Neck–shoulder pain and depressive symptoms: A cohort study with a 7-year follow-up

1. Introduction 

Chronic pain is associated with depression (Fishbain et al., 1997, Ohayon and Schatzberg, 2003, Mäntyselkä et al., 2003), and it is possible that pain is a predisposing factor for depression (Gureje et al., 2001, Ohayon and Schatzberg, 2003). Depression is the leading cause of disability worldwide as measured by years lived with the disability. By the year 2020, depression is estimated to achieve second place in terms of rank order of disability adjusted life years calculated for all ages of both sexes (WHO, 2009). A recent study from Finland found the prevalence of depression to be 6.5% in the population (Pirkola et al., 2005).

Associations between low back pain and depression have been found in many studies (Pincus et al., 2002, Carroll et al., 2004, Nakao and Yano, 2006, Demyttenaere et al., 2007). In general, certain somatic symptoms like joint pain, low back pain and abdominal pain have been suggested to play a role in development of depression (Hein et al., 2003, Haavisto et al., 2004, Nakao and Yano, 2006), or psychological distress (McBeth et al., 2002).

In a large Finnish population-based study, 24% of men and 37% of women aged at least 30years had suffered from neck pain during the preceding month (Kaila-Kangas, 2007) and even in adolescents neck pain has become more common in recent years (Hakala et al., 2002). Neck pain (Mäntyselkä et al., 2002), and depression (Salokangas et al., 1996) are two common symptoms of patients seeking assistance from primary care physicians. Subjects with neck pain have been postulated to be more commonly depressed than other subjects (Carroll et al., 2004, Demyttenaere et al., 2007). Most of the studies which have examined the relationship between depression and neck and shoulder pain have been cross-sectional. These studies do not clarify whether depression is a precursor or a consequence of neck and shoulder pain, or whether depression and neck and shoulder pain are comorbidities without temporal sequencing or any causal relationship. Carroll et al. (2003) found that spinal pain (neck or low back pain) is associated with onset of a new episode of depression.

We wanted to study predictors of a new onset of depressive symptoms in a 7-year population-based follow-up study. Our hypothesis was that neck and shoulder pain (neck–shoulder pain) is a predisposing factor for increased depressive symptoms. In the present study, our aim was to analyse neck and shoulder pain as a potential predicting factor for a new onset of depressive symptomatology.

2. Methods 

3. Results 

Of 604 subjects who had no depressive symptomatology at the initial baseline health check-up, 330 subjects had infrequent neck–shoulder pain and 74 subjects had daily neck–shoulder pain. A total of 77 subjects (13%) had onset of depressive symptomatology by the second health check-up. Depressive symptoms in the second health check-up were more prevalent in those subjects with daily neck–shoulder pain at the baseline (21/74, 28%) compared to those with infrequent neck–shoulder pain (37/330, 11%) or to those without neck–shoulder pain (19/200. 9.5%) (Chi2=18.9, df=2, P<0.001). A total of 86 subjects had daily neck–shoulder pain at the follow-up. Of these, 26 (30%) had depressive symptomatology at the follow-up. Of the 74 subjects who had daily neck–shoulder pain at baseline, 34 (46%) had daily neck–shoulder pain also at the follow-up (persistent neck–shoulder pain). A non-significant trend showed that onset of depressive symptomatology was more common among these subjects than among those who had daily neck–shoulder at the baseline but not at the follow-up (12/34, 35% versus 9/40, 23%, Chi2=1.48, df=1, P=0.22). The corresponding proportions were 9% (42/478) for those who had neck–shoulder pain neither at the baseline nor at the follow-up, and 27% (14/42) for those who had daily neck–shoulder pain only at the follow-up. At the follow-up the prevalence of daily neck–shoulder pain was 34% (26/77) among subjects with depressive symptomatology and 11% (60/527) among those without depressive symptomatology (Chi2=27.6, df=1, P<0.001).

The subjects with depressive symptomatology in the follow-up were slightly older and used slightly less often alcohol than those without depressive symptoms (Table 1). When adjusted for age and gender, the neck–shoulder pain status at the baseline was strongly associated with depressive symptomatology at the follow-up (P=0.009).

