Eur Spine J (2006) 15: 1671–1685
DOI 10.1007/s00586-005-0046-z
Thomas Andersen
Finn B. Christensen
Cody Bünger
Received: 15 November 2004
Revised: 21 October 2005
Accepted: 8 December 2005
Published online: 10 February 2006
 Springer-Verlag 2006
T. Andersen (&) Æ F. B. Christensen
C. Bünger
Spine Unit, Department of Orthopaedics E,
Aarhus University Hospital, Aarhus,
Denmark
T. Andersen
Orthopaedic Research Laboratory Building
1A, Aarhus University Hospital,
Nørrebrogade 44, 8000 Aarhus C,
Denmark
E-mail: tba@dadlnet.dk
Tel.: +45-89494122
ORIGINAL ARTICLE
Evaluation of a Dallas Pain Questionnaire
classification in relation to outcome in lumbar
spinal fusion
Abstract Several studies have sug-
gested that psychosocial distress is
an important predictor of negative
outcome in lumbar spine surgery.
Ozguler et al. [Spine 27:1783–1789,
2002, 18] described a classification
tool for low back pain patients using
the Dallas Pain Questionnaire
(DPQ) which included a measure of
distress. We wanted to evaluate the
ability of this classification tool to
predict the outcome in spinal fusion
patients. Five hundred and sixty-six
patients (239 men, 327 women; mean
age 46, range 18–81) operated be-
tween 1992 and 2002, with a com-
plete DPQ preoperatively and after a
minimum of 1-year follow-up, were
included. They were classified pre-
operatively and at follow-up into
four groups: group 1 (slight disabil-
ity), group 2 (intermediate disabil-
ity), group 3 (major disability) and
group 4 (major disability and emo-
tional distress). Using logistic
regression, seven predictor variables
were investigated: age ()39 years/
40–59 years/60+ years), Gender
(male/female), Indication (spondyl-
olisthesis/primary degeneration/sec-
ondary degeneration), Work status
(working/without work or on sick
leave/retired or pensioned), Dura-
tion of pain (less than 1 year/1–
2 years/more than 2 years), Presence
of radiating pain (yes/no) and dis-
ability/distress [intermediate disabil-
ity (group 1–2)/major disability
(group 3)/major disability and dis-
tress (group 4)]. The outcome vari-
able was disability at follow-up
(low = group1 + 2/high = group
3 + 4). Preoperative classification
was group 1, 1%; group 2, 14%;
group 3, 37%; group 4, 48%. Vari-
ables found to predict high disability
at follow-up were secondary degen-
eration Odds Ratio (OR) 1.61
(P=0.020), being retired/pensioned
OR 3.48 (P<0.0005), age between
40–59 years OR 1.68 (P=0.011),
belonging to group 3 OR 2.69
(P=0.003) or belonging to group 4
OR 5.53 (P<0.0005). The classifi-
cation based on the DPQ were able
to identify lumbar spinal fusion pa-
tients with a considerable amount of
psychological distress in their symp-
tomatology. Furthermore, the pres-
ence of distress, as determined by
this classification, was a highly sig-
nificant risk factor for inferior out-
come.
Keywords Dallas Pain
Questionnaire Æ Classification Æ
Spinal fusion Æ Low back pain Æ
Outcome Æ Distress
1672
Introduction
Outcome predictors in spinal surgery have been the
subject of intensive research throughout the past two
decades. A variety of questionnaires have been utilized,
some designed to explore health-related aspects and
social phenomena, while a number of questionnaires
deal with various aspects of the patients’ psychological
profile. Because a single, isolated questionnaire, while
able to reveal risk factors, often does not provide
adequate predictive value, several studies have investi-
gated the efficacy of employing batteries of different
questionnaires in order to increase predictive value.
The drawback of this approach is time consumption
and intimidation of the respondee. This could be re-
duced by utilizing questionnaires whose assessment
abilities in relation to specific aspects of the patient’s
life are already proven and by devising for these
questionnaires a new scoring method with expanded
variables.
One of the disease specific questionnaires used in
research on low back pain is the Dallas Pain Ques-
tionnaire (DPQ). It consists of 16 questions and as-
sesses the functional impact of chronic spinal pain in
four categories: daily activities, work-leisure activities,
anxiety and depression, and social concerns [15]. It has
been used as the main outcome instrument in several
studies on the treatment of low back pain [1–6, 8, 13,
19, 23, 24]. In 2002 Ozguler et al. [18] published a new
way of scoring this questionnaire. In a population of
individuals still in the work force despite low back
pain, they were able to classify each individual into
four groups based on their degree of disability and the
presence of emotional distress. The degree of disability
and emotional distress obtained from the classification
showed strong correlation to questions from the Fear-
Avoidance Beliefs Questionnaire [25], the Coping
Strategies Questionnaire [20], and the Nottingham
Health Profile [12].
As the degree of disability, and especially degree of
emotional distress, have been shown to be strong pre-
dictors of negative outcome after lumbar surgery, [9, 16,
24] we wanted to test whether or not this new classifi-
cation would demonstrate predictive value in relation to
outcome after lumbar spinal fusion surgery.
Materials and methods
Classification
The DPQ was the main outcome instrument used.
Initially, the development of the classification was
sought to be reproduced using the same methods as in
the original publication [18] (Appendix). The result of
these analyses led to the decision to continue the
investigation using the original classification. The
classification was calculated as described by Ozguler
et al. [18] (Appendix). The classification defines four
groups characterized as follows: group 1, slight dis-
ability; group 2, intermediate disability; group 3, major
disability; group 4, major disability and emotional
distress.
In order for the classification to be calculated, it is a
prerequisite that all 16 questions in the DPQ have been
answered.
Patients
Patients with fully answered preoperative and postop-
erative DPQ were identified in the department’s patient
database. Five hundred and sixty-six patients operated
between March 1992 and January 2002 were identified.
Mean age was 46 years, ranging from 18 to 81 years.
They consisted of 239 men (mean age 46 years ranging
from 18 to 81 years) and 327 women (mean age
47 years, ranging from 18 to 80 years). Indications for
surgery were spondylolisthesis (32%) or degenerative
disorders (78%). Fifty-eight percent of those with
degenerative disorders had had previous spinal surgery.
Eighty-one percent had had low back pain for more
than 2 years. Twenty-one percent of the patients had
radicular symptoms. A full description of the study
sample is seen in Table 1. Fully answered preoperative
DPQ was available for all patients. A fully answered
DPQ was available after a minimum of 1-year follow-
up in all patients. Seventy percent had completed
2 years of follow-up (400 patients). However, 59 of
those required use of 1-year data due to incomplete
2-year data.
Information on age, gender, operative indication, and
work status was provided in 100% of the cases, on pain
duration in 97% (n=550) and on the presence of radi-
ating symptoms in 98% (n=552). This results in 550
patients eligible for entry into logistic regression analy-
sis.
A total of 55 patients (24 men and 31 women, mean
age 49 years, range 23–71 years) were ineligible for
inclusion because of incomplete preoperative DPQ.
They were compared to the included group with respect
to age, gender, operative indication and work status in
order to investigate any patient characteristic associated
with not understanding/answering all questions in the
DPQ. For the same purpose, included patients with an
incomplete DPQ at one of their two follow-ups were
compared with those patients with no missing DPQ re-
sponses from the included group. This group consisted
of 21 patients (9 men and 12 women, mean age 52 years,
range 24–75 years).
 1673

