A SYSTEMATIC REVIEW AND META-ANALYSIS COMPARING CARDIOPULMONARY EXERCISE TEST VALUES OBTAINED FROM THE ARM CYCLE AND THE LEG CYCLE RESPECTIVELY IN HEALTHY ADULTS

Abstract

Introduction

The cardiopulmonary exercise test (CPET) assesses maximal oxygen uptake (VO_2max) and is commonly performed on a leg cycle ergometer (LC). However, some individuals would rather perform the CPET on an arm cycle ergometer (AC).

Objective

The objectives of this study were to undertake a systematic review and meta-analysis of the difference in VO_2max achieved by AC compared to LC in healthy adults and to explore factors that may be predictive of this difference.

Methods

MEDLINE, EMBASE, CINAHL, and PEDro were searched in April 2015. The differences in VO_2max (ACLC_diff) were pooled across studies using random effects meta-analysis and three different methods were used to estimate the ratio between the values obtained from the tests (ACLC_ratio).

Results

This paper included 41 studies with a total of 581 participants. The mean ACLC_diff across studies was 12.5 ml/kg/min and 0.89 l/min with a mean ACLC_ratio of 0.70. The ACLC_diff was lower in studies with higher mean age and lower aerobic capacity.

Conclusion

There is linear association between the AC and LC values in healthy adults. The AC values were on average 70% of the LC values. The magnitude of this difference appeared to be reduced in studies on older and less active populations.

Level of evidence

3a

INTRODUCTION

The cardiopulmonary exercise test (CPET) is the gold standard for the direct assessment of maximal oxygen uptake (VO_2max).^1-5 VO_2max determines the maximal ability for the human body to deliver, obtain and consume oxygen during maximal exercise and is a measure of maximum aerobic capacity.⁴ Assessments of aerobic capacity are used by healthcare professionals to evaluate exercise capacity,⁵ exercise intolerance⁶ and functional aerobic impairment,⁷ which all provide important information on health status and prognosis in various populations.^2,8-11

CPET is commonly performed on a treadmill or on a leg cycle ergometer (LC).^3,5 However, due to disability, co-morbidity, preference or athletic discipline there is a need to investigate alternatives to the LC test.¹² In some cases, it could be more important to assess aerobic fitness using the arms when leg exercise is not feasible or possible.^13-15 A potential alternative is to perform the test with the upper body using an arm cycle ergometer (AC).¹³ However, the AC has limitations as studies have shown that untrained individuals will achieve a lower level of VO_2max on the AC, due to a reduced stress on the cardiovascular system, compared to the LC test.^12,15,16 Having a smaller amount of muscle mass being active during the test, AC is likely to result in an earlier termination of the CPET due to peripheral factors such as an earlier onset of lactate threshold, rather than central cardiovascular limitations.^12,17 While individual studies have directly assessed the difference in VO_2max of a CPET conducted using AC compared to LC in healthy adults, no previous systematic review of these studies has been published.

The objectives of this study were to undertake a systematic review and meta-analysis of the difference in VO_2max achieved by AC compared to LC in healthy adults and to explore factors that may be predictive of this difference. The determination of this factor would allow the direct comparison of data obtained on the two tests.

METHODS

This review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.¹⁸

DATA SOURCES AND SEARCHES

Preliminary searches were conducted and relevant search terms identified. A formal search of the databases MEDLINE, EMBASE, CINAHL, and PEDro was undertaken in April 2015. References of the identified studies in the preliminary searches were screened and relevant search terms were added to the search strategy. The search strategy consisted of a combination of relevant keywords and MeSH/Thesaurus terms for: 1) direct assessment of VO_2max (or VO_2peak), 2) a CPET performed on an AC and 3) a CPET performed on an LC. No language or publication limits were applied. The reference lists of identified studies were checked and the authors of unobtainable studies were contacted. Papers suggested by experts in the field were evaluated. Search strategies specified for MEDLINE are presented in Appendix 1.

STUDY SELECTION

Study selection was undertaken based on a priori defined criteria. Only original research papers reporting within comparison (AC and LC) VO_2max (or VO_2peak), as milliliter oxygen per kilogram per minute (ml/min/kg) or as liters per minute (l/min), were considered eligible for inclusion in this systematic review. The CPET had to be non-assisted on AC and LC. Studies that reported values for healthy adults (age > 18 years) with a level of physical activity < 300 minutes per week were included. People with higher physical activity levels were considered athletes and where therefore excluded.¹⁹

Two authors (RTL, CK) independently screened titles and abstracts and assessed eligible articles in full-text. Any inconsistencies between authors were discussed and disagreement was solved by consultation with a third author (JC).

DATA EXTRACTION AND RISK OF BIAS ASSESSMENT

The following information was extracted: sample size, gender distribution, mean age, mean height, body mass index (BMI) together with the VO_2max values, peak respiratory exchange ratio (RER), CPET starting watt, and watt increment for both the AC and LC test.

The Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies (QAT)²⁰ was used to assess the methodological quality of all included studies. Six items (6-10 and 13) were considered not applicable for the studies included in this review and thus did not contribute to the quality rating total score (SumQAT). Two authors (RTL and CK) independently extracted data and undertook the quality assessment. Inconsistencies between reviewers were discussed and in cases of disagreement, a third reviewer (JC) was consulted.

DATA ANALYSIS

The mean VO_2max difference between the AC test and the LC test (ACLC_diff) was calculated for each study. Given the within subject nature of these comparisons the standard deviation (SD) of this difference for the within subject correlation was adjusted, using the method described in chapter 16.4.6.1 of the Cochrane Handbook.²¹ Hereby the SD was calculated using an imputed r-value of 0.5. The level of statistical heterogeneity was assessed using the I² score. Given the variation of participant characteristics across included studies the ACLC_diff, for ml/kg/min and l/min, were pooled across studies using a random effects meta-analysis. Summary of the characteristics of included studies are expressed as median values and interquartile range (IQR).

Meta-regressions were used to perform sub-group analyses to clarify which variables were affecting the main analysis on the ACLC_diff. The sub-groups included were: aerobic capacity (as a categorical variable based on the Åstrand classification -“low”, “fair”, “average”, “good” or “high”),²² participant mean age (in years), gender (percentage of males), study risk of bias (SumQAT), and the difference in obtained peak RER values during test.

Three different approaches were used to find the ratio between AC and LC (ACLC_ratio). First a meta-analysis of the ACLC_ratio was undertaken using the studies presenting the group mean ± SD of the within comparison ratio (%). Second, a linear regression model was determined using the group mean values. The linear regression analysis was weighted by sample size. Third the reported AC values were divided with the reported LC values, giving an estimate of the ratio in each study, which are expressed as a total mean ratio.

