Physical Function in Patients with Cancer: Psychometric Characteristics and Clinical Usefulness of a Physical Performance Test Battery

Transcription

1 404 Journal of Pain and Symptom Management Vol. 24 No. 4 October 2002 Original Article Physical Function in Patients with Cancer: Psychometric Characteristics and Clinical Usefulness of a Physical Performance Test Battery Maureen J. Simmonds, PT, PhD School of Physical Therapy, Texas Woman s University, Houston, Texas, USA Abstract This study investigated the psychometric properties of a battery of physical performance tests, characterized physical function in patients with cancer referred for rehabilitation, and provided normative standards against which to compare disease progression and/or future treatment effectiveness. A total of 109 patients with cancer (55 women and 54 men) and 105 control subjects (66 women and 39 men) participated. Subjects completed self-report questionnaires regarding pain, physical function, and fatigue, and also performed nine physical performance tests: the time taken to complete various tasks (picking up coins, tying a belt, reaching up, putting on a sock, standing from sitting, a 50-foot fast walk, a 50-foot walk at preferred speed), the distance walked in 6 minutes, and the distance reached forward while standing were measured. Inter-tester and test retest reliability was good to excellent for all tests (ICC to 0.99). Known group analyses controlling for age were significant (P 0.001) for all physical performance tests. Control subjects significantly and systematically outperformed those with cancer by a factor of two or three. Examination of the correlation matrices showed relatively low correlations between performance and external measures (r 0.01 to 0.45). In contrast, correlations among performance measures were generally in the range of r 0.25 to Correlations between self-report of function and performance of functional tests were moderate, suggesting that the two methods of measuring function are complementary and both should be used for assessment and as outcome measures. J Pain Symptom Manage 2002;24: U.S. Cancer Pain Relief Committee, Key Words Physical function, reliability, validity, cancer, rehabilitation Address reprint requests to: Maureen J. Simmonds, PT, PhD, School of Physical Therapy, Texas Woman s University, 1130 John Freeman Blvd., Houston, Texas 77030, USA. Accepted for publication: December 22, Introduction Diagnostic and treatment advances for many different types of cancers have led to increased survival. More than 50% of individuals newly diagnosed with cancer will survive for 5 years and many will have a normal life expectancy. Unfortunately, residual symptoms and impairments from the disease or its treatment often U.S. Cancer Pain Relief Committee, /02/$ see front matter Published by Elsevier, New York, New York PII S (02)00502-X

2 Vol. 24 No. 4 October 2002 Physical Performance in Patients with Cancer 405 persist. These health problems compromise physical function, thereby adding to the distress level and disease burden for individuals and their families. Unfortunately, quantitative assessment of physical function is not well documented in patients with cancer, in part because of limitations in assessment methods. Physical function status is often inferred from standard clinical tests of impairment (e.g., pain, joint range of motion, and muscle strength). However, because function is influenced by psychological (cognitive and emotional), social, and environmental factors, as well as physical factors, the relationship between impairment measures and functional status may be quite trivial. A more direct method of measuring function is through the use of self-report questionnaires or actual task performance. A variety of questionnaires are in clinical use, ranging from simple undimensional or global questionnaires that take seconds to complete 1,2 to complex multidimensional questionnaires that can sample a wide range of activities and thoughts but also may be very time-consuming and challenging for patients to complete. 3 The latter is an important consideration given the high prevalence of cancer-related fatigue. Simple questionnaires, such as the Karnofsky performance status 2,4 and the Eastern Cooperative Oncology Group Scale, 1 are in common clinical use. They provide a simple global estimate of function and medical care requirements, and may even predict early death. 4 Other questionnaires, such as the Brief Fatigue Inventory (BFI) 5 and Functional Assessment of Cancer Therapy (FACT) scale 6 address functional abilities more specifically, but do so in the context of other symptoms or constructs (i.e., fatigue and quality of life, respectively). Although standardized questionnaires of physical function have clinical utility, they may not be a valid reflection of a patient s actual functional status, 7 especially when an external reference is unavailable. Self-report measures are more closely associated with other self-report measures than with physical capabilities observed or measured by clinicians. Even with the use of standardized and validated instruments, patient self-reports and clinician-measured tests of the same functions are moderately correlated at best. 8 This can be partly explained by the fact that self-reports and performance tasks are differentially influenced by a variety of factors. 