The Journal of Pain, Vol 13, No 12 (December), 2012: pp 1188-1197

Available online at www.jpain.org and www.sciencedirect.com

Depressive and Anxiety Symptoms as Risk Factors for

Temporomandibular Joint Pain: A Prospective Cohort Study
in the General Population

Stefan Kindler,* Stefanie Samietz,y Mohammad Houshmand,z Hans Jo

€rgen Grabe,x,{
z y z z
€lzke,**
Olaf Bernhardt, Reiner Biffar, Thomas Kocher, Georg Meyer, Henry Vo
y
Hans-Robert Metelmann,* and Christian Schwahn
*Department of Oral and Maxillofacial Surgery/Plastic Surgery, University of Greifswald, Greifswald, Germany.
y
Center of Oral Health, Department of Prosthetic Dentistry, Gerodontology and Biomaterials, University of Greifswald,
Greifswald, Germany.
z
Department of Restorative Dentistry, Periodontology, and Endodontology, University of Greifswald, Greifswald,
Germany.
x
Department of Psychiatry and Psychotherapy, University Medicine Greifswald, Greifswald, Germany.
{
Department of Psychiatry and Psychotherapy, HELIOS Hospital Stralsund, Stralsund, Germany.
**Institute of Community Medicine, Section Epidemiology of Health Care and Community Health, Greifswald,
Germany.

Abstract: Previous studies have associated depression and temporomandibular joint disorders
(TMDs). The temporality, however, remains to be clarified. Most patient studies have selected sub-
jects from treatment facilities, whereas in epidemiological studies a clinical examination has not
been performed. In this study the 5-year follow-up data of the population-based Study of Health
in Pomerania (SHIP) were analyzed. To estimate the effect of symptoms of depression and those of
anxiety on the risk of TMD pain, the Composite International Diagnostic-Screener (CID-S) and a clinical
functional examination with palpation of the temporomandibular joint and the masticatory muscles
were used. After exclusion of subjects having joint pain at baseline, a sample of 3,006 Caucasian par-
ticipants with a mean age of 49 years resulted. Of those, 122 participants had signs of TMD joint pain
upon palpation. Subjects with symptoms of depression had an increased risk of TMD joint pain upon
palpation (rate ratio: 2.1; 95% confidence interval: 1.5–3.0; P < .001). Anxiety symptoms were associ-
ated with joint and with muscle pain. The diagnosis, prevention, and therapy of TMD pain should
also consider symptoms of depression and those of anxiety, and appropriate therapies if necessary.
Perspective: Depressive and anxiety symptoms should be considered as risk factors for TMD pain. De-
pressive symptoms are specific for joint pain whereas anxiety symptoms are specific for muscle pain, find-
ings that deserve detailed examination. These findings may support decision-making in treating TMD.
ª 2012 by the American Pain Society
Key words: Temporomandibular disorders, orofacial pain, depression, anxiety, risk factors.

T
emporomandibular disorders (TMDs) are defined as multifactorial5,10; biologic, behavioral, environmental,
a subgroup of craniofacial pain problems. The tem- social, emotional, and cognitive factors seem to play
poromandibular joint, the masticatory muscles, and a role.7,10 The fundamental etiology of TMD, however,
the associated head and neck musculoskeletal structures remains poorly understood.31 Malocclusion,10 parafunc-
are involved.31 The etiology of TMD is considered to be tional habits such as tooth grinding and clenching,10,23

Received April 17, 2012; Revised August 28, 2012; Accepted September Address reprint requests to Stefan Kindler, MD, DDS, Department of
13, 2012. Oral and Maxillofacial Surgery/Plastic Surgery, Ernst Moritz Arndt
SHIP is part of the Community Medicine Research net of the University of University, Sauerbruchstr. D-17475 Greifswald, Germany. E-mail:
Greifswald, Germany, which is funded by the Federal Ministry of Educa- [email protected]
tion and Research (grants no. 01ZZ9603, 01ZZ0103, and 01ZZ0403), the 1526-5900/$36.00
Ministry of Cultural Affairs and the Social Ministry of the Federal State
of Mecklenburg-West Pomerania. ª 2012 by the American Pain Society
None of the authors reported potential conflicts of interest. http://dx.doi.org/10.1016/j.jpain.2012.09.004