All analyses were performed using Review Manager 5.3 (Cochrane collaboration) software and Stata 14.0 software (StataCorp. 2013. Stata Statistical Software: Release 14.9 College Station, TX: StataCorp LP). A p-value ≤ 0.05 was considered statistically significant.

RESULTS

Results of the search

The electronic searches identified 3,300 records. After removal of 617 duplicates, 2,683 unique studies remained. 2,510 studies were excluded by screening their title and abstract and 171 studies were considered eligible for full text review. Of these, 130 did not meet the inclusion criteria. Thus, 41 studies (published between 1973 and 2014) were included in the review.^{12,15,17,23-60} The study selection process is summarized as a flow chart in Figure 1.

Figure 1.

Flowchart illustrating the systematic literature search, screening of studies and full text assessment.

Description of studies

A summary of the characteristics of the 41 included studies is provided in Table 1. Some of the included studies reported results from several groups and the data extraction and risk of bias were therefore performed on 53 groups. The full details of included studies are listed in Appendix 2.

Table 1.

Study Characteristics of the 53 included groups

Continent of publication	(%)
North America	56.6 %
Europe	35.8 %
South America	3.8 %
Asia	3.8 %

Study Design
RCT	17.0 %
Non-RCT	3.8 %
Cross-sectional	79.2 %

Study risk of bias	Median (IQR)
SumQAT	4 points (3 to 5)

Participant characteristics
Gender	(%)
Male only	66 %
Female only	15.1 %
Mixed	15.1 %
Not reported	3.8 %

	Median (IQR)
Mean age years	28.4 years (25 to 32.3)
Mean BMI, kg/m2	23.65 kg/m2 (22.7 to 25)

Aerobic capacity	(%)
Low	3.8 %
Average	28.7 %
Good	5.6 %
High	3.8 %
Did not report	58.1 %

Test characteristics
Order on AC/LC test	(%)
AC first	3.8 %
LC first	18.9 %
Random order	45.3 %
Not reported	32 %

	Median (IQR)
Time between tests (hours)	72 (24 to 168)
AC start level (watts)	25 (15 to 40)
LC start levels (watts)	50 (30 to 50)
AC increase/min (watt)	10.7 (5 to 17)
LC increase/min (watt)	30 (20.7 to 30)

IQR: Interquartile range, SumQAT: sum of quality assessment tool score, AC: Arm cycle, LC: Leg cycle

Risk of bias in included studies

Figure 2 presents a summary of the risk of bias in the included groups. The median SumQAT was 4 points, (IQR: 3 to 5). A detailed risk of bias figure is presented in Appendix 3.

Figure 2.

Risk of bias graph. Review authors’ judgements about each risk of bias item presented as percentages across all included studies.

Research question and study population

Although all included groups were judged to have a well-defined research question (Item 1), 13^{15,17,25,30,34-38,44,47,52,53} had insufficient description of the population (Item 2). One group⁴³ described the participation rate of eligible subjects (Item 3) and 13 ^{15,24,26,29,32,33,35,36,41,43,46,55,57} had a well-defined description of the subject selection (Item 4). Four groups ^15,38,39,46 included sample size justification (item 5).

Outcome measures

Five groups ^{17,34,40,45,53} did not report the VO_2max as ml/kg/min but as l/min (item 11) and therefore did not adjust the outcome for subject weight.

Blinding and statistical analysis

One group¹² blinded the outcome assessor (Item 12) and 12^{25,34,37,38,40,42,43,45,48,51,52,55} did not provide report a description of their statistical analysis methods (Item 14).

META-ANALYSIS OF VO_2MAX DIFFERENCE BETWEEN AC AND LC

A total of 36 groups (413 participants) reported data for the ACLC_diff measured in ml/kg/min. The meta-analysis for the ACLC_diff is shown in Figure 3 with a pooled mean ACLC_diff of 12.5 ml/kg/min, (95% CI: 10.3 to 14.7, I² = 59.9%, p > 0.001) favouring the LC test values. A total of 37 groups (415 participants) presented data of the ACLC_diff in l/min with a pooled mean ACLC_diff of 0.89 l/min, (95% CI: 0.78 to 1.00, I² = 30.5%, p = 0.043) favouring the LC values as shown in Figure 4.

Figure 3.

Random effects meta-analysis of the difference between group means for arm cycle (AC) and leg cycle (LC) in ml/kg/min.

Figure 4.

Random effects meta-analysis of the difference between group means for arm cycle (AC) and leg cycle (LC) in l/min.

SUBGROUP ANALYSES

In univariate meta-regression and multivariate meta-regression lower participant mean age and higher aerobic capacity were found to be significantly associated to an increased ACLC_diff. The meta-regressions are shown in Table 2.

Table 2.

Meta-regression analyses performed on each variable (univariate) and adjusted for all variables (multivariate)

Univariate meta-regression on ACLCdiff	Groups included in analysis	Mean coefficient (95% CI)	p-value
Aerobic capacity	27	4.1 (95% CI: 1.5 to 6.6)	p = 0.003
Gender distribution (% male)	33	−1.25 (95% CI: −7.4 to 4.9)	p = 0.684
Mean age	29	−2.1 (95% CI: −0.3 to −0.1)	p<0.001
Mean difference in peak RER values	24	−12.1 (95% CI: −68.8 to 44.6)	p = 0.663
Risk of bias (SumQAT score)	34	−0.19 (95% CI: −2.6 to 2.2)	p = 0.875

Multivariate meta-regression on ACLCdiff
Aerobic capacity	16	4.0 (95% CI: 0.81 to 7.2)	p = 0.019
Gender distribution (% male)	16	4.5 (95% CI: −4.1 to 13.2)	p = 0.268
Mean age (years)	16	−0.25 (95% CI: −0.4 to −0.06)	p = 0.014
Mean difference in peak RER values	16	7.9 (95% CI: −59.0 to 74.8)	p = 0.797
Risk of bias (SumQAT score)	16	0.9 (95% CI: −3.8 to 5.6)	p = 0.682

ACLC_diff: difference between obtained AC VO_2max and obtained LC VO_2max, 95% CI: 95% confidence intervals, RER: respiratory exchange ratio, SumQAT: sum of quality assessment tool score

ANALYSES OF THE AC/LC RATIO

The mean ratio between the AC test and the LC test for the 37 groups (n = 413 participants) reporting VO_2max in ml/kg/min was 0.70 (95% CI: 0.66 to 0.73) in favour of the LC. The corresponding value of the 37 groups (n = 415 participants) reporting VO_2max in l/min, the mean ACLC_ratio was 0.71 (95% CI: 0.66 to 0.75). The meta-analysis (n = 46 participants) for the ACLC_ratio across groups was 71%, (95% CI: 68 to 74, I² = 0%, p = 0.530) (figure 5). The coefficient for the linear regression in Figure 6, between the AC and the LC mean VO_2max was 0.65 ml/kg/min (95% CI: 0.48 to 0.81) with an r² of 0.689.