9 Current mood, attitudes, anxieties, expectations, memory, reporting biases, and situational demands can differentially influence self-reports or proxy reports of function and measured function. Moreover, it also is apparent that patients as well as clinicians or healthy subjects often misjudge estimates of physical ability based on inaccurate estimates of distance walked or time taken to complete an activity. Inaccuracy of reporting is not purposeful nor is it confined to patients. Most individuals, including physicians, have difficulty estimating the time it takes to complete a task, their tolerance for different activities, or the distance they can walk. 10 Finally, scores on questionnaires are influenced by the number of items and the manner in which the question is phrased (e.g., Do you do... vs. Can you do... ). But regardless of the accuracy and response issues regarding functional status questionnaires, they do not provide enough information on physical performance to direct or evaluate rehabilitation outcome. In recent years, standardized performancebased tests have been developed and used in patients with chronic pain and with spinal problems, as well as in the elderly. 7,12 14 These test batteries sample and quantify patients abilities to perform simple everyday tasks. The reliabilities (intra- and inter-rater as well as day-to-day stability) are generally established and are good to excellent (Intraclass correlations [ICCs] 0.7). Construct validity (convergent and discriminant) also is fairly well established in specific populations. 7,12 However, the clinical utility and psychometric characteristics of performance tests have not been tested in patients with cancer. The purpose of this investigation was to test the psychometric properties and clinical utility of a battery of physical performance tests and to characterize physical performance in patients with cancer referred for rehabilitation due to physical dysfunction. In addition to using data from control subjects as a comparison group, the control group data, stratified by age, provide normative standards for judging disease progression, and/or future treatment effectiveness. Methods Overview Two hundred-fourteen subjects (109 patients with cancer [55 women and 54 men] and

3 406 Simmonds Vol. 24 No. 4 October control subjects [66 women and 39 men]) participated in this series of studies. Intertester reliability was established first, using 20 consecutive patients and five testers. Test retest reliability (internal consistency) was evaluated using data from 50 consecutive patients. Construct validity (discriminant and convergent) was assessed using data from the total patient group (n 109) and control group (n 105), controlling for age. Finally, data from the control group, stratified by age, were used to establish performance norms. Subjects Informed consent was required and obtained prior to participation. All patients had a diagnosis of cancer and were recruited from the Rehabilitation Service at the University of Texas M.D. Anderson Cancer Center. Individuals who had symptomatic cardiovascular disease or who were unable to walk (walking aids were permitted) were excluded. Control subjects were a convenience sample that primarily comprised health care workers and faculty and graduate students from the M.D. Anderson Cancer Center and Texas Woman s University, respectively. Control subjects also were excluded if they had symptomatic cardiovascular disease; if they had musculoskeletal problems that would influence their physical performance; or if they were unable to walk for other reasons (e.g., neurological problems). There were no significant differences between the patient and control groups based on sex ( , P 0.056). However, the patient group was significantly older (t 2.2, P 0.05). The mean ages of the patient and control groups were and years, respectively. Therefore, in subsequent analyses, age was statistically controlled. The entire study group also was stratified on the basis of age (i.e., younger, 30; middle, 30 59; older, 59) to provide more meaningful comparisons between groups, as well as normative data. There were no significant differences in age between the control and patient groups in these subgroups. Performance Measures The performance battery comprised a series of nine tasks that challenged upper limb movements and total body movements in both seated and standing positions. Most of the tasks were timed because decline in task performance, whether due to age, disease, or treatment, is generally evidenced by a decline in reaction time or in velocity of movement, 11 and it is an easy and meaningful measure for both patient and practitioner. The tasks were as follows: foot walk at preferred speed. For this test, subjects were timed as they walked 25 feet, turned around, and walked back to the starting position at their preferred walking speed foot walk at fastest speed. Subjects were again timed as they walked 25 feet, turned around, and walked back to the start as fast as they could. 3. Forward reach. For this test, subjects stood adjacent to a wall on which a yardstick/tape measure was positioned horizontally at shoulder height. Subjects reached forward as far as they could and the distance reached was measured in cm Timed, repeated sit-to-stand. Subjects sat in a standard chair and were then timed as they stood up and then sat back down, twice. The test was repeated after a brief rest and the average time of the two trials was used. 5. Sock test. Subjects sat in a standard chair. They were timed as they put on one loose-fitting sock. 6. Timed, repeated reach-up. For this test, subjects stood facing a wall and reached up as high as they could with both hands. A mark was placed on the wall at the reach distance. Subjects then reached up and returned their hands to their sides three times, as fast as they could. 7. Timed belt tie. For this test, subjects were seated. Using an elastic wrap bandage approximately 4 feet long, they were timed as they wrapped the bandage around their waists and tied it in front of them. 8. Distance walked in 6 minutes. Subjects walked as far and as fast as they could for 6 minutes. The distance walked was measured at 6 minutes. (Subjects were allowed to rest if and as necessary during the 6-minute period.)

