Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2023 Sep 8;18(9):e0291295. doi: 10.1371/journal.pone.0291295

Risk factors for the progression to multimorbidity among UK urban working-age adults. A community cohort study

Anne L Stagg 1,2,*, Lisa Harber-Aschan 1,3, Stephani L Hatch 1, Nicola T Fear 4,5, Sarah Dorrington 1,6, Ira Madan 2, Sharon A M Stevelink 1,5
Editor: Andrea Dell’Isola7
PMCID: PMC10490989  PMID: 37682940

Abstract

Objectives

The progression of long-term conditions (LTCs) from zero-to-one (initiation), and from one-to-many (progression)are common trajectories that impact a person’s quality of life including their ability to work. This study aimed to explore the demographic, socioeconomic, psychosocial, and health-related determinants of LTC initiation and progression, with a focus on work participation.

Methods

Data from 622 working-age adults who had completed two waves (baseline and follow-up) of the South-East London Community Health survey were analysed. Chi square tests and multinomial logistic regression were used to describe the associations between self-reported demographic, socioeconomic, psychosocial, and health-related variables, and the progression of LTCs.

Results

Small social networks, an increased number of stressful life events, low self-rated health, functional impairment, and increased somatic symptom severity were all associated with both the progression from zero-to-one LTC and from one LTC to multimorbidity (two or more LTCs). Renting accommodation (RRR 1.73 [95% CI 1.03–2.90]), smoking (RRR 1.91 [95% CI 1.16–3.14]) and being overweight (RRR 1.88 [95% CL 1.12–3.16]) were unique risk factors of developing initial LTCs, whereas low income (RRR 2.53 [95% CI 1.11–5.80]), working part-time (RRR 2.82 ([95% CL 1.12–7.10]), being unemployed (RRR 4.83 [95% CI 1.69–13.84]), and making an early work exit (RRR 16.86 [95% CI 3.99–71.30]) all increased the risk of progressing from one LTC to multimorbidity compared to being employed full-time. At follow-up, depression was the most prevalent LTC in the unemployed group whereas musculoskeletal conditions were the most prevalent in those working.

Conclusions

The journey to multimorbidity is complex, with both common and unique risk factors. Non-full-time employment was associated with an increased risk of progression to multimorbidity. Future research should explore the risk and benefit pathways between employment and progression of LTCs. Interventions to prevent progression of LTCs should include mitigation of modifiable risk factors such as social isolation.

Introduction

The number of people living with multiple long-term health conditions (MLTCs, multimorbidity) is rising, with older age, socioeconomic deprivation and poor health behaviours being key risk factors [1–4]. The journey to deteriorating health often starts with the acquisition of an initial long-term condition (LTC), with additional LTCs developing over time (negative health transitions), leading to an individual living with the impact of multimorbidity [5]. Multimorbidity is now developing earlier in the life course; studies in England and Scotland have shown that in absolute numbers, there are more people with multimorbidity below the age of 65 than over 65 [2, 6]. Although the development of LTCs increases with age, the current research focus on LTCs in the older age groups does not explore the risk factors for LTC acquisition or the impact of living with MLTCs in younger, working-age adults. Indeed, it has been reported that more than a third of patients with three-or-more LTCs that lead to severe functional impairment and high health and social care requirements, are between the ages of 18 and 64 years [5].

There is a close relationship between quality of life and functional status, with one key indicator of functional status being the ability to engage in employment. The association between employment and long-term conditions is complex and known to have a bi-directional relationship [7]. Health conditions can impact workers’ ability to engage in employment, but unemployment and poor working conditions among those in work, can also impact health [8, 9]. Unemployment is associated with poorer physical and mental health, with the risk of having both one and multiple LTCs greater in those who are unemployed compared to employed [8, 10]. However, the impact of employment on the development of initial and multiple LTCs is under-researched, and few studies focus exclusively on working-age adults [8].

Among working-age adults, known risk factors for multimorbidity not only include increasing age and socioeconomic deprivation, but also obesity and smoking [6, 11]. However, whilst there is evidence that socioeconomic deprivation influences the transition to multimorbidity [4], there is a lack of information about individual socioeconomic, psychosocial, behavioural, and health-related influencers of developing a first LTC, and of progression from one-to-many LTCs. The objective of the present study is to address this research gap by exploring a wide range of potential risk factors and LTC development and progression within a working-age community sample. By using two time points, the primary aim of this study is to explore relationships between demographic, socioeconomic (including employment status), psychosocial (social isolation and stressful life events), and health-related determinants at the baseline, and two negative health transitions between baseline and follow-up of (1) initiation (0 to one-or-more LTCs), and (2) progression of LTCs (one LTC to multimorbidity). This objective also aims to gain a better understanding of demographic and socioeconomic inequalities that influence health disparities in the development and progression of LTCs [12]. The secondary aim is to provide information about the distribution of the most common LTCs at follow-up within the different categories of employment status.

Methods

Study design, setting and participants

This paper utilises data from adults aged 16 to 64 years who participated in the South East London Community Health (SELCoH) study [13, 14]. The SELCoH study assessed physical and psychiatric morbidity in adults aged 16 and overliving in randomly selected households in Lambeth and Southwark. Survey data from the first (baseline) and second waves (follow-up) of the SELCoH study (S1 and S2) were used in which household interviews were carried out in 2008–2010 and 2011–2013. Demographic characteristics of participants who responded at both waves, and reasons for non-response at the second wave have previously been reported [14, 15]. Of the 1052 adults who completed both surveys, the present study used data from working-age adults who completed face-to-face interviews at both baseline and follow-up, and were within the age range of 16 years at baseline and 64 years at follow-up.

Ethical approval for the baseline SELCoH survey was given by King’s College London Research Ethics Committee (CREC/07/08-152), and the follow-up survey wave by King’s College London Psychiatry Nursing and Midwifery Research Ethics Committee (PNM/10/11-106). Participants gave written informed consent to participate in the study before taking part.

Study variables

Long-term health conditions

In the SELCoH questionnaire, a long-term condition was defined as a long-standing illness, disability or infirmity troubling a participant over a period of time, or considered likely to in the future. Participants were able to select a specific LTC from a list of 15 health conditions, and in addition, specify any other condition not on the list. The final list of LTCs was constructed by allocating conditions either into specific condition groups (for example, two condition groups were asthma and high blood pressure), or into groups of similar conditions where individual case numbers were <5 (for example, gastrointestinal conditions and respiratory conditions). No participant had more than one condition within any of the individual combined condition groups. A final list of 17 conditions/groups of conditions was derived from all self-reported conditions in addition to a general non-specific LTC group for conditions with less than 5 cases that could not be combined with other similar conditions. A summary description of the LTC groups has been previously published [10].

Independent variables

All variables were measured at the baseline (S1) except for ethnicity which was assessed at the follow-up interviews (S2). Demographic variables consisted of gender (male; female), age, relationship status, ethnicity, and migrant status. Due to small counts for LTCs in younger participants, age was categorised into four groups: 16–34, 35–44, 45–54 and 55–64 years. Relationship status was categorised into three groups comprising married or cohabiting, single and never married, and divorced, separated, or widowed. Ethnicity was categorised into three groups comprising White British, Black African / Black Caribbean, and Other ethnic groups. Migrant status was determined using country of birth and number of years residing in the UK and was categorised into 3 groups: non-migrants (born in the UK), migrants living in the UK for ≤ 10 years, and migrants living in the UK for ≥ 11 years.

Socioeconomic variables included housing tenure, highest education level attained, gross annual household income, employment, and benefit receipt. Housing tenure was reported as owning a property with or without a mortgage, paying rent or part rent, and rent-free accommodation. Educational attainment was categorised into 4 categories: degree level or above, A level or equivalent, GCSE level or equivalent, and no qualifications. Gross annual household income was based on four income bands: (1) £31,495 or more (2) £12,098-£31,494 (3) £0–12,097 and (4) Don’t know. Employment was categorised into working full time, working part time or casual work, unemployed, students, and two economically inactive groups: early work exit and ‘at home’. Early work exit included those who were not available for work due to either having chosen to retire early (before age 64) or could not work due to being registered disabled or permanently sick. ‘At home’ indicated those who were at home looking after children. Benefit receipt was categorised into a binary variable of benefit receipt/no benefit receipt and excluded benefits relating to being sick or disabled due to potential increased predictor-outcome associations between benefit receipt and long-term condition status.

Psychosocial variables included social network size and stressful life events. Social network size was determined from the total number of weekly contacts with different groups of people, and categorised into three groups: ≥ 5, 3–4, and ≤ 2. The number of stressful life events was assessed using a checklist of childhood and lifetime events considered relevant to urban populations [16, 17], and the cumulative score categorised into ≤ 2, 2–5 and ≥ 6.

Health risk factors were body mass index (BMI), smoking and alcohol intake. BMI was calculated from weight and height measurements taken at the initial interview. Categorisation of BMI into four groups were: underweight (<18.4 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), and obese (≥30.0 kg/m2). Smoking was categorised as non-smokers (including past smokers), and current smokers (regular and sporadic). Hazardous alcohol intake was determined from a cumulative score of 8–40 on the Alcohol Use Disorders Identification Test (AUDIT) questionnaire, and moderate intake indicated by a score of 1–7 [13, 18, 19].

Health-related quality of life was measured using three individual items on the 12-item Short Form (SF-12) questionnaire to assess self-rated health, and subjective measures of functional limitations affecting the ability to work or other activities due to physical and emotional health (anxiety or depression) during the previous four weeks [20]. The 5-point measure of self-rated health was categorised into three groups: excellent/very good, good, and fair/poor, whereas limitations due to physical and emotional health were reported as binary responses of no/yes.

Somatic symptom severity was measured using the cumulative score from 14 of the 15 items in the Patient Health Questionnaire (PHQ-15, [21], excluding the question relating to menstrual cramps due to only being applicable to females [15]. Consistent with a previous study using the SELCoH community population, the cumulative score was categorised into low (<5), medium (5–9) and high (10–28) somatic symptom severity [15]).