Figure 5.

Random effects meta-analysis comparing studies that reports ratio values for the within person comparison between arm cycle (AC) and leg cycle (LC) as group mean ± SD.

Figure 6.

Two-way scatter plot and best fitted line between the mean arm cycle (AC) maximal oxygen uptake (VO2_max) values and mean VO2_max leg cycle (LC) values.

DISCUSSION

This systematic review and meta-analysis brings together data from 41 studies and 53 groups in 581 healthy individuals directly comparing VO_2max values obtained from the AC compared to LC. The LC values were found to be substantively higher (mean difference: 12.5 ml/kg/min and 0.89 l/min) than the AC values. But with an I² value of 59.9% for the ACLC_diff in ml/kg/min these results could be affected by substantial heterogeneity. The results support the belief that the AC test achieves lower oxygen uptake values as it involves a smaller amount of muscle mass and places less stress on the cardiovascular system.^12,15,16

Both age and the aerobic capacity appear to be associated with the ACLC_diff. The difference is decreased with increasing age and increased with better aerobic capacity. This was expected, due to the fact that aerobic capacity decreases with age.²²

The RER represents the relationship between the volume of carbon dioxide and the volume of oxygen in every breath and it is recommended to continue VO_2max tests until RER values above 1.1 are reached in order to obtain a valid CPET.²³ The majority of groups reporting RER values reported values in both tests to be above 1.1.^{23,24,26-29,32,36,38,46,60} Only one group reported RER values above 1.1 for the AC and below 1.1 for the LC,²³ and three groups reported RER for both test to be below 1.1.^33,39,49 The difference in the obtained RER values were expected to affect the ACLC_diff. However, this relationship was not found, which could be due to a lack of power, as not all included groups reported the values for the meta-regressions. The level of aerobic capacity is affected by gender.²² However, a correlation between gender distribution and the ACLC_diff was not found. This makes our results applicable for future research and clinical use in single gender groups as well as mixed gender groups.

The ACLC_diff does not seem to be affected by the risk of bias in the studies as low quality studies are reporting the same ACLC_diff as high quality studies. This may be explained by the precise and accurate equipment used during CPET,⁶¹ and thereby limits the possibility of imprecise testing in different settings, which increases the clinical applicability.

The most accurate estimate of the ratio is the meta-analysis of the reported ratios, but only four groups ^33,39,46,54 reported mean ± SD (%) values for the ratio between the tests. The meta-analysis revealed a linear relationship between the AC test values and the LC test values with an ACLC_ratio of 0.7. This analysis should be seen as the main expression for the ratio between the values of the AC and the LC, where no important heterogeneity was found.⁶² Three different methods were used to estimate the ACLC_ratio. The calculation and the linear regression of the ACLC_ratio should only be used as a prediction, since they do not incorporate variation. Despite different approaches to estimate the ratio, the results are very similar and the ACLC_ratio of 0.7 is similar to the ones described in the literature.^33,39,46,54 To investigate if the 0.7 is a valid estimate for the population mean ACLC_ratio, future research should report within comparison ratios between the AC and the LC, making them applicable for inclusion in meta-analysis.

This is the first systematic review and meta-analysis of the literature comparing arm and leg exercise, and it is thus important to stress that this paper has a number of limitations. First, some groups did not report ACLC_diff SD. Thus, imputation was performed of the value based on an assumed within participant correlation coefficient (r-value) between AC and LC VO_2max. This method is recommended by the Cochrane Handbook²¹ but may still influence the accuracy of the findings. The only way to avoid these limitations in a meta-analysis is for future research to report the correlation coefficients between the two tests. However, sensitivity analyses were undertaken to assess the impact of this estimation on the findings. A small number of groups have reported a range of correlation coefficients between the AC test and the LC test (0.78, 0.94, 0.77, 0.32, 0.70).^{12,17,31,37,54} A sensitivity analysis was performed on the r-values and the pooled ACLC_diff was found to be 12.52 ml/kg/min (95% CI: 10.2 to 14.6) based on the lowest of the reported r-values (0.32) and 12.6 ml/kg/min (95% CI: 10.6 to 14.7) based on the highest reported r-value (0.94). Thus, it was shown that this imputation method made no difference to the pooled results. Future studies should report the standard deviation (or equivalent) of the mean difference between AC ad LC VO_2max or the within person correlation coefficient.

Second, the quality of the included studies was variable. This review sought to assess study risk of bias using the QAT as it can be applied to cross-sectional studies.²⁰ However, to make this tool relevant for this review some of the original QAT elements (items 6-10 and item 13) were dropped, as they were inapplicable to the research question. This could affect the reliability of the tool, and hereby the method of assessing the risk of bias, but the remaining items should be a good measure specific to the method of this paper.

Third, this review was limited to non-athlete healthy adults and limits generalizability of the current findings. Non-athlete healthy adults are expected to have a larger aerobic capacity when doing CPET using the legs compared to the arms due to everyday use and large lower limb muscle mass.²⁹ However, in athletic populations, particularly arm-trained populations, the ACLC_diff is expected to be smaller than shown in this review.⁶³ To avoid systematic bias, 18 comparisons in individuals performing more than 300 minutes per week of physical activity or involved in competitive exercise were excluded.¹⁹ Those groups contained ‘well trained subjects’, ‘triathletes’, ‘swimmers’, ‘cross-country skiers’ or ‘highly arm-trained’. However, no exclusion sedentary individuals was made. Two of the groups included extremely sedentary or sedentary subjects.^39,47 But having an ACLC_ratio of 0.76 and 0.64 these groups are not likely to have had a systematic affect on the final results. A sensitivity analysis was performed without the two groups and showed no impact on the result as the pooled ACLC_diff was found to be 12.7 ml/kg/min (95% CI: 10.4 to 15.0). Future well-conducted studies that directly compare AC and LC in other populations are needed. Future research should also include disease populations with limitations by lower limb disability such as peripheral vascular disease or osteoarthritis, or trained and upper extremity dominant athletes.

CONCLUSION

This systematic review and meta-analysis showed that the VO_2max achieved by the AC tests were on average 70% of the VO_2max achieved by the LC test, in studies on healthy non-athletic individuals. There was a linear association between the VO_2max for the AC test and the LC test. The magnitude of this difference appeared to be reduced in studies with older and less active populations.

APPENDIX 1.

