This is the first analysis to compare the efficacy of two IMT protocols that used different devices and training programs in the general population of people with chronic tetraplegia. While both training protocols improved PImax after 4 weeks of training, the findings of this secondary analysis do not identify a superior IMT protocol based on ΔPImax or %ΔPImax. However, the findings do identify factors that may impact overall IMT protocol efficiency and show relationships that are clinically important. We found a positive correlation between Work-% performed and %ΔPImax, regardless of the device or protocol used. The Work-based findings are evidence that the Work-% construct may be used when comparing IMT protocols. Further, the relationship between perceived exertion and intensity of training may help clinicians grade IMT training intensity in the absence of facilities to assess inspiratory pressure or PImax.
After 4 weeks training, no group difference was found in %ΔPImax or ΔPImax, despite the large difference in PImax between the F-IMT and T-IMT groups at baseline, where the PImax of the F-IMT group was more than three times that of the T-IMT group. However, the post-intervention PImax, ΔPImax and %ΔPImax, although not different between groups, in the F-IMT group were 1.8, 0.6, and 0.4 times the median values for the T-IMT group, respectively. The lack of significant differences may be due to the low numbers of participants included in the comparison analysis, as well as large inter-participant variability in PImax resulting in an underpowered analysis and possible Type 2 error. Similar to our current findings in SCI, IMT training in people with chronic obstructive pulmonary disease (COPD) increased PImax, but with no difference in ΔPImax after F-IMT compared to T-IMT . Additionally, between group differences in ΔPImax were not found in a study investigating T-IMT vs F-IMT vs no IMT in elite rugby athletes, most of whom had SCI . However, it is unclear whether a significant increase in PImax occurred within groups based on the analyses reported and the small numbers in each group (T-IMT group, n = 4; F-IMT group, n = 5; no IMT control group, n = 7). Therefore, while IMT in either form can increase PImax, further research is needed to identify if there is an optimal mode of training or training protocol for each diagnostic population.
The similar improvements in ΔPImax or %ΔPImax across F-IMT and T-IMT in the current study suggest that supervision of every training session may not be mandatory in every case to obtain positive results. All IMT sessions were supervised in the T-IMT group, while only one session per week was supervised in the F-IMT group. There are financial, transportation, and staffing barriers that may limit the clinical translation of fully supervised training protocols [20, 21]. However, supervision may be required for individuals with hand function impairments if adaptations are not available to allow for independent use of devices. Some form of clinical supervision is likely beneficial even for individuals with the ability to perform IMT independently or with assistance from a carer. Both the original F-IMT and T-IMT studies included follow-up phases where no supervision was provided. During the non-supervised phase, only 1 of 3 active participants (33%) and 16 of 62 participants (26%) continued to train in the F-IMT and T-IMT studies, respectively [4, 7]. No supervision seems to be detrimental to continued training after an exposure to supervised IMT in people with SCI. Similarly, a meta-analysis investigated the impact of supervision on adherence of people with non-neurologic chronic disease during follow-up exercise programs after participants had completed a 4–6 week supervised exercise program . A pooled and weighted analysis of two studies found the proportion of people who were “partially adherent” to a home exercise program without any supervision was low (29%). This reported proportion is very similar to the unsupervised adherence rates reported by the studies included in this secondary analysis [4, 7, 22]. Future studies should continue to investigate supervision models, including remote supervision, that are sustainable and tailored to the individual abilities of people with SCI and clinicians, more importantly so in the era of telehealth consultations and treatments.
Beyond supervision, other training factors differed between the F-IMT and T-IMT groups including a higher number of breaths (#Breaths) taken in the T-IMT group with a lower training intensity. Overall, these differences did not result in significantly different ΔPImax or %ΔPImax outcomes in the current study. Raab et al. have reported on the predictive relationship of training Intensity-% and ΔPImax in a retrospective study of an inpatient cohort (n = 67) with SCI ranging from C4-T12 levels of injury (AIS A-D) . They found that median training Intensity-% and PImax at baseline, but not #Breaths, were predictive of the ΔPImax after a median of 6 weeks (interquartile range of 5–8 weeks) of training . The expected ΔPImax for each group in the current study (based on the Raab et al (2019) equation) indicated that the T-IMT group performed much better than expected (167%) while the F-IMT group performed much worse than expected (26%) . These differences from expected changes in each group raise doubt about the utility of this predictive equation. The equation may be inaccurate when comparing community dwelling individuals with variable injury characteristics, training devices, protocols including number of sessions, intensities of training, duration of training and levels of supervision. Our limited data cannot determine the utility of the equation and variability of sample and training factors is not accounted for in previous meta-analyses [12, 13].
Across all 14 participants in the current study, there was a strong positive correlation between Work-% (calculated from #Breaths × Intensity-%) and %ΔPImax, regardless of the training paradigm. Work was not included in the model by Raab et al.  despite work being a predictor of increased strength in limb resistance training [23, 24]. However, work is not commonly reported in IMT trials [12, 13], and in the current study, we have calculated work relative to baseline PImax across the first 4 weeks of training only. The relationship of Work-% and %ΔPImax is unknown beyond this. Nevertheless, the strong correlation suggests that higher levels of work produce higher %ΔPImax even if the Intensity-% of training is reduced and the #Breaths is increased to compensate. This occurred for the T-IMT group, in which the Intensity of training relative to baseline (Intensity-%) was 30% of that for the F-IMT group. The calculation of work relative to PImax at baseline (Work-%) could allow comparison of the impact of different IMT protocols to a common outcome (%ΔPImax) and may offer a more complete analysis of the effect of IMT on respiratory function in people with SCI. Future studies could report Work-% to improve between-protocol comparison and translation to clinical practice as well as allow for tailored approaches to IMT.
The data from the current study also showed that across both groups, training intensity (both Intensity-% and Intensity-absolute) are positively correlated with RPE, a measure of the effort required to do the training. Similar relationships between effort and pressure have been reported previously in able-bodied people , people with COPD with and without anxiety , and people with chronic tetraplegia . Further, a meta-analysis reported that peak oxygen uptake and peak power output improve when individuals with SCI complete perceptually regulated exercise protocols . Although IMT protocols were not included in that meta-analysis, our findings support that perceptually regulated IMT may be effective at determining the dose for intensity of IMT via RPE scores since a respiratory pressure meter is not always available in clinical settings. Clinicians may be able to prescribe the Intensity-% of IMT based on RPE; for example, from our limited data in Fig. 3, training at an RPE of 5 is likely to represent a training intensity of at least 65% PImax. The relationship between Intensity% and RPE suggests that individuals with tetraplegia can generally perceive the intensity at which they are performing IMT and warrants further investigation to confirm.
The introduction of the Work-% variable, calculated from training Intensity-% and #Breaths, and the recognition of the correlation between perceived exertion and training intensity are clinically important and should be considered in future research. However, this secondary analysis is limited by its small sample size (and low power) and the existing baseline differences in study participant groups. In general, the participants in the F-IMT study were younger and had a shorter injury duration than participants in the T-IMT groups. Older age and longer injury duration are related to poorer respiratory function [29,30,31] which may have contributed to the lower average baseline PImax of participants in the T-IMT study. The higher baseline PImax of the F-IMT group may have resulted in a ceiling effect in ΔPImax.