The influence of multisite pain and psychological comorbidity on prognosis of chronic low back pain: longitudinal data from the Norwegian HUNT Study.

OBJECTIVES: This study aimed to investigate the prospective influence of multisite pain, depression, anxiety, self-rated health and pain-related disability on recovery from chronic low back pain (LBP).
SETTING: The data is derived from the second (1995-1997) and third (2006-2008) wave of the Nord-Trøndelag Health Study (HUNT) in Norway. PARTICIPANTS: The study population comprises 4484 women and 3039 men in the Norwegian HUNT Study who reported chronic LBP at baseline in 1995-1997. PRIMARY OUTCOME MEASURES: The primary outcome was recovery from chronic LBP at the 11-year follow-up. Persons not reporting pain and/or stiffness for at least three consecutive months during the last year were defined as recovered. A Poisson regression model was used to estimate adjusted risk ratios (RRs) with 95% CIs.
RESULTS: At follow-up, 1822 (40.6%) women and 1578 (51.9%) men reported recovery from chronic LBP. The probability of recovery was inversely associated with number of pain sites (P-trend<0.001). Compared with reporting 2-3 pain sites, persons with only LBP had a slightly higher probability of recovery (RR 1.10, 95% CI 0.98 to 1.22 in women and RR 1.10, 95% CI 1.01 to 1.21 in men), whereas people reporting 6-9 pain sites had substantially lower probability of recovery (RR 0.58, 95% CI 0.52 to 0.63 in women and RR 0.70, 95% CI 0.63 to 0.79 in men). Poor/not so good self-rated general health, symptoms of anxiety and depression, and pain-related disability in work and leisure were all associated with reduced probability of recovery, but there was no statistical interaction between multisite pain and these comorbidities.
CONCLUSIONS: Increasing number of pain sites was inversely associated with recovery from chronic LBP. In addition, factors such as poor self-rated health, psychological symptoms and pain-related disability may further reduce the probability of recovery from chronic LBP. © Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2017. All rights reserved. No commercial use is permitted unless otherwise expressly granted.

The strengths of the current study are the large and unselected population of women and men with chronic low back pain (LBP), the prospective design and the possibility of adjusting for several potential confounding factors.

A limitation is the lack of information about the course of LBP and the other variables between the Nord-Trøndelag Health 2 (HUNT2) and HUNT3 Study.

Furthermore, we cannot rule out that changes in lifestyle differed between those who experienced remission of symptoms and those who did not, for example, individuals with a high number of pain sites at baseline could be less prone to adopt a healthy lifestyle during the follow-up period because of the pain-related disability.

Introduction

Low back pain (LBP) is a common cause of disability and is ranked as the most burdensome disease globally.1 2 LBP is the fourth most common diagnosis (after upper respiratory infection, hypertension and coughing) seen in primary care3 and approximately every fifth adult suffers from chronic LBP.4 Thus, in addition to the suffering experienced by affected individuals, LBP represents a substantial economic burden to the society. This underscore the importance of increased knowledge about factors that can improve the prevention and management of chronic LBP.

Chronic LBP rarely exists as a separate entity and co-occurrence of multisite pain and other comorbidities are common.5–9 A large case-control study comprising more than 1 00 000 people showed that individuals with chronic LBP had higher occurrence of other musculoskeletal conditions, depression, anxiety and sleep disorders compared with controls without chronic LBP.10 In particular, other chronic pain conditions are very prevalent among people with chronic LBP.5 Number of pain sites by itself has been suggested to be dose-dependently related to reduced physical and mental function11 12 and there are data to support the notion that generalised pain differs markedly from conditions with only one or a few pain sites with respect to other risk factors.13 Currently, there is a lack of longitudinal studies addressing how the extent of multisite pain influences the prognosis of chronic LBP. Moreover, it is unclear to what extent multisite pain interacts with other comorbid factors such as poor self-rated general health, pain-related disability and poor mental health to influence the prognosis of chronic LBP.

The main objective of this study was therefore to prospectively investigate the influence of common somatic and psychological comorbidities on prognosis of chronic LBP. We hypothesised (1) that multisite chronic pain, poor self-rated general health, pain-related disability and poor psychological health are factors that are inversely and independently related to the probability of recovery from chronic LBP and (2) that the possible association between number of pain sites and prognosis of LBP is modified by other somatic and psychological comorbidities.