Table 1 The preoperative distribution of patients in the four classification groups in relation to the predictor variables

Preoperative classification Total

Group 1 Group 2 Group 3 Group 4

Gendera
Male 1% (3) 18% (44) 36% (85) 45% (107) 42% (239)
Female 1% (3) 11% (35) 37% (122) 52% (167) 58% (327)
Indicationb
Listhesis 2% (4) 25% (44) 39% (69) 35% (62) 32% (179)
Primary degeneration 1% (2) 16% (26) 33% (54) 49% (80) 29% (162)
Secondary degeneration 0% (0) 4% (9) 37% (84) 59% (132) 40% (225)
Pain durationc
< 1 year 5% (1) 0% (0) 20% (4) 75% (15) 4% (20)
1–2 years 1% (1) 12% (10) 42% (35) 46% (37) 15% (83)
> 2 years 1% (4) 15% (67) 36% (161) 49% (215) 81% (447)
Radiating paind
No 3% (4) 25% (29) 28% (33) 43% (50) 21% (116)
Yes 1% (2) 11% (48) 39% (168) 50% (218) 79% (436)
Work statuse
Working 3% (5) 24% (40) 45% (75) 28% (46) 29% (166)
Without work/sick leave 0% (1) 11% (25) 34% (77) 55% (125) 40% (227)
Retired/pensioned 0% (0) 9% (16) 32% (55) 60% (103) 31% (173)
Agef
)39 years 2% (3) 23% (34) 31% (46) 44% (66) 26% (149)
40–59 years 1% (3) 9% (31) 40% (141) 50% (176) 62% (351)
60 years 0% (0) 21% (14) 30% (20) 49% (32) 12% (66)
Mean age (range)g 36 (18–53) 44 (18–71) 47 (19–81) 47 (18–80) 46 (18–81)
Total 1% (6) 14% (79) 37% (207) 48% (274)
a
There is a near-significance association between gender and preoperative classification (P=0.064) with a larger proportion of women in
the more disabled groups (P=0.035)
b
There is significant association between indication and preoperative classification group (P<0.0005) with a larger proportion of
degenerated in the more disabled groups (P<0.0005)
c
There is a near-significance association between duration of pain and preoperative classification (P=0.068)
d
Significant association between radiating pain and classification (P<0.0005) with a larger proportion of disabled among those with
radiating pain (P=0.007)
e
Significant association between work status and classification (P<0.0005) with more disabled among retired/without work (P<0.0005)
f
Significant association between age group and classification (P<0.0005) but with no trend (P=0.113)
g
Significant difference in age between classification groups (P=0.038)

Outcome
Outcome parameters were the classification at follow-up
as well as the scores of the four categories of the DPQ.
For logistic regression, the classification at follow-up
was dichotomized in a high disability category, consist-
ing of groups 3 and 4, and a low disability category,
consisting of groups 1 and 2. The high disability cate-
gory was considered the ‘‘negative outcome’’ category.
To investigate factors with special importance to the
presence of emotional distress at follow-up, the classifi-
cation at follow-up was also dichotomized into a ±
distress variable (group 4 vs. groups 1–3) with the
presence of distress being the ‘‘negative outcome’’ cate-
gory. Also, the improvement in classification group was
calculated; this was done by subtracting the follow-up
group (1–4) from the preoperative group (1–4).
Furthermore, the patients were asked at follow-up
whether they were willing to go through the operation
again now that they knew the result. They could answer
either ‘yes’ or ‘no’. This was considered a subjective
evaluation of the outcome by the patient. A negative
answer to this question was considered a negative out-
come. This variable was also used for logistic regression
analysis. Ninety-eight percent (n=552) had answered
this question.
Predictive variables
Predictive variables of interest were gender, operative
indication, duration of pain, presence of radiating pain,
work status, age and preoperative classification. Fur-
thermore, the length of follow-up, either 1 or 2 years,
and the type of operation, were checked for any
influence. The type of operation was either uninstru-
mented posteolateral fusion, instrumented posterolat-
eral fusion or 360-fusion. The operative indication was
1674

spondylolisthesis and primary or secondary degenera- Statistics

tion as defined by Christensen et al. [6]. Primary and
secondary degeneration denote degenerative changes to
the spine with secondary degeneration being degener-
ative changes due to previous back surgery. Pain
duration was recorded as less than 1 year, from 1 to
2 years or more than 2 years. Work status was defined
as either working, being without work or on sick leave,
or being retired or on pension. Age was grouped as 18–
39, 40–59, or 60 years and above. As there were only
six patients in the classification group 1 preoperatively,
groups 1 and 2 were combined for the logistic regres-
sion analysis.
Association among groups was tested with chi-square
statistics, after analysis of standardized residuals. Test
for trends in proportions was performed using Spear-
mans q. Difference in means between groups was tested
with two-sample t-test, one-way ANOVA with Bonfer-
roni post hoc analysis or Kruskal–Wallis test, whichever
was appropriate.
Logistic regression was performed as stepwise back-
ward regression using likelihood ratio and a cut-point of
0.1 for staying in the model. Model fit and covariate
patterns were analyzed according to Hosmer and

Table 2 The postoperative distribution of patients in the four classification groups in relation to the seven predictor variables

