802868 BJP British Journal of PainSmith et al.

Article

British Journal of Pain

Association between musculoskeletal

2019, Vol 13(2) 82­–90
© The British Pain Society 2018
Article reuse guidelines:
pain with social isolation and loneliness: [Link]/journals-permissions
DOI: 10.1177/2049463718802868
[Link]

analysis of the English Longitudinal [Link]/home/bjp

Study of Ageing

Toby O Smith1 , Jack R Dainty2, Esther Williamson1 and Kathryn R Martin3

Abstract
Introduction: Musculoskeletal pain is a prevalent health challenge for all age groups worldwide, but
most notably in older adults. Social isolation is the consequence of a decrease in social network size with
a reduction in the number of social contacts. Loneliness is the psychological embodiment of social isola-
tion and represents an individual’s perception of dissatisfaction in the quality or quantity of their social
contacts. This study aims to determine whether a relationship exists between musculoskeletal pain and
social isolation and loneliness.
Methods: A cross-sectional analysis of the English Longitudinal Study of Ageing (ELSA) cohort was under-
taken. ELSA is a nationally representative sample of the non-institutionalised population of individuals
aged 50 years and over based in England. Data were gathered on social isolation through the ELSA Social
Isolation Index, loneliness through the University of California, Los Angeles (UCLA) Loneliness Scale and
musculoskeletal pain. Data for covariates included physical activity, depression score, socioeconomic
status, access to transport and demographic characteristics. Logistic regression analyses were under-
taken to determine the relationship between social isolation and loneliness with pain and the additional
covariates.
Results: A total of 9299 participants were included in the analysis. This included 4125 (44.4%) males, with
a mean age of 65.8 years. There was a significant association where social isolation was lower for those
in pain (odd ratio (OR): 0.87; 95% confidence intervals (CI): 0.75 to 0.99), whereas the converse occurred
for loneliness where this was higher for those in pain (OR: 1.15; 95% CI: 1.01 to 1.31). Age, occupation,
physical activity and depression were all associated with increased social isolation and loneliness.
Conclusion: People who experience chronic musculoskeletal pain are at greater risk of being lonely,
but at less risk of being socially isolated. Health professionals should consider the wider implications of
musculoskeletal pain on individuals, to reduce the risk of negative health implications associated with
loneliness from impacting on individual’s health and well-being.

Keywords
Pain, elderly, community, lonely, isolated, health outcomes

1Botnar Research Centre, Nuffield Department of Orthopaedics, Corresponding author:

Rheumatology and Musculoskeletal Sciences, University of Toby O Smith, Botnar Research Centre, Nuffield Department
Oxford, Nuffield Orthopaedic Centre, Oxford, UK of Orthopaedics, Rheumatology and Musculoskeletal Sciences,
2Norwich Medical School, University of East Anglia, Norwich, UK University of Oxford, Nuffield Orthopaedic Centre, Windmill Road,
3Institute of Applied Health Sciences, School of Medicine, Medical Oxford OX3 7LD, UK.
Sciences and Nutrition, University of Aberdeen, Aberdeen, UK Email: [Link]@[Link]
Smith et al. 83

