Comparison of two Borg exertion scales for monitoring exercise intensity in able-bodied participants, and those with paraplegia and tetraplegia

The study addresses how best to prescribe relative exercise intensity to ensure homogeneous stimuli across individuals. Conventional prescriptions using %VO₂peak or %HRpeak can yield heterogeneous metabolic stimuli at comparable thresholds and require costly, invasive testing. Metabolic thresholds (e.g., LT₁, LT₂/MLSS/CP) better demarcate intensity domains but also entail methodological and practical challenges. Ratings of perceived exertion (RPE) offer a simple, scalable alternative; prior work suggests RPE at LT₁ is independent of age, sex, training status, and exercise mode, but evidence is largely confined to able-bodied individuals performing lower-body exercise. Further, multiple RPE scales exist—Borg’s 6–20 scale and CR10—with differing scaling properties, and there is limited, lower-body-only evidence comparing them quantitatively. The study’s aims were: (1) to compare RPE from Borg’s 6–20 scale and CR10 in able-bodied individuals during upper and lower body exercise and in individuals with spinal cord injury (SCI; paraplegia and tetraplegia) during upper body exercise; and (2) to determine RPE at LT₁ and LT₂ across these groups and modes. The hypotheses were that the scales would be strongly related and that RPE at LT₁ and LT₂ would be independent of exercise mode and participant group.

The literature highlights limitations of prescribing intensity via %VO₂peak or %HRpeak due to large inter-individual variability at metabolic thresholds, potentially leading to non-uniform training stimuli. Metabolic thresholds (LT₁, LT₂, MLSS, CP) better reflect physiological domains but require invasive assessment and trained personnel, impeding broad implementation. RPE has been proposed as a practical solution with evidence that RPE at LT₁ is consistent across age, sex, training status, and mode in able-bodied treadmill and cycle studies. In SCI populations, both Borg’s 6–20 and CR10 scales have been used to prescribe and regulate exercise; Borg’s 6–20 often yields linear associations with intensity markers, whereas CR10 shows nonlinear growth functions. An original theoretical transformation table exists for mapping between the scales, but robust empirical comparisons are scarce, with only one lower-body study in able-bodied adults using statistical modeling. This gap motivates direct comparison in upper-body exercise and SCI cohorts to enable interchangeable use of RPE scales.

Design: Cross-sectional cohort study approved by institutional ethics committees in the UK and the Netherlands. Participants: N=24 adults divided equally into three groups (n=8 each): able-bodied (AB; recreationally active, untrained in upper-body endurance), paraplegia (PARA; recreationally active), and tetraplegia (TETRA; trained wheelchair rugby athletes). Participant characteristics included sex, age, height, body mass, neurological injury level (PARA: T4–L2; TETRA: C5–C7), AIS grade distribution, and time since injury. Testing sites: AB and TETRA tested in the UK; PARA tested in the Netherlands; same investigators at each respective site. Exercise protocols: AB completed two randomized sessions separated by 2–7 days: cycle ergometry (AB-CYC) and handcycle ergometry (AB-HC). PARA completed handcycling; TETRA completed wheelchair propulsion on a motorized treadmill. Equipment: Cyclus 2 ergometer for cycle/handcycle; motorized treadmill for wheelchair propulsion. AB used standardized cycle and adjustable handcycle rigs; PARA and TETRA used personal handcycle/wheelchair rugby chairs. Familiarization was not provided for AB upper-body testing, relying on prior evidence for reliable peak responses without familiarization. Graded tests: After a 5-min self-selected warm-up, participants performed continuous 3-min stages with individualized starting workloads and increments: AB-CYC 50 W start with +15 W stages; AB-HC 10 W start with +10 W stages; PARA 20–45 W start with +20 W stages; TETRA 1.2–1.7 m·s⁻¹ start with +0.2 m·s⁻¹ stages. Termination: For AB and TETRA, the initial test ended when blood lactate [BLa] exceeded 4 mmol·L⁻¹ or Borg 6–20 RPE reached 17 (blunted lactate response anticipated in TETRA). PARA continued to volitional exhaustion. A second graded test to exhaustion followed after 15 min rest/low-intensity recovery for AB and TETRA, with starting loads set to the stage where [BLa] first rose ≥0.5 mmol·L⁻¹ above rest; increments: AB-CYC 110–180 W start +15 W·min⁻¹; AB-HC 30–60 W start +10 W·min⁻¹; TETRA 1.3–2.0 m·s⁻¹ start +0.1 m·s⁻¹·min⁻¹. Measurements: Continuous breath-by-breath gas exchange (Metalyzer 3B) and HR (Polar RS400), processed as 30-s rolling averages with peak values identified. RER and VE were recorded. RPE was collected verbally during the final minute of each stage on both Borg 6–20 and CR10 scales, with scale order randomized between participants but fixed within participant; only the active scale was visible to prevent bias. Capillary earlobe blood samples were collected during the final 30 s of each stage for [BLa] measurement (Biosen C-line for AB and TETRA; Lactate Pro 2 for PARA). Outcomes and derived variables: Peak workload (power or speed) calculated from the final completed and partial stages using Peak workload = F + [(t/d) × I] where F is final completed stage workload, t is time spent in the uncompleted stage, d is stage duration, and I is increment. LT₁ identified from the intersection of horizontal and ascending segments in the plot of log-[BLa] vs. log-VO₂. LT₂ defined as [BLa] equal to LT₁ + 1.5 mmol·L⁻¹. VO₂ at LT₁ and LT₂ obtained by inverting the corresponding log-VO₂ values at those points. For each participant, RPE (both scales) was fit against [BLa] using a quadratic model (RPE = a[BLa]² + b[BLa] + c), and the fitted functions were used to estimate RPE at LT₁ and LT₂. Statistical analysis: Normality assessed via Shapiro–Wilk. Model comparisons for mapping Borg 6–20 to CR10 employed individual curve analyses testing linear, quadratic, exponential, and power functions, with F-tests for significance and selection by highest R². Follow-up two-level random-intercept multilevel models (stages nested within participants) were generated per group using the best-fitting (quadratic) function to derive group equations and R². Group differences in peak responses and in VO₂ (absolute, relative, %VO₂peak) and RPE at LT₁ and LT₂ were tested using one-way ANOVA with Bonferroni post hoc corrections. Significance threshold P<0.05; standardized effect sizes classified as trivial to very large.