Follow-up classification Total

Group 1 Group 2 Group 3 Group 4

Gendera
Male 34% (81) 17% (40) 17% (40) 33% (78) 42% (239)
Female 20% (66) 20% (65) 29% (94) 31% (102) 58% (327)
Indicationb
Listhesis 39% (69) 20% (36) 16% (29) 25% (45) 32% (179)
Primary degeneration 25% (41) 21% (34) 23% (37) 31% (50) 29% (162)
Secondary degeneration 16% (37) 16% (35) 30% (68) 38% (85) 40% (225)
Pain duration
< 1 year 35% (7) 15% (3) 14% (3) 38% (8) 4% (21)
1–2 years 25% (21) 21% (17) 31% (26) 24% (20) 15% (85)
> 2 years 25% (113) 19% (84) 23% (103) 33% (149) 81% (450)
Radiating painc
No 31% (36) 25% (29) 18% (21) 26% (30) 21% (116)
Yes 24% (105) 17% (75) 26% (111) 33% (145) 79% (436)
Work statusd
Working 42% (70) 25% (42) 16% (26) 17% (28) 29% (166)
Without work/sick leave 23% (51) 16% (36) 25% (56) 37% (84) 40% (227)
Retired/pensioned 15% (26) 16% (27) 30% (52) 39% (68) 31% (173)
Preoperative classificatione
Group 1 100% (6) 0% (0) 0% (0) 0% (0)
Group 2 61% (48) 19% (15) 11% (9) 9% (7)
Group 3 28% (58) 22% (45) 30% (61) 21% (43)
Group 4 13% (35) 16% (45) 23% (64) 47% (130)
Length of follow-up
1 year 26% (59) 19% (43) 21% (47) 34% (76)
2 years 26% (88) 18% (62) 26% (87) 31% (104)
Agef
)39 years 35% (52) 20% (30) 20% (29) 25% (38) 26% (149)
40–59 years 22% (76) 18% (63) 26% (91) 35% (121) 62% (351)
60 years 29% (19) 18% (12) 21% (14) 32% (21) 12% (66)
Mean age (range) 45 (18–81) 46 (20–71) 47 (24–80) 47 (18–78) 46 (18–81)
Total 26% (147) 19% (105) 24% (134) 32% (180)
a
There is a significant association between gender and follow-up classification (P<0.0005) with a larger proportion of women being
slightly more disabled (P=0.039)
b
There is significant association between indication and classification group (P<0.0005) with a larger proportion of degenerated in the
more disabled groups (P<0.0005)
c
Significant association between radiating pain and classification (P=0.039) with a larger proportion of disabled among those with
radiating pain (P=0.020)
d
Significant association between work status and classification (P<0.0005) with more disabled among retired/without work (P<0.0005)
e
Significant linear association between pre- and post-operative classification (P<0.0005)
f
Near-significant association between age group and classification (P=0.057) with a larger proportion of severely disabled at follow-up
with increasing age (P=0.027)
 1675

Lemeshow [7, 11]. If one or more design variables were Results

excluded, the remaining variable is recoded into a binary
variable. The OR given is for the comparison of that
category compared to the total group of patients not
fitting into that category.
Effect sizes were calculated using Cohen’s method as
described by Taylor et al. [22]. As some data tended to
be skewed, median and interquartile range were used
instead of mean and standard deviation [22], thus effect
size = (median preoperative score ) median postoper-
ative score)/interquartile range of the change score of the
group. Effect sizes of 0.2 are considered small, 0.5
moderate, and 0.8 or greater large.
Classification
Distribution of the patients in the different classification
groups is seen in Table 1. This table also shows the
relation between some of the predictive values and the
preoperative classification. As can be seen, 85% of the
patients were classified as heavily disabled (groups 3 and
4), preoperatively. There was significant or near signifi-
cant association between belonging to the more disabled
groups and: being female, suffering from degeneration
(especially secondary degeneration), having had pain for

120 120
Daily activity Work/leisure activity
100 100
Follow-up DPQ score

Follow-up DPQ score

80 80

60 60

40 40

20 20

0 p<0.0005 p<0.0005 0 p<0.0005 p<0.0005

p<0.0005 p<0.0005

1+2 3 4 1+2 3 4
Preoperative classification group Preoperative classification group

120 120
Anxiety/depression Social interest
100 100
Follow-up DPQ score

Follow-up DPQ score

80 80

60 60

40 40

20 20

0 p=0.002 p<0.0005 0 p<0.0005 p<0.0005

p<0.0005 p<0.0005

1+2 3 4 1+2 3 4
Preoperative classification group Preoperative classification group
Fig. 1 Boxplot showing the follow-up score in the four categories defines the interquartile range with line at the median. Error bars
of the DPQ according to the preoperative classification. All groups define the 10th and 90th percentiles, circles are outliers of this
are significantly different in all four DPQ-categories. The box interval
Table 3 Risk factors for disability at follow-up
1676

Risk factor OR 95% CI P valuee

(a) Indication
Secondary degenerationd 1.61 1.08 – 2.40 0.020
Pain duration
< 1 year 1c
1–2 years 2.70 0.92 – 7.95 0.071
> 2 years 2.32 0.87 – 6.17 0.093
Radiating pain
Yes 1.51 0.94 – 2.43 0.091
Work status
c
Working 1
Without work/sick-leave 2.84 1.78 – 4.53 <0.0005
Retired/pensioned 3.48 2.08 – 5.82 <0.0005
Preoperative classification
Group 1+2 1c
Group 3 2.69 1.40 – 5.18 0.003
Group 4 5.53 2.89 – 10.59 <0.0005
Age
40–59 yearsd 1.68 1.13 – 2.49 0.011
Sensitivity 83.3%, specificity 56.7% Positive predictive False positive predictions if true False positive if predicted
value 70.6% negative 43.3% positive 29.4%
Cut-off 0.5, ROC-area 0.76, Negative predictive False negative predictions if true False negative if predicted
correctly classified 71.5% value 73.2% positive 16.7% negative 26.8%
(b) Work status
Working 1c
Without work/sick-leave 1.99 1.19 – 3.35 0.009
Retired/pensioned 2.06 1.20 – 3.54 0.009
Preoperative classification
Group 1+2 1c
Group 3 2.51 1.07 – 5.90 0.034
Group 4 7.82 3.43 – 17.82 <0.0005
Sensitivity 30.5% Specificity 87.8% Positive predictive value 53.5% False positive predictions if true False positive if predicted
negative 12.2% positive 46.5%
Cut-off 0.5, ROC-area 0.72, Negative predictive value 73.2% False negative predictions if true False negative if predicted
correctly classified 69.6% positive 69.5% negative 26.8%
a
Odds ratios for negative outcome, measured as belonging to group 3 or 4 at follow-up, for the predictor variables, which remained in the model, using a cut-point of 0.1 for
staying in the model. Indication and age has been recoded into binary variables (secondary degeneration vs. listhesis/primary degeneration and age 40–59 years vs. age below
40 years or 60 years and above). Summary statistics for the logistic regression model is presented at the bottom
b
Odds ratios for negative outcome, measured as the presence of distress at follow-up (belonging to group 4), for the predictor variables, which remained in the model, using a
cut-point of 0.1 for staying in the model. Summary statistics for the logistic regression model is presented at the bottom
c
Denominator (reference group) of following odds ratios
d
Recoded into binary variable
e
Associated with the two-tailed test that OR=1
OR Odds ratio; C confidence Interval
 1677

