You are on page 1of 9

SPINE Volume 24, Number 23, pp 2497–2505

©1999, Lippincott Williams & Wilkins, Inc.

Personal Risk Factors for First-Time Low Back Pain

Michael A. Adams, PhD,* Anne F. Mannion, PhD,† and Patricia Dolan, PhD*

gene60) biochemistry (abnormal collagen), physical fac-


Study Design. A prospective study of personal risk tors (a long back), or psychology (depressive tendencies).
factors for first-time low back pain. Many previous investigations have concentrated on
Objectives. To construct and validate a multivariate
model to predict low back pain.
physical risk factors such as lumbar posture, mobility,
Summary of Background Data. Various physical and and anthropometry. Only the more recent studies have
psychological factors have been reported to increase the incorporated the important biasing effects of psychoso-
risk of low back pain, but conflicting results may be at- cial factors, which have a profound influence on all as-
tributable to inaccurate “clinical” measures and to poorly
pects of LBP behavior, including the reporting of it and
validated statistical models.
Methods. A total of 403 health care workers aged response to treatment.10,15,16,27,41,46
18 – 40 years volunteered for the study. None had any Even when all relevant factors are studied, it may be
history of “serious” back pain requiring medical attention difficult to prove that they actually cause LBP, because
or time off work. The volunteers completed the following personal characteristics such as reduced mobility and de-
questionnaires: the modified somatic perception ques-
tionnaire, the Zung depression scale, and the Health Lo- pression may themselves be a result of pain and ineffec-
cus of Control. Anthropometric factors were quantified tive treatment. The task of determining cause and effect
using standard techniques. The 3Space Isotrak device can be tackled only in prospective studies that link per-
(Polhemus, VT) was used to measure lumbar curvature sonal characteristics with future back pain. Several pro-
and hip and lumbar spine mobility. Leg and back strength
spective studies already have been performed,7,9,10,14 but
and back muscle fatiguability were measured in func-
tional postures. Postal follow-up questionnaires, sent af- their inconsistent and variable results may reflect meth-
ter 6, 12, 18, 24, 30, and 36 months, inquired about back odologic difficulties. Firstly, if a high proportion of study
pain, and multivariate logistic regression was used to patients do not return the questionnaires, then the results
identify risk factors at each follow-up. may be applicable only to a self-selected group. Sec-
Results. The response rate fell from 99% at 12 months
to 90% at 36 months, at which time 90 volunteers re- ondly, the sensitivity of a multivariate analysis to small
ported “serious” back pain and 266 reported “any” back risk factors can be reduced by the inclusion of exception-
pain. The following were consistent predictors of serious ally large risk factors such as previous back pain9,10,46 or
back pain: reduced range of lumbar lateral bending, a an occupation that demands heavy and awkward lift-
long back, reduced lumbar lordosis, increased psycholog-
ical distress, and previous nonserious low back pain. Only
ing.31,45,55,56,61 Merely omitting such factors from a
the latter three were consistent predictors of “any” back multivariate analysis would exaggerate the importance
pain. Physical factors had the most influence in a sub- of any minor factors that depend on them. Thirdly, in
population of volunteers who were new to the job. several large prospective studies, investigators have used
Conclusions. Personal risk factors explained up to 12%
quick but inaccurate “clinical” techniques to measure
of first-time low back pain. [Key words: anthropometry,
back pain, mobility, psychometric questionnaires, risk physical characteristics such as lumbar mobility and
factors] Spine 1999;24:2497–2505 back muscle fatigability7,9,14 or have relied on individual
questions, rather than validated questionnaires, to assess
psychosocial factors. Fourthly, the authors of most pre-
Recent epidemiologic studies have shown that many vious studies were not able to check the reproducibility
cases of disc degeneration and low back pain (LBP) de- of the multivariate statistical model used to identify risk
velop because of patients’ personal characteristics, rather factors for LBP. Unexpected and possibly chance inter-
than patients’ exposure to physical labor.8,9,10,54 Mod- actions between variables may suggest a significant risk
erate physical labor may even strengthen the factor in a given analysis, even though analyses on sub-
spine,1,34,49,61 and only arduous work significantly in- sets of the data or at different time points fail to support
creases the risks of LBP.31,45,56,57,61 the association.
Personal risk factors may be genetic or may be ac- The purpose of the current study was to overcome
quired during the lifetime of the individual. Other factors these problems and to evaluate the true relative impor-
such as lumbar mobility may represent an interaction tance of physical and psychological personal risk factors
between genetics and environment. The genetic influence for LBP. Follow-up rates were kept above 90% by using
could involve metabolism (e.g., an abnormal vitamin D only volunteers who gave the authors permission to trace
them after any change of address during the 3-year fol-
From the *Department of Anatomy, University of Bristol, United King- low-up period. The effects of potential dominant risk
dom, and the †Schulthess Klinik, Zurich, Switzerland. factors were minimized by restricting recruitment to 403
Supported by the Medical Research Council and the Arthritis and
Rheumatism Council (United Kingdom), and by the Schulthess Klinik volunteers of similar age and occupation, who had no
Research Fund (Switzerland). previous history of serious LBP. All physical risk factors

2497
2498 Spine • Volume 24 • Number 23 • 1999

were measured by using precise quantitative techniques,


and five validated questionnaires were used to evaluate
psychological characteristics. The statistical model was
validated at five time points and on various subpopula-
tions. With these checkpoints, the authors hoped to
avoid the difficulties that beset previous investigations.

Methods
Study Design. Healthy volunteers were recruited into a pro-
spective study of risk factors for LBP. On entering the study,
each volunteer underwent a functional assessment that in-
cluded anthropometry and measures of strength, endurance,
mobility, and posture. Five psychometric questionnaires were
completed at the time of entry and at 6, 12, 18, 24, 30, and 36
months thereafter. Accompanying questionnaires asked about
any back-related pain experienced in the previous 6 months.
Various multiple logistic regression models were used to iden-
tify risk factors for future LBP.