Methods

Study population

In Nord-Trøndelag county, Norway, all inhabitants aged 20 years or older were invited to participate in three health surveys (the Nord-Trøndelag Health Study (the HUNT Study)), t the second in 1995–1997 (HUNT2) and the last in 2006–2008 (HUNT3). The current study is based on data from HUNT2 and HUNT3. Of 93 898 eligible participants, 65 237 (65.5%) accepted the invitation to participate in HUNT2. In HUNT3, a total of 93 860 participants were invited, and 50 807 (54.1%) accepted the invitation. More detailed information about selection procedures, participation and questionnaires used in the HUNT Study can be found at http://www.ntnu.edu/hunt.

Information on lifestyle and health-related factors were collected by questionnaires and a clinical examination at both HUNT2 and HUNT3. For the purpose of this study, we included data from the 37 070 people who participated at both surveys. We excluded 15 062 women and 12 861 men who reported to be free from chronic LBP at HUNT2. Moreover, we excluded 1557 persons with missing information on musculoskeletal pain at HUNT3 and 23 persons without complete values on body mass index (BMI) from the clinical examination. Further, 44 persons defined as underweight (BMI <18.5 kg/m2) were additionally excluded from the analyses to reduce the possibility for reverse causation due to undetected disease. Thus, the prospective analyses were based on 4484 women and 3039 men. Each participant signed a written consent, and the study was approved by the Regional Committee for Ethics in Medical Research (project no. 2014/2044 REK midt, Norway).

Chronic LBP

The questions about musculoskeletal pain were adopted from the Standardised Nordic Questionnaire.14 The participants were asked “During the last year, have you had pain and/or stiffness in your muscles and joints that lasted for at least three consecutive months?” Response options were ‘yes’ and ‘no’. If the participants were asked to indicate the affected body area(s), that is, up to nine body areas (neck, shoulders/upper arms, upper back, elbows, low back, wrists/hands, hips, knees and ankles/feet). Chronic LBP was in both surveys defined by ‘yes’ to the first question and low back indicated as an affected body area by the second question. Persons who responded ‘yes’ to the first question but did not indicate low back as an affected body area were considered to be free from chronic LBP. Number of chronic pain sites were estimated by adding together pain-afflicted body areas, of which the total number of pain sites includes LBP. The primary outcome was recovery from chronic LBP at the 11-year follow-up. Persons categorised with chronic LBP at HUNT2 responding ‘no’ at HUNT3 to the question “During the last year, have you had pain and/or stiffness in your muscles and joints that lasted for at least three consecutive months?” were defined as recovered.

Pain-related comorbidities

The participants’ self-rated general health was evaluated using the question “How is your health at the moment?”, with response options ‘very good’, ‘good’, ‘not so good’ and ‘poor’. The answers were dichotomised into two groups: ‘very good/good’ and ‘not so good/poor’ in line with previous studies.15

Pain-related disability was evaluated separately for work ability and leisure time activity. The question about work ability was: “Have the pain and/or stiffness reduced your ability to work during the last year?” with four possible responses: ‘no’, ‘not significantly’, ‘to some degree’, ‘significantly’ and ‘don’t know’. The first and last response options were merged and categorised as ‘no disability’, and the two middle categories as ‘work disability’. For leisure time activity, the question was: “Have the pain/or stiffness reduced your leisure activity?” with possible responses: ‘yes’ and ‘no’. The responses on disability due to musculoskeletal symptoms were then categorised into four groups: ‘no disability’, ‘work disability’, ‘leisure disability’ and ‘work and leisure disability’.

Symptoms of anxiety and depression were measured using the Hospital Anxiety and Depression Scale (HADS). HADS is a validated and well-established self-rating questionnaire including seven questions on anxiety and seven questions on depression.16 score value was set to ≥8 on both anxiety and depression and were dichotomised as presence or no presence of anxiety and/or depression.16 17 In addition, a mixed HADS variable was constructed consisting of four groups: ‘no depression or anxiety’, ‘only depression’, ‘only anxiety’ and ‘both depression and anxiety’.18 Symptoms of only depression or only anxiety were defined by a HADS score ≥8 on the respective subscales, whereas symptoms of both depression and anxiety were defined by a HADS score ≥8 on both subscales.