- Strong mapping between Borg’s 6–20 RPE and CR10 scales across all groups and exercise modes, best described by a quadratic function; group multilevel model R² values: AB-CYC 0.970; AB-HC 0.968; PARA 0.965; TETRA 0.967 (all P<0.005). - Group-specific quadratic equations (x = Borg 6–20; y = CR10): AB-CYC y = 0.023x² + 0.067x − 0.754; AB-HC y = 0.024x² + 0.085x − 1.087; PARA y = 0.019x² + 0.240x − 2.212; TETRA y = 0.015x² + 0.306x − 1.989. - Transformation of Borg values to CR10 produced similar mapped values across groups (e.g., Borg 13 ≈ CR10 4–5; Borg 17 ≈ CR10 7–8; Borg 19–20 ≈ CR10 9–10). - No significant differences in RPE at LT₁ or LT₂ between groups on either scale (Borg: F(3)=0.02, P=0.99 at LT₁; F(3)=0.86, P=0.47 at LT₂. CR10: F(3)=0.36, P=0.78 at LT₁; F(3)=2.34, P=0.10 at LT₂). - Peak exercise responses: AB-CYC exhibited significantly higher absolute and relative VO₂peak than AB-HC, PARA, and TETRA (all P<0.005; ES up to 3.1). TETRA showed markedly lower HRpeak and [BLa]peak than other groups. - At LT₁ and LT₂, absolute and relative VO₂ were significantly greater in AB-CYC versus AB-HC, PARA, and TETRA (all P<0.01; ES typically large). Conversely, %VO₂peak at LT₁ and LT₂ was significantly greater in TETRA than AB-CYC, AB-HC, and PARA (P<0.005; ES ≈ 1.9–3.0), indicating they operated at higher fractions of VO₂peak at threshold. - Despite these physiological differences, perceived exertion at lactate thresholds was consistent across modes and groups, supporting the interchangeability of RPE scales and the potential for RPE to demarcate threshold-based intensity.

The findings directly address the research questions by demonstrating a robust, quadratic relationship between Borg’s 6–20 and CR10 RPE scales that generalizes across lower- and upper-body exercise and across able-bodied and SCI cohorts. This supports interchangeable use of the two scales in research and practice, aided by the provided transformation equations/values. Furthermore, RPE at LT₁ and LT₂ did not differ by group or exercise mode, despite substantial differences in absolute and relative VO₂ at these thresholds and in peak responses, particularly for TETRA. This indicates that subjective exertion around metabolic thresholds is perceived similarly across populations and modalities, reinforcing RPE’s utility for intensity demarcation when direct physiological testing is impractical. However, the observed inter-individual variability cautions against prescribing homogeneous training intensities solely by fixed RPE targets without considering individual responses. For SCI athletes, especially those with tetraplegia who exhibit distinct physiological profiles (e.g., lower [BLa], HRpeak, higher %VO₂peak at thresholds), the scale mapping remains valid, providing a pragmatic tool for monitoring and guiding training loads in upper-body exercise contexts.

This study establishes a strong, generalizable quadratic relationship between Borg’s 6–20 RPE and CR10 scales across exercise modes and populations, including individuals with paraplegia and tetraplegia. RPE at LT₁ and LT₂ were independent of mode and injury level, supporting the use of RPE to identify threshold-related intensities. The derived transformation equations enable interchangeable application of the two scales in practice and research. Nonetheless, considerable inter-individual variability limits the ability to prescribe uniformly homogeneous intensities using RPE alone. Future work should investigate strategies to account for individual variability when using RPE for prescription, validate these findings in larger and more diverse SCI cohorts and upper-body modalities, and explore integration of RPE with simple physiological markers to refine individualized intensity guidance.

Inter-individual variation in RPE responses limits the ability to make firm recommendations for prescribing homogeneous exercise intensity using RPE alone. The study’s cross-sectional design and small group sizes further constrain generalizability. Additionally, different testing sites and analyzers for blood lactate across groups may introduce measurement variability, and familiarization for upper-body ergometry in able-bodied participants was not provided.