Study Population. A total of 403 health care workers volun-


teered for the study. Most were nurses, but physiotherapists,
health care assistants, radiographers, and other therapists were
included. There were 371 women and 32 men. The numbers of
men and women simply reflect the numbers who volun- Figure 1. “Sacral angle” (Ø) was the angle between the vertical
teered—no deliberate selection bias was used. Only those aged and the tangent to the skin surface at S1–S2. Lumbar curvature, or
between 18 and 40 years were admitted (average age on entry, lordosis, was the angle (u) between the tangent to the skin surface
27 years). None had any history of “serious” LBP (defined as at L1 and the tangent at S1–S2.
LBP requiring medical attention or time off work), but when
prompted, 141 volunteers recalled having experienced “nonse- measured maximal strength for up to 20 seconds. Fatiguability
rious” LBP. Throughout the study, “back pain” was taken to was quantified from the endurance time (in the Biering-
include symptoms from the region of the back between L1 and Sorensen test only) and from the rate of change of median
the gluteal folds. Further details of the study population and frequency of the electromyographic signal from the erector spi-
recruitment methods have been reported previously.41 No vol- nae muscles at T10 and L3.23,40,59
unteers were known to be pregnant at the time of the func-
tional assessment. Lumbar Lordosis and the Ranges of Lumbar Flexion, Ex-
tension, and Lateral Bending. These values were measured us-
Functional Assessment. This evaluation lasted approxi- ing the 3Space Isotrak device (Polhemus, VT).19 A source of
mately 2 to 3 hours. To generate interest in the study, the pulsed electromagnetic waves was attached to the skin surface
authors explained each test thoroughly and gave a verbal sum- overlying the sacrum, and a sensor of these waves was attached
mary of each person’s performance in relation to the average. to the back overlying the L1 spinous process (Figure 1). The
Refreshments were offered and expenses reimbursed, but no Isotrak then recorded the angle between source and sensor in
other payments were made. Recruitment and testing of the 403 the sagittal and frontal planes at 28 Hz. This angle is the lum-
volunteers took 30 months. bar lordosis, and the change in this angle between erect stand-
ing and extreme full flexion (measured while standing and
Back Muscle Strength. For this assessment, volunteers were while sitting with the legs out straight to find a true maximum)
asked to adopt a “functional” stooped posture and to pull is the lumbar range of flexion. Ranges of movement in exten-
upward with maximum force on a load cell attached to the sion (standing and lying prone) and side-to-side lateral bending
floor, for a period of 3 seconds.23,40 (standing) were similarly quantified. In all cases, the volunteers
had several attempts to reach an extreme position, and the
Quadriceps Strength. To assess this strength, the volunteer maximum value was noted. The range of lumbar flexion move-
was asked to flex the knee of the dominant leg by 90° while ment also was measured by using the modified Schober (skin-
sitting in a Tornvall chair.39 stretching) test.7
Back Muscle Fatiguability. The Biering-Sorensen test was Sacral Inclination and the Ranges of Hip Flexion and Ex-
used to determine fatiguability. The volunteer lay prone on the tension. The sacral angle is defined in Figure 1. It was measured
edge of a couch, with the torso unsupported, and attempted to by comparing the angle between the Isotrak source, mounted
maintain the trunk in a horizontal position for as long as pos- on the sacrum, and the sensor, which was attached to a vertical
sible. This test is widely used,9,35 but it does not simulate pro- surface. Measurements were repeated with the volunteer in the
longed physical labor, because the back muscles are not loaded standing fully flexed and fully extended postures, to obtain the
severely, and the lumbar lordosis is preserved or exaggerated. range of hip flexion and extension, respectively.19,22
Therefore, an additional fatigue test was devised in which vol-
unteers adopted a standardized stooped position and pulled up Peak Spinal Loading During Standardized Lifts. Peak spinal
on a floor-mounted load cell with 80% of their previously loading was quantified21 to measure each volunteer’s “clumsi-
Personal Risk Factors • Adams et al 2499