more than 2 years, having radiating symptoms, not
being in the work force, and being in the age group from
40 to 59 years.
Association between follow-up classification and
predictive variables is seen in Table 2. Forty-five percent
of patients were now in groups 1 and 2 as compared to
only 15% before operation. There were slightly more
women than men in the more disabled groups after
surgery. Surgical indication, presence of radiating
symptoms, and work status were also associated with
the postoperative classification. The preoperative clas-
sification was also strongly associated with the postop-
erative classification. There was no difference in
postoperative classification between those with only 1-
year follow-up as compared to those with a full 2-year
follow-up.
Those preoperatively classified in the more disabled
groups also had significantly higher DPQ-scores at fol-
low-up (Fig. 1).
Risk factors for disability and distress
Variables found to be of significant or near-significant
influence in the logistic regression models are seen in
Table 3. With disability as outcome, the most important
variables were preoperative classification and work sta-
tus. As compared with the patients in groups 1 and 2
preoperatively, placement in groups 3 or 4 increased the
risk of being severely disabled at follow-up with OR 2.69
and OR 5.53, respectively. In comparison with the
working patients, those without work or on sick leave

Table 4 The patients subjective evaluation at follow-up according to the predictor variables.

Subjective evaluation Total

‘‘Yes’’ ‘‘No’’

Gendera
Male 80% (188) 20% (46) 42% (234)
Female 73% (233) 27% (85) 58% (318)
Indicationb
Listhesis 79% (140) 21% (37) 32% (177)
Primary degeneration 82% (130) 18% (28) 29% (158)
Secundary degeneration 70% (151) 30% (66) 39% (217)
Pain duration
< 1 year 84% (16) 16% (3) 4% (19)
1–2 years 83% (67) 17% (14) 15% (81)
> 2 years 75% (329) 25% (109) 81% (438)
Irradiating pain
No 81% (94) 19% (22) 21% (116)
Yes 75% (319) 25% (105) 79% (424)
Work statusc
Working 85% (138) 15% (25) 30% (163)
Without work/sick-leave 72% (160) 28% (62) 40% (222)
Retired/pensioned 74% (123) 26% (44) 30% (167)
Preoperative classificationd
Group 1 100% (6) 0% (0) 1% (6)
Group 2 85% (67) 15% (12) 14% (79)
Group 3 78% (156) 22% (45) 36% (201)
Group 4 72% (192) 28% (74) 48% (266)
Length of follow-up
1 year 72% (155) 28% (59) 39% (214)
2 years 79% (266) 21% (72) 61% (338)
Age
)39 years 77% (114) 23% (34) 27% (148)
40–59 years 75% (257) 25% (84) 62% (341)
60- years 79% (50) 21% (13) 11% (63)
Mean age (range) 46 (18–81) 46 (18–80) 46 (18–81)
Total 76% (421) 24% (131)
a
There is a near-significant association between gender and answer (P=0.054)
b
There is significant association between indication and answer (P=0.010) with a larger proportion of ‘‘No’’ answers in the secondary
degeneration group (P=0.0174)
c
Significant association between work status and answer (P=0.010) with a larger proportion of ‘‘Yes’’ answers in the working group
(P=0.0198)
d
There is a near-significant association between answer and preoperative classification (P=0.052) with more ‘‘No’’ answers in the more
disabled groups (P=0.0101). This becomes significant if group 1 and 2 is pooled together (P=0.030/P=0.0109) (data not shown)
1678
and those retired/pensioned had an increased risk of
heavy disability at follow-up with OR 2.84 and OR 3.48,
respectively. Other significant risk factors were second-
ary degeneration and age between 40 and 59 years. Both
lengthier pain duration and the presence of radiating
symptoms increased the risk of being severely disabled at
follow-up, but only with near-significant ORs (Table 3).
Using emotional distress as an outcome variable, the
only variables remaining in the logistic regression model
were preoperative classification and work status (Ta-
ble 3). Being emotionally distressed preoperatively was
the largest risk factor for being so at follow-up. Major
disability and emotional distress (group 4) with OR 7.82
as compared to major disability only (group 3) with OR
2.51. Being out of the work force doubled the risk of
being distressed at follow-up (Table 3).
Risk factors for negative patient evaluation
Seventy-six percent of the canvassed patients stated that
they, now familiar with the result of their procedures,
would be willing to undergo the procedure again.
Associations between the predictive variables and the
subjective evaluation are seen in Table 4. Gender,
operative indication, work status and preoperative
classifications are the variables with significant or near
significant influence on the response. Using the logistic
regression model, operative indication, work status and
preoperative classification stayed in the model together
with duration of pain and length of follow-up; all ORs
were, however, modest in size. Secondary degeneration
increased the risk of a negative answer as did being
without work or on sick leave at the time of operation
and being classified as having distress (belonging to
group 4), preoperatively. Follow-up length of 1 year
compared to 2 years increased the risk of a negative
answer (Table 5).
Change in classification group
The classification of seven patients (1%) from group 2
dropped to group 4. A total of 52 patients (9%) dropped
1 group: (9 patients from group 2 to group 3 and 43
patients from group 3 to 4). Two hundred and twelve
patients (28%) did not change classification group. They
consisted of the 6 patients from group 1 (3%), 15 pa-
tients from group 2 (7%), 61 patients from group 3
(29%), and 130 patients from group 4 (61%). One
hundred and fifty-seven patients (28%) improved one
classification group: 48 from group 2, 45 from group 3,
and 64 from group 4. One hundred and three patients
(18%) improved two classification groups: 58 from
group 3 and 45 from group 4. A total of 35 patients (6%)
improved three groups from group 4 to group 1. The
only variables significantly associated with group
improvement were work status (P<0.0005) and preop-
erative classification (P<0.0005). In particular, there
was no association between either operation type or
length of follow-up and improvement in classification
group.
Comparison of improvement in classification group
with improvement in the continuous DPQ-score in the
four categories showed moderate to large effect sizes for
the changes in classification group. However, small ef-
fect sizes were observed in some of the categories in the
group with unchanged classification (Fig. 2). Using the
subjective evaluation of the patient as outcome, the four
continuous DPQ-scales demonstrated larger effect sizes
than the change in classification group demonstrated
(Table 6). ROC-analysis with one as cut point for
improvement and the subjective evaluation as true out-
come, results in a sensitivity of 62.7%, a specificity of
77.9% and 66.3% being correctly classified. Area under
the curve is 0.75.
Analysis of excluded patients
Characteristics with regard to age, gender, indication
and work status of those patients lacking a fully com-
pleted DPQ, either preoperatively or at follow-up, are
seen in Table 7. The only significant difference among
the groups is that the group with missing data is chro-
nologically older. However, when looking at the distri-
bution in the three age groups, this seems to be
attributable to a larger proportion of patients in the age
group 40–59 (71 vs. 62%). Although this was a nonsig-
nificant finding, the incomplete DPQ groups, both those
lacking preoperative data and those lacking follow-up
data, were characterised by a larger proportion of pa-
tients with primary degeneration as well as a larger
proportion of patients who had retired or were pen-
sioners.
Discussion
In this study, we evaluated the classification of the
DPQ devised by Ozguler et al. [18] both as a predictive
and as an outcome variable in a large sample of
lumbar spinal fusion patients. We chose the original
classification, although we could not replicate the re-
sults used for the development in our patient sample,
as it was developed in a large cohort with a broader
range of disability (Appendix). Used as a predictive
parameter, its key value is, in our opinion, the ability
to isolate patients with a considerable degree of
emotional distress as a component of their overall
disability. Although we had no questionnaire
concerning psychological characteristics for direct
 1679