Introduction guideline to report this comparative prospective

cohort study.16
Musculoskeletal pain is a prevalent health challenge
for all age groups worldwide, but most notably in older
adults.1,2 It is associated with increased disability, Participants
frailty, loss of independence and reduced quality of Data were identified from the English Longitudinal
life.3 The burden of musculoskeletal disease has Study of Ageing (ELSA) cohort. The ELSA study is an
increased, with the disability-adjusted life years increas- ongoing national cohort of English community-dwell-
ing from 20.6 to 30.9 million between 1990 and 2010.1 ing adults born on or before 29 February 1952. It is a
Social isolation is the consequence of a decrease in nationally representative sample of the non-institution-
social network size with a reduction in the number of alised general population.17 In this present cross-
social contacts.4 It can be either active, that is, withdrawal sectional analysis, data were initially identified from all
from one’s network, or passive where an individual’s 9432 people included in Wave 2 (2004/2005).
social network moves or dies for example.5 Loneliness is Original ethical approval was given by the London
the psychological embodiment of social isolation5 and Multi-Centre Research Ethics Service (MREC/01/2/91)
represents an individual’s perception of dissatisfaction in and written informed consent obtained from all par-
the quality or quantity of their social contacts. Loneliness, ticipants. Anonymised unlinked data for this study
therefore, incorporates the discrepancy between the rela- were provided by the UK Data Service.
tionships an individual has and the relationships they
would like to have.4 There are significant negative health
consequences associated with social isolation and loneli- Measurements
ness. These include an increased risk of cardiovascular All data were collected during the routine face-to-face
disease,6 infectious diseases,7 cognitive decline8 and all- follow-up interval.
cause mortality.9 Both social isolation and loneliness
impair quality of life, physical and mental health.4,5 Both Social isolation. Social isolation was measured using
are particularly prevalent with increasing age, with up to the validated and previously reported ELSA Social
50% of older people at risk of social isolation10 and Isolation Index.17,18 This index is derived from five
approximately a third of older individuals experiencing questions related to: marital/cohabiting status;
some degree of loneliness.11,12 monthly contact (including face-to-face, telephone, or
People with chronic musculoskeletal pain may written/e-mail contact) with children, other family
actively reduce contact with friends, family and other members and friends; and participation in organisa-
social networks.1,13 This has been attributed to patients tions such as social clubs or residents groups, religious
decreasing their social networks through friends ‘dis- groups or committees.17,18 Scores range from 0 to 5
appearing’ having not understood the pain or the qual- where higher scores indicate greater social isolation.
ity of contacts being reduced as patients feel that have The cut-point for the existence of social isolation was
to hide their true state of being.4 This may lead to ⩾2.19
greater social isolation and loneliness with reduced
opportunities for physical activity impacting on physi- Loneliness. Loneliness was measured using the vali-
cal and mental health. dated three-item short form of the Revised University
Previous literature has highlighted the association of California, Los Angeles (UCLA) Loneliness Scale.20
between social isolation and loneliness with mortality,5,14 Participants indicated how frequently they ‘felt left
However, it remains unclear whether there is a relation- out’, ‘isolated from others’ and ‘felt lonely’, with
ship between social isolation or loneliness and musculo- response options of ‘hardly ever’ or ‘never’, ‘some of
skeletal pain.15 Given the impact musculoskeletal pain the time’ and ‘often’. Scores range from 3 to 9 where
has on an individual’s health and well-being, and the higher scores indicate greater loneliness.20 The cut-
potential complex nature which musculoskeletal disease point for the existence of loneliness was ⩾6.19
has with social isolation and loneliness, it is important to
understand how these may or may not relate to one Pain. Musculoskeletal pain was assessed through par-
another. The purpose of this study was therefore to ticipants indicating whether they were often troubled
determine whether a relationship exists between muscu- by bone/joint/muscle pain or not (yes/no).
loskeletal pain, social isolation and loneliness.
Participant characteristics. Data included age, gender,
BMI (body mass index), ethnic classification (ELSA
Methods
defined as White/non-White) and occupational status
We have followed the Strengthening the Reporting of (as measured with the National Statistics Socio-
Observational Studies in Epidemiology (STROBE) economic Classification-3, NS-SEC3).
84 British Journal of Pain 13(2)