Search strategy for MEDLINE

Search item 1	Search item 2	Search item 3
Bicycling [MeSH]	Arm [MeSH]	Oxygen consumption [MeSH]
Ergometry [MeSH]	CPET arm	Physical endurance [MeSH]
Leg [MeSH]	Arm cycle ergometry	Exercise test [MeSH]
CPET leg	Arm-crank ergometry	Fatigue [MeSH]
CPETleg	Arm ergometry	Physical exertion [MeSH]
Cycle ergometry	CPETarm	Oxygen uptake
Leg cycle ergometry	Arm crank	Physical fitness
cycle ergometer	Arm crank ergometer	VO2
bicycle ergometer	Arm ergometer	VO2 max
Electromagnetically braked ergometer	Arm cycle ergometer	VO2max
Lower body exercise	Arm cycling	Oxygen consumption
Cardiopulmonary exercise testing leg	Upper body exercise	Peak oxygen consumption
Cardiopulmonary exercise testing	Arm cranking	Anaerobic threshold
ETT	Cardiopulmonary exercise testing arm	VO2 peak
Exercise tolerance test	Arm exercise	VO2peak
Leg exercise	Cranking	Maximal aerobic power
Cycling	Arm work	Aerobic power
Cycle exercise		Aerobic capacity
Leg cycling		Work capacity
Leg ergometry		Peak pulmonary O2 metabolic efficiency
		Maximal oxygen uptake
		Oxygen uptake
		Peak exercise
		Peak physiologic responses
		Cardiorespiratory responses
		Physiological comparison

APPENDIX 2.

