Energy Cost of Propulsion in Standard and Ultralight Wheelchairs in People With Spinal Cord Injuries.

Background and Purpose. Wheelchair- and subject-related factors influence the efficiency of wheelchair propulsion. The purpose of this study was to compare wheelchair propulsion in ultralight and standard wheelchairs in people with different levels of spinal cord injury. Subjects. Seventy-four subjects (mean age = 26.2 years, SD = 7.14, range = 17-50) with spinal cord injury resulting in motor loss (30 with tetraplegia and 44 with paraplegia) were studied. Method. Each subject propelled standard and ultralight wheelchairs around an outdoor track at self-selected speeds, while data were collected at 4 predetermined intervals. Speed, distance traveled, and oxygen cost ([VO.sub.2] mL/kg/m) were compared by wheelchair, group, and over time, using a Bonferroni correction. Results. In the ultralight wheelchair, speed and distance traveled were greater for both subjects with paraplegia and subjects with tetraplegia, whereas [VO.sub.2] was less only for subjects with paraplegia. Subjects with paraplegia propelled faster and farther than did subjects with tetraplegia. Conclusion and Discussion. The ultralight wheelchair improved the efficiency of propulsion in the tested subjects. Subjects with tetraplegia, especially at the C6 level, are limited in their ability to propel a wheelchair. [Beekman CE, Miller-Porter L, Schoneberger M. Energy cost of propulsion in standard and ultralight wheelchairs in people with spinal cord injuries. Phys Ther. 1999;79:146-158.]

Key Words: Energy cost, Paraplegia, Spinal cord injury, Tetraplegia, Wheelchair.

For most people, walking is a relatively efficient method of moving from place to place, particularly if a self-selected speed is used.[1,2] If serious damage to the neurological or musculoskeletal system occurs, as with people who have sustained a spinal cord injury (SCI), walking becomes less efficient.[2,3] When the energy cost of ambulation is too high, other methods of mobility are sought.

Most commonly, the alternative mode of mobility chosen is a wheelchair (WC). More than half of the 183,000 to 230,000 people with SCI in the United States[4] are nonambulatory and are presumed to use WCs. Wheelchair propulsion (WCP) is more efficient than walking for people with extensive paralysis,[5-8] but less efficient than normal walking. The inefficiency of WCP is due, in part, to the small muscle mass of the arms and the biomechanical disadvantages of using handrims to propel the WC.[9] In people with SCI, other factors such as impaired sympathetic vascular responses,[10-12] respiratory compromise,[12] and trunk instability[13] further impair the effectiveness of WCP. Because of the relative efficiency of WCP for people with SCI and the large number of people with SCI in WCs, considerable efforts have been made to identify the WC models, components, and propulsion techniques that maximize the efficiency of WCP.[11,14-21]

Energy cost studies have been used to compare the efficiency of different WCs and to document the demands of WCP in subjects with various impairments. Researchers have compared WCP using arm cranks and handrims,[9,11,17,18] front and rear location of the large wheels,[20] various surfaces,[18,22,23] different starting techniques,[24] standard and sports or lightweight WCs,[14,15] and various speeds of WCP.[21] Other researchers have investigated the kinematics of WCP[16,25] and muscle activity during WCP.[26-28] Investigations of maximal aerobic capacity in subjects with SCI[12,29,30] have added further information about physiological responses during WCP.

Limitations exist in a number of these studies. One limitation is the use of a constant, preestablished speed, rather than a self-selected speed.[11,14,21] A self-selected speed is thought to better reflect demands during daily propulsion, when the patient is able to change his or her energy output to one that is most efficient[31] for him or her. Another limitation is the use of subjects who do not have an impairment affecting their ability to propel a WC.[11,17,20] Subjects without impairment have a more stable sitting position[13] and are unable to mimic the postures that must be assumed by people with tetraplegia or paraplegia to maximize stability and muscular output. Although some research comparing lightweight and standard WCs has been published,[14,15] additional studies are needed to corroborate those findings and to investigate other variables of WCP. Aspects of WCP that we believed warranted further study were the effects of level of injury, use of a self-selected speed, and data collection during a longer bout of WCP than had been used by previous researchers.

The purpose of our study was to determine the energy cost of WCP in people with SCI, comparing ultralight wheelchairs (UWCs) with standard wheelchairs (SWCs), people with different levels of injury, and changes over a 20-minute trial.

Method

Sample

Seventy-four subjects, 44 with paraplegia and 30 with tetraplegia, who were chosen by convenience, participated in the study. Descriptive data for the sample are shown in Table 1. Sixty-nine subjects were men, and 5 subjects were women. The average age was 26.2 years (SD = 7.14, range = 17-50); 89% of the subjects were 35 years of age or younger. All subjects were tested near the end of their inpatient rehabilitation program and had at least 8 hours of sitting tolerance. The time from onset of SCI to testing averaged 5.1 months for all subjects. All subjects met the following criteria: had a complete injury as defined in the international standards of the American Spinal Injury Association (ASIA)[32] or sparing of only light touch; had a stable spine at the time of testing and were free of any rigid external trunk or neck support; were able to propel a manual WC for at least 20 minutes at a self-selected speed; and were in good health except for the sequelae of their SCI. All subjects had lesions at neurological level C6[32] or below. Subjects with C5 tetraplegia were not included because use of manual WCP is typically not a goal for these patients.

Table 1.
Sample Characteristics

                          Group

                           C6           C7-8            T2-8

Sample size                 14             16               25

Sex
 Male                       14             14               24
 Female                      0              2                1

No. of subjects by                      C7:14
neurological level          14          C8: 2

Age (y)(a)
 [bar]X                     24.57          26.19          24.88
 SD                          4.57           8.67           6.27
 Range                      17-32          17-50          17-40

Months since onset(b)
 [bar]X                      6.75           6.15           4.62
 SD                          2.07           1.72           3.60
 Range                    3.23-12.07     3.24-9.84      1.64-19.16

Vital Capacity (mL)(c)
 [bar]X                  3,085.7        3,060.00       3,768.0
 SD                       323.01         759.51          884.46
 Range                   2,200-3,900    2,000-4,800    2,200-5,500

Body Weight (lb)(d)
 [bar]X                   140.43         143.5           153.04
 SD                        13.29          25.37           22.29
 Range                    122-175        115-173         124-213

                          T10-L1        Primary Group
                                        Tetraplegia     Paraplegia

Sample size                 19             30             44

Sex
 Male                       17             29             38
 Female                      2              1              6

No. of subjects by
neurological level

Age (y)(a)
 [bar]X                     29.16          25.43          26.73
 SD                          7.95           7.0            7.28
 Range                      18-46

Months since onset(b)
 [bar]X                      3.73           6.43           4.24
 SD                          1.89           1.89           2.99
 Range                    0.90-9.13

Vital Capacity (mL)(c)
 [bar]X                  4,147.37       3,072.41       3,931.82
 SD                        892.37         609.98         898.00
 Range                   2,500-5,500

Body Weight (lb)(d)
 [bar]X                    146.21         142.07         150.10
 SD                         22.63          20.36          22.44
 Range                     116-196

(a) Range = 18-50 y.