ness” during simple manual handling tasks. Details of these Cross-sectional Data
experiments will be reported elsewhere. Details of the physical and psychological measurements
Anthropometry. Measurements were made using standard on the 403 healthy volunteers will be published sepa-
physiologic techniques. They included the length of trunk and rately. A summary can be obtained from the authors.
limbs, body weight and height, body mass index (weight/
height2), and body fat from skinfold measurements.
Univariate Analyses
These revealed several consistent predictors of LBP, in-
Psychological Factors. Five validated psychometric ques- cluding the sum of the scores from the MSPQ and ZUNG
tionnaires were used, as justified previously.41 These were: the questionnaires, which was the best psychometric predic-
modified somatic perception questionnaire (MSPQ18,37), the tor (Table 1). Physical characteristics such as a pro-
Zung depression scale (ZUNG67), and the Health Locus of nounced lordosis, a short back, and a large range of
Control (HLC64) which is divided into three subsections con-
lateral bending reduced the risk of serious LBP, but not
cerned with “internal” factors (IHLC), “powerful others”
(PHLC), and “chance” (CHLC). These questionnaires were
of “any” LBP. Conversely, increasing age and the num-
chosen because they assess personal characteristics such as de- ber of years spent working in health care both tended to
pressive feelings and anxiety, rather than work-related factors, reduce the risk of “any” back pain, but had little effect on
and because they are all suitable for use by nonspecialists. serious LBP. Being a qualified (and therefore experi-
Scores from the MSPQ and the ZUNG questionnaires can be enced) nurse or physiotherapist reduced the risk of seri-
added together to provide a total score (MSPQ 1 ZUNG), ous LBP, but otherwise occupational categories had little
which has a particularly high specificity and sensitivity for pre- effect, possibly because most volunteers were involved in
dicting psychological disturbance in LBP.28,38 The psychomet- similar work. Variables that consistently failed to predict
ric scores used in the final analysis were those obtained at the LBP of any description included: back strength, leg
initial assessment. strength, body weight, arm length, leg length, Schober
The five psychometric questionnaires were sent to each vol-
value, hip mobility, clinical hypermobility, IHLC, and
unteer as soon as they volunteered. Forms were checked for
completeness when they were brought to the functional assess-
PHLC. Lumbar sagittal mobility, body height, smoking
ment. If necessary, advice was given concerning the completion status, and pregnancy during the 3-year follow-up pe-
of future questionnaires. Subsequently, all five psychological riod were of variable significance. Most risk factors be-
questionnaires were sent and returned by post every 6 months. came more significant with increasing follow-up time,
An accompanying questionnaire asked about any back-related presumably because of the greater numbers of volunteers
pain experienced during the previous 6 months, including de- with LBP, but several risk factors, including trunk length
tails of any medical consultations and time taken off work. and lumbar lordosis, were more significant at 12 months.
Returned questionnaires were checked, and scores were en-
tered into a database. If questionnaires were not returned or Correlations Between Variables
were incomplete, then the volunteer was contacted repeatedly Despite the narrow age range of the population, the
on the telephone or by letter until a response was forthcoming. range of lumbar extension (P , 0.001) and that of lateral
Even if a questionnaire was not returned, the volunteer was bending (P , 0.01) decreased with age, whereas back
pursued with equal vigor at the next 6-month interval. muscle strength and endurance capacity increased (P ,
0.05). Some other variables were obviously highly corre-
Statistical Methods. The ability of each measured factor to lated, such as trunk length and body height (P , 0.001),
predict either serious or “any” LBP (i.e., serious or nonserious)
back muscle strength and fatigability (P , 0.05), and
was assessed by univariate linear logistic regression. Univariate
analyses were repeated using data from each of the five fol-
lumbar lordosis and sacral angle when standing (P ,
low-up questionnaires to identify consistent risk factors for 0.001). Schober value (a measure of skin stretching) was
LBP. The independence of variables was assessed using corre- related to the lumbar range of flexion (P , 0.001) but
lation analysis, and the linearity of risk associated with certain accounted for less than 9% of it, perhaps because it was
variables was checked by comparing subsets of their range. influenced by other factors such as trunk length (P ,
Additional univariate analyses were performed on various sub- 0.01). The generally small interactions between physical
sets of the study group. Results from the univariate analyses and psychometric variables have been described previ-
were used to construct multiple regression models for predict- ously.41
ing LBP, using stepwise forward multiple logistic regression.
Results were defined as significant at P , 0.05, P , 0.01, or P Multivariate Models
, 0.001. For each of the five follow-up times, stepwise forward
multiple regression models were constructed from all
Results
those variables that were significant predictors of LBP at
Response Rates and Incidence of Low Back Pain two or more time points and that were not correlated to
The follow-up response fell from 99% at 12 months to each other. Age and gender also were entered. The best
90% at 36 months. The number of volunteers reporting models for all volunteers combined are summarized in
serious LBP increased from 46 to 90 at these times, and Table 2. The best predictors of serious LBP were reduced
those reporting “any” LBP increased from 159 to 266. range of lumbar lateral bending, reduced lumbar lordo-
There were no significant differences between men sis, a long back, increased psychological distress (as in-
and women. dicated by the combined MSPQ 1 ZUNG score), and
2500 Spine • Volume 24 • Number 23 • 1999

Table 1. Results of Univariate Analyses Indicating Risk ceptably linear). For example, reducing lumbar range of
Factors for Low Back Pain (LBP) lateral bending by 2 standard deviations would (for a
student nurse) increase the risk of serious LBP by 550%
12 months 36 months
(odds ratio 5 6.5).
Serious Any Serious Any
Variable LBP (P) LBP (P) LBP (P) LBP (P) Discussion

Age 0.721 (0.042)* 0.283 (0.073) The purpose of the study was to assess the relative im-
Women (vs. men) (0.016)* 0.897 (0.093) 0.993 portance of physical and psychological risk factors for
Previous LBP (0.011)* (,0.001)* (,0.001)* (,0.001)* first-time LBP, while avoiding the methodologic difficul-
Smoking 0.861 0.203 (0.072) (0.117)
Years in occupation 0.245 (0.002)* 0.363 (0.025)* ties that have beset previous investigations. The high sta-
Qualified nurse (0.011)* 0.261 (0.034)* 0.318 tistical significance and robustness of the multivariate
Qualified physiotherapist 0.267 0.522 (0.033)* 0.215 models (Table 2) suggest that this objective has been
CHLC (0.050)* 0.731 (0.061) 0.310
MSPQ 0.788 ,0.001* 0.155 ,0.001* achieved. Evidently, a long, stiff, or flat back appears to
ZUNG 0.082 0.028* 0.010* 0.014* predispose one to LBP, as do certain psychological fac-
MSPQ1ZUNG 0.090 0.001* 0.006* ,0.001* tors and previous LBP. The following sections will con-
Height 0.017* 0.428 0.129 0.874
Trunk length 0.008* 0.492 0.037* 0.633 sider the validity of the techniques used and the range of
Fat-free body mass 0.050* 0.600 0.133 0.761 applicability of the results obtained. Each identified risk
Sacral angle—standing (0.004)* 0.800 (0.019)* 0.157 factor will be discussed in turn, and the final section will
Sacral angle—full flexion 0.146 0.197 (0.029)* 0.237
Lumbar lordosis—standing (0.002)* (0.048)* (0.027)* 0.144 tackle the question: What did cause back pain in most of
Lumbar ROE (0.770) 0.310 0.576 0.971 the volunteers?
Lumbar ROF 0.131 0.462 0.252 0.170
Lumbar ROLF (0.012)* 0.279 (0.007)* 0.290
Modified Schober 0.846 0.946 0.619 0.405
Validity of Techniques
Back muscle endurance time (0.133) 0.486 (0.010)* (0.058) The use of self-reported pain as the outcome measure is
MF-gradient (max) 0.200 0.739 0.113 0.956 justified by a recent study in which 120 of 131 individu-
* Significant. als who reported previous LBP in a postal questionnaire
Serious LBP 5 required medical attention or time off work; Any LBP 5 could gave the same response 12 months later.65 Furthermore,
be serious or not. Previous LBP 5 an episode of nonserious LBP prior to
entering the study; ROE 5 range of sagittal extension; ROF 5 range of sagittal the reliability of self-reported LBP increases when the
flexion; ROLF 5 range of lateral flexion; MF 5 median frequency. pain is recent. The volunteers in the current study re-
Note: Values in parentheses indicate a negative association.
ported pain during the previous 6 months, so the average
elapsed time between pain onset and reporting was only
3 months.
previous nonserious LBP. This last factor dominated the The five psychometric questionnaires have been vali-
prediction of “any” LBP. dated and were devised for use by nonspecialists. Each
Similar results were obtained if women were ana- volunteer completed each questionnaire on seven sepa-
lyzed as a separate group (Table 2). Results among the rate occasions during their 3-year involvement with the
men were qualitatively similar, but few predictors study, and a preliminary assessment showed that re-
reached significance because of the small numbers (not peated scores were reproducible over long periods.41
shown in Table 2). A separate analysis was performed Techniques used to measure the physical risk fac-
on the 130 student nurses in the study, because they tors have been evaluated previously. When mounted
were very homogenous for age, occupation, and job on the back as described above, Isotrak measurements
experience, yet were required to perform the same are reproducible,19 and when used to quantify the
work as the more experienced nurses. Within this range of lumbar flexion movement, they provide mean
group, there was a greatly increased influence (as in- values and standard deviations similar to those ob-
dicated by R2) from the physical factors such as trunk tained using bilateral radiographs.24,47 This suggests
length and lumbar mobility on future serious LBP (Ta- that the Isotrack measurements in the current study do
ble 2). Finally, a separate analysis was performed on not contain larger, systematic errors. Random errors
those 262 volunteers who did not report any previous may arise from small skin movements, but these will
nonserious LBP when they entered the study. Remov- serve only to reduce statistical correlations with future
ing this important risk factor caused age to become LBP. Isotrak measurements of lateral bending have not
significant; increasing age tended to protect this subset been validated against radiographs because of the dif-
of volunteers from LBP. ficulty in obtaining ethical committee approval for ra-
Table 3 summarizes the most important results, which diation exposure. Muscles are strongest when
were evaluated at the central time point (24 months) stretched slightly from their resting position, so back
when response rates were still almost 100%. The odds muscle strength and fatiguability need to be measured
ratios quoted in this table show the increased risk of LBP in a moderately flexed functional posture if they are to
associated with changing the predictor variable from the be applicable to real life. The current authors’ electro-
mean value minus 1 standard deviation to the mean plus myographic techniques for quantifying muscle fatigue
1 standard deviation (within this range, the risk was ac- have been validated extensively, 23,40 and a recent
Personal Risk Factors • Adams et al 2501

