Depression augments activity-related pain in women but not in men with chronic musculoskeletal conditions

OBJECTIVES: The primary objective of the present study was to examine the role of sex as a moderator of the relation between depression and activity-related pain. METHODS: The study sample consisted of 83 participants (42 women, 41 men) with musculoskeletal conditions. Participants were askeJ to lift a series of 18 canisters that varied in weight (2.9 kg, 3.4 kg and 3.9 kg) anJ Jistance from the body. Participants were askeJ to rate their pain while they lifteJ each canister and estimate the weight of the canisters. RESULTS: Consistent with previous research, the relations among Jepression, pain intensity anJ disability were stronger for women than for men. ANOVA revealed. that depression was associated with more imense activity-relateJ pain in women only. For b0th women anJ men, the intensity of pain increaseJ with each trial, although the weight of the objects lifted remained constant. Neither sex nor depression had an effect on participants' weight estimates. CONCLUSIONS: The present Jiscussion aJdresses the mecha­ nisms through which Jepression may Jifferentially affect pain in women and men. It also addresses the potential clinicaI implications of pain-augmenting effects of Jepression in women.

S urveys indicate that approximately 20% to 50% of indi-depressive symptoms and tehabilitation outcomes.Schultz et al viduals with musculoskeletal conqitions show evidence of (14) teported that levels of deptession in tecently injured wotk elevated depressive symptoms (1)(2)(3)(4).Recent studies suggest ets wete not significant predictots of follow-up return-to-work that deptessive symptoms associated with musculoskeletal status.At the time of writing, thete have been few efforts to disotdets may increase the tisk fot prolonged work disability explore the factors that determine when depression will be a (5)(6)(7)(8).Individuals with pain-telated musculoskeletal condi-significant determinant of pain and disabilitY outcomes.tions and elevated depressive symptoms have a sick leave There is a large body of literature indicating that women duration twice as long as individuals with musculoskeletal are more likely to show elevated symptoms of depression conditions who do not have depressive symptoms (9,10).(15,16).Women are not only at greater risk for developing Depressive symptoms in individuals with musculoskeletal depression, they are also at greater risk for relapse following conditions have also been associated with longer duration of treatment (17).Numerous investigations have also shown that wage replacement benefits following work injury or surgical the prevalence of pain-telated musculoskeletal conditions is intervention (11,12).
highet in women (18)(19)(20)(21).Reseatch suggests that women ate at Nevertheless, not all studies found a telation between a highet tisk for prolonged disability following the onset of a deptession and recovery outcomes.Fot example, in a study musculoskeletal condition (18)(19)(20)(21).These data suggest that the examining the outcome of tehabilitation treatment in pain burden of disability associated with pain and depression may be patients, Gatchel et al (13) teported no telation between gteater for women.
There are indications that sex may be an important moder ator of the relationship between depression and disability (22).In a recent study of chronic pain patients, Keogh et al (23) found that depression was associated with increased disability in women, but not in men.Based on their findings, the authors suggested that depression may leave women more vu,lnerable to disability.Keogh et al (24) also suggested that their findings may explain why women are at greater risk for the develop ment of pain-related problems.
Albeit intriguing, the findings of Keogh et al (24) do not point to the processes that would link depression and disability in women, but not in men.The absence of information about the processes that differentially link depression and disability in women and men compromises efforts to develop rehabilita tion interventions specifically targeting the determinants of disability in women and men.Indeed, there are indications that treatment gains following multidiSciplinary pain programs are less likely to be maintained by women than by men (23).
The primary objective of the present study was to explore the processes that may differentially link depre.ssionand dis ability in women and men.One of the processes by which depression may affect pain-related disability is by accentuating the negative impact of the activity-related pain symptoms (25).Because disability is typically construed as a reduction in activity participation, measures of activity-related pain, instead of spontaneous or condition-related pain, may be the most relevant pain variables to assess.To our knowledge, the relations among sex, depression and activity-related pain have not been previously studied.
In the present study, women and men with chronic muscu loskeletal conditions were asked to lift a series of canisters that varied according to weight and distance from the body.Participants were asked to rate their pain while they lifted each canister and to estimate the weight of the canisters they lifted.In this manner, it was pOSSible to examine whether sex differ ences in the relation between depression and disability could be accounted for by differences in the perceived effort associ ated with the lifting task.Secondary analyses were conducted to assess whether pain ratings and weight estimates changed over repeated trials, and whether these changes varied as a function of sex.

