Feasibility, safety, and reliability of exercise testing using the combined arm-leg (Cruiser) ergometer in subjects with a lower limb amputation

Background Physical fitness of patients with a lower limb amputation predicts their walking ability and may be improved by physical exercise and training. A maximal exercise test is recommended prior to training in order to determine cardiovascular risks and design exercise programs. A potentially suitable ergometer for maximal exercise testing in patients with a lower limb amputation is the combined arm-leg (Cruiser) ergometer. The aim of this study was to determine feasibility, safety, and reliability of (sub)maximal exercise testing on the Cruiser ergometer in subjects with a lower limb amputation. Methods and findings Subjects with a lower limb amputation performed 1 submaximal exercise test and 3 maximal exercise tests on the Cruiser ergometer. Feasibility was determined by examining whether key variables such as power output, heart rate and oxygen uptake were correctly and reliably measured, by determining whether a test was a maximal aerobic performance, by studying reasons for non-completion, and by measuring gross efficiency. Safety was analyzed by recording complications, electrocardiogram results, and blood pressure. Reliability was tested by comparing the results of the second and third maximal exercise test. Seventeen subjects (14 men and 3 women) out of 21 preselected subjects completed the study. In general, the maximal Cruiser exercise test was feasible. Almost 75% of the subjects reached a maximal aerobic performance. The test was also safe because no complications occurred, although electrocardiogram and blood pressure could only be reliably recorded in most subjects just before and after the test. Reliability was good: Intraclass correlation was 0.84 for peak oxygen uptake. Conclusions The Cruiser ergometer is a feasible, safe, and reliable ergometer for measuring physical fitness of subjects with a lower limb amputation.


Introduction
Patients who require a lower limb amputation (LLA) are often elderly and have a high prevalence of comorbidities, especially cardiovascular diseases [1].The presence of cardiovascular diseases reduces the chance of being able to walk with a prosthesis and negatively influences mobility outcomes after LLA [2].Other factors that influence the ability to walk with a prosthesis are amputation level, age and physical fitness [3].Most patients with a LLA experience a decline in physical fitness, which in turn negatively influences their functional activity level [4,5,6].In addition, energy costs of walking with a prosthesis are much higher compared with normal walking and increase proportionally with the level of amputation [4].It has been demonstrated that maximal aerobic capacity, which is a major constituent of physical fitness, is an important predictor for walking ability in patients with a LLA due to vascular disease [5].Furthermore, exercise training can improve walking ability [6,7].
Before starting exercise training, a maximal exercise test is not only recommended [8] for reasons of safety, especially with regard to cardiovascular risks, but also for developing individually tailored exercise programs.To achieve the best possible outcomes, the cardiovascular system has to be maximally stressed by using the largest possible muscle mass, i.e., by obtaining the highest possible oxygen uptake (peak VO 2 ) and/or work capacity [9].Patients with a LLA have lower functional muscle mass by definition.Previous studies have researched the use of different ergometers by LLA patients, including the arm ergometer [10] and the unilateral bicycle ergometer [5,7].A disadvantage of the arm ergometer is that only the arms and part of the upper body are used.Similar to the arm ergometer, the bicycle ergometer also only involves the muscle mass of one extremity (the leg) and part of the upper body.Patients with a LLA often need help to make the cycling movement with one leg and have difficulty maintaining balance.A combined arm-leg ergometer, the Cruiser ergometer, is a suitable alternative for testing the physical fitness of patients with a unilateral LLA [11].The Cruiser ergometer has several advantages for its users: they are seated on the ergometer and their back and residual limb are supported; they can exercise without the help of a therapist; and they use the muscle mass of the trunk and 3 extremities, i.e., 1 leg and 2 arms.The use of relatively high muscle mass during exercise on a combined arm-leg ergometer may lead to a higher peak VO 2 [12][13][14].Previous research has demonstrated that the Cruiser ergometer is a valid, reliable, and safe instrument for measuring the physical fitness of healthy volunteers [12].The next step is to study the Cruiser in subjects with a LLA.Therefore, the aims of the current study are 1) to explore the feasibility of the Cruiser ergometer in maximal exercise testing in subjects with a LLA, 2) to evaluate the safety of the Cruiser ergometer, and 3) to study the test-retest reliability of a repeated maximal exercise test on the Cruiser ergometer.