Depression. Depressive symptoms were assessed using total of 133 (1.4%) participants were missing from
the eight-item version of Center for Epidemiologic the analyses.
Studies Depression (CES-D) scale,21 with a cut-off The characteristics of the cohort are illustrated in
value of ⩾3 used to classify someone with depressive Table 1. The overall study sample included 4125
symptoms.19 (44.4%) males, with a mean age of 65.8 years (SD:
10.8) and mean BMI of 27.9 kg/m2 (SD: 4.9 kg/m2). Of
Physical activity. Participants were asked how often these, 97.6% were White ethnicity. Twenty-three per-
they engaged in mild, moderate or vigorous physical cent of the cohort reported depressive symptoms based
activity. For each level of activity, participants on the CES-D threshold. The cohort consisted of 2871
responded as being ‘very active’ (more than once a (30.9%) being managerial/professional occupations
week), ‘active’ (once a week), ‘moderately active’ (one while 4059 (43.6%) were routine and manual
to three times per month) and ‘inactive’ (hardly ever/ occupations.
never). We derived a summary index of physical activ-
ity by summing responses to the three physical activity Social isolation
items which were dichotomised using a cut-point of
once a week or more often.22 This physical activity There was no statistically significant relationship
assessment method has been previously used to deter- between pain and social isolation in the initial regres-
mine the level of physical activity participation under- sion model (OR: 0.88; 95% CI: 0.76 to 1.03). Of those
taken by older people.22,23 It has demonstrated who were socially isolated, 42% were often troubled by
excellent convergent validity within this population.24 pain compared to 36% of those who were not socially
isolated (Table 2) but this was not statistically signifi-
cant. There was, however, a significant association
Data analysis between social isolation and age (OR: 1.01; 95% CI:
All data were initially analysed with descriptive statis- 1.00 to 1.02), BMI (OR: 0.98; 95% CI: 0.97 to 0.99),
tics to present the frequency (%) or mean and standard gender (OR: 0.81; 95% CI: 0.70 to 0.94), occupation
deviation (SD) values. (OR: 1.61; 95% CI: 1.36 to 1.91), the frequency to
Given the potential relationship between pain and which participants undertook moderate (OR: 1.60;
depression on social isolation and loneliness, the asso- 95% CI: 1.28 to 1.99) and vigorous levels of physical
ciation between patient characteristics, pain and activity (OR: 1.51; 95% CI: 1.22 to 1.89) and depres-
depression with the response variables of social isola- sion (OR: 1.68; 95% CI: 1.43 to 1.99).
tion or loneliness were assessed using logistic regres- When analysed in the step-wise regression model,
sion. In these models, ‘pain’ was analysed using the the variables which remained statistically significant
‘often troubled by pain’ variable (binary: yes/no). This with social isolation are presented in Table 3.
modelling strategy used all the explanatory variables of Participants who reported being often troubled by pain
a priori interest in an initial logistic regression model to were 13% less likely to report being socially isolated
examine the univariate relationship between each vari- (OR: 0.87; 95% CI: 0.75 to 0.99). Increasing age (OR:
able and the outcome. Following this, we eliminated 1.02; 95% CI: 1.01 to 1.02), occupations which are
(using a backward step-wise regression approach) each more intermediate or manual in nature (OR: 1.70; 95
variable (in turn) that was least significant until a final CI: 1.43 to 1.99), those who were more sedentary
multivariable model was arrived at which only included when assessed by moderate physical activity (OR: 1.60;
explanatory variables that met the significance criteria 95% CI: 1.34 to 1.91) and vigorous physical activity
(p < 0.05). Regression model data were presented as (OR: 1.60; 95% CI: 1.30 to 1.96) and those with self-
odd ratio (OR) and 95% confidence intervals (CIs). reported depression (OR: 1.80; 95% CI: 1.56 to 2.09)
All analyses were performed in the R statistics pack- had a greater probability of being socially isolated.
age (R Foundation for Statistical Computing, Vienna, Females had a reduced risk of being socially isolated
Austria). (OR: 0.82; 95% CI: 0.72 to 0.93).

Results Loneliness
Cohort characteristics There was a statistically significant relationship
between pain and loneliness on the initial regression
A total of 9432 participants were identified from model (OR: 1.23; 95% CI: 1.06 to 1.43). Of those who
Wave 2 of the ELSA cohort. Of these, there were were lonely, 47% were often troubled by pain com-
available data from 9299 participants who did or did pared to 34% of those who were not lonely (Table 2).
not report being often troubled by pain (Table 1). A There were significant associations between loneliness
Smith et al. 85

Table 1. Descriptive characteristics of the analysed cohort.

Often troubled with pain

Yes (N = 3513) No (N = 5786)

Age, mean (SD) 66.7 (10.6) 65.1 (10.6)
BMI, mean (SD) 28.9 (5.4) 27.4 (4.5)
Gender
Male 38.5% 46.7%
Female 61.5% 53.3%
Ethnicity
White 96.8% 98.1%
Non-White 3.2% 1.9%
Occupation (NS-SEC3)
Managerial/professional 25.0% 35.0%
Intermediate 24.3% 25.1%
Routine and manual 50.7% 39.8%
Loneliness
Not lonely 74.3% 85.0%
Lonely 25.7% 15.0%
Social isolation
Not socially isolated 83.3% 86.1%
Socially isolated 16.7% 13.9%
Mild physical activity
More than once a week 71.3% 81.7%
Once a week 11.2% 9.2%
One to three times a month 3.8% 3.3%
Hardly ever or never 13.7% 5.8%
Moderate physical activity
More than once a week 46.4% 68.2%
Once a week 14.4% 14.8%
One to three times a month 10.2% 6.1%
Hardly ever or never 29.0% 10.9%
Vigorous physical activity
More than once a week 12.3% 21.2%
Once a week 6.4% 11.8%
One to three times a month 8.3% 12.0%
Hardly ever or never 73.0% 55.0%
Self-reported depression (CES-D score)
Not depressed 62.8% 85.2%
Depressed 37.2% 14.8%

BMI: body mass index; CES-D: Center for Epidemiologic Studies Depression; N: number of participants; NS-SEC3: National Statistics
Socio-economic Classification-3; SD: standard deviation.