Detailed characteristics of included studies

Aminoff, T. et al. (1996)
Title	Physical work capacity in dynamic exercise?with differing muscle masses in healthy young and older men
Methods	Cross-sectional
Participant status	19 healthy, non-smokers, and physically active men. Subjects participated in conditioning exercises, on average, two to four times a week. Group ”old”: n = 9 Group ”young”: n = 10 Age: ”old” 56.9 ± 1.5 years, ”young” 26.3 ± 2.3 years Gender (M/F): Only males Height: ”old”: 178 ± 6.1 cm, ”young” 184.2 ± 4.8 cm Bodyweight: ”old”: 82.8 ± 7.2 kg, ”young” 81.7 ± 11.1 kg BMI: ”old” 26.2 ± 3.1 kg/m2, ”young” 26.3 ± 2.3 kg/m2 AC RER: ”old” 1.15 ± 0.04, ”young” 1.15 ± 0.08 LC RER: ”old” 1.09 ± 0.04, ”young” 1.13 ± 0.05
Study protocol
	AC or LC tested first: Random Time between tests: 1 day AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: 5.35 W/min LC Watt increase: 10.7 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 4
Notes
Aminoff, T. et al. (1999)
Title	Physiological strain during kitchen work in relation to maximal and task-specific peak values
Methods	Cross-sectional
Participant status	Nine kitchen workers from a large hospital kitchen with a conveyor belt, collecting and sorting dirty plates. Group ”female”: n = 6 Group ”male”: n = 3 Age: ”female” 32.3 ± 10.3 years, ” male” 28.7 ± 8.6 years Gender (M/F): “female”: only females, “male”: only males Height: ” female”: 171.8 ± 7.2 cm, ” male” 179.4 ± 5.4 cm Bodyweight: ” female”: 62.9 ± 6.1 kg, ” male” 75.1 ± 5.9 kg BMI: ” female” 21.3 ± 1.7 kg/m2, ” male” 23.4 ± 3.1 kg/m2 AC RER: ” female” 1.10,” male” 1.12 LC RER: ” female” 1.10 ” male” 1.12
Study protocol
	AC or LC tested first: LC Time between tests: 1 day AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: 5.5 W/min LC Watt increase: “25 or 50” Watts every third minute
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 5
Notes
Bhambhani, Y, et al, (1998)
Title	Muscle oxygenation during incremental arm and leg exercise in men and women
Methods	Cross-sectional
Participant status	Fifteen men and 10 women who were free from metabolic and cardiorespiratory diseases. The volunteers were university students and members of local sports club. Group ”female”: n = 10 Group ”male”: n = 15 Age: ”female” 27.3 ± 4.9 years, ” male” 25.2 ± 5.3years Gender (M/F): “female”: only females, “male”: only males Height: ” female”: 166 ± 4.1 cm, ” male” 176 ± 8.0 cm Bodyweight: ” female”: 64.5 ± 5.4 kg, ” male” 72.8 ± 8.5 kg BMI: ” female” 22.1 ± 3.9 kg/m2, ” male” 24.8 ± 2.1 kg/m2 AC RER: ” female” 1.21 ± 0.08 ” male” 1.2 ± 0.09 LC RER: ” female” 1.2 ± 0.08 ” male” 1.19 ± 0.07
Study protocol
	AC or LC tested first: Random Time between tests: Within one week AC Start Watt: 25 LC Start Watt: 30 AC Watt increase/min: 12.5 W/min LC Watt increase: 15 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 5
Notes
Bhambhani, Y, et al, (1995)
Title	Prediction of stroke volume during upper and lower body exercise in men and women
Methods	Cross-sectional
Participant status	37 recreationally active subjects not involved in any particular exercise-training programme. Group ”female”: n = 12 Group ”male”: n = 25 Age: ”female” 32.1 ± 7.7 years, ” male” 35.0 ± 7.3 years Gender (M/F): “female”: only females, “male”: only males Height: ” female”: 161.1 ± 8.1 cm, ” male” 177.8 ± 6.6 cm Bodyweight: ” female”: 59.5 ± 6.0 kg, ” male” 82.9 ± 9.1 kg BMI: ” female” 22.9 kg/m2, ” male” 26.2 kg/m2 AC RER: ” female” 1.31 ± 0.11 ” male” 1.2 ± 0.09 LC RER: ” female” 1.33 ± 0.07 ” male” 1.22 ± 0.06
Study protocol
	AC or LC tested first: Random Time between tests: Within two weeks AC Start Watt: 12.5 LC Start Watt: 30 AC Watt increase/min: 6.25 W/min LC Watt increase: 15 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 5
Notes
Bhambhani, Y, et al, (1991)
Title	Transfer effects of endurance training with the arms and legs
Methods	RCT
Participant status	16 healthy middle-aged male subjects in the arm group, and 8 healthy middle-aged male subjects in the leg group. Group ”arm”: n = 16 Group ”leg”: n = 8 Age: ” arm” 35.2 ± 6.6 years, ” leg” 41.0 ± 4.7 years Gender (M/F): “arm” only males, “leg” only males Height: ” arm”: 176 ± 7.1 cm, ” leg” 177.6 ± 5.0 cm Bodyweight: ” arm”: 85.5 ± 9.9 kg, ” leg” 80.4 ± 11.1 kg BMI: ” arm” 27.6 kg/m2, ” leg” 25.49 kg/m2 AC RER: ” arm” 1.15 ± 0.05 ” leg” 1.21 ± 0.09 LC RER: ” arm” 1.21 ± 0.03 ” leg” 1.21 ± 0.07
Study protocol
	AC or LC tested first: Random Time between tests: Within two weeks AC Start Watt: 12.5 LC Start Watt: 30 AC Watt increase/min: 6.25 W/min LC Watt increase: 15 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 4
Notes
Barstow, T. J. et al. (1993)
Title	O2 uptake kinetics and the O2 deficit as related to exercise intensity and blood lactate
Methods	Cross-sectional
Participant status	Four untrained subjects aged 24-38 years, weight 59-89 kg n = 4 Age: 24-38 years Gender (M/F): Three males and one female Height: ” arm”: 176 ± 7.1 cm, ” leg” 177.6 ± 5.0 cm Bodyweight: 59-89 kg
Study protocol
	AC or LC tested first: Cannot determine (CD) Time between tests: not reported AC Start Watt: Not reported LC Start Watt: Not reported AC Watt increase/min: Not reported LC Watt increase: Not reported
Outcomes of interest	VO2max
Risk of bias	SumQAT: 3
Notes
Boileau, R. A. et al. (1984)
Title	Cardiovascular and metaboli contributions to the maximal power of the arms and legs
Methods	Cross-sectional
Participant status	Moderately active, nonathletic male college students. n = 40 Age: 18-25 years Gender (M/F): only males Bodyweight: 75.8 ± 7.8 kg AC RER: 1.16 ± 0.08 LC RER: 1.16 ± 0.08
Study protocol
	AC or LC tested first: Random Time between tests: Cannot determine AC Start Watt: 60 LC Start Watt: 150 AC Watt increase/min: 6 LC Watt increase: 15
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 6
Notes
Bond, V. et al. (1986)
Titel	Aerobic capacity during two-arm and one-leg exercise
Methods	Cross-sectional
Participant status	Eight healthy males. The subjects had not participated in any upper or lower body conditioning for 12 months prior to the tests n = 8 Age: 24 ± 4.7 years Gender (M/F): only males Height: 177 ± 6.5 cm Bodyweight: 77 ± 9.8 kg BMI: 24.58 kg/m2
Study protocol
	AC or LC tested first: Random Time between tests: Within one weeks AC Start Watt: 30 LC Start Watt: 30 AC Watt increase/min: 30 W/min LC Watt increase: 30 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 4
Notes
Bouchard, C. et al. (1979)
Title	Specificity of maximal aerobic power
Methods	Cross-sectional
Participant status	30 moderately active male subjects. n = 30 Age: 28 ± 4.5 years Gender (M/F): only males Bodyweight: 73.0 ± 6.6
Study protocol
	AC or LC tested first: Random Time between tests: Within two weeks AC Start Watt: 75 LC Start Watt: 125 AC Watt increase/min: 10 W/min LC Watt increase: 25 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 5
Notes	The study reports a correlation coefficient between AC and LC on r = 0.70.
Castro, R. et al. (2011)
Title	Different ventilatory responses to progressive maximal exercise test performed with either the arms or legs
Methods	Cross-sectional