Participants
The study sample consisted of 83 participants (42 women, 41 men) wi th musculoskeletal conditions.All participants were referred for assessment at one of three pain treatment centres in Montreal, Quebec.A t the time of the assessment, all participants were work-disabled due to their musculoskeletal condition and were receiving compensation.The mean age of the sample was 40.8 years, with a range of 20 to 58 years.The mean (± SO) number of years spent in pain was 6.8±6.6 years.All participants underwent a medical evaluation to ensure there were no medical contraindications to performing the physical manoeuvres involved in the lifting task.

Measures
Depression: The Seck Depression Inventory II (SOl-II) (26) was used to measure severity of depressive symptoms.The BOI-ii consists of 21 items describing various symptoms of depression.Respondents are asked to endorse phrases that best describe how they felt during the past two weeks.The BDI-II Pain Res Manage vor 13 No 3 May/June 2008 has been shown to be a reliable and vaLid index of depressive symptoms in chronic pain patients (5,27,28).
Participants were classified as 'high' or 'low' depression based on their BDI-II scores.Participants who scored nine or less on the SOl-II were assigned to the low depression group; participants who scored 16 or greater on the BDl-1i were assigned to the high depression group.Participants with SDI-II scores between 10 and IS, inclusive, were not included in the analyses.Extreme groups were used to maximize the clinical relevance of the findings.Previous research (29,30) has shown that, in individuals with persistent pain conditions, SDI-II scores above 16 are associated with the best sensitivity and specificity values for a clinical diagnosis of depression.Pain severity: The McGill Pain Questionnaire (MPQ) (31) was used as a measure of spontaneous (ie, condition-related) pain.On this measure, participants are asked to endorse adjec tives that best describe their current pain experience.The MPQ Pain Rating Index (MPQ-PRI) was computed as the weighted sum of an adjectives endorsed.The MPQ-PRI has been shown to be a reliable and valid measure of an individ ual's chronic pain experience (32).Disability: The Pain Disability Index (POI) (33) was used as a self-report measure of pain-related disability.On this measure, participants are asked to rate their level of disability in seven different areas of daily Living (home, social, recreational, occu pational, sexual, self-care, life support).The PDI has been shown to be internally reliable and significantly correlated with objective measures of disability (34).

Procedure and apparatus
The present study received ethical approval from the Ethics Review Committee of the Universite de Montreal and the Centre de recherche interdisciplinaire en readaptfltion du Montreal metropolitain.Participants signed a consent form as a condition of participation in the research.Participants were asked to complete the MPQ, the PDI and the BOI-II as part of their ini tial assessment.Participants were informed that the study aimed to develop a new assessment procedure for individuals suffering from persistent pain.They were made aware that the lifting task may lead to temporary increases in discomfort and they were free to discontinue at any point.
The lifting task was the same as that described by Sullivan et al (35), and consisted of 18 canisters (4 L paint canisters) partially filled with sand and placed on a table (surface 80 cm x 120 cm) adjusted to waist height for each participant.The canisters weighed 2.9 kg, 3.4 kg or 3.9 kg and were arranged in three rows of six canisters.Within the 3 x 6 matrix, each weight was represented twice in each location of a double latin square.The selection of loads was based on research suggesting a 12% weight difference as a detection threshold, and National Institute for Occupational Safety and Health recommenda tions for safe weight limits (36,37).The movements required to lift the canisters recruit musculature of the shoulders and lower back, areas of the body likely to be associated with pain in individuals with musculoskeletal conditions.As shown in Figure I, the canister locations required the adoption of three different functional anthropometric pos tural positions: normal, maximum and extreme reaches (38).In the normal reach position (position 1), the participant stands erect with his or her elbow bent at 90°; in the maximum reach position (position 2), the participant stands erect with his or her arm fully extended; in the extreme reach condition, Figure 1) Postural positions required to tJerform the lifting task the participant is forward flexed with his or her arm fully extended (38).The participants were asked to perform two different tasks -a pain rating task and a weight estimation task.Pain rating task: Participants were asked to lift the 18 canisters, with their dominant arm, in a predetermined sequence (ie, col umn 1, first, second, third position; column.2,first, second, third position; etc).Participants were asked to provide a verbal rating of their pain as they lifted each canister, on an II-point scale with the end points 0 (no pain) and 10 (extreme pain).Weight estimation task: Participants lifted the same canisters, with their dominant arm, in the same sequence as they were lifted in the pain rating task.Participants were asked to estimate the weight of each canister and to provide a verbal estimate in imperial or metric units.All weight estimates were later con verted to metric units.For the purposes of the present study, weight estimates were used as a proxy for perceived effort.