and age (OR: 1.01; 95% CI: 1.00 to 1.02), BMI (OR: age (OR: 1.01; 95% CI: 1.00 to 1.02), females (OR:
0.98; 95% CI: 0.97 to 0.99), gender (OR: 1.31; 95% 1.28; 95% CI: 1.13 to 1.45), non-White ethnicity (OR:
CI: 1.13 to 1.52), ethnicity (OR: 1.77; 95% CI: 1.04 to 1.91; 95% CI: 1.34 to 2.90), occupations which are
2.94), occupation (OR: 1.50; 95% CI: 1.27 to 1.78) more intermediate or manual in nature (OR: 1.52; 95
and depression (OR: 5.46; 95% CI: 4.71 to 6.34). CI: 1.31 to 1.77), and greater sedentary behaviours
When analysed using the step-wise regression when assessed by moderate physical activity (OR: 1.51;
model, the variables which remained statistically sig- 95% CI: 1.28 to 1.78) had a greater probability of
nificant with loneliness are presented in Table 4. reporting loneliness. Of particular note, there was a
Participants who were often troubled by pain were fivefold greater probability of reporting loneliness
15% more likely to report being lonely (OR: 1.15; 95% when individuals were depressed compared to those
CI: 1.01 to 1.31). The data indicated that increasing who were not (OR: 5.23; 95% CI: 4.59 to 5.96).
Table 2. Results of the univariate regression analysis.
86

Social isolation Loneliness

Not socially Socially Odd ratio (95% Not lonely Lonely Odd ratio (95%
isolated (N = 7908) isolated (N = 1391) CI) p-value (N = 7528) (N = 1771) CI) p-value
Often troubled with pain
No 64% 58% Reference Reference 66% 53% Reference Reference
Yes 36% 42% 0.88 (0.76, 1.03) 0.118 34% 47% 1.23 (1.06, 1.43) 0.006
Age, mean (SD) 64.8 (10.0) 68.6 (12.0) 1.01 (1.00, 1.02) 0.002 64.7 (10) 68.1 (12) 1.01 (1.00, 1.02) 0.043
BMI, mean (SD) 27.9 (4.8) 27.7 (5.1) 0.98 (0.97, 0.99) 0.031 27.9 (4.7) 27.9 (5.1) 0.98 (0.97, 0.99) 0.012
Gender
Male 43% 45% Reference Reference 46% 39% Reference Reference
Female 57% 55% 0.81 (0.70, 0.94) 0.006 54% 61% 1.31 (1.13, 1.52) <0.001
Ethnicity
White 98% 95% Reference Reference 99% 95% Reference Reference
Non-White 2% 5% 1.25 (0.68, 2.16) 0.452 1% 5% 1.77 (1.04, 2.94) 0.032
Occupation (NS-SEC3)
Managerial/professional occupations 34% 23% Reference Reference 35% 23% Reference Reference
Intermediate occupations 26% 22% 1.14 (0.94, 1.40) 0.192 25% 24% 1.40 (1.16, 1.70) <0.001
Routine and manual occupations 41% 54% 1.61 (1.36, 1.91) <0.001 40% 53% 1.50 (1.27, 1.78) <0.001
Mild physical activity
More than once a week 80% 69% Reference Reference 81% 70% Reference Reference
Once a week 10% 11% 0.99 (0.78, 1.25) 0.916 9% 12% 1.17 (0.93, 1.47) 0.176
One to three times a month 3% 4% 1.11 (0.76, 1.58) 0.579 4% 3% 0.73 (0.47, 1.11) 0.150
Hardly ever or never 7% 15% 1.22 (0.93, 1.60) 0.142 6% 15% 0.97 (0.72, 1.28) 0.798
Moderate physical activity
More than once a week 64% 47% Reference Reference 65% 47% Reference Reference
Once a week 15% 13% 1.08 (0.88, 1.33) 0.460 15% 14% 1.15 (0.94, 1.40) 0.173
One to three times a month 7% 9% 1.47 (1.14, 1.89) 0.003 7% 9% 1.08 (0.83, 1.40) 0.548
Hardly ever or never 14% 30% 1.60 (1.28, 1.99) <0.001 13% 30% 1.23 (0.99, 1.54) 0.061
Vigorous physical activity
More than once a week 20% 12% Reference Reference 20% 12% Reference Reference
Once a week 11% 6% 1.00 (0.72, 1.36) 0.977 11% 7% 1.194 (0.90, 1.59) 0.224
One to three times a month 11% 9% 1.47 (1.11, 2.00) 0.007 11% 9% 1.212 (0.92, 1.60) 0.176
Hardly ever or never 58% 73% 1.51 (1.22, 1.89) <0.001 57% 72% 1.228 (1.00, 1.52) 0.054
Self-reported depression (CES-D
score)
Not depressed 81% 65% Reference Reference 86% 55% Reference Reference
Depressed 19% 35% 1.68 (1.43, 1.99) <0.001 14% 45% 5.46 (4.71, 6.34) <0.001

BMI: body mass index; CES-D: Center for Epidemiologic Studies Depression; CI: confidence interval; N: number of participants; NS-SEC3: National Statistics Socio-economic
British Journal of Pain 13(2)

Classification-3; OR: odd ratio; SD: standard deviation.