comparison included, we reached similar numbers and

False negative if predicted

Odds ratios for negative outcome, using the subjective evaluation of the patient, for the predictor variables, which remained in the model, using a cut-point of 0.1 for staying in

degeneration, duration of pain >2 years vs. duration <1 year/1–2 years, being without work or on sick-leave vs. working/retired and preoperatively classified in group 4 vs.
the model. Indication, duration of pain, work status and preoperative classification has been recoded into binary variables (secondary degeneration vs. listhesis/primary
False positive if predicted
results with regards to distress as compared with sev-
eral other studies using different psychological ques-
tionnaires for evaluation of spinal surgery patients.

negative 18.4%
positive 69.7%
The Distress and Risk Assessment Method (DRAM)
[16] demonstrates that distress is present preoperatively
P valuea

in 40–48% of spinal surgery patients [10, 21]. As we

0.007

0.083

0.028

0.092

0.070
had 48% of the patients in the distressed group pre-
operatively, we are within the range of the published
data that employs the DRAM classification.
We found that the presence of distress had significant
predictive value in our logistic regression models,
regardless of how outcome was defined. In both studies
using the DRAM [10, 21], classification as distressed was
associated with higher disability, as measured with the
Oswestry Disability Index (ODI), but it did not have any
False negative predictions
False positive predictions

predictive value in terms of outcome. In the original

if true negative 38.6%

if true positive 45.2%

study on the DRAM by Main et al. [16] it was found to

be a strong predictor for outcome in lumbar surgery.
This is also supported by Trief et al. [24] in a study
comprising 102 patients undergoing various forms of
1.17 – 2.70

0.94 – 2.96

1.05 – 2.39

0.94 – 2.17

0.97 – 2.23

lumbar surgery. Using the DPQ as an outcome instru-

95% CI

ment to investigate psychological predictors of lumbar

surgery outcome, they found that pain duration and
DRAM classification were significant predictors of
improvement in the daily activity and work-leisure
activity categories. Using data from the Swedish Lum-
bar Spine Study, Hagg et al. [9], scrutinised several risk
factors for poor outcome. In their study, the only sig-
nificant risk factors found in the surgically treated group
were neurotic personality and >50% reduction in disc
Summary statistics for the logistic regression model is presented at the bottom
Negative predictive

height prior to operation when patient improvement was

Positive predictive

used as an outcome variable. When using reduction in

value 30.3%

value 81.6%

the ODI as an outcome variable, the only significant risk

Table 5 Risk factors for a negative patient evaluation at follow-up

factor was neuroticism. Their variable neuroticism/neu-

rotic personality classification is based on a Swedish-
1.78

1.66

1.58

1.43

1.47

developed personality inventory and it might very well

OR

reflect many of the same psychological parameters as

those usually incorporated in the term ‘‘distress’’. Simi-
larly, in a study of 109 inpatients with chronic low back
pain treated both conservatively and surgically, Nickel
Associated with the two-tailed test that OR=1

OR = Odds Ratio; CI = Confidence Interval

et al. [17] found ‘‘somatisation’’ and ‘‘doctor shopping’’

to be the only two variables with significant influence on
outcome. The variable ‘‘somatisation’’ was based on a
questionnaire screening for somatoform disorders,
Sensitivity 54.8% Specificity 61.4%

whereas the ‘‘doctor shopping’’ variable represented the

Recoded into binary variable

number of physicians consulted before the present

Cut-off 0.25, ROC-area 0.63
Correctly classified 63.3%

treatment. However, only 38% of the patients were

Without work/sick-leaveb
Preoperative classification
Secundary degenerationb

treated surgically, which might explain the lack of other

classified in group 1–3)

explanatory variables. In a study of 163 patients, Junge

et al. [14] investigated a large number of variables as
possible predictors of good and bad outcome in lumbar
Pain duration

Work status
> 2 yearsb
Risk factor

spinal surgery. They found a combination of the vari-

FU length

Group 4b
Indication

ables: Laségue test, number of previous back operations,

1 year

number of other pain locations and ‘‘general avoidance’’

behaviour to provide good predictability. Several of
b
a
1680

120 120
Daily activity Work/leisure activity
100 100
80 80
Improvement in DPQ

Improvement in DPQ
60 60
40 40
20 20
0 0
-20 -20
-40 -40
-60 -60
-80 -80
-1.5 -0.4 0.3 1.1 2.0 2.3 -0.7 -0.6 0.0 1.5 1.7 1.6
-100 -100
-2 -1 0 1 2 3 -2 -1 0 1 2 3
Improvement in classification group Improvement in classification group

