Perceived social support impacts on exercise adherence in patients with chronic low back pain

Abstract

BACKGROUND:

Initiating and maintaining physical activity is particularly difficult for at-risk groups such as those with chronic low back pain (LBP).

OBJECTIVE:

This study aimed to assess whether there is a relationship between perceived social support (e.g., emotional and physical support) in individuals recently discharged from treatment for chronic LBP and the number of exercise sessions and total amount of exercise they perform over a 6-month period. This study also aimed to investigate a possible mediation effect of exercise self-efficacy on the relationship between perceived social support and exercise adherence (number of exercise sessions and total amount of exercise).

METHODS:

This prognostic study employed a secondary analysis of data collected for The Buddy Study; Data was collected through online weekly diaries over a 6-month period. Poisson regression analyses were used to quantify the relationship between social support at baseline and total number of exercise sessions, and total amount of exercise performed (frequency and duration) in the 6-month follow-up period. Where a relationship was observed, the Stata SEM command was used for the mediation analysis.

RESULTS:

A positive relationship was found between participants’ perceived levels of social support at baseline and the total number of exercise sessions they performed (IRR $=$ 1.56, 95%CI: 1.18 to 2.06) and the total amount of exercise they performed (IRR $=$ 1.57, 95%CI: 1.05 to 2.35) during the 6-month follow-up period. No mediating relationship was found between exercise self-efficacy at 3-months and the total number of exercise sessions performed ($\beta$ 3.96, 95% CI: $-$4.91 to 12.84), or the total amount of exercise performed ($\beta$ 243.96, 95% CI: $-$258.08 to 746.01).

CONCLUSION:

Social support is potentially an important aspect of exercise adherence, following discharge from treatment, for those with chronic LBP. People’s self-efficacy to exercise does not appear to mediate this relationship. Harnessing social support following physiotherapy treatment may increase exercise adherence and may therefore improve long term outcomes for those with chronic LBP.

1. Introduction

Low back pain (LBP) is the leading cause of years lived with disability (YLD) globally [1], with the percentage of YLDs associated with LBP increasing by 13% from 2010 to 2019 [2]. In response to this increasing global burden, a recent call for action has recommended clinicians, researchers, and policy makers to consider shifting from the current medical management of LBP towards the promotion of patients’ self-management and adoption of healthy lifestyles to support people to live well with LBP [1].

The World Health Organisation (WHO) recommends adults to perform at least 150 to 300 minutes of moderate to vigorous physical activity (PA) throughout the week [3], yet only one-third of adults in the United States of America [4] and less than one-quarter in Australia [5] meet the current PA guidelines. This is of major concern for people diagnosed with chronic LBP, a condition that commonly results in decline of PA [6] engagement. Not surprisingly, engaging in adequate dosage and type of PA – for example vigorous leisure-time PA – is an effective approach [7] to improving function [8] and pain resilience [9] in chronic LBP [6].

Promoting the behaviour changes essential for initiating and maintaining PA (including participation in exercise) is particularly difficult for groups at risk of poor PA compliance, such as those with chronic LBP [10, 11]. The success or failure of discharge programs patients receive following their treatment for LBP often depends on how patients adhere to the PA-based advice [12]. This can be particularly challenging for patients navigating persistent pain and the functional and emotional limitations brought on by LBP [13]. Behaviour change theories [14, 15]have identified social support [16] (e.g., perceived emotional and physical support, or a sense of belonging), and self-efficacy [10] (confidence in one’s ability to perform a particular task) as being integral to initiating, maintaining, and gradually supporting increased levels of PA in healthy and older adults [10, 16].

The WHO has also recognised social support as a key determinant of maintaining adequate levels of PA [17] and a consistent predictor of exercise initiation and adherence [17, 18]. Increased levels of general social support [18] and PA-specific social support (e.g., encouragement to exercise or demonstrating positive exercise behaviours) are associated with younger, [19] and older adults’ [16] increased likelihood of engaging in PA.