Smith et al. 87

Table 3. Results from the step-wise regression (backward elimination) analysis for social isolation.

Odd ratio (95% CI) p-value

Often troubled with pain
No Reference Reference
Yes 0.87 (0.75, 0.99) 0.038
Age 1.02 (1.01, 1.02) <0.001
Gender
Male Reference Reference
Female 0.82 (0.72, 0.93) 0.002
Occupation (NS-SEC3)
Managerial/professional occupations Reference Reference
Intermediate occupations 1.25 (1.04, 1.50) 0.015
Routine and manual occupations 1.70 (1.45, 1.99) <0.001
Moderate physical activity
More than once a week Reference Reference
Once a week 1.03 (0.85, 1.24) 0.756
One to three times a month 1.43 (1.136, 1.781) 0.002
Hardly ever or never 1.60 (1.34, 1.91) <0.001
Vigorous physical activity
More than once a week Reference Reference
Once a week 1.03 (0.77, 1.38) 0.846
One to three times a month 1.46 (1.12, 1.91) 0.004
Hardly ever or never 1.59 (1.30, 1.96) <0.001
Self-reported depression (CES-D score)
Not depressed Reference Reference
Depressed 1.80 (1.56, 2.09) <0.001

CES-D: Center for Epidemiologic Studies Depression; NS-SEC3: National Statistics Socio-economic Classification-3; SD: standard
deviation; CI: confidence interval.

Table 4. Results from the step-wise regression (backward elimination) analysis for loneliness.

Odd ratio (95%CI) p-value

Often troubled with pain
No Reference Reference
Yes 1.15 (1.01, 1.31) 0.031
Age 1.01 (1.00, 1.02) 0.004
Gender
Male Reference Reference
Female 1.28 (1.13, 1.45) <0.001
Ethnicity
White Reference Reference
Non-White 1.91 (1.24, 2.90) 0.003
Occupation (NS-SEC3)
Managerial/professional occupations Reference Reference
Intermediate occupations 1.36 (1.14, 1.61) <0.001
Routine and manual occupations 1.52 (1.31, 1.77) <0.001
Moderate physical activity
More than once a week Reference Reference
Once a week 1.21 (1.02, 1.44) 0.030
One to three times a month 1.22 (0.97, 1.52) 0.079
Hardly ever or never 1.51 (1.28, 1.78) <0.001
Self-reported depression (CES-D score)
Not depressed Reference Reference
Depressed 5.23 (4.59, 5.96) <0.001

CES-D: Center for Epidemiologic Studies Depression; NS-SEC3: National Statistics Socio-economic Classification-3; SD: standard
deviation; CI: confidence interval.
88 British Journal of Pain 13(2)