Progression of LTCs

The number of LTCs were categorised at both baseline and follow-up to enable the investigation of negative health transitions. This comprised a baseline comparable ‘healthy’ group with no LTCs at baseline or follow-up, and two groups that showed negative health transitions; a group who progressed from no LTCs at baseline to one-or-more LTCs at follow-up (initiation group), and a group who progressed from one LTC at baseline to two-or-more LTCs at follow-up (progression group). Data from those who had more than one LTCs at baseline, or who had one-or-more LTCs at baseline but either remained stable or reported fewer or no LTCs at follow-up (n = 271), were excluded from the statistical analysis since the focus of this study was on factors associated with negative health transitions. The final LTC trajectory sample was 622 and the percentage of missing data was below 1.5%.

Statistical analyses

All analyses were carried out using Stata/MP 16.1, applying robust standard errors using survey commands (svy) for prevalence estimates and associations. Weighting was applied to adjust for household clustering, non-response, attrition, and changes in household composition between baseline and follow-up. Descriptive statistics are reported using unweighted frequencies and weighted percentages, and Pearson’s χ2 tests with Rao & Scott corrections for survey data. Multinomial logistic regressions were used to analyse associations between each individual indicator variable measured at baseline with the three categories of (1) no LTC at either baseline or follow-up (the reference category), (2) development from no LTCs at baseline to 1-or-more LTCs at follow-up (initiation) and (3) progression from 1 LTC at baseline and multiple LTCs at follow-up. Results from multinomial logistic regression report relative risk ratios (RRR) and 95% confidence intervals. for LTC initiation and progression between baseline and follow-up (S1 and S2). Two models were tested; unadjusted and also adjusted for possible confounding by gender and age (as a continuous variable, in years). The overall distribution of the eight most prevalent LTCs within each employment category at follow-up are reported and compared descriptively.

Results

LTCs and health transitions

In the whole sample (n = 893), the healthy group of participants with no LTCs at either baseline or follow-up accounted for 56% (n = 479), whereas 15.4% (n = 143) of the sample reported negative health transitions with 10.5% (n = 97) developing LTCs at follow-up, and 4.9% (n = 46) progressed from one-to-many LTCs between baseline and follow-up. Positive health transitions were also reported, with 52 participants (5.9%) reporting at least one LTC at baseline but no LTCs at follow-up, and 31 participants (3.0%) reporting two or more LTCs at baseline had an improved LTC status at follow-up, only reporting one LTC. 77 participants reported having no LTCs at baseline but a first LTC at follow-up, with depression (n = 15, 21.5%), musculoskeletal conditions (n = 13, 14.3%), asthma (n = 10, 12.8%) and high blood pressure (n = 8, 9.7%) being the most prevalent first conditions to be reported. The six most prevalent conditions in those reporting progressing from one to MLTCs at follow-up were musculoskeletal conditions (n = 21, 42.6%), depression (n = 16, 36.5%), asthma (n = 10, 22.6%), gastrointestinal conditions (n = 10, 24.7%), high blood pressure (n = 10, 20.8%) and migraine (n = 10, 22.1%).

The distribution of the stable and negative LTC trajectories used in the analysis are shown in Table 1 (n = 622). For these three LTC trajectories, the average time between S1 and S2 was 2.46 years (SD ± 0.55).

Table 1. Baseline demographic, socioeconomic, employment, psychosocial and health-related factor distribution of LTC trajectories.

No LTCs a 0-to- ≥ 1 LTC b 1-to-MLTCs c
(n = 479) (n = 97) (n = 46)
n (row %) n (row%) n (row %) χ2 p-value d
Demographic variables
Gender 0.390
 Female 269 (76.2) 62 (16.4) 28 (7.4)
 Male 210 (80.8) 35 (12.9) 18 (6.3)
Age (years) <0.001
 16–34 284 (85.1) 38 (11.5) 11 (3.5)
 35–44 106 (76.0) 22 (15.7) 11 (8.3)
 45–54 68 (61.9) 26 (23.0) 16 (15.1)
 55–64 21 (53.1) 11 (26.5) 8 (20.4)
Marital status 0.072
 Married/Cohabiting 234 (77.5) 53 (16.1) 21 (6.4)
 Single 209 (81.3) 33 (12.5) 17 (6.2)
 Divorced/Separated/Widowed 36 (65.1) 11 (20.0) 8 (15.0)
Ethnicity 0.180
 White British 229 (78.5) 44 (14.1) 23 (7.4)
 Black Caribbean & Black African 119 (84.1) 20 (12.5) 5 (3.4)
 Other 131 (73.7) 33 (17.4) 18 (8.9)
Migrant Status 0.066
 Born in UK 303 (77.8) 62 (14.8) 32 (7.4)
 0–10 years in the UK 107 (85.9) 15 (11.2) 4 (2.9)
 11 or more years in the UK 68 (70.5) 20 (19.4) 10 (10.2)
Socioeconomic variables
Household income (gross, annual) 0.169
 £31,495 or more 261 (80.8) 47 (13.6) 20 (5.6)
 £12,098–31,494 93 (75.3) 28 (20.0) 7 (4.7)
 £0–12,097 68 (75.1) 12 (12.8) 11 (12.2)
 Do not know 56 (77.7) 10 (13.1) 8 (9.2)
Housing tenure
 Own/ mortgage 164 (78.5) 30 (12.4) 20 (9.2) 0.110
 Rent/part rent 265 (76.5) 62 (17.0) 25 (6.6)
 Rent free 49 (88.9) 5 (9.3) 1 (1.9)
Education (highest level attained) 0.231
 Degree level or above 249 (81.9) 40 (12.5) 19 (5.6)
 Up to A level 123 (77.1) 29 (16.1) 13 (6.9)
 Up to GCSE 78 (76.3) 17 (14.8) 9 (8.9)
 No qualifications 27 (64.0) 11 (24.6) 5 (11.4)
Employment <0.001
 Employed full-time 235 (81.5) 46 (15.1) 11 (3.3)
 Employed part-time 83 (79.8) 19 (15.9) 12 (9.3)
 Unemployed 39 (67.5) 13 (20.5) 7 (12.1)
 Student 97 (87.9) 8 (7.2) 6 (4.9)
 Early work exite 5 (31.8) 4 (24.1) 7 (44.0)
 At home 20 (67.4) 7 (23.4) 3 (9.2)
Benefit receipt (not health-related) 0.023
 No benefits 400 (80.5) 72 (13.5) 34 (5.9)
 Benefits 79 (69.1) 25 (19.8) 12 (11.1)
Psychosocial variables
Size of social network 0.002
 ≥5 358 (81.3) 62 (13.0) 29 (5.8)
 3–4 104 (76.1) 25 (16.7) 11 (7.2)
 ≤2 17 (51.6) 10 (28.5) 6 (20.0)
Stressful life events 0.003
 ≤2 202 (83.8) 32 (12.6) 9 (3.6)
 3–5 207 (78.5) 40 (14.2) 22 (7.4)
 ≥6 70 (66.1) 25 (20.8) 15 (13.0)
Health Risk Factors
Body Mass Index 0.031
 Normal weight 254 (81.7) 35 (10.8) 25 (7.6)
 Underweight 14 (100) 0 (0) 0 (0)
 Overweight 140 (73.3) 41 (19.9) 14 (6.8)
 Obese 63 (73.7) 19 (19.8) 7 (7.0)
Smoking
 Non-smoker 376 (80.9) 66 (12.9) 32 (6.2) 0.033
 Current Smoker 103 (70.8) 31 (20.2) 14 (8.9)
Alcohol Use 0.708
 Non-drinkers (AUDIT score 0) 91 (81.4) 15 (12.7) 7 (5.9)
 Moderate alcohol use (1–7) 289 (78.8) 62 (15.0) 26 (6.2)
 Hazardous alcohol use (8–40) 99 (75.2) 20 (15.4) 13 (9.4)
Health-Related Quality of Life
Self-rated health <0.001
 Excellent/very good 299 (86.9) 39 (10.6) 10 (2.6)
 Good 153 (72.6) 43 (18.5) 21 (8.9)
 Fair/Poor 27 (50.4) 15 (24.7) 15 (25.0)
Functional Impairment due to physical health <0.001
 No 446 (80.6) 84 (14.0) 34 (5.4)
 Yes 33 (56.6) 13 (21.8) 12 (21.6)
Functional impairment due to emotional health 0.002
 No 429 (80.9) 74 (13.0) 37 (6.1)
 Yes 50 (62.7) 23 (25.3) 9 (12.1)
Somatic symptoms
Somatic symptom severity <0.001
 Low (0/4) 347 (83.9) 54 (12.1) 19 (4.1)
 Medium (5/9) 107 (70.3) 29 (17.2) 20 (12.5)
 High (≥10) 25 (56.6) 14 (29.6) 7 (13.8)
Open in a new tab

Frequencies are unweighted and weighted percentages account for survey design. Numbers may not add up to the total sample (N = 622) due to missing data. All variables were measured at baseline (S1) except ethnicity which was determined at follow-up (S2).

a No LTCs: S1 = 0, S2 = 0

b 0-to-≥ 1 LTC: S1 = 0 LTC, S2 ≥1 LTCs

c 1-to-MLTCs: S1 = 1 LTC, S2 ≥2 LTCs

d Pearson’s χ2 test with Rao & Scott correction for survey data. Bold indicates significance (p < 0.05)

e Early work exit: permanently sick/disabled or early retirement

Risk factors of negative health transitions

The unadjusted and adjusted relative risk ratios and 95% confidence intervals are shown in Table 2, with values adjusted for age and gender. Age was the primary demographic indicator of negative health transitions, with the 45–54 and 55–64 age group having an increased the risk of developing initial LTCs between baseline and follow-up. All age groups were at increased risk of progressing from one-to-many LTCs when compared to the youngest age band of 16–34 years. Once age and gender were accounted for in the adjusted model, being single was also associated with an increased risk of progressing from one-to-many LTCs.

Table 2. Unadjusted and adjusteda multinomial regression analysis estimating associations between demographic, socioeconomic, psychosocial and health factors at baseline, and LTC trajectory groups.