Self-efficacy represents an individual’s belief in their ability to successfully perform a particular skill [14]. It is specific to the behaviour (e.g. exercise) as well as the environment [20] (e.g. home or outdoor setting), and is an important determinant of the amount of effort an individual will expend commencing a behaviour, and their persistence in the face of barriers and difficulties. [20]. Exercise self-efficacy (i.e. confidence to exercise in various conditions and situations) has been associated with exercise adherence in older adults [21, 22], adults undergoing cardiac rehabilitation [23], and those with chronic pain [24]. Social support and self-efficacy may be interconnected [14, 20] with self-efficacy as a mediator in the relationship between people’s perceived level of social support and their exercise adherence [10, 25]. However, the role of social support and self-efficacy has not been evaluated in patients with chronic LBP.

This study aimed to assess whether there is a relationship between perceived social support in individuals who have recently been discharged from treatment for chronic LBP and the number of exercise sessions and total amount of exercise they perform over a 6-month period. Secondly, we also aimed to examine if exercise self-efficacy acts as a mediator between perceived social support and engagement in exercise.

2. Methods

2.1. Study design

This prognostic study employed a secondary analysis of data collected as part of The Buddy Study. The Buddy Study was a pilot randomised control trial (RCT) investigating the feasibility of health coaching interventions to increase PA in people with chronic LBP following discharge from treatment. Details of The Buddy Study can be found in Supplementary Digital Content. In brief, individuals from the general community who reported recently being discharged from care by their physiotherapist, chiropractor, or GP were invited to participate. If eligible, they were randomly allocated into three groups following consent – standard discharge care, standard discharge care plus individual telephone health coaching, or standard discharge care plus telephone health coaching and the support of an exercise partner/buddy. The study collected data including pain measures (pain intensity, duration of pain, number of episodes of LBP), health care utilisation, disability, exercise self-efficacy, and perceived social support at baseline, 3-month, and 6-month follow-ups. Weekly data were also collected via an online diary which was accessed via a link that was emailed to participants every week. The diary included questions on PA (frequency and duration), type of exercise performed, whether they exercised alone or with another person and health care utilisation for LBP. Data were collected from August 2020 to May 2022.

To be included in The Buddy Study participants had to be over 50 years old and be discharged within the previous 6 months from care for chronic LBP by their treating physiotherapist, general practitioner, or chiropractor. They were eligible to be included if they reported having non-specific LBP (LBP not attributable to a recognised serious or neuropathic pathology) for at least 3 months, experiencing LBP intensity of $⩾$ 2/10 on a numeric rating scale and performed less than five hours of light to moderate intensity PA in the average week (assessed with the Active Australia Questionnaire [26]). At baseline, a trial coordinator contacted each study participant by phone, and they were reminded to continue with the standard discharge care that was provided by their treating therapist. Those randomised to the individual health coaching group were enrolled in a telephone health coaching program and were also recommended to continue with the standard discharge care. The health coaching program was provided by New South Wales Get Healthy Service^®, [27] and was designed to support the participant in setting healthy goals, ranging from increasing PA, or decreasing alcohol consumption, to improving dietary intake. Those randomised to the buddy-assisted group were asked to nominate a buddy as an exercise partner whom they could exercise with at least once a week and also recommended to continue their standard discharge care.

Figure 1.

Predicted total number of exercise sessions and total duration of exercise. Margins plot showing the total number of exercise sessions and the total duration of exercise over a 6-month period with confidence intervals, predicted at each score on the Duke UNC Functional Social Support Scale (following discharge from treatment for low back pain).

2.2. Independent variable

Perceived social support was considered the predictor of prognosis in this study, and was measured at baseline with the Duke UNC Functional Social Support Questionnaire (DUFSS) [28]. The DUFSS is an eight-item functional support instrument, which has been proven to be a reliable and valid measure of a person’s perception of the available support within their social support network. The DUFSS questionnaire asks questions regarding confidant support (e.g., chances to speak with others about personal or work concerns and receive advice about important things in life), as well as affective support (e.g., having people around them who care for them if they are unwell) [28].

2.3. Outcome variables

We assessed two prognostic outcomes: (i) the number of exercise sessions engaged in, and (ii) the total amount of exercise performed (frequency and duration) in the 6-month follow-up period. An online weekly diary was used to collect these measures over the 6-month follow-up period. Each week participants were inquired about how many days they had exercised alone or with another person, and the type of PA they had performed. They were also asked to report the frequency and the duration (in minutes) of each exercise session.