1188
Kindler et al The Journal of Pain 1189
6,43 27,30
trauma, and genetic factors may initiate TMD. nonparticipation were health-related reasons (n = 161)
There is an ongoing debate about the importance of and no interest (n = 154).
psychological disorders as risk factors for TMD pain.20 Cog-
nitive behavioral treatment,19 biopsychosocial interven-
tion,9 biofeedback, and minimal therapist contact Outcome and Exposure Measures
interventions32 are options to treat TMD pain. Antidepres- We assessed the outcome variable, TMD pain, using
sants are often used in treatment of chronic pain of TMD data from the oral clinical examination, which was per-
patients in the head and neck region.31 Biobehavioral formed by 8 calibrated and certificated dentists. Inter-
studies suggest an association between TMD pain and co- examiner variability for TMD signs was measured in 5
existing psychopathology including depression, anxiety, calibrated sessions using a total of 22 volunteers. Train-
and posttraumatic stress disorders.8,19 Mongini et al25 in- ing of the dentists and consensus discussions were per-
quired about a careful screening for depression and anxi- formed before the study started and twice a year while
ety to be able to clinically evaluate facial pain. However, the study was running.15 TMD examinations were per-
findings on the association between those psychological formed following the guidelines for diagnosis of TMD
traits and TMD pain are inconsistent: some studies have by the Academy of Orofacial Pain.28
demonstrated such associations34,42 but others have To reach a higher degree of specificity we split up our
not.21 Methodological differences might explain diver- outcome variable—TMD pain—into the end points joint
gent findings. Most patient studies have selected rela- pain (arthrodynia) and muscle pain. We defined joint
tively small numbers of participants from treatment pain by pain upon lateral pressure and by pressure pain
facilities,21,36 whereas in observational studies a clinical referred cranially. These palpation sites were assessed
examination has not been performed.1,33 Only a few bilaterally, yielding a total of 4 measures. The assessment
studies analyzed the role of depression36 and anxiety1 us- of joint pain in more detail was as follows: 1) pain upon
ing a prospective cohort design. In most patient studies dorsocranial compression of the joints; here, a compres-
the Research Diagnostic Criteria for TMDs (RDC-TMD) are sion test of the condyles was conducted in the dorsocra-
based on clusters of signs and symptoms that may not de- nial direction with the patient’s mandible in the relaxed
fine a common pathophysiology.18 position; and 2) joint pain upon lateral pressure of about
Here, we aimed to estimate the relative risk of symp- 2 kg/cm2. While the mouth was slightly open, the con-
toms of depression and those of anxiety on TMD pain dyles were palpated; subjects were asked to describe their
over a 5-year follow-up. We used a population-based perception as ‘‘indolent/painless,’’ ‘‘uncomfortable,’’ or
study in which symptoms of depression and anxiety ‘‘painful.’’ For definition of pressure pain referred crani-
were assessed by an established screening instrument ally, a compression test of the condyles was conducted
for mental disorders and in which signs of TMD were de- with the patient’s mandible in a relaxed position.
fined by a standardized clinical examination. Pain in the masseter and temporalis muscles was bilat-
erally assessed, yielding a total of 4 measures. These
masticatory muscles were palpated extraorally under
Methods pressure of about 1 kg/cm2. The patients’ resulting per-
ceptions of pain using the 3 descriptions specified above
Participants were documented.
The Study of Health in Pomerania (SHIP) is a longitudi- To differentiate pain from discomfort, each outcome
nal population-based study in West Pomerania, the (joint pain and muscle pain) was defined as present if there
northeastern region of Germany with a total population was at least 1 site of pain upon palpation. The category ‘‘un-
of 212,157 inhabitants. At baseline examination, a 2- comfortable’’ was not excluded but coded as ‘‘indolent/
stage cluster sampling method using population regis- painless.’’ Likewise, the TMD cases at baseline were defined
tries was used to select German citizens residing in the only by pain. In ancillary analyses, however, we defined
study area who were 20 to 79 years of age at the time both prevalent and incident cases via pain or discomfort.
of invitation. Seven thousand eight persons were invited, For the definition of the 2 exposures—symptoms of de-
including 292 of each gender in each of 12 5-year age pression and those of anxiety—we used the Composite
strata.17 The net sample (without migrated or deceased International Diagnostic-Screener (CID-S), which is an es-
persons) comprised 6,265 eligible subjects, of whom tablished screening tool for psychological disorders. The
4,308 (68.8%) agreed to participate. The study was ap- CID-S was administered as a self-report questionnaire.
proved by the local ethics committee. All participants The CID-S is a 12-item screening instrument for mental
gave informed written consent. Data collection was con- disorders representing key symptoms of mental disorders
ducted from October 1997 until May 2001 and included 4 defined by Diagnostic and Statistical Manual of Mental
parts: a medical examination, an oral health examina- Disorders, 4th Edition, and International Classification
tion, a health-related interview, and a self-administered of Diseases, 10th Revision criteria.41 Eight questions of
health- and risk-factor-related questionnaire. The the CID-S are designed to assess lifetime occurrence of so-
follow-up examination was carried out from October matoform, anxiety, and depressive disorders. We used
2002 until September 2006. After exclusion of subjects the screening questions for depressive disorders (‘‘Did
who died or moved away, the follow-up response was you ever suffer from feelings of sadness or depressed
83.6% (Fig 1). Among those 541 subjects who refused mood for a period of at least two weeks?’’ and ‘‘Did
the follow-up examination, the main reasons for you ever suffer from lack of interest, tiredness or loss of
1190 The Journal of Pain Effect of Depressive and Anxiety Symptoms on TMD Pain

2192 females and 2116 males aged 20 - 81 years at the time of baseline examination

Lost to follow-up: 231 died

129 moved away
541 refused
107 for reasons other than above

1711 females and 1589 males after 5 years

Joint: pain on palpation Muscles: pain on palpation

60 subjects having joint pain at 71 subjects having muscle pain at

baseline baseline
17 with missing baseline data 14 with missing baseline data

3223 without pain on palpation 3215 without pain on palpation

(joint) at baseline (muscles) at baseline

29 excluded: 26 excluded:
19 with epilepsy, 16 with epilepsy,
4 with multiple sclerosis, 4 with multiple sclerosis,
6 with Parkinson’s disease 6 with Parkinson’s disease
16 no data for exclusion criteria 14 no data for exclusion criteria

3178 met inclusion criteria 3175 met inclusion criteria

172 excluded for missing data: 141 excluded for missing data:
127 missing outcome data, 96 missing outcome data,
45 missing exposure data, 45 missing exposure data,
1 missing confounder data 1 missing confounder data
(overlapping) (overlapping)

3006 were included in analysis 3034 were included in analysis

Figure 1. Flow diagram of the study population.