Table 1. Subjects without depressive symptomatology at the baseline (N=604). Baseline characteristics in subjects without and with depressive symptomatology at the follow-up.

		No depressive symptomatology at the follow-up (BDI-score<10), N=527	Depressive symptomatology present at the follow- up (BDI-score⩾10), N=77	P valuea
	Females, N (%)	288 (55)	48 (62)	0.21
	Age, mean (SD)	46 (6)	48 (5)	0.012
	BMI, mean (SD)	26.3 (4.3)	27.2 (3.9)	0.15
	Marital status, cohabiting, N (%)	449 (85)	66(86)	0.97
	Education, primary school or lower N (%)	270 (51)	47 (61.0)	0.58
	Working, N (%)	460 (87)	62 (81)	0.32
	Alcohol use, N (%)	445 (84)	55 (71)	0.013
	Smoking, N (%)	131 (25)	14 (18)	0.36
	Exercise three times a week, N (%)	151 (29)	23 (30)	0.86
	BDI-score at baseline, mean (SD)	2.4 (2.5)	4.6 (2.9)	<0.001
	Other musculoskeletal symptoms	66 (13)	14 (18)	0.25
	Neck–shoulder pain N, (%)			0.009
	None	181 (34)	19 (25)
	Infrequently	293 (56)	37 (48)
	Daily	53 (10)	21 (27)

a Age and sex adjusted, except females and age.

Table 2 shows the results of the univariate and multivariate logistic regression analysis. Neck–shoulder pain status, age, alcohol use (inverse association) and BDI-score (0–9) at the baseline were associated with the onset of depressive symptomatology. In the multivariate model, neck–shoulder pain, alcohol use (inverse association) and BDI-score at the baseline were associated with the new onset of depressive symptomatology at the follow-up. Compared to subjects without neck–shoulder pain, subjects with daily neck–shoulder pain had almost threefold odds (OR, 2.64, 95% CI, 1.27–5.48) for the new onset of depressive symptomatology. The BDI-score itself at the baseline was also strongly associated (OR, 1.32, 95% CI, 1.21–1.44) with depressive symptoms at the follow-up. The baseline alcohol use was inversely associated (OR 0.46, 95% CI, 0.27–0.80) with depressive symptoms at the follow-up.

Table 2. Univariate and multivariate logistic regression analysis for depressive symptomatology at the follow-up.

	Variable	Univariate		Multivariate (a)
		OR (95% CI)	P value	OR (95% CI)	P value
	Neck–shoulder pain
	No	1 (referenceb)	0.004c	1 (referenceb)	0.019c
	Infrequently	1.20 (0.67–2.16)		1.02 (0.55–1.87)
	Daily	3.77 (1.89–7.54)		2.64 (1.27–5.48)
	Gender
	Male	1 (referenceb)
	Female	1.37 (0.84–2.25)	0.21
	Age, years	1.05 (1.01–1.10)	0.013
	BMI, kg/m2
	<30	1 (referenceb)
	⩾30	1.52 (0.86–2.71)	0.15
	Marital status
	Not cohabiting	1 (referenceb)
	Cohabiting	1.04 (0.53–2.06)	0.91
	Education
	Primary	1 (referenceb)	0.14c
	Secondary	0.68 (0.38 to1.22)
	Tertiary	0.65 (0.33–1.27)
	Working status
	Not working	1 (referenceb)
	Working	0.60 (0.32–1.12)	0.11
	Alcohol used
	No	1 (referenceb)		1 (referenceb)
	Yes	0.46 (0.27–0.80)	0.006	0.50 (0.28–0.89)	0.015
	Smoking
	No	1 (referenceb)
	Yes	0.67 (0.36–1.24)	0.20
	Exercise, units per week
	<3	1 (referenceb)
	⩾3	1.06 (0.63–1.79)	0.83
	Other musculoskeletal symptoms
	Not present	1 (referenceb)
	Present	1.55 (0.82–2.93)	0.15
	BDI-score (0–9) at baseline	1.32 (1.21–1.44)	<0.001	1.31 (1.20–1.43)	<0.001

a Forward selection. Only those variables are shown which entered the model. b Denominator (reference group) of following odds ratios. c P for linearity. d Alcohol use in the preceding 12months.