4 Vol. 24 No. 4 October 2002 Physical Performance in Patients with Cancer Coin test. For this test, subjects sat at a table. They were timed as they picked up four coins (a quarter, a dime, a nickel, and a penny) and placed them in a cup. They were required to pick up each coin individually. Self-Report Measures External measures were included to better characterize the patients and further assess the validity of the tests. The patient group completed a series of self-report measures that assessed pain, function, and fatigue. The measures included: 1. Pain. Two numerical rating scales (0 10) were used to measure pain. One scale measured pain intensity and one measured pain unpleasantness Function. The Functional Status Index (FSI) is a generic self-report measure. Subjects rated their performance as well as the degree of pain and difficulty in accomplishing a series of tasks Fatigue. The Brief Fatigue Inventory (BFI) was used to characterize the presence and impact of fatigue on function. Its reliability and validity have been established in patients with cancer. 5 Procedure Patients referred to the Rehabilitation Service who met the criteria were advised about the purpose of the study and invited to participate. Those who agreed to participate then signed an informed consent form and completed the self-report questionnaires. A total of five testers (physical therapists or physical therapy students) participated in data collection. Two testers simultaneously but independently measured test performance of the subjects for the inter-rater reliability phase; otherwise, one tester usually measured test performance. With the exception of the 50-foot walk at the patient s preferred speed, subjects were asked to perform the tasks as quickly as they could, or to reach or walk as far as they could. To reduce the impact of fatigue or any other symptoms such as nausea, subjects were advised that they could rest between each task. They also were advised that they could decline to perform or complete any of the tasks without prejudice. The testing order of the physical performance tasks was counterbalanced and each test was repeated twice. The exception was the 6-minute walk; this always was completed last and only was completed once. Data Analysis Descriptive statistics (mean and standard deviation) of performance on the tests were calculated for the total group (patient and control) and for specific subgroups (i.e., stratified by age and cancer type). Standard scores also were computed to facilitate performance comparisons between groups and across tasks. Finally, performance norms for each of the age groups were calculated in the control group to allow for future comparison. Reliability was analyzed in the patient group only. Inter-tester and test retest reliability was determined using the ICC 11 model 17 and the first 20 and 50 patients, respectively. In the reliability phase the results of the two trials for each test were comparable; therefore, an arbitrary decision was made to use the first trial in all subsequent analyses. Finally, because of the variability in performance between the first and second trial of the 50-foot walk at preferred speed, it was dropped from the test battery and further analysis. Validity was tested in several ways. Discriminant validity was first evaluated by examining the correlations between the performance tests and external measures. It was further evaluated by comparison of performance between patients and control subjects using the known group method and a multivariate analysis of variance (MANOVA) controlling for age. Convergent and discriminant validities were next assessed through examination of the correlation matrix among the performance measures, controlling for age. Stronger correlations were anticipated between similar tests. Finally, factor analysis was used to identify the underlying constructs of the tests. Principle components was used as the common factor extraction method and quartimax was used as the rotation method to derive correlated constructs. Alpha was set at P Results Characteristics of the study subjects and descriptive statistics of patients self-reports of pain, fatigue, and function are presented in

