Journal of Occupational Rehabilitation, Vol. 5, No. 2, 1995
Isokinetic Performance in Low Back Pain Patients: The Predictive Power of the Self-Efficacy Scale K. K. Kaivanto, 1,4 A-M. Estlander, 2 G. B. Moneta, 1 and H. Vanharanta 3
The Self-Efficacy Scale (SES) has been found to predict isokinetic performance better than anthropometric variables. This study tests the predictive power of SES further against other measures of efficacy expectancies as well as measures of depression and perceived disability. A group of 105 chronic back pain patients was administered Beck's Depression Inventory (BDI), SES, the Pain Self-Efficacy Questionnaire (PSEQ), and the Oswestry low back pain disability questionnaire (OSWESTRY). Total isokinetic work done was measured at slow, medium and high speeds, for which multiple regression models were fitted controlling for sex, ag~ weight and height. The results confirmed SES to be the best overall predictor of isokinetic performance. BDI was not significant as a predictor of isokinetic performance. The models also revealed that SES predicts less well with increases in the test speed, particularly in extension. These results provide further evidence of the diagnostic value of SES relative to OSWESTRY and PSEQ. KEY WORDS: isokinetic performance; low back pain; self-efficacy expectancies; disability; prediction.
INTRODUCTION In the ever greater number of studies being performed on Low Back Pain (LBP) the role attributed to nonsomatic factors is becoming increasingly pronounced (1-3). As with LBP in general, evidence is mounting that behavioral and psychological factors also affect the results of dynamic strength iso-measurement (4). It has been suggested that the significant motor impairment of chronic LBP patients can be at least partially accounted for by avoidance learning, in which the negative reinforcement of experienced pain causes patients to avoid particular types of motor activity. Expectations come to be formed with respect to each tInstitute of Occupational Health, Helsinki, Finland. 2Finnish Back Institute, Helsinki, Finland. 3Oulu University Central Hospital, Oulth Finland. 4Correspondence should be directed to Kim Kaivanto, Institute of Occupational Health, H-department, Topeliuksenkatu 41 a A, FIN-00250 Helsinki, Finland. 87 1053-0487/95/0600-0087507.50/001995PlenumPublishingCorporation
88
Kaivanto, Estlander, Moneta, and Vanharanta
kind of activity, guiding the choice of what will be attempted and what will be avoided. But in addition to being differentiated by activity, expectancies of differing definitional qualities have also been distinguished. Yet while diverse measurement instruments have been implemented, self-efficacy has been the principal expectancy construct addressed by many theoretical discussions and psychometric instruments. Self-efficacy was originally conceived as the central concept within a theoretical framework for analyzing changes in avoidant and fearful behavior. As proposed by Bandura (5), this theoretical framework submits that behavioral changes are mediated by cognitive processes irrespective of the particular methods used to bring them about, and that cognitive events are primarily modulated by experiences of mastery emanating from effective performance. Within this context Bandura found it necessary to distinguish sharply between outcome expectancies and efficacy expectancies. Definitionally, an outcome expectancy is "a person's estimate that a given behavior will lead to certain outcomes." In contrast, an efficacy expectancy is "the conviction that one can successfully execute the behavior required to produce the outcomes" (5). There is arguably a degree of overlap between these two forms of expectancies, in that a subjective "estimate" is in some ways as much a belief as a "conviction" is. Subsequent work on expectancies has spawned an array of variations in the definition of expectancies, some of them referring to comparable constructs and others to differing constructs (6-9). Within the present paper, this "nebulous and elusive" (9) conceptual field will be avoided by adhering to Bandura's original formulation. A number of new instruments for the measurement of dynamic muscle strength were developed and commercialized in the 1980s. These isomachines were thought to provide relevant, accurate, and reliable information for diagnostic, screening, and rehabilitation follow-up purposes. Current evidence, however, points in quite the opposite direction (4). If iso-machines are to have any practical clinical use, a more profound understanding of the factors which affect performance as measured by them is necessary. In current practice, it is common to adjust isokinetic performance for the anthropometric variables of sex, body weight and age, including perhaps height at most; remaining performance variations are ascribed to supposed somatic factors. A recent study performed by Estlander et al. indicates, however, that anthropometric variables are not in fact very good predictors of isokinetic performance, and that self-efficacy beliefs as measured by the Self-Efficacy Scale (SES) are a much more powerful predictor (10). The purpose of this study is to test the predictive power of SES further against other competing alternatives suggested by the literature, including measures of depression, efficacy expectancies and subjective disability. Depression is nominated by the epidemiologically detected association between psychological or emotional problems and diagnosed back disorders (11). Efficacy expectancies are nominated by the documented predictive power of SES (10). Finally, subjective disability is included because iso-measures have been claimed to provide information relevant to the diagnosis and identification of back conditions. The present study focuses solely on the predictive power of the selected attribute measures. Questions concerning the connection between the attributes and isoki-
lsokinetic Performance in LBP Patients: The Predictive Power of SES
89
netic performance will be taken up elsewhere. The operational task undertaken here is an assessment of the predictive power of Beck's Depression Inventory (BDI), the Mood Disorders scale, the Pain Self-Efficacy Questionnaire (PSEQ), the SelfEfficacy Scale (SES), the Oswestry low back pain disability questionnaire (OSWESTRY), and a number of single-question items.
METHODS Subjects
Between November 1990 and September 1991, 105 patients from the Oulu region were referred to the Oulu University Central Hospital by their doctors for undergoing further diagnostic and treatment procedures. As evaluated by their doctors, these patients had evident degenerative disc disease and recurrent or persistent low back pain with or without root impairment. All 105 patients were submitted to a thorough medical evaluation, a part of which consisted of the administration of an array of question-series prior to isokinetic testing. None of the patients failed to meet the range of movement inclusion criteria of 90 degrees of flexion and 15 degrees of extension as measured in the iso-machine's fixed position. On average low back pain had been troubling the patients for 4.7 years, ranging from 0.5 to 30 years; thus there were no acute injury patients with less than 6 weeks since the onset of pain. Fifty-six percent were male. Forty-nine percent were working, and 48% were sick listed or enjoying retirement benefits. There were no statistically significant differences detected between these two groups on any of the six measures of isokinetic performance (10). Measurements
After general questions inquiring demographic, background, and anthropometric information, the questionnaire included the following series' Finnish-language renditions: BDI, the Symptom Questionnaire, PSEQ, SES, and OSWESTRY (10, 12-16). Both BDI and the Oswestry series were in their standard and widely utilized forms. Individual questions not part of any particular established series were also posed, bringing the total number of nonsomatic variables to ten. Finally, subjects' isokinetic performance was tested. Depression
BDI was in its widely known and utilized form. The Symptom Questionnaire on the other hand was originally constructed with the Finnish population in mind. The 31 items of the Symptom Questionnaire yield several subscales upon coding, from among which the MOOD DISORDERS (14) scale was chosen for the purposes of this study. This particular scale is composed of six items which assess the
90
Kaivanto, Estlander, Moneta, and Vanlmranta
frequency of mood changes, depressive episodes, heightened irritability, restlessness, fear, and apathy.