Participant status	12 subjects of hospital staff and university students. They were considered healthy upon a clinical evaluation (physical examination and clinical history) and a maximal exercise test performed on a cycle ergometer. None of the subjects were engaged in regular physical exercise. None of the subjects were accustomed to arm-crank exercise. Group n = 12 Age: 27 ± 1 years Gender (M/F): Six males and six females BMI: 22.7 ± 0.7 kg/m2 AC RER: 1.37 ± 0.03 LC RER: 1.26 ± 0.03
Study protocol
	AC or LC tested first: Not reported Time between tests: Not reported AC Start Watt: 30 LC Start Watt: 30 AC Watt increase/min: 20 W/min LC Watt increase: 15 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 6
Notes
Charbonnier, J. P. et al. (1975)
Title	Experimental study on the performance of competition swimmers.
Methods	Cross-sectional
Participant status	Six non-swimmers with a mean age of 31 ± 4 years, a mean height of 178 ± 3 cm and a mean bodyweight of 71 ± 5 kg. The swimmers were among the best in the country and the non-swimmers were members of the laboratory staff. Group n = 6 Age: 31 ± 4 years Gender (M/F): Three males and three females Height: 178 ± 3 cm Bodyweight: 71 ± 5 kg BMI: 22.41 kg/m2 AC RER: 1.06 ± 0.01 LC RER: 1.04 ± 0.03
Study protocol
	AC or LC tested first: Not reported Time between tests: Not reported AC Start Watt: Not reported LC Start Watt: Not reported AC Watt increase/min: Not reported LC Watt increase: Not reported
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 5
Notes
Davies, C. T. M., & Sargeant, A. J. (1974)
Title	Indirect determination of maximal aerobic power output during work with one or two limbs.
Methods	Cross-sectional
Participant status	12 healthy male subjects. All except four subjects were accustomed to physical investigations. Group n = 12 Age: 30.5 ± 6.2years Gender (M/F): only males Height: 178.2 ± 5.1 cm Bodyweight: 75.1 ± 10.1 kg BMI: 23.65 kg/m2
Study protocol
	AC or LC tested first: Not reported Time between tests: Not reported AC Start Watt: Not reported LC Start Watt: Not reported AC Watt increase/min: Not reported LC Watt increase: Not reported
Outcomes of interest	VO2max
Risk of bias	SumQAT: 1
Notes
Davis, J. A. et al (1976)
Title	Anaerobic threshold and maximal aerobic power for three modes of exercise.
Methods	Cross-sectional
Participant status	39 healthy university male students. None of them had been endurance training four months prior to the experiment. Nine of the 39 subjects participated only in the validation period. Group n = 30 Age: 22.5 ± 2.6 years Gender (M/F): only males Height: 179.8 ± 6.9 cm Bodyweight: 75.5 ± 9.0 kg BMI: 23.35 kg/m2
Study protocol
	AC or LC tested first: Not reported Time between tests: One day AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max
Risk of bias	SumQAT: 4
Notes
Dekerle, J. et al. (2002)
Title	Ventilatory thresholds in arm and leg exercices with spontaneously chosen crank and pedal rates.
Methods	Cross-sectional
Participant status	20 male students in physical education with a mean age of years, a mean height of cm and a mean bodyweight of kg. Group n = 12 Age: 22 ± 2.2 years Gender (M/F): only males Height: 180 ± 6 cm Bodyweight: 73.5 ± 5.3 kg BMI: 22.69 kg/m2 AC RER: 1.2 ± 0.1 LC RER: 1.2 ± 0.1
Study protocol
	AC or LC tested first: not reported Time between tests: Within one week AC Start Watt: 30 LC Start Watt: 60 AC Watt increase/min: 15 LC Watt increase: 30
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 4
Notes
Franklin, B. A. et al. (1983)
Titel	Aerobic requirements of arm ergometry: implications for exercise testing and training
Methods	Cross-sectional
Participant status	10 healthy male subjects. Group n = 10 Age: 28 ± 2.4 years Gender (M/F): only males Height: 170.7 ± 7.5 cm Bodyweight: 69.3 ± 7.1 kg BMI: 23.78 kg/m2 AC RER: 1.06 ± 0.09 LC RER: 1.12 ± 0.11
Study protocol
	AC or LC tested first: not reported Time between tests: One day AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 2
Notes	The study reports a correlation coefficient between AC and LC on r = 0.32.
Franssen, F. M. et al. (2002)
Title	Arm mechanical efficiency and arm exercise capacity are relatively preserved in chronic obstructive pulmonary diseas
Methods	Cross-sectional
Participant status	Controls, male/female (14/6) did not participate in any exercise-training program and were found through an advertisement in the local newspaper Group n = 20 Age: 61 ± 1 years Gender (M/F): 14 males and six females BMI: 25.9 ± 0.6 kg/m2 AC RER: 1.22 ± 0.02 LC RER: 1.23 ± 0.02
Study protocol
	AC or LC tested first: not reported Time between tests: One week AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 4
Notes
Javierre, C. et al. (2007)
Title	Physiological responses to arm and leg exercise in women with chronic fatigue syndrome.
Methods	Cross-sectional
Participant status	15 healthy controls that were extremely sedentary. Their occupation did not require physical effort, they did not perform physical activity and or their hobbies were sedentary. Group n = 15 Age: 61 ± 1 years Gender (M/F): only females Bodyweight: 57.5 ± 5.1 kg Height: 159 ± 5 cm BMI: 22.74 kg/m2 AC RER: 1.05 ± 0.11 LC RER: 1.08 ± 0.09
Study protocol
	AC or LC tested first: AC Time between tests: 10 minutes AC Start Watt: 10 LC Start Watt: 0 AC Watt increase/min: 10 W/min LC Watt increase: 12.5 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 6
Notes
Keteyian, S. et al. (1994)
Title	Cardiovascular responses of cardiac transplant patients to arm and leg exercise.
Methods	Cross-sectional
Participant status	10 healthy normal men. Five healthy subjects performed leg exercise three or more times per week and one healthy adult performed arm and leg exercise three times per week. Group n = 15 Age: 51 ± 5 years Gender (M/F): only males Bodyweight: 80.4 ± 15.6 kg
Study protocol
	AC or LC tested first: Random Time between tests: One hour AC Start Watt: 15 LC Start Watt: 30 AC Watt increase/min: 15 W/min LC Watt increase: 10 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 3
Notes
Lewis, S. et al. (1980)
Title	Transfer effects of endurance training to exercise with untrained limbs
Methods	RCT
Participant status	Five healthy male college students. Group n = 5 Age: 22 ± 2 years Gender (M/F): only females Bodyweight: 79 ± 13 kg Height: 186 ± 10 cm BMI: 22.84 kg/m2 AC RER: 1.27 ± 0.06 LC RER: 1.28 ± 0.07
Study protocol
	AC or LC tested first: Random Time between tests: Within one week AC Start Watt: 25 LC Start Watt: 50 AC Watt increase/min: 7 W/min LC Watt increase: 33 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 6
Notes
Loughney, L. et al. (2014)
Title	Comparison of oxygen uptake during arm or leg cardiopulmonary exercise testing in vascular surgery patients and control subjects
Methods	Cross-sectional
Participant status	Twenty healthy control subjects . Group n = 20 Age: 31 (24–42) years Gender (M/F): 10 males and 10 females BMI: 26 (24–28) kg/m2
Study protocol
	AC or LC tested first: not reported Time between tests: not reported AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 5
Notes	The study reports a correlation coefficient between AC and LC on r = 0.77.