5 408 Simmonds Vol. 24 No. 4 October 2002 Table 1. Mean levels of pain and fatigue were relatively low, but variability was high. Patient physical performance measures are presented in Table 2. Data are presented for the group as a whole and for age-stratified subgroups in Table 3. Clearly, subjects in the control group systematically outperformed those with cancer, often by a factor of two or three. Figures 1 and 2 present the 95% confidence intervals of standardized performance scores for the speed and distance tests, respectively, and show clear separation in performance between patients and controls. Predictably, greater separation in performance is apparent in tasks that are more physically challenging (e.g., the sit-to-stand compared to the coin test). Performance norms grouped according to age are presented in Table 4. Reliability Inter-tester agreement was excellent for all tests (ICC and 0.99). Based on the high level of reliability and the equivalent results between testers, the measures from the first rater were used for all subsequent analyses. Test retest reliability was good to excellent for all tests (ICC to 0.99). Performance on the more complex exercises coin test and belt tie was the least consistent (ICC and 0.78, respectively). Validity Several different methods were used to assess construct validity. Discriminant and convergent validity were assessed through examination of the correlations among the performance tests and between performance tests and external measures (Table 2). Correlations between performance and external measures were trivial to moderate (r 0.01 to 0.51). In general, relatively low correlations were apparent between pain and performance measures (r 0.03 to 0.35) except when the performance tasks involved simultaneous upper limb and trunk movements. Correlations between self-report of function and performance were generally higher, especially when the FSI total score was used (r 0.25 to 0.51). Correlations between fatigue and performance (r 0.14 to 0.45) were generally moder- Table 1 Subject Characteristics Characteristics Patients (n 109) Controls (n 105) Sex Male Female Age Mean (SD) 50.1 yrs (19.1) 44.9 (14.7) Ethnicity Caucasian 76 Hispanic 17 African American 12 Cancer Frequency Hematologic Leukemia (Acute) 9 Leukemia (Chronic) 8 Lymphoma 12 Solid Tumors Breast 19 Bone 6 Neurological 15 Pulmonary 3 Other 33 Patient Status Inpatient 42 Outpatient 63 Mean (SD) Min Max Fatigue: Brief Fatigue Inventory Fatigue Now (0 10) Fatigue Worst in Last 24 Hours (0 10) Fatigue Total Functional Status Index Total Difficulty (0 90) Pain Intensity (0 10) Pain Affect (0 10)

6 Vol. 24 No. 4 October 2002 Physical Performance in Patients with Cancer 409 Test Table 2 Correlations Between Performance Measures and External Measures in Patient Group (n 109) Pain Fatigue Functional Status Intensity Affect Now Worst Total Total Personal Care Mobility Sock test 0.27 b 0.19 a a 0.21 a 0.43 d 0.21 a 0.27 b Belt tie 0.33 c 0.35 c 0.24 a a 0.51 d 0.20 a 0.33 c Coin test b c 0.33 c 0.15 Reach-up a 0.19 a a 0.51 d 0.32 c 0.19 a Forward reach a 0.23 a 0.33 c 0.32 c 0.25 b c Sit-to-stand a 0.30 c 0.44 d 0.44 d 0.39 d 50-foot fast walk a 0.26 b 0.47 d 0.36 d 0.31 c 6-minute walk c 0.50 d 0.45 d 0.49 d 0.53 d 0.44 d a P b P c P d P ate. Noteworthy was the strong correlation between fatigue and the 6-minute walk (r 0.45), a task that challenges endurance. The MANOVA controlling for age was significant (Wilks lambda 0.452, F 8, , P ), showing that physical performance differences between subjects with and without cancer were significantly different. Subsequent univariate analyses showed that, with the exception of the coin test which was significant at the P 0.05 level, the patient group was systematically and significantly outperformed by the control group on all performance tasks at the P level (Table 5). Construct validity was further examined using the multitrait-multimethod matrix. Pearsons product moment correlation coefficients, controlling for age (Table 6), were computed for each group (patient and control) and the correlation matrix was examined for convergent and discriminant validity. Strong correlations were apparent among the upper limb tests (reach-up, belt tie, sock test, and coin test) in both groups. Large body movement tests (e.g., sit-to-stand and walk tests) also were strongly correlated in both groups. The correlations between tests of upper limb function (e.g., reach-up, belt tie, and coin test) and those of lower limb function (e.g., walk tests) were generally lower. Also, the forward reach test a test of balance as well as of function and the sit-to-stand test had generally weaker relationships with all other performance tests in the control group but moderate to strong relationships with almost all other tests in the patient group. Table 3 Comparison of Performance Test Results Among Patient and Control Groups Total Group Younger Group ( 30 years) Middle Group (30 59 years) Older Group ( 59 years) Performance Test Group n Mean SD n Mean SD n Mean SD n Mean SD Reach-up (sec) Patient Control Coin test (sec) Patient Control Sock test (sec) Patient Control Belt tie (sec) Patient Control foot walk (sec) Patient Control Sit-to-stand (sec) Patient Control Forward reach (cm) Patient Control minute walk (m) Patient Control