energy for a period of at least two weeks?’’) and for anx- divided into 4 categories with respect to the mean daily
iety (panic attacks, generalized anxiety, agoraphobia, so- alcohol consumption: none, 0 to 10 g alcohol/day; 10 to
cial phobia, specific phobia). The variables for depression 20 g alcohol/day; 20 to 30 g alcohol/day; and >30 g
and anxiety were dichotomized (none versus at least 1 alcohol/day. We also controlled for relevant general
positive answer). The performance of the CID-S to detect diseases including arthritis, degenerative disc disease,
lifetime occurrence of mental disorders has been shown osteoporosis, myopathy, migraine, and cancer from
to be good, as indicated by an overall sensitivity of 80.7% the interview. Household income was defined by
and a negative predictive value of 85.1%.41 questionnaire. The monthly household income (in
German Marks; 1 Euro = 1.956 German Marks [DM]) was
Confounding divided by the square root of the number of people
To reduce confounding we excluded participants with living in the household and was grouped into quintiles
epilepsy, multiple sclerosis, and Parkinson’s disease (Fig (limits: 1,200, 1,678, 2,130, and 2,650 DM).
1). Analyses were adjusted for other potential con- Dental occlusion was classified according to the Eich-
founders. Sociodemographic characteristics and alcohol ner index, a classification system based on occluding
drinking habits were assessed by computer-assisted per- pairs of teeth.26 The dental status was examined for
sonal interviews. School education was categorized in 3 each tooth. The dental part of the interview was con-
levels (<10 years, 10 years, and >10 years of schooling). ducted by 2 trained dental assistants.15
The current marital status comprised 5 categories (mar-
ried, married but separated, single, divorced, and wid-
owed). Alcohol intake during the previous weekend was Pain Conditions That Play an Unclear Role
used as a proxy for general intake, and the mean daily al- of Confounder or Intermediate
cohol consumption was calculated using beverage-specific Suboccipital and sternocleidomastoid muscles as loci
pure ethanol volume proportions.2,26 Participants were of cervical pain were palpated as the masticatory muscles
Kindler et al The Journal of Pain 1191
were palpated. The question about neck and shoulder mum 8.6 years), 50 subjects had muscle pain upon palpa-
pain was used from the von Zerssen’s complaints scale. tion in the masseter or in the temporalis muscle.
Because the question about headache of the von Zers-
sen’s complaints scale included facial pain (‘‘headache
or facial pain’’), headache was included from the dental Baseline Characteristics
part of the interview in order to increase the specificity. The baseline characteristics of the study participants
with respect to joint and muscle pain are presented in
Table 1. Both outcomes occurred in 17 subjects, yielding
Statistics an association of .19 (Crame r’s V). Besides sex, only symp-
Data on quantitative characteristics are expressed as toms of depression, anxiety symptoms, and migraine
a mean and standard deviation. Data on qualitative char- were related to both outcomes. Sex, school education,
acteristics are expressed as percent values or absolute marital status, arthritis, degenerative disc disease, osteo-
numbers, as indicated. For continuous data, groups porosis, and migraine were associated either with symp-
were compared using t-test and nominal data were com- toms of depression or with those of anxiety.
pared using chi-square test. To estimate the rate ratio ap-
propriately we used a modified Poisson regression
approach in our main analyses.44 We adjusted 1) for Main Analyses
age and sex; 2) additionally for school education; and Adjusting for age and sex, the rate ratio of joint pain
3) additionally for arthritis and degenerative disc dis- was 2.13 (95% confidence interval [CI]: 1.50–3.01) for sub-
ease. To estimate the effect of further potential con- jects with depressive symptoms. Further adjustment did
founders, we used the criterion of the change in the not lead to a change >10% in the rate ratio (Table 2).
coefficient of interest. A substantial change was present The additional inclusion of migraine into the final model
if the inclusion in the model led to $10% change in the presented in Table 2 changed the results only slightly. By
rate ratio of depression or anxiety. A value of P < .05 was additionally including anxiety symptoms, the rate ratio
considered statistically significant. Analyses were con- of joint pain for subjects with depressive symptoms de-
ducted with STATA/MP software, v.10.1 (StataCorp LP, creased by 11.0% (rate ratio from 2.09 to 1.86; Table 2).
College Station, TX). The natural logarithm of the length The change in the rate ratio from the final model to the
of the observation period was included as an offset to model adjusting for pain conditions including headache,
handle different observation lengths. In sensitivity anal- neck and shoulder pain, pain in suboccipital muscles,
yses, the rate ratio was estimated in a count model, pain in sternocleidomastoid muscles, and migraine, was
namely the negative binomial regression. 5.7% (rate ratio from 2.09 of the final model to 1.97).
The age- and sex-adjusted rate ratio of joint pain was
1.79 (95% CI: 1.24–2.58) for subjects with anxiety symp-
Sensitivity Analyses toms. No further potential confounder of this relationship
Because of the difficulty of differentiating pain medi- met our criterion for a confounder (Table 2). The inclusion
cation taken for TMD-specific pain from pain medication of depressive symptoms, however, decreased the rate ratio
taken for other reasons, we assessed pain medication by of joint pain for subjects with anxiety symptoms by 18.9%,
using the Anatomical-Therapeutic-Chemical (ATC) code which dissolved the statistical relationship between joint
only in sensitivity analyses. We excluded subjects who pain and anxiety symptoms. The change in the rate ratio
used nonsteroidal antiinflammatory drugs (NSAIDs), from the final model to the model adjusting for pain con-
(M01A), NSAIDs in combination with steroids, anti- ditions was 9.7% (rate ratio from 1.75 to 1.58).
inflammatory/antirheumatic agents in combination The age- and sex-adjusted rate ratio of muscle pain was
(M01B), opioids (N02A), and other analgesics and antipy- 1.83 (95% CI: 1.05–3.19) for subjects with depressive symp-
retics (N02B) at baseline or at the follow-up examination. toms (Table 2). After including arthritis and degenerative
Moreover, the ATC codes N06A (antidepressants), N05B disc disease, the relationship between muscle pain and
(anxiolytics), and N06C (psycholeptics and psychoanalep- depressive symptoms was no longer statistically signifi-
tics in combination) were considered as confounders. cant. When additionally adjusting for anxiety symptoms,
To increase the number of events with muscle pain, we the rate ratio of muscle pain for subjects with depressive
counted pain or discomfort in the masseter and tempora- symptoms decreased by 28.8% (rate ratio from 1.70 to
lis muscles instead of only pain. 1.21). The change in the statistically not significant rate ra-
tio from the final model to the model adjusting for pain
conditions was 11.2% (rate ratio from 1.70 to 1.51).
Results For subjects with anxiety symptoms, the age- and sex-
Participants having epilepsy, multiple sclerosis, Parkin- adjusted rate ratio of pain upon palpation in the masti-
son’s disease, and joint or muscle pain at baseline were ex- catory muscles was 3.46 (95% CI: 1.75–6.78) (Table 2).
cluded. Thus, 3,006 participants remained for the analysis The inclusion of further variables did not lead to
of joint pain and 3,034 participants remained for the anal- a change >10%. In particular, the inclusion of depressive
ysis of muscle pain (Fig 1). Counting the 15,682 person-years symptoms reduced the rate ratio of interest only slightly.
of follow-up (median 5.0, maximum 8.6 years), 122 subjects The change in the rate ratio from the final model to the
had at least 1 site of joint pain upon palpation. Counting model adjusting for pain conditions was 5.3% (rate ratio
the 15,835 person-years of follow-up (median 5.0, maxi- from 3.37 to 3.19).
1192 The Journal of Pain Effect of Depressive and Anxiety Symptoms on TMD Pain
Characteristics of Study Participants From the Study of Health in Pomerania at Baseline in
Table 1.
1997–2001
JOINT: PAIN ON PALPATION MUSCLES: PAIN ON PALPATION