120 120
Anxiety/depression Social interest
100 100
80 80
Improvement in DPQ

Improvement in DPQ

60 60
40 40
20 20
0 0
-20 -20
-40 -40
-60 -60
-80 -80
-0.4 -0.9 0.2 0.6 1.8 2.0 -1.0 -0.9 -0.1 0.5 1.2 1.8
-100 -100
-2 -1 0 1 2 3 -2 -1 0 1 2 3
Improvement in classification group Improvement in classification group
Fig. 2 Boxplot showing the improvement in the DPQ score from change in classification group. The box defines the interquartile
preoperatively to follow-up according to the improvement in range with line at the median. Error bars define the 10th and 90th
classification group. Numbers in bold are the effect sizes for the percentiles, circles are outliers of this interval

these variables are similar to those we have found to be
significantly influential. We did not perform the Laségue
test, but we did find the presence of radiating pain to be
a variable with near-significant influence in our logistic
regression model. Similarly, we found the diagnosis
‘‘secondary degeneration’’ to be significant. Most
importantly, they found psychological characteristics,
manifested in reported aggregates of pain locations and
‘‘general avoidance’’ behaviour, to be of significant
influence. The extent to which these variables overlap
with what is denoted ‘‘distress’’ in this study is uncertain,
but it seems reasonable to infer that there is a consid-
erable overlap. In this context, our data supports the
hypothesis that emotional distress increases the risk of a
poor outcome in patients undergoing lumbar spinal
surgery.
The variable which we found to be the next strongest
predictor after classification was work status: the com-
parable variable in the Swedish study [9], was found to
be influential only when considering return to work and
not when investigating outcome. ‘‘Worker compensa-
tion’’ was, however, significant as a univariate predictor
when looking at patient global assessment, but unem-
ployment, sick leave due to low back pain and disability
pension were not. Unfortunately it was not possible for
us to pinpoint which of the patients who had retired or
 1681

Table 6 Effect sizes for DPQ-scores and change in classification group.

’’Yes’’ Effect size ’’No’’ Effect size

Improvement Median (IQR) Improvement Median (IQR)

Daily activity 24 (6–39) 0.73 )3 ()15–12) )0.11

Work-leisure activity 25 (5–45) 0.88 0 ()10 –15) 0
Anxiety-depression 15 (0–35) 0.71 0 ()20–15) )0.14
Social interest 10 (0–30) 0.67 )5 ()25–15) 0
Improvement in classification group 1 (0–2) 0.5 0 ()1–0) 0

The range of improvement (median and interquartile range) in the four DPQ categories and in classification group according to the
subjective evaluation of the patient (answer to the question ‘‘would you be willing to go through the operation again now that you know
the result?’’) with the corresponding effect sizes

were on pension had had ongoing or terminated legal
cases related to their back problems. Thus the explana-
tion for the variable ‘‘retired/pensioned’’ being such a
strong predictor might be an overweight of legal cases in
this category.
The four-group classification, when used as an out-
come parameter, has the advantage of providing a
combined score for the four categories, both when the
classification is used directly and when improvement in
classification group is used. This gives an ordinal out-
come parameter that facilitates dichotomisation and
thus entry into, e.g. logistic regression models, as in this
study. The change in classification group was very good
at identifying patients who had improved or worsened in
all of the four continuous DPQ-scores with effect sizes
that were moderate to large. Small effect sizes were,
however, observed in some of the categories in the group
with unchanged classification. This suggests that
improvement in classification group is not sensitive en-
ough to detect small improvements. Using the subjective
evaluation as standard, the change in classification was
not as responsive as the four continuous DPQ-scores in
determining improvement. Using a cohort of patients
treated for low back pain, Taylor et al. [22] investigated
the responsiveness of several questionnaires, including
the ODI and the SF-36. They also found small effect
sizes in the group of patients rating themselves un-
changed. However, only a subgroup of the cohort had
undergone surgical treatment. The effect sizes for
improvement were similar or slightly better in compar-
ison to those obtained for the four DPQ-scales. In a
similar study, Walsh et al. [26] obtained effect sizes
within the same range as those we obtained.
As this study uses a sample of patients characterized
by the fact that they have completed a full DPQ, bias
cannot be excluded. Reasons for bias could be specu-
lated to be either selection of a patient sample with a
high degree of compliance or selection of patients with a
substantial level of education and/or intelligence, the
latter being inferred from the fact that they were able to
answer all of the questions in the DPQ. However, our
analysis of incomplete cases seems to demonstrate that
the patients most likely to skip some of the questions in
the DPQ were those between 40 and 59 years and who

Table 7 Differences between the included patients and those excluded because of lacking information in their DPQ.

