Aging into and throughout the adult years with cerebral palsy (CP) is associated with accelerated health declines, including loss of function,¹ increased risk of chronic diseases and mental health disorders, and premature mortality.^2-8 Commonly used clinical methods to screen for and monitor risk of disease have limited value in their current form (e.g. ranges of body mass index for obesity status) for adults with CP⁹ because of individuals' different anatomy and physiology (for a review, see Whitney et al.¹⁰). Classification scales to characterize CP, namely the Gross Motor Function Classification System (GMFCS)¹¹ and the Manual Ability Classification System (MACS),¹² may broadly differentiate risk of disease outcomes.^{2, 3} However, the GMFCS and MACS were developed for clinical description and monitoring of high-priority domains for the pediatric population, and may be less useful for higher-priority domains for adults with CP, including individual-level monitoring of disease risk and progression. Thus, research is needed to identify predictors that can differentiate risk (i.e. enhance predictive value) across and within levels of functional abilities such as the GMFCS among adults with CP and adolescents nearing adulthood.

Hand grip strength is a proxy of overall muscle strength/weakness and function, easily obtainable in the clinic setting at low cost, and has excellent predictive value for functional loss and disease outcomes among non-CP cohorts.^13-15 Moreover, inclusion of grip strength in risk prediction models improves prediction of morbidity and mortality beyond established office-based risk scores.¹⁶ Sarcopenia and frailty are a proposed contributing mechanism to early and accelerated loss of function and morbidity onset in adults with CP, making grip strength an attractive option for disease screening in CP.¹⁷ Previous work in children with CP has shown validity of grip strength as a predictor of motor function.^{18, 19} So far, the relation between grip strength and demographics, CP-related classification scales (e.g. the GMFCS, MACS), anthropometrics, and disease risk has not been adequately assessed for adults with CP. Thus, it is unknown whether grip strength can serve as a proxy of function to assist in risk prediction on its own or as part of a risk prediction model with other predictors (e.g. the GMFCS) for adults with CP, especially considering the diversity of distribution and types of motor impairment. More fundamentally, the extent to which grip strength can be reliably and feasibly assessed during a routine clinical visit for adults with CP remains unknown, thus limiting clinical translation.

To inform future work that develops methods for disease screening and monitoring specific to adults with CP, this study focused on grip strength. The first aim was to determine the feasibility and reliability of obtaining grip strength during a routine clinical visit among adults with CP classified in GMFCS levels I to V. The second aim was to determine the association between grip strength and demographic variables, GMFCS, MACS, and anthropometrics. The third aim was to determine whether grip strength on its own or in addition to GMFCS, compared with GMFCS alone, improved prediction of relevant disease outcomes among adults with CP.

METHOD

RESULTS

DISCUSSION

This single-center study suggests that grip strength is a reliable and feasible measure to administer to adults with CP across a range of motor function abilities during a routine clinical visit. In this cohort, grip strength was associated with sex, GMFCS, MACS, and some anthropometrics. Furthermore, adding grip strength to a model with age, sex, and GMFCS improved the model's predictive ability, although slightly, to detect risk of certain chronic diseases for adults with CP.

It was not surprising that males had higher grip strength than females, more involved GMFCS levels had lower grip strength, and that body mass had a positive association with grip strength in this study. MACS was also found to be associated with grip strength, such that more involved MACS levels were associated with significant decreases in grip strength, even after accounting for GMFCS. Interestingly, in model 1 + MACS, GMFCS was no longer associated with grip strength, where the parameter estimates were close to 0 (1.5–2.6) for each GMFCS level compared with GMFCS level I (all p > 0.400) (data not shown). However, caution in interpretation is warranted as the GMFCS level in that model was considered a possible confounder as opposed to a secondary exposure of interest. The variable inflation factor was also at its peak in this model (up to 3.4), but still lower than the commonly used threshold of less than 5.0. Thus, biased parameter estimates due to multicollinearity are unlikely to be strong. Findings for MACS were similar when GMFCS was removed from the model (data not shown), providing evidence that MACS may be a stronger predictor of grip strength than GMFCS, and may be the more important classification system when trying to understand the clinical significance of grip strength in adults with CP.