Unadjusted Adjusted
0 to ≥1 LTCsb 1-to-Many LTCsc 0 to ≥1 LTCsb 1-to-Many LTCsc
RRR (95% CI) RRR (95% CI) RRR (95% CI) RRR (95% CI)
Demographic variables
Gender (n = 622)
 Female 1.00 1.00 1.00 1.00
 Male 0.74 (0.47–1.18) 0.80 (0.42–1.52) 0.71 (0.44–1.14) 0.72 (0.38–1.38)
Age (years) (n = 622)
 16–34 1.00 1.00 1.00 1.00
 35–44 1.53 (0.85–2.75) 2.71 (1.15–6.37) 1.54 (0.85–2.76) 2.72 (1.16–6.39)
 45–54 2.74 (1.54–4.90) 6.03 (2.62–13.89) 2.76 (1.54–4.94) 6.06 (2.62–14.00)
 55–64 3.70 (1.60–8.54) 9.46 (3.33–26.84) 3.90 (1.68–9.07) 9.93 (3.47–28.42)
Marital status (n = 622)
 Married/Cohabiting 1.00 1.00 1.00 1.00
 Single 0.74 0.45–1.21 0.92 (0.47–1.79) 1.16 (0.66–2.03) 2.24 (1.02–4.92)
 Divorced/Separated/Widowed 1.48 (0.70–3.12) 2.79 (1.13–6.88) 1.13 (0.51–2.50) 1.83 (0.73–4.57)
Ethnicity (n = 622)
 White British 1.00 1.00 1.00 1.00
 Black Caribbean & Black African 0.83 (0.46–1.50) 0.42 (0.15–1.16) 0.94 (0.51–1.73) 0.55 (0.19–1.55)
 Other 1.31 (0.78–2.20) 1.27 (0.63–2.59) 1.41 (0.83–2.40) 1.47 (0.70–3.05)
Migrant status (n = 621)
 Born in UK 1.00 1.00 1.00 1.00
 0–10 years in the UK 0.69 (0.37–1.27) 0.36 (0.12–1.06) 0.77 (0.41–1.44) 0.47 (0.16–1.38)
 11 or more years in the UK 1.45 (0.82–2.55) 1.50 (0.68–3.31) 1.18 (0.64–2.16) 1.05 (0.44–2.49)
Socioeconomic variables
Housing Tenure (n = 621)
 Own/ mortgage 1.00 1.00 1.00 1.00
 Rent/part rent 1.41 (0.86–2.31) 0.73 (0.38–1.40) 1.73 (1.03–2.90) 1.02 (0.50–2.08)
 Rent free 0.67 (0.24–1.86) 0.18 (0.24–1.38) 1.08 (0.36–3.19 0.39(0.05–3.39)
Education (highest level attained) (n = 620)
 Degree level or above 1.00 1.00 1.00 1.00
 Up to A-Level 1.37 (0.80–2.35) 1.30 (0.61–2.77) 1.64 (0.94–2.87) 1.74 (0.82–3.71)
 Up to GCSE 1.27 (0.66–2.44) 1.69 (0.75–3.81) 1.53 (0.79–2.99) 2.17 (0.95–4.94)
 No qualifications 2.52 (1.14–5.57) 2.60 (0.88–7.68) 2.15 (0.95–4.87) 1.87 (0.61–5.70)
Household Income (gross, annual) (n = 621)
 £31,495 or more 1.00 1.00 1.00 1.00
 £12,098–31,494 1.58 (0.92–2.71) 0.91 (0.34–2.47) 1.60 (0.93–2.75) 0.91(0.33–2.53)
 £0–12,097 1.01 (0.50–2.04) 2.35 (1.05–5.24) 1.07 (0.52–2.18) 2.53 (1.11–5.80)
 Do not know 1.00 (0.47–2.14) 1.72 (0.69–4.26) 1.30 (0.60–2.85) 2.69 1.07–6.73)
Employment (n = 618)
 Employed Full time 1.00 1.00 1.00 1.00
 Employed part time 1.14 (0.63–2.1) 3.06 (1.27–7.35) 1.09 (0.59–2.01) 2.82 (1.12–7.10)
 Student 0.44 (0.20–0.99) 1.36 (0.51–3.63) 0.61 (0.26–1.42) 2.84 (0.93–8.73)
 Unemployed 1.64 (0.80–3.36) 4.37 (1.56–12.22) 1.73 (0.84–3.58) 4.83 (1.69–13.84)
 Early work exit 4.08 (1.04–16.11) 33.84 (9.14–125.29) 2.99 (0.69–12.94) 16.86 (3.99–71.30)
 At home 1.87 (0.73–4.79) 3.34 (0.84–13.23) 1.63 (0.36–4.26) 2.91 (0.70–12.13)
Benefit receipt (n = 618)
 No benefits 1.00 1.00 1.00 1.00
 Benefits 1.71 (1.01–2.87) 2.19 (1.05–4.55) 1.58 (0.93–2.69) 2.03 (0.93–4.42)
Psychosocial variables
Size of social network (n = 622)
 ≥5 1.00 1.00 1.00 1.00
 3–4 1.38 (0.82–2.32) 1.32 (0.62–2.79) 1.50 (0.87–2.56) 1.52 (0.71–3.26)
 ≤2 3.46 (1.46–8.24) 5.42 (1.94–15.13) 3.31 (1.38–7.95) 5.02 1.69–14.94)
Stressful life events (n = 622)
 ≤2 1.00 1.00 1.00 1.00
 3–5 1.20 (0.72–2.00) 2.17 (0.94–5.01) 1.19 (0.70–2.02) 2.05 (0.88–4.81)
 ≥6 2.10 (1.16–3.81) 4.54 (1.86–11.07) 1.88 (1.01–3.51) 3.49 (1.44–8.46)
Health risk factors
Body Mass Index (n = 609)
 Normal weight 1.00 1.00 1.00 1.00
 Overweight 2.06 (1.24–3.42) 1.01 (0.51–1.99) 1.88 (1.12–3.16) 0.79 (0.39–1.58)
 Obese 2.03 (1.09–3.78) 0.95 (0.39–2.35) 1.76 (0.94–3.28) 0.71 (0.29–1.70)
Smoking status (n = 622)
 Non-smoker 1.00 1.00 1.00 1.00
 Current Smoker 1.80 (1.1–2.93) 1.64 (0.82–3.27) 1.91 (1.16–3.14) 1.74 (0.83–3.66)
Alcohol Use (n = 622)
 Non-drinkers 1.00 1.00 1.00 1.00
 Moderate drinkers 1.22 (0.65–2.30) 1.08 (0.4–2.90) 1.12 (0.59–2.11) 0.90 (0.33–2.47)
 Hazardous alcohol use 1.31 (0.62–2.78) 1.71 (0.58–5.07) 1.43 (0.65–3.11) 1.93 (0.59–6.31)
Health-related quality of life
Self-rated health (n = 622)
 Excellent/very good 1.00 1.00 1.00 1.00
 Good 2.10 (1.28–3.44) 4.08 (1.84–9.09) 2.22 (1.33–3.68) 4.44 (1.99–9.91)
 Fair/Poor 4.03 (1.93–8.44) 16.59 (6.89–39.95) 4.34 (2.04–9.23) 18.70 (7.71–45.35)
Functional impairment due to physical health (n = 622)
 No 1.00 1.00 1.00 1.00
 Yes 2.22 (1.09–4.52) 5.66 (2.63–12.17) 2.20 (1.05–4.62) 5.61 (2.47–12.75)
Functional impairment due to emotional health (n = 622)
 No 1.00 1.00 1.00 1.00
 Yes 2.51 (1.39–4.52) 2.56 (1.11–5.86) 2.76 (1.50–5.08) 3.10 (1.25–7.69)
Somatic symptoms
Somatic symptom severity (n = 621)
 Low (0/4) 1.00 1.00 1.00 1.00
 Medium (5/9) 1.70 (1.01–2.87) 3.66 (1.90–7.03) 1.69 (1.00–2.87) 3.76 (1.85–7.63)
 High (≥10) 3.63 (1.74–7.60) 4.98 (1.87–13.28) 4.35 (2.03–9.32) 7.49 (2.61–21.46)
Open in a new tab

S1 and S2 represent the two SELCoH sample surveys at timepoints 1 and 2, respectively. Relative risk, RRR, Is weighted to account for survey design. “No reported LTC at either S1 or S2” represents the reference category in the multinomial regressions. All variables are measured at the S1 timepoint except ethnicity which was only measured at the S2 timepoint. CI = 95% confidence interval for the relative risk ratio. Bold indicates statistical significance. There were no participants in the underweight group in either the initiation or progression trajectory categories, so the underweight category is not shown in the table.

aAdjusted by age (continuous) and gender

b 0-to-≥ 1 LTC: S1 = 0 LTC, S2 ≥1 LTCs

c 1-to-MLTCs: S1 = 1 LTC, S2 ≥2 LTCs

Psychosocial factors increased the risk of both negative health transitions. Having the smallest social network (≤ 2 known contacts per week) and being in the group with the greatest number of stressful life events (≥ 6) at baseline, increased the risk of developing initial LTCs and also progression from one-to-many LTCs (Table 2). Health risk factors included being overweight and smoking which were both determinants of an increased risk of developing initial LTCs. Whilst there was an increased risk of developing initial LTCs in those who were obese (RRR 1.76, 95% CI 0.94–3.28), and also of progressing from one-to-many LTCs in those who were smokers (RRR 1.74, 95% CI 0.83–3.66), neither of these were statistically significant.

Baseline measures related to health-related quality of life (self-rated health and functional impairment due to both physical and emotional health) and somatic symptom severity, were all determinants of both LTC initiation and progression, with poor/fair self-rated health at baseline having the strongest association with progressing from one-to-many LTCs (RRR 18.70, 95% CI 7.71–45.35).

Socioeconomic determinants of negative health transitions included housing tenure, income and employment status (Table 2). After adjusting for age and gender, renting/part-renting a property was associated with an increased the risk of developing initial LTCs, and having a low income or not knowing the household income, were both associated with an increased risk of progression from one-to-many LTCs. Employment status at baseline increased the risk of progressing from one-to-many LTCs for those who were employed part-time, unemployed or having taken an early work exit, compared to those who were employed full-time.

LTCs and employment

The average number of LTCs at follow-up was greatest in the group that had made an early work exit (2.2 LTCs), with the full-time employed and student group reporting the lowest average number of LTCs (0.5 and 0.4 LTCs respectively). Compared to the full-time employee group, those who were employed part-time and the unemployed both had a higher average number of LTCs (0.6 and 0.8 respectively). Those who were at home also had an average of 0.6 LTCs. The distribution of the eight most prevalent LTCs within each employment category at follow-up (S2) is shown in Table 3.