7 410 Simmonds Vol. 24 No. 4 October 2002 indicators for the first factor were the timed sock test, belt tie, coin test, reach up, sit-to-stand, and 50-foot speed walk. These tasks are all associated with speed and coordination. Salient indicators for the second factor were the forward reach and 6-minute walk. These indicators are associated with endurance and balance. Fig % confidence intervals of standardized performance scores for the speed tests in patient and control groups. Higher scores ( 0) indicate lower performance. A scree plot was used to capture the correct number of factors. Eigen values of 4.3, 1.3, 0.7, and 0.5 indicated that most of the data (70%) could be explained by two factors. The first and second factors accounted for 54% and 16%, respectively, of the variance in the data set. Salient Clinical Utility The tests were generally acceptable to most patients. Ten and 12 patients, respectively, declined or were unable to complete the 6-minute distance walk and 50-foot speed walk. Seven declined or were unable to complete the sock test; six declined or were unable to complete the forward reach test. The time taken to finish the task battery varied greatly among subjects. Those individuals with greater functional deficits were generally slower to complete all tests and needed more rest between tests. In contrast, individuals with minimal functional deficits completed each task very quickly and needed little if any rest between tests. The range of time taken to finish the total test battery, including explanations and demonstrations of the tests, varied between 20 and 40 minutes. This time would be reduced with subsequent testing as the time taken to explain and/or demonstrate the tests would be greatly reduced. Fig % confidence intervals of standardized performance scores for the distance tests in patient and control groups. Higher scores ( 0) indicate higher performance. Discussion Physical dysfunction is a frequent consequence of cancer and its treatments. The problem contributes to psychosocial distress, compromises quality of life, and adds to the disease burden for patients and their families. Improvement of physical function therefore should be a primary goal of treatment; the use of appropriate assessment and outcome measures is fundamental to that process. Physical performance assessments have the potential to contribute to a more complete understanding and therefore better management of the physical difficulties encountered by patients with cancer. The results of this study confirm the fact that good to excellent reliability of simple physical performance measures is a robust finding. 7,12 14 The simplicity of the tests and their method of measurement (e.g., stopwatch) essentially account for this finding. The tests are simple

8 Vol. 24 No. 4 October 2002 Physical Performance in Patients with Cancer 411 Table 4 Percentile Norms for Physical Performance Tests Across Age Groups (n 105) Percentile 25 Percentile 50 Percentile years (n 20) Sock test Belt tie (sec) Coin test (sec) Sit-to-stand (sec) foot walk (sec) Reach-up (sec) Foward reach (sec) minute walk (m) years (n 70) Sock test (sec) Belt tie (sec) Coin test (sec) Sit-to-stand (sec) foot walk (sec) Reach-up (sec) Forward reach (sec) minute walk (m) years (n 15) Sock test (sec) Belt tie (sec) Coin test (sec) Sit-to-stand (sec) foot walk (sec) Reach-up (sec) Forward reach (sec) minute walk (m) for the patient to perform because they mimic familiar everyday tasks. Thus, these measures can be mastered without difficulty by both patients and practitioners and should be incorporated into standard clinical practice easily, without undue concern regarding their reliability. The stability of the test retest results shows that the tests need only be performed once in a clinical assessment. The stability suggests that test performance is associated with minimal learning, probably because of the simplicity and familiarity of the tests. The stability also suggests that performance of the tasks did not aggravate the patients symptoms, most likely because the patient determines the level of performance. The tests that were least stable (i.e., the sock test, belt tie, and coin test) still had acceptable reliability coefficients (0.89, 0.78, and 0.69, respectively). The lower level of stability is probably due to the fact that these tests involved a more complex series of movements that challenged both gross and fine motor abilities. Although the tests are familiar in everyday use, they are not usually performed as quickly as possible. Reliability of performance tests is moot if the tests are not valid. Face validity is not in question because the tests are direct measures of physical performance. In general, and compared to the control group, the patients were compromised to some extent across all physical performance tests. The patient group took approximately twice as long as the control group to complete most tasks and only were able to walk half the distance in 6 minutes. The sitto-stand task was even more of a challenge. For this task the patient group took about four times as long as the control group. This is an important point because it begins to quantify the time and effort burden with which individuals living with cancer have to contend. The tasks used in this battery were either basic tasks of self-care or of mobility. Individuals would normally complete these tasks many times a day; this would have an additive effect on the time taken for basic care activities, thereby leaving less time and perhaps energy available for nonessential or recreational activities. Slower movement is not necessarily associated with less effort and may in fact require more effort. For example, individuals with back pain walk a significantly shorter distance dur-