Discussion (defined as musculoskeletal pain for ⩾3 months). The

relationship between loneliness and negative emotions
This is the first study to investigate the relationship has been reported in Dures et al.33 for those with
between social isolation, loneliness and musculoskele- inflammatory arthritis. While causation between mus-
tal pain using validated measures at a population-based culoskeletal pain, loneliness and depression cannot be
level. These findings indicate that individuals with ascertained in this analysis given its cross-sectional
musculoskeletal pain have a greater probability of design, there appears an important relationship
experiencing loneliness, but are less likely to experi- between these variables which should be emphasised
ence social isolation. However, factors such as age, for clinical consideration. It has been suggested that
occupation, level of physical activity and depression are depression and loneliness may be amenable to change
consistently associated with the probability of individ- through psychological interventions and support.34,35
uals experiencing social isolation or loneliness. For specific musculoskeletal conditions such as rheu-
While the data indicate that there is an association matoid arthritis, international and national guidelines
where musculoskeletal pain had a negative impact to such as the European and UK treatment guidelines
increase loneliness, the opposite occurred in social iso- and the National Institute for Health and Care
lation where musculoskeletal pain was associated with Excellence (NICE) have recommended that patients
a decrease in social isolation. This was unexpected. It is should be offered psychological interventions as part
hypothesised that, for this population, being in pain of multidisciplinary care.36–38 However, the adoption
may result in an increase in contact with friends, family of such recommendations has been reported as
members or social networks potentially in a caring or variable.39,40 Dures et al.33 suggested that patient–clini-
support role or taking them to healthcare appoint- cian interaction can positively or negative influence
ments. While this is an increase in social contact, indi- patient’s psychological status depending on their per-
viduals, based on this data, still perceived themselves to ceived willingness and ability to acknowledge emo-
be lonely. This may be attributed to the difference in tional and social challenges. Consideration of the
these two constructs. While social isolation is the fre- psychological distress and global well-being which
quency of contact between individuals and society, individuals with chronic pain have is therefore a key
loneliness is a perception of feeling isolated regardless recommendation which may positively influence self-
of the breadth of actual social networks.25,26 The find- efficacy and self-management strategies.
ings suggest that while pain may not hinder the degree This analysis presented with three key limitations.
to which people have or engage in society, they seem to First, these data were not linked to hospital or medical
perceive being less connected. This can have a detri- records. It was therefore not possible to determine the
mental health consequence including anxiety, depres- musculoskeletal pathologies which this cohort pre-
sion, atrophy and overall physical deconditioning.27–29 sented. However, it may be surmised that a large pro-
Consideration as to how to enhance this perception of portion of participants will present with osteoarthritis,
pre-existing social networks with cognitive behaviour given the age and joints affected.41 Nonetheless, future
interventions may help prevent loneliness from nega- subgroup analyses based on the type of musculoskele-
tively impacting on these individual’s lives. tal disease would be valuable to be able to determine
There was a consistent difference in the relationship whether there is a difference, at least, between inflam-
between pain with loneliness by age. This has been pre- matory and non-inflammatory musculoskeletal dis-
viously reported in other painful conditions, demon- eases given their differences in pathological mechanisms
strating differences in pain response, pain-related and drivers.42 Second, while the data provide a national
attitudes and stoicism.30 This has been attributed to representation from England, facilitated by the ‘low-
older people under-reporting pain with age-related tech’ data collection approaches, the data were largely
increases in the degree of reticence to pain and reluc- self-reported, requiring participant recall. Accordingly,
tance to label a sensation as painful.31 Based on this there is a potential risk that the data may have been
study, and previous literature in other pain-related influenced by both recall error and social desirability
conditions, targeting interventions to address loneli- bias which may have inflated or suppressed the effect
ness among those with musculoskeletal pain may be depending on the respondent’s perception of the ques-
most effective if messages are tailored to specific age tions asked. Nonetheless, this dataset provides a signal
groups. from a large number of participants, from differing
The association between musculoskeletal pain and social circumstances and demographics, therefore pro-
loneliness and depression is supported by previous lit- viding valuable data to better understand the relation-
erature. Rapo-Pylkkö et al.32 reported that older adults ship between musculoskeletal pain, social isolation and
with chronic pain more frequently presented with loneliness. Finally, due to considerable issues with
poorer perceived function, felt sadder, lonelier and missing data, pain was measured using whether indi-
more fatigued compared to those without chronic pain viduals were ‘often troubled by pain’ as opposed to
Smith et al. 89