Table 3. Frequency and prevalence distribution of LTCs within each employment category at follow-up.

Employed FT (n = 431) Employed PT (n = 161) Unemployed (n = 90) Early work exit (n = 78) Student (n = 82) At home (n = 51)
N n (%) n (%) n (%) n (%) n (%) n (%)
Asthma 82 34 (7.5) 13 (9.3) 9 (9.2) 19 (24.3) 5 (5.6) 2 (4.2)
Depression 79 16 (3.5) 11 (6.4) 12 (12.7) 29 (37.8) 9 (9.7) 2 (3.7)
Diabetes 30 8 (1.5) 5 (2.9) 5 (4.6) 9 (11.1) 1 (0.8) 2 (2.9)
Gastrointestinal conditions 38 13 (2.8) 4 (2.2) 7 (7.9) 7 (9.9) 7 (8.2) 0
Gynaecological conditions 26 9 (2.1) 5 (2.9) 2 (1.7) 4 (5.0) 4 (4.1) 2 (2.8)
High blood pressure 81 31 (6.2) 12 (6.2) 8 (7.2) 26 (33.1) 0 4 (6.3)
Migraine 34 14 (3.1) 5 (3.3) 6 (6.9) 7 (10.3) 1 (1.1) 1 (1.8)
Musculoskeletal conditions 106 37 (7.5) 20 (10.4) 8 (7.8)) 36 (47.1) 1 (1.6) 4 (6.7)
Open in a new tab

S2 represents data collected at the follow-up SELCoH interviews. Frequencies are unweighted. Percentages are weighted to account for survey design. Column percentages may add up to >100% for each employment category due to multimorbidity. N = 893 and includes all follow-up data.

Musculoskeletal conditions were the most prevalent condition in those who were employed full-time (7.5%), part-time (10.4%), the early work exit (41.7%) and at home (6.7%) groups, with asthma additionally being the most prevalent in the full-time employee group (7.5%). In contrast, depression was the most prevalent condition in those who were unemployed (12.7%) and students (9.7%). The early work exit group (which includes permanently sick, disabled and the early-retired), had the highest overall prevalence of LTCs. The four most prevalent conditions in this group were musculoskeletal conditions, depression, high blood pressure and asthma, which were also the most prevalent conditions in the employed full-time and part-time groups. Of those who were economically active (employed or unemployed but available for work), depression was reported by 12.7% of the unemployed group compared to 4% and 6% of the full-time and part-time employee groups respectively.

Discussion

This study provides evidence that the journey to multimorbidity may have both common and unique risk factors according to the stage of MLTC progression. At the early stage of developing initial LTCs, health behaviours appear to be important, with smoking, being overweight and renting being unique to this initial stage. In contrast, the progression from one to multiple LTCs was associated with socioeconomic factors including low household income and the three employment categories of being employed part-time, being unemployed or having made an early work exit. This could be explained by the fundamental social cause theory which states that those with access to more and better quality socioeconomic resources (such as, better employment circumstances and income) and beneficial social connections, will have a lower mortality risk due to increased capacity to manage their health [12]. Employers may also provide specific health resources such as health screening in workplaces and health insurance [8, 12]. Having an income as a result of being employed also facilitates participation in society (and thereby supporting social support networks), with the additional psychological benefits of providing a secure social context and a sense of identity [8]. In contrast, both unemployment and part-time employment, were linked to negative health transitions (specifically, progressing from one to multiple LTCs). Whilst unemployment is known to be associated with poorer physical and mental health and is a risk factor for negative health transitions [8, 22], the role of part-time employment in negative health transitions is less clear. One possibility is that there is a higher proportion of people with one LTC at baseline who found the need to change from full-time to part-time employment, and so this group may already be at increased risk. The link between unemployment and poor mental health was supported by our findings that whilst depression was the most prevalent LTC in the unemployed group, it was the fourth most prevalent condition in those who were employed full-time and the third most prevalent condition in those who were employed part-time. Although musculoskeletal conditions and asthma were the two most prevalent physical conditions in those who were employed both full- and part-time, they were also the most prevalent across the whole sample irrespective of employment status, which may reflect a generally high occurrence in working age adults in this inner-city cohort.

Whilst the risk of progression from good health to multimorbidity increased with age, this study showed that even in the younger age group (35 to 44 years), there was a greater (and statistically significant) risk of progressing from one to multiple LTCs, even when compared to developing a first LTC. This may be linked to socioeconomic factors that were also found to increase the risk of progression to multimorbidity. Previous evidence for multimorbidity in younger age groups has been shown to occur in areas of high socioeconomic deprivation where the prevalence of multimorbidity in young and middle-aged adults living in socioeconomically deprived areas was equivalent to those 10–15 years older living in more affluent areas [6]. Since the present study uses data from an inner-city population known to have high levels of socioeconomic deprivation overall [23, 24], the specific socioeconomic factors of low household income and not being employed full-time being linked to an earlier progression to multimorbidity, provides more detailed supporting evidence.

It is of interest that whilst there may be different risk factors linked to LTC initiation and then progression to multimorbidity, there were common factors such as small social network size, increased number of stressful life events, poor/fair self-rated health, increased functional impairments due to physical and mental health, and increased somatic symptom severity. Social networks are linked to social support, with the latter being defined as the subjective appraisal of being able to draw on the resources of a social network [25]. Insufficient social support has been linked to higher rates of physical and mental health conditions, and in the present study, if smaller social networks decreased the opportunity for social support, this may have increased the risk of developing LTCs [26]. Similarly, higher levels of perceived social support have been shown to have a positive impact on health-related quality of life in patients with both single and multiple LTCs [25, 27].

A strength of this study is its longitudinal design with the independent variables being measured at baseline (S1) indicating potential risk factors for the two negative health transition trajectories. Additionally, variables relating to health perceptions (self-rated health, functional impairment due to both physical and mental health and somatic symptom severity) were measured prior to the increased prevalence of LTCs measured at follow-up (S2) for each of the trajectories. If negative health perceptions at S1 are associated with the development of initial LTCs at S2, the onset of negative health perceptions may represent an early stage of the diagnostic journey between the onset of symptoms and a formal diagnosis of an initial LTC. Since this study showed an increased risk of developing an initial LTC in those who were overweight and smokers, a better understanding of the socioeconomic and behavioural profile of individuals with early self-reported negative health perceptions may assist in designing early targeted interventions.

This study is limited by the small sample size and would benefit from repeating in a larger population to confirm the results, and carry out comparisons in a wider setting including rural communities. In contrast to many studies using patient records, all self-reported conditions were included in this study. However, a larger sample size would have enabled a larger number of LTCs to be analysed separately rather than being combined into similar condition groups. Despite the advantages of including all self-reported conditions, it should also be noted that without access to medical records, we were unable to confirm whether a formal diagnosis had been received for all conditions, or if there were un-reported conditions that had not yet received a diagnosis or were being successfully managed by the individual. A larger sample size and the addition of medical record data would also enable factors supporting positive health transitions to be investigated. By expanding the scope of future studies to include positive health transitions, this would provide guidance for health promotion interventions aimed at slowing the journey to multimorbidity. It should also be acknowledged that whilst this study is longitudinal in design with two timepoints, the testing of baseline factors that increase the likelihood of the two negative LTC trajectories should be interpreted with caution, and can only imply causality rather than prove it. Also, reverse causality cannot be excluded, particularly in the case of employment where health status may lead to reduced working hours or unemployment.

In conclusion, this study demonstrates that the journey to multimorbidity is complex, and suggests there are both common and unique factors linked to the two stages of initiation and progression of LTCs. Further research focusing on developing health promotion interventions in younger, healthy participants, should aim to increase understanding of the role of precursors to LTC in the most prevalent functionally impairing LTCs that impact the ability to engage in employment. Whilst we have shown that not being employed full-time may have a negative influence on health, future research should consider the quality of employment (employment conditions and relationships) and job characteristics (the nature of the job). Investigation of a wide range of employment characteristics could elaborate on specific causal relationships between employment and the development of multiple long-term health conditions.

Data Availability

This study used third party data from the SELCoH survey. SELCoH data is available to researchers upon request. Please contact selcoh@kcl.ac.uk to apply for access. The authors confirm they did not receive any special privileges in accessing the data that other researchers would not have. For more information visit www.kcl.ac.uk/research/selcoh.