Possible confounders

All estimated associations were adjusted for possible confounders. Age was categorised in

Statistical analysis

We used a generalised linear model of the Poisson family to estimate the relative probability of recovery from chronic LBP as risk ratios (RRs) w.0 indicates higher probability of recovery compared with the reference category, whereas a RR <1.0 indicates a reduced probability of recovery. All estimated associations were adjusted for age, BMI, physical activity, education, smoking and physical work demands. All main analyses were conducted separately for men and women. Furthermore, a test for linear trend (ie, dose response) across categories of number of pain sites was conducted by treating the categories as an ordinal variable in the regression model.

In addition, we conducted analyses combining number of pain sites (<4 vs 4–9 sites) and comorbid conditions in relation to the probability of recovery from chronic LBP. Previous studies have shown that reporting of ≥4 pain sites is associated with a markedly poorer prognosis of pain relief,20 as well as increasing likelihood of healthcare utilisation and sickness absence.21 Statistical interaction was evaluated by likelihood ratio tests of a product term of number of pain sites and each of the comorbid factors (self-reported health, pain-related disability and HADS). All statistical analyses were performed using Stata for Windows V.13.1.

Results

Table 1 presents the baseline characteristics of the study population according to number of chronic pain sites. At baseline, 66.4% of the women and 47.2% of the men reported ≥4 pain sites. Of the 4484 women and 3039 men who reported chronic LBP at baseline (HUNT2), 1822 (40.6%) women and 1578 (51.9%) men were reported recovered from chronic LBP at the 11-year follow-up (HUNT3).

Table 1

Baseline characteristics of the study population stratified by gender and number of chronic pain sites

	Women		Men
	<4 pain sites	4–9 pain sites	<4 pain sites	4–9 pain sites
No of persons (%)	1506 (33.6)	2978 (66.4)	1605 (52.8)	1434 (47.2)
Age (years), mean (SD)	47.9 (13.6)	50.7 (11.9)	48.4 (12.1)	51.8 (11.4)
Body mass index (kg/m2), mean (SD)	26.1 (4.1)	27.0 (4.5)	26.5 (3.3)	27.0 (3.4)
Physically inactive, n (%)	82 (5.4)	208 (7.0)	96 (6.0)	103 (7.2)
Education ≤13 years, n (%)	1142 (75.8)	2470 (82.9)	1244 (77.5)	1220 (85.1)
Current smoker, n (%)	427 (28.4)	1021 (34.3)	416 (25.9)	412 (28.7)
Poor/not so good self-rated health, n (%)	443 (29.4)	1786 (60.0)	461 (28.7)	831 (57.9)
Pain-related disability, work and leisure, n (%)	726 (48.2)	2034 (68.3)	784 (48.8)	970 (67.6)
HADS score depression >8, n (%)	65 (4.3)	187 (6.3)	96 (6.0)	124 (8.6)
HADS score anxiety >8, n (%)	149 (9.9)	425 (14.3)	110 (6.9)	147 (10.3)

HADS, Hospital Anxiety and Depression Scale.

Table 2 shows the association between number of pain sites, pain-related disability, psychological symptoms and self-rated general health with the probability of recovery from chronic LBP at follow-up. Increasing number of pain sites was inversely associated with the probability of recovery (P-trend <0.001 in both women and men). In specific, women and men who reported 6–9 pain sites had substantially lower probability of recovery (RR 0.58, 95% CI 0.52 to 0.63 and RR 0.70, 95% CI 0.63 to 0.79, respectively), compared with women and men who reported 2–3 pain sites. People with only LBP had a slightly higher probability of recovery (RR 1.10, 95% CI 0.98 to 1.22 in women and RR 1.10, 95% CI 1.01 to 1.21 in men) compared with women and men who reported 2–3 pain sites. Pain-related disability that influenced both work ability and leisure activity was associated with reduced probability of recovery in both women (RR 0.68, 95% CI 0.62 to 0.74)) and men (RR 0.76, 95% CI 0.70 to 0.83). HADS score ≥8 on both depression and anxiety subscales was associated with reduced probability of recovery in both women (RR 0.77, 95% CI 0.66 to 0.91) and men (RR 0.79, .67 to 0.94). Persons reporting poor or not so good general health had a markedly reduced probability of recovery, both in women (RR 0.66, 95% CI 0.61 to 0.71) and men (RR 0.72, 95% CI 0.67 to 0.78), compared with those reporting good or very good general health.