Table 2. Results of Multivariate Analyses Indicating Risk Factors for Low Back Pain (LBP) in Different Groups
of Subjects
12 months 18 months 24 months 30 months 36 months
(R2 %) (R2 %) (R2 %) (R2 %) (R2 %)

Serious LBP (all subjects) n 5 386


Previous LBP 1.2* 1.6† 1.8† 2.6‡ 2.7‡
MSPQ1ZUNG 1.3* 0.8* 1.5† 0.9*
Trunk length 0.7* 0.5* 0.7*
Lumbar lordosis (2.0†) (0.7*) (0.7*) (1.2*) (0.9*)
Lumbar ROLF (1.4*) (1.8†) (2.1†) (1.5†) (1.4†)
Any LBP (all subjects) n 5 386
Age (0.9†) (0.8*) (0.6*) (0.9*) (0.8*)
Previous LBP 3.5‡ 4.3‡ 6.0‡ 4.9‡ 4.0‡
MSPQ1ZUNG 0.6* 0.8* 0.6* 1.3† 1.4†
Lumbar lordosis (0.6*) (2.0‡) (1.0†) (0.6*)
Serious LBP (women) n 5 355
Previous LBP 1.5* 1.3* 3.0‡ 2.8‡ 3.1‡
MSPQ1ZUNG 1.0* 1.0* 0.5*
Trunk length 0.9*
Lumbar lordosis (1.7*) (0.6*) (0.7*) (0.5*)
Lumbar ROLF (2.0†) (2.3†) (2.3†) (1.5†) (1.4†)
Serious LBP (no previous LBP) n 5 248
MSPQ1ZUNG 2.3†
Lumbar lordosis (1.7*) (1.4*) (2.1*) (1.8*)
Lumbar ROLF (2.8*) (3.5†) (2.0*) (1.5*) (2.1†)
Age (1.7*) (2.0*) (1.3*)
Serious LBP (students) n 5 125
Trunk length 6.4† 4.2† 6.3† 2.2*
Lumbar lordosis (2.4*)
Lumbar ROLF (2.8*) (5.3†) (3.6*) (2.3*)
Lumbar ROF (2.1*)
* P , 0.05.
† P , 0.01.
‡ P , 0.001.
R2 5 the % of reports of LBP that are attributable to each risk factor; Serious LBP 5 required medical attention or time off work; Any LBP 5 could be serious
or not; Previous LBP 5 an episode of nonserious LBP prior to entering the study; No previous LBP 5 those subjects who, on entry into the study, had no
recollection of having suffered from any LBP in the past; Lumbar lordosis 5 lordosis in standing; ROLF 5 range of lateral flexion; ROF 5 range of sagittal flexion;
n 5 total number of subjects included in model at 12 months follow up.
Note: Values in parentheses indicate a negative association.