Funding Statement

This study was funded by Guy's and St Thomas' Charity, EIC180702: IM and SAMS. SAMS is supported by the National Institute for Health Research (NIHR) Maudsley Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and the National Institute for Health Research, NIHR Advanced Fellowship, Dr Sharon A M Stevelink, NIHR300592. SH is supported by the ESRC Centre for Society and Mental Health at King’s College London (ESRC Reference: ES/S012567/1) and the NIHR Biomedical Research Centre at South London and Maudsley NHS Foundation Trust. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Whitty CJM, MacEwen C, Goddard A, Alderson D, Marshall M. Calderwood C et al. Rising to the challenge of multimorbidity. British Medical Journal, 2020. 368:l6964. doi: 10.1136/bmj.l6964 . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Aiden H. Multimorbidity: Understanding the Challenge. A Report for the Richmond Group of Charities. 2018. https://richmondgroupofcharities.org.uk/sites/default/files/multimorbidity_-_understanding_the_challenge.pdf.
  • 3.Canizares M, Hogg-Johnson S, Gignac MAM, Glazier RH, Badley EM. Increasing Trajectories of Multimorbidity Over Time: Birth Cohort Differences and the Role of Changes in Obesity and Income. J Gerontol B Psychol Sci Soc Sci. 2018; 73(7):1303–1314. doi: 10.1093/geronb/gbx004 . [DOI] [PubMed] [Google Scholar]
  • 4.Stafford M, Thorlby R, Fisher R. Understanding the health care needs of people with multiple health conditions. 2018: Health Foundation London. https://www.health.org.uk/publications/understanding-the-health-care-needs-of-people-with-multiple-health-conditions.
  • 5.Ashworth M, Durbaba S, Whitney D, Crompton J, Wright M, Dodhia H. Journey to multimorbidity: longitudinal analysis exploring cardiovascular risk factors and sociodemographic determinants in an urban setting. BMJ Open, 2019. 9(12):e031649. doi: 10.1136/bmjopen-2019-031649 . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.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: 37–43. doi: 10.1016/S0140-6736(12)60240-2 . [DOI] [PubMed] [Google Scholar]
  • 7.Nathwani T, Birkin G, Cheng Z, Brazier JE, Roberts J, Rice NE, et al. Employment outcomes for people with long-term conditions: a rapid evidence assessment. 2015, CFE Research: Leicester. http://cfe.org.uk/app/uploads/2016/09/Emp-outcomes-for-people-with-LT-conditions_REA.pdf.
  • 8.Waddell G. Burton AK. Is work good for your health and well-being? London: The Stationary Office, 2006. https://huddersfield.box.com/s/kmyygl6clmsap4m3femp6wb7joaxd87d.
  • 9.Bambra C. Work, worklessness and the political economy of health inequalities. J Epidemiol Community Health. 2011;65(9):746–50. doi: 10.1136/jech.2009.102103 Epub 2011 Jan 30. . [DOI] [PubMed] [Google Scholar]
  • 10.Stagg AL, Hatch S, Fear NT, Dorrington S, Madan I, Stevelink SAM. Long-term health conditions in UK working-age adults: a cross-sectional analysis of associations with demographic, socioeconomic, psychosocial and health-related factors in an inner-city population. BMJ Open. 2022;12:e062115. doi: 10.1136/bmjopen-2022-062115 . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Wikström K, Lindström J, Harald K, Peltonen M, Laatikainen T. Clinical and lifestyle-related risk factors for incident multimorbidity: 10-year follow-up of Finnish population-based cohorts 1982–2012. Eur J Intern Med. 2015;26(3):211–6. doi: 10.1016/j.ejim.2015.02.012 Epub 2015 Mar 3. . [DOI] [PubMed] [Google Scholar]
  • 12.Phelan JC, Link BG, Tehranifar P. Social conditions as fundamental causes of health inequalities: theory, evidence, and policy implications. J Health Soc Behav. 2010;51 Suppl:S28–40. doi: 10.1177/0022146510383498 . [DOI] [PubMed] [Google Scholar]
  • 13.Hatch SL, Frissa S, Verdecchia M, Stewart R, Fear NT, Reichenberg A, et al. Identifying socio-demographic and socioeconomic determinants of health inequalities in a diverse London community: the South East London Community Health (SELCoH) study. BMC Public Health. 2011;11:861. doi: 10.1186/1471-2458-11-861 . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Hatch SL, Gazard B, Williams DR, Frissa S, Goodwin L; SELCoH Study Team, et al. Discrimination and common mental disorder among migrant and ethnic groups: findings from a South East London Community sample. Soc Psychiatry Psychiatr Epidemiol. 2016; 51(5):689–701. doi: 10.1007/s00127-016-1191-x Epub 2016 Feb 13. . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Harber-Aschan L, Hotopf M, Brown JSL, Henderson M, Hatch SL. Longitudinal patterns of mental health service utilisation by those with mental-physical comorbidity in the community. J Psychosom Res. 2019;117:10–19. doi: 10.1016/j.jpsychores.2018.11.005 Epub 2018 Nov 27. . [DOI] [PubMed] [Google Scholar]
  • 16.Meyer IH. Prejudice as stress: conceptual and measurement problems. Am J Public Health. 2003;93(2):262–5. doi: 10.2105/ajph.93.2.262 . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Turner RJ, Lloyd DA. Lifetime traumas and mental health: the significance of cumulative adversity. J Health Soc Behav. 1995;36(4):360–76. . [PubMed] [Google Scholar]
  • 18.Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption—II. Addiction. 1993;88(6):791–804. doi: 10.1111/j.1360-0443.1993.tb02093.x . [DOI] [PubMed] [Google Scholar]
  • 19.Gazard B, Frissa S, Nellums L, Hotopf M, Hatch SL. Challenges in researching migration status, health and health service use: an intersectional analysis of a South London community. Ethn Health. 2015;20(6):564–93. doi: 10.1080/13557858.2014.961410 Epub 2014 Oct 1. . [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Ware J Jr, Kosinski M, Keller SD. A 12-Item Short-Form Health Survey: construction of scales and preliminary tests of reliability and validity. Med Care. 1996;34(3):220–33. doi: 10.1097/00005650-199603000-00003 . [DOI] [PubMed] [Google Scholar]
  • 21.Kroenke K, Spitzer RL, Williams JB. The PHQ-15: validity of a new measure for evaluating the severity of somatic symptoms. Psychosom Med. 2002;64(2):258–66. doi: 10.1097/00006842-200203000-00008 . [DOI] [PubMed] [Google Scholar]
  • 22.Salis S, Speckesser S, Samek L, Bivand P. A study of work and health transitions. Analysis of Understanding Society, DWP. 2021. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/1003424/a-study-of-work-and-health-transitions-analysis-of-understanding-society.pdf.
  • 23.Lambeth Council. Lambeth Demographic Factsheet. May 2014: London. 2014. https://www.lambeth.gov.uk/sites/default/files/ssh-Lambeth-Demography-2014.pdf.
  • 24.Indices of Deprivation 2019. Southwark’s Joint Strategic Needs Assessment. Southwark Council: London. 2019. https://moderngov.southwark.gov.uk/documents/s85676/Document%20Indices%20of%20Deprivation%202019%20-%20JSNA%20Factsheet.pdf.
  • 25.Vogel I, Miksch A, Goetz K, Ose D, Szecsenyi J, Freund T. The impact of perceived social support and sense of coherence on health-related quality of life in multimorbid primary care patients. Chronic Illn. 2012;8(4):296–307. doi: 10.1177/1742395312445935 Epub 2012 Apr 19. . [DOI] [PubMed] [Google Scholar]
  • 26.Cadzow RB, Servoss TJ. The association between perceived social support and health among patients at a free urban clinic. J Natl Med Assoc. 2009; 101(3):243–50. doi: 10.1016/s0027-9684(15)30852-x . [DOI] [PubMed] [Google Scholar]
  • 27.Göz F, Karaoz S, Goz M, Ekiz S, Cetin I. Effects of the diabetic patients’ perceived social support on their quality-of-life. J Clin Nurs. 2007. Jul;16(7):1353–60. doi: 10.1111/j.1365-2702.2007.01472.x . [DOI] [PubMed] [Google Scholar]

Decision Letter 0

Andrea Dell'Isola

4 Apr 2023

PONE-D-22-33742Risk factors for the progression to multimorbidity among UK urban working-age adults. A community cohort studyPLOS ONE

Dear Dr. Stagg,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

 In particular, the analysis requires extensive and careful revision as pointed out by reviewer 1. Please ensure that your decision is justified on PLOS ONE’s publication criteria and not, for example, on novelty or perceived impact.

For Lab, Study and Registered Report Protocols: These article types are not expected to include results but may include pilot data. 

==============================

Please submit your revised manuscript by May 19 2023 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Andrea Dell'Isola

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Competing Interests section:

"This paper represents independent research part funded by NIHR Maudsley Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London, and the NIHR. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care."

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to  PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests).  If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

3. We noted in your submission details that a portion of your manuscript may have been presented or published elsewhere. "The supporting information table listing the prevalence of health conditions at both timepoints has been published in a preliminary study which has recently been published in BMJ Open.

However, it is also key information for this paper in relation to the number of individual long-term conditions at baseline and at follow-up as it was used to create the long-term condition variables showing acquisition of a first LTC between the two timepoints, and also development of multimorbidity over the same timeframe. Since it has been published before, it was included only as a supplementary table so readers could easily refer to it." Please clarify whether this [conference proceeding or publication] was peer-reviewed and formally published. If this work was previously peer-reviewed and published, in the cover letter please provide the reason that this work does not constitute dual publication and should be included in the current manuscript.

4. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This is an interesting and well-written paper. Its strengths include the longitudinal data and the data on employment - which is particularly valuable in the context of multimorbidity among working-age people. Specific comments below.

1. The model description is not 100% clear. It would be good to know the parameterisation and any assumptions. I also wonder if it would be easier to run two separate logistic regression models. One for any LTC and then another, conditional on having a LTC what is the odds of developing further LTCs. This would make it difficult to model improvement, but the trade off is it would be simpler.

2. There seem to be a lot of parameters for the relatively small number of events (I counted 39 even though there were only 46 people who transitioned from 1 to many LTCs). I know that the rule of thumb 10 events per variable is an overstatement, but 39 still seems excessive. Moreover, age could be modelled using splines or fractional polynomials to reduce the number of parameters that need to be fitted.

3. Related to this, if I understand correctly these coefficients are all mutually adjusted (even for the unadjusted model). This leads to the table 2 fallacy (https://pubmed.ncbi.nlm.nih.gov/23371353/). Instead it would be better to draw some DAGS and decide on which covariates should go into models asking questions which are relevant for clinical practice or for determining policy.

4. Notwithstanding the model fitting some post-processing to obtain absolute risks of transitioning to each state for people with different covariate values would be useful. The odds ratio scale is fine for relative effects, but less good for describing patterns of disease and risks, which is the purpose here.

5. It would be useful to discuss why people were not followed-up in the second round. Could this be due to ill health preventing them from completing the wave? It would be good to report the number and characteristics of people who were invited for a second wave but did not participate.

6. In the discussion the statement is made that the odds ratios are different for 0/1 versus 1/many LTCs. If this is to be asserted the difference in the odds ratios should be estimated. Again, however, some thought needs to be given to scale. Some characteristics may increase risk MORE or LESS on the absolute scale while being the same on the relative scale (and vice versa).

7. Would be good to see some numbers in the abstract results section.

8. The introduction states that having a single LTC occurs before multiple LTCs - isn't this almost true by definition (at least for sufficiently granular and complete data)?

9. nice to see the number of events in table 1. Would be good to see this at the top of Table 2 also.

10. Not sure of the usefulness of the unadjusted coefficients in table 2

11. Would be good to share code and (if possible) aggregate level data (I know IPD cannot be shared).

Reviewer #2: The manuscript “Risk factors for the progression to multimorbidity among UK urban working-age adults. A community cohort study” explores factors associated with the initiation and progression of long-term chronic conditions among the working population in the UK. This is a very nice paper with an important contribution to the literature. The manuscript is well written and organized. The methodology and analytical methods used are appropriate for the study goals and results are presented with clarity. I only have minor questions/clarifications as follows.