Louhevaara, V. et al. (1990)
Title	Differences in cardiorespiratory responses during and after arm crank and cycle exercise.
Methods	Cross-sectional
Participant status	21 untrained healthy men. Group n = 21 Age: 33.3 ± 5.9 years Gender (M/F): only males Bodyweight: 78.3 ± 12.7 kg Height: 178.4 ± 7.2 cm BMI: 24.6 kg/m2 AC RER: 1.09 ± 0.08 LC RER: 1.17 ± 0.01
Study protocol
	AC or LC tested first: LC Time between tests: 2-7 days between AC Start Watt: 25 LC Start Watt: 50 AC Watt increase/min: 25 W/min LC Watt increase: 50 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 3
Notes
Lyons, S. et al. (2007)
Title	Excess post-exercise oxygen consumption in untrained men following exercise of equal energy expenditure: comparisons of upper and lower body exercise
Methods	Cross-sectional
Participant status	The subjects were recruited from the local university and city community, and consisted of individuals who were already participating in at least 30 min of moderate recreational physical activity on most days of the week. Group n = 10 Age: 25.7 ± 5.83 years Gender (M/F): only males Bodyweight: 104.9 ± 18.71 kg Height: 183.6 ± 6.73 cm BMI: 31.12 kg/m2 AC RER: 1.09 ± 0.08 LC RER: 1.17 ± 0.01
Study protocol
	AC or LC tested first: Random Time between tests: not reported AC Start Watt: 12 LC Start Watt: 50 AC Watt increase/min: 6.25 W/min LC Watt increase: 17.5 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 6
Notes
McConnell, T. R. et al. (1984)
Title	The hemodynamic and physiologic differences between exercise modalities
Methods	Cross-sectional
Participant status	Healthy subjects. Group n = 10 Age: 29.4 years Gender (M/F): Five males and five females Bodyweight: 63.5 AC RER: 1.09 ± 0.08 LC RER: 1.17 ± 0.01
Study protocol
	AC or LC tested first: Random Time between tests: Within one week AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 3
Notes
Nag, P. K. (1984)
Title	Circulo-respiratory responses to different muscular exercises
Methods	Cross-sectional
Participant status	Five young men, free from any cardiovascular. They actively participated habitually in moderately heavi agricultural work, none were the limb and trunk muscles specially trained. Group n = 5 Age: 22.7 ± 3.4 years Gender (M/F): only males Bodyweight: 48.9 ± 0.9 kg Height: 160.2 ± 2.0 cm BMI: 19.05 kg/m2 AC RER: 1.2 LC RER: 1.01
Study protocol
	AC or LC tested first: Random Time between tests: not reported AC Start Watt: 50-75 LC Start Watt: 50-75 AC Watt increase/min: 25-50 LC Watt increase: 25-50
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 3
Notes
Orr, J. L. et al. (2013)
Title	Cardiopulmonary exercise testing: arm crank vs cycle ergometry
Methods	Cross-sectional
Participant status	Fifteen healthy women were recruited from the University of Dundee. Group n = 15 Age: 23.5 ± 3.7 years Gender (M/F): Only females Bodyweight: 60.6 ± 7.8 kg Height: 167 ± 5 cm BMI: 21.73 kg/m2
Study protocol
	AC or LC tested first: Random Time between tests: 30 minutes AC Start Watt: 15 LC Start Watt: 25 AC Watt increase/min: 25-50 LC Watt increase: 25
Outcomes of interest	VO2max
Risk of bias	SumQAT: 5
Notes
Pogliaghi, S. et al. (2006)
Title	Adaptations to endurance training in the healthy elderly: arm cranking versus leg cycling
Methods	RCT study
Participant status	18 men were recruited by local advertisements in the metropolitan area of Verona (Italy). Group “arm” n = 6, “leg” n = 6, “c” n = 6 Age: “arm” 68 ± 4, “leg” 66 ± 5, “c” 73 ± 4 years Gender (M/F): only males Bodyweight: “arm” 74 ± 6, “leg” 76 ± 11, “c” 80 ± 8 kg Height: “arm” 172 ± 4, “leg” 169 ± 6, “c” 173 ± 8 cm BMI: “arm” 25 ± 2 kg/m2, “leg” 27 ± 3 kg/m2, “c” 27 ± 2 kg/m2 AC RER: 1.2 ± 0.1, “leg” 1.2 ± 0.1, “c” 1.1 ± 0.07 LC RER: 1.2 ± 0.07, “leg” 1.2 ± 0.06, “c” 1.2 ± 0.05
Study protocol
	AC or LC tested first: Random Time between tests: 60 minutes AC Start Watt: 40 LC Start Watt: 50 AC Watt increase/min: 5 LC Watt increase: 10
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 7
Notes
Protas, E. J. et al. (1996)
Title	Cardiovascular and metabolic responses to upper- and lower-extremety exercise in men with idiopathic Parkinson's disease.
Methods	Cross-sectional study
Participant status	7 control subjects from the local community with a mean age of 65 (53-71) were recruited. The controls were more sedentary than the PD group. Group n = 7 Age: 53-71 Gender (M/F): only males AC RER: 1.07 (1.0-1.18) LC RER: 1.12 (1.01-1.23)
Study protocol
	AC or LC tested first: Random Time between tests: 20 minutes AC Start Watt: 10 LC Start Watt: 20 AC Watt increase/min: 5 LC Watt increase: 10
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 3
Notes
Rathnow, K. M., & Mangum, M. (1990)
Title	Cardiopulmonary exercise testing: arm crank vs. cycle ergometry
Methods	Cross-sectional
Participant status	The single mode treatment group consisted of six males and three females. The mixed mode treatment group consisted of five males and three females. Due to the presentation of data in the study, and mixture of sexes in the groups this study will appear with no demographic data. Group “Mixed” n = 8, “Single” n = 9, “control” n = 7 Gender (M/F): “Mixed” five males and three females, “Single” six males and three females, “control” three males and four females. AC RER: 1.09 ± 0.08, “Single” 1.08 ± 0.08, “control” 1.07 ± 0.07 LC RER: 1.15 ± 0.08, “Single” 1.16 ± 0.08, “control” 1.07 ± 0.06
Study protocol
	AC or LC tested first: LC Time between tests: at least 20 minutes AC Start Watt: 15 LC Start Watt: 50-150 AC Watt increase/min: 25-50 LC Watt increase: 50
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 5
Notes
Reybrouck, T. et al. (1975)
Title	Limitations to maximum oxygen uptake in arm, leg and combined arm and leg ergometry.
Methods	Cross-sectional
Participant status	The untrained subject was 25 years old. Group n = 1 Age: 25 years Gender (M/F): only males AC RER: 0.95 ± 0.03 LC RER: 1.16 ± 0.05
Study protocol
	AC or LC tested first: not reported Time between tests: not reported AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 4
Notes
Ramonatxo, M. er al. (1996).
Title	Differences in mouth occlusion pressure and breathing pattern between arm and leg incremental exercise.
Methods	Cross-sectional study
Participant status	Eight normal male subjects. No subjects were involved in exercise training but all maintained their accustomed exercise training. Group n = 8 Gender (M/F): only males Age: 20-35 Heigh: 167-183 cm Weight: 60-80 kg
Study protocol
	AC or LC tested first: Random Time between tests: Within one week AC Start Watt: 25 LC Start Watt: 50 AC Watt increase/min: 6.25 LC Watt increase: 12.5
Outcomes of interest	VO2max
Risk of bias	SumQAT: 3
Notes
Rösler, K. et al (1985)
Title	Transfer effects in endurance exercise.
Methods	Trial
Participant status	Ten healthy male subjects with a mean age of 30.5 (23-37) years, a mean height of 178 (172-182) cm and a mean bodyweight of 70.8 (64-83) kg. None of them had been involved in regular training during the preceding two years, although some of then did some recreational jogging or cycling. Group n = 10 Age: 30.5 (23-37) years Gender (M/F): only males Bodyweight: 70.8 (64-83) kg. Height: 178 (172-182) BMI: 22.35 kg/m2