study has confirmed that a decline in median frequency Repeated contacts with the volunteers and the very
does reflect a real loss of maximal force-generating fact that they were volunteers could have resulted in a
capacity, or “fatigue.”39 well-motivated and self-selected group. However, the
fact that the annual incidence of serious LBP in the first
Generalizability of Results
12 months of the current study (13%) was similar to the
This study focused on healthy volunteers of similar age
16% reported previously in a large prospective study on
and occupation to prevent the regression model from
a general population9 suggests that this was not the case.
being dominated by previous LBP, age, and occupation
It may not be valid to compare incidence rates with those
loading. However, this makes it more difficult to extrap-
from previous studies on nurses, because the volunteers
olate to other ages and occupations. Each of the risk
in the current study benefited from the recent introduc-
factors considered in Tables 1 and 2 may be more or less
tion of manual handling aids to counter the previously
important at predicting LBP in heavy laborers, office
reported high incidence of LBP in the profes-
workers, or retired people. The influence of physical risk
sion.32,48,62,66
factors such as trunk length and ranges of lumbar move-
Male and female data were pooled because gender
ment probably can be explained in terms of increased
had only a marginal influence on future LBP (Table 2)
mechanical loading on certain spinal tissues (see below),
and because pooling the results increased the range of
so these factors might be expected to have more influence
several predictor variables such as body weight and
in occupations that are physically demanding. Similarly,
height. Results were very similar among women, which
psychological factors may predict more LBP in mentally
suggests that the inclusion of men could not have dis-
stressful occupations.
torted the model. Although the data on men showed a
A closely-related problem is that the homogeneity of
few similar trends, there were too few men to identify
the current study population may have restricted the
any real gender differences in these risk factors.
range of certain predictor variables, such as strength and
lumbar mobility, and may have made it more difficult for Predictors of First-Time Low Back Pain
them to influence future LBP. R2 values for these factors Several factors unexpectedly failed to reach significance.
may be higher in the general population. The effect of smoking is dose-related,17 and the volun-
2502 Spine • Volume 24 • Number 23 • 1999

Table 3. Relative Risks of Serious Low Back Pain (LBP) for the 24-months Follow-up Data
Serious LBP (all subjects) 24 months (n 5 382) Serious LBP (students) 24 months (n 5 125)
2 2
R % OR 95% CI R % OR 95% CI

Previous LBP 1.8* 2.3 1.3–4.0


MSPQ1ZUNG 0.8† 1.3 1.0–1.5
Trunk length 0.7† 1.9 1.1–3.3 6.3* 4.7 1.7–13.1
Lumbar lordosis 0.7† (1.8) 1.0–3.2
Lumbar ROLF 2.1* (2.5) 1.4–4.5 5.3* (6.5) 2.0–20.8

Total 6.1% 11.6%


* P , 0.01.
† P , 0.05.
CI 5 confidence interval; Lumbar lordosis 5 lordosis in standing; ROLF 5 range of lateral flexion; n 5 total number of subjects included in model.
Note: For continuous variables, the odds ratio (OR) represents the increased risk associated with increasing the variable from the mean less one SD to the mean
plus one SD. Where the association is negative (OR in parentheses), the risk is for reducing the variable by this amount. For example, for student nurses, increasing
trunk length by two SD increases the risk of LBP by a factor of 4.7 (i.e., by 370%).

teers in the current study may not have smoked many Although psychological factors are not very impor-
cigarettes per day. Age and occupational category may tant predictors of first-time LBP, they do explain how
have had little influence, because all volunteers were people respond to their pain, e.g., by seeking treatment,
aged 18 – 40 years and worked in similar occupations. and how they improve. Questionnaires concerning fear-
Previous Low Back Pain. This has been a dominant risk
avoidance beliefs63 are particularly good at identifying
factor in several previous investigations,7,9,10 and it pre- patients whose LBP is likely to become chronic,12,62 and
vailed again in the current study, even though the previ- it remains to be seen whether these or other question-
ous pain recalled by volunteers in this study was too naires can explain more first-time LBP than the question-
trivial to require medical attention or even time off work. naires used in the current study.
This association may be attributable to the cyclic nature Lumbar Mobility. Intuitively, one would suspect that a
of LBP. Also, it is possible that individuals who were supple back should be more difficult to injure in bending,
prepared to report such trivial LBP on entry to the study and a study on healthy volunteers performing simple
were also more prepared to report pain that occurred bending tasks in the laboratory has confirmed that sup-
during the 3-year follow-up period. When those report- ple individuals subject their lumbar spine to bending mo-
ing previous LBP were excluded from the analysis (Table ments that are lower than those of stiff individuals.19
2), the risk of LBP declined with increasing age. These previous results referred to sagittal-plane flexion,
Psychological Factors. The MSPQ 1 ZUNG was a major whereas in the current study the greatest risk for LBP was
predictor of serious back pain, and several psychometric associated with poor lumbar mobility in the frontal
scores predicted “any” back pain in the univariate anal- plane (i.e., a reduced range of lateral bending). This may
yses (Table 1). The generally higher significance levels for be attributable to the increased stiffness of the lumbar
“any” LBP may be attributable to the higher number of spine in the frontal plane compared with that in the sag-
people who reported it, or it may reflect a greater psy- ittal plane.52 Many bending movements incorporate a
chological involvement in the reporting of minor ail- mixture of forward and lateral flexion,25 and the in-
ments.10 The current authors previously have suggested creased risk of disc injury under these circumstances has
that future first-time back pain can best be predicted by been demonstrated by cadaveric experiments,3,4 mathe-
means of two questionnaires, the ZUNG and the matical models,53 and epidemiologic studies.25,42 In the
MSPQ.41 The latter is better for predicting relatively current study, the ranges of lumbar motion in the two
nonserious back pain whereas the combined score is best planes were so highly correlated (P , 0.001) that only
for predicting more serious back pain. The data for seri- the range of lateral bending was included in most multi-
ous back pain in Table 2 are comparable with those of variate models.
the Boeing study,10 in which the authors found that med- Previous retrospective surveys have linked stiff backs
ical history (including previous back pain) predicted with LBP,13 whereas prospective studies have yielded in-
3.3% of future back injury reports, and that psycholog- conclusive or contradictory results,7,9 which may well be
ical factors explained 1.9% of them. Of the psychomet- attributable to the use of the Schober (skin stretching)
ric scores, only the MSPQ was changed (slightly) by a test to measure mobility. A study claiming a high corre-
first attack of serious back pain, and none was changed lation between Schober measurements and the angular
by a first attack of “any” back pain,41 so it seems likely rotation of vertebrae is seriously flawed because it in-
that a high MSPQ 1 ZUNG score is a true predictor of cludes healthy volunteers and patients with ankylosing
future LBP, rather than a consequence of previous non- spondylitis in a single population for the purposes of
serious LBP. statistical correlation.36 In fact, the Schober test never
Personal Risk Factors • Adams et al 2503