Table 2

Relative probability of recovery from chronic low back pain at 11-ow-up according to number of chronic pain sites, pain-related disability, the HADS score and self-rated general health at HUNT2

	Women				Men
	No of persons	No of cases	Crude RR	Multi adjusted* RR (95% CI)	No of persons	No of cases	Crude RR	Multi adjusted* RR (95% CI)
No of pain sites
1	326	189	1.13	1.10 (0.98 to 1.22)	454	284	1.11	1.10 (1.01 to 1.21)
2–3	1180	608	1.00	1.00 (reference)	1151	651	1.00	1.00 (reference)
4–5	1330	557	0.81	0.83 (0.76 to 0.90)	873	422	0.85	0.86 (0.79 to 0.94)
6–9	1648	468	0.55	0.58 (0.52 to 0.63)	561	221	0.70	0.70 (0.63 to 0.79)
Pain-related disability
No disability	649	355	1.00	1.00 (reference)	487	304	1.00	1.00 (reference)
Work disability	591	271	0.84	0.87 (0.78 to 0.98)	430	249	0.93	0.94 (0.85 to 1.05)
Leisure disability	257	131	0.93	0.94 (0.82 to 1.08)	250	143	0.92	0.90 (0.79 to 1.02)
Work and leisure disability	2760	964	0.64	0.68 (0.62 to 0.74)	1754	818	0.75	0.76 (0.70 to 0.83)
HADS
No depression or anxiety	2572	1103	1.00	1.00 (reference)	2050	1110	1.00	1.00 (reference)
Depression	252	90	0.83	0.90 (0.76 to 1.07)	220	101	0.85	0.85 (0.73 to 0.99)
Anxiety	574	215	0.87	0.88 (0.78 to 0.98)	257	120	0.86	0.88 (0.77 to 1.01)
Depression and anxiety	342	107	0.73	0.77 (0.66 to 0.91)	195	80	0.76	0.79 (0.67 to 0.94)
Self-rated general health
Very good/good	2208	1099	1.00	1.00 (reference)	1730	1017	1.00	1.00 (reference)
Poor/not so good	2229	704	0.64	0.66 (0.61 to 0.71)	1292	551	0.73	0.72 (0.67 to 0.78)

*Adjusted for age (19–29, 30–39, 40–49, 50–59 and >60 years), education (primary school and lower secondary school, upper secondary school, higher education <4 years, higher education >4 years and unknown), body mass index (normal weight, overweight and obesity), physical activity (inactive, low activity, moderate activity, high activity and unknown), smoking (never-smoker, previous smoker, current smoker and unknown) and physical work demands (mostly sedentary, much walking, much walking and lifting, heavy physical work and unknown).

HADS, Hospital Anxiety and Depression Scale; HUNT, Nord-Trøndelag Health Study; RR, risk ratio.

Table 3 presents the combined effect of number of pain sites and pain-related disability, psychological symptoms and self-rated general health on the probability of recovering for chronic LBP. We did not observe any statistical interaction between number of pain sites and pain-related disability, psychological symptoms or self-rated health (p≥0.24 for all tests). However, stratified analysis within categories of the exposure variables showed that reporting ≥4 pain sites was associated with lower probability of recovery independently of level of pain-related disability and psychological symptoms. Within strata of pain-related disability, persons who reported ≥4 pain sites had 16%–27% lower probability of remission compared with persons with 1–3 pain sites in the same pain-related disability categories. Likewise, within the different strata of psychological symptoms, persons with ≥4 pain sites had 25%–35% lower probability of recovery compared with persons with 1–3 pain sites.