Population
Twenty-one subjects with a LLA (18 men and 3 women) living in the North of the Netherlands were screened for participation this study.The principal investigator (ES) informed specialists in Rehabilitation Medicine and certified prosthetists working in the North of the Netherlands about this study.These specialists and prosthetists subsequently asked subjects with a LLA to participate in the study and provided them with written information.Following the subjects' agreement to participate, the principal investigator (ES) contacted the subjects to screen for inclusion and exclusion criteria.Inclusion criteria were: age between 18 and 75 years and a LLA (unilateral transfemoral amputation, knee disarticulation, or transtibial amputation).Exclusion criteria were: coronary heart disease, clinically relevant arrhythmia, hypertension (diastolic blood pressure > 100 mm Hg or systolic blood pressure >180 mm Hg), recently diagnosed pulmonary embolism, bilateral LLA, upper limb amputation, and cognitive impairments leading to inability to cooperate or inability to obtain consent [8].
Prior to their participation, subjects signed an informed consent form.All tests were conducted in accordance with the Declaration of Helsinki.The local Medical Ethics Committee (METc UMC Groningen) approved the study (file number METc 2011/123).

Instruments
The Cruiser ergometer (Enraf-Nonius serial number: 3800EN014, ultimo number 10-3013, Delft, The Netherlands) was standardized for use in exercise testing.It is a combined arm-leg ergometer equipped with a comfortable seat (Fig 1).The foot of the user is placed against a fixed footrest, which can be adjusted to the subjects' length.Users can perform tests on the ergometer without their prosthesis.The residual limb is supported by a special stump support connected to the seat.The footrest is used to push off and make the seat move backward.Users can move the seat forward again by pulling the handlebars.In this way, arms and leg are simultaneously used to overcome the resistance provided by the ergometer.The position of the footrest of the ergometer was adjusted to a fixed setting for each subject during the tests.The ergometer could only be set in a constant power mode of between 35 and 60 revolutions per minute (rpm) and subjects were instructed to maintain 50 rpm [12][13][14].The accuracy of the Cruiser ergometer is ± 10% power output (PO in W) and ± 2 rpm speed.Cardiorespiratory parameters were recorded using an Oxycon Delta (Jaeger, Bunnik, the Netherlands).Subjects wore a face mask and ventilation (VE in l/min), oxygen uptake (VO 2 in l/min), and carbon dioxide output (VCO 2 in l/min) were continuously measured.Peak VO 2 and peak VCO 2 were defined as the highest average values obtained over a 30-s period.Blood pressure was measured manually at the beginning of the test, immediately after the test was completed, and after the cooling down period.Heart rate (HR in b/min) was continuously recorded with a 12-lead electrocardiogram (ECG).

Study design
All subjects performed 4 exercise tests (1 submaximal and 3 maximal tests) on 3 different days, with an interval of more than 1 week between each test day.On day 1, the principal investigator screened the subjects for contraindications by means of a questionnaire, ECG, and measurement of blood pressure.Next, the subjects started the submaximal test to become acquainted with the Cruiser ergometer and to determine the gross mechanical efficiency (GE).The submaximal exercise test consisted of 3 minutes rest on the Cruiser followed by 3 minutes exercise at 20 W and 3 minutes exercise at 30 W at 50 rpm.GE of the participants was measured during the final 30 seconds of the submaximal blocks of 20 and 30 W. GE was determined to analyze the mechanical efficiency of the movement on the Cruiser ergometer.GE is an important measure because it can be used to evaluate future training effects as well as motor learning effects of the Cruiser ergometer [13,14].After the submaximal test, subjects had a rest period of more than 10 minutes, which was followed by the first maximal exercise test.The maximal exercise test was repeated on days 2 and 3.The first maximal exercise test was seen as a familiarization test, and tests 2 and 3 were used to determine test-retest reliability.Each maximal exercise test started with 3 minutes rest on the Cruiser and was followed by a 3 minute warm-up at 20 W at 50 rpm.After the warm-up, work load was increased by 10 W per minute, keeping speed at 50 rpm, until the point of exhaustion was reached or until the physician stopped the test.Reasons for terminating the test were inability to maintain 50 rpm, pain in arms or legs, chest pain, dizziness or faintness, severe dyspnea, pallor, cyanosis, or cold and clammy skin.The test was also stopped by the investigator in case of ECG abnormalities.After completing the test, subjects were observed for another 3 minutes.The protocol was derived from an earlier study in healthy volunteers [12].Taking the lower exercise capacity of subjects with LLA into account, appropriate adaptations in workload were made.The maximal exercise test protocol was consistent in all subjects over the 3 test days.