The total number of exercise sessions each participant performed was calculated as the total number of times (counts) the participant exercised alone or with another person in the 6-month period. The total amount of exercise (in minutes) performed by each participant was calculated by multiplying the total duration and frequency of exercise sessions performed alone or with another person each week to create a count total in the 6-month period.

2.4. Covariates

The following variables were considered for inclusion in the prognostic models given their association with exercise adherence and social support: age [29, 30], disability [31, 32], duration of LBP [33], and exercise self-efficacy [30].

2.5. Potential mediators

Exercise self-efficacy was assessed at baseline and at 3-months (as the potential mediating variable) using the Exercise Self-efficacy Scale [34] – a 15-item scale that measures an individual’s self-efficacy to exercise. Participants were asked to score their confidence to exercise (from 0%–100%) under various situations (e.g., when tired, when feeling anxious, when they have not reached their exercise goals or when they have no one to exercise with).

If a positive significant association (at the 0.05 level) was observed between social support and the total number of exercise sessions, or the total amount of exercise performed (frequency and duration) in the 6-month follow-up period, we investigated the possible mediating effect of exercise self-efficacy.

2.6. Statistical analysis

Descriptive analysis was performed using means and standard deviations or numbers and percentages to summarise the demographic data including age, gender, body mass index (BMI), employment status, and education level. LBP-specific descriptors were also included: LBP duration (years), pain score on a 0–100 numeric rating scale, and disability measured with the Roland Morris Disability Questionnaire (RMDQ) [35].

Poisson regression analyses were used to assess the relationship between participants’ perceived level of social support at baseline and (1) the total number of exercise sessions; and (2) the total amount of exercise performed (frequency and duration) in the 6-month period. Assumptions of count data, independence of variables, dispersion and linearity were initially assessed. Collinearity was then assessed (using the Stata vif command). Univariable and multivariable analyses were adjusted for exposure (using the Stata exposure command) to adjust for the number of weeks that participants were followed up with complete weekly diaries. Margins analyses were utilised to predict the outcomes (total number of exercise sessions and total amount of exercise performed in the six-month period) for each score on the social support scale. Finally, where a relationship ( 0.05) was observed between social support and the total number of exercise sessions, or the total amount of exercise performed, a mediation analysis was conducted to assess the possible mediating effect of exercise self-efficacy. The Stata SEM command was used for the mediation analysis using mlmv (maximum likelihood with missing values) and estat teffects (average treatment effects) and all mediation analyses were adjusted for baseline exercise self-efficacy as well as the covariates of age, duration of LBP and disability [29, 31, 36]. Ratios and proportions were then calculated to provide the proportion of total effects explained by social support, the indirect (mediating) effect explained by exercise self-efficacy, and the total effects (sum of direct and indirect effects). Robust standard errors were calculated (using the Stata vce robust command) for both the poisson regression analysis and the mediation analysis. All statistical analyses were performed using Stata SE 16.1 [37]. All $p$-values were two-sided, confidence intervals were set at 95% and associations were described with incidence risk ratios and coefficients where appropriate.

3. Results

3.1. Baseline characteristics

All participants from the Buddy Study were included in this analysis with data not included for those who did not complete the 3-month follow-up survey, resulting in 20 participants and a total of 441 data points (Table 1). The mean age of the sample was 63.3 years and 70.0% were female. Almost half of the participants were retired (47.4%) and 80.0% had achieved qualifications following high school. All participants reported experiencing chronic LBP with a mean duration of 15.5 years (Table 1), mean pain intensity of 49.3/100 (Table 2), mean disability score (RMDQ) of 11.9/24, and mean social support score (DUFSS) of 4/8. The mean exercise self-efficacy score (ESE) at baseline was 55/100 and at the 3-month follow up was 54/100 (Table 2).

Table 1.

