderated to reflect this also.

Statistical analyses

3. High quality imputation approach has been used (which is rare - and therefore commended)! - the approach follows good practice guidelines for MICE such as including the outcome variable and all covariates and their formulations, as well as a comparison with complete cases. Please state whether any additional auxiliary variables were included in the imputation models or not, and include a detailed description of missing data for all variables (e.g. by an extra column in Table 1). Missing data are reported for some but not all variables in the table - e.g. it is hard to believe there were no missing data at all for BMI and alcohol intake - yet missing data for these are not included in Table 1.

4. Table 2 presents details of those with and without depression according to the conditions at baseline and those accrued during the follow up period. It is not clear however what the timing of those baseline conditions in relation to depression is? They can be before or after the depression as far as I can tell - which makes interpretation of this table difficult. Could the authors elaborate on this? How does this impact on your results?

5. "Baseline depression" has been identified as any occurrence of depression recorded during an 8 year medical history - however this is likely to lead to a heterogenous population with differential impact on outcomes for e.g. individuals who had one short episode of depression related to an acute event vs. those with longer term/persistent depression or mental health conditions over the length of follow up. Has this been accounted for as these two scenarios are likely to influence the impact of depression physical health conditions very differently.

6. Observational studies of this type should at minimum use some form of matching between cohorts to minimise additional sources of bias. I would expect to see either a propensity score analyses, or other form of matching (e.g. risk-set matching approach) minimise bias between the groups (esp. given concerns about ascertainment highlighted below).

7. Given that ascertainment of depression and physical health conditions will vary hugely between primary, secondary care and the UK biobank baseline data it is important to understand how many cases included in the study were linked with the various datasets and over what time periods these linkages were present. Those with fully linked data may have an increase in physical health conditions recorded and a higher chance of a record for depression vs those who have fewer linked datasets and it would be important to know the degree with which this varies for individuals in the cohort. How does this impact on your results, and has this been accounted for in any way?

8. The use of centring and quadratics in the modelling is a little outdated and requires either further justification plus clear model diagnostics to be presented (e.g. why was age divided by 10 specifically? How much improvement in model fit were you looking for in the quadratics? centring is usually done to aid interpretation, but as these are just included as confounders and should not to be directly assessed (See point later about "table 2 fallacy" - there seems little point). This approach overall is a little inefficient and requires a number of steps and judgements to be made in the process. Accounting for non-linearity in variables is better done via the use of restricted cubic splines which estimates all transformations simultaneously (instead of adding quadratic terms for all predictors and then determining which ones should be kept). This approach would also lead to appropriate confidence intervals of your estimates - which are not influenced by scale of the data and therefore do not require centring.

9. The modelling approach does not appear to take different follow up periods for individuals and any censoring into account (as far as I could discern from the paper). Lack of accounting for this will cause major biases in your model which should not be published unless addressed (e.g. at minimum through inclusion of an appropriate offset term in the Poisson model, but ideally through use of a time to event modelling framework). In the current analyses it is far more likely to identify someone in the "depression" group if they have more follow up time (i.e. there is more opportunity to identify their existing diagnoses) and therefore any comparisons between a depressed and non-depressed group is likely to be inherently biased. In addition to that - the outcome (number of conditions accrued) is heavily influenced by length of follow up time and censoring - which therefore must be addressed in the modelling directly.

10. The inclusion of a good number of confounders is commended - however - further thought needs to be provided as to why those variables are indeed included. Were other data available and discounted, are all these variables definitely confounders and not mediators? The use of a DAG in this instance of many variables collected for a dataset such as these is highly recommended, which would also identify any further potentially important confounders which were not available to be listed in the discussion.

11. The authors present their modelling data in such a way that falls foul of the so called "Table 2 Fallacy" - which is when estimates of confounders are presented and interpreted directly as exposures in their own right whilst adjusting for all other confounders. The only relevant estimates to be interpreted and presented for this particular model is the RRs for depression vs no depression. It's fine to include a model with several "layers" of confounders, although STROBE/RECORD guidance recommends the reporting of unadjusted RRs in addition to these also. The RRs for all the confounders do not have / should not be interpreted independently in any way and therefore should not be presented in the table or described as they are in the results section. The only act as confounders in this case. If the authors are additionally interested in the effects of age on morbidity accrual - then a new model should be generated with age as the main exposure and a bespoke set of confounders relevant to the age-outcome relationship should be identified, and this should also be added to the original aims of the presented work.

12. Figure 2 shows a plot of mean number of long term conditions - this is a simplistic view of these data which may be biased by not accounting for differential follow up time for individuals contributing to these data, and confidence intervals should be displayed for these curves also. The use of formal time to event analyses accounting directly for censoring and length of follow up time is advised here (e.g. calculating cumulative incidence functions) and displayed with confidence intervals. It is also good practice to list the numbers at risk in time intervals (e.g. for each year of follow up) for each of these plots so that the accuracy of the longer term data (which is a stated strength of this paper). There will be fewer individuals with longer term follow up data - and the drop off rate of this should be clearly presented.

13. Aside from the table 2 fallacy as already discussed - the strength of association cannot be compared in this way with respect to size of estimate; therefore this statement and others similar throughout the manuscript should be removed : " The association between depression and the accrual of physical conditions was of a similar magnitude to associations between other characteristics and condition accrual, although weaker than associations with age, smoking, and BMI in the fully adjusted model".

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

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

General editorial requests:

(Note: not all will apply to your paper, but please check each item carefully)

* 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. Please do not add or remove authors without first discussing this with the handling editor. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

* Please 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 PLOSMedicine@plos.org.