Efficacy Expectancies Experience in the use of PSEQ is not diminutive (15, 17). The PSEQ consists of ten items in which respondents are asked to gauge how confident they are that they can do each of the ten activities or functions despite their pain, where rating takes place on 7-point scales ranging from "not at all confident" at 0 to "completely confident" at 7. Items refer to activities and functions which chronic pain patients generally find problematic. It includes statements such as: "I can still ac-
W e w o u l d l i k e to k n o w w h a t y o u t h i n k y o u r p e r f o r m a n c e c a p a b i l i t y is a__~t this v e r y m o m e n t . F o r e a c h q u e s t i o n p l e a s e c i r c l e the a n s w e r w h i c h describes your situation.
H o w long a r e y o u a b l e to walk?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
How long are y o u a b l e to run?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
How long are under 2 2-5 y o u a b l e to c a r r y minutes shopping bags (2 x 4-5 k i l o s ) ?
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long a r e under 2 2-5 y o u a b l e to s t a n d minutes eg. in q u e u e ?
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long a r e y o u a b l e to r i d e a bicycle?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long are y o u a b l e to sit in an a r m c h a i r ?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long a r e y o u a b l e to sit at a desk?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
How long are under 2 2-5 y o u a b l e to w o r k minutes bending over (eg. w a s h i n g dishes, vacuum-cleaning, fixing a car)?
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
Fig. 1. The Serf-Efficacy Scale (SES) questionnaire.
Isoldnetic Perfomnnee in LBP Patients: The Predictive Power of SES
91
complish most of my goals in life, despite the pain," and "I can still do many of the things I enjoy doing, such as hobbies or leisure activities, despite the pain" (15). The SES is in turn composed of eight items--one for each of the activities of walking, running, carrying shopping bags, standing, bicycling, sitting in an armchair, sitting at a desk, and working in a forward leaning position--in which the respondent is asked to identify the duration of time he/she thinks he/she is able to endure each activity on an 8 point scale (less than 2 minutes, 2-5, 5-10, 10-15, 15-25, 25-35, 35-45, more than 45 minutes). SES is computed as the sum of these eight items (range 8-64). Much of the contemporary work in designing and modifying self-efficacy measures has concerned the choice of activities to be assessed and the specificity or generality thus obtained. While the SES refers to common, everyday activities, it gains its distinctiveness from framing its items in terms of endurance-time (see Fig. 1) (10). In addition, three efficacy expectancy proxy items were posed to the subjects just before isokinetic testing. The variable CERTA/N consists of responses to the question "How certain are you that you can complete the lifting test as instructed?", where responses range from 1 (absolutely certain) to 10 (I don't believe I can). The variable E N D U R E CONCERN consists of responses to the question "How concerned are you that your back won't be able to endure this test?", where responses range from 1 (not concerned at all) to 10 (extremely concerned). The variable PAIN CONCERN consists of responses to the question "How concerned are you that trying will be rewarded with pain?", where responses range from 1 (not at all concerned) to 10 (extremely concerned).
Subjective Disability The OSWESTRY low back pain disability questionnaire was in its widely known and utilized form, consisting of ten sections ranging over topics as varied as pain intensity, sleeping, sex life, social life, and traveling. Two additional items were obtained from the first group of self-response questions presented to the subjects: IMPAIRMENT and PAIN INTENSITY. The variable IMPAIRMENT consists of responses to the question "To what extent does your back problem impair your daily activities?", where responses range from 1 (not at all) to 10 (almost entirely). The variable PAIN INTENSITY consists of responses to the question "How intense is your back pain at this very moment?", where responses range from 1 (no pain at all) to 10 (worst pain possible).
Isokinetic Performance Isokinetic performance was measured with a Lidoback Isokinetic Dynamometer (Lidoback System, Loredan Biomedical Inc.). Following a verbal description of the procedure, subjects were placed into the Lidoback device following the standard protocol for measurement in the standing position (10). Pulse rate measurements were obtained with a Sport-Tester r electronic ECG-detector. After three initial
92
Kaivanteb Estlander, Moneta, and Vanharanta
warm-up trials, subjects performed five consecutive flexion-extension movements at the speeds of 50 degrees/second (~ and 100~ and 20 consecutive flexion-extension movements at the speed of 150~ The uninitiated subject may initially experience this latter speed as being uncomfortably fast, but in general the procedure is considered safe.5 Before each increase in speed, subjects rested for at least four minutes, or as long as it took for their pulse rate to decrease to an arbitrary 110 beats/minute. Separate measures of isokinetic total work done were obtained for flexion and extension at each respective speed, resulting in a total of 6 outcome measures of isokinetic performance. Statistical Procedures
In spite of the unwieldiness of considering numerous models simultaneously, individual models were fitted for all six measures of isokinetic performance in order
to reveal the effects of increasing speed and the qualitative differences between flexion and extension that would remain obscured if only a single measure were considered. Using SPSS/PC+ TM, models of the form Y = 130 + 131AGE + 132SEX + [33WEIGHT + [34HEIGHT + [~sX + E were fitted for each outcome variable of total work done, where the fifth term, x, represents an independent measure of depression, expectancies or disability, and is the residual error term. The block of four anthropometric variables was included for control purposes only. By substituting the ten independent measures into the model one at a time, ten separate models were obtained for each speed. Predictive power was operationalized in terms of variance explained. Fitted models were screened on the basis of the t-test significance of the fifth predictor where the utilized alpha-level was derived from the Bonferroni inequality (a = 0.05/10 = 0.005), and were compared on the basis of their adjusted coefficient of determination (R2). Thus the research task reduces to a comparison of portions of total variance explained by models in which the nonsomatic variable is significant (18, 19).
RESULTS The results of the regression models are presented in Tables I and II. The most immediate and evident finding which these numerous models afford concerns the depression measures; these variables are systematically insignificant in both flexion and extension, indicating that there was no association between these depression measures and isokinetic performance in this population. Furthermore, it is apparent that the anthropometric variables predict performance better, though inconsistently, at elevated speeds. In flexion the anthropometric variables account for .183 of the variance at 50~ only .134 of the variance at 100~ but .200 of the variance at 5One of this paper's anonymousreferees reported havingpersonallywitnessedan LBP patient herniate a disc on the Lidoback.The patient had no objective indicationof a disc problem prior to the test.