Data reduction
As described above, the lifting task required participants to lift 18 canisters arranged in six columns of three canisters.Each weight was represented in each column, thus equalizing columns in terms of the total weight lifted.For the purposes of the present study, the average pain rating or weight estimate within each column was used as the unit of analysis.
The pain rating and weight estimation data from the lifting task were analyzed as a three-way mixed factorial with sex (women, men) and level of depression (high, low) as between groups factors, and trial (collumns 1 to 6) as the within-groups factor.

Sample characteristics
Means and SOs for sample demographics and pain condition characteristics are presented in Table 1.Scores on measures of pain severity and self-reported disability are comparable with those that were reported in previous research on pain behav iour in chronic pain patients (39,40).Women and men did not differ significantly in age, t(81)=1.0,not significant (ns), or pain duration, t(81)=0.37,ns.A higher proportion of women had completed high school than men, X 2 =204, P<0.05.

Corrdations among dependent measures
Correlations among measures of pain, depression and self reported disability were computed separately for women and men.In Table 2, correlations computed for women are dis played above the diagonal and correlations for men are dis played below the diagonal.The results partially replicate previous research on the relations among pain, depression and disability.The correlation between depression and self reported disability was slightly (but not significantly) higher for women (r=0.58,P<O.OI) than for men (r=Oo4l, P<O.OI).
Depression was significantly correlated with pain ratings (MPQ-PRI and activity-related pain) for women but not for men.Sex differences in the magnitude of correlations between depression and pain were significant for pain ratings made dur ing the lifting task, z=1.7,P<0.05, but not for MPQ-PRI scores.
Measures of spontaneous pain (MPQ-PRl) and activity related pain were differentially associated with self-reported disability in women and in men.When both pain measures were included in a direct regression analysis, activity-related pain W=Oo4O, P<O.OI), but not the MPQ-PRI W=O.Results are presented as mean ± SO.Trial 1 is the average pain rating pro vided for the three canisters in column 1, Irial 2 is the average pain rating for the three canisters in column 2, etc self-reported disabiIity.For men, the MPQ-PRI W=0.38, P<O.OI), but not activity-related pain W=0.18, ns), contributed significant unique variance to the prediction of self-reported disability.
Activity-related pain ratings A three-way sex x level of depression x trials ANOVA was conducted on activity-related pain ratings.The analysis yielded significant main effectsfor sex, F(I, 79)= 11.4,P<O.OOI, and trials, F(5,395)=14.9,P<O.OO1.The main effect for level of depression failed to reach statistical significance, F(l,79)=2.8,P=0,08.A significant sex x level of depression interaction was obtained, F(l,79)=4,J, P<0.05, indicating that level of depression moderated the relation between sex and activity-related pain, Mean and standard deviations for the main effect of sex on activity-related pain ratings are presented in Table 3. Across all trials, women rated their activity-related pain as more severe than men.For both men and women, there was a significant linear trend reflecting increasing pain over repeated trials, F(l,79)=25,4, P<O.OO1. Figure 2 displays the mean activity-related pain ratings for women and men sepa rately, as a function of level of depression and trials.Depression contributed to a significant increase in activity-related pain ratings for women, but not for men.
Given that women and men in the high depression group dif fered with respect to their scores on the BDI-II, it is possible that sex-dependent differences in the relation between depression and activity-related pain may be the result of women's more extreme depression scores.A hierarchical regression was con ducted to examine whether the sex x level of depression inter action remained significant even when controlling for scores on the BDI-I!.In this analysis, mean activity-related pain was used as the dependent variable.Sex and level of depression were entered in the first step of the analysis and contributed signifi cantly to the prediction of pain, R2=0.21,F(2,80)=9,4, P<O.OO1.BD1-II scores were entered in the second step of the analysis but did not contribute significantly to the prediction of pain ratings beyond the variance accounted for by sex and level of depres sion, R2change=0.02,F(l,79)=1.8,P=0.16.The sex x level of depression interaction was entered in the third step of the analy sis and contributed significantly to the prediction of pain ratings, R2 chan e=F(I,78)=3.5,P<0.05.The results of this analysis indi cate t~at women's more extr•eme scores on the BD1-n do not account for the sex-dependent relation between depression and activity-related pain.
An analysis of covariance was conducted to address whether sex differences in activity-related pain were simply a • ..• = ...

Figure 2) Activity-related pain ratings as a function of level of depres sion and trials
reflection of more severe spontaneous (ie, condition-related) pain.In this analysis, the main effect of sex on activity-related pain was examined using the MPQ-PRI as a covariate.The results of this analysis indicated that the main effect of sex on activity-related pain remained significant even when MPQ PRI scores were statistically controlled, F(l ,80)=8,J, P<O.O 1.