(b) Range = 0.9-19.16 mo.

(c) Range = 2,000-5,500 mL.

(d) Range = 115-214 lb.

Procedure

All WC testing took place at the Pathokinesiology Laboratory, Rancho Los Amigos Medical Center, following approval of the research by the hospital institutional review board. Written consent to participate was obtained from all subjects after explanation of the test procedures and the rights of human subjects. Each subject performed 2 trials of WCP, 1 in an SWC and 1 in a UWC. The order of testing was randomized for each subject, and the 2 tests were performed no less than 24 hours and no more than 72 hours apart.

Subjects propelled each WC for 20 minutes around a level outdoor track, 60.5 m in circumference. The UWC was a 43.2-cm (17-in) Everest and Jennings (E&J) Lightning,(*) with 12.7-cm (5-in) casters and a 3-degree rear wheel camber. The Lightning had a rigid frame and no armrests and weighed 12.2 kg (27 lb). Its seat back height was not adjustable. A WC with a rigid frame was selected over a WC with a folding frame because the treating therapists believed, based on their experience, that the subjects would have more efficient WCP in a rigid chair. The E&J Lightning was one of the few rigid-frame WCs available when the study was initiated. The SWC the subjects used was either a 40.6- or 45.7-cm-wide (16- or 18-in-wide) E&J Premier,(*) with 20.3-cm (8-in) casters and vertical rear wheels. The 40.6-cm SWC weighed 20 kg (44 lb), and the 45.7-cm SWC weighed 22.2 kg (49 lb). Both the UWC and the SWCs had 61-cm (24-in) rear wheels with pneumatic tires, which were maintained at 60 psi. Rubber tubing was spirally wrapped around the handrims of the chairs for subjects with C6 tetraplegia to help maintain hand contact. Each subject sat on a pressure-relieving cushion, which, in most cases, was of 10.2-cm (4-in) foam with a cutout for the ischial tuberosities. Each subject transferred to the selected WC, and the footrests were adjusted. A soft cloth strap was secured around the chest of subjects with C6 tetraplegia. The strap was loose enough to allow arm movement and some trunk movement, but tight enough to prevent excessive forward excursion of the trunk.

Test equipment was fitted to each subject, and he or she was given 3 to 5 minutes to become familiar with breathing into the mouthpiece and propelling while wearing the apparatus. After a 3-minute rest, baseline resting data for oxygen uptake were recorded for 2 minutes.

Each subject propelled around the track at a self-selected speed for 20 minutes in either a clockwise or counter-clockwise direction. The direction of travel allowed the subject's stronger or dominant arm to be to the outside of the track. Expired air samples were collected without interrupting propulsion for 2 minutes at minutes 3 through 5 and for 1 minute at minutes 9 through 10, 14 through 15, and 19 through 20. The distance traveled was recorded during each period. Total distance traveled during 20 minutes or the maximum time pushed was determined by counting the number of laps completed.

The oxygen consumed by subjects was measured using a modified Douglas bag technique.[33] The inner diameter of the 2-way respiratory valve was 40 mm. A thermistor sensitive to change in the temperature of inspired and expired air was placed in the airway just beyond the mouthpiece and was used to monitor respiratory rate. Heart rate was measured using bipolar surface electrodes placed over the manubrium and the left seventh intercostal space. Heart rate and respiratory rate were transmitted to a strip chart recorder by a battery-powered FM-FM telemetry system.([dagger]) Telemetry eliminated the need for wires or cables that might impede normal WCP. Airway valves, hoses, and the telemetry system were supported by a lightweight shoulder harness. A frame of polyvinyl tubing, which weighed 1.02 kg (2.25 lb), extended behind the chair to support the 5-way valve and the 5 air collection bags (Fig. 1). Oxygen content of the expired air was measured using a paramagnetic oxygen analyzer.([dagger]) Carbon dioxide content was determined with an infrared analyzer.([double dagger]) The analyzer was calibrated daily prior to each test to gases of known content. A dry gas flowmeter was used to determine expired air volume. All volumes were corrected to standard values for temperature, pressure, and water vapor (body temperature, pressure, saturated units). Vital capacity was measured using a Breon Laboratories model 2400 spirometer.([sections])

[Figure 1 ILLUSTRATION OMITTED]

To establish validity and reliability of measurements obtained with the system, cylinders of gases of known oxygen and carbon dioxide concentrations were used to calibrate the gas analyzer sensors. A reading was taken from a gas sample in the midrange for both gases, and the sensors were adjusted to the known levels. Following this calibration, repeated samples were taken from 4 known gas samples with oxygen concentrations ranging from 2.99% to 18.02% and carbon dioxide concentrations ranging from 1.0% to 8.9%. Intraclass correlation coefficients (ICC[1,1]), as described by Shrout and Fleiss,[34] were calculated between the known values and the first reading to establish the validity of readings obtained for the sensors and between 2 repeated readings to determine the reliability of measurements obtained with the system (Tab. 2). To establish the reliability of readings for the sensors with expired air, gases were collected from 4 subjects at rest. Two readings taken 5 minutes apart were made from each bag of expired air, and ICCs were calculated (Tab. 2). The ICCs indicated high validity and reliability of measurements.

Table 2.
Validity and Reliability of Measurements From the Gas Analyzers

Known Gas Concentration         Mean Reading (%)   ICC(a)

Validity
 Oxygen
   Known concentration          11.505
   Measured concentration       11.625              .998

 Carbon Dioxide
   Known concentration          4.475
   Measured concentration       4.625               .999

Reliability
 Oxygen
   Reading 1                    11.625
   Reading 2                    11.775              .999

 Carbon dioxide
   Reading 1                    4.475
   Reading 2                    4.475               .999

Expired Gas (n=4)

Reliability
 Oxygen
   Reading 1                    18.125
   Reading 2                    18.150              .960

 Carbon dioxide
   Reading 1                    3.1075
   Reading 2                    3.1025              .998

(a) ICC=intraclass correlation coefficient.