Demographic variables

$n=$ 20	$n$, %
Age (mean, SD)	63.3 (8.0)
Gender (female)	14 (70.0%)
BMI (mean, SD)	28.34 (5.48)
Nationality ($n$, %)
Australian	16 (80.0)
French	1 (5.0)
New Zealander	3 (15.0)
Employment status ($n$, %)
Unemployed	1 (5.3)
Part time work	5 (26.3)
Full time work	3 (15.8)
Volunteer	1 (5.3)
Retired	9 (47.4)
Highest level of education attained ($n$, %)
Year 11 or below (equivalent)	0
Year 12 or equivalent	4 (20.0)
Certificate III/IV	3 (15.0)
Advanced diploma, diploma	5 (25.0)
Bachelor degree	1 (5.0)
Post graduate diploma/graduate certificate	3 (15.0)
Masters	4 (20.0)
Living arrangements ($n$, %)
Single	5 (25.0)
Married	10 (50.0)
De-facto	2 (10.0)
Rather not say	3 (15.0)
Number of other msk conditions (m, SD)	1 (1.1)
Min – Max	0–4
Do you have back pain (yes)	20 (100)
Duration of back pain in years (mean, SD)	15.5 (15)
Min – max	0.25–50
SD $=$ standard deviation

Table 2.

Clinical variables

	Total cohort	Usual care group	Individual group	Buddy group
	$n=$ 20	$n=$ 8	$n=$ 4	$n=$ 8
VAS at baseline (mean, SD)	49.30 (22.75)	49.38 (20.37)	61 (17.44)	43.38 (27.22)
Min – max	11–87	20–70	39–75	11–87
VAS at 3 months (mean, SD)	45.21 (25.54)	48.25 (22.69)	43.25 (27.97)	42.86 (30.73)
Min – max	10–86	10–75	20–80	10–86
VAS at 6 months (mean, SD)	45.25 (28.77)	51.88 (23.90)	40.50 (43.65)	41.00 (27.85)
Min – max	0–86	22–86	0–84	7–70
Disability at baseline (RMDQ) (mean, SD)	11.90 (4.54)	12 (3.89)	13.5 (4.20)	11 (5.55)
Min – max	1–19	5–17	8–18	1–19
DUFSS at baseline (mean, SD)	4.02 (0.81)	3.86 (1.044)	4.03 (0.58)	4.17 (0.69)
Min – max	2.37–5	2.38–5	3.25–4.63	2.75–5
Exercise self-efficacy at baseline (mean, SD)	55% (23.56)	58% (19.27)	52% (18.49)	53% (31.08)
Min – max	9%–89%	33%–86%	27%–72%	9%–89%
Exercise self-efficacy at 3-months (mean, SD) (n=19)	54% (15.99)	63% (16.43)	54% (13.85)	45% (11.97) ($n=$ 7)
Min – max	30%–81%	37%–81%	35%–68%	30%–62%
Exercise self-efficacy at 6-months (mean, SD)	59% (18.05)	61% (14.84)	51% (18.83)	62% (21.44)
Min – Max	30%–91%	37%–77%	35%–77%	30%–91%

SD $=$ standard deviation; msk $=$ musculoskeletal; DUFSS $=$ Duke UNC Functional Social Support Questionnaire; VAS $=$ Visual Analogue Scale; RMDQ $=$ Roland Morris Disability Questionnaire; LBP $=$ low back pain.

3.2. The relationship between perceived social support at baseline and total number of exercise sessions performed in the six-month period

After adjusting for covariates (age, disability, duration of LBP, and exercise self-efficacy at baseline), a positive relationship was found between participants’ perceived levels of social support at baseline and the total number of exercise sessions they performed in the 6-month follow up period (Incidence Rate Ratio [IRR]: 1.56, 95%CI: 1.18 to 2.06, $p=$ 0.002) (Table 3). For individuals with a social support score of 2 (minimum score for the sample), the predicted total number of exercise sessions performed in the 6-month follow up period was 28 (95% CI: 11.90 to 44.95, $p=$ 0.001), while in those with a social support score of 5 (maximum score for this cohort), the predicted total number of exercise sessions performed in the 6-month follow-up period was 107 (95% CI: 66.78 to 147.70, $p=$ 0.000) (Table 4). No mediating relationship was found between exercise self-efficacy at 3-months and the total number of exercise sessions performed ($\beta$ 3.96, 95% CI: $-$4.91 to 12.84, $p=$ 0.381), with the percentage of total effect mediated by exercise being 12% (Table 5).