has been shown to be a good measure of the range of quency.23,40 In future work, the current authors will an-
angular movement of lumbar vertebrae during flexion. alyze electromyographic data in more detail to determine
which aspects of the fatigue tests best predicts LBP.
Lumbar Lordosis. A lordotic lumbar spine acts as a
shock-absorber during locomotion, in a manner similar
What Caused Most of the Back Pain?
to that of a thin metal rod that is bent to form a helical Although several personal characteristics were highly
spring. In both cases, the curved shape allows the loaded significant predictors of future back pain, the R2 values
structure to bend, and this in turn allows it to absorb (Table 3) indicate that they explained less than 12% of it.
more strain energy than if it were simply compressed. In Either these factors play only a minor role in helping to
the case of the lumbar spine, even more energy may be determine when most young people first report back
absorbed by muscles and their tendons as they attempt to pain, or else they play a decisive role, but only in a small
resist changes in lordosis. This shock-absorbing function minority of people. A risk factor can be highly significant
of a lordosis may explain why it appears to protect (small P value) but unimportant (low R2 value) if it is
against future LBP. Other mechanisms have been pro- rare in the population studied, but has a decisive effect
posed to explain how a lordosis might reduce spinal when it does occur. As discussed above, the personal
loading during manual handling,51,58 but they are of factors included in the current study might have pre-
doubtful relevance because it is impossible to lift an ob- dicted more LBP in physically or mentally stressful occu-
ject from the ground without flattening the lower back pations, especially if a more variable group of individuals
and stretching the lumbodorsal fascia.21,22 In the current had been considered.
study, lumbar lordosis was inversely correlated to range Nevertheless, less than 12% of the volunteers studied
of extension (P , 0.05) and positively correlated to could attribute their back pain to any of the psycholog-
range of flexion (P , 0.001), suggesting that a particu- ical or physical risk factors considered. What, then,
larly lordotic posture is equivalent to standing in a posi- caused LBP in the remainder? Some might have been
tion closer to full extension than is normal. This may attributable to personal risk factors not considered in
explain in part the protective effect of a pronounced lor- this study, such as a deleterious gene for vitamin D60 or
dosis, because increases in lordosis are associated with collagen, short internal lever-arms for the back muscles,
reduced intradiscal pressures6 and with a transfer of or specific work-related factors such as job satisfaction,
loading from the nucleus to the posterior anulus and which were not assessed in this study’s question-
apophyseal joints.5 naires.10,46 Also, purely random accidents and emergen-
cies may have played a part, especially among nurses
Trunk Length. When objects are lifted from the floor, a
who deal with unpredictable loads (patients), or who
long back would tend to keep both object and upper
have poor handling skills.62 Chance quarrels with col-
body weight further from the center of rotation in the
leagues may similarly precipitate pain with a high psy-
intervertebral discs. The back muscles would have to
chosomatic component.
work harder, and the discs would be compressed more
Despite these possibilities, it is becoming apparent
severely. This may explain why individuals with a long
that there are no simple explanations for LBP, only a
trunk (or body—the two are closely related) were predis-
large number of interacting causes that must be identified
posed to low back problems in the current and previous
and pieced together like a jig-saw puzzle. As far as rela-
studies.11,29,30,33 The influence of external levers, such as
tively trivial LBP is concerned, the effort to pursue these
trunk length, may be mirrored by a similar influence
causes may simply not be worthwhile, because the failure
from the length of internal levers, such as the spinous
of all the physical factors to predict “any” LBP in the
processes. The inherited length of internal and external
current study (Table 2) surely indicates that mechanical
levers may explain in part the strong influence of genetics
influences in the reporting of such pain are negligible. In
on back pain50 and disc degeneration.8
this sense, mechanical influences would include any met-
Back Muscle Strength and Fatigability. The results of this abolic or biochemical factors that might weaken tissues
study confirm those in previous reports that back muscle and leave them more vulnerable to injury. Perhaps the
strength has little influence on LBP,9,33,44 whereas the problem of nonserious LBP should be left to psycholo-
rate at which the muscles fatigue does have some ef- gists and work managers.
fect.9,35 This influence may be explained in terms of the As far as serious LBP is concerned, the results of the
protection offered the spine by the back muscles during current study offer some encouragement to those who
repetitive or sustained exertions. Fatigable muscles soon seek to understand it in mechanistic terms. The posses-
lose their ability to generate maximal force, and this may sion of a long, flat, or stiff back substantially increases
leave the underlying spine more vulnerable to bending the risk of serious LBP, and it is difficult to explain this in
injury20 or to excessive “creep” in sustained postures.43 other than mechanical terms. Furthermore, the finding
Typical fatigue tests involve a combination of physical, that these factors have a much greater influence in stu-
metabolic, and psychological factors that can be difficult dent nurses who are new to the job (Table 3) provides
to untangle, and several variable factors can influence the additional insight into the role of mechanical loading in
rate of decrease in the electromyographic median fre- LBP. It suggests that recent increases in physical activity
2504 Spine • Volume 24 • Number 23 • 1999