Data Analysis

The sample was divided into 4 groups by injury levels: C6, C7-8, T2-8, and T10-L1. These groups were considered functionally distinct because of the different muscles innervated. The variables studied were those identified as important in other energy cost studies in which subjects walked or propelled their WCs at self-selected speeds.[1,5] These variables were speed, which is the best indicator of oxygen cost during walking[1] and which is independent of age or sex; total distance traveled; and oxygen cost per distance traveled ([VO.sub.2] mL/kg/m). The oxygen consumed over a given period of time ([VO.sub.2] mL/kg/min) has not been shown to differentiate among groups during WCP when a self-selected speed is used[5] and therefore was not analyzed.

Means and standard deviations were calculated for descriptive data and the energy cost variables. Comparative analyses were done by condition (WC), by group (SCI level), and over time (4 data collection periods) using a mixed-design analysis of variance (ANOVA), with 1 between-group factor (the 4 groups) and 2 within-group factors (WC and time). To help identify sources of differences, 2 X 4 (WC X time) ANOVAs for repeated measures were performed. Data were analyzed using a Crunch interactive statistical software package.([parallel]) When the computer program could not analyze data from subjects with missing data, mean group values were substituted for the missing data. This substitution was made for 2 subjects who did not complete the fourth data collection period and for 3 other subjects during one data collection period due to technical difficulties. Scheffe post hoc tests were used to determine significant comparisons. Because of the presumed interrelationship of the dependent variables, the number of variables studied was limited to the 3 identified above. In addition, the level of significance (alpha level) was lowered from .05 to .0167, using a Bonferroni correction for 3 dependent variables.[35]

The purpose of the Bonferroni correction is to compensate for the probability of a Type I error, which is increased when multiple interdependent variables are statistically analyzed. Type I errors occur when the null hypothesis is erroneously rejected. This erroneous rejection of the null hypothesis means that differences have been statistically identified, when they do not actually exist. Ottenbacher's[35] discussion of problems associated with analyzing multiple dependent variables in rehabilitation research forms the rationale for the statistical procedures used in this study. Methods for dealing with the problems of multiplicity (multiple interrelated dependent variables) include lowering the significance level, using multivariate statistical procedures, and applying appropriate post hoc tests. We were able to adopt the recommendation for the use of post hoc comparisons, but we were unable to perform multivariate analyses with the Crunch statistical program. We, therefore, chose to decrease the alpha level by using the Bonferroni correction. It is calculated by dividing the chosen alpha by the number of dependent variables. In our case, an alpha of .05 was divided by 3, which resulted in a revised alpha level of .0167.

Because of the small group sizes, subjects with C6 and C7-8 tetraplegia were combined and described as being in the "primary group with tetraplegia." The subjects with T2-8 and T10-L1 paraplegia were combined and described as the "primary group with paraplegia." The purpose of this combination of subjects was to increase the sample size and, thus, the statistical power. Power is the probability of correctly rejecting the null hypothesis. Beta, or the probability of a Type II error, is 1 -- power. A Type II error occurs when the null hypothesis is erroneously accepted. This erroneous acceptance of the null hypothesis means that differences have not been statistically identified, when they actually exist. Power is affected by sample size, the size of the effect being detected, and the chosen probability value.

A power analysis was performed, using the PASS computer program,(#) with a large effect size (ie, .4) and P=.05 (Tab. 3). A power of .8 or .85 is considered adequate to reduce the probability of a Type II error. When the subjects with C6 tetraplegia and the subjects with C7-8 tetraplegia were analyzed separately, the power was low (ie, .67 and .74, respectively). When the 2 groups were combined as primary group with tetraplegia, a power of .86, which is in the desired range, was achieved. Additional mixed-design ANOVAs were performed using tetraplegia and paraplegia as the between-group factors.

Table 3.
Power Analysis: Comparison by Sample Size(a)

Groups          Sample Size   Power(b)     [Beta](c)
Four groups
 C6             14              0.67          .33
 C7-8           16              0.74          .26
 T2-8           19              0.82          .18
 T10-L1         25              0.92          .08

Two groups
 Tetraplegia    30              0.86          .14
 Paraplegia     44              0.96          .04

(a) Effect size of 0.4 and P=.05.

(b) Power is the probability of making a correct rejection of the null hypothesis.

(c) Beta is the probability of a Type II error and is 1 -- power.

Results

Data were analyzed using WC, group, and time as independent variables and total distance traveled, speed, and [VO.sub.2] mL/kg/m as dependent variables. This section is organized so that data for the 3 dependent variables are presented for each of the independent variables.

All subjects completed the first 3 data collection periods in both WCs. Two subjects failed to complete the final data collection period, 1 subject with C6 tetraplegia in the UWC and 1 subject with T2-8 paraplegia in the SWC.

Comparison by Wheelchair

Total distance traveled (in meters). The mean distance propelled in 20 minutes in the UWC was greatest for the subjects with T2-8 paraplegia (1,517.48 m, P [is less than] .0000) and the subjects with T10-L1 paraplegia (1,557.72 m, P=.0030) (Tab. 4, Fig. 2). It was also greater for subjects in the primary group with tetraplegia (958.65 m, P=.0132) (Fig. 3), but not for the subjects with C6 tetraplegia (807.8 m) or the subjects with C7-8 tetraplegia (1,090.7 m) when analyzed separately.