Preoperative P value Follow-up P value

Missing Complete Missing Complete

Gender
Male 44% (24) 42% (239) 0.840 43% (9) 42% (230) 0.952
Female 56% (31) 58% (327) 57% (12) 58% (315)
Indication
Listhesis 24% (13) 32% (179) 0.116 19% (4) 32% (175) 0.131
Primary degeneration 42% (23) 29% (162) 48% (10) 28% (152)
Secundary degeneration 35% (19) 40% (225) 33% (7) 40% (218)
Work status
Working 29% (16) 29% (166) 0.286 33% (7) 29% (159) 0.536
Without work/sick-leave 31% (17) 40% (227) 29% (6) 41% (221)
Retired/pensioned 40% (22) 31% (173) 38% (8) 30% (165)
Age
)39 years 15% (8) 26% (149) 0.155 10% (2) 27% (147) 0.160
40–59 years 71% (39) 62% (351) 71% (15) 62% (336)
60 years 15% (8) 12% (66) 19% (4) 11% (62)
Mean age (range) 49 (23–71) 46 (18–81) 0.0532 52 (24–75) 46 (18–81) 0.0304
1682
have already retired or were on early pension. It is often
among this group that we find the worst cases regarding
outcome; so, a bias towards a patient group with better
results cannot be totally excluded. However, in our point
of view, the numbers are so relatively small that the
possibility of bias should not discourage use of the
classification.
In conclusion, we have evaluated a classification
based on the DPQ. As an outcome parameter the clas-
sification is not superior to the original DPQ-scales, but
as it transforms the four continuous scales into a com-
bined ordinal outcome parameter, it is, in our opinion, a
useful adjunct to the four original scales. We have
demonstrated that the classification, as a predictive
parameter, is able to identify lumbar spinal fusion pa-
tients with a considerable amount of psychological dis-
tress in their symptomatology. Furthermore, the
presence of distress, as determined by this classification,
was a highly significant risk factor for inferior outcome.
We believe the classification could be a tool in the pre-
operative evaluation of patients, not as a stand-alone
selection criterion, but as additional information serving
to assist in the identification of patients who could
benefit from further exploration of eventual psychoso-
cial problems before they undergo surgery.
Appendix: Development, reproducibility and choice
of classification scheme
Original scoring of the classification
First, the original continuous visual analogue scale
ranging from 0 to 140 was divided and transformed into
a numerical scale ranging from 0 to 10. Next, two
summary variables, Sum1 and Sum2, were created. The
first is calculated as the sum of the numerical scores of
question 1 to question 10. The second was calculated as
the sum of the numerical scores of question 11 to
question 16. Based on these two sum scores, the groups
are defined as:
Group 4: Sum1 + (3·Sum2) ‡ 132
Group 3: Not belonging to group 4 and Sum1 +
(1/2)·Sum2 ‡ 49
Group 2: Not belonging to group 4 or 3 and
Sum1+(1/3)·Sum2 ‡ 22
Group 1: Sum1 + (1/3)Sum2 < 22.
Principal component and hierarchical cluster analysis
Using SPSS for Windows 11.0, a principal component
analysis of the 16 items from the DPQ was performed
extracting two factors. After the VAS-based score had
been transformed, we used the raw data with no
weighting so that the scores would mimic a 0–10
numerical scale.
The loading plot derived from the analysis is seen in
Fig. 3. It closely resembles the loading plot shown in the
article by Ozguler et al. with the exception of item 15,
which is the question regarding social support.
Hierarchical cluster analysis was performed using the
individual values of the two extracted factors. The
analysis was set to extract four clusters using the
ward method. The four clusters were relatively dis-
tinct and could be separated visually by three straight
lines (Fig. 4). This resulted in classification of the
groups in accordance with the following rules:
Group 1: Sum1 2 23 Sum2 > 14 23 & Sum1 56 Sum2\43 13
Group 2: Sum1 2 23 Sum2 > 14 23 & Sum1 56 Sum2>43 13
Group 3: Sum1 2 23 Sum26 14 23 & Sum1 56 Sum2 \35
Group 4: Sum1 2 23 Sum26 14 23 & Sum1 56 Sum2>35
This classification resulted in a misclassification of
11.5% of the patients, but with good agreement between
the cluster analysis classes and the practical classifica-
tion (Table 8). When compared with the classification
by Ozguler et al. there was very poor agreement
(Table 9).
Differences between the two classification schemes
As compared to the result obtained by Ozguler et al.,
we found a similar result in the principal component
Fig. 3 Loading plot of the results from the principal component
analysis. Each item in the DPQ (d1–d16) is plotted against its
correlation with the two extracted factors
 1683

Fig. 4 Distribution of the pa-

tients in relation to the two
summary scores. The symbols
denote which cluster class the
patient belongs to. The straight
lines represent the rules defining
the practical classification

analysis with one exception: while question 15 regarding
social support correlated positively with both compo-
nents, it correlated weakly with component 2. Ozguler
et al. found it to correlate negatively with the second
component. This implies that it might not reflect emo-
tional capacities but more likely physical capacities. As
it is included in the calculation of the second summary
score, this might be one reason for the difference ob-
tained in our hierarchical cluster analysis as compared to
the result of Ozguler et al. The distribution of the four
clusters was, however, very similar to that seen in Fig. 4
if the analysis was done based only on the other 15
questions in the DPQ (data not shown). Similarly, no
difference in the distribution was seen if the clusters were
plotted not according to Sum1 and Sum2 but instead to
the sums of questions 1–10 + 15 and questions 11–14 +
16 (data not shown). Thus, it seems unlikely that the
difference between the two classification schemes is fully
attributable to the different loading of question 15 in the
principal component analysis.
Another explanation for the difference could be that
Ozguler et al. developed their classification on a patient
sample with a broader range of disability. Comparing
Fig. 4 with the corresponding figure from the paper by
Ozguler et al., it is evident that our patient sample is
somewhat skewed toward the higher values of the two
sum scores. The lack of patients with low values of the two
summary scores in our sample could explain the different
results obtained by the hierarchical cluster analysis in our
study. Additionally, Ozguler et al. used the DPQ with
numerical 0–10 scales instead of the original VAS-scales
used in our version of the questionnaire.
Comparison of classification schemes
Using the classification by Ozguler et al., the four groups
differ due to a graduated increase in the value of the first
summary score (increased level of physical disability)
whereas the increase in the second summary score

Table 8 Distribution of patients according to classification and Table 9 Distribution of patients according to the classification by
cluster analysis based on our patient sample Ozguler et al. and the classification based on our analysis

Group 1 Group 2 Group 3 Group 4 Total Group 1 Group 2 Group 3 Group 4 Total

Cluster analysis, class 1 152 12 6 3 173 Ozguler et al., Group 16 0 0 0 6

Cluster analysis, class 2 15 181 0 1 197 Ozguler et al., Group 2 67 0 12 0 79
Cluster analysis, class 37 0 82 5 94 Ozguler et al., Group 3 82 106 19 0 207
Cluster analysis, class 40 10 6 86 102 Ozguler et al., Group 4 19 97 63 95 274
Total 174 203 94 95 566 Total 174 203 94 95 566

Agreement 88.5% Kappa coefficient 0.84 Agreement 21.2% Kappa coefficient 0.02
1684

(emotional capacities) is much lower except for those in
group 4 (In Fig. 2 in the paper by Ozguler et al., they are
classified by the line labelled C, which is almost hori-
zontal, displaying the addition of emotional disturbance
characterising this group). Using the classification based
on our sample, there is a more linear increase in both the
sum scores. Thus, it is not possible to isolate a group that
differs much in relation to emotional disturbance.
Using the subjective evaluation of the patient as
outcome, logistic regression was performed using the
same predictive variables as in the article, except for the
variable ‘‘preoperative classification’’, which was now
based on the classification developed using our patient
sample. The resulting ORs were almost identical, except
that when using the classification based on our patient
sample, preoperative classification did not remain in the
model (data not shown).
Choice of classification scheme
We chose to use the original classification despite the
fact that we could not reproduce the classification in
our patient sample. We did so because the original
classification was developed on a larger patient sample
with a broader range of disability; furthermore, we find
the interpretation of the original classification straigh-
ter and more intuitive in comparison with the classifi-
cation based on our analysis. We observed no
predictive value of the classification based on our
analysis with respect to subjective outcome after fusion
surgery. We are of the opinion that the original clas-
sification is robust, that it can handle differences due to
different language versions and that it gives meaningful
and valuable information also, in relation to spinal
fusion patients.