9 412 Simmonds Vol. 24 No. 4 October 2002 Task Table 5 Results of Univariate Analyses on the Effect of Patient Group and Control Group Df Hypothesis MS Error MS F Value P Value Sock test 1, Belt tie 1, Coin test 1, Reach-up 1, Foward reach 1, Sit-to-stand 1, foot fast walk 1, minute walk 1, ing 5 minutes than those in an age- and sexmatched control group, but the effort required for walking that distance is similar between groups. 12 This may be because slow movements have little forward momentum and thus require more muscle activity and effort to stabilize, control, and coordinate the body segments during an activity. 11 This is akin to the fact that it is far more difficult and requires more skill and effort to ride a bicycle slowly compared to riding it fast. The association between fatigue and movement efficiency warrants further investigation. In general, stronger correlations were evident among the performance measures in the patient group compared to the control group. This finding agrees with the results of our previous study that showed stronger correlations among performance measures in a patient group compared to healthy, pain-free individuals. 12 It may be explained in part by the higher level of variance in the patient group. However, it also may be due to the fact that cancer leads to a generalized decline in physical performance that is nonspecific. Although speculative at present, it suggests that central mechanisms are involved in functional deficits. This generic (nonspecific) aspect of physical performance is not apparent in the control group. For example, performance on the forward reach test (which is essentially a test of balance) was trivially related to performance on most other tests. Similarly, the relationships between the coin test (i.e., a test of fine motor activity) and other tests were selectively and appropriately strong or weak. The strength and pattern of correlations between performance tests and external measures support the validity of the test battery. In regard to pain and performance, it was interesting that relatively stronger correlations were apparent between pain and tasks that involved simultaneous upper limb, trunk, and lower limb movements (e.g., sock test). Thus, it is possible that pain was aggravated or evoked by the sizable ranges of movement specifically as- Table 6 Correlations Controlling for Age Among Performance Measures in Patient (n 109) and Control Groups (n 105) Test Sock Test Belt Tie Coin Test Reach Up Foward Reach Sit-to-Stand 50-Foot Walk 6-Minute Walk Sock test d 0.34 d 0.49 d 0.26 b 0.19 a 0.24 b 0.35 d Belt tie 0.32 c d 0.35 d b 0.23 a Coin test 0.22 a 0.85 d d Reach-up 0.26 b 0.56 d 0.56 d b 0.31 c 0.08 Foward reach 0.21 a 0.28 b 0.25 b 0.33 c Sit-to-stand 0.44 d 0.56 d 0.47 d 0.59 d 0.43 d d 0.53 d 50-foot fast walk d 0.38 d d 6-minute walk 0.41 d 0.39 d 0.23 a 0.45 d 0.56 d 0.62 d 0.39 d 1.00 Bold lower left-hand matrix cancer patient group; upper right-hand matrix control group. a P b P c P d P

10 Vol. 24 No. 4 October 2002 Physical Performance in Patients with Cancer 413 sociated with the types of movement that these tasks require. It is plausible that specific cancers (e.g., breast cancer) were associated with more pain and more difficulty with these tasks. However, it was not possible to answer that question with this subject sample. In a post hoc exploration the patient group was subdivided according to whether the patient s cancer was a solid tumor. However, no statistical differences were apparent on this basis. It is intuitive and further confirmation of the validity of the test battery that fatigue was most closely related to physically challenging and endurance tasks. Moderate relationships between self-report of function and performance tests were expected and are in agreement with previous reports. 