Study protocol
	AC or LC tested first: not reported Time between tests: not reported AC Start Watt: 40 LC Start Watt: 100 AC Watt increase/min: 6.667 W/min LC Watt increase: 17.5 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 4
Notes
Sargeant, A. J., & Davies, C. T. M. (1973).
Title	Perceived exertion during rhythmic exercise involving different muscle mass
Methods	Cross-sectional study
Participant status	Six healthy male subjects. All of the subjects had taken part of physiological investigations before and were habituated to the exercise modalities in the study. All of the physical characteristics are given individually. The age ranges from 24-39 years, the height ranges from 171.8-189.0 cm and the bodyweight from 63.0-98.0 kg. Group n = 6 Sex (M/F) = 6/0 Age: 30.6 ± 5.53 years Gender (M/F): only males Bodyweight: 79.1 ± 11.68 kg Height: 179.4 ± 5.98 cm BMI: 24.58 kg/m2
Study protocol
	AC or LC tested first: not reported Time between tests: not reported AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max
Risk of bias	SumQAT: 2
Notes
Sawka, M. N. Et al. (1983)
Title	Physiological factors affecting upper body aerobic exercise
Methods	Cross-sectional study
Participant status	Nine male subjects. Group n = 9 Sex (M/F): 9/0 Gender (M/F): only males Gender (M/F): 1/0 Bodyweight: 79.8 ± 7.8 kg
Study protocol
	AC or LC tested first: AC Time between tests: Two weeks AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max
Risk of bias	SumQAT: 2
Notes	The study reports a correlation coefficient between AC and LC on r = 0.94.
Sharp, M. A. et al. (1988)
Title	Maximal aerobic capacity for repetitive lifting: comparison with three standard exercise testing modes
Methods	Cross-sectional study
Participant status	18 male subjects. Group n = 18 Gender (M/F): only males Age: 23.9 ± 3.7 years Bodyweight: 75.9 ± 8.8 kg Height: 177.7 ± 8.9 kg BMI: 24.04 kg/m2 AC RER: 1.19 ± 0.11 LC RER: 1.21 ± 0.09
Study protocol
	AC or LC tested first: not reported Time between tests: not reported AC Start Watt: not reported LC Start Watt: not reported AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 2
Notes
Shiomi, T. et al. (2000)
Title	Physiological responses and mechanical efficiency during different types of ergometric exercises.
Methods	Cross-sectional study
Participant status	Seven healthy male. No subjects were performing regular exercise and none had orthopaedic diseases. The subjects performed the tests at least two hours after the last meal and did not exercise before the tests. Group n = 7 Gender (M/F): only males Age: 32.1 (27-36) years Bodyweight: 64.1 (54-73) kg Height: 169.9 (163-180) cm BMI: 22.21 kg/m2
Study protocol
	AC or LC tested first: not reported Time between tests: not reported AC Start Watt: 20 LC Start Watt: 20 AC Watt increase/min: 20 W/min LC Watt increase: 20 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 4
Notes	The study reports a correlation coefficient between AC and LC on r = 0.78.
Sporer, B. C. et al. (2007)
Title	Entrainment of breathing in cyclists and non-cyclists during arm and leg exercise
Methods	Cross-sectional study
Participant status	Eight control subjects. Group n = 8 Gender (M/F): only males Age: 26 ± 5 years years Bodyweight: 82.8 ± 8.1 kg Height: 184.5 ± 6.5 cm BMI: 24.32 kg/m2
Study protocol
	AC or LC tested first: Random Time between tests: 60 minutes AC Start Watt: 15 LC Start Watt: 30 AC Watt increase/min: 15 W/min LC Watt increase: 30 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 5
Notes
Swensen, T. C., & Howley, E. T. (1993)
Title	Effect of one- and two-leg training on arm and two-leg maximum aerobic power
Methods	RCT
Participant status	21 untrained college-age men participated. They had a mean age of 22.8 (19-32) years and a mean bodyweight of 73 (45-91) kg. Group “control” n = 5, “two leg” n = 7, “one leg” n = 9 Age: not reported Gender (M/F): only males AC RER: “control” 1.16 ± 0.02, “two leg” 1.13 ± 0.07, “one leg” 1.16 ± 0.06 LC RER: “control” 1.16 ± 0.04, “two leg” 1.18 ± 0.07, “one leg” 1.22 ± 0.05
Study protocol
	AC or LC tested first: LC Time between tests: at least 20 minutes AC Start Watt: 70 LC Start Watt: 120 AC Watt increase/min: 10 W/min LC Watt increase: 15 W/min
Outcomes of interest	VO2max and RER
Risk of bias	SumQAT: 5
Notes
Turner, D. L. et al. (1997)
Title	Effects of endurance training on oxidative capacity and structural composition of human arm and leg muscles.
Methods	Trial
Participant status	Six healthy male subjects. None of the subjects trained systematically although they were recreationally active. Group n = 6 Gender (M/F): only males Age: 23 ± 1 years Bodyweight: 75 ± 2 kg Height: 179 ± 1 cm BMI: 23.41 kg/m2
Study protocol
	AC or LC tested first: not reported Time between tests: not reported AC Start Watt: 54 LC Start Watt: 50 AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max
Risk of bias	SumQAT: 6
Notes
Warren, G. L. et al. (1990)
Title	Is gender difference in peak VO2 greater for arm than leg exercise?
Methods	Cross sectional study
Participant status	The untrained subjects had not participated in any form of regular aerobic exercise or strength training three months prior to the study. Group n = 10 Gender (M/F): only females Age: 25.1 ± 2.8 years Bodyweight: 54.3 ± 7.1 kg Height: 161 ± 4 cm BMI: 20.95 kg/m2
Study protocol
	AC or LC tested first: Random Time between tests: two days AC Start Watt: 6-12 LC Start Watt: 30-47 AC Watt increase/min: not reported LC Watt increase: not reported
Outcomes of interest	VO2max
Risk of bias	SumQAT: 5
Notes
Yasuda, N., et al. (2008)
Title	No gender-specific differences in mechanical efficiency during arm or leg exercise relative to ventilatory threshold
Methods	Cross sectional study
Participant status	The women had a mean age of 23.4 ± 3.6 years, mean BMI 22.8 ± 2.1 units, a mean height of 167.3 ± 6.2 cm and a mean bodyweight of 63.6 ± 5.5 kg. Group n = 9 Gender (M/F): only females Age: 23.4 ± 3.6 years Bodyweight: 63.6 ± 5.5 kg Height: 167.3 ± 6.2 cm BMI: 22.28 ± 2.1 kg/m2 AC RER: 1.08 ± 0.05 LC RER: 1.12 ± 0.06
Study protocol
	AC or LC tested first: Random Time between tests: Within 14 days AC Start Watt: 20 LC Start Watt: 97 AC Watt increase/min: 5 W/min LC Watt increase: 18 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 5
Notes
Yasuda, N., et al. (2006)
Title	Substrate oxidation during incremental arm and leg exercise in men and women matched for ventilator threshold
Methods	Cross sectional study
Participant status	The subject performed low intensity exercises, such as running og cycling, for less than 1 hour per week for a maximum of 4 times per week. Group n = 10 Gender (M/F): 0/10 Age: 23.4 ± 3.4 years Bodyweight: 62.5 ± 6.2 kg Height: 166 ± 7 cm BMI: 22.68 kg/m2 AC RER: 1.10 ± 0.07 LC RER: 1.12 ± 0.06
Study protocol
	AC or LC tested first: Random Time between tests: Within 7 days AC Start Watt: 20 LC Start Watt: 97 AC Watt increase/min: 5 W/min LC Watt increase: 18 W/min
Outcomes of interest	VO2max
Risk of bias	SumQAT: 5
Notes

APPENDIX 3.

Detailed risk of bias assessment