Table 4. Mean Distance Traveled During 20-minute Trials in Standard and Ultralight Wheelchairs, Independent of Data Collection Period

                                  Total Distance Traveled (m)
                                  Group

                                  C6                  C7-8

No. of subjects per group         14                  16

Standard wheelchair
 [bar]X                          784.75              985.66
 SD                              155.15              222.30
 Range                         527.0-1,006.5       647.0-1,404.5

Ultralight wheelchair
 [bar]X                          807.79            1,090.66
 SD                              128.69              270.66
 Range                         631.0-997.5         758.0-1,622.5

Independent of wheelchair
(both wheelchairs combined)
 [bar]X                          796.27(f)         1,038.16(g)
 SD                              140.36              249.51
 Range                           595-986           763.5-1,513.5

                                  T2-8                T10-L1

No. of subjects per group         25                  19

Standard wheelchair
 [bar]X                        1,333.64(a)          1,439.50(b)
 SD                              265.18               218.29
 Range                         717.5-1,820.5       907.5-1,699.5

Ultralight wheelchair
 [bar]X                        1,517.48            1,557.72
 SD                              305.14              237.4
 Range                         955.0-2,024.0     1,037.5-1,867.0

Independent of wheelchair
(both wheelchairs
combined)
 [bar]X                        1,425.56            1,498.61
 SD                              297.77              232.69
 Range                         868.25-1,886.5       972.5-1,758.25

                                           Primary Group

                                  Tetraplegia         Paraplegia

No. of subjects per group         30                  44

Standard wheelchair
 [bar]X                          891.90(c)         1,379.35(d,e)
 SD                              216.21              249.02
 Range

Ultralight wheelchair
 [bar]X                          958.65            1,534.86
 SD                              256.75              275.55
 Range

Independent of wheelchair
(both wheelchairs
combined)
 [bar]X                          925.28(h)         1,457.11
 SD                              237.72              272.56
 Range

(a) Difference between wheelchairs, T2-8, P < .0000.

(b) Difference between wheelchairs, T10-L1, P=0030.

(c) Difference between wheelchairs, subjects with tetraplegia, P=.0132.

(d) Difference between wheelchairs, subjects with paraplegia, P < .0000.

(e) Interaction of group X wheelchair, subject with paraplegia, P=.0108.

(f) Difference between groups, C6 < T2-8, P < .0000; C6 < T10-L1, P < .0000.

(g) Difference between groups, C7-8 < T2-8, P < .0000; C7-8 < T10-L1, P < .0000.

(h) Difference between subjects with tetraplegia and subjects with paraplegia, P < .0000.

[Figures 2 and 3 ILLUSTRATION OMITTED]

Speed (in meters per minute). Mean speed was greater in the UWC than in the SWC for the subjects with T2-8 paraplegia (77.85 versus 67.85 m/min, P [is less than] .0000) and the subjects with T10-L1 paraplegia (80.91 versus 72.52 m/min, P=.0002) (Tab. 5). When the groups were combined, the difference in speed was 9.3 m/min (P [is greater than] .0000) for subjects in the primary group with paraplegia and 3.2 m/min (P=.0135) for subjects in the primary group with tetraplegia.

Table 5. Mean Speed During 20-Minute Trials in Standard and Ultralight Wheelchairs, Independent of Data Collection Period.

                                  Speed (m/min)

                                  Group

                                  C6             C7-8

No. of subjects per group         14              16

Standard wheelchair
 [bar]X                           40.38            50.57
 SD                                8.15            11.25
 Range                         28.13-52.13       35.7-73.38

Ultralight wheelchair
 [bar]X                           41.73            55.36
 SD                                7.70            14.06
 Range                         31.60-52.13      37.50-86.08

Independent of wheelchair
(both wheelchairs combined)
 [bar]X                           41.05            52.97
 SD                                7.92            12.91
 Range                            29.88-52.13   37.33-79.73

                                  T2-8             T10-L1

No. of subjects per group         25               19

Standard wheelchair
 [bar]X                           65.85(a)         72.52(b)
 SD                               13.33            10.74
 Range                         35.70-91.83      47.00-85.50

Ultralight wheelchair
 [bar]X                           77.85            80.91
 SD                               15.94            11.62
 Range                         43.83-92.88      55.38-94.33

Independent of wheelchair
(both wheelchairs combined)
 [bar]X                           72.85            76.72
 SD                               15.49            11.92
 Range                         46.64-92.35      51.19-88.85

                                  Primary Group

                                 Tetraplegia       Paraplegia

No. of subjects per group         30               44

Standard wheelchair
 [bar]X                           45.81(c)         69.87(d,e)
 SD                               11.13            12.47
 Range

Ultralight wheelchair
 [bar]X                           49.00            79.17
 SD                              213.37            14.28
Range

Independent of wheelchair
(both wheelchairs combined)
 [bar]X                           47.41            74.52(f)
 SD                               12.38            14.17
 Range

(a) Difference between wheelchairs, T2-8, P=.0000.

(b) Difference between wheelchairs, T10-L1, P=.0023.

(c) Difference between wheelchairs, subjects with tetraplegia, P=.0135.

(d) Difference between wheelchairs, subjects with paraplegia, P < .0005.

(e) Interaction group X wheelchair, subjects with paraplegia, P=.0005.

(f) Difference between subjects with tetraplegia and subjects with paraplegia, P < .0000.

Oxygen cost per distance traveled ([VO.sub.2] mL/kg/m). Mean [VO.sub.2] mL/kg/m was less in the UWC than in the SWC only for subjects with T2-8 paraplegia (0.13 versus 0.16 mL/kg/m, P [is less than] .0000) and subjects with T10-L1 paraplegia (0.13 versus 0.15 mL/kg/m, P=.009) (Tab. 6).

Table 6. Mean Oxygen Cost ([VO.sub.2])(in Millimeters per Kilogram per Meter) During 20-Minute Trials in Standard and Ultralight Wheelchairs, Independent of Data Collection Period

                                [VO.sub.2]
                                Group

                                C6              C7-8

Standard wheelchair
 [bar]X                         0.19            0.16
 SD                             0.05            0.03
 Range                          0.143-0.279     0.112-0.196

Ultralight wheelchair
 [bar]X                         0.18            0.15
 SD                             0.05            0.04
 Range                          0.132-0.284     0.107-0.197

Independent of wheelchair
(both wheelchairs combined)
 [bar]X                         0.19(d)         0.15
 SD                             0.05            0.03
 Range                          0.148-0.282     0.109-0.196

                                T2-8            T10-L1

Standard wheelchair
 [bar]X                         0.16(a)         0.15(b)
 SD                             0.03            0.03
 Range                          0.126-0.229     0.116-0.192

Ultralight wheelchair
 [bar]X                         0.13            0.13
 SD                             0.02            0.02
 Range                          0.099-0.185     0.116-0.180

Independent of wheelchair
(both wheelchairs combined)
 [bar]X                         0.15            0.14
 SD                             0.03            0.25
 Range                          0.119-0.196     0.122-0.186

                                Tetraplegia     Paraplegia

Standard wheelchair
 [bar]X                         0.17            0.16(c)
 SD                             0.04            0.02
 Range

Ultralight wheelchair
 [bar]X                         0.16            0.13
 SD                             0.05            0.02
 Range

Independent of wheelchair
(both wheelchairs combined)
 [bar]X                         0.17(e)         0.15
 SD                             0.04            0.03
Range

(a) Difference between wheelchairs, T2-8, P <.0000.