NO YES NO YES
Number of participants 2,884 122 2,984 50
Age (years) 49.2 6 15.1 46.7 6 14.1 49.1 6 15.1 48.7 6 14.6
Length of observation period (years) 5.2 6 .5 5.1 6 .4y 5.2 6 .5 5.2 6 .5
Sex (female) 1,456 (50.5) 85 (69.7)z 1,510 (50.6) 36 (72.0)y
Anxiety 1,358 (47.1) 79 (64.8)z 1,410 (47.3) 39 (78.0)z
Panic attacks 642 (22.3) 44 (36.1)z 671 (22.5) 21 (42.0)y
Generalized anxiety 748 (25.9) 42 (34.4)* 776 (26.0) 25 (50.0)z
Social phobia 324 (11.2) 17 (13.9) 335 (11.2) 10 (20.0)
Agoraphobia 144 (5.0) 8 (6.6) 149 (5.0) 5 (10.0)
Specific phobia 535 (18.6) 32 (26.2)* 552 (18.5) 22 (44.0)z
Depression 817 (28.3) 60 (49.2)z 866 (29.0) 23 (46.0)*
Daily sadness 631 (21.9) 46 (37.7)z 671 (22.5) 17 (34.0)
Loss of energy 532 (18.5) 40 (32.8)z 567 (19.0) 18 (36.0)y
Anxiety or depression 1,570 (54.4) 88 (72.1)z 1,633 (54.7) 40 (80.0)z
School education
8 years 1,019 (35.3) 31 (25.4) 1,038 (34.8) 20 (40.0)
10 years (reference) 1,342 (46.5) 69 (56.6) 1,411 (47.3) 20 (40.0)
12 years 523 (18.1) 22 (18.0) 535 (17.9) 10 (10.0)
Household income (German Marks - DM) 1,981 6 966 2,098 6 1,028 1,986 6 964 1,833 6 1,062
Household income, quintiles
1st quintile, #1,200 DM 572 (20.7) 21 (18.6) 583 (20.4) 14 (29.8)*
2nd quintile, >1,200–1,678 DM 553 (20.0) 23 (20.4) 574 (20.1) 9 (19.1)
3rd quintile, >1678–2,130 DM 548 (17.6) 27 (23.9) 562 (19.7) 13 (27.7)*
4th quintile, >2,130–2,650 DM (ref.) 485 (17.6) 12 (10.6) 503 (17.6) 1 (2.1)
5th quintile, >2,650 DM 604 (21.8) 30 (26.5) 629 (22.1) 10 (21.3)*
Marital status
Married (reference) 1,937 (67.2) 91 (74.6) 1,904 (66.8) 29 (61.7)
Married but separated 34 (1.2) 3 (2.5) 37 (1.3) 0 (.0)
Single 531 (18.4) 12 (9.8)* 523 (18.3) 10 (21.3)
Divorced 207 (7.2) 11 (9.0) 211 (7.4) 6 (12.8)
Widowed 175 (6.1) 5 (4.1) 176 (6.2) 2 (4.3)
Arthritis 202 (7.0) 10 (8.2) 195 (6.9) 6 (12.8)
Degenerative disc disease 1,308 (45.4) 60 (49.2) 1,261 (44.8) 26 (55.3)
Osteoporosis 125 (4.5) 3 (2.5) 114 (4.2) 6 (12.8)*
Myopathy 18 (0.6) 2 (1.6) 21 (.7) 0 (.0)
Migraine 388 (13.5) 26 (21.3)* 373 (13.3) 14 (30.4)y
Cancer 23 (.8) 0 (.0) 22 (.8) 1 (2.1)
Alcohol consumption
0 g/day 1,032 (35.8) 44 (36.1) 994 (35.3) 19 (40.4)
>0–10 g/day 356 (12.3) 20 (16.4) 351 (12.5) 7 (14.9)
>10–20 g/day 424 (14.7) 17 (13.9) 421 (15.0) 3 (6.4)
>20–30 g/day 279 (9.7) 13 (10.7) 278 (9.9) 4 (8.5)
>30 g/day 792 (27.5) 28 (23.0) 769 (27.3) 14 (29.8)
Edentulism (related to 32 teeth) 246 (8.5) 6 (4.9) 245 (8.7) 2 (4.3)
Number of teeth in dentate subjects 21.7 6 8.0 22.5 6 7.3 21.8 6 8.0 20.6 6 8.7
Eichner classification
A (reference) 1,490 (51.7) 69 (56.6) 1,463 (52.0) 26 (55.3)
B1 316 (11.0) 14 (11.5) 307 (10.9) 2 (4.3)
B2 213 (7.4) 10 (8.2) 203 (7.2) 1 (2.1)
B3 159 (5.5) 7 (5.7) 160 (5.7) 4 (8.5)
B4 174 (6.0) 8 (6.6) 161 (5.7) 8 (17.0)*
C 532 (18.4) 14 (11.5) 520 (18.5) 6 (12.8)
Suboccipital muscles (right or left side)
Inconspicuous (reference) 2,835 (98.3) 115 (94.3) 2,838 (98.4) 113 (94.5)
Uncomfortable 25 (.9) 3 (2.5) 26 (.9) 3 (2.5)
Painful 23 (.8) 4 (3.3)y 21 (.7) 3 (2.5)*
Sternocleidomastoid muscles (right or left side)
Inconspicuous (reference) 2,877 (99.9) 122 (100) 2,878 (99.9) 120 (100)
Kindler et al The Journal of Pain 1193
Table 1. Continued
JOINT: PAIN ON PALPATION MUSCLES: PAIN ON PALPATION