pain scores such as numerical rating scores. There is Contributorship

therefore no estimation on the severity of pain. We can confirm that the following authors have made sub-
Understanding the relationship between pain severity stantial contributions to the following:
and social isolation and loneliness would be valuable Conception and design: TS, JD, EW, KM
for further research. Acquisition of data: TS, JD
Analysis and interpretation of data: TS, JD, EW, KM
Drafting the article: TS, JD, EW, KM
Conclusion Revising it critically for important intellectual content: TS,
The findings of this analysis indicate that there is a JD, EW, KM
Final approval of the version to be published: TS, JD, EW, KM
relationship between musculoskeletal pain and loneli-
ness and social isolation, where pain is associated with
increased loneliness but decreased social isolation in ORCID iD
community-dwelling older adults in England. Health Toby O Smith [Link]
professionals are recommended to consider the wider
implications of pain on individuals to reduce the risks References
of negative health implications associated with loneli- 1. Smith TO, Purdy R, Lister S, et al. Living with osteoar-
ness from impacting on individual’s health and thritis: a systematic review and meta-ethnography. Scand
well-being. J Rheumatol 2014; 43: 441–452.
2. Mody GM and Brooks PM. Improving musculoskeletal
Acknowledgements health: global issues. Best Pract Res Clin Rheumatol 2012;
26: 237–249.
The English Longitudinal Study of Ageing (ELSA) was
3. Briggs AM, Cross MJ, Hoy DG, et al. Musculoskeletal
developed by researchers based at University Colleague
health conditions represent a global threat to healthy
London, the Institute of Fiscal Studies and the National
aging: a report for the 2015 World Health Organization
Centre for Social Research. We can confirm that the follow-
world report on ageing and health. Gerontologist 2016;
ing authors have made substantial contributions to the fol-
56(Suppl 2): S243–S255.
lowing: conception and design: T.O.S., J.R.D., E.W. and
4. Poscia A, Stojanovic J, La Milia DI, et al. Interventions
K.R.M. Acquisition of data: T.O.S. and J.R.D. Analysis and
targeting loneliness and social isolation among the older
interpretation of data: T.O.S., J.R.D., E.W. and K.R.M.
people: an update systematic review. Exp Gerontol 2017;
Drafting the article: T.O.S., J.R.D., E.W. and K.R.M.
102: 133–144.
Revising it critically for important intellectual content:
5. Steptoe A, Shankar A, Demakakos P, et al. Social isola-
T.O.S., J.R.D., E.W. and K.R.M. Final approval of the ver-
tion, loneliness, and all-cause mortality in older men and
sion to be published: T.O.S., J.R.D., E.W. and K.R.M.
women. Proc Natl Acad Sci U S A 2013; 110: 5797–5801.
6. Tillmann T, Pikhart H, Peasey A, et al. Psychosocial and
Conflict of interest socioeconomic determinants of cardiovascular mortal-
The author(s) declared no potential conflicts of interest with ity in Eastern Europe: a multicentre prospective cohort
respect to the research, authorship and/or publication of this study. PLoS Med 2017; 14: e1002459.
article. 7. Cohen S, Doyle WJ, Turner RB, et al. Emotional style
and susceptibility to the common cold. Psychosom Med
2003; 65: 652–657.
Ethical approval
8. Kotwal AA, Kim J,Waite L, et al. Social function and cogni-
Ethical approval for the ELSA cohort was obtained from tive status: results from a US nationally representative sur-
London Multi-Centre Research Ethics Service vey of older adults. J Gen Intern Med 2016; 31: 854–862.
(MREC/01/2/91). 9. Smith SG, Jackson SE, Kobayashi LC, et al. Social iso-
lation, health literacy, and mortality risk: findings from
Funding the English Longitudinal Study of Ageing. Health Psy-
chol 2018; 37: 160–169.
T.O.S. and E.W. are supported by funding from the National
10. Ibrahim R, Abolfathi Momtaz Y and Hamid TA. Social
Institute for Health Research (NIHR) Oxford Health
isolation in older Malaysians: prevalence and risk fac-
Biomedical Research Centre. The views expressed are those
tors. Psychogeriatrics 2013; 13: 71–79.
of the author(s) and not necessarily those of the NIHR.
11. Grenade L and Boldy D. Social isolation and loneli-
ness among older people: issues and future challenges
Guarantor in community and residential settings. Aust Health Rev
T.O.S. is the Guarantor of this article. 2008; 32: 468–478.
12. Victor CR and Yang K. The prevalence of loneliness
among adults: a case study of the United Kingdom. J
Informed consent
Psychol 2012; 146(1–2): 85–104.
Informed consent was provided by each individual partici- 13. Nyvang J, Hedström M and Gleissman SA. It’s not just
pant involved in the cohort study. a knee, but a whole life: a qualitative d
­ escriptive study on
90 British Journal of Pain 13(2)