(b) Difference between wheelchairs, T10-L1, P=.009.

(c) Difference between wheelchairs, subjects with paraplegia, P < .0000.

(d) Difference between groups, C6 > C7-8, P=.0066; C6 > T2-8, P=.007; C6>T10-L1, P=.0002.

(f) Difference between subjects with tetraplegia and subjects with paraplegia, P=.0017.

Comparison by Groups

Total distance traveled (in meters). Subjects with C6 and C7-8 tetraplegia did not differ from each other in distance traveled (796.3 and 1,038.2 m), but they propelled a shorter distance than did subjects with T2-8 injuries (1,425.6 m) or subjects with T10-L1 injuries (1,498.6 m), independent of WC or time (P [is greater than] .0000) (Tab. 4, Fig. 2).

Speed (in meters per minute). Speed was less for the subjects in the primary group with tetraplegia (47.4 m/min) than for the subjects in the primary group with paraplegia (74.5 m/min), independent of WC or time (P [is less than] .0000) (Tab. 5).

Oxygen cost per distance traveled ([Vo.sub.2] mL/kg/m). The oxygen cost was greater for the subjects with C6 tetraplegia (0.19 mL/kg/m) than for the subjects in all other groups, independent of WC or time (C6 [is less than] C7-8 [0.15 mL/kg/m], P=.0066; C6 [is less than] T2-8 [0.15 mL/kg/ m],P=.0007; C6 [is less than] T10-L1 [0.14 mL/kg/m], P=.0002) (Tab. 6).

Comparison Over Time

Total distance traveled (in meters). Distance traveled was analyzed for each trial, but not for each data collection period.

Speed (in meters per minute). Speed increased over the 4 data collection periods for all groups except for the subjects with C6 tetraplegia, independent of the type of WC used (Fig. 3). Subjects with C7-8 tetraplegia had greater speeds during the last 2 periods (53..70 and 55.02 m/min) compared with the first 2 periods (50.43 and 52.72 m/min). Subjects with T2-8 and T10-L1 paraplegia had greater speeds during only the last period (75.21 and 78.44 m/min) compared with the first period (70.54 and 75.35 m/min). When the WC used was taken into account, the increase in speed over time in the UWC was different only for the subjects with T2-8 paraplegia (P=.0003). For the primary group with tetraplegia, speed, independent of WC, was less for the first period (45.04 m/min) than for the 3 subsequent periods (47.44, 48.23, and 48.91 m/min) (P [is less than] .0000).

Oxygen cost per distance traveled ([VO.sub.2] mL/kg/m). No change over time in oxygen cost was noted for any group (Fig. 4).

[Figure 4 ILLUSTRATION OMITTED]

Discussion

Comparison by Wheelchair

Subjects with paraplegia. For subjects with paraplegia, efficiency in the UWC was greater for all 3 variables, with greater distance traveled, increased speed, and less oxygen cost, than in the SWC. The improved WCP demonstrated by our subjects with paraplegia in the UWC was similar to that observed by Parziale,[15] who reported that adults with paraplegia increased their speed and decreased their expenditure of energy when they propelled a UWC as compared with an SWC.

Subjects with tetraplegia. The UWC was superior to the SWC for distance traveled and speed for the subjects with C6 and C7-8 tetraplegia only when they were combined into the primary group with tetraplegia. Parziale,[15] in contrast, found no difference between the distance traveled in the UWC and the distance traveled in the SWC during a 4-minute trial, a finding that might have been related to his small sample size or the short period of time his subjects propelled the WC, as compared with our subjects. He did, however, find the UWC to be superior to the SWC during a sprint push.[15] None of our groups with tetraplegia (ie, subjects with C6 tetraplegia, subjects with C7-8 tetraplegia, and primary group with tetraplegia) demonstrated a greater energy efficiency (decreased [VO.sub.2] mL/kg/m) in the UWC.

A number of WC-related factors could account for the relative efficiency of the UWC compared with the SWC. These factors include the geometry and stiffness of the frame[19]; rolling,[19,36] air,[19] and bearing[19,37] resistance; wheel stiffness[19]; push ring size[38]; static stability[36]; fore-aft and vertical location of the seat[36,37]; and wheel camber.[37] The effect of the WC weight on efficiency is unclear. One view is that WC weight has little effect on efficiency when propulsion is on Level ground.[36] Another view is that a WC of lighter weight is easier to push and, therefore, more energy-efficient.[37] Weight seems unlikely to be the sole factor accounting for the increased efficiency of one WC over another, based on the limited studies available. In one study,[16] for example, the addition of 5- to 10-kg weights to low-weight WC systems did not change wheeling kinematics, at least during the short distances used.[16] Specific WC features that might account for the efficiency of the UWC were not studied.

A UWC is not appropriate for all people with SCI, but it can be considered for most people who use WCs. The UWCs are available in both folding and rigid models, with each model having advantages over the other. A rigid UWC, which is the type of UWC used in our study, is considered by many clinicians to be more durable and more efficient to push. The folding UWC is thought to be easier for some people to put into a car because it is more compact and, unlike the rigid WC, does not require removal of the wheels for transport. Rigid WCs are frequently ordered with a lower back height and no armrests, which could compromise sitting posture in some people. The WC does not need to have those components, however, as numerous options are available. Rigid WCs can be adjusted to assist with balance and accessories can be added to produce very effective seating systems, even for people with tetraplegia. Folding UWCs, as well as currently available SWCs, are typically lighter in weight and easier to push than the SWCs used in our study.

Comparison by Group

We chose 4 groups based on level of injury to represent varying functional capabilities. The 2 groups with paraplegia differed from each other in the extent of trunk and intercostal muscle innervation. Subjects with T10-L1 paraplegia had complete or almost complete innervation of the trunk and intercostal muscles, whereas subjects with higher thoracic lesions (ie, subjects with T2-8 paraplegia) had few of these muscles innervated. Subjects with paraplegia lacked motor control of their lower-extremity musculature. Both groups with paraplegia also had normal hand function, but neither of the groups with tetraplegia did. The primary muscle groups that differentiated the subjects with C6 tetraplegia from the subjects with C7-8 tetraplegia were the triceps muscles, the latissimus dorsi muscle, the sternal portion of the pectoralis major muscle (which was not innervated in the subjects with C6 tetraplegia), and the serratus anterior muscle (which was weak in the subjects with C6 tetraplegia).