are more likely to lead to LBP than a high but constant 12. Burton AK, Tillotson KM, Main CJ, Hollis S. Psychosocial predictors of
outcome in acute and subchronic low back trouble. Spine 1995;20:722– 8.
level of activity. Previously, the current authors have ar- 13. Burton AK, Tillotson KM, Troup JDG. Variation in lumbar sagittal mobility
gued that sudden increases in physical activity can make with low back trouble. Spine 1989;14:584 –90.
the large avascular discs the “weak link” in a rapidly 14. Cady LD, Bischoff DP, O’Connell ER, Thomas PC, Allan JH. Strength and
fitness, and subsequent back injuries in firefighters. J Occup Med 1979:269 –72.
strengthening spine, which is loaded with increasing se-
15. Coste J, Delecoeuillerie G, Cohen de Lara A, Le Parc JM, Paolaggi JB.
verity by its own strengthening muscles.1 This hypothesis Clinical course and prognostic factors in acute low back pain: An inception
has received some support from the recent evidence that cohort study in primary care practice. Br Med J 1994;308:577– 80.
student nurses experience more back pain during their 16. Croft PR, Papageorgiou AC, Ferry S, Thomas E, Jayson MIV, Silman
AJ. Psychologic distress and low back pain: Evidence from a prospective study in
first year working on the wards.32 The results of the the general population. Spine 1995;20:2731–7.
current study lend further support to the hypothesis, be- 17. Deyo RA, Bass JE. Lifestyle and low back pain: The influence of smoking and
cause they suggest that the increased LBP in young nurses obesity. Spine 1989;14:501– 6.
18. Deyo RA, Walsh NE, Schoenfeld LS, Ramamworthy S. Studies of the mod-
is associated with mechanical factors. ified somatic perceptions questionnaire (MSPQ) in patients with back pain.
Spine 1989;14:507–10.
Conclusions 19. Dolan P, Adams MA. Influence of lumbar and hip mobility on the bending
stresses acting on the lumbar spine. Clin Biomech 1993;8:185–92.
20. Dolan P, Adams MA. Repetitive lifting tasks fatigue the back muscles and
1. In multivariate analyses, none of the physical risk
increase the bending moment acting on the lumbar spine. J Biomech 1998;31:
factors was an important predictor of “any” LBP. It 713–21.
may not be possible to explain trivial LBP in mechan- 21. Dolan P, Earley M, Adams MA. Bending and compressive stresses acting on
the lumbar spine during lifting activities. J Biomech 1994;27:1237– 48.
ical terms, at least in young health care workers.
22. Dolan P, Mannion AF, Adams MA. Passive tissues help the back muscles to
2. The following were consistent predictors of serious generate extensor moments during lifting. J Biomech 1994;27:1077– 85.
first-time LBP: reduced lumbar mobility, reduced 23. Dolan P, Mannion AF, Adams MA. Fatigue of the erector spinae muscles: A
quantitative assessment using “frequency banding” of the surface EMG signal.
lumbar lordosis, a long back, increased psychological
Spine 1995;20:149 –59.
distress, and previous nonserious LBP. 24. Dolan P, Mannion AF, Adams MA. Schober test measurements do not cor-
3. Up to 12% of serious LBP could be explained by relate well with angular movements of the lumbar spine. Presented to the ISSLS,
Helsinki, Finland, June 1995.
various combinations of the physical and psycholog-
25. Fathallah FA, Marras WS, Parnianpour M. The role of complex simulta-
ical factors considered in the current study. Other fac- neous trunk motions in the risk of occupation-related low back disoeders.
tors may be more important, and the factors consid- Spine 1998;23:1035– 42.
26. Frymoyer JW, Pope MH, Clements JH, Wilder DG, MacPhearson B, Ash-
ered in this study may have more influence in other
ikaga T. Risk factors in low back pain. J Bone Joint Surg [Am] 1983;65:213– 8.
populations and work environments. 27. Greenough CG, Fraser RD. The effects of compensation on recovery from
4. Serious LBP was more common among those who low-back injury. Spine 1989;14:947–55.
28. Greenough CG, Fraser RD. Comparison of eight psychometric instruments
were new to the job, and the influence of physical risk
in unselected patients with back pain. Spine 1991;16:1068 –74.
factors was much greater in this group. 29. Heliovaara M. Body height, obesity and risk of herniated lumbar interver-
tebral disc. Spine 1987;12:469 –72.
Acknowledgments 30. Heliovaara M. Risk factors for low back pain and sciatica. Ann Med 1989;
21:257– 64.
The authors thank Christine Standell for helping with all 31. Kelsey JL, Githens PB, White AA, et al. An epidemiologic study of lifting and
aspects of this work, and Tony Hughes for statisti- twisting on the job and risk for acute prolapsed lumbar intervertebral disc. J Or-
thop Res 1984;2:61– 6.
cal advice. 32. Klaber Moffett JA, Hughes GI, Griffiths P. A longitudinal study of low back
pain in student nurses. Int J Nursing Studies 1993;30:197–212.
References 33. Kujala UM, Taimela S, Viljanen T, et al. Physical loading and performance as
predictors of back pain in healthy adults: A 5-year prospective study. Eur J Appl
1. Adams MA, Dolan P. Could sudden increases in physical activity cause Physiol 1996;73:452– 8.
intervertebral disc degeneration? Lancet 1997;350:734 –5. 34. Leino PI. Does leisure time physical activity prevent low back disorders?
2. Adams MA, Dolan P, Marx C, Hutton WC. An electronic inclinometer Spine 1993;18:863–71.
technique for measuring lumbar curvature. Clin Biomech 1986;1:130 – 4. 35. Luoto S, Heliovaara M, Hurri H, Alaranta H. Static back endurance and the
3. Adams MA, Hutton WC. Prolapsed intervertebral disc: A hyperflexion in- risk of low back pain. Clin Biomech 1995;10:323– 4.
jury. Spine 1982;7:184 –91. 36. Macrae IF, Wright V. Measurement of back movement. Ann Rheum
4. Adams MA, Hutton WC. Gradual disc prolapse. Spine 1985;10:524 –31. Dis 1969;28:584 –9.
5. Adams MA, McNally DS, Chinn H, Dolan P. Posture and the compressive 37. Main CJ. The modified somatic perception questionnaire (MSPQ). J Psycho-
strength of the lumbar spine: International Society of Biomechanics Award Paper. somatic Research 1983;27:503–14.
Clin Biomech 1994;9:5–14. 38. Main CJ, Wood PLR, Hollis S, Spanswick CC, Waddell G. The distress and
6. Andersson GBJ, Ortengren R, Nachemson A. Intradiscal pressure, intra- risk assessment method: A simple patient classification to identify distress and
abdominal pressure and myoelectric back muscle activity related to posture and evaluate the risk of poor outcome. Spine 1992;17:42–52.
loading. Clin Orthop 1977;129:156 – 64. 39. Mannion AF, Dolan P. Relationship between myoelectric and mechanical
7. Battie MC, Bigos SJ, Fisher LD, et al. The role of spinal flexibility in back manifestations of fatigue in the quadriceps femoris muscle group. Eur J Appl
pain complaints within industry: A prospective study. Spine 1990;15:768 –73. Physiol 1996;74:411–19.
8. Battie MC, Videman T, Gibbons LE, et al. Determinants of lumbar disc 40. Mannion AF, Dolan P. Electromyographic median frequency changes during
degeneration: A study relating lifetime exposures and MRI findings in identical isometric contraction of the back extensors to fatigue. Spine 1994;19:1223–9.
twins. Spine 1995;20:2601–12. 41. Mannion AF, Dolan P, Adams MA. Psychological questionnaires: Do “ab-
9. Biering-Sorensen F. Physical measurements as risk indicators for low back normal” scores precede or follow first-time low back pain? Spine 1996;21:2603–
trouble over a one-year period. Spine 1984;9:106 –18. 11.
10. Bigos SJ, Battie MC, Spengler DM, Fisher LD, et al. A prospective study of 42. Marras WS, Lavender SA, Leurgans SE, et al. The role of dynamic three-
work perceptions and psychosocial factors affecting the report of back injury. dimensional trunk motion in occupationally-related low back disorders.
Spine 1991;16:1– 6. Spine 1993;18:617–28.
11. Bostman OM. Body mass index and height in patients requiring surgery for 43. McGill SM, Brown S. Creep response of the lumbar spine to prolonged full
lumbar intervertebral disc herniation. Spine 1993;18:851– 4. flexion. Clin Biomech 1992;7:43– 6.
Personal Risk Factors • Adams et al 2505