Table 3.

Adjusted poisson regression analysis

	Coef.	IRR	SE	P $>$ z	95% CI
DUFSS ($n=$ 20)
Total number of exercise sessions	0.44	1.56	0.22	0.002*	1.18 to 2.06
Total duration of exercise sessions	0.45	1.57	0.32	0.027*	1.05 to 2.35

Poisson regression analysis of the relationship between perceived social support at baseline and count total number of exercise sessions and count total exercise duration in a six-month period. Coef. $=$ coefficient; IRR $=$ Incidence rate ratio; SE $=$ Standard error; P $>$ z $=$$p$ value; CI $=$ confidence interval; DUFSS $=$ Duke UNC Functional Social Support Questionnaire; Adjusted for exercise self-efficacy, age, length of time of low back pain and disability at baseline.

Table 4.

Margins predictions

DUFSS	Margin	SE	P $>$ z	95% CI
Total number of exercises sessions
1	18.26	7.88	0.020	2.81 to 33.70
2	28.43	8.43	0.001	11.90 to 44.95
3	44.25	7.71	0.000	29.14 to 59.36
4	68.89	7.96	0.000	53.28 to 84.50
5	107.24	20.64	0.000	66.78 to 147.70
6	166.95	53.16	0.002	62.76 to 271.14
7	259.90	118.07	0.028	28.48 to 491.31
8	404.60	240.06	0.092	$-$65.91 to 875.10
Total amount of exercise performed
1	660.63	460.91	0.152	$-$242.74 to 1564.01
2	1038.50	517.78	0.045	23.68 to2053.32
3	1632.49	499.76	0.001	652.98 to 2612.00
4	2566.22	382.06	0.000	1817.41 to 3315.04
5	4034.02	743.01	0.000	2577.75 to 5490.29
6	6341.36	2277.57	0.005	1877.41 to 10805.31
7	9968.42	5526.69	0.071	$-$863.69 to 20800.53
8	15670.06	11821.06	0.185	$-$7498.80 to 38838.92

Predicted margins for total number of exercise sessions and total amount of exercise performed over the 6-month follow-up period. SE $=$ Standard error; P $>$ z $=$$p$ value; CI $=$ confidence interval; DUFSS $=$ Duke UNC Functional Social Support Questionnaire; Adjusted for exercise self-efficacy, age, length of time of low back pain and disability at baseline.

Table 5.

Mediation analysis

	Coef	SE	P $>$ Z	95% CI
Total number of ex sessions performed alone or with another person
Direct effect
ESE (midpoint)	1.42	0.89	0.111	$-$0.33 to 3.17
DUFSS	28.41	13.65	0.037*	1.65 to 55.16
Indirect effect
DUFSS	3.96	4.53	0.381	$-$4.91 to 12.84
Total effect
ESE (midpoint)	1.42	0.89	0.111	$-$0.33 to 3.17
DUFSS	32.37	13.35	0.015*	6.21 to 58.53
Proportion of total effect mediated (indirect/total effect) $=$ 12%
Ratio of direct effect:total effect $=$ 88%
Total duration of exercise performed alone or with another person
Direct effect
ESE (midpoint)	92.42	29.39	0.002*	34.82 to 150.03
DUFSS	962.85	525.51	0.067*	$-$67.12 to 1992.83
Indirect effect
DUFSS	243.96	256.15	0.341	$-$258.08 to 746.01
Total effect
ESE (midpoint)	92.42	29.39	0.002*	34.82 to 150.03
DUFSS	1206.82	565.19	0.033*	99.07 to 2314.57

Proportion of total effect mediated (indirect/total effect) $=$ 20%; Ratio of direct effect: total effect $=$ 80%; Mediation analysis of the effect of perceived social support at baseline on count number of exercise sessions performed over a 6-month period with the mediation analysis of exercise self-efficacy at 3-month. Coef. $=$ coefficient; SE $=$ Standard error; Z $=$ z score; P $>$ z $=$$p$ value; CI $=$ confidence interval; ESE $=$ Exercise self-efficacy; DUFSS $=$ Duke UNC Functional Social Support Questionnaire; Adjusted for exercise self-efficacy, age, length of time of low back pain and disability at baseline.