Outcome
Feasibility.To evaluate the feasibility of the Cruiser ergometer as an instrument for measuring the physical fitness of subjects with a LLA, this study specifically focused on the indicators acceptability, demand, and practicality [15].To this end, a 5-step approach was used.First, it was investigated whether all relevant variables could be measured during the test.Second, subjects who dropped out of the study were analyzed, and this analysis served to refine exclusion criteria for an exercise test on the Cruiser ergometer.Third, when subjects experienced symptoms of dyspnea and fatigue, these symptoms were assessed using a 10-point Borg scale at peak load for dyspnea and for arm and leg muscle fatigue so as to determine the impact of the test [16].Fourth, the maximal aerobic performance was evaluated for each test.Since no reference data are available of the maximal aerobic performance on the Cruiser ergometer using one leg and two arms, the criteria for the bicycle ergometer were used.A performance was regarded as a maximal aerobic performance when 1 or more of 3 criteria were fulfilled: a heart rate of more than 85% of predicted, maximal ventilation of more than 75% of predicted, or a respiratory exchange rate (RER) of more than 1.1 [17].In addition, it was studied to what extent VO 2 peak was related to predicted VO 2 peak, which is age and gender dependent (calculated for the bicycle ergometer).Fifth, the GE of movement during the submaximal period of 20 and 30 W was calculated and compared to previous research of Simmelink et.al in healthy volunteers [13,14].GE (in %) is derived from the ratio between the mechanical power output (Po) and the metabolic power (P met ), as shown in Eq 1 [18].
The metabolic power (Pmet) was calculated using Eq 2.
Safety.Subjects became acquainted with the Cruiser ergometer during the submaximal exercise test.When no complications occurred, they started the maximal exercise test.It was studied whether ECG and blood pressure could be reliably recorded prior to, during, and after the (sub)maximal exercise tests.Outcome measures for safety were ECG abnormalities or the occurrence of adverse events during the tests.
Reliability.Test-retest reliability of the maximal exercise test on the Cruiser was studied by comparing outcome measures of the second and third tests.Outcome measures for reliability were peak oxygen uptake (VO 2 peak), peak heart rate (HR), peak power output (PO), peak ventilation (VE), peak breathing frequency (BF), and respiratory exchange rate (RER).

Statistics
Data analyses were performed using SPSS version 23.0 for Windows (SPSS, Chicago, Illinois, USA).Descriptive statistics were generated for all variables (mean ± SD).
Reliability of the second and third maximal exercise tests was assessed by means of paired ttests, the one-way random effect model and single measure-intraclass correlation coefficient (ICC), and Bland-Altman plots.Level of significance was set at p <0.05.

Results
Between May 2012 and September 2015 a total of 21 subjects were screened for inclusion.One subject was excluded because of a Syme amputation, which made it impossible to perform the tests without using a prosthesis.Three other subjects dropped out of the study due to medical reasons.Consequently, the complete test results of 17 subjects (14 men and 3 women) could be obtained.Mean age was 54.5 y (SD 18.6, range 25-80), mean BMI 25.2 kg/m 2 (SD 4.0, range 18.4-31.2),and the mean time since amputation was 96.6 months (SD 111.2, range 2-372).Subject characteristics and outcome measures are shown in Table 1.

Feasibility
Three subjects dropped out of the study due to medical reasons.Subject 1 developed an allergic skin reaction located at the site of the ECG suction cups after the first day.The subject recovered within three days but did not perform days 2 and 3 of the protocol.Subject 2 showed signs of cardiac ischemia during the first maximal exercise test and was referred to a cardiologist.This subject withdrew from the study.Subject 3 dropped out after day 2 because of an increase in stump pain.Another subject had hypertension on day 1 (196/103 mmHg) and was referred to his general practitioner.This subject, however, could re-enter the study after treatment.All subjects who completed the protocol (n = 17) were able to perform the movement on the Cruiser ergometer.In most cases (second test 9/17 subjects, third test 11/17 subjects), the chosen protocol led to an exercise time of 10-15 minutes, which is considered to be the ideal duration of a maximal exercise test.Monitoring of cardiopulmonary parameters was possible in all subjects.
The Borg scores, though relatively low, were fairly consistent for all of the tests (Table 2).The Borg score for arm fatigue was significantly higher for the first maximal test than for the second test.There were no significant differences between the tests with regard to other Borg scores.
The majority of the subjects had a maximal aerobic performance: 11 of 16 subjects performed maximal at test 1, 12 of 17 subjects performed maximal at test 2, and 14 of 17 subjects performed maximal at test 3. VO 2 peak was 75% of predicted (SD 30, range 40-140%) for test 1; for test 2 it was 73% of predicted (SD 26, range 36-118%); and for test 3 it was also 73% of predicted (SD 25, range 40-141%) (Table 1).In most cases, the test was stopped because coordination problems of arms and leg, which resulted in an inability to maintain the speed of 50 rpm (Table 3).