• Page 6, line 116-124: The analytical sample included participants who completed both surveys. This is understandable given the objectives. However, I wonder if not responding at wave 2 may partially related to chronic conditions or some important participant characteristics. Could the authors comment on the differences in # of conditions and characteristics of those excluded due to loss to follow up? I think a brief comment in the discussion/limitations should suffice.

• Page 10, lines 204-210: It is not clear how adjusted models were fit. For example, for BMI, the adjusted model included BMI, age, sex only or the RRR is from a model with all variables? Please clarify. If the latter, please comment on the possible impact of collinearity to the results.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: David A McAllister

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (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. Registration is free. 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 PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 Sep 8;18(9):e0291295. doi: 10.1371/journal.pone.0291295.r002

Author response to Decision Letter 0

1 Jun 2023

1 Editor comments / Journal requirements

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Response from authors:

We have checked PLOS ONE’s style requirements are applied throughout the document and one adjustment has been made:

1) Table 2 legend has been altered so it is in one section with the footnotes following the legend.

2) There is no longer a supplementary table, so the re-naming of that table is no longer required.

3) Please could you ensure that Prof Ira Madan and Dr Sharon Stevelink are joint last/senior authors. Thank you.

2 Editor comments / Journal requirements

Thank you for stating the following in the Competing Interests section:

"This paper represents independent research part funded by NIHR Maudsley Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London, and the NIHR. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care."

Please confirm that this does not alter your adherence to all PLOS ONE policies on sharing data and materials, by including the following statement: "This does not alter our adherence to PLOS ONE policies on sharing data and materials.” (as detailed online in our guide for authors http://journals.plos.org/plosone/s/competing-interests). If there are restrictions on sharing of data and/or materials, please state these. Please note that we cannot proceed with consideration of your article until this information has been declared.

Please include your updated Competing Interests statement in your cover letter; we will change the online submission form on your behalf.

Response from authors:

Please could you update our Competing Interests statement with the following statement which includes the reference to PLOS ONE policies on sharing data and materials:

“This paper represents independent research part funded by NIHR Maudsley Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London, and the NIHR. The views expressed are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health and Social Care. This does not alter our adherence to PLOS ONE policies on sharing data and materials.”

3 Editor comments / Journal requirements

We noted in your submission details that a portion of your manuscript may have been presented or published elsewhere. "The supporting information table listing the prevalence of health conditions at both timepoints has been published in a preliminary study which has recently been published in BMJ Open.

However, it is also key information for this paper in relation to the number of individual long-term conditions at baseline and at follow-up as it was used to create the long-term condition variables showing acquisition of a first LTC between the two timepoints, and also development of multimorbidity over the same timeframe. Since it has been published before, it was included only as a supplementary table so readers could easily refer to it." Please clarify whether this [conference proceeding or publication] was peer-reviewed and formally published. If this work was previously peer-reviewed and published, in the cover letter please provide the reason that this work does not constitute dual publication and should be included in the current manuscript.

Response from authors:

We can confirm that the supplementary Table S1 has been published in the following peer-reviewed preliminary study: Stagg AL, Hatch S, Fear NT, et al. Long-term Health conditions in UK working-age adults: a cross-sectional analysis of associations with demographic, socioeconomic, psychosocial and health-related factors in an inner-city population. BMJ Open 2022;12:e062115. doi:10.1136/bmjopen-2022-062115

The reason this table was included as a supplementary table was as previously stated, for ease of access. However, this has now been amended so that reference is now made to the initial publication where readers can access the full information about the numbers of conditions at both timepoints (see Revised Manuscript with Track Changes, page 7, lines 140 and 141). The original supplementary file is therefore no longer necessary. The supporting information section on page 27, lines 506-517 has also been removed.

4 Editor comments / Journal requirements

In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

Response from authors on Data Availability:

This study used third party data from the SELCoH survey. The following data availability statement should be included:

“SELCoH data is available to researchers upon request. Please contact selcoh@kcl.ac.uk to apply for access. For more information visit www.kcl.ac.uk/research/selcoh”

Further Editor comments / Journal requirements (Email: 26 05 23)

1 and 2 Editor comments / Journal requirements

We note that your uploaded Cover Letter is not a Cover Letter but a Response to Reviewers file. It is important that you include a cover letter with your manuscript. Please ensure that this letter is addressed specifically to PLoS ONE. Please also include

* why this manuscript is suitable for publication in PLoS ONE.

* how does your paper provide a worthwhile addition to the scientific literature?

* how does your paper relate to previously published work?

* which types of scientists do you believe will be most interested in your study?

Please also note that the Cover Letter and Response to Reviewers file should be uploaded as two separate files.

2. Please upload a Response to Reviewers letter which should include a point by point response to each of the points made by the Editor and / or Reviewers. (This should be uploaded as a 'Response to Reviewers' file type.) Please follow this link for more information: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fblogs.plos.org%2Feveryone%2F2011%2F05%2F10%2Fhow-to-submit-your-revised-manuscript%2F&data=05%7C01%7Canne.l.stagg%40kcl.ac.uk%7C0cb97bc2fa5d4fcb1fb908db5d96cff3%7C8370cf1416f34c16b83c724071654356%7C0%7C0%7C638206674534014930%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=upX54CLXyfauygwUSl3vmgiwyj368IvaEQrw5KAAuCY%3D&reserved=0

Response from authors:

The cover letter and Response to Reviewers letters have now been uploaded as two separate files.

3 Editor comments / Journal requirements

Your ethics statement should only appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please ensure that your ethics statement is included in your manuscript, as the ethics statement entered into the online submission form will not be published alongside your manuscript.

Response from authors:

The ethics statement has now been moved within the Methods from the Statistical analyses section to the Study design, setting and participants section (see Revised Manuscript with Track Changes page 6 lines 125-128). Thank you for pointing this out as this is a more appropriate section for the ethics statement which is linked to the SELCoH data collection.

4 Editor comments / Journal requirements

We note your current Data Availability Statement: "This study used third party data from the SELCoH survey. The following data availability statement should be included:

SELCoH data is available to researchers upon request. Please contact selcoh@kcl.ac.uk to apply for access. For more information visit https://www.kcl.ac.uk/research/selcoh

Regarding third party data, please confirm whether or not the authors received any special privileges in accessing the data that other researchers would not have.

Response from authors:

We can confirm that the authors did not receive any special privileges in accessing the data that other researchers would not have.

1 Reviewer 1 Comments to the Author

The model description is not 100% clear. It would be good to know the parameterisation and any assumptions. I also wonder if it would be easier to run two separate logistic regression models. One for any LTC and then another, conditional on having a LTC what is the odds of developing further LTCs. This would make it difficult to model improvement, but the trade off is it would be simpler

Response from authors:

We agree with the reviewer model description should be made clearer, and a similar comment on clarity was also raised by the second reviewer. To address this, the model description in the transcript has been revised in the statistical analyses section to make it clearer (see document ‘Revised Manuscript with Track Changes’, page 10 lines 209-217) and page 14 lines 253-254. In explanation to the reviewers, we used multinomial logistic regressions and with the dependent outcome variable consisting of the following categories: (1) the odds of developing any number of LTCs at the second timepoint having not had any LTCs at baseline, and (2) when there is one LTC at baseline, the odds of developing further LTCs by the second timepoint. (3) No reported LTCs at either S1 or S2 was used as the reference category in the regressions. The outcome is therefore similar to the two models the reviewer suggests but has the advantage of including the 479 individuals in the healthy group who had no LTCs at either timepoint (the reference group), is that it increases the strength of the analysis compared to running two separate regression models where we would lose data from these 479 individuals. Since we note that the second reviewer was satisfied with the statistics used, we have not changed our modelling, but are happy to reconsider if the editor believes that re-modelling the analysis is essential.

2 Reviewer 1 Comments to the Author

There seem to be a lot of parameters for the relatively small number of events (I counted 39 even though there were only 46 people who transitioned from 1 to many LTCs). I know that the rule of thumb 10 events per variable is an overstatement, but 39 still seems excessive. Moreover, age could be modelled using splines or fractional polynomials to reduce the number of parameters that need to be fitted.

Response from authors:

This comment regarding the balance between the number of parameters for the number of participants appears to be linked to the previous comment about lack of clarity over the model description as it would be highly relevant where all the predictor variables were tested simultaneously within in the same model. However, we did not carry out a multiple regression, and each predictor variable was tested independently for the multinomial LTC outcomes, apart from the adjusted model which included age and gender, so a maximum of 3 predictor variables in the adjusted model.

3 Reviewer 1 Comments to the Author

Related to this, if I understand correctly these coefficients are all mutually adjusted (even for the unadjusted model). This leads to the table 2 fallacy (https://pubmed.ncbi.nlm.nih.gov/23371353/). Instead it would be better to draw some DAGS and decide on which covariates should go into models asking questions which are relevant for clinical practice or for determining policy.

Response from authors:

This comment has been addressed in the response to this reviewer’s the first comment as it relates to the model description, which we have now clarified and explained that the coefficients are not all mutually adjusted.

4 Reviewer 1 Comments to the Author

Notwithstanding the model fitting some post-processing to obtain absolute risks of transitioning to each state for people with different covariate values would be useful. The odds ratio scale is fine for relative effects, but less good for describing patterns of disease and risks, which is the purpose here.

Response from authors:

Again, we think that this comment is linked to the lack of clarity about the regression modelling we used. We hope that this has now been addressed with the revisions that we have made.

5 Reviewer 1 Comments to the Author

It would be useful to discuss why people were not followed-up in the second round. Could this be due to ill health preventing them from completing the wave? It would be good to report the number and characteristics of people who were invited for a second wave but did not participate.

Response from authors:

The second reviewer has also raised a similar comment about reasons for lack of response at follow-up particularly in relation to ill health, which we have addressed below in response to their comment.

6 Reviewer 1 Comments to the Author

In the discussion the statement is made that the odds ratios are different for 0/1 versus 1/many LTCs. If this is to be asserted the difference in the odds ratios should be estimated. Again, however, some thought needs to be given to scale. Some characteristics may increase risk MORE or LESS on the absolute scale while being the same on the relative scale (and vice versa).

Response from authors:

We were a little uncertain as to the interpretation of this comment, but assume that it is in reference to the following sentence (page 19, lines 333-336): “Whilst the risk of progression from good health to multimorbidity increased with age, this study showed that even in the younger age group (35 to 44 years), there was a greater risk of progressing from one to multiple LTCs, even when compared to developing a first LTC.”