3.3. The relationship between perceived social support at baseline and total amount of exercise performed in the six-month period

After adjusting for covariates (age, disability, duration of LBP, and exercise self-efficacy at baseline), a positive relationship was found between the participants’ perceived social support at baseline and the total amount of exercise they performed in the 6-month follow-up period (IRR: 1.57, 95%CI: 1.05 to 2.35, $p=$ 0.027) (Table 3). For individuals with a social support score of 2 (minimum score for the sample), the predicted total amount of exercise performed in the 6-month follow-up period was 1038.50 minutes (95% CI: 23.68 to 2053.32, $p=$ 0.045) (Table 4), and for individuals with a social support score of 5 (maximum score for the sample), the predicted total amount of exercise performed in the 6-month follow-up period was 4034.02 minutes (95% CI: 2577.75 to 5490.29, $p=$ 0.000) (Table 4). No mediating relationship was found between exercise self-efficacy at 3 months and the total amount of exercise performed ($\beta$ 243.96, 95% CI: $-$258.08 to 746.01, $p=$ 0.341), with the percentage of total effect mediated by exercise being 20% (Table 5).

4. Discussion

Results of this study showed that there is a positive relationship between people’s perceived levels of social support and their adherence to exercise, assessed as the number of exercise sessions and the total amount of exercise during a 6-month period after discharge from conservative treatment for LBP. No mediating effect was found of exercise self-efficacy at 3 months on the total number of exercise sessions or the total amount of exercise performed in the 6-month follow-up period. Long-term success of chronic LBP management relies on patient adherence to the exercise advice provided by their treating practitioner [38], and yet adherence to home exercise is undeniably poor [13, 25, 39], with approximately 50% of patients with chronic LBP being non-compliant with home exercise programs [12, 40]. The results of the current study indicate a possible pathway to overcoming the low compliance of those with chronic LBP to home exercise programs and potentially improving long-term outcomes by improving their perceived social support.

Our findings of the positive association between perceived social support and exercise adherence are in line with previous research. Research of older adults with chronic illness (excluding LBP) found those with higher perceived social support were more physically active at a 6-month follow-up than those with low levels of perceived social support [41]. They also found social support to be a better predictor of physical activity levels if combined with moderate to high levels of exercise self-efficacy. A larger social support network (and high levels of exercise self-efficacy) have been found to predict exercise participation in men and women with fibromyalgia irrespective of pain levels [42] and exercise participation in people with rheumatoid arthritis and osteoarthritis at 9-months following an exercise program [43].

Within social cognitive theory, social support and self-efficacy are both involved in regulating motivation and behaviour, and are thus intricately connected [14, 20]. This study did not find a mediating effect of exercise self-efficacy on the relationship between perceived social support and exercise (neither total number of exercise sessions nor total amount of exercise). The mean social support score for this cohort was 4 (out of a total score of 8) indicating moderate levels of perceived social support at baseline, potentially explaining the lack of mediating effect of self-efficacy as it is possible that self-efficacy is more impactful when people’s perceived levels of social support are low [41]. Further to this, social cognitive theory suggests that individuals with higher self-efficacy are more likely to leverage social support for health benefits [14]. As the mean exercise self-efficacy score for this cohort was moderate, this may further explain the lack of impact of self-efficacy on the relationship between social support and PA.

Social support may be provided through partners, friends, family, and social networks [44], and may be more important in the early phases of exercise initiation when intrinsic motivation (e.g., personal satisfaction or enjoyment) [45] is likely to be low [18]. Social support may be associated with increased PA in people with chronic pain by attenuating their pain levels [46], however, low levels of perceived social support are frequently recognised as a barrier to PA engagement in people with LBP [11, 13, 47] and chronic pain [13]. Social support is a multifaceted construct consisting of structural and functional dimensions [48]. Functional dimensions (such as receiving emotional and practical help) are most likely to correlate with positive health outcomes [49, 50], and have been measured in the current study via the DUFSS questionnaire [28]. While the current results indicate an association between general social support and PA performed within 6 months after discharge from treatment, we cannot rule out the role of PA-specific social support participants received during the study. For example, when participants reported that they exercised with another person, 40% reported exercising with their spouse, and 15% reported exercising with a friend – with these possibly representing forms of PA specific social support. The health coach involved in the health coaching program may have also provided PA specific support through, for example, setting PA specific goals with participants. This suggests there may also be a degree of PA-specific support provided, although this was not specifically assessed in this study.