Safety
Most of the tests were performed without complications.As mentioned earlier, three adverse events occurred, resulting in attrition of three subjects.These adverse events were an allergic skin reaction due to the ECG suction cups, an abnormal ECG, and an increase in pain of the stump without physical signs after exercising on the Cruiser ergometer.The ECG could be reliably recorded in all subjects, at least during low to medium workloads.The investigator had to stop a number of tests because the ECG was affected by muscle activity of the arms and thorax (Table 3).Blood pressure could not be reliably measured during exercise, but only before and after the test.

Reliability
The first maximal exercise test was a familiarization trial.Tests 2 and 3 were used for test-retest reliability analysis.No significant differences were present, except in RER (Table 1).A slight learning/ practice effect was seen, especially with regard to the peak workload in the 3 sequential maximal exercise tests.The ICCs are presented in Table 4.It is advocated that an ICC of > 0.75 indicates good agreement [19,20].This means there is a good agreement for the outcome measures VO 2 peak and POpeak between tests 2 and 3.
Finally, Bland-Altman plots were constructed.The plot for VO 2 shows a bias close to zero with two outliers (

Discussion
In this study, feasibility, safety, and reliability of exercise testing on the combined arm-leg Cruiser ergometer were evaluated in subjects with a LLA.The test procedure was feasible because almost all subjects were able to perform the combined arm-leg movement and managed to reach adequate symptom scores.Furthermore, a large majority of subjects reached a maximal aerobic performance and most variables of interest could be measured appropriately.The test procedure was safe and ECGs could be reliably recorded in most cases.Approximately 20% of ECG recordings were severely influenced by muscle activity of arms and thorax.In addition, blood pressure could not be measured during the tests on the upper arm or on the wrist.Finally, the test-retest reliability was good.

Feasibility
Three subjects dropped out of this study.One subject had an allergic skin reaction to the ECG suction cups.Although the skin reaction healed spontaneously within a few days, the subject was withdrawn from the study.ECGs are now in clinical practice recorded using self-adhesive electrodes.Another subject dropped out because his pre-existent stump pain worsened after test 2, despite the use of a special stump support attached to the Cruiser.Most subjects, however, felt the stump support was sufficiently comfortable and did not experience pain or discomfort in their stump.Borg scores for dyspnea and fatigue of the arms and leg were relatively low (mean scores between 2.8 and 3.9).In a study on healthy volunteers, Borg scores for fatigue of the arms (4.9-5.0) and legs (4.3-4.5) were higher [12].It could be hypothesized that the real maximal aerobic capacity was not measured in our study population.In the study of Wezenberg [21], 93% of subjects reached a RER peak value of more than 1.1.In our study, at least 1 of the following 3 criteria had to be fulfilled in order to consider a test as a maximal aerobic performance: a heart rate of more than 85% of predicted, maximal ventilation of more than 75% of predicted, and a RER of more than 1.1 [17].Following these criteria, at the first test 69% (n = 11/16) of subjects had a maximal exercise test and at the second and thirds test 71% (n = 12/17) and 82% (n = 14/17) of subjects had a maximal exercise test, respectively.These scores may be explained by several factors.Subjects may have experienced increased symptom perception due to the unfamiliar type of exercise, to coordination problems, and to the relatively large impact of arm exercise in relation to leg exercise.In addition, a few tests were discontinued due to ECG disruptions.Finally, the fixed protocol with relatively large increments of workload may have been less appropriate.Some learning effects were found, including higher VO2 and greater power output, which have also been demonstrated in previous research [12].The GE varied greatly between subjects and ranged between 2.7-18.1% at 20 W and 5.5-14.0%. at 30 W. In earlier research in healthy young subjects, a GE of 13.0 in men and 15.0% in women at 45 W was found [13].When using the Cruiser ergometer in the future for testing and training, it is important to realize that motor learning effects can vary among subjects.