lsokinetic Performance in LBP Patients: The Predictive Power of SES
93
Table I. Portion of Variance Explained and t-Test p-Values ( a = 0.005) Obtained from Regressing Total Isokinetie Work Done on Anthropometric Variables and Nonsomatie Measures in Flexion Anthropometric variables' block R-2 BDI Mood disorders PSEQ R-2 SES ~-2 Certain ~-2 Endure concern Pain concern Oswestry k-2 _~pairment Pain intensity R-2
Flexion 50*Is .18383
100~ .13449
150*/s .19991
ns. ns. p = .0003 .27865 p = .0000 .47467 p = .0000 .33068 as. ns. p = .0000 .39218 p = .0001 .30612 p = .0000 .31627
ns. ns. p = .0011 .21699 p = .0000 .35642 p = .0001 .25511 ns. Its. p = .0000 .31734 p = .0013 .21566 p = .0005 .23164
ns. ns. p = .0029 .26276 p = .0000 .36865 p = .0005 .28660 ns. ns. p = .0000 .35419 p = .0017 .27151 p = .0005 .28916
Table 1L Portion of Variance Explained and t-Test p-Values ( a = 0.005) Obtained from Regressing Total Isokinetic Work Done on Anthropometric Variables and Nonsomatic Measures in Extension Anthropometric variables' block R-z BDI Mood disorders PSEQ SES Certain R-'2 Endure concern Pain concern Oswestry ~2 I_~pairment Pain intensity R-2
150~
In extension
counting
the pattern
for .114 of the variance
to .151 of the variance
Extension
50*/s .11402
100*/s .15954
ns. ns. p = .0009 .20115 p -- .0000 .33387 p = .0001 .23648 ns. ns. p = .0000 .29658 p = .0000 .25845 p = .0000 .25045
ns. ns. p = .0031 .22437 p = .0000 .29931 p = .0003 .25622 ns. ns. p = .0000 .31983 p = .0003 .26094 p = .0001 .27911
150*/s .15149 ns. ns. ns. p = .0001 .27101 p = .0020 .22197 ns. ns. p = .0001 .27325 p = .0026 .22057 p = .0003 .25402
is r e v e r s e d , w i t h t h e a n t h r o p o m e t r i c a t 50~
a t 150*/s. I n o r d e r
.160 of the variance
variables
a t 100~
to focus on nonsomatic
ac-
dropping
variables alone
94
Kaivanto, Estlander, Moneta, and Vanharanta
it is necessary to filter out this variance accounted for by the anthropometric variables, particularly because of its irregularity across speeds and reversed directionality in extension with respect to flexion; this was accomplished by subtracting the of the anthropometric model from the t/2 of the model including the nonsomatic variable at each speed. Thus taken alone, the non-somatic variables consistently become poorer predictors as the testing speed is increased. The models fitted for PSEQ, SES, CERTAIN, OSWESTRY, IMPAIRMEN'I; and PAIN INTENSITY confirm this without exception in both flexion and extension. Moreover, it would be tempting to add further that both the anthropometric and the nonsomatic variables predict flexion better than extension. In the case of PSEQ, SES, CERTAIN, and OSWESTRY, this is true at all speeds. However, a reversal occurs at the speed 100~ in the predictivity of the anthropometric variables, as occurs likewise in the case of the variable IMPAIRMENT at the speeds 50~ and 100~ Finally, the variable PAIN INTENSITY systematically predicts extension better than it predicts flexion. But on the whole, flexion is predicted to a greater extent; among the portions of variance explained by the anthropometric variables' block and the portions of variance explained by the significant nonsomatic variables, the number of exceptions amount to only six, while concordant instances number 14. That the mode of the ~2 figures occurs in flexion provides a further justification of this synopsis. As can be seen from the k 2 figures, SES is the best predictor of isokinetic performance in flexion, after which OSWESTRY is next best. Considering the variance explained by SES and OSWESTRY alone, it becomes apparent that SES loses predictive power with respect to OSWESTRY as the test speed is increased (see Figs. 2 and 3). In other words, the rate at which SES loses predictive power (with 0.3
~ SES
69
.~ .t-
E~ E s
r
t-
0.2
-(
88 =_
>
.~-
> o.1
!
50
10o
150
degrees/second
Fig. 2. Fexion: Additional variance explained across speeds by adding SES and OSWESTRY to the model containing only anthropometric variables.
Isokinetic Performance in LBP Patients: The Predictive Power of SES
95
0.3 ~SES
ffl .--
O" OSWESTRY
~
E~
~2
0.2-
0.1 50
I 1oo
150
degrees/second Fig. 3. Extension: Additional variance explained across speeds by adding SES and OSWESTRY to the model containing only anthropometrie variables.
increasing speed) is greater than the rate at which OSWESTRY loses predictive power (with increasing speed). In flexion at 500/s the initial difference in variance explained by the two predictors is so great (.47467-.39218 = .08249) that SES remains the better predictor at 150~ in spite of a higher predictivity loss-rate. While in extension at 50~ SES accounts for moderately more variance than OSWESTRY does, the difference is quickly eroded by the relative loss-rates, leaving OSWESTRY the superior predictor at the speeds 100~ and 1500/s. Overall, however, SES appears to be a better predictor of total work done than the other nine nonsomatic measures. This must be qualified, since the results also indicate that SES' predictive power deteriorates with increased test speeds at a greater rate than OSWESTRY's predictive power does, and, that this deterioration becomes particularly manifest in extension.
DISCUSSION There is only one clear result which this study may presume to present: given the characteristics of the population conditioned by the prevailing institutional environment, SES predicted isokinetic total work done better than the nine other measures considered. As defined by this criterion, the predictive criterion validity of SES surpassed that of the remaining nine measures. To date, there is reasonable evidence confirming the reliability of trunk flexion-extension iso-measurement up to 120~ above which there is a general paucity of satisfactory evidence (4). Recent evidence suggests that differences across speeds
96
Kaivanto, Estlander, Moneta, and Vanharanta
in chronic LBP patients isokinetic performance may be attributable to differences in type 1 (slow twitch) and type 2 (fast twitch) muscle fibers (20). The decrease in R"2 observed at the higher speeds (100~ and 150~ used in this study could potentially be thought of as the result of a speed-related deterioration in the reliability of iso-measurement. Nevertheless, the anthropometric variables explain a larger proportion of variance at 150~ than they do at 500/s. Presumably this would not take place if the highest speed produced the most unreliable iso-measurements. Conversely, the decrease in the regression models' R"2 can be ascribed to the properties of the independent measures (the predictors) themselves. They simply do not remain as informative when considering isokinetic performance at the higher speeds. It would be desirable to furnish this downward trend in the predictive value of SES with a particular interpretation in terms of one of its dimensions: magnitude, generality, or strength. The magnitude or level of efficacy expectancies conveys the level of task difficulty to which they are extended. The generality of efficacy expectancies refers to the breadth of activities and situations which they encompass. Finally, the strength of efficacy expectancies indicates the tenacity of their persistence in the face of disconfirming experiences (5). Stll, a unique interpretation in terms of only one of these three presently known dimensions of efficacy expectancies does not take shape. If the three different speeds of isokinetic trunk flexion-extension are equated as examples of a single task differing only in difficulty, then an interpretation in terms of the magnitude-dimension follows naturally. Here, as above, a counter-argument based on the anthropometric variables carries considerable force. Can disparate speeds truly be equated to a single "task" if the contribution of anthropometric factors to performance is unique to a particular speed and different from other speeds? If, on the other hand, each test speed is taken to be a qualitatively peculiar "task" unto itself, an interpretation in terms of the generality-dimension would be forthcoming. However, efficacy expectancies are defined as convictions or beliefs; nothing in the definition stipulates that the beliefs must be born out as accurate and justified. An efficacy expectancy may initially extend or generalize to a particular task well enough, but this will not necessarily imply corresponding performance. The task may surprise the subject by being more difficult or more easy than he/she had expected it to be, with performance adjusting accordingly. After the fact (testing completed) it is reasonable to conclude that the expectancy-measurements obtained with SES generalize better to 50~ than to 150~ but there is no information on whether, during the test itself, the subjects' efficacy expectancies actually encompassed 1500/s or not. The third dimension, strength, further complicates the picture. It would appear that each of the dimensions is involved in some way, but more refined study designs, for example accounting for the differential atrophication of type 1 and type 2 muscle fibers, are required to resolve the issue. Nevertheless, the results indicate that SES is more informative with respect to the slower speeds. Although previous studies have found isokinetic measures to be highly correlated across speeds, the regression analytic approach utilized here suggests that there are indeed interesting distinctions to be made between speeds. In light of the size principle and the specificity of training theory, these differences
Isoldnetic Performance in LBP Patients: The Predictive Power of SES
97
were not anticipated. Based on SES' predictive power at slower speeds, one might hypothesize that isokinetic testing at slower speeds is more relevant to everyday functioning in activities such as walking, running, carrying shopping bags, standing in queue, sitting in an armchair, sitting at a desk, or bending over the sink. Conversely, this should be remembered when using SES as a diagnostic or evaluational tool. If isokinetic performance at slower speeds is representative of actual work capability, then the major implication of this study is that efficacy expectancies of motorial activities occurring in nonwork environments are an important predictor of work capability. The implication for evaluation is: what happens outside the workplace is very important. Perhaps by focusing treatment programs on natural behaviors like walking, running, or biking, we may be able to improve the most important type of efficacy expectancies; in turn, this improvement may lead to increased work capability. Problems and Limitations
Numerous factors contribute to the case for holding reservations as to the generalizability of the results obtained presently. To begin with, the study population consisted of a diverse group of Finns living in the region surrounding Oulu, i.e., between the 64th parallel and the arctic circle. This population may well sport idiosyncrasies relative to the southern, urban Finnish population, to which cultural and institutional factors peculiar to Finland must be added when international comparability is sought. Additionally, it must be conceded that the subjects were studied at a very special historical moment characterized by a particular social atmosphere, which is not expected to repeat itself. In this context, the issue of the distinction between overt and covert behaviors raised by Dolce becomes especially relevant (6). Likewise the outcome measure deserves qualification. In their review of the iso-measurement literature, Newton and Waddell observed that "there is strong evidence of a learning effect between the first and second test sessions and of the need to use a second iso-test session on a separate day as the baseline to allow for initial learning effects" (4). In spite of this, only a single session was used for isokinetic testing in this study. Furthermore, of all the possible output variables that isokinetic testing commonly produces, total work done was selected and utilized in this study. In general there is very little information on variables computed from the values of torque and angle, work included. Newton and Waddell added, however, that the single study which addressed this matter found the isokinetic measures of torque, work, and power in trunk flexion-extension to be highly correlated (4).
CONCLUSION Owing to the nature of the study population, conclusions entertained here must necessarily be only tentative and limited to chronic low back pain patients. Within these limits, the predictive value of SES was confirmed with respect to the
98
Kaivanto, Estlander, Moneta, and Vanharanta
other measures included in this study. SES displayed the strongest association with isokinetic total work done. At the higher speeds however--particularly in extension-the relative predictive power of OSWESTRY gained over SES. Ensuing from this observation is the suggestion that there may be qualitative differences across test speeds and flexion-extension calling for special attention and accordingly requiring special adjustment when interpreting the results of isokinetic testing.
ACKNOWLEDGMENTS This study was supported by a grant from the Gyllenberg Foundation. The authors would also like to thank the Editor-in-Chief as well as three anonymous referees for their assistance in bringing about the final version of this paper.
REFERENCES 1. Bigos SJ, Batti6 M, Spengler DM, et al. A prospective study of work perceptions and psychosocial factors affecting the report of back injury. Spine 1991; 16: 1-6. 2. Frymoyer JW, Cats-Baril W. Predictors of low back pain disability. Clin Orthop Re/at Res 1987; 221: 89-98. 3. Riihim~dd H. Low-back pain, its origin and risk indicators. Scand J Work Env Health 1991; 17: 81-90. 4. Newton M, Waddell G. Trunk strength testing with iso-maehines: Part 1: Review of a decade of scientific evidence. Spine 1993; 18: 801-811. 5. Bandura A. Self-efficacy: Toward a unifying theory of behavioral change. Psychol Rev 1977; 84: 191-215. 6. Dolce JJ. Self-efficacy and disability beliefs in behavioral treatment of pain. Behav Res Ther 1987; 25: 289-299. 7. Council JR, Ahem DK, Follick MJ, Kline CL Expectancies and functional impairment in chronic low back pain. Pain 1988; 33: 323-331. 8. Kitsch I. Response expectancy as a determinant of experience and behavior. Am Psychol 1985; 40: 1189-1202. 9. Palenzuela DL. The expectancy construct within the social learning theories of Rotter and Bandura: A reply to Kirseh's approach. J Soc Behav Person 1987; 2: 437-452. 10. Estlander A-M, Vanharanta H, Moneta GB, Kaivanto K. Anthropometric variables, self-efficacy beliefs, and pain and disability ratings on the isokinetic performance of low back pain patients. Spine 1994; 19: 941-947. 11. Bongers PM, de Winter CR, Kompier MAJ, Hildebrandt VH. Psychosocial factors at work and muscoloskeletal disease. Scand J Work Env t-l-ealth 1993; 19: 297-312. 12. Beck AT, Ward CH, Mendelson M, Mock J, Erbangh J. An inventory for measuring depression. Arch Gen Psychiat 1961; 4: 561-571. 13. Husman K. Symptoms of car-painters with long-term exposure to a mixture of organic solvents. Scand J Work Env Health 1980; 6: 19-32. 14. H~irminen H. Neurotoksisten Haittojen Seulonta: Oirekyselyt ja Psykologiset Testit (The screening of neuro-to~dc effects: Symptom questionnaires and psychological tests). Helsinki: TyOterveyslaitos, 1989. 15. Nicholas MK. Self-Efficacy and Chronic Pain. Paper presented at the Annual Conference of the British Psychological Society, St. Andrews, March 1989. 16. Fairbanks JCT, Couper J, Davies JB, O'Brien JP. The Oswestry Low Back Pain Disability Questionnaire. Physiotherapy 1980; 66: 271-273. 17. Nicholas MK, Wilson PH, Goyen J. Comparison of cognitive-behavioral group treatment and an alternative non-psycholo~!cal treatment for chronic low back pain. Pain 1992; 48: 339-347. 18. Norusis MJ. SPSS/PC+ ~ V2.0 base manual for the IBM PC/XT/AT and PS/2 Chicago: SPSS, 1988. 19. Weisberg S. Applied Linear Regression (2nd Ed.). New York: John Wiley & Sons, 1985.
lsokinetic Performance in LBP Patients: The Predictive Power of SES
99
20. Rissanen A, Kalimo H, Alaranta H. Muscle histomorphology and isokinetic strength: A prospective study of intensive training in chronic low back pain patients. In: Book of abstracts of the international society for the study of the lumbar spine; 1993 June 15-19. Marseilles: International Society for the Study of the Lumbar Spine, 1993.
Isokinetic Performance in Low Back Pain Patients: The Predictive Power of the Self-Efficacy Scale K. K. Kaivanto, 1,4 A-M. Estlander, 2 G. B. Moneta, 1 and H. Vanharanta 3
The Self-Efficacy Scale (SES) has been found to predict isokinetic performance better than anthropometric variables. This study tests the predictive power of SES further against other measures of efficacy expectancies as well as measures of depression and perceived disability. A group of 105 chronic back pain patients was administered Beck's Depression Inventory (BDI), SES, the Pain Self-Efficacy Questionnaire (PSEQ), and the Oswestry low back pain disability questionnaire (OSWESTRY). Total isokinetic work done was measured at slow, medium and high speeds, for which multiple regression models were fitted controlling for sex, ag~ weight and height. The results confirmed SES to be the best overall predictor of isokinetic performance. BDI was not significant as a predictor of isokinetic performance. The models also revealed that SES predicts less well with increases in the test speed, particularly in extension. These results provide further evidence of the diagnostic value of SES relative to OSWESTRY and PSEQ. KEY WORDS: isokinetic performance; low back pain; self-efficacy expectancies; disability; prediction.
INTRODUCTION In the ever greater number of studies being performed on Low Back Pain (LBP) the role attributed to nonsomatic factors is becoming increasingly pronounced (1-3). As with LBP in general, evidence is mounting that behavioral and psychological factors also affect the results of dynamic strength iso-measurement (4). It has been suggested that the significant motor impairment of chronic LBP patients can be at least partially accounted for by avoidance learning, in which the negative reinforcement of experienced pain causes patients to avoid particular types of motor activity. Expectations come to be formed with respect to each tInstitute of Occupational Health, Helsinki, Finland. 2Finnish Back Institute, Helsinki, Finland. 3Oulu University Central Hospital, Oulth Finland. 4Correspondence should be directed to Kim Kaivanto, Institute of Occupational Health, H-department, Topeliuksenkatu 41 a A, FIN-00250 Helsinki, Finland. 87 1053-0487/95/0600-0087507.50/001995PlenumPublishingCorporation
88
Kaivanto, Estlander, Moneta, and Vanharanta
kind of activity, guiding the choice of what will be attempted and what will be avoided. But in addition to being differentiated by activity, expectancies of differing definitional qualities have also been distinguished. Yet while diverse measurement instruments have been implemented, self-efficacy has been the principal expectancy construct addressed by many theoretical discussions and psychometric instruments. Self-efficacy was originally conceived as the central concept within a theoretical framework for analyzing changes in avoidant and fearful behavior. As proposed by Bandura (5), this theoretical framework submits that behavioral changes are mediated by cognitive processes irrespective of the particular methods used to bring them about, and that cognitive events are primarily modulated by experiences of mastery emanating from effective performance. Within this context Bandura found it necessary to distinguish sharply between outcome expectancies and efficacy expectancies. Definitionally, an outcome expectancy is "a person's estimate that a given behavior will lead to certain outcomes." In contrast, an efficacy expectancy is "the conviction that one can successfully execute the behavior required to produce the outcomes" (5). There is arguably a degree of overlap between these two forms of expectancies, in that a subjective "estimate" is in some ways as much a belief as a "conviction" is. Subsequent work on expectancies has spawned an array of variations in the definition of expectancies, some of them referring to comparable constructs and others to differing constructs (6-9). Within the present paper, this "nebulous and elusive" (9) conceptual field will be avoided by adhering to Bandura's original formulation. A number of new instruments for the measurement of dynamic muscle strength were developed and commercialized in the 1980s. These isomachines were thought to provide relevant, accurate, and reliable information for diagnostic, screening, and rehabilitation follow-up purposes. Current evidence, however, points in quite the opposite direction (4). If iso-machines are to have any practical clinical use, a more profound understanding of the factors which affect performance as measured by them is necessary. In current practice, it is common to adjust isokinetic performance for the anthropometric variables of sex, body weight and age, including perhaps height at most; remaining performance variations are ascribed to supposed somatic factors. A recent study performed by Estlander et al. indicates, however, that anthropometric variables are not in fact very good predictors of isokinetic performance, and that self-efficacy beliefs as measured by the Self-Efficacy Scale (SES) are a much more powerful predictor (10). The purpose of this study is to test the predictive power of SES further against other competing alternatives suggested by the literature, including measures of depression, efficacy expectancies and subjective disability. Depression is nominated by the epidemiologically detected association between psychological or emotional problems and diagnosed back disorders (11). Efficacy expectancies are nominated by the documented predictive power of SES (10). Finally, subjective disability is included because iso-measures have been claimed to provide information relevant to the diagnosis and identification of back conditions. The present study focuses solely on the predictive power of the selected attribute measures. Questions concerning the connection between the attributes and isoki-
lsokinetic Performance in LBP Patients: The Predictive Power of SES
89
netic performance will be taken up elsewhere. The operational task undertaken here is an assessment of the predictive power of Beck's Depression Inventory (BDI), the Mood Disorders scale, the Pain Self-Efficacy Questionnaire (PSEQ), the SelfEfficacy Scale (SES), the Oswestry low back pain disability questionnaire (OSWESTRY), and a number of single-question items.
METHODS Subjects
Between November 1990 and September 1991, 105 patients from the Oulu region were referred to the Oulu University Central Hospital by their doctors for undergoing further diagnostic and treatment procedures. As evaluated by their doctors, these patients had evident degenerative disc disease and recurrent or persistent low back pain with or without root impairment. All 105 patients were submitted to a thorough medical evaluation, a part of which consisted of the administration of an array of question-series prior to isokinetic testing. None of the patients failed to meet the range of movement inclusion criteria of 90 degrees of flexion and 15 degrees of extension as measured in the iso-machine's fixed position. On average low back pain had been troubling the patients for 4.7 years, ranging from 0.5 to 30 years; thus there were no acute injury patients with less than 6 weeks since the onset of pain. Fifty-six percent were male. Forty-nine percent were working, and 48% were sick listed or enjoying retirement benefits. There were no statistically significant differences detected between these two groups on any of the six measures of isokinetic performance (10). Measurements
After general questions inquiring demographic, background, and anthropometric information, the questionnaire included the following series' Finnish-language renditions: BDI, the Symptom Questionnaire, PSEQ, SES, and OSWESTRY (10, 12-16). Both BDI and the Oswestry series were in their standard and widely utilized forms. Individual questions not part of any particular established series were also posed, bringing the total number of nonsomatic variables to ten. Finally, subjects' isokinetic performance was tested. Depression
BDI was in its widely known and utilized form. The Symptom Questionnaire on the other hand was originally constructed with the Finnish population in mind. The 31 items of the Symptom Questionnaire yield several subscales upon coding, from among which the MOOD DISORDERS (14) scale was chosen for the purposes of this study. This particular scale is composed of six items which assess the
90
Kaivanto, Estlander, Moneta, and Vanlmranta
frequency of mood changes, depressive episodes, heightened irritability, restlessness, fear, and apathy.
Efficacy Expectancies Experience in the use of PSEQ is not diminutive (15, 17). The PSEQ consists of ten items in which respondents are asked to gauge how confident they are that they can do each of the ten activities or functions despite their pain, where rating takes place on 7-point scales ranging from "not at all confident" at 0 to "completely confident" at 7. Items refer to activities and functions which chronic pain patients generally find problematic. It includes statements such as: "I can still ac-
W e w o u l d l i k e to k n o w w h a t y o u t h i n k y o u r p e r f o r m a n c e c a p a b i l i t y is a__~t this v e r y m o m e n t . F o r e a c h q u e s t i o n p l e a s e c i r c l e the a n s w e r w h i c h describes your situation.
H o w long a r e y o u a b l e to walk?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
How long are y o u a b l e to run?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
How long are under 2 2-5 y o u a b l e to c a r r y minutes shopping bags (2 x 4-5 k i l o s ) ?
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long a r e under 2 2-5 y o u a b l e to s t a n d minutes eg. in q u e u e ?
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long a r e y o u a b l e to r i d e a bicycle?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long are y o u a b l e to sit in an a r m c h a i r ?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
H o w long a r e y o u a b l e to sit at a desk?
under 2 2-5 minutes
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
How long are under 2 2-5 y o u a b l e to w o r k minutes bending over (eg. w a s h i n g dishes, vacuum-cleaning, fixing a car)?
5-10
10-15
15-25
25-35
25-45
o v e r 45 minutes
Fig. 1. The Serf-Efficacy Scale (SES) questionnaire.
Isoldnetic Perfomnnee in LBP Patients: The Predictive Power of SES
91
complish most of my goals in life, despite the pain," and "I can still do many of the things I enjoy doing, such as hobbies or leisure activities, despite the pain" (15). The SES is in turn composed of eight items--one for each of the activities of walking, running, carrying shopping bags, standing, bicycling, sitting in an armchair, sitting at a desk, and working in a forward leaning position--in which the respondent is asked to identify the duration of time he/she thinks he/she is able to endure each activity on an 8 point scale (less than 2 minutes, 2-5, 5-10, 10-15, 15-25, 25-35, 35-45, more than 45 minutes). SES is computed as the sum of these eight items (range 8-64). Much of the contemporary work in designing and modifying self-efficacy measures has concerned the choice of activities to be assessed and the specificity or generality thus obtained. While the SES refers to common, everyday activities, it gains its distinctiveness from framing its items in terms of endurance-time (see Fig. 1) (10). In addition, three efficacy expectancy proxy items were posed to the subjects just before isokinetic testing. The variable CERTA/N consists of responses to the question "How certain are you that you can complete the lifting test as instructed?", where responses range from 1 (absolutely certain) to 10 (I don't believe I can). The variable E N D U R E CONCERN consists of responses to the question "How concerned are you that your back won't be able to endure this test?", where responses range from 1 (not concerned at all) to 10 (extremely concerned). The variable PAIN CONCERN consists of responses to the question "How concerned are you that trying will be rewarded with pain?", where responses range from 1 (not at all concerned) to 10 (extremely concerned).
Subjective Disability The OSWESTRY low back pain disability questionnaire was in its widely known and utilized form, consisting of ten sections ranging over topics as varied as pain intensity, sleeping, sex life, social life, and traveling. Two additional items were obtained from the first group of self-response questions presented to the subjects: IMPAIRMENT and PAIN INTENSITY. The variable IMPAIRMENT consists of responses to the question "To what extent does your back problem impair your daily activities?", where responses range from 1 (not at all) to 10 (almost entirely). The variable PAIN INTENSITY consists of responses to the question "How intense is your back pain at this very moment?", where responses range from 1 (no pain at all) to 10 (worst pain possible).
Isokinetic Performance Isokinetic performance was measured with a Lidoback Isokinetic Dynamometer (Lidoback System, Loredan Biomedical Inc.). Following a verbal description of the procedure, subjects were placed into the Lidoback device following the standard protocol for measurement in the standing position (10). Pulse rate measurements were obtained with a Sport-Tester r electronic ECG-detector. After three initial
92
Kaivanteb Estlander, Moneta, and Vanharanta
warm-up trials, subjects performed five consecutive flexion-extension movements at the speeds of 50 degrees/second (~ and 100~ and 20 consecutive flexion-extension movements at the speed of 150~ The uninitiated subject may initially experience this latter speed as being uncomfortably fast, but in general the procedure is considered safe.5 Before each increase in speed, subjects rested for at least four minutes, or as long as it took for their pulse rate to decrease to an arbitrary 110 beats/minute. Separate measures of isokinetic total work done were obtained for flexion and extension at each respective speed, resulting in a total of 6 outcome measures of isokinetic performance. Statistical Procedures
In spite of the unwieldiness of considering numerous models simultaneously, individual models were fitted for all six measures of isokinetic performance in order
to reveal the effects of increasing speed and the qualitative differences between flexion and extension that would remain obscured if only a single measure were considered. Using SPSS/PC+ TM, models of the form Y = 130 + 131AGE + 132SEX + [33WEIGHT + [34HEIGHT + [~sX + E were fitted for each outcome variable of total work done, where the fifth term, x, represents an independent measure of depression, expectancies or disability, and is the residual error term. The block of four anthropometric variables was included for control purposes only. By substituting the ten independent measures into the model one at a time, ten separate models were obtained for each speed. Predictive power was operationalized in terms of variance explained. Fitted models were screened on the basis of the t-test significance of the fifth predictor where the utilized alpha-level was derived from the Bonferroni inequality (a = 0.05/10 = 0.005), and were compared on the basis of their adjusted coefficient of determination (R2). Thus the research task reduces to a comparison of portions of total variance explained by models in which the nonsomatic variable is significant (18, 19).
RESULTS The results of the regression models are presented in Tables I and II. The most immediate and evident finding which these numerous models afford concerns the depression measures; these variables are systematically insignificant in both flexion and extension, indicating that there was no association between these depression measures and isokinetic performance in this population. Furthermore, it is apparent that the anthropometric variables predict performance better, though inconsistently, at elevated speeds. In flexion the anthropometric variables account for .183 of the variance at 50~ only .134 of the variance at 100~ but .200 of the variance at 5One of this paper's anonymousreferees reported havingpersonallywitnessedan LBP patient herniate a disc on the Lidoback.The patient had no objective indicationof a disc problem prior to the test.