4.1. Clinical implications

People with chronic LBP are generally less physically active than those without pain [51] due to their pain or physical impairment [52, 53]. People with chronic LBP are also likely to suffer comorbidities such as such as asthma, arthritis, and diabetes [54, 55, 56]. Achieving adequate levels of PA however is associated with many health benefits including decreased incidence of cardiovascular disease and cancer [57], decreased risk of falls [58], greater bone strength [3, 59] and a decreased incidence of LBP [60] as well as improved function and lower pain intensity associated with LBP [8, 48]. Research also indicates that clinical decline is common following discharge from treatment for chronic LBP and adherence to home exercise programs is poor [10, 40, 61, 62]. The current results indicate that improving social support may improve adherence to home-based PA over a 6-month period following discharge from treatment for chronic LBP. Using the margins predictions, increasing an individual’s perception of social support, may improve the number of exercise sessions they engage in from 29 to 155 over a 6-month period post discharge from conservative care. This equates to improving exercise adherence from one session per week to almost six sessions per week. Increasing an individual’s perception of social support, may also increase the total duration of exercise patients engage in from 1267 minutes to 5123 minutes. This represents an increase in their PA from an average of 48 minutes per week to an average of 197 minutes per week. The current results indicate that for people with LBP, increasing their social support could help them meet the WHO PA recommendation of aiming for 150–300 minutes of exercise per week [3].

4.2. Strengths and limitations

This study has several strengths. The results support the benefits of harnessing social support to improve exercise adherence following discharge from care for individuals with LBP and suggest further research is needed in this area. The collection of regular weekly data decreases recall bias and strengthens the results and the vast majority of participants completed 100% of their weekly diaries, with only three participants following up for less than 20 weeks. Additionally, the statistical analysis allowed us to maximise the data available with the use of over 400 data points. This study also has some limitations. Data collection for The Buddy Study was heavily impacted by the COVID-19 pandemic requiring an adjustment to original sample size calculations. Participants were also encouraged to exercise with a buddy or exercise partner; however, this was not feasible for many participants due to social distancing restrictions during the pandemic. This meant that when exercising with a buddy, participants largely exercised with their spouse or a close friend, which in fact, strengthens the importance of social support in LBP. The higher-than-normal education level of the participants and the narrow representation of nationalities may affect the generalisability of the results. However, participants in this study were representative of a group of older males and females who are likely to be impacted by chronic LBP. Future studies should include a larger sample size and as exercise adherence was self-reported by participants, in future studies this could be improved through the use of wearable devices such as a Fitbit or pedometer.

5. Conclusion

In the current study, people’s perceived level of social support was associated with an increased total number of exercise sessions and total amount of exercise performed by individuals with chronic LBP within 6 months following discharge from conservative treatment for their LBP. Adherence to advice to increase PA for LBP is usually poor, and harnessing social support to increase exercise adherence may improve long-term outcomes for those with chronic LBP.

Author contributions

KR and CGF made substantial contributions to the data acquisition and interpretation. All authors made substantial contributions to the drafted manuscript. All authors read, edited, and approved the final version of the manuscript.

Data availability

The datasets used and/or analysed are available on reasonable request however are not publicly available due to participants’ consent to their data being shared by the University of Sydney and the Sydney Local Health District.

Ethical approval

The trial was approved by the Sydney Local Health District Human Ethics Committee (2019/ETH13224).

Funding

This study was partially funded by the Sydney Local Health District. KR is a PhD student who is funded by a Sydney University low back pain scholarship. CGF was a PhD student with scholarship of Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil (Finance Code 001). The funders played no role in the design, conduct, or reporting of this study.

Informed consent

Informed consent was appropriately obtained from all participants in The Buddy Study.

Trial registration

The trial was registered at the Australian New Zealand Clinical Trial Registry (ACTRN12620001118998).

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/BMR-230239.

Conflict of interest

The authors declare that they have no competing interests.