NO YES NO YES
Uncomfortable 2 (.1) 0 (0) 2 (.1) 0 (0)
Painful 2 (.1) 0 (0) 2 (.1) 0 (0)
Neck and shoulder pain
No (reference) 918 (32.0) 32 (26.7) 925 (32.3) 33 (28.0)
Mild 817 (28.5) 21 (17.5) 811 (28.3) 23 (19.5)
Moderate 886 (30.9) 54 (45.0)* 887 (30.9) 49 (41.5)
Severe 244 (8.5) 13 (10.8) 243 (8.5) 13 (11.0)
Headache
No (reference) 2,064 (71.6) 68 (55.7) 2,067 (71.7) 66 (55.0)
Sometimes 575 (20.0) 29 (23.8) 576 (20.0) 31 (25.8)*
Often 231 (8.0) 22 (18.0)z 226 (7.8) 20 (16.7)z
Always 11 (.4) 3 (2.5)z 12 (.4) 3 (2.5)z

*P < .05.
yP < .01.
zP < .001.

Sensitivity Analyses 2,909 subjects remained after exclusion of prevalent

After excluding subjects taking pain medication ac- cases, among them 124 incident cases of masticatory
cording to the ATC code, the findings of the main analyses muscle pain or discomfort (4.3%). For subjects with de-
were confirmed. For subjects with depressive symptoms, pressive symptoms, the change in the rate ratio from
the rate ratio of joint pain was 1.95 (95% CI: 1.29–2.96; the final model to the model adjusting for pain condi-
P = .002, n = 2,279, 85 events) adjusted for sex, age, school tions was 11.7% (rate ratio from 1.79 to 1.58) and those
education, arthritis, and degenerative disc disease. This to the model including anxiety was 16.2% (rate ratio
rate ratio decreased slightly to 1.88 (95% CI: 1.23–2.86; from 1.79 to 1.50). For subjects with anxiety symptoms,
P = .003) after additional adjustment for treatments of the change in the rate ratio from the final model to
depression and anxiety. For subjects with anxiety symp- the model adjusting for pain conditions was 10.8%
toms, the rate ratio of joint pain adjusted for sex, age, (rate ratio from 1.95 to 1.74) and those to the model in-
school education, arthritis, and degenerative disc disease cluding depressive symptoms was 12.3% (rate ratio from
was 1.56 (95% CI: 1.02–2.38; P = .039), and it was 1.51 1.95 to 1.71).
(95% CI: .99–2.31, P = .056) when adjusted additionally
for treatments of depression and anxiety. Discussion
For analyses of muscle pain in the masseter or temporalis Depressive and anxiety symptoms emerged as risk fac-
muscle, only 28 events were available (n = 2,301). There- tors for TMD pain in this population-based follow-up
fore, the following rate ratios should be regarded with study. We found moderate-to-strong relationships that
caution. For subjects with depressive symptoms, the rate were independent of relevant confounders and re-
ratio adjusted for sex, age, school education, arthritis, mained stable over a number of sensitivity analyses.
and degenerative disc disease was 1.95 (95% CI: 0.95– Our findings show some specificity with respect to the
4.00; P = .068), and it was 2.01 (.99–4.08, P = .054) when exposures used. In the main analyses, depressive symp-
adjusted additionally for treatments of depression and toms were more strongly related to joint pain compared
anxiety. For subjects with anxiety symptoms, the rate ratio to muscle pain, whereas anxiety symptoms were more
adjusted for sex, age, school education, arthritis, and de- strongly related to muscle pain compared to joint pain.
generative disc disease was 5.19 (95% CI: 1.95–13.8; To our knowledge, this is the first population-based fol-
P = .001), and it was 5.22 (1.96–13.9, P = .001) when adjusted low-up study that investigated these associations.
additionally for treatments of depression and anxiety. A relationship between depressive or anxiety symptoms
and an increased risk for muscle pain or joint pain is plau-
sible for at least 2 reasons. First, depressive and anxiety
Ancillary Analyses Including Pain and symptoms may initiate muscular hyperactivity followed
Discomfort for Defining Prevalent and by muscle abnormality and altered muscle mechanics, a se-
Incident Cases quence in which each state can provoke muscle pain.31
The ancillary analyses confirmed the main analyses They may also initiate joint inflammation followed by
(Table 2). The exposure effects were smaller than in the biomechanical alterations, which provoke joint pain.
main analyses but nevertheless statistically significant. Second, TMD might be related to abnormal pain process-
For analysis of joint pain or discomfort, 2,872 subjects ing in the trigeminal system31 caused by imbalances22
remained after exclusion of prevalent cases, among in common neurotransmitters such as serotonin and cate-
them 320 incident cases of joint pain or discomfort cholamines.3,4,31 Moreover, TMD pain may reflect
(11.1%). For analysis of muscle pain or discomfort, physical manifestation of anxiety or depression.37
 1194
The Journal of Pain
Relationship Between Joint Pain and Muscle Pain (Outcomes) and Depressive Symptoms and Anxiety Symptoms (Exposures) in
Table 2.
a 5-Year Follow-Up
PAIN VERSUS DISCOMFORT OR INDOLENT/PAINLESS (REFERENCE GROUP) PAIN OR DISCOMFORT VERSUS INDOLENT/PAINLESS (REFERENCE GROUP)