4. Discussion 

Our study highlights that frequent neck–shoulder pain is a potential and clinically important predisposing factor of depression. This relationship was independent from other covariates (alcohol use and BDI-score at the baseline) included in the final multivariate analysis. The results of the present study revealed almost a three fold odds for the new onset of depressive symptomatology over 7 years in subjects with daily neck–shoulder pain compared to those without neck–shoulder pain at the baseline.

A recent Finnish population-based study showed a clear association of low back pain with neck pain (Kääriä et al., 2009). Subjects with low back tend to be more depressed than other subjects (e.g. Pincus et al., 2002). Hence it could be possible that the co-occurrence of other types of musculoskeletal disturbances might explain the occurrence of depression. However, in the present study the association between neck–shoulder pain and depressive symptomatology remained statistically significant even after including other musculoskeletal complaints into the multivariate analysis.

This study showed that frequent neck–shoulder pain tends to be persistent. Almost half of those subjects who had frequent neck pain at the baseline also had it at the follow-up. These subjects seemed to have more often depressive symptomatology at the follow-up than those subjects who had neck–shoulder pain at the baseline but not at the follow-up although this difference was not statistically significant. It is possible that daily pain increases concurrent depressive symptoms. This may on a long-term basis enhance the odds of depressive symptomatology. We suggest that frequent neck–shoulder pain is a predisposing factor for the depressive symptomatology but we cannot explicitly state that frequent neck–shoulder pain is a causative factor for depressive symptomatology.

The previous studies have found an association between neck pain and psychological distress (with or without depression) (McBeth et al., 2002, van der Windt et al., 2002). In our study, BDI-score at the baseline in the non-depressive individuals (i.e. BDI-score<10) was associated with the onset of depressive symptomatology. Slightly elevated BDI-score at the baseline could have been related to psychological distress which is defined as a reaction to external or internal stress with various symptoms, such as sadness, anxiety, and psychophysiological symptoms (Dohrenwend et al., 1980). Therefore it may be possible that psychological distress (without depressive symptomatology) may be a predisposing factor for both neck–shoulder pain and onset of depression. Stress and impaired function of stress related hormones are associated with both depression (Holsboer and Ising, 2008) and chronic pain (McBeth et al., 2007). Thus, stress related dysfunction may partly explain the relationship between neck–shoulder pain and depressive symptoms in the present study. The slightly elevated depression score at the baseline may also indicate a possible depressive trait in these individuals, which may result in a more manifest depressive symptomatology at the follow-up. However, after controlling for baseline BDI-score in the multivariate analysis the association between baseline daily neck–shoulder pain and depressive symptomatology was still significant.

Alcohol use was inversely associated with the onset of depressive symptomatology which is slightly surprising. However, we were not able to distinguish the excessive alcohol use from the non-excessive use. Therefore the present result may mean that only low or moderate alcohol consumption is associated with decreased risk of developing depressive symptoms.

Definition of the depressive symptomatology was based on BDI-score obtained from the self-report. We did not undertake a diagnostic interview, which may be seen as a limitation of our study. However, BDI with a cut-off score of ten points has been shown to represent a feasible instrument for depression detection (Steer et al., 1999, Timonen et al., 2005). The period between the health check-ups was 7 years. Therefore we were not able to detect all depressive episodes occurring during these years. Also the assessment of neck–shoulder pain was based on the self-report and we did not assess more deeply the diagnostic features of these symptoms, which also is a potential limitation of our study. The subject participating in the follow-up had more often daily neck–shoulder pain but had lower BDI-score than those who were lost to follow-up. This can indicate a potential non-participating bias. However, this may not affect the results of this study because the study cohort consisted of those subjects who did not have depressive symptomatology at the baseline. In general, people with depressive symptoms may be less willing to participate in surveys. The study population was from a limited area in Finland which may affect the possibility of generalizing these results to the whole population. Future research should strengthen our results in other populations.

We suggest that frequent neck–shoulder pain predicts future depressive symptoms. In particular, patients with persistent non-specific neck–shoulder pain could benefit from screening for depression.