Primary group with paraplegia. Subjects with paraplegia benefited more from use of the UWC, although they moved farther and faster than subjects with tetraplegia independent of the WC used. In either WC, subjects with paraplegia could maintain speeds of almost 70 to 80 m/min, which is very close to the normal walking speed of 80 m/min for similar distances.[2,5,6] These WC speeds are similar to those identified by Lerner-Frankel et al[39] as being necessary for "community" ambulation.

In contrast to Waters and Lunsford,[5] who found differences in speed between subjects with high- and low-level paraplegia, our subjects with paraplegia, like those of Newsam et al[23] were not different from each other. The discrepancy between our finding and that of Waters and Lunsford[5] may be related to the time postinjury when testing occurred, which was earlier in our study. The importance of time following injury was demonstrated by Yakura et al,[40] who found that gait variables and walking energy cost improved in the year following initial rehabilitation for SCI. In both our study and the study by Waters and Lunsford,[5] oxygen cost per distance traveled ([VO.sub.2] mL/kg/m) was equivalent for people with both high- and low-level paraplegia.

Primary group with tetraplegia. Subjects with tetraplegia exhibited lower speeds, less total distance traveled, and higher oxygen cost than did subjects with paraplegia. Newsam et al[23] found similar differences for cycle distance and speed on carpet between subjects with tetraplegia and those with paraplegia. They also found, as we did, that subjects with C6 and C7 injuries were functionally similar to each other and that subjects with high- and low-level paraplegia were similar.

The subjects with C6 tetraplegia were least efficient of all the groups, as demonstrated by their higher oxygen cost in both the SWC and the UWC and their inability to increase their distance traveled. They were 25% slower than the subjects with C7-8 tetraplegia, 46% slower than the subjects with T2-8 paraplegia, and 48% slower than the subjects with T10-L1 paraplegia, even when using the UWC. The WCP speed of the subjects with C6 tetraplegia was only about half as great (41 m/min) as the normal walking speed. Using the criteria established by Lerner-Frankel et al[39] for community ambulation, our subjects with C6 tetraplegia would be categorized as having only marginal community mobility in their WCs because of their slow speeds and the risks to safety that their slowness would impose. Newsam et al[23] concluded similarly that most people with C6 tetraplegia are functioning near their maximum capability and demonstrate even more difficulty with community environments, such as ramps and carpeting, than they do with level propulsion on smooth surfaces.

Although configuration of the WC will influence the way the subject propels the WC, subject-related factors have an even greater effect on the efficiency of WCP. People with tetraplegia have limited upper-extremity force production, restricted handgrip strength, poor trunk balance, and small respiratory reserves. They also have lower tidal volumes, vital capacities, and peak heart rates than do subjects with paraplegia,[12] and their maximal oxygen uptake during WC ergometry is less.[29]

The extent of weakness in the upper-extremity musculature of people with tetraplegia has probably been underestimated. According to international classification standards,[32] key muscle groups that distinguish neurological levels of injury need only have a manual muscle test (MMT) grade of Fair (3/5), provided that more rostral muscles test as 5/5 on the numerical scale. Only limited forces can be generated, however, by muscles with an MMT grade of Fair. Upper-extremity weakness in people with tetraplegia has been confirmed by Powers et al,[41] especially in the medial (internal) rotators, where subjects with tetraplegia were able to generate only 38% as much force as as that generated by subjects without impairment. The sternal portion of the pectoralis major muscle, a medial rotator of the shoulder, is not innervated in a person with C6 tetraplegia and is weak in someone with a C7-level injury. This muscle has been shown to produce high-intensity electyromyographic activity in subjects with paraplegia during the push phase of WCP and is responsible for providing the primary propulsive force.[27] Weakness or absence of motor control of this muscle affects the ability of most people with tetraplegia to generate effective propulsive forces[27,28] and results in increased intensity[26,28] and duration[28] of activity in other shoulder muscles to substitute for the lost function.

The changes in speed and total distance traveled by subjects in the primary group with tetraplegia were small, although statistically significant. Given the very limited ability of these subjects to physically and physiologically meet the demands of WCP, however, even small benefits, in our opinion, are likely to be of clinical importance. The clinical importance of using the UWC may also relate to factors that we did not investigate. For our study, subjects propelled only on a level, paved surface. This environment did not replicate the variety of conditions that would be encountered during daily living because energy efficiency, speed, and distance traveled in a given period of time are not the only variables that might be altered by using an UWC. Other factors such as demands on the shoulder musculature that can lead to pain[42,43] might be positively influenced by use of UWCs. The beneficial effect of UWC use by people with tetraplegia might also be enhanced by attention to seat position, wheel size, and other features that have been shown to affect the efficiency of WCP,[19,36] but were not examined in our study.

A number of factors contribute to differences in responses between people with tetraplegia and people with paraplegia. All of these factors are related to the additional motor control that people with paraplegia gain by sparing of lower segments of the spinal cord. People with paraplegia have use of additional upper-extremity muscles, increased force production in those muscles, and more trunk and shoulder muscle stability. Although subjects with paraplegia do not perform as well as control subjects,[12,13,26] Powers et al[41] found that the torque of tested upper-extremity muscles in subjects with paraplegia was not different than that of control subjects, except in medial rotation, which was 65% of normal. In individuals with SCI, the trunk muscles are important in allowing a greater forward lean, which permits them to apply a greater force to the WC handrim.[25] In addition, activity of the intercostal and abdominal muscles, when present, improves respiratory function. Finally, people with injuries below T6 have normal sympathetic activity, which improves cardiac output and the vascular response to activity.[29]

Comparison Over Time

Speed increased over the 20-minute period for all groups, independent of the WC used, except for the subjects with C6 injuries. The subjects with C7-8 tetraplegia had different speeds between the first and last 2 periods. The subjects with paraplegia had different speeds only between the first and last periods. The change in speed over time was a reflection, in part, of the time to achieve a physiologic steady state and, in part, of the effort to put the WC into motion initially. This commencement of movement would be more difficult for people with tetraplegia with a smaller upper-extremity muscle mass and particularly difficult for people with C6 tetraplegia who lack elbow extensor force production, which is thought to be important in initiating a WC start.[15,24]

Limitations

We identified the following limitations of our study. Subjects were more familiar with the SWC than the UWC, despite practice trials with the UWC. An SWC was used daily for mobility around the hospital, and some subjects had never propelled a UWC before the study. Seat-to-wheel position, wheel camber, and other factors that might influence WC efficiency were not adjusted for each subject in the UWC. These adjustments were not possible for the SWC. Many more UWCs are now available. It is possible that the results would be different, especially for the subjects with tetraplegia, if the study were replicated with other UWCs. The study was carried out somewhat early in the period following injury. Although subjects had been participating in structured endurance and strengthening activities, they were relatively untrained. Because either training programs[30] or daily activities[40] can improve force production[30] maximum aerobic power[30] speed,[40] and oxygen cost per meter traveled,[40] the performances of the subjects, particularly those with tetraplegia, would probably have been better if more time had elapsed since injury or the subjects had been more highly trained.