patients’ experiences of living with knee osteoarthritis outcomes among the elderly. J Aging Health 2006; 18:
and their expectations for knee arthroplasty. Int J Qual 359–384.
Stud Health Well-Being 2016; 11: 30193. 29. Jessen MAB, Pallesen AVJ, Kriegbaum M, et al.
14. Elovainio M, Hakulinen C, Pulkki-Råback L, et al. Con- The association between loneliness and health – a
tribution of risk factors to excess mortality in isolated survey-based study among middle-aged and older
and lonely individuals: an analysis of data from the UK adults in Denmark. Aging Ment Health 2017; 1: 48480.
Biobank cohort study. Lancet Public Health 2017; 2: 30. Mah K, Tran KT, Gauthier LR, et al. Do correlates of
e260–e226. pain-related stoicism and cautiousness differ in younger
15. Peat G, Thomas E, Handy J, et al. Social networks and and older people with advanced cancer? J Pain 2018; 19:
pain interference with daily activities in middle and old 301–316.
age. Pain 2004; 112: 397–405. 31. Yong HH. Can attitudes of stoicism and cautiousness
16. Von Elm E, Altman DG, Egger M, et al. The strengthen- explain observed age-related variation in levels of self-
ing the reporting of observational studies in epidemiology rated pain, mood disturbance and functional inter-
(STROBE) statement: guidelines for reporting observa- ference in chronic pain patients? Eur J Pain 2006; 10:
tional studies. J Clin Epidemiol 2008; 61: 344–349. 399–407.
17. Steptoe A, Breeze A, Banks J, et al. Cohort profile: the 32. Rapo-Pylkkö S, Haanpää M and Liira H. Chronic pain
English Longitudinal Study of Ageing. Int J Epidemiol among community-dwelling elderly: a population-based
2013; 42: 1640–1648. clinical study. Scand J Prim Health Care 2016; 34: 159–
18. Shankar A, McMunn A, Banks J, et al. Loneliness, social 164.
isolation, and behavioral and biological health indicators 33. Dures E, Fraser I, Almeida C, et al. Patients’ perspectives
in older adults. Health Psychol 2011; 30: 377–385. on the psychological impact of inflammatory arthritis
19. Reinhard E, Courtin E, Van Lenthe F, et al. Public trans- and meeting the associated support needs: open-ended
port policy, social engagement and mental health in older responses in a Multi-Centre Survey. Musculoskelet Care
age: a quasi-experimental evaluation of free bus passes in 2017; 15: 175–185.
England. J Epidemiol Commun Health 2018; 72: 5. 34. Cruwys T, Alexander Haslam S, Dingle GA, et al. Feeling
20. Hughes ME, Waite LJ, Hawkley LC, et al. A short scale connected again: interventions that increase social identi-
for measuring loneliness in large surveys: results from two fication reduce depression symptoms in community and
population-based studies. Res Aging 2004; 26: 655–672. clinical settings. J Affect Disord 2014; 159: 139–146.
21. Radloff LS. The CES-D scale: a self-report depression 35. VanderWeele TJ, Hawkley LC, Thisted RA, et al. A
scale for research in the general population. Appl Psychol marginal structural model analysis for loneliness: impli-
Meas 1977; 1: 385–401. cations for intervention trials and clinical practice. J
22. Garfield V, Llewellyn CH and Kumari M. The rela- Consult Clin Psychol 2011; 79: 225–235.
tionship between physical activity, sleep duration and 36. Luqmani R, Hennell S, Estrach C, et al. British Society
depressive symptoms in older adults: the English Longi- for Rheumatology and British Health Professionals in
tudinal Study of Ageing (ELSA). Prev Med Rep 2016; 4: Rheumatology guideline for the management of rheu-
512–516. matoid arthritis (after the first 2 years). Rheumatology
23. Demakakos P, Hamer M, Stamatakis E, et al. Low- 2009; 48: 436–439.
intensity physical activity is associated with reduced risk 37. National Institute for Health Care Excellence. Rheuma-
of incident type 2 diabetes in older adults: evidence from toid arthritis: the management of rheumatoid arthritis
the English Longitudinal Study of Ageing. Diabetologia in adults [CG79]. Available at: [Link]
2010; 53: 1877–1885. CG79 (2009, accessed 13 March 2018).
24. Hamer M, Molloy GJ, de Oliveira C, et al. Leisure time 38. Forestier R, Andre-Vert J, Guillez P, et al. Non-drug
physical activity, risk of depressive symptoms, and inflam- treatment (excluding surgery) in rheumatoid arthritis:
matory mediators: the English Longitudinal Study of clinical practice guidelines. Joint Bone Spine 2009; 79:
Ageing. Psychoneuroendocrinology 2009; 34: 1050–1055. 691–698.
25. Holwerda TJ, Beekman AT, Deeg DJ, et al. Increased 39. Firth J, Snowden N, Ledingham J, et al. The first national
risk of mortality associated with social isolation in older clinical audit for rheumatoid arthritis. Br J Nurs 2016;
men: only when feeling lonely? Results from the Amster- 25: 613–617.
dam Study of the Elderly (AMSTEL). Psychol Med 40. Ledingham JM, Snowden N, Rivett A, et al. Achievement
2012; 42: 843–853. of NICE quality standards for patients with new presen-
26. Stickley A and Koyanagi A. Physical multimorbidity and tation of inflammatory arthritis: observations from the
loneliness: a population-based study. PLoS One 2018; national clinical audit for rheumatoid and early inflam-
13: e0191651. matory arthritis. Rheumatology 2017; 56: 223–230.
27. Luo Y, Hawkley LC, Waite LJ, et al. Loneliness, health, 41. Glyn-Jones S, Palmer AJ, Agricola R, et al. Osteoarthri-
and mortality in old age: a national longitudinal study. tis. Lancet 2015; 386: 376–387.
Soc Sci Med 2012; 74: 907–914. 42. Berenbaum F. Osteoarthritis as an inflammatory dis-
28. Tomaka J, Thompson S and Palacios R. The relation of ease (osteoarthritis is not osteoarthrosis!). Osteoarthritis
social isolation, loneliness, and social support to disease ­Cartilage 2013; 21: 16–21.