DEPRESSIVE SYMPTOMS ANXIETY SYMPTOMS DEPRESSIVE SYMPTOMS ANXIETY SYMPTOMS

JOINT PAIN OR MUSCLE PAIN OR JOINT PAIN OR MUSCLE PAIN OR

JOINT PAIN MUSCLE PAIN JOINT PAIN MUSCLE PAIN DISCOMFORT DISCOMFORT DISCOMFORT DISCOMFORT
n = 3,006 n = 3,034 n = 3,006 n = 3,034 n = 2,872 n = 2,909 n = 2,872 n = 2,909
SUCCESSIVE ADJUSTMENT FOR
CONFOUNDERS RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI) RR (95% CI)
Unadjusted 2.37z (1.67–3.35) 2.07y (1.19–3.59) 2.01z (1.40–2.90) 3.89z (2.00–7.57) 1.91z (1.55–2.35) 2.25z (1.60–3.18) 1.62z (1.32–2.01) 2.35z (1.63–3.41)
Age and sex 2.13z (1.50–3.01) 1.83* (1.05–3.19) 1.79y (1.24–2.58) 3.46z (1.75–6.78) 1.80z (1.46–2.22) 1.93z (1.37–2.73) 1.55z (1.25–1.92) 2.04z (1.41–2.97)
School education 2.13z (1.51–3.01) 1.82* (1.03–3.19) 1.78y (1.24–2.57) 3.49z (1.78–6.82) 1.77z (1.43–2.18) 1.90z (1.34–2.69) 1.52z (1.23–1.88) 2.03z (1.40–2.94)
Arthritis and degenerative 2.09z (1.47–2.98) 1.70 (.96–3.00) 1.75y (1.21–2.52) 3.37z (1.70–6.66) 1.72z (1.39–2.13) 1.79y (1.26–2.54) 1.49z (1.20–1.84) 1.95z (1.34–2.83)
disc disease (final model)
Migraine 2.08z (1.46–2.96) 1.57 (.87–2.86) 1.74y (1.20–2.52) 3.18y (1.59–6.38) 1.69z (1.37–2.10) 1.75y (1.32–2.79) 1.47z (1.19–1.83) 1.92z (1.32–2.79)

Effect of Depressive and Anxiety Symptoms on TMD Pain

n = 2,996 n = 3,024 n = 2,996 n = 3,024 n = 2,862 n = 2,900 n = 2,862 n = 2900
Further adjustment of the final model for alternatively included covariates that play an unclear role of confounder or intermediate
Anxiety 1.86y (1.28–2.69) 1.21 (.69–2.12) 1.58z (1.26–1.98) 1.50* (1.03–2.18)
Depression 1.42 (.97–2.10) 3.18z (1.61–6.27) 1.28* (1.02–1.61) 1.71y (1.15–2.55)
Pain conditionsx 1.97z (1.37–2.83) 1.51 (.85–2.71) 1.58* (1.10–2.29) 3.19z (1.65–6.13) 1.62z (1.31–2.02) 1.58* (1.10–2.27) 1.40z (1.12–1.75) 1.74y (1.19–2.54)
n = 2,968 n = 3,004 n = 2,968 n = 3,004 n = 2,835 n = 2,872 n = 2,835 n = 2,872