Use of the Bonferroni correction increases the possibility of a Type II error, while correspondingly decreasing the possibility of a Type I error.[35] Type II errors might have occurred in this study, particularly with data obtained from the subjects with C6 and C7-8 tetraplegia. For example, changes in speed over time in the subjects with C6 tetraplegia (P=.0173), differences in speed between the 2 WCs for the subjects with C7-8 tetraplegia (P=.0351), and differences in distance traveled for the subjects with C7-8 tetraplegia (P=.023) would have been significant at the .05 level, but not at the more stringent .0167 level used. The small sample sizes for the subjects with C6 and C7-8 tetraplegia necessitated combining them to increase the statistical power, which may have resulted in masking differences that might exist between the 2 groups.

Conclusions

Based on the findings of this study, we conclude that the use of UWCs instead of SWCs for people with paraplegia is supported by the increased speed, increased distance traveled, and decreased oxygen cost ([VO.sub.2] mL/kg/m) achieved. The use of UWCs instead of SWCs for people with C6 and C7-8 tetraplegia is also supported by our data, although the changes were not as great as for the subjects with paraplegia. Oxygen cost ([VO.sub.2] mL/kg/m) was not different between the 2 WCs for subjects with tetraplegia.

People with tetraplegia have less efficient WCP than do people with paraplegia. All subjects with tetraplegia, especially the subjects with C6 injuries, were extremely inefficient in WCP and were so limited by weakness that no method of manual propulsion was efficient.

A 5-minute data collection period can be used to determine both 5-minute and 20-minute speeds for people with C6 tetraplegia because no change in speed was seen between the first and last data collection periods. For people with injuries at other levels, however, a shorter (5-minute) test of speed will not be as accurate as a 20-minute test in determining speed.

Future studies should be aimed at determining additional benefits of the UWC not investigated in our study and WC features that affect efficiency of propulsion. Other UWCs not on the market at the time of this study should also be evaluated.

(*) Everest and Jennings Co, 1100 Corporate Square Dr, St Louis, MO 63132.

([dagger]) Biosentry Telemetry Inc, 20720 Earl St, Torrance, CA 90503.

([double dagger]) Sensormedics Corp, 16305 State College Blvd, Anaheim, CA 93806.

([sections]) Breon Laboratories Inc, Park Ave, New York, NY 10016.

([parallel]) Crunch Software Corp, 2966 Diamond St, #292, San Francisco, CA 94131.

(#) NCSS, 329 N 1000 E, Kaysville, UT 84037.

References

[1] Blessey RL, Hislop HJ, Waters RL, Antonelli D. Metabolic energy cost of unrestrained walking. Phys Ther. 1976;56:1019-1024.

[2] Waters RL, Hislop HJ, Perry J, Antonelli D. Energetics: application to the study and management of locomotor disabilities -- energy cost of normal and pathologic gait. Orthop Clin North Am. 1978;9:351-356.

[3] Waters RL, Yakura JS. The energy expenditure of normal and pathologic gait. Physical and Rehabilitation Medicine. 1989;1:183-209.

[4] Stover SL, DeLisa JA, Whiteneck GG, eds. Spinal Cord Injury Clinical Outcomes From the Model System. Gaithersburg, Md: Aspen Publishers Inc; 1995.

[5] Waters RL, Lunsford BR. Energy cost of paraplegic locomotion. J Bone Joint Surg Am. 1985;67:1245-1250.

[6] Cerny K, Waters RL, Hislop HJ, Perry J. Walking and wheelchair energetics in persons with paraplegia. Phys Ther. 1980;60:1133-1139.

[7] Gordon EE, Vanderwalde H. Energy requirements in paraplegic ambulation. Arch Phys Med Rehabil. 1956;37:276-285.

[8] Hussey RW, Stauffer ES. Spinal cord injury: requirements for ambulation. Arch Phys Med Rehabil. 1973;54:544-547.

[9] Sawka MN, Glaser RM, Wilde SW, von Luhrte TC. Metabolic and circulatory responses to wheelchair and arm crank exercise. J Appl Physiol. 1980;49:784-788.

[10] Figoni SF, Bolleau RA, Massey BH, Larsen JR. Physiological responses of quadriplegic and able-bodied men during exercise at the same [VO.sub.2]. Adapted Physical Activity Quarterly. 1988;5:130-139.

[11] Brattgard S-O, Grimby G, Hook O. Energy expenditure and heart rate in driving a wheel-chair ergometer. Scand J Rehabil Med. 1970;22:143-148.

[12] Van Loan MD, McCluer S, Loftin JM, Boileau RA. Comparison of physiological responses to maximal arm exercise among able-bodied, paraplegics, and quadriplegics. Paraplegia. 1987;25:397-405.

[13] Curtis KA, Kindlin CM, Reich KM, White DE. Functional reach in wheelchair users: the effects of trunk and lower extremity stabilization. Arch Phys Med Rehabil. 1995;76:360-367.

[14] Hilbers PA, White TP. Effects of wheelchair design on metabolic and heart rate responses during propulsion by persons with paraplegia. Phys Ther. 1987;67:1355-1358.

[15] Parziale JR. Standard versus lightweight wheelchair propulsion in spinal cord injured patients. Am J Phys Med Rehabil. 1991;70:76-80.

[16] Bednarczyk JH, Sanderson DJ. Limitations of kinematics in the assessment of wheelchair propulsion in adults and children with spinal cord injury. Phys Ther. 1995;75:281-289.

[17] Glaser RM, Sawka MN, Young RE, Suryaprasad AG. Applied physiology for wheelchair design. J Appl Physiol. 1980;48:41-44.

[18] Smith PA, Glaser RM, Petrofsky JS, et al. Arm crank versus handrim wheelchair propulsion: metabolic and cardiopulmonary responses. Arch Phys Med Rehabil. 1983;64:249-254.