References

1. Christensen FB, Bunger CE (1997)
Retrograde ejaculation after retroperi-
toneal lower lumbar interbody fusion.
Int Orthop 21:176–180
2. Christensen FB, Karlsmose B, Hansen
ES, Bunger CE (1996) Radiological and
functional outcome after anterior lum-
bar interbody spinal fusion. Eur Spine J
5:293–298
3. Christensen FB, Thomsen K, Eiskjaer
SP, Gelinick J, Bunger CE (1998)
Functional outcome after posterolateral
spinal fusion using pedicle screws:
comparison between primary and sal-
vage procedure. Eur Spine J 7:321–327
4. Christensen FB, Laursen M, Gelineck J,
Hansen ES, Bunger CE (2001) Pos-
terolateral spinal fusion at unintended
levels due to bone-graft migration: no
effect on clinical outcome in 19/130
patients. Acta Orthop Scand 72:354–
358
5. Christensen FB, Hansen ES, Eiskjaer
SP, Hoy K, Helmig P, Neumann P,
Niedermann B, Bunger CE (2002)
Circumferential lumbar spinal fusion
with brantigan cage versus
posterolateral fusion with titanium
cotrel-dubousset instrumentation: a
prospective, randomized clinical study
of 146 patients. Spine 27:2674–2683
6. Christensen FB, Hansen ES, Laursen
M, Thomsen K, Bunger CE (2002)
Long-term functional outcome of pedi-
cle screw instrumentation as a support
for posterolateral spinal fusion: ran-
domized clinical study with a 5-year
follow-up. Spine 27:1269–1277
7. Dupont WD (2002) Statistical modeling
for biomedical researchers. A simple
introduction to the analysis of complex
data. University Press, Cambridge
8. Haas M, Jacobs GE, Raphael R,
Petzing K (1995) Low back pain
outcome measurement assessment in
chiropractic teaching clinics:
responsiveness and applicability of two
functional disability questionnaires.
J Manipulative Physiol Ther 18:79–87
9. Hagg O, Fritzell P, Ekselius L,
Nordwall A (2003) Predictors of
outcome in fusion surgery for chronic
low back pain. A report from the
Swedish Lumbar Spine Study. Eur
Spine J 12:22–33
10. Hobby JL, Lutchman LN, Powell JM,
Sharp DJ (2001) The distress and risk
assessment method (DRAM). J Bone
Joint Surg Br 83:19–21
11. Hosmer DW, Lemeshow S (2000)
Applied logistic regression. Wiley,
New York
12. Hunt SM, McKenna SP, McEwen J,
Backett EM, Williams J, Papp E (1980)
A quantitative approach to perceived
health status: a validation study.
J Epidemiol Community Health
34:281–286
13. Jamison J (2002) Chiropractic
management: beyond manual care.
J Manipulative Physiol Ther 25:332
14. Junge A, Frohlich M, Ahrens S,
Hasenbring M, Sandler A, Grob D,
Dvorak J (1996) Predictors of bad and
good outcome of lumbar spine surgery.
A prospective clinical study with
2 years’ follow up. Spine 21:1056–1064
15. Lawlis GF, Cuencas R, Selby D,
McCoy CE (1989) The development of
the Dallas Pain Questionnaire. An
assessment of the impact of spinal pain
on behavior. Spine 14:511–516
16. Main CJ, Wood PL, Hollis S,
Spanswick CC, Waddell G (1992) The
distress and risk assessment method. A
simple patient classification to identify
distress and evaluate the risk of poor
outcome. Spine 17:42–52
17. Nickel R, Egle UT, Rompe J, Eysel P,
Hoffmann SO (2002) Somatisation
predicts the outcome of treatment in
patients with low back pain. J Bone
Joint Surg Br 84:189–195
18. Ozguler A, Gueguen A, Leclerc A,
Landre MF, Piciotti M, Le Gall S,
Morel-Fatio M, Boureau F (2002)
Using the dallas pain questionnaire to
classify individuals with low back pain
in a working population. Spine
27:1783–1789
19. Piperno M, Hellio le Graverand MP,
Reboul P, Mathieu P, Tron AM, Perrin
G, Peschard MJ, Richard M, Vignon E
(1997) Phospholipase A2 activity in
herniated lumbar discs. Clinical corre-
lations and inhibition by piroxicam.
Spine 22:2061–2065

20. Rosenstiel AK, Keefe FJ (1983) The use
of coping strategies in chronic low back
pain patients: relationship to patient
characteristics and current adjustment.
Pain 17:33–44
21. Tandon V, Campbell F, Ross ER (1999)
Posterior lumbar interbody fusion
Association between disability and
psychological disturbance in noncom-
pensation patients. Spine 24:1833–1838
22. Taylor SJ, Taylor AE, Foy MA, Fogg
AJ (1999) Responsiveness of common
outcome measures for patients with low
back pain. Spine 24:1805–1812
23. Thomsen K, Christensen FB, Eiskjaer
SP, Hansen ES, Fruensgaard S, Bunger
CE (1997) 1997 Volvo Award winner in
clinical studies. The effect of pedicle
screw instrumentation on functional
outcome and fusion rates in postero-
lateral lumbar spinal fusion: a prospec-
tive, randomized clinical study. Spine
22:2813–2822
24. Trief PM, Grant W, Fredrickson B
(2000) A prospective study of psycho-
logical predictors of lumbar surgery
outcome. Spine 25:2616–2621
1685
25. Waddell G, Newton M, Henderson I,
Somerville D, Main CJ (1993) A fear-
avoidance beliefs questionnaire (FABQ)
and the role of fear-avoidance beliefs in
chronic low back pain and disability.
Pain 52:157–168
26. Walsh TL, Hanscom B, Lurie JD,
Weinstein JN (2003) Is a
condition-specific instrument for
patients with low back pain/leg
symptoms really necessary? The
responsiveness of the Oswestry
Disability Index, MODEMS, and the
SF-36. Spine 28:607–615