44. Mostardi RA, Noe DA, Kovacik MW, Porterfield JA. Isokinetic lifting HM. Trunk extensor endurance and its relationship to electromyogram param-
strength and occupational injury: A prospective study. Spine 1992;17:189 –93. eters. Euro J Appl Physiol 1993;66:388 –96.
45. Mundt DJ, Kelsey JL, Golden AL, et al. An epidemiologic study of non- 60. Videman T, Leppavuori J, Kaprio J, et al. Intragenic polymorphisms of the
occupational lifting as a risk factor for herniated lumbar intervertebral disc. Vitamin D receptor gene associated with intervertebral disc degeneration.
Spine 1993;18:595– 602. Spine 1998;23:2477– 85.
46. Papageorgiou AC, MacFarlane GJ, Thomas E, et al. Psychosocial factors in 61. Videman T, Nurminen M, Troup JDG. Lumbar spinal pathology in cadav-
the workplace— do they predict new episodes of low back pain? Spine 1997;22: eric material in relation to history of back pain, occupation and physical loading.
1137– 42. Spine 1990;15:728 – 40.
47. Pearcy MJ, Tibrewal SB. Axial rotation and lateral bending in the normal 62. Videman T, Rauhala H, Asp S, et al. Patient-handling skill, back injuries, and
lumbar spine measured by three-dimensional radiography. Spine 1984;9:582–7. back pain: An intervention study in nursing. Spine 1989;14:148 –56.
48. Pheasant S, Stubbs D. Back pain in nurses: Epidemiology and risk assess- 63. Waddell G, Newton M, Henderson I, Somerville D, Main CJ. A Fear-
ment. Appl Ergonom 1992;23:226 –32. Avoidance Beliefs Questionnaire (FABQ) and the role of fear-avoidance beliefs in
49. Porter RW, Adams MA, Hutton WC. Physical activity and the strength of the chronic low back pain and disability. Pain 1993;5:157– 68.
lumbar spine. Spine 1989;14:201–3. 64. Wallston KA, Wallston BS, DeVellis R. Development of the Multidimen-
50. Postacchini F, Lami R, Pugliese O. Familial predisposition to discogenic sional Health Locus of Control (MHLC) scales. Health Education Mono-
low-back pain: An epidemiologic and immunogenetic study. Spine 1991;13: graph 1978;6:160 –70.
1403– 6. 65. Walsh K, Coggon D. Reproducibility of histories of low-back pain obtained
51. Potvin JR, McGill SM, Norman RW. Trunk muscle and lumbar ligament by self-administered questionnaire. Spine 1991;16:1075–77.
contributions to dynamic lifts with varying degrees of trunk flexion. Spine 1991; 66. Yassi A, Khokhar JB, Tate R, Cooper JE, Snow C, Vallentyne S. The epide-
16:1099 –108. miology of back injuries in nurses at a large Canadian tertiary care hospital:
52. Schultz AB, Warwick DN, Berkson MH, Nachemson AL. Mechanical prop- Implications for prevention. Occup Med 1995;45:215–20.
erties of human lumbar spine segments: Part 1. Response in flexion, extension, 67. Zung WWK. A self-rating depression scale. Arch Gen Psychiatry 1965;12:
lateral bending and torsion. J Biomech Eng 1979;101:46 –52. 63–70.
53. Shirazi-Adl A. Biomechanics of the lumbar spine in sagittal/lateral moments.
Spine 1994;19:2407–14.
54. Simmons ED, Guntupalli M, Kowalski JM, Seidel T. Familial predisposition
for degenerative disc disease: A case-control study. Spine 1996;21:1527–9.
Address reprint requests to
55. Svensson H-O, Andersson GBJ. Low back pain in 40- to 47-year-old men:
Work history and work environment factors. Spine 1983;8:272– 6. Michael A. Adams, PhD
56. Svensson H-O, Andersson GBJ. The relationship of low-back pain, work
history, work environment and stress. Spine 1989;14:517–22.
University of Bristol, Department of Anatomy
57. Troup JDG, Martin JW, Lloyd DCEF. Back pain in industry. Spine 1981;6: Southwell Street, Bristol BS2 8EJ
61–9. United Kingdom
58. Tveit P, Daggfeldt K, Hetland S, Thorstensson A. Erector spinae lever arm E-mail: M.A.Adams@bris.ac.uk.
length variations with changes in spinal curvature. Spine 1994;19:199 –204.
59. van Dieen JH, Oude Vrielink HHE, Housheer AF, Lotters FBJ, Toussaint

You might also like