[19] Cooper RD. Wheelchair racing sports science: a review. J Rehabil Res Dev. 1990;27:295-312.

[20] Stoboy H, Rich BW, Lee M. Workload and energy expenditure during wheelchair propelling. Paraplegia. 1971;8:223-230.

[21] Hildebrandt G, Voight ED, Bahn D, et al. Energy cost of propelling wheelchair at various speeds: cardiac response and effect on steering accuracy. Arch Phys Med Rehabil. 1970;51:131-136.

[22] Wolfe GA, Waters RL, Hislop HJ. Influence of floor surface on the energy cost of wheelchair propulsion. Phys Ther. 1977;57:1022-1027.

[23] Newsam CJ, Mulroy SJ, Gronley JK, et al. Temporal-spatial characteristics of wheelchair propulsion: effects of level of spinal cord injury, terrain, and propulsion rate. Am J Phys Med Rehabil. 1996;75:292-299.

[24] Tupling SJ, Davis GM, Pierrynowski MR, Shephard RJ. Arm strength and impulse generation: initiation of wheelchair movement by the physically disabled. Ergonomics. 1986;29:303-311.

[25] Sanderson DJ, Sommer HJ 3d. Kinematic features of wheelchair propulsion. J Biomech. 1985;18:423-429.

[26] Harburn KL, Spaulding SJ. Muscle activity in the spinal cord-injured during wheelchair ambulation. Am J Occup Ther. 1986;40:629-636.

[27] Mulroy SJ, Gronley JK, Newsam CJ, Perry J. Electromyographic activity of shoulder muscles during wheelchair propulsion by persons with paraplegia. Arch Phys Med Rehabil. 1996;77:187-193.

[28] Mulroy SJ, Newsam CJ, Gronley JK, Perry J. The effect of level of spinal cord injury on shoulder muscle electromyography during wheelchair propulsion. Gait and Posture. 1996;4:183.

[29] Coutts KD, Rhodes EC, McKenzie DC. Maximal exercise responses of tetraplegics and paraplegics. J Appl Physiol. 1983;55:479-482.

[30] Eriksson P, Lofstrom L, Ekblom B. Aerobic power during maximal exercise in untrained and well-trained persons with quadriplegia and paraplegia. Scand J Rehabil Med. 1988;20:141-147.

[31] Waters RL, Lunsford BR, Perry J, Byrd R. Energy-speed relationship of walking: standard tables. J Orthop Res. 1988;6:215-222.

[32] International Standards for Neurological and Functional Classification of Spinal Cord Injury, Revised 1996. Chicago, Ill: American Spinal Injury Association; 1996.

[33] Waters RL. Energy expenditure. In: Perry J. Gait Analysis: Normal and Abnormal Function. Thorofare, NJ: SLACK Inc; 1992:451.

[34] Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

[35] Ottenbacher K. Multiplicity in rehabilitation research: a quantitative assessment. Arch Phys Med Rehabil. 1988;69:170-173.

[36] Brubaker C. Technical considerations, ergonomic considerations: choosing a wheelchair system. J Rehabil Res Dev Clin Suppl. 1990;(2):37-48.

[37] Ragnarsson KT. Prescription considerations and a comparison of conventional and lightweight wheelchairs. J Rehabil Res Dev Clin Suppl. 1990;(2):8-16.

[38] van der Woude LHV, Veeger HEJ, Rozendal RH, Sargeant TJ. Wheelchair racing: effects of rim diameter and speed on physiology and technique. Med Sci Sports Exerc. 1988;20:492-500.

[39] Lerner-Frankel MB, Vargus S, Brown M, et al. Functional community ambulation: What are your criteria? Clinical Management in Physical Therapy. 1986;6(2):12-15.

[40] Yakura JS, Waters RL, Adkins RH. Changes in ambulation parameters in spinal cord injury individuals following rehabilitation. Paraplegia. 1990;28:364-370.

[41] Powers CM, Newsam CJ, Gronley JK, et al. Isometric shoulder torque in subjects with spinal cord injury. Arch Phys Med Rehabil. 1994;75:761-765.

[42] Sie IH, Waters RL, Adkins RH, Gellman H. Upper extremity pain in the postrehabilitation spinal cord injured patient. Arch Phys Med Rehabil. 1992;73:44-48.

[43] Subbarao JV, Klopfstein J, Turpin R. Prevalence and impact of wrist and shoulder pain in patients with spinal cord injury. J Spinal Cord Med. 1995;18:9-13.

CE Beekman, PT, is Clinical Manager, Spinal Injury and Pediatric Services, Physical Therapy Department, Rancho Los Amigos Medical Center, 7601 E Imperial Hwy, Downey, CA 90242. Address all correspondence to Ms Beekman.

L Miller-Porter, PT, is Clinical Manager, Adult Orthopedic and Outpatient Services, Physical Therapy Department, Rancho Los Amigos Medical Center.

M Schoneberger, PT, is Physical Therapist, Santa Barbara Visiting Nurses Association, Santa Barbara, Calif. She was Physical Therapy Instructor, Spinal Injury Service, Rancho Los Amigos Medical Center, when the study was conducted.

Concept and research design were provided by Beekman and Schoneberger; writing, by Beekman and Miller-Porter; data collection and management, by Miller-Porter and Schoneberger; data analysis, by Beekman; project management, by Miller-Porter and Schoneberger; fund procurement, by Beekman; provision of subjects, by the Spinal Injury Service, Rancho Los Amigos Medical Center; consultation, by Brenda Lunsford and Rodney Atkins, PhD (data entry and analysis), and Sara Mulroy, PhD, PT. Sandy Hardy, PTA, and the staff of the Pathokinesiology Lab, Rancho Los Amigos, assisted with testing; Jacquelin Perry, MD, gave support and encouragement. Janet Konecne, PT, OCS, and James Harrison, DPT, OCS, provided the original idea and pilot work for this study.

This research was supported in part by a grant from Everest and Jennings, through the Foundation for Physical Therapy Inc.

The study protocol was approved by the Institutional Review Board of Rancho Los Amigos Medical Center.

Preliminary results of this research were presented at the Annual Conference of the American Physical Therapy Association (APTA), June 11-15, 1989, Nashville, Tenn; at the Annual Conference of the California Chapter of APTA, October 21-25, 1987, Palm Springs, Calif; and at the Combined Sections Meeting of APTA, February 1-4, 1990, New Orleans, La.

This article was submitted December 26, 1997, and was accepted October 8, 1998.