*P < .05.
yP < .01.
zP < .001.
xPain conditions that play an unclear role of confounder or intermediate include headache, neck and shoulder pain, pain on suboccipital muscles, pain on sternocleidomastoid muscles. To complete the pain-related adjustment, migraine was
also included.
Kindler et al The Journal of Pain 1195
Undifferentiated somatization may affect as much as alent cases by using pain and discomfort, resulting in
20% of the population.13,37 Anxiety and depression 124 incident cases for ancillary analyses. Here, the
are the most frequent mental disorders in Europe.40 change in rate ratio of about 12% by including pain
Associations of TMD with depression and anxiety have conditions is something larger than the 10% change
been previously described.33,38 The temporality of that considered important in selecting relevant con-
relation, however, has not been demonstrated.33 In founders. These pain conditions, however, may also
addition a clinical examination in epidemiological stud- be affected by the outcome (joint or muscle pain) or
ies has not been performed; and in most patient studies by the exposure (depressive or anxiety symptoms) or
with signs and symptoms of TMD a small number of sub- play a role as an intermediate step in the relationship
jects participated. Furthermore, the association was not of interest. In each of these 3 causal models, the pain
always adjusted for important confounding factors like conditions must not be included in the analysis. Be-
pain medication and migraine.16 In concordance with cause of the unclear role of the pain conditions in the
other studies,33,38 we found a moderate-to-strong rela- causal pathway of interest, it seems reasonable to
tionship between symptoms of depression or anxiety ease the usual but arbitrary 10% change criterion to
and signs of TMD. In contrast, patients with complete some degree. Nevertheless, the effect of anxiety symp-
temporomandibular joint intracapsular disorder have toms on muscle pain was statistically significant in each
a low proportion of accompanying symptoms and are main and ancillary analysis even when adjusting for
apt to have a normal personality profile.24 Unlike our pain conditions. In addition, this study shares with
approach, in most studies the diagnosis of TMD is based others the limitations inherent to multiple event data.
on clusters of signs and symptoms that may not define To model multiple events appropriately, data to answer
a common pathophysiology.18 In addition, ‘‘disease 4 points are needed.35 The 4 points are related to the
should be defined on the basis of criteria that have uniformity of the events, to the temporality of the
nothing to do with exposure’’ in order to avoid bias in events, to the state dependence of the events, and to
epidemiologic evaluation.29 To achieve a higher speci- the time origin of the events.
ficity of the outcome, we focused only on clinical signs Even though CID-S is an established screening tool
of TMD rather than on subjective symptoms of TMD. for depression and anxiety, it does not generate diag-
To further increase the specificity of the outcome, we noses of mental disorders but rather captures typical
separated joint from muscle pain. Moreover, to increase symptoms. It is especially noteworthy that anxiety
the specificity of the exposure, we separated symptoms symptoms and depression frequently co-occur and
of depression from those of anxiety. Therefore, our that the complete distinction between both syn-
main results indicate a specificity in the relationship dromes is not feasible in our study. Furthermore, we
between depressive symptoms and joint pain and the did not analyze stressful life events and traumatiza-
relationship between anxiety symptoms and muscle tion as a possible confounder or trigger of mental dis-
pain. We do not have a good explanation for this spec- orders. Traumatic events such as sexual or physical
ificity. Although depressive and anxiety symptoms are abuse may have an acute but also chronic effect, lead-
associated, it is possible that depressive symptoms, but ing to a sensitization of pain perception. After exclud-
not anxiety symptoms, modify the effect of gene poly- ing subjects taking pain medication, the association
morphisms on both muscle and joint pain (or pain became even stronger, whereas the number of cases
sensitivity), which was recently demonstrated for the decreased substantially. However, it was not possible
COMT gene.30 Second, depressive and anxiety symptoms for us to differentiate between permanent or acute in-
may differently alter the stress and alarm systems of the take of pain medication. Also, information bias and
body as the result of differential regulations of the unmeasured confounding could have influenced our
hypothalamus-pituitary-adrenal axis.14 If, in addition, results. Finally, subjects in this study were Caucasian,
certain risk factors including gene polymorphisms limiting to some degree the generalizability of our
diverge in their effects on muscle and joint pain, findings.
then the observed specificity could be due to Notwithstanding these limitations there are several
differential effect modifications by depressive and anx- important strengths of our study. In fact, it is an un-
iety symptoms. usual quality beyond patient studies that signs of
A major limitation of our study is the low incidence TMD were diagnosed by palpation. Compared to
of muscle pain, although the proportion of cases is patient studies of more modest size, the population-
in good agreement with other studies given the based design ensures at least 3 advantages. First, the
population-based study design.11 In the final model, sampling procedure avoids self-selection by patients.
50 new cases were related to 8 variables. However, 6 Second, large samples considerably reduce the random
events per variable do not necessarily lead to biased error inherent in each study. Third, the sampling
estimates.39 Thus, by using the change-in-estimate cri- procedure based on population registries allows for
terion the general rule of 10 events per variable can a high degree of generalizability of the findings for
be relaxed. Moreover, the number of potential con- Caucasians. Moreover, the 2 items for depressive symp-
founders that were statistically related to both the out- toms have already proven their power in predicting
come and the exposure was low. To overcome the low health care utilization and costs 5 years from base-
number of new cases for muscle pain in the masseter line12 and can be recommended in everyday clinical
and temporalis muscles, we defined incident and prev- settings.
1196 The Journal of Pain Effect of Depressive and Anxiety Symptoms on TMD Pain
References painful temporomandibular disorders (TMD). J Dent Res
81:284-288, 2002
1. Aggarwal VR, Macfarlane GJ, Farragher TM, McBeth J: 17. John U, Greiner B, Hensel E, Ludemann J, Piek M,
Risk factors for onset of chronic oro-facial pain–results of Sauer S, Adam C, Born G, Alte D, Greiser E, Haertel U,
the North Cheshire oro-facial pain prospective population Hense HW, Haerting J, Willich S, Kessler C: Study of Health
study. Pain 149:354-359, 2010 In Pomerania (SHIP): A health examination survey in an
east German region: Objectives and design. Soz Praven-
2. Alte D, Ludemann J, Piek M, Adam C, Rose HJ, John U: Dis- tivmed 46:186-194, 2001
tribution and dose response of laboratory markers to alco-
hol consumption in a general population: Results of the 18. Kim H, Clark D, Dionne RA: Genetic contributions to clin-
study of health in Pomerania (SHIP). J Stud Alcohol 64: ical pain and analgesia: Avoiding pitfalls in genetic research.
75-82, 2003 J Pain 10:663-693, 2009
3. Apkarian AV, Baliki MN, Geha PY: Towards a theory of 19. Litt MD, Shafer DM, Ibanez CR, Kreutzer DL, Tawfik-
chronic pain. Prog Neurobiol 87:81-97, 2009 Yonkers Z: Momentary pain and coping in temporoman-
dibular disorder pain: Exploring mechanisms of cognitive
4. Bair MJ, Robinson RL, Katon W, Kroenke K: Depression behavioral treatment for chronic pain. Pain 145:160-168,
and pain comorbidity: A literature review. Arch Intern 2009
Med 163:2433-2445, 2003
20. Madland G, Feinmann C, Newman S: Factors associated
5. Clark GT: Etiologic theory and the prevention of tempo- with anxiety and depression in facial arthromyalgia. Pain
romandibular disorders. Adv Dent Res 5:60-66, 1991 84:225-232, 2000
6. De Boever JA, Keersmaekers K: Trauma in patients with 21. Marbach JJ, Lund P: Depression, anhedonia and anxiety
temporomandibular disorders: Frequency and treatment in temporomandibular joint and other facial pain syn-
outcome. J Oral Rehabil 23:91-96, 1996 dromes. Pain 11:73-84, 1981
7. Dworkin SF, Massoth DL: Temporomandibular disorders 22. McEwen BS: Physiology and neurobiology of stress and
and chronic pain: Disease or illness? J Prosthet Dent 72: adaptation: Central role of the brain. Physiol Rev 87:
29-38, 1994 873-904, 2007
8. Gameiro GH, da Silva Andrade A, Nouer DF, Ferraz de 23. Michelotti A, Cioffi I, Festa P, Scala G, Farella M: Oral par-
Arruda Veiga MC: How may stressful experiences con- afunctions as risk factors for diagnostic TMD subgroups. J
tribute to the development of temporomandibular dis- Oral Rehabil 37:157-162, 2010
orders? Clin Oral Investig 10:261-268, 2006
24. Mongini F, Ciccone G, Ibertis F, Negro C: Personality char-
9. Gatchel RJ, Stowell AW, Buschang P: The relationships acteristics and accompanying symptoms in temporomandib-
among depression, pain, and masticatory functioning in ular joint dysfunction, headache, and facial pain. J Orofac
temporomandibular disorder patients. J Orofac Pain 20: Pain 14:52-58, 2000
288-296, 2006
25. Mongini F, Ciccone G, Ceccarelli M, Baldi I, Ferrero L:
10. Gesch D, Bernhardt O, Mack F, John U, Kocher T, Alte D: Muscle tenderness in different types of facial pain and its re-
[Dental occlusion and subjective temporomandibular joint lation to anxiety and depression: A cross-sectional study on
symptoms in men and women. Results of the Study of Health 649 patients. Pain 131:106-111, 2007
in Pomerania (SHIP)]. Schweiz Monatsschr Zahnmed 114:
573-580, 2004 26. Mundt T, Mack F, Schwahn C, Bernhardt O, Kocher T,
John U, Biffar R: Gender differences in associations be-
11. Gesch D, Bernhardt O, Alte D, Schwahn C, Kocher T, tween occlusal support and signs of temporomandibular
John U, Hensel E: Prevalence of signs and symptoms of tem- disorders: Results of the population-based Study of
poromandibular disorders in an urban and rural German Health in Pomerania (SHIP). Int J Prosthodont 18:
population: Results of a population-based Study of Health 232-239, 2005
in Pomerania. Quintessence Int 35:143-150, 2004
27. Oakley M, Vieira AR: The many faces of the genetics con-
12. Grabe HJ, Baumeister SE, John U, Freyberger HJ, tribution to temporomandibular joint disorder. Orthod Cra-
Vo€ lzke H: Association of mental distress with health niofac Res 11:125-135, 2008
care utilization and costs: A 5-year observation in a gen-
eral population. Soc Psychiatry Psychiatr Epidemiol 44: 28. Okeson JP: Current terminology and diagnostic classifi-
835-844, 2009 cation schemes. Oral Surg Oral Med Oral Pathol Oral Radiol
Endod 83:61-64, 1997
13. Grabe HJ, Meyer C, Hapke U, Rumpf HJ, Freyberger HJ,
Dilling H, John U: Specific somatoform disorder in the gen- 29. Rothman KJ: Epidemiology: An Introduction. New York,
eral population. Psychosomatics 44:304-311, 2003 NY, Oxford University Press, 2002
14. Heim C, Newport DJ, Mletzko T, Miller AH, Nemeroff CB: 30. Schwahn C, Grabe HJ, Meyer zu Schwabedissen H,
The link between childhood trauma and depression: Insights Teumer A, Schmidt CO, Brinkman C, Kocher T, Nauck M,
from HPA axis studies in humans. Psychoneuroendocrinol- € lzke H, Biffar R, Bernhardt O: The effect of cate-
Vo
ogy 33:693-710, 2008 chol-O-methyltransferase polymorphisms on pain is
modified by depressive symptoms. Eur J Pain 16:878-889,
15. Hensel E, Gesch D, Biffar R, Bernhardt O, Kocher T, 2012
Splieth C, Born G, John U: Study of Health in Pomerania
(SHIP): A health survey in an East German region. Objectives 31. Scrivani SJ, Keith DA, Kaban LB: Temporomandibular
and design of the oral health section. Quintessence Int 34: disorders. N Engl J Med 359:2693-2705, 2008
370-378, 2003
32. Sherman JJ, Turk DC: Nonpharmacologic approaches to
16. Huang GJ, LeResche L, Critchlow CW, Martin MD, the management of myofascial temporomandibular disor-
Drangsholt MT: Risk factors for diagnostic subgroups of ders. Curr Pain Headache Rep 5:421-431, 2001
Kindler et al The Journal of Pain 1197
33. Sipila K, Veijola J, Jokelainen J, Jarvelin MR, dibular joint pain and dysfunction syndrome. Psychosom
Oikarinen KS, Raustia AM, Joukamaa M: Association Med 57:439-444, 1995
between symptoms of temporomandibular disorders and
depression: An epidemiological study of the Northern Fin- 39. Vittinghoff E, McCulloch CE: Relaxing the rule of ten
land 1966 Birth Cohort. Cranio 19:183-187, 2001 events per variable in logistic and Cox regression. Am J Epi-
demiol 165:710-718, 2007
34. Sipila K, Veijola J, Jokelainen J, Jarvelin MR,
Oikarinen KS, Raustia AM, Joukamaa M: Association of 40. Wittchen HU, Jacobi F: Size and burden of mental disor-
symptoms of TMD and orofacial pain with alexithymia: An ders in Europe–a critical review and appraisal of 27 studies.
epidemiological study of the Northern Finland 1966 Birth Eur Neuropsychopharmacol 15:357-376, 2005
Cohort. Cranio 19:246-251, 2001
41. Wittchen HU, Ho € fler M, Gander F, Pfister H, Storz S,
35. Skrondal A, Rabe-Hesketh S: Generalized latent variable Bedirhan U,€ Mu € ller N, Kessler RC: Screening for mental
modelling: Multilevel, longitudinal and structural equation disorders: Performance of the Composite International Di-
models. Boca Raton, FL, Chapman & Hall/CRC, 2004 agnostic - Screener (CID-S). Int J Meth Psychiat Res 8:
59-70, 1999
36. Slade GD, Diatchenko L, Bhalang K, Sigurdsson A,
Fillingim RB, Belfer I, Max MB, Goldman D, Maixner W: Influ- 42. Yap AU, Chua EK, Tan KB, Chan YH: Relationships be-
ence of psychological factors on risk of temporomandibular tween depression/somatization and self-reports of pain
disorders. J Dent Res 86:1120-1125, 2007 and disability. J Orofac Pain 18:220-225, 2004

37. van der Windt DA, Dunn KM, Spies-Dorgelo MN, 43. Zhang ZK, Ma XC, Gao S, Gu ZY, Fu KY: Studies on con-
Mallen CD, Blankenstein AH, Stalman WA: Impact of physi- tributing factors in temporomandibular disorders. Chin J
cal symptoms on perceived health in the community. J Psy- Dent Res 2:7-20, 1999
chosom Res 64:265-274, 2008
44. Zou G: A modified poisson regression approach to pro-
38. Vimpari SS, Knuuttila ML, Sakki TK, Kivela SL: Depressive spective studies with binary data. Am J Epidemiol 159:
symptoms associated with symptoms of the temporoman- 702-706, 2004