Waist circumference was also positively associated with grip strength but may present multicollinearity in the presence of body mass, which was not assessed in this study. This may be important in certain research and clinic settings that wish to develop risk prediction models, as multicollinearity can bias regression coefficients and the model's performance. The measurement of body mass and waist circumference can be challenging in CP, depending on the clinical presentation and the clinic set-up (presence of a wheelchair scale). Future risk prediction models should incorporate measures that are reliable and widely available so that the models can be used by a broad audience of the intended end users (e.g. primary care physicians for disease risk assessment). This is one reason why hand grip strength is attractive for adults with CP, given the ease of implementing the protocol, recording and interpreting values, and having little effect on the values from the tester (e.g. handheld dynamometer producing back-end force from the tester). It is worth noting that the lower extremities may be more affected in the context of CP. Children with CP have greater deficits in the lower versus upper extremities, and lower extremity strength may better correlate with their gross motor function.²⁶ However, many adults with CP use walkers, crutches, or manual wheelchairs for mobility, all of which require upper extremity strength and function to maintain activities of daily living. Speculation follows that grip strength may be a more proportionally sensitive measure of overall functioning beyond activity alone (e.g. occupational activities, caring for dependents/living area) for adults than for children with CP.

This study did not focus on developing models of risk prediction, but it does provide needed preliminary information to inform such studies. Grip strength enhanced the prediction of salient disease outcomes for adults with CP above the model with age, sex, and GMFCS level. Although there were no statistically significant increases in the predictive ability, future work may wish to opt for a higher predictive potential and then scale back on the number of predictors to avoid model overfitting if needed, which can diminish effective clinical translation of prediction models.²⁷ Our findings help to identify that grip strength may be useful for risk prediction, but how it can be used requires further investigation. For example, cut-off thresholds to differentiate high versus low risk are attractive because of their simplicity. However, the population of adults with CP is heterogenous, such that a single measure and a binary cut-off or value ranges may not adequately capture varied risks in a clinically meaningful way. Thus, risk prediction models incorporating grip strength, as opposed to cut-offs, may be better suited for screening and monitoring disease risk for adults with CP. We did not include distribution and type of movement disorder in the analyses, as there is an absence of evidence that these factors influence or predict health outcomes. If future studies continue to suggest that grip strength is an important prediction measure, it will be interesting to examine these factors.

This study had limitations. First, it was performed at a single medical center and measures and interpretations may not be fully generalizable. Second, it should be noted that this study did not cover the diversity of CP, as it excluded those whose hand function was not sufficient for grip strength measurement (limiting, for example, the involvement of individuals classified in MACS level V) and those with severe cognitive impairment from the perspective of the grip strength measurement methodology. In fact, even though we had good reliability between trials for GMFCS levels IV and V, we had many results of zero and a low variability of results with a majority less than 10. This indicates that grip strength may not be the best measure in this population because of the complexity of their motor and health impairments. Other measures should be considered to replace or supplement grip strength. Third, the disease outcomes were self-reported which is subject to recall bias. Further, self-report disease included history or current presence as opposed to incidence, which is better suited for risk prediction. The modelling used to assess the third aim should therefore be viewed as proof of concept that grip strength may be useful to help differentiate the presence or history of a disease, while future work will need to identify predictive ability for new-onset disease or disease progression. Fourth, other common anthropometrics were not collected, such as height or body mass index. This was intentional as our broad goal was to identify methods, probably in the form of risk prediction models, to identify risk of certain diseases. For risk prediction models to be useful with optimal external validity, a key component is having reliable measures that can be easily obtained across anticipated settings where the model is intended for use. Height is challenging to measure for those with contractures or who use mobility devices. Furthermore, many adults with CP receive care from physicians who are unfamiliar or lack appropriate knowledge and equipment for accurate anthropometric measurements.^{28, 29} Fifth, this study lacked biological factors, such as serum biomarkers, which may help to better understand how grip strength relates to (patho)physiology in CP. In addition, we used the GMFCS as a status assessment of the patient in the clinic, without considering whether their GMFCS level had changed since childhood. However, it has been noted that the GMFCS is very stable into late adolescence and adulthood, so noting changes would probably have only minimal effects on the outcomes.³⁰

CONCLUSION

This study found that grip strength is a feasible and reliable measurement among adults with CP that can be collected during a routine clinical visit. Grip strength is associated with sex, body mass, waist circumference, GMFCS, and especially MACS. Furthermore, grip strength may enhance prediction of salient diseases when added to a model including age, sex, and GMFCS, providing novel information to inform future studies in developing risk prediction models.

FUNDING INFORMATION

Foundation for Physical Medicine and Rehabilitation: E. Molnar-Swafford Pediatric Rehabilitation Grant.