This sentence primarily relates to the greater risk of progressing from one to multiple LTCs at the younger age (35-44 years) group which was shown to be statistically significant, whilst the risk of the initial development from none to one-or-more LTCs in the same age group was not statistically significant in both the unadjusted and adjusted (by gender) models. In addition, the relative risk ratios increase as age increases and this indicates an increased risk of developing LTCs with age. It may be that the reviewer was again considering the model to be a multiple logistic regression, which has now been clarified to not be the case.

We have added in a comment on page 19 line 334-335 that this increased risk is statistically significant. However, if this is not the correct interpretation of the reviewer’s comment, please could you provide more clarification so that we can make the appropriate revision.

7 Reviewer 1 Comments to the Author

Would be good to see some numbers in the abstract results section.

Response from authors:

The authors would have like to have a included some numbers in the abstract section, but the word count limitation prevented this.

8 Reviewer 1 Comments to the Author

The introduction states that having a single LTC occurs before multiple LTCs - isn't this almost true by definition (at least for sufficiently granular and complete data)?

Response from authors:

The authors agree with the comment that the development of multiple LTCs from 1 LTC is a common fact. The reason it is included in the introduction is because the paper is looking at risk factors for progressing from one to MLTCs, so this comment was made as the initial part of the argument justifying this study which identifies risk factors linked to this progression.

9 Reviewer 1 Comments to the Author

nice to see the number of events in table 1. Would be good to see this at the top of Table 2 also.

Response from authors:

If the numbers for Table 2 were put at the top of the table, it would not show the reference group of 479 participants, so we have inserted the numbers against each of the indicator variables (pages 14-16). This also then indicates where the small number of missing records lie (see also the comment about missing data on page 9, line 201-202 in the document ‘Revised Manuscript with Track Changes’).

10 Reviewer 1 Comments to the Author

Not sure of the usefulness of the unadjusted coefficients in table 2

Response from authors:

This comment is indeed understandable if all the coefficients had been tested together, as they would not be unadjusted results. However, having revised the script to provide a clearer explanation of the modelling used, we hope that now the inclusion of the unadjusted values in this table makes sense since all the predictor variables were tested individually. The adjusted value columns then show results where each predictor variable was adjusted by age and gender so that a direct comparison can be made between the unadjusted and adjusted relative risk ratios.

11 Reviewer 1 Comments to the Author

Would be good to share code and (if possible) aggregate level data (I know IPD cannot be shared).

Response from authors:

As acknowledged by the reviewer, individual participant data cannot be directly shared. However, access to the SELCoH data is available and details have now been provided (see the updated data availability statement).

Whilst the entire coding process covers data cleaning and another preliminary study, so not all relevant for this paper, useful examples of the code used for the unadjusted and adjusted regressions (using marital status as an example) are:

svy:mlogit ltc_change_3gps i.relationship_s1,rrr

svy:mlogit ltc_change_3gps i.relationship_s1 i.sex age_s1cont ,rrr

This code should also help to clarify for the reviewers the modelling we used for the regressions.

1 Reviewer 2 Comments to the Author

Page 6, line 116-124: The analytical sample included participants who completed both surveys. This is understandable given the objectives. However, I wonder if not responding at wave 2 may partially related to chronic conditions or some important participant characteristics. Could the authors comment on the differences in # of conditions and characteristics of those excluded due to loss to follow up? I think a brief comment in the discussion/limitations should suffice.

Response from authors:

This is an interesting comment, and whilst the cleaned data set used in this study did not include this group, the authors have investigated the SELCoH data for health-related information about those who did not respond at follow-up. 21 participants were ineligible for the second wave due to death or poor health. However, the data used for the current study was restricted to participants under the age of 64 at follow-up as it focused on working-age participants, and none of these 21 participants were in this age group. Consequently, we would be unable to comment further on this as an actual limitation within our study.

2 Reviewer 2 Comments to the Author

Page 10, lines 204-210: It is not clear how adjusted models were fit. For example, for BMI, the adjusted model included BMI, age, sex only or the RRR is from a model with all variables? Please clarify. If the latter, please comment on the possible impact of collinearity to the results.

Response from authors:

The comment by this reviewer highlighting the need for clarity regarding the modelling has been addressed in the response to the first reviewer who also raised this (please see the response to reviewer 1, comment 1). The description in the methods section has now been altered (see Revised Manuscript with Track Changes page 10 lines 209-217). In summary, the multinomial LTC trajectory variable was regressed on each indicator variable individually. Two models were tested 1) these regressions were carried out unadjusted and 2) then adjusted by gender and age.

Attachment

Submitted filename: Response to reviewers_01.docx

Click here for additional data file. (120.1KB, docx)

Decision Letter 1

Andrea Dell'Isola

4 Jul 2023

PONE-D-22-33742R1Risk factors for the progression to multimorbidity among UK urban working-age adults. A community cohort studyPLOS ONE

Dear Dr. Stagg,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.  Reviewer 2 was satisfied with your response while reviewer one was not available to review the manuscript. This prompted us to review your manuscript again. This further review highlighted some issues that were not pointed out during the previous round of review.

Please submit your revised manuscript by Aug 18 2023 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Andrea Dell'Isola

Academic Editor

PLOS ONE

Additional Editor Comments:

General: Please accommodate the request from Reviewer One to provide estimates in the abstract. Currently, the abstract contains excess information that could be condensed to make space for these required amendments.

We also ask that you revise your manuscript to refrain from using causal language (e.g., "determine"), as the present study design does not permit assertions of causality.

Introduction: We advise that you abbreviate the introduction to keep its focus consistent with the aim of the study. Given that the study's objective is exploratory, it appears excessive space is allocated to specific factors, such as occupation. Please shorten the introduction accordingly and realign it with the study's aims.

Methods: It is necessary to provide a clearer account of those individuals who did not participate in the second wave of the study and were subsequently excluded. We recommend including a flowchart as a supplementary file, outlining the participant trajectory. Moreover, we ask that you supply a table comparing the baseline characteristics of participants who met the inclusion criteria but did not participate in the follow-ups.

We noted that you group certain conditions together that are treated as a single condition in your analysis, as per Stagg et al. (2022). In this regard, please provide a rationale for the validity of classifying subjects as having a single condition when they may have multiple LTCs. Addressing this issue is crucial, as it may necessitate significant alterations in your analysis.

Your exclusion of individuals whose health status improves may introduce significant selection bias. The current emphasis on studying individuals with a deteriorating health status does not sufficiently justify your choice of analysis. We strongly suggest revising your analysis to include all participants.

Discussion: Please refine the structure of your discussion for improved readability. For instance, the correlation between employment categories is discussed at two different points (lines 318 and 338). Consolidating this discussion into a single paragraph would greatly enhance readability. Also, we recommend limiting speculative conclusions, as your study merely observes associations (e.g., employment can both be a potential causal factor for multimorbidity, or employment could reflect health status at baseline which is the real risk factor for multimorbidity).

Alo, please reconsider carefully your limitation statement. Consider for example the limited number of LTCs, the grouping of LTCs, the lack of information on propensity to seek care (and thus receive a diagnosis).

We appreciate your attention to these details and look forward to receiving your revised manuscript.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: I have no further comments/suggestions. The authors have addressed all my questions and comments.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (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. Registration is free. 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 PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2023 Sep 8;18(9):e0291295. doi: 10.1371/journal.pone.0291295.r004

Author response to Decision Letter 1

17 Aug 2023

Response to the additional comments from the editors

Note: All references to the location of revisions made to the manuscript relate to the tracked changes version.

1. Editor comment:

General: Please accommodate the request from Reviewer One to provide estimates in the abstract. Currently, the abstract contains excess information that could be condensed to make space for these required amendments.

Response from authors:

The abstract has been condensed and relative risk ratios (RRRs) with 95% confidence intervals have been provided for all the significant unique risk factors for both initiation and progression pathways (Abstract pages 3 and 4, lines 49-73). The abstract is still within the maximum 300 word count requirements.

2. Editor comment:

We also ask that you revise your manuscript to refrain from using causal language (e.g., "determine"), as the present study design does not permit assertions of causality.

Response from authors:

The use of ‘determinants’ is related to the longitudinal design of the study which has two timepoints with the baseline variables being tested against two trajectories of risk of developing LTCs (This was acknowledged by Reviewer 1 in their comments to the authors who cited the longitudinal data as a strength of the study). However, it is acknowledged that there is an over-use of ‘determinants’ in the paper and a lack of caution in interpreting the results, so the following revisions have been made:

Revisions so that there is the appropriate use of the term ‘variables’ rather than ‘determinants’:

• Abstract, page 3 line 53: ‘determinants’ changed to ‘variables’

• Methods, page 7 line 149: The heading has been revised from ‘Determinant variables’ to ‘Independent variables’.

• Methods page 8 line 175: ‘determinants’ changed to ‘variables’

• Table 1 pages 11-14: ‘determinants’ changed to ‘variables’ and Table 2 (pages 14-17) has been changed to match the terminology used in Table 1.

• Discussion: Page 22 line 397: ‘determinant variables’ has been amended to ‘independent variables’

• Discussion: Page 22, line 399 ‘determinants’ has been changed to ‘variables’.

Additional changes:

• Abstract page 4 line 70: ‘determinants’ has been changed to 'risk factors’.

• Methods page 14 line 256: ‘Determinants’ has been changed to ‘Risk factors’

• Discussion page 19 lines 321-322: The following sentence has been changed from ‘This study indicates that the journey to multimorbidity has both common and unique determinants ..’ to ‘This study provides evidence that the journey to multimorbidity may have both common and unique risk factors..’

• Discussion page 22 lines 385-387: The following sentence has been changed from ‘Whilst there may be different determinants of LTC initiation and then progression to multimorbidity, it is of interest that there were common determinants of small social support networks …’ to ‘It is of interest that whilst there may be different risk factors linked to LTC initiation and then progression to multimorbidity, there were common factors such as small social support networks...’

• Discussion page 22 lines 402-403: The following sentence has been changed from ‘If negative health perceptions at S1 are determinants of developing initial LTCs at S2,..’ to ‘If negative health perceptions at S1 are associated with the development of initial LTCs at S2,..’

