Under‐recognition of leg dystonia in people with cerebral palsy

Abstract

Objective

To determine the rates of clinical under‐documentation of leg dystonia in people with cerebral palsy (CP).

Methods

In this prospective cohort study, we identified independently ambulatory people age 10–20 years with CP‐associated spasticity seen in a tertiary care CP center between 1/1/20 and 11/4/21. Three pediatric movement disorders specialists assessed gait videos from these visits for leg dystonia using the Global Dystonia Rating Scale. We compared the gold standard expert consensus assessment for each patient with the clinical documentation of dystonia during a contemporaneous CP center clinic visit and also with dystonia documentation longitudinally in their medical record.

Results

Of 116 people with CP‐associated spasticity assessed in this study, 70 were found to have leg dystonia in their gait videos. Only 13% of these 70 individuals (n = 9/70) had leg dystonia documented in their contemporaneous CP center clinic visit, even though they were assessed during this visit by clinicians well trained in CP and dystonia assessment. Even with repeated assessment, only 54% (n = 38/70) of these individuals had leg dystonia documented in their medical record.

Conclusions

Leg dystonia is clinically under‐documented in people with CP‐associated spasticity, even when these people are evaluated by well‐trained clinicians. Longitudinal evaluation and vigilance for leg dystonia is critical to address this diagnostic gap.

Introduction

Dystonia is notoriously underdiagnosed in people with cerebral palsy (CP), with years‐long delays to eventual clinical diagnosis. ¹ , ² This is problematic because of the potential time sensitivity of dystonia treatment. For example, the responsiveness of CP‐related dystonia deep brain stimulation is proportional to the duration of a person's dystonia. ³ Dystonia also responds best to targeted pharmacologic and surgical treatments, requiring its differentiation from other motor phenotypes in people with CP. ⁴ Finally, dystonia is a relative contraindication for some surgical treatments like selective dorsal rhizotomy. ⁵ Therefore, the accurate diagnosis of dystonia in people with CP is critical for appropriate management.

When systematically assessed for research purposes, dystonia has been commonly identified in people with CP. Up to 62% of people with spasticity‐predominant CP have dystonia in their upper extremities and up to 80% have dystonia in their lower extremities. ⁶ However, the rates at which upper or lower extremity dystonia is clinically identified in a single clinic visit or longitudinally in people with spastic CP is less clear. Furthermore, it is unclear whether underdiagnosis of dystonia in people with spastic CP occurs even when these people are assessed by clinicians well trained in dystonia diagnosis.

We had three pediatric movement disorders experts systematically assess gait videos of people with CP‐associated spasticity using a validated dystonia rating scale. We then compared these assessments to their own clinical documentation (and that of their colleagues) at a large tertiary care CP center during clinic visits contemporaneous with video capture. We hypothesized that dystonia may be missed in any given clinic visit, even when people with CP are assessed by clinicians well trained in dystonia diagnosis.

Methods

This study was approved by the Washington University in St. Louis Institutional Review Board (IRB) (IRB ID# 202102101, 6/15/21).

Participants

Participants were recruited from those seen in the St. Louis Children's Hospital Cerebral Palsy Center between 1‐1‐2020 and 11‐4‐2021. Videos of gait are recorded at the time of clinic visits for all independently ambulatory people with CP seen in clinic using an Apple iPad A1432. Participants were included if they: (1) were aged 10–20 years at the time of CP clinic evaluation (to ensure that gait pattern had matured ⁷ , ⁸ and that gross motor function had stabilized ⁹ ); (2) carried a CP diagnosis as confirmed by a CP center medical provider and as per the 2006 consensus diagnostic criteria ¹⁰ ); (3) had spasticity as confirmed by a CP center medical provider; and (4) were independently ambulatory to allow for assessment of all participants across a common functionally relevant task (gait). The exclusion criterion was based on gait video quality: if gait videos did not have a clear visualization of the torso through feet recorded as a part of the clinic evaluation, the participant was excluded.

Expert review

To allow for anonymous review, videos were labeled with a random letter and number ID and faces were blurred. These anonymized gait videos were independently reviewed by three fellowship‐trained movement disorders neurologists using the 10‐point Likert‐style Global Dystonia Severity Rating Scale (GDRS) to grade dystonia severity only in the legs (0—no dystonia, 1—mild dystonia, 10—severe dystonia). ¹¹ Arm dystonia was not rated during the gait task as some subjects used handheld assistive devices like canes and walkers, while others did not, thus confounding upper extremity dystonia assessment. After videos were rated, subjects were placed into three categories: those without leg dystonia (summed leg GDRS subscore of 0 to <1 on average across all three expert raters), those with mild leg dystonia (GDRS of 1 to <4), and those with at least moderate leg dystonia (GDRS ≥ 4).

Abstraction of dystonia documentation from the medical record

All gait videos were recorded during a CP center clinical visit. Given this, we could assess CP center clinical documentation of dystonia at the same time of each gait video. Documentation from six clinicians was assessed (two nurse practitioners under the supervision of a pediatric movement disorders specialist, and four pediatric movement disorders specialists, which included the three experts who assessed gait videos for dystonia). We noted whether dystonia was marked as present and the body region in which it was identified (legs, arms, or other). We also noted when dystonia was mentioned as being present anywhere in the medical record, if at all. That is, in addition to determining whether dystonia was identified in the clinical documentation on the day of gait video capture, we also determined whether dystonia had ever been identified before or after the day of gait video capture as a part of the subject's longitudinal clinical evaluation.

Statistical analysis

Chi‐squared tests were used to assess whether rates of clinical documentation of dystonia differed between practitioner types (physician vs. nurse practitioner) across different dystonia severity categories (none, mild, moderate). McNemar's test was used to determine whether rates of clinical dystonia documentation differed from rates of dystonia identified by expert consensus video review. Subject demographics were compared between different dystonia severity categories using t‐tests, comparison of proportion tests, one‐way analysis of variance, and χ ² tests, as appropriate. Statistical analyses were performed using GraphPad Prism (version 8; GraphPad Software). Statistical significance levels were set a priori at p < 0.05.

Results

Participants

One hundred and sixteen participants were ultimately included in the study (180 participants met inclusion criteria with 64 excluded due to poor gait video quality). Participant demographics are in Table 1, separated by dystonia severity category. There was no significant difference between groups with regards to age at the time of evaluation, sex, gross motor functional status, or gestational age at the time of birth.

Table 1.

Participant demographics

	Dystonia assessment based on expert consensus review of gait videos
	No dystonia (GDRS 0 to <1) n = 46	Mild dystonia (GDRS 1 to <4) n = 54	Moderate dystonia (GDRS ≥ 4) n = 16	p
Age (years, 95% CI)	14.1, 13.1–15.1	14.2, 13.5–14.8	13.1, 12.0–14.1	0.4a
Sex (male n, %)	23, 50	26, 48	9, 56	0.9b
GMFCS
I (n, %)	21, 46	24, 44	7, 44
II (n, %)	22, 48	25, 46	7, 44	1b
III (n, %)	3, 7	4, 7	2, 13
Gestational age (weeks, 95% CI)	32.1, 30.5–33.7	33.1, 31.6–34.7	32.8, 29.8–35.7	0.7a

Abbreviations: ANOVA, analysis of variance; GDRS, Global Dystonia Rating Scale; GMFCS, gross motor function classification system level (I, can climb stairs without using a railing; II, can climb stairs using the railing; III, can walk using a handheld mobility device).

^a One‐way ANOVA.

^b Chi‐squared test.

Rates of dystonia documentation in the clinical record on the same day of gait video capture

A gait video from a single time point may underestimate the frequency of leg dystonia. Eleven percent (n = 5/46) of people who did not display leg dystonia in their gait video did have leg dystonia identified during their clinic visit on the same day. The converse is also true. Clinicians underestimate, or at least under‐document, leg dystonia in a single clinic visit. People who did definitively have leg dystonia identified in their gait videos by all three expert assessors only had leg dystonia documented as a part of their contemporaneous clinic visit 13% of the time (n = 9/70). The rate of clinical documentation of leg dystonia was lowest for people with mild leg dystonia (7%, n = 4/54) but remained low even for people with moderate leg dystonia (31%, n = 5/16) (Figure 1A). Overall, rates of leg dystonia documentation in a single clinic visit were significantly lower than rates of leg dystonia identified by expert consensus video review (McNemar's test, p < 0.001).

Figure 1.

Clinical documentation of dystonia compared to dystonia observed by expert consensus review of gait videos. (A) Clinical documentation of leg dystonia. (B) Clinical documentation of arm dystonia. (C) Clinical documentation of any dystonia. GDRS, Global Dystonia Rating Scale.

Although we did not assess arm dystonia during expert consensus review of gait videos, we did find that arm dystonia was more frequently documented than was leg dystonia during the assessed single clinic visit. Arm dystonia documentation frequency increased with increasing severity of leg dystonia as identified on gait video review (Figure 1B).

Documentation of any dystonia at all in a single clinic visit occurred for only 44% (n = 31/70) of people who were found by all three experts to have displayed at least some degree of leg dystonia in the gait video acquired on the same day as the visit (Figure 1C). Rates of documentation did increase for people with greater leg dystonia severity: 69% (n = 11/16) of people who displayed moderate leg dystonia in their gait video had documentation of dystonia in some part of their body during the contemporaneous clinic visit.

Of note, there was no significant difference in dystonia documentation rates between physicians and nurse practitioners across dystonia severity groups (Chi‐squared test, p = 0.5).

Rates of dystonia documentation longitudinally in the clinical record

Rates of dystonia documentation increased when considering the entire clinical record, particularly for leg dystonia. Rates of documentation of leg dystonia increased from 13% in a single clinic visit to 54% (n = 38/70) with longitudinal assessment of people who displayed leg dystonia in their gait videos. For those who did not display leg dystonia in the assessed gait video, rates of leg dystonia documentation increased from 11% in a single clinic visit to 50% (n = 23/46) with longitudinal assessment (Figure 1A).

Though increases in rates of arm dystonia documentation were also observed with longitudinal assessment compared to assessment during a single clinic visit, these differences were not as striking as they were for leg dystonia. Rates of arm dystonia documentation increased from 41% in a single clinic visit to 50% (n = 35/70) with longitudinal assessment of people who displayed any leg dystonia in their gait videos. For those in the moderate leg dystonia category, arm dystonia documentation was stable at 63% (n = 10/16) regardless of whether this assessment was done during a single clinic visit or longitudinally (Figure 1B).

Discussion

We demonstrate that assessment of dystonia during a single time point, even by skilled clinicians, likely underestimates the true prevalence of dystonia in people with CP‐associated spasticity. This is true with isolated gait video analysis or clinical evaluation during a single visit and is most striking for leg dystonia, which we found was increasingly identified with repeated assessment. People with moderate leg dystonia, as identified following expert consensus review of gait videos, are most likely to have any dystonia identified with longitudinal clinical assessment (94% have dystonia identified). However, only 69% of those with moderate leg dystonia have any dystonia documented in a single clinic visit. Perhaps most sobering, of people with CP‐associated spasticity who have leg dystonia unanimously identified on their gait videos by three experts, the rate of leg dystonia documentation in a single visit by well‐trained clinicians is 13%.

Potential reasons for under‐documentation of leg dystonia

Our results suggest that arm dystonia may be more apparent during routine clinical assessment than leg dystonia, though this should be confirmed by directly comparing video‐based arm dystonia assessment with clinical documentation. Gait assessments can be difficult to evaluate in real time and may require assessment outside of the clinic room, making real‐time leg dystonia identification difficult. In contrast, upper extremity tasks are easier to perform in a confined clinic space and may result in a bias to observation of upper extremity dystonia. It is also possible that clinicians are not documenting all dystonia, but only the dystonia that they perceive to be functionally limiting. Noting that all of the included subjects also had spasticity, it is possible that clinicians could ascribe leg‐related functional limitations in the person primarily to spasticity or weakness and thus neglect to document the co‐existing leg dystonia. Though this is a possible contributor, the rates of leg dystonia under‐documentation in people with even moderate degrees of leg dystonia observed on gait videos suggests that several of the above factors are at play.