8 Self-report measures of function and performance measures assess different components of the function construct. The methods are complementary, and both should be used in clinical practice and clinical research. Clinical acceptability of the measures is important not only for the patients but also for the clinician or investigator. It is important to note that the majority of patients completed most of the tests, indicating that they found the tests acceptable. It also was interesting to observe that some patients found it less burdensome to complete the physical performance tests than to complete the self-report questionnaires. Clearly, some patients found cognitive effort more troublesome than physical effort. The fact that the tasks were simple and familiar and required little cognitive effort probably contributed to the individual patient s willingness to participate. Clinical utility is apparent because the tests are simple, need little equipment, and can provide the clinician with information about movement difficulties that are encountered by the patient and may be amenable to treatment. The tests should be able to be used individually or as a battery, depending on the extent of the patient s problem and on the specific questions being asked. Based on these data and the magnitude of performance difference between patients and control subjects, the tasks are differentially challenging (e.g., sit-to-stand compared to the belt tie). It also is clear from these data that the only task associated with endurance is the 6-minute walk. Clearly, further study is necessary to test specific groups of patients with different types of cancers in order to determine empirically the most appropriate test battery. Further study also is necessary to determine the responsiveness and predictive validity of the tests in patients with a variety of different cancers. Conclusion The psychometric properties of nine physical performance tests (subsequently reduced to eight tests) were investigated in 214 subjects (109 patients with cancer and 105 healthy control subjects). Most tasks had excellent inter-tester and test retest reliability and discriminative validity. They were simple and easy for clinicians to use and acceptable to patients. The tests can be used individually or as a battery and are recommended as a complement to patient selfreport and clinician judgements of function. Acknowledgments Thanks to Diane Novy and Christina Meyers (who helped with protocol procedures); and Eileen Donovan, Pam Latham, Julie Felderman, and Tracey Powell (who helped with data collection). This work was supported, in part, by The Fleming Davenport Foundation, Texas Medical Center, Houston, Texas. References 1. Zubrod C, Schneiderman M, Frei E, et al. Appraisal of methods for the study of chemotherapy in man: comparative therapeutic trial of nitrogen mustard and triethylene thiophosphoramide. J Chron Dis 1960;11: Karnofsky D, Abelmann WH, Craver LF, et al. The use of the nitrogen mustards in the palliative treatment of carcinoma. Cancer 1948;1: Bergner M, Bobbitt RA, Carter WB, Gilson BS. The Sickness Impact Profile: development and final revision of a health status measure. Med Care 1981; 19: Yates JW, Chalmer B, McKegney FP. Evaluation of patients with advanced cancer using the Karnofsky performance status. Cancer 1980;45: Mendoza TR, Wang XS, Cleeland CS, et al. The rapid assessment of fatigue severity in cancer patients: use of the Brief Fatigue Inventory. Cancer 1999;85: Cella DF, Tulsky DS, Gray G, et al. The Func-

11 414 Simmonds Vol. 24 No. 4 October 2002 tional Assessment of Cancer Therapy scale: development and validation of the general measure. J Clin Oncol 1993;11: Harding VR, Williams AC, Richardson PH, et al. The development of a battery of measures for assessing physical functioning of chronic pain patients. Pain 1994;58: Lee C, Simmonds MJ, Novy DM, Jones S. Selfreports and clinician-measured physical function among patients with low back pain: a comparison. Arch Phys Med Rehabil 2001;82: Reuben DB, Siu AL, Kimpau S.The predictive validity of self-report and performance-based measures of function and health. J Gerontol 1992;47(4): M106 M Sharrack B, Hughes RA. Reliability of distance estimation by doctors and patients: cross sectional study. BMJ 1997;315: Simmonds MJ. Pain and performance: what are the measures and what do they mean? In: Max M, ed. Pain clinical update. Seattle, WA: IASP Press, 1999: Simmonds MJ, Olson SL, Jones S, et al. Psychometric characteristics and clinical usefulness of physical performance tests in patients with low back pain. Spine 1998;23: Reuben DB, Siu AL. An objective measure of physical function of elderly outpatients: the Physical Performance Test. J Am Geriatr Soc 1990;38: Duncan P, Weiner DK, Chandler J, Studenski S. Functional reach: a new clinical measure of balance. J Gerontol 1990;45:M192 M Price DD, Harkins SW, Baker C. Sensory-affective relationships among different types of clinical and experimental pain. Pain 1987;28: Jette AM. The Functional Status Index: reliability and validity of a self-report functional disability measure. J Rheumatol 1987;14: Shrout P, Fleiss J. Intraclass correlation: uses in assessing rater reliability. Psychol Bull 1979;86: