Validation and Distribution Characteristics of the Grading of Lateropulsion in Patients with Infratentorial Stroke: A Preliminary Study Authors: Tatsuya Igarashi, Shingo Hirano, Taisei Inoue, Shogo Fukumoto, Mamoru Igawa, Reika Yoshida, Toshiya Nezu, Osuke Shibata, Shota Hayashi, Kozo Ueta, Tsubasa Mitsutake Published: 2026 ABSTRACT ABSTRACTObjectives:Body lateropulsion (BL) is a postural orientation disorder after stroke, characterized by an involuntary tilt of the body to one side. We aimed to evaluate the validity of the Grading of Lateropulsion (GoL)—a measure of BL severity—by examining its associations with balance, ataxia, and walking ability in patients with infratentorial stroke. We also assessed the distribution characteristics of GoL scores.Methods:We evaluated 21 patients with infratentorial stroke who presented with BL using the Berg Balance Scale (BBS), subjective visual vertical, Scale for the Assessment and Rating of Ataxia (SARA), and Functional Ambulation Category (FAC). We then analyzed the correlations between these measures and GoL. We also calculated the proportions of patients with the lowest and highest GoL scores. In patients who completed the 1-month follow-up, FAC scores were compared between baseline and follow-up.Results:Significant correlations with GoL were observed for BBS, SARA, and FAC. GoL Grade 1 was observed in 57.1% of the patients and Grade 4 in 28.6%, suggesting the presence of both floor and ceiling effects. The median FAC score increased from 4 at baseline to 5 at follow-up, demonstrating a significant improvement.Conclusions:GoL severity was strongly associated with ataxia, balance function, and walking independence, indicating a complex and multifactorial pathophysiology of BL. However, notable floor and ceiling effects as well as the negatively skewed distribution of GoL scores indicated limited sensitivity for detecting subtle changes in BL severity. INTRODUCTION Body lateropulsion (BL) is a postural orientation disorder after stroke and is characterized by involuntary tilting of the body to one side.1) BL is typically caused by infratentorial lesions (brainstem or cerebellum) and is characterized by a lack of resistance when the nonparalyzed limbs are pushed or when posture is corrected.1,2,3,4) A related phenomenon is lateropulsion, also called pusher behavior, which is defined as actively pushing the body across the midline toward the affected side or resisting weight transfer to the unaffected side.5,6,7) BL occurs mainly after infratentorial stroke, whereas lateropulsion is more often associated with supratentorial lesions.6) Although both present with body tilt, BL lacks the active pushing or resistance that characterizes lateropulsion.6) Distinguishing between these conditions is therefore essential, and our study has focused on the assessment of BL severity. Because BL delays the acquisition of basic motor functions such as standing and walking,8) early clinical assessment and rehabilitation are important. Although BL can lead to profound clinical disability, its prevalence remains extremely low—only 2.0% of consecutively registered stroke patients met the diagnostic criteria over a 1-year period.8) This rarity has hindered the establishment of a consensus regarding its evaluation or treatment. The Grading of Lateropulsion (GoL) scale rates BL based on the degree of head and trunk tilt and imbalance during quiet standing, as well as the presence or absence of falls.1) GoL has been used in many studies to assess BL severity,1,2,3,9,10,11,12,13,14) and it has been correlated with subjective visual vertical (SVV).1,2,15) However, evaluation scales used in clinical practice must be examined for concurrent validity with multiple clinical indicators and whether their score distributions adequately reflect severity.16) Currently, it is difficult to ascertain whether the GoL satisfies these important aspects. BL pathogenesis after infratentorial stroke is complex. BL is thought to be caused by damage to the lateral vestibulospinal tract (LVST) and the dorsal spinocerebellar tract (DSCT).1,15,17,18,19) Damage to the LVST leads to balance disorders because of impaired vestibulospinal reflexes,10,17) whereas damage to the DSCT leads to limb ataxia owing to reduced unconscious proprioceptive input.4,17,19,20,21) Moreover, damage to the uncrossed and crossed vestibulo-ocular pathways passing through the midbrain is also thought to contribute to BL.22) Although regaining walking ability is a major rehabilitation goal for patients with stroke, this is delayed in patients with BL.8) Based on these findings, we hypothesized that GoL-assessed BL severity would be associated with balance function, motor ataxia, SVV, and walking ability. We hypothesized that these associations would have convergent validity with BL assessment captured by GoL. We aimed to examine the clinical usefulness of GoL as an indicator for assessing BL severity in patients with infratentorial stroke. Specifically, we first evaluated GoL validity in relation to other functional indices and described the distribution of GoL scores. Furthermore, we demonstrated the usefulness of GoL by examining time-dependent changes in GoL and walking ability from preliminary longitudinal data. MATERIALS AND METHODS Study Design and Participants This was a multicenter, longitudinal observational study. We included a consecutive sample of patients with stroke admitted to eight medical institutions with acute and convalescent wards from December 2024 to June 2025. Patients were required to meet all of the following inclusion criteria: (1) aged 18 years or older at the time of obtaining informed consent; (2) first-ever onset of either cerebral infarction or intracerebral hemorrhage; (3) presence of a lesion in the infratentorial region; (4) independent ambulation before stroke onset, defined as a Functional Ambulation Category (FAC) score of 4 or greater23); (5) Stroke Impairment Assessment Set (SIAS) lower-limb motor items (hip, knee, and ankle) each scored at least 4,24) indicating preserved voluntary movement comparable to that on the nonparetic side; (6) SIAS lower-limb sensory items (superficial sensation and position sense) each scored at least 2, reflecting normal sensation or only mild subjective paresthesia; (7) Scale for Contraversive Pushing25) subitems for “pushing behavior” and “resistance to correction” both scored 0, indicating absence of lateropulsion; and (8) GoL score of 1 or greater.1) These criteria excluded lateropulsion and ensured the inclusion BL cases. The exclusion criteria were as follows: (1) inability to provide informed consent; (2) history of orthopedic or neurological disorders that could affect balance function; (3) history of peripheral vestibular disorders; (4) markedly impaired cognitive function; and (5) determination by the attending physician or the responsible rehabilitation staff that participation in the study was not feasible. All patients received routine physical therapy, including balance training, basic movement practice, and activities of daily living (ADL) training. The study protocol was approved by the Clinical Research Ethics Committee of Saga University Hospital (approval number: