Behavioral Sciences, University of Texas-Houston Medical School and University. Center for Pain Medicine and Rehabilitation at Hermann Hospital (Dr. Novy); the. Deputment of ... isokinetic devices to quantify trunk strength and motion) call.
195
Physical Performance: Differences in Men and Women With and Without Low Back Pain Diane M. Novy, PhD, Maureen J. Simmonds, PhD, PT, Sharon L. Olson, PhD, Plj C. Ellen Lee, MS, Stanley C. Jones, MD ABSTRACT. Novy DM, Simmonds MJ, Olson SL, Lee CE, Jones SC. Physical performance: differences in men and women with and without low back pain. Arch Phys Med Rehabil 1998;80: 195-8. Objective: To determine the extent to which there may be major differences in scoreson a battery of physical performance tasks among men with nonspecific, mechanical low back pain (LBP), women with LBP, healthy men, and healthy women. Design: Case series survey. Setting: A referral-based orthopedic clinic. Patients: Thirty-three men and 46 women with LBP Control Subjects: Twenty-one men and 25 women healthy controls. Intervention: Completion of six clinician-assessedphysical performance tasks and self-report inventories. Main Outcome Measure: Performance scores on distance walked in 5 minutes, 50-foot walk at fastest speed, repeated sit-to-stand, repeated trunk flexion, loaded forward reach, and the Sorensen fatigue tasks. Results: Discriminant function analysis revealed that the four groups of subjects performed the physical tasks significantly different in two major ways: (1) healthy control subjects outperformed LBP patients, irrespective of gender, on tasks involving trunk control, coordination, and stability while withstanding heavy or quickly changing loads on the spine; (2) men outperformed women, irrespective of patient or nonpatient status, on tasks involving anthropometric features of limb length. The findings provide guidance on reasonable performance expectations for men and women patients with LBP. Future studies of treatment effectiveness also will be able to assessphysical performance change in terms of the intersection between standards set by the men and women healthy control subjects and those of men and women patients. However, whether a return to nonpatient status is an appropriate treatment goal is left to future research. 0 1999 by the American Congress of Rehabilitation Meditine and the American Academy of Physical Medicine and Rehabilitation
W
ITHIN THE CONTEXT of clinical rehabilitation assessment, there has been recent interest in including simple, standardized physical performance tasks along with traditional
From the Department of Anesthesiology and Department of Psychiatry and Behavioral Sciences, University of Texas-Houston Medical School and University Center for Pain Medicine and Rehabilitation at Hermann Hospital (Dr. Novy); the Deputment of Physical Therapy, Texas Woman’s University (Dr. Simmonds, Dr. Olson, Ms. Lee); and Spine Care Southwest (Dr. Jones), Houston, TX. Submitted for publication June 1, 1998. Accepted in revised form September 8, 1998. Supported in part by Texas Woman’s University Research Enhancement Award grant 10-0131382 and NIH EARDAResearch Pilot Project grant 1997-8. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. &print requests to Diane M. Navy, PbD, Department ofAnesthesiology, University of Texas-Houston Medical School, 643 1 Fannin, MSB 5.020, Houston, TX 77030. 0003-9993/99/8002-5052$3.00/O
measures of physical impairment (eg, range of motion and muscle strength) and patients’ perceptions of their ability to perform certain activities and tasks (eg, Roland and Morris Disability Questionnaire).’ In part, this interest stems from a current thrust for cost-effective and time-efficient clinical assessment. However, before physical performances tasks are used to compare patients against a standard set by healthy control subjects, or are included in pretreatment and posttreatment assessments, the potential for apparent differences (ie, slighter anthropometric features, including a lower percentage of muscle mass, in women as compared to men) to influence performance on physical tasks needs to be investigated systematically.2 Although a variety of physical performance tasks have long been used to complement standard clinical assessmentsin many rehabilitation facilities, several problems with those tasks have been identified. Some tasks (eg, those using specially designed equipment) tend to be center-specific and cannot be replicated by independent teams of researchers.Other noted problems are that some assessments tend to be equipment- and timeintensive, as well as expensive.3 Further, unassessedor inadequate reliability or validity of many tasks (eg, those involving isokinetic devices to quantify trunk strength and motion) call into question the data from measures of those tasks.4,5 Only recently have the psychometric properties and clinical utility of a battery of simple, objectively measurable physical performance tasks been investigated thoroughly.‘*‘j Specifically, in a two-group sample comprising patients with nonspecific, mechanical low back pain (LBP) (n = 44) and individuals who served as healthy controls (n = 48), good interrater and testretest reliability was found for assessmentof distance walked in 5 minutes, 50-foot walk at fastest speed, 5 repetitions of a sit-to-stand task, 10 repetitions of a repeated trunk Aexion task, loaded forward reach task, and the Sorensen fatigue test. Although the healthy control subjects outperformed the patients on these tasks, patterns of major differences (ie, uncorrelated linear combinations of the original variables) between the groups were not investigated.’ Hence, it is difficult to formulate integrated conclusions about the differences. Other researchershave found performance differences across patient groups and healthy control subjects. Specifically, healthy control subjects outperformed LBP patients on tasks involving trunk control under a condition of speed.7s8Time to fatigue on those tasks also was reported to be shorter in healthy control subjects than in patients9 Previous investigations have done little to clarify the underlying pattern of major differences between patients and healthy control subjects. In studies of physical performance tasks with chronic LBP patients that have focused on gender differences, findings across independent teams of researchers have consistently supported the expectation that men outperform women on tasks involving lifting. l”,ll Regretfully, gender differences on other types of tasks have been sparsely investigated by independent teams of researchers.Specifically, Gronblad’s teamlo found that men outperformed women on repetitive squatting, and Lackner and Carosellarl found that men outperformed women on pulling Arch
Phys
Med
Rehabil
Vol 80, February
1999
196
DIFFERENCES
IN PHYSICAL
and carrying tasks but not on a static pushing task. Although anthropometric features appear to be major differentiating factors in gender differences on performance tasks, previous investigations have done little to clarify this. Taken together, the evidence, albeit relatively sparse, supports the inference that healthy control subjects should outperform patients with LBP on tasks involving trunk strength. Another inference is that men should outperform women on physical performance tasks that are influenced by enhanced anthropometric features. To enable further investigation of major gender and patient versus nonpatient status differences on a battery of physical performance tasks, data are presented and analyzed by group: men with nonspecific, mechanical LBP; women with LBP; healthy men; and healthy women. In addition to using data from men and women healthy control subjects as comparison groups, healthy control subject data are presented in an effort to provide early normative standards for judging future treatment effectiveness. Specifically, these data will make it possible to assessthe extent to which rehabilitation is able to improve the performance of patients with LBP. METHODS Subjects Physical performance data on 33 men and 46 women with LBP were collected. Seventy of these subjects were recruited from a large metropolitan orthopedic clinic in the southwestern United States, and the remaining nine subjects were faculty and graduate students who had LBP symptoms and had sought medical attention. Sixteen men and 28 women with LBP comprised the interim sample used in Simmonds’ collaborative psychometric study.’ The additional subjects in the current study represent the end-point in data collection for the project. The mean age of the women with LBP was 42.87 years (SD = 11.24yrs); their mean weight was 72.25kg (SD = 20.60kg) and their mean height was 164.98cm (SD = 17.65cm). The women with LBP rated their average pain intensity as a 3.67 (SD = 2.86) on a IOcm visual analogue line, and their average Roland and Morris Disability score was 8.95 (SD = 5.42).12 The average duration of the women’s LBP was 147.62 months (SD = 116.42), and approximately 50% of these individuals had at least one surgical procedure for their LBP. The mean age of the men with LBP was 42.45 years (SD = 11.46yrs); their mean weight was 89.27kg (SD = 13.27kg) and their mean height was 182.21cm (SD = 6.67cm). The men with LBP rated their average pain intensity as a 3.34 (SD = 2.34) on a 1Ocmvisual analogue line, and their average Roland and Morris Disability score was 10.47 (SD = 7.62). The average duration of the men’s LBP was 65.75 months (SD = 97.38), and approximately 50% of these individuals had at least one surgical procedure for their LBP The physical performance data on 21 men and 25 women healthy control subjects from the collaborative psychometric study by Simmonds and colleagues were used in the current study.* However, because a casewise deletion for missing data was used for the analyses that follow, two of the 28 healthy control subjects with missing data in Simmonds’ study were deleted. The mean age of the healthy control women was 34.24 years (SD = 10.08yrs); their mean weight was 64.82kg (SD = 12.37kg) and their mean height was 167.82cm (SD = 7.08cm). The mean age of the healthy control men was 36.57 years (SD = 8.41yrs); their mean weight was SO.OSkg (SD = 12.53kg) and their mean height was 176.46cm (SD = 7.37cm). Arch
Phys
Med
Rehabil
Vol 80, February
1999
PERFORMANCE,
Novy
Measures The measures of primary focus were the physical performance tasks that yielded the best psychometric properties from Simmonds’ study.’ Those measures and the procedures associated with them are described as follows. Repeated trunk Jlexion. Three warm-up trunk flexion movements preceded the actual test. A piece of tape was used to mark the target distance that the subjects were required to reach as they performed the timed repeated flexion test. For the actual test, subjects were required to flex to the limit of their range and return to the upright position as quickly as tolerated. For this test, the described activity was repeated 10 times and the total procedure was timed with a stopwatch. Repeated sit-to-stand. Subjects were required to rise to a standing position and return to a sitting position as quickly as possible five times. Loaded reach. Subjects were required to stand next to a wall on which a meter rule was mounted at shoulder height. Keeping their heels on the floor, subjects reached forward with hands at shoulder height while holding a 4.5-kg weight. The maximum reach distance was recorded in centimeters. F$y-foot walk at fastest speed. Subjects were required to walk 25 feet and turn around and walk back to the starting line as fast as possible. Distance walked in 5 minutes. Subjects were required to walk as far as they could in 5 minutes. The distance was measured. Sorensen fatigue. Subjects were required to lay prone on a standard treatment table. They were positioned with their hips at the edge of the table. Straps were placed across the thighs and calves to provide stability. With arms flat against the body, subjects were required to lift their upper body so that it was in a horizontal plane with the table and to hold that position as long as possible. Time taken to fatigue was then recorded. Analysis A descriptive discriminant analysis was performed to reveal major differences on the six physical performance tasks among the four groups of subjects. In this analysis the total association of the physical performance tasks is actually broken down into additive pieces through the use of uncorrelated linear combinations of the original variables (ie, the discriminant functions). The additive breakdown is obtained because the discriminant functions are derived to be uncorrelated.13 RESULTS The univariate F ratios with 3 and 121 degrees of freedom, along with their associated statistical significance levels, for distance walked in 5 minutes, 50-foot walk, sit-to-stand, trunk flexion, loaded forward reach, and Sorensen fatigue were 10.07 (p < .OOl), 3.91 (p = .OlO), 7.37 (p < .OOl), 7.88 (p < .OOl), 5.66 (p = .OOl), and 16.11 (p < .OOl), respectively. Thus, the four groups of subjects did perform differently on the six physical performance tasks. Descriptive statistics on the groups’ performance on the individual tasks are listed in table 1. A stepwise discriminant analysis breakdown of the overall association of the six tasks showed that there were two significant discriminant functions (p < .OOl and p = .003, respectively) that separated the groups. Interpretation of the discriminant functions was based on the discriminant functionvariable (ie, data on the performance tasks) correlations (table 2) and the standardized coefficients (table 3). Specifically, the correlations were used for substantive interpretation (ie, to identify the underlying construct which the discriminant function represents). The first discriminant function is primarily
DIFFERENCES
Table
IN PHYSICAL
1: Description
Statistics
PERFORMANCE,
on the
Individual
197
Novy
Performance
Tasks
Group Performance Task Distance
Women With LBP (n=46)
walked
418.41 9.95
50-Foot walk Sit-to-stand Trunk flexion Loaded reach Sorensen fatigue Values
reported
as mean
Men With LBP (n=33)
(106.78) (2.17)
460.39 9.91
2: Discrimination
Performance
Tasks
Distance walked 50-Foot walk Sit-to-stand Trunk flexion Loaded reach Sorensen fatigue
501.19
(88.16) (3.94)
11.03 20.49
(4.42) (7.20)
12.75 21.11
(8.67) (9.26)
58.89 33.57
(10.59) (37.10)
64.74 31.07
(12.72) (27.07)
Men Controls (n = 21)
(45.74)
536.03
(84.49)
8.60 7.56
(0.93) (1.14)
8.13 (0.74) 7.14(1.72)
15.44 67.13
(1.80) (6.66)
14.21 67.55
(2.18) (6.48)
76.28
(33.29)
73.96
(32.53)
(SD).
defined by the Sorensen fatigue task (correlation of .87), with trunk flexion, sit-to-stand, distance walked, and 50-foot walk secondarily involved (correlations of -.61, -.58, .58, and -.42, respectively). The negative values for trunk flexion, sit-to-stand, and 50-foot walk mean that the groups who scored higher on the Sorensen fatigue and distance walked (ie, they were better able to withstand and control heavy trunk loads, both static and dynamic) performed the trunk flexion, sit-tostand, and 50-foot walk tasks in less time (ie, their trunks accommodated complex challenges more quickly). The second function is primarily defined by the loaded reach and distance walked (correlations of .57 and .56, respectively), with other tasks contributing trivial amounts of information. The standardized coefficients were used to determine which of the performance tasks were redundant given others in the set. In the first function, the Sorensen fatigue and sit-to-stand tasks were not redundant (coefficients of .73 and -.55, respectively), whereas the other tasks were redundant. In the second function, sit-to-stand, loaded reach, and distance walked were not redundant (coefficients of 1.12, .80, and .68, respectively), whereas the other tasks were redundant. Given that the sit-tostand task had only a trivial correlation with the second function (correlation of. 12), it appears that this task is quite distinct from the tasks that imparted substantive meaning to the second function (ie, loaded reach and distance walked) and, hence, is not redundant to them. Combining the information from the discriminant functionvariable correlations and the standardized coefficients, the first function appears to characterize trunk control. Specifically, this function comprises tasks that require the trunk to withstand and coordinate heavy or quickly changing loads on the spine. The second function appears to characterize anthropometric features of limb length. From the group centroid means in table 4, it can be noted that the groups differ according to their patient status/nonstatus (ie, LBP versus healthy control subjects) on the first discriminant function (ie, trunk control) and by gender on the second discriminant function (ie, anthropometric features of limb length). Specifically, the healthy men and women control subjects outperformed the men and women LBP patients on tasks involving trunk control. Men with and without LBP outperformed women with and without LBP on tasks involving anthropometric features of limb length. Given the adequate Table
Women Controls (I?= 25)
Function-Variable Function .58 -.42 -.58 -.61 .35 .87
1
subject/variable ratio (ie, 20 to l), there can be a reasonable degree of confidence in the reliability of these resultsi DISCUSSION With the psychometric support for a battery of simple physical performance tasks established, attention was given to investigating major performance differences in men and women patients and healthy control subjects. Although previous research has noted significant differences in patients and healthy control subjects irrespective of gender and in men and women irrespective of patient/nonpatient status, the major patterns of differences among these groups have received sparse attention. Specifically, trunk control appeared to be a major differentiating factor in the performance of patient and health control groups, with healthy control subjects outperforming patients. In regard to gender differences, anthropometric features appeared to be a major differentiating factor in the performance of men and women, with men outperforming women. To empirically verify the potential differentiating factors, four groups of subjects (ie, women with LBP; men with LBP; healthy women; and healthy men) were assessedwith a battery of physical performance tasks. On tasks in which the trunk muscles were challenged, healthy subjects outperformed the patients, regardless of gender. On tasks involving anthropometric features of limb length, men outperformed women, regardless of their patient or nonpatient status. This pattern of different physical performance abilities has important implications. Empirical evidence accrued here supports the expectation of nonequivalent trunk control capabilities across patients with LBP and healthy control subjects. Specifically, it seems reasonable to expect men and women LBP patients to have more compromised trunk function than men or women without LBP Whether or not patients’ lower trunk control results from fear, functional impairment, or lack of exercise and activity secondary to disability and pain was not examined here. Indeed, given that a clear pattern of difference was found between patients with LBP and healthy subjects, such a topic is one for future research. Whether performance on tasks involving stress on the trunk can be improved to levels associated with healthy subjects also is a topic for future investigation. Various treatments (eg, quota-based physical therapy or regular walking programs) may be contenders to