Weight estimates
As noted earlier, weight estimates were used as a proxy for per ceived effort.A three-way (sex x level of depression x trials) mixed ANOVA was conducted on participants' weight esti mates of the canisters they lifted.The analysis yielded only a main effect for trials, F(5,395 )=21.9,P<O.OO1.As shown in Table 4, there was a significant linear trend reflecting increas ing weight estimates over trials, F(I,79)=32.9,P<O.OOI.The main effects for sex, F(I, 79)= 1.0, ns, and level of depression, F{ 1,79 )=0.06, ns, did not approach statistical significance.The absence of significant effects for sex and level of depression on weight estimates indicates that variations in perceived effort do not account for the sex-dependent re'lations between depression and activity-related pain.

DISCUSSION
The findings of the present research join a growing body of liter ature showing that women experience more severe pain than men (18,20).In the present study, sex differences in pain expe rience were found not only for spontaneous pain (ie, using the MPQ-PRI) but for activity-related pain as well.In addition, sex Results are presented as mean ± SO.Trial 1 is the average weight estimate provided for the three canisters in column 1, Trial 2 is the weight estimate for the three canisters in column 2, etc differences in activity-related pain remained significant even when controlling for MPQ-PRI scores.The latter finding sug gests that activity-related pain is a dimension of pain experi ence not captured by self-report measures of spontaneous or condition-related pain.
The results of correlational analyses on self-report measures of pain, depression and disability were partially consistent with those reported in previous research.Bolton (22) and Keogh et al (24) reported that depression was correlated with self-reported disability in women but not in men.In the present study, depres sion was correlated with self-reported disability for both men and women; however, the correlation was slightly higher in women than in men.Consistent with Keogh et al (24), the pres ent study did show that depression was correlated with pain rat ings (activity-related pain) for women, but not for men.
A few studies (41)(42)(43) have reported that the relation between depression and pain is greater in men than in women, and some studies have reported no sex differences in the rela tion between depression and pain ( 44).An explanation for these discrepant findings is not readily apparent.Two of the studies that have reported stronger depression-pain relations in men have been conducted with older adults (41,42).It is pos sible that life stage-related changes in the prevalence of depression, or sex-specific changes in hormonal mechanisms implicated in depression, may account for differences in depression-pain relations in older adults (45)(46)(47).More research is needed to clarify the sample characteristics that will determine the nature of sex differences in relations between depression and pain.
The results of the present study extend previous research in showing that depression augments activity-related pain in women, but not in men.In women, high levels of depression, compared with low levels of depression, were associated with approximately 40% more intense pain while performing the lifting task.Depression had a near-zero impact on activity related pain in men.Given that depression occurs more fre quently in women, its more negative impact on pain experience in women may help explain why women may be at greater risk for prolonged disability following musculoskeletal injury.If depressive symptoms act to amplify pain symptoms in women, they may be more likely to reach tolerance limits that affect their participation in activity.
In the depressed group, women obtained higher scores on the BDI-II than men.It could, therefore, be argued that the absence of a pain-augmenting effect from depression in men is due to the fact that the depression scores of men were not suffi ciently elevated to influence their pain experience.The results do not support such an explanation.Even when controlling for scores on the BDI-H, the interaction between sex and level of depression remained significant.It is also unlikely that the pain-augmenting effect of depression in women was due to sex differences in perceived effort.Analyses revealed no signifi cant effects due to sex or level of depression on participants' weight estimates.It is important to note, however, that weight estimation is an indirect measure of perceived effort (48).More direct measures of perceived effort or fatigue will be required to conclusively rule out the potential role of sex dif ferences in perceptions of effort as an account of the pain augmenting effects of depression in women.
In a previous study using the same lifting paradigm, Sullivan et al (35) reported significant sex differences in activity-related pain ratings and weight estimates.In the present study, women consistently provided higher weight estimates than men; however, unlike Sullivan et al (35), the differences did not attain statistical significance.The results of the present study suggest that the relation between sex and weight esti mates is associated with a small effect size and may not be reli ably reproducible.
Previous discussions of the relation between sex and dis ability have appealed to social ro~e explanations.It has been suggested that women may hold beliefs about the relation between depression and pain that impact negatively on their experience (24).It has also been suggested that the multiple role demands ofwomen may require that they establish priori ties of involvement.Under conditions of compromised resources, family role responsibilities may be accorded higher priority status than occupational role responsibilities.Although the present findings do not question the tenability of social role explanations of disability in women with muscu loskeletal conditions, they suggest that factors at a more basic level may also be playing a role.
Recent studies suggest that the relation between depression and pain sensitivity may vary as a function of the modality of pain stimulation (49).ln a sample predominantly composed of women, Bar et al (50) reported that depression was associated with 'hypoalgesic' responses to heat and electrical stimulation, and 'hyperalgesic' responses to musde ischemic pain.It has been suggested that hypoalgesic responses in depression may be the result of diminished spinal or brainstem transmission, while hyperalgesic responses in depression may be the result of insufficient activation of endogenous opioid pain inhibitory systems (49).Findings showing reduced efficacy of opioids in women further suggest the possibility that central pain inhibitory systems may be activated differently in women and men (51).
Recent research suggests that hormonal factors may under lie observed sex differences in the frequency and mechanisms of pain and depression in women and men (52)(53)(54).It has been suggested that hormonal factors may reduce the sensitivity of ~-opioid receptors in women (55).It has been shown that luteinizing hormone desensitizes opioid receptors, leading to decreased effectiveness of endogenous and exogenous opioids (56).It is pOSSible that similar hormonal processes may con tribute to the differential impact of depression on activity related pain in women and men.
Women's greater tendency to ruminate, or to become internally or emotionally focused when depressed, may also explain why depression may affect women's pain experiences in a manner different from that of men (57)(58)(59)(60).If women are more likely than men to become interna ly focused when Pain Res Manage Vol 13 No 3 May/June 2008 depressed, the increased attention to pain symptoms may inad vertently lead to a more severe pain experience.It has also been shown that depression potentiates an information pro cessing bias for pain-related stimuli.Although the question has not been investigated, it is pOSSible that such a processing bias is more pronounced in women than in men (61).
Unfortunately, the results of the present study do not per mit favouring one of the above potential explanations of the sex-dependent relation between depression and activity related pain.Future research examin•ing attentional or coping related processes may elucidate underlying mechanisms.Research incorporating assessment of descending modulatory mechanisms, such as the nociceptive flexion reflex, may also reveal a basis for the differential impact of depression on activity-related pain in women and men (62).
From a clinical perspective, the present findings point to the importance of effectively treating depression in women with persistent pain.In the past, some authors have suggested using antidepressants as a primary treatment for persistent pain (63).Although research has not supported the use of antide pressants as a primary treatment for persistent pain, there are indications that reductions in depression may be a prerequisite to pain reduction in individuals with concomitant symptoms of pain and depression (64).Unfortunately, depression contin ues to be underdetected in patients with persistent pain, and pain patients with depressive symptoms are more likely to be prescribed opioids instead of receiving treatment for depression (64,65).Even when antidepressant medication is considered, many pain patients with high levels of depressive symptoms are prescribed antidepressants at doses below the therapeutic range (66).Surprisingly, there have been few clinical trials examin ing the efficacy of antidepressants in depressed patients with musculoskeletal pain (4,67).
Some caution is warranted in the interpretation of the find ings of the present study.First, depression was operationalized as a high score on a self-report measure of depressive symptoms as opposed to a diagnostic interview.To date, the bulk of