Table 3

Relative probability of recovery from chronic low back pain at 11-year follow-up according to the combined effect of number of chronic pain sites and pain-related disability, score on the HADS and self-rated general health at HUNT2

	1–3 pain sites			4–9 pain sites
	No of persons	No of cases	Multi adjusted* RR (95% CI)	No of persons	No of cases	Multi adjusted* RR (95% CI)	p Value†
Pain-related disability
No disability	714	448	1.00 (reference)	422	211	0.84 (0.75 to 0.94)	0.002
Work disability	466	270	0.94 (0.85 to 1.03)	555	256	0.77 (0.69 to 0.86)	0.002
Leisure disability	271	172	0.98 (0.88 to 1.09)	236	102	0.71 (0.61 to 0.83)	<0.001
Work and leisure disability	1510	756	0.81 (0.75 to 0.87)	3004	1026	0.59 (0.54 to 0.64)	<0.001
HADS
No depression or anxiety	2151	1225	1.00 (reference)	2471	988	0.75 (0.70 to 0.80)	<0.001
Depression	161	84	0.92 (0.79 to 1.07)	311	107	0.66 (0.56 to 0.77)	0.002
Anxiety	259	142	1.00 (0.89 to 1.12)	572	193	0.65 (0.58 to 0.73)	<0.001
Depression and anxiety	138	69	0.92 (0.77 to 1.09)	399	118	0.58 (0.49 to 0.68)	<0.001
Self-rated general health
Very good/good	2187	1302	1.00 (reference)	2617	838	0.57 (0.53 to 0.61)	<0.001
Not all good/poor	904	417	0.78 (0.72 to 0.85)	1751	814	0.82 (0.77 to 0.87)	0.301

*Adjusted for age (19–29, 30–39, 40–49, 50–59 and >60 years), education (primary school and lower secondary school, upper secondary school, higher education <4 years, higher education >4 years and unknown), body mass index (normal weight, overweight and obesity), physical activity (inactive, low activity, moderate activity, high activity and unknown), gender, smoking (never-smoker, previous smoker, current smoker and unknown) and physical work demands (mostly sedentary, much walking, much walking and lifting, heavy physical work and unknown).

†p Value from stratified analysis of number of pain sites by general health, physical disability and HADS score.

HADS, Hospital Anxiety and Depression Scale; HUNT, Nord-Trøndelag Health Study; RR, risk ratio.

Discussion

In this large population-based study, we found that musculoskeletal comorbidity, reduced self-rated general health and psychological symptoms were independently associated with reduced probability of recovery from chronic LBP at 11-year follow-up. The factors with the strongest association with poor prognosis were widespread chronic pain (6–9 pain sites) and poor or not so good self-rated general health. The strength of the associations between the various comorbidities and pain prognosis was fairly similar for women and men. Probability of recovery from chronic LBP was inversely associated with increasing number of chronic pain sites. Although there was no interaction between number of chronic pain sites and other comorbidities, we observed in the combined analysis that persons with ≥4 pain sites were associated with lower probability of recovery from chronic LBP within all strata of pain-related disability and symptoms of depression and/or anxiety. The current findings indicate that musculoskeletal comorbidity has a strong and independent influence on long-term prognosis of chronic LBP.

It is noteworthy that about 66% of the women and 47% of the men in this study reported ≥4 chronic pain sites at baseline. This supports the view that co-occurrence of musculoskeletal pain is very common in chronic LBP.5 6 To our knowledge, this is the first population-based study to investigate the prospective influence of graded musculoskeletal comorbidity on the prognosis of chronic LBP. The dose–response association between number of chronic pain sites and reduced probability of recovery from chronic LBP suggests that musculoskeletal comorbidity should be considered an important predictor in identifying target groups for public health secondary prevention. This was also supported by our combined analysis, showing that number of pain sites was the main driving factor for predicting persistence of chronic LBP.

More than 40% of the women and 50% of the men in the current study reported recovery from chronic LBP at 11-year follow-up. Interestingly, a previous study showed that the prevalence of chronic LBP was relatively stable from HUNT2 to HUNT3 with about 26% of women and 20% of men reporting chronic LBP at both surveys.22 Thus, our results indicate that during an 11- year period a substantial proportion of the population shift from having chronic LBP to remission, but that a substantial proportion also develops pain in the same period. Similar large fluctuations in reporting of chronic LBP at the individual level have also been observed by others.23 24 Thus, our findings lend further support to the notion that chronic LBP on the individual level may fluctuate substantially over time e population prevalence remains relatively stable. The current study adds to this knowledge by showing that individuals who shift from having chronic LBP symptoms to remission of symptoms are more likely to have fewer chronic pain sites, less pain-related disability, better self-rated health and no major symptoms of anxiety or depression.