Correlations Function
Table 2
Performance
3: Standardized
Tasks
Coefficients
Function
1
Function
.56 -.I3 .I2
Distance walked BO-Foot walk Sit-to-stand
.31 .46 -.55
.68 -.22 1.12
-.08 .57 .I2
Trunkflexion Loaded reach Sorensen fatigue
-.25 -.26 .73
-.Ol .80 1.14
Arch
Phys
Med
Rehabil
Vol 80, February
2
1999
198
DIFFERENCES
Table Group
4: Group
Centroid Function
LBP Women Men
Means 1
Function
-.43 -.63
Healthy Control Women Men
IN PHYSICAL
2
-.53 .64
.67
-.05
.91
.21
achieve improved performance on tasks involving the trunk muscles. With regard to the implications of gender differences on performance tasks, it will be important to consider the potential impact of limb length. Indeed, the average performances by the men and women healthy subjects in this sample may be considered early normative data on the individual physical performance tasks. Directions for future research include replication of analyses on performance differences in men and women with larger samples. Also, individual differences due to factors other than patient statuslnonstatus and gender, such as age, activity levels, and affective status, are worthy of future investigation. Further, it will be important to assesswhether the gender differences found here are robust among men whose stature resembles that of women in our sample (ie, shorter and lighter in weight). Although the subject/variable ratio in this study supports the expectation that our results should be robust, other features in the design of this study were limitations. For one, the groups were not matched on potentially relevant variables. Specifically, the patients were approximately 7 years older and approximately 15kg heavier than the healthy subjects.As would be expected, the men were approximately 30kg heavier and approximately 25cm taller than the women. There also were differences in variability on the individual performance tasks across the groups. It is possible that these differences in variability affected the results of the study. Hence, it is recommended that the methodology from this study be replicated by an independent team of researchers before a degree of certainty can be associated with our results.
Arch
Phys
Med
Rehabil
Vol 80, February
1999
PERFORMANCE,
Novy
References 1. Simmonds MJ, Olson SL, Jones SC, Hussein T, Lee CE, Radwan H, et al. Physical performance tests: are they psychometrically sound and clinically useful for patients with low back pain? Spine. Forthcoming. 2. Astrand PO, Rodahl H. Textbook of work physiology: physiological bases of exercise. New York: McGraw-Hill; 1986. 3. Mayer TG, Gatchel RJ. Functional restoration for spinal disorders: the sports medicine approach. Philadelphia (PA): Lea and Febiger; 1988. 4. Newton M, Thow M, Somerville D, Henderson I, Waddell G. Trunk strength testing with Iso-machines. Part 2: Experimental evaluation of the Cybex II back testing system in normal subjects and patients with chronic low back pain. Spine 1993;18:812-24. 5. Pope MH. A critical evaluation of functional muscle testing. In: Weinstein JN, editor. Clinical efficacy and outcome in the diagnosis and treatment of low back pain. New York: Raven Press; 1992. p. 101-13. 6. Simmonds MJ, Claveau Y. Measures of pain and physical function in patients with low back pain. Physiother Theory Pratt 1997;13: 53-65. 7. Marras WS, Parnianpour M, Ferguson SA, Kim JY, Crowell RR, Bose S, et al. The classification of anatomic- and symptom-based low back disorders using motion measure models. Spine 1995;20: 2531-46. 8. Marras WS, Wongsamm PE. Flexibility and velocity of normal and impaired lumbar spine. Arch Phys Meh Rehabil 1986;67:213-7. 9. Ito T. Shirado 0. Suzuki H. Takahashi M. Kaneda K. Strax TE. Lumbar trunk muscle endurance testing: An inexpensive alternative to a machine for evaluation. Arch Phys Med Rehabil 1996;17:75-9. 10. Gronblad M, Hurri H, Kouri J-P. Relationships between spinal mobility, physical performance tests, pain intensity and disability assessments in chronic low back pain patients. Stand J Rehabil Med 1997;29:17-24. 11. Lackner JM, Carosella AM. Pain expectancies, pain, and functional self-efficacy expectancies as determinants of disability in patients with chronic low back disorders. J Consult Clinical Psycho1 1996;64:212-20. 12. Roland M, Morris R. A study of the natural history of back pain. Part 1: development of a reliable and sensitive measure of disability in low-back pain. Spine 1983;8:141-4. 13. Huberty CJ. Applied discriminant analysis. New York Wiley; 1994.
14. Barcikowski R, Stevens JP.A Monte Carlo study of the stability of canonical correlations, canonical weights and canonical variatevariable correlation. Multivariate Behav Res 1975;10:353-64.