lsokinetic Performance in LBP Patients: The Predictive Power of SES
93
Table I. Portion of Variance Explained and t-Test p-Values ( a = 0.005) Obtained from Regressing Total Isokinetie Work Done on Anthropometric Variables and Nonsomatie Measures in Flexion Anthropometric variables' block R-2 BDI Mood disorders PSEQ R-2 SES ~-2 Certain ~-2 Endure concern Pain concern Oswestry k-2 _~pairment Pain intensity R-2
Flexion 50*Is .18383
100~ .13449
150*/s .19991
ns. ns. p = .0003 .27865 p = .0000 .47467 p = .0000 .33068 as. ns. p = .0000 .39218 p = .0001 .30612 p = .0000 .31627
ns. ns. p = .0011 .21699 p = .0000 .35642 p = .0001 .25511 ns. Its. p = .0000 .31734 p = .0013 .21566 p = .0005 .23164
ns. ns. p = .0029 .26276 p = .0000 .36865 p = .0005 .28660 ns. ns. p = .0000 .35419 p = .0017 .27151 p = .0005 .28916
Table 1L Portion of Variance Explained and t-Test p-Values ( a = 0.005) Obtained from Regressing Total Isokinetic Work Done on Anthropometric Variables and Nonsomatic Measures in Extension Anthropometric variables' block R-z BDI Mood disorders PSEQ SES Certain R-'2 Endure concern Pain concern Oswestry ~2 I_~pairment Pain intensity R-2
150~
In extension
counting
the pattern
for .114 of the variance
to .151 of the variance
Extension
50*/s .11402
100*/s .15954
ns. ns. p = .0009 .20115 p -- .0000 .33387 p = .0001 .23648 ns. ns. p = .0000 .29658 p = .0000 .25845 p = .0000 .25045
ns. ns. p = .0031 .22437 p = .0000 .29931 p = .0003 .25622 ns. ns. p = .0000 .31983 p = .0003 .26094 p = .0001 .27911
150*/s .15149 ns. ns. ns. p = .0001 .27101 p = .0020 .22197 ns. ns. p = .0001 .27325 p = .0026 .22057 p = .0003 .25402
is r e v e r s e d , w i t h t h e a n t h r o p o m e t r i c a t 50~
a t 150*/s. I n o r d e r
.160 of the variance
variables
a t 100~
to focus on nonsomatic
ac-
dropping
variables alone
94
Kaivanto, Estlander, Moneta, and Vanharanta
it is necessary to filter out this variance accounted for by the anthropometric variables, particularly because of its irregularity across speeds and reversed directionality in extension with respect to flexion; this was accomplished by subtracting the of the anthropometric model from the t/2 of the model including the nonsomatic variable at each speed. Thus taken alone, the non-somatic variables consistently become poorer predictors as the testing speed is increased. The models fitted for PSEQ, SES, CERTAIN, OSWESTRY, IMPAIRMEN'I; and PAIN INTENSITY confirm this without exception in both flexion and extension. Moreover, it would be tempting to add further that both the anthropometric and the nonsomatic variables predict flexion better than extension. In the case of PSEQ, SES, CERTAIN, and OSWESTRY, this is true at all speeds. However, a reversal occurs at the speed 100~ in the predictivity of the anthropometric variables, as occurs likewise in the case of the variable IMPAIRMENT at the speeds 50~ and 100~ Finally, the variable PAIN INTENSITY systematically predicts extension better than it predicts flexion. But on the whole, flexion is predicted to a greater extent; among the portions of variance explained by the anthropometric variables' block and the portions of variance explained by the significant nonsomatic variables, the number of exceptions amount to only six, while concordant instances number 14. That the mode of the ~2 figures occurs in flexion provides a further justification of this synopsis. As can be seen from the k 2 figures, SES is the best predictor of isokinetic performance in flexion, after which OSWESTRY is next best. Considering the variance explained by SES and OSWESTRY alone, it becomes apparent that SES loses predictive power with respect to OSWESTRY as the test speed is increased (see Figs. 2 and 3). In other words, the rate at which SES loses predictive power (with 0.3
~ SES
69
.~ .t-
E~ E s
r
t-
0.2
-(
88 =_
>
.~-
> o.1
!
50
10o
150
degrees/second
Fig. 2. Fexion: Additional variance explained across speeds by adding SES and OSWESTRY to the model containing only anthropometric variables.
Isokinetic Performance in LBP Patients: The Predictive Power of SES
95
0.3 ~SES
ffl .--
O" OSWESTRY
~
E~
~2
0.2-
0.1 50
I 1oo
150
degrees/second Fig. 3. Extension: Additional variance explained across speeds by adding SES and OSWESTRY to the model containing only anthropometrie variables.
increasing speed) is greater than the rate at which OSWESTRY loses predictive power (with increasing speed). In flexion at 500/s the initial difference in variance explained by the two predictors is so great (.47467-.39218 = .08249) that SES remains the better predictor at 150~ in spite of a higher predictivity loss-rate. While in extension at 50~ SES accounts for moderately more variance than OSWESTRY does, the difference is quickly eroded by the relative loss-rates, leaving OSWESTRY the superior predictor at the speeds 100~ and 1500/s. Overall, however, SES appears to be a better predictor of total work done than the other nine nonsomatic measures. This must be qualified, since the results also indicate that SES' predictive power deteriorates with increased test speeds at a greater rate than OSWESTRY's predictive power does, and, that this deterioration becomes particularly manifest in extension.
DISCUSSION There is only one clear result which this study may presume to present: given the characteristics of the population conditioned by the prevailing institutional environment, SES predicted isokinetic total work done better than the nine other measures considered. As defined by this criterion, the predictive criterion validity of SES surpassed that of the remaining nine measures. To date, there is reasonable evidence confirming the reliability of trunk flexion-extension iso-measurement up to 120~ above which there is a general paucity of satisfactory evidence (4). Recent evidence suggests that differences across speeds
96
Kaivanto, Estlander, Moneta, and Vanharanta
in chronic LBP patients isokinetic performance may be attributable to differences in type 1 (slow twitch) and type 2 (fast twitch) muscle fibers (20). The decrease in R"2 observed at the higher speeds (100~ and 150~ used in this study could potentially be thought of as the result of a speed-related deterioration in the reliability of iso-measurement. Nevertheless, the anthropometric variables explain a larger proportion of variance at 150~ than they do at 500/s. Presumably this would not take place if the highest speed produced the most unreliable iso-measurements. Conversely, the decrease in the regression models' R"2 can be ascribed to the properties of the independent measures (the predictors) themselves. They simply do not remain as informative when considering isokinetic performance at the higher speeds. It would be desirable to furnish this downward trend in the predictive value of SES with a particular interpretation in terms of one of its dimensions: magnitude, generality, or strength. The magnitude or level of efficacy expectancies conveys the level of task difficulty to which they are extended. The generality of efficacy expectancies refers to the breadth of activities and situations which they encompass. Finally, the strength of efficacy expectancies indicates the tenacity of their persistence in the face of disconfirming experiences (5). Stll, a unique interpretation in terms of only one of these three presently known dimensions of efficacy expectancies does not take shape. If the three different speeds of isokinetic trunk flexion-extension are equated as examples of a single task differing only in difficulty, then an interpretation in terms of the magnitude-dimension follows naturally. Here, as above, a counter-argument based on the anthropometric variables carries considerable force. Can disparate speeds truly be equated to a single "task" if the contribution of anthropometric factors to performance is unique to a particular speed and different from other speeds? If, on the other hand, each test speed is taken to be a qualitatively peculiar "task" unto itself, an interpretation in terms of the generality-dimension would be forthcoming. However, efficacy expectancies are defined as convictions or beliefs; nothing in the definition stipulates that the beliefs must be born out as accurate and justified. An