Safety
Although ECG recordings were possible during the exercise test on the Cruiser, they were hampered by muscle activity of arms and thorax.Nevertheless, ECG abnormalities were found in one subject, who was referred to a cardiologist.Exercise stress testing on the bicycle ergometer has a sensitivity of 33-50% and a specificity of 84-96% for detecting coronary artery disease in patients with suspected myocardial ischemia.ECG abnormalities are more easily detected after stopping the exercise.Therefore, in clinical practice, ECGs are recorded 2 minutes after the exercise phase [22,23].In addition, blood pressure could not be measured during the test, neither with an upper arm monitor or manually measured nor with a pulse monitor This is a limitation of the Cruiser test with respect to safety especially for patients with LLA with a high risk of cardiovascular disease.In this study, however, no subject had a systolic blood pressure of > 225 mm Hg or a diastolic blood pressure of > 99 mmHg directly after the exercise phase.In future research measurement of blood pressure may be possible, e.g. by using a discontinuous protocol.In addition, further research into the question whether blood pressure measurement before and directly after the test is sufficient, has to be performed.Finally, with the exception of the three subjects who dropped out of the study due to medical reasons, no other subjects experienced medical problems during or after the tests.

Reliability
Reliability was measured by comparing the second and third test on the Cruiser ergometer and was found to be satisfactory in general, as can be deducted from the ICC and Bland-Altman plots.However, the Bland-Altman plots revealed some outliers.The outliers may be explained by a difference in effort of some subjects between the second and third test and by prematurely stopping the test in 7 subjects because of suspected ECG abnormalities.In retrospect, the premature stop of the tests was not necessary, for the abnormalities could be explained by muscle activity in all but one subject.It is recommended that both subject and investigator learn how to use the Cruiser ergometer prior to testing so as to improve reliability.

Limitations of this study
As shown in Table 1, physical fitness of the study population varied greatly.This is to be expected in patients with LLA, who follow a rehabilitation program to learn to walk with a prosthesis.Most subjects in our study were trauma patients, which is unusual given that peripheral arterial disease is the main reason for performing a LLA [24].It is likely that selection bias occurred in this study.Specialists in Rehabilitation Medicine and certified prosthetists located in the North of the Netherlands were asked by the principal investigator to recruit subjects with a LLA for this study.This might have led to a selection bias of motivated and relatively healthy subjects.Consequently, the findings of this study might overestimate the physical fitness of this population The study population was not large enough to perform a comparative analysis between amputees with different causes of the amputation such as trauma, vascular disease, cancer, pain syndrome and neurofibromatosis.The time since amputation varied between 2 to 372 months, which is a large variation.This variation, however, enabled us to study the Cruiser in both inexperienced and experienced subjects.No differences were seen with regard to ease of use and becoming acquainted with the Cruiser ergometer.One protocol was used in all subjects.For some subjects, the submaximal level of 30 W was already the maximal power output they could reach on the Cruiser ergometer.Two subjects only reached a peak power output of 20 W, which was still a maximal aerobic performance when considering the criteria for maximal aerobic exercise.In future studies, individually tailored exercise protocols are needed instead of a one-size-fits-all protocol.These protocols should aim at a test duration of 10-15 minutes [19].

Recommendations for further research
An implementation study is recommended to design patient-specific protocols for using the Cruiser ergometer at the start of the rehabilitation period of patients with a LLA.Also, future research is needed to design exercise training protocols using the data from the baseline Cruiser exercise test.

Conclusions
The Cruiser ergometer is a feasible, safe, and reliable testing instrument for measuring the physical fitness of this study population with LLA.Adequate practice before actual testing, and the development of individually tailored testing protocols are advised.

Fig 3 .Fig 4 .
Fig 3. Bland-Altman plot for peak heart rate of the exercise test on the Cruiser ergometer of 17 subjects with a lower limb amputation.diffHRt2t3 = difference between peak heart rate (b/min) between test 2 and test 3 on the Cruiser ergometer.mean HRt2t3 = mean peak heart rate (b/min) of test 2 and test 3 on the Cruiser ergometer.https://doi.org/10.1371/journal.pone.0202264.g003