We also note that, when evaluated on a research basis, leg dystonia was found in 80% of people with CP‐associated spasticity. ⁶ However, longitudinal clinical documentation of leg dystonia in this cohort was lower at 53% (n = 61/116, Figure 1A). If we include all people who had leg dystonia identified in their gait videos (n = 70) and those who had leg dystonia identified with longitudinal clinical evaluation who did not have leg dystonia identified in their single gait video assessment (n = 23), perhaps a more accurate estimation of the presence of leg dystonia in this cohort of people with CP‐associated spasticity is 80% (n = 93/116), matching the research‐based estimation. This clinical diagnostic gap should be a key area for further investigation. Namely, it will be important to identify whether this diagnostic gap represents people with functionally limiting dystonia who go undiagnosed even when they have access to clinicians well trained in dystonia diagnosis. Utilization of function‐focused dystonia scales like the Dyskinetic Cerebral Palsy Functional Impact Scale can help inform this effort. ¹²

Strategies to mitigate leg dystonia under‐documentation

Though lack of clinician expertise is sometimes blamed for dystonia under‐recognition, ² , ¹³ , ¹⁴ our results suggest that is not the sole, or perhaps even the primary, contributor to under‐recognition of dystonia in people with CP. Efforts to mitigate leg dystonia under‐documentation could include: (1) increased awareness of the need to observe for leg dystonia during a routine clinical evaluation (e.g., looking for dystonic overflow activation of the legs that may be elicited during upper extremity tasks), (2) a dedicated notes template prompting the clinician to explicitly document the presence or absence of leg dystonia during the exam, including during gait evaluation, and (3) development of objective dystonia biomarkers to facilitate clinical dystonia diagnosis and severity assessments. Quality improvement initiatives can help with such under‐documentation concerns.

Perhaps most importantly, clinicians need to recognize that the lack of observation of dystonia in a single clinic visit could still mean that the person being evaluated has dystonia. Remaining vigilant for dystonia in every CP clinical evaluation will be critical for mitigating dystonia under‐documentation.

Limitations

This was a single‐center study inherently limiting generalizability. However, we also note that this is a large center with a multistate catchment area that cares for over 1000 unique people with CP. Furthermore, because this study was conducted in a large‐center tertiary‐care surgical referral center, the spectrum of CP seen at this center may not reflect the spectrum of CP seen at smaller centers. We hypothesize that the dystonia diagnosis results presented here are likely to be mirrored at other large centers with potentially higher rates of dystonia under‐documentation in centers without access to CP‐focused clinicians.

We did not assess how orthopedic or neurosurgical interventions affected dystonia diagnosis. It is possible that certain procedures, like a selective dorsal rhizotomy, may make dystonia more or less challenging to assess. This should be an area of future study.

We assessed only independently ambulatory individuals. Dystonia is more common in people with greater gross motor functional impairment. Future studies should include people across all functional abilities and compare rates of clinical dystonia documentation to expert video review.

We did not account for how long each individual was followed in the CP center when assessing rates of dystonia identification by longitudinal assessment. It is possible that rates of dystonia documentation may be higher the longer each child was evaluated in our center. However, this does not diminish the main call‐to‐action of these results: that clinical dystonia identification, particularly in the legs, requires longitudinal assessment and cannot be definitively determined with a single clinical evaluation alone.

We used expert consensus‐based gait video assessments as a putative gold standard for dystonia diagnosis. However, even this assessment, in isolation, missed dystonia that was identified clinically. This highlights not only the importance of longitudinal assessment for dystonia identification, but also that dystonia under‐documentation may be more striking than what we have reported here.

Conclusions

Leg dystonia is under‐documented in single clinic visits in people with CP‐associated spasticity, even when these people are evaluated by well‐trained clinicians. Longitudinal evaluation and vigilance for leg dystonia is critical to address this diagnostic gap.

Author Contributions

Bhooma Aravamuthan: Conceptualization; data curation; formal analysis; funding acquisition; methodology; project administration; resources; writing—original draft; writing—review and editing. Toni S. Pearson: Formal analysis; writing—review and editing. Keerthana Chintalapati: Formal analysis; writing—review and editing. Keisuke Ueda: Formal analysis; writing—review and editing.

Data Availability Statement

Data from this study will be made available to investigators upon request.