• Discussion page 23 lines 418-419: ‘determinants of’ has been changed to ‘factors supporting’.

• Discussion page 23 line 428: ‘determinants of’ has been changed to ‘factors linked to’.

Revisions highlighting caution in interpreting the results:

• The insertion of the following two sentences in the discussion, page 23, lines 421-426: ’It should also be acknowledged that whilst this study is longitudinal in design with two timepoints, the testing of baseline determinants of the two negative LTC trajectories should be interpreted with caution, and can only imply causality rather than prove it. Also, reverse causality cannot be excluded, particularly in the case of employment where health status may lead to reduced working hours or unemployment.’

• Discussion page 23 lines 427-428: this sentence has been amended so the conclusions are not stated so strongly. It has been changed from: ‘In conclusion, this study demonstrates that the journey to multimorbidity is complex, with both common and unique determinants of the two stages of …’ to ‘In conclusion, this study demonstrates that the journey to multimorbidity is complex, and suggests there are both common and unique factors linked to the two stages of …’.

• Discussion page 24, lines 432-33: ‘Whilst we have shown that not being employed full-time can have a negative influence on health, ..’ has been changed to ‘Whilst we have shown that not being employed full-time may have a negative influence on health,..’.

INTRODUCTION

3. Editor comment:

We advise that you abbreviate the introduction to keep its focus consistent with the aim of the study. Given that the study's objective is exploratory, it appears excessive space is allocated to specific factors, such as occupation. Please shorten the introduction accordingly and realign it with the study's aims.

Response from authors:

This manuscript is the outcome of a project supported by a research grant that was very specific in its objectives - to explore risk factors for developing LTCs and progression to multimorbidity in a working-age population. Employment status was specifically included within a wide range of exploratory variables with the additional objective for employment status to receive specific focus as a secondary aim. The current introduction reflects these objectives, and we cannot see how any further amendments would benefit this section. We would like to highlight the comment made by the first reviewer in relation to the usefulness of including employment data (“This is an interesting and well-written paper. Its strengths include the longitudinal data and the data on employment - which is particularly valuable in the context of multimorbidity among working-age people”).

METHODS

4. Editor comment:

It is necessary to provide a clearer account of those individuals who did not participate in the second wave of the study and were subsequently excluded. We recommend including a flowchart as a supplementary file, outlining the participant trajectory. Moreover, we ask that you supply a table comparing the baseline characteristics of participants who met the inclusion criteria but did not participate in the follow-ups.

Response from authors:

The data set used for this secondary data analysis study only included participants who had completed both the SELCoH 1 and SELCoH 2 surveys and were within a working-age range. Analysis of the SELCoH data not included in our dataset was outside the aims and objectives of this study. However, there have been many previous SELCoH studies published with information on the original data, and the participant trajectory between SELCoH 1 and SELCoH 2 has been previously reported in studies using the complete SELCoH dataset (For example, Hatch et al., 2016; doi:10.1007/s00127-016-1191-x, and Harber-Aschan et al., 2019; doi: 10.1016/j.jpsychores.2018.11.005). Hatch et al., (2016) also has a supplementary table comparing baseline demographic characteristics of the SELCoH 1 and SELCoH 2 participants. We have now inserted the following sentence into the methods so that these papers are now referred to (the references have also been adjusted accordingly):

Methods page 6, lines 125-127:

Demographic characteristics of participants who responded at both waves, and reasons for non-response at the second wave have previously been reported [14, 15].

Please also see our previous response to the second reviewer in regard to participants who did not complete SELCoH 2, as we were able to establish that the 21 non-responders for health reasons or death (referred to in Harber-Aschan et al., 2019) were outside the age range of our study so would not have been included in any analysis of non-responders

5. Editor comment:

We noted that you group certain conditions together that are treated as a single condition in your analysis, as per Stagg et al. (2022). In this regard, please provide a rationale for the validity of classifying subjects as having a single condition when they may have multiple LTCs. Addressing this issue is crucial, as it may necessitate significant alterations in your analysis.

Response from authors:

This is a very good point, and we agree that this needs further clarification as no subject was classified as having one condition when they had multiple conditions, and this is not clearly stated in the methods section. During the lengthy LTC data coding process, careful attention was given to the number of individual self-reported conditions when developing the final groupings of conditions to ensure that an individual with two or more conditions always had these conditions listed separately, and not within a single condition group. Once the list of LTC groups were established, a further check was made within Stata by viewing all the condition groups against each participant ID to confirm that the number of conditions within each group for each participant was never more than 1. This process was essential to determining the correct number of LTCs self-reported by each participant.

To ensure clarity on this point, we have amended the long-term conditions section of the methods as follows:

Methods page 7, lines 140-144. This sentence has been amended so it now reads: ‘The final list of LTCs was constructed by allocating conditions either into specific condition groups (for example, two condition groups were asthma and high blood pressure), or into groups of similar conditions where individual case numbers were <5 (for example, gastrointestinal conditions and respiratory conditions).’

Methods page 7, line 144-145: The following sentence was also inserted: ‘No participant had more than one condition within any of the individual combined condition groups.’

6. Editor comment:

Your exclusion of individuals whose health status improves may introduce significant selection bias. The current emphasis on studying individuals with a deteriorating health status does not sufficiently justify your choice of analysis. We strongly suggest revising your analysis to include all participants.

Response from authors:

Whilst ‘improvers’ were identified in the data and was discussed in the research team as it is interesting, they were not included as an additional LTC trajectory group primarily as this group was never included in the study protocol, and we were confident that excluding it was not detrimental to the key aims of the study focusing on participants reporting increased numbers of LTCs (this was acknowledged in the methods section, page 9, lines 207-209). The ‘improvers ‘group was also complex and comprised sub-groups with some participants reporting 1 LTC at S1 and none at S2, and another group reporting more than 1 LTC at S1 and fewer LTCs at S2. There were additional potential limitations in analysing this group as we did not know whether they improved due to successful on-going treatments, or had experienced a complete recovery, or were in remission (this included cancer) which would have been a major limitation to the findings.

From a statistical perspective, the multinomial regression used in the study comprises the two discrete categorical outcome groups (0 to ≥1 LTCs and 1-to-many LTCs), and inclusion of the improvers group would not have impacted the results of these two categorical outcomes, or imply an error in the results of these two categories due to selection bias.

However, had this study focused more generally on all LTC trajectories (and included comparative patient medical records), a significantly larger sample size would have enabled an in-depth stratified analysis of the ‘improved’ groups of participants. We have acknowledged the omission of this group in the limitation section of the discussion, and have added in ‘and the addition of medical record data’:

Discussion page 23, lines 417-421: “ A larger sample size and the addition of medical record data would also enable factors supporting positive health transitions to be investigated. By expanding the scope of future studies to include positive health transitions, this would provide guidance for health promotion interventions aimed at slowing the journey to multimorbidity”.

DISCUSSION

7. Editor comment:

Please refine the structure of your discussion for improved readability. For instance, the correlation between employment categories is discussed at two different points (lines 318 and 338). Consolidating this discussion into a single paragraph would greatly enhance readability.

Response from authors:

The discussion has now been extensively revised, with the employment section consolidated into a single paragraph and various additional edits made to enhance readability through this section. Please see the tracked changes in the discussion pages 19-23.

8. Editor comment:

Also, we recommend limiting speculative conclusions, as your study merely observes associations (e.g., employment can both be a potential causal factor for multimorbidity, or employment could reflect health status at baseline which is the real risk factor for multimorbidity).

Response from authors:

The revisions to the discussion have taken this point into consideration and made appropriate changes so that speculative conclusions are limited. Please see the changes made to the discussion listed in our response to editor comment 2, as these two comments are linked.

The revisions to the discussion also specifically ensure that there is no suggestion of causality in relation to the regression results between employment and the LTC trajectories. We are aware that there may be a bi-directional association between employment and health and acknowledged this in the following sentence in the discussion:

Discussion page 21, lines 358-360: ‘One possibility is that there is a higher proportion of people with one LTC at baseline who found the need to change from full-time to part-time employment, and so this group may already be at increased risk.’

This has been further highlighted in the limitations section of the discussion with the insertion of the following sentence:

Discussion page 23 lines 424-426: ‘Also, reverse causality cannot be excluded, particularly in the case of employment where health status may lead to reduced working hours or unemployment’.

9. Editor comment:

Also, please reconsider carefully your limitation statement. Consider for example the limited number of LTCs, the grouping of LTCs, the lack of information on propensity to seek care (and thus receive a diagnosis).

Response from authors:

The limitations section has been expanded and the following comment inserted into the discussion on page 23, lines 411-417.

‘In contrast to many studies using patient records, all self-reported conditions were included in this study. However, a larger sample size would have enabled a larger number of LTCs to be analysed separately rather than being combined into similar condition groups. Additionally, despite the advantages of including all self-reported conditions, it should also be noted that without access to medical records, we were unable to confirm whether a formal diagnosis had been received for all conditions, or if there were un-reported conditions that had not yet received a diagnosis or were being successfully managed by the individual.’

Attachment

Submitted filename: Response to Editors.docx

Click here for additional data file. (33KB, docx)

Decision Letter 2

Andrea Dell'Isola

29 Aug 2023

Risk factors for the progression to multimorbidity among UK urban working-age adults. A community cohort study

PONE-D-22-33742R2

Dear Dr. Stagg,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Andrea Dell'Isola

Academic Editor

PLOS ONE

Acceptance letter

Andrea Dell'Isola

1 Sep 2023

PONE-D-22-33742R2

Risk factors for the progression to multimorbidity among UK urban working-age adults. A community cohort study.

Dear Dr. Stagg:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Assoc Prof Andrea Dell'Isola

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Attachment

    Submitted filename: Response to reviewers_01.docx

    Click here for additional data file. (120.1KB, docx)
    Attachment

    Submitted filename: Response to Editors.docx

    Click here for additional data file. (33KB, docx)

    Data Availability Statement

    This study used third party data from the SELCoH survey. SELCoH data is available to researchers upon request. Please contact selcoh@kcl.ac.uk to apply for access. The authors confirm they did not receive any special privileges in accessing the data that other researchers would not have. For more information visit www.kcl.ac.uk/research/selcoh.

    Articles from PLOS ONE are provided here courtesy of PLOS

    RESOURCES