* Thank you for including a completed STROBE checklist. Please move this document to ‘Supporting Information’ and add the following statement, or similar, to the Methods: "This study is reported as per [XXXX] guideline (S1 Checklist)."

FORMATTING - GENERAL

* 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. Ideally each sub-heading should contain 2-3 single sentence, concise bullet points containing the most salient points from your study. In the final bullet point of 'What Do These Findings Mean?', please include the main limitations of the study in non-technical language. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary.

* When reporting 95% CIs please separate upper and lower bounds with commas instead of hyphens as the latter can be confused with reporting of negative values.

* Please include page numbers and line numbers in the manuscript file. Use continuous line numbers (do not restart the numbering on each page).

FIGURES AND TABLES

* Please provide titles and legends for all figures and tables (including those in Supporting Information files).

* Please define all abbreviations used in each figure/table (including those in Supporting Information files).

* Please consider avoiding the use of red and green in order to make your figure more accessible to those with color blindness.

SUPPLEMENTARY MATERIAL

* Please cite your Supporting Information as outlined here: https://journals.plos.org/plosmedicine/s/supporting-information

REFERENCES

* PLOS uses the numbered citation (citation-sequence) method and first six authors, et al.

* Please ensure that journal name abbreviations match those found in the National Center for Biotechnology Information (NCBI) databases (http://www.ncbi.nlm.nih.gov/nlmcatalog/journals), and are appropriately formatted and capitalised.

* Where website addresses are cited, please include the complete URL and specify the date of access (e.g. [accessed: 12/06/2023]).

* Please also see https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references for further details on reference formatting.

OBSERVATIONAL STUDIES

* Abstract: We ask that Abstracts include the following information: the study design, population and setting, number of participants, years during which the study took place (enrollment and follow up), length of follow up, and main outcome measures. Could you kindly revise your Abstract to ensure all criteria are met, please?

* For all observational studies, in the manuscript text, please indicate: (1) the specific hypotheses you intended to test, (2) the analytical methods by which you planned to test them, (3) the analyses you actually performed, and (4) when reported analyses differ from those that were planned, transparent explanations for differences that affect the reliability of the study's results. If a reported analysis was performed based on an interesting but unanticipated pattern in the data, please be clear that the analysis was data driven.

* Please state in the Methods section whether the study had a prospective protocol or analysis plan. If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant document(s) with your revised manuscript as a Supporting Information file to be published alongside your study and cite it in the Methods section. A legend for this file should be included at the end of your manuscript. If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place. Changes in the analysis, including those made in response to peer review comments, should be identified as such in the Methods section of the paper, with rationale.

Submitted filename: fleetwood.pdf

 10.1371/journal.pmed.1004532.r005 Author response to Decision Letter 2 Submission Version 2

3 Dec 2024

Submitted filename: Response_to_reviewers_29Nov2024.docx

 10.1371/journal.pmed.1004532.r006 Decision Letter 2 Van Epps Heather Senior Editor 2025 Heather Van Epps 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. Submission Version 2

7 Jan 2025

Dear Dr. Fleetwood,

Thank you very much for re-submitting your manuscript "Depression and physical multimorbidity: a cohort study of physical health condition accrual in UK Biobank" (PMEDICINE-D-24-02561R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by three of the original reviewers, whose comments are included below. I am pleased to say that provided the remaining editorial and production issues are dealt with, we plan 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.***

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.

A reminder that 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.

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 plosmedicine@plos.org.

We expect to receive your revised manuscript within 1 week. Please email me directly (hvanepps@plos.org) if you have any questions or concerns. Otherwise, we look forward to receiving the revised manuscript by Wed, Jan 15th,

Kind regards,

Heather

Heather Van Epps, PhD

Executive Editor

PLOS Medicine

hvanepps@plos.org

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

Requests from Editors:

1. Abstract, line 34-35 and line 36: Please refrain from expressing RRs as fold (or times)-increase; suggest modifying to “…participants with depression accrued physical morbidities at a faster rate than those without depression (RR 1.32…)…”

2. Abstract. Please include a sentence describing the main limitation(s) of the study at the end of the Methods and findings section.

3. For references to URLs (eg, ref 7, 21), please add the date accessed in brackets.

4. Please add a URL to the sponsor’s website to your funding statement.

5. Please add a URL for UK Biobank to your Data availability statement.

Comments from Reviewers:

Reviewer #1:

The authors have addressed my points

Michael Dewey

Reviewer #2:

Authors have adequately addressed my comments and the paper has been much improved. I have no further comments.

Reviewer #4:

All my comments have been thoughtfully and thoroughly addressed or clarified , and the manuscript is much improved overall. I do not have further recommendations to make and wish the authors success with this work.

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

[LINK]

 10.1371/journal.pmed.1004532.r007 Author response to Decision Letter 3 Submission Version 3

9 Jan 2025

Submitted filename: Response to requests from editors_9Jan2025.docx

 10.1371/journal.pmed.1004532.r008 Decision Letter 3 Van Epps Heather Senior Editor 2025 Heather Van Epps 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. Submission Version 3

13 Jan 2025

Dear Dr Fleetwood, 

On behalf of my colleagues and the Academic Editor, Vikram Patel, I am pleased to inform you that we have agreed to publish your manuscript "Depression and physical multimorbidity: a cohort study of physical health condition accrual in UK Biobank" (PMEDICINE-D-24-02561R3) 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 medicinepress@plos.org. 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/.

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

Kind regards,

Heather

Heather Van Epps, PhD 

Executive Editor 

PLOS Medicine