efficacy expectancy may initially extend or generalize to a particular task well enough, but this will not necessarily imply corresponding performance. The task may surprise the subject by being more difficult or more easy than he/she had expected it to be, with performance adjusting accordingly. After the fact (testing completed) it is reasonable to conclude that the expectancy-measurements obtained with SES generalize better to 50~ than to 150~ but there is no information on whether, during the test itself, the subjects' efficacy expectancies actually encompassed 1500/s or not. The third dimension, strength, further complicates the picture. It would appear that each of the dimensions is involved in some way, but more refined study designs, for example accounting for the differential atrophication of type 1 and type 2 muscle fibers, are required to resolve the issue. Nevertheless, the results indicate that SES is more informative with respect to the slower speeds. Although previous studies have found isokinetic measures to be highly correlated across speeds, the regression analytic approach utilized here suggests that there are indeed interesting distinctions to be made between speeds. In light of the size principle and the specificity of training theory, these differences
Isoldnetic Performance in LBP Patients: The Predictive Power of SES
97
were not anticipated. Based on SES' predictive power at slower speeds, one might hypothesize that isokinetic testing at slower speeds is more relevant to everyday functioning in activities such as walking, running, carrying shopping bags, standing in queue, sitting in an armchair, sitting at a desk, or bending over the sink. Conversely, this should be remembered when using SES as a diagnostic or evaluational tool. If isokinetic performance at slower speeds is representative of actual work capability, then the major implication of this study is that efficacy expectancies of motorial activities occurring in nonwork environments are an important predictor of work capability. The implication for evaluation is: what happens outside the workplace is very important. Perhaps by focusing treatment programs on natural behaviors like walking, running, or biking, we may be able to improve the most important type of efficacy expectancies; in turn, this improvement may lead to increased work capability. Problems and Limitations
Numerous factors contribute to the case for holding reservations as to the generalizability of the results obtained presently. To begin with, the study population consisted of a diverse group of Finns living in the region surrounding Oulu, i.e., between the 64th parallel and the arctic circle. This population may well sport idiosyncrasies relative to the southern, urban Finnish population, to which cultural and institutional factors peculiar to Finland must be added when international comparability is sought. Additionally, it must be conceded that the subjects were studied at a very special historical moment characterized by a particular social atmosphere, which is not expected to repeat itself. In this context, the issue of the distinction between overt and covert behaviors raised by Dolce becomes especially relevant (6). Likewise the outcome measure deserves qualification. In their review of the iso-measurement literature, Newton and Waddell observed that "there is strong evidence of a learning effect between the first and second test sessions and of the need to use a second iso-test session on a separate day as the baseline to allow for initial learning effects" (4). In spite of this, only a single session was used for isokinetic testing in this study. Furthermore, of all the possible output variables that isokinetic testing commonly produces, total work done was selected and utilized in this study. In general there is very little information on variables computed from the values of torque and angle, work included. Newton and Waddell added, however, that the single study which addressed this matter found the isokinetic measures of torque, work, and power in trunk flexion-extension to be highly correlated (4).
CONCLUSION Owing to the nature of the study population, conclusions entertained here must necessarily be only tentative and limited to chronic low back pain patients. Within these limits, the predictive value of SES was confirmed with respect to the
98
Kaivanto, Estlander, Moneta, and Vanharanta
other measures included in this study. SES displayed the strongest association with isokinetic total work done. At the higher speeds however--particularly in extension-the relative predictive power of OSWESTRY gained over SES. Ensuing from this observation is the suggestion that there may be qualitative differences across test speeds and flexion-extension calling for special attention and accordingly requiring special adjustment when interpreting the results of isokinetic testing.
ACKNOWLEDGMENTS This study was supported by a grant from the Gyllenberg Foundation. The authors would also like to thank the Editor-in-Chief as well as three anonymous referees for their assistance in bringing about the final version of this paper.
REFERENCES 1. Bigos SJ, Batti6 M, Spengler DM, et al. A prospective study of work perceptions and psychosocial factors affecting the report of back injury. Spine 1991; 16: 1-6. 2. Frymoyer JW, Cats-Baril W. Predictors of low back pain disability. Clin Orthop Re/at Res 1987; 221: 89-98. 3. Riihim~dd H. Low-back pain, its origin and risk indicators. Scand J Work Env Health 1991; 17: 81-90. 4. Newton M, Waddell G. Trunk strength testing with iso-maehines: Part 1: Review of a decade of scientific evidence. Spine 1993; 18: 801-811. 5. Bandura A. Self-efficacy: Toward a unifying theory of behavioral change. Psychol Rev 1977; 84: 191-215. 6. Dolce JJ. Self-efficacy and disability beliefs in behavioral treatment of pain. Behav Res Ther 1987; 25: 289-299. 7. Council JR, Ahem DK, Follick MJ, Kline CL Expectancies and functional impairment in chronic low back pain. Pain 1988; 33: 323-331. 8. Kitsch I. Response expectancy as a determinant of experience and behavior. Am Psychol 1985; 40: 1189-1202. 9. Palenzuela DL. The expectancy construct within the social learning theories of Rotter and Bandura: A reply to Kirseh's approach. J Soc Behav Person 1987; 2: 437-452. 10. Estlander A-M, Vanharanta H, Moneta GB, Kaivanto K. Anthropometric variables, self-efficacy beliefs, and pain and disability ratings on the isokinetic performance of low back pain patients. Spine 1994; 19: 941-947. 11. Bongers PM, de Winter CR, Kompier MAJ, Hildebrandt VH. Psychosocial factors at work and muscoloskeletal disease. Scand J Work Env t-l-ealth 1993; 19: 297-312. 12. Beck AT, Ward CH, Mendelson M, Mock J, Erbangh J. An inventory for measuring depression. Arch Gen Psychiat 1961; 4: 561-571. 13. Husman K. Symptoms of car-painters with long-term exposure to a mixture of organic solvents. Scand J Work Env Health 1980; 6: 19-32. 14. H~irminen H. Neurotoksisten Haittojen Seulonta: Oirekyselyt ja Psykologiset Testit (The screening of neuro-to~dc effects: Symptom questionnaires and psychological tests). Helsinki: TyOterveyslaitos, 1989. 15. Nicholas MK. Self-Efficacy and Chronic Pain. Paper presented at the Annual Conference of the British Psychological Society, St. Andrews, March 1989. 16. Fairbanks JCT, Couper J, Davies JB, O'Brien JP. The Oswestry Low Back Pain Disability Questionnaire. Physiotherapy 1980; 66: 271-273. 17. Nicholas MK, Wilson PH, Goyen J. Comparison of cognitive-behavioral group treatment and an alternative non-psycholo~!cal treatment for chronic low back pain. Pain 1992; 48: 339-347. 18. Norusis MJ. SPSS/PC+ ~ V2.0 base manual for the IBM PC/XT/AT and PS/2 Chicago: SPSS, 1988. 19. Weisberg S. Applied Linear Regression (2nd Ed.). New York: John Wiley & Sons, 1985.
lsokinetic Performance in LBP Patients: The Predictive Power of SES
99
20. Rissanen A, Kalimo H, Alaranta H. Muscle histomorphology and isokinetic strength: A prospective study of intensive training in chronic low back pain patients. In: Book of abstracts of the international society for the study of the lumbar spine; 1993 June 15-19. Marseilles: International Society for the Study of the Lumbar Spine, 1993.