Increasing number of chronic pain sites were inversely associated with probability of recovery, that is, women and men who reported ≥6 pain sites had about 30%–40% lower probability of recovery from chronic LBP compared with women and men with 2–3 pain sites. Previous cross-sectional studies have indicated a dose–response association between number of pain sites and a range of negative health outcomes such as psychological distress, poor sleep, poor self-rated health, reduced social and functional ability,11 as well as increased sickness absence and healthcare utilisation.25The current prospective study extends this body of knowledge showing that number of chronic pain sites have a strong dose–response influence on prognosis of chronic LBP. Although we observed no interaction between number of chronic pain sites and other comorbid factors, the probability of relief from chronic LBP was consistently lower for the group with multisite pain within all strata of pain-related disability and psychological symptoms scores. These findings support the long-held view that it may be useful to classify patients with chronic LBP into ‘back pain alone’ or ‘back pain plus other pain’ to improve clinical decision-making.26

The current finding of a dose–response association between number of chronic pain sites and prognosis of chronic LBP may indicate that the extent of musculoskeletal comorbidity could provide additional complementary information to improve classification in stratified care approaches. The idea that assessment of multisite pain can assist clinical judgement of prognosis and improve targeted treatment has been proposed before,6 and the current data lend further support to this idea. Furthermore, since number of chronic pain sites per se seem to be a strong prognostic factor in chronic LBP it may also be useful to consider this variable when recruiting subjects into research studies to facilitate baseline comparisons.

ious data indicate that psychological symptoms are more common in patients with LBP than in comparable controls,10 our results do not indicate that such symptoms strongly influence the prognosis of chronic LBP. However, another study of subjects with neck and/or LBP in HUNT3 showed that symptoms of mental distress were significant determinants for seeking healthcare, which could have moderated the associations.27 Our findings are in line with Dunn et al 28 who found no significant association between depression, and only a modest association between anxiety, and the risk of disabling LBP at 12 months follow-up in patients presenting with LBP in general practice. In the same study, it was observed that self-rated health had a relatively strong impact on prognosis of LBP with patients who rated their health as poor having more than twofold increased risk of disabling back pain. Very few individuals in our study population rated their health as poor and we were, therefore, not able to estimate probability for recovery among these individuals. However, we observed that women and men who rated their health as less than good (ie, poor or not so good) had about 30% lower probability of recovery from chronic LBP compared with those who rated their health as good or very good.

The strengths of the current study are the large and unselected population of women and men with chronic LBP, the prospective design and the possibility of adjusting for several potential confounding factors. The questions on chronic musculoskeletal pain used in HUNT2 have acceptable reliability and validity.14 29 30 Likewise, the HADS Scale has been shown to be at a valid indicator of possible depression and anxiety in clinical practice as well as in the general population.16 17 31 A limitation is the lack of follow-up information about the course of LBP and the other variables between the HUNT2 and HUNT3 Study. Thus, any changes occurring during the follow-up period could not be taken into account in the analyses. For example, information regarding treatment during the follow-up period or information on changes in lifestyle could be of interest. A healthy lifestyle has been associated with improved long-term outcome in individuals with recurrent LBP episodes.32 Thus, it may be possible that individuals who changed their lifestyle during the follow-up period also altered their course of chronic LBP. Furthermore, we cannot rule out that such changes in lifestyle were differential between participants who experienced remission of symptoms versus those who did not, for example, individuals with a high number of pain sites at baseline could be less prone to adopt a healthy lifestyle during the follow-up period because of pain-related disability.

In conclusion, the current study indicates that multisite chronic pain is independently and inversely associated with the probability of recovery from chronic LBP. Poor self-rated health, psychological symptoms and pain-related disability might further reduce the probability of recovery from chronic LBP. There was no interaction between number of chronic pain sites and other comorbidities, including pain-related disability, psychological symptoms and self-rated general health. These findings underscore the importance of taking comorbid symptoms into account, and in particular number of chronic pain sites, when designing management programmes or treatment for secondary prevention of chronic LBP.