Pain Res Manage Vol 13 No 3 May/June 2008
Influences on pain behaviour research on depression associated with musculoskeletal condi tions has been conducted with self-report measures (68).High scores on self-report measures of depressive symptoms do not imply that the diagnostic criteria for a depressive disorder have been met.More research is needed to address the prognostic value of diagnoses of depressive conditions verified by struc tured diagnostic interviews.
Another limitation of the present findings is that pain was induced under laboratory conditions that may not resemble the conditions under which activity-related pain is typically experienced.In the present study, participants were reassured that the lifting task was safe and would not lead to adverse neg ative health consequences, Under experimental conditions where safety concerns are paramount and the potential nega tive consequences of participation must be minimized, the lab oratory context may differ in important ways from the day-to-day experiences of pain patients, and the degree of gen eralizability of findings may be compromised to some extent.Finally, the quasi-experimental design, using pre-existing Ilev e1s of depression, limits the confidence that can be had in con clusions about tll.e causal status of depression on activity-related pain.
In spite of these limitations, the present findings provide preliminary evidence that depression augments activity related pain in women, but not in men.These findings may help explain why women experience more pronounced and prolonged pain-related disability than men.The findings may also bring attention to the detection and treatment of depres sive symptoms in women with persistent pain conditions.
23, ns), contributed significant unique variance to the prediction of Pain Res Manage Vol 13 No 3 May/June 2008 Pain Res Manage Vol 13 No 3 May/June 2008

TABLE 3
Pain ratings during canister lifts as a function of sex and trial

TABLE 4
Weight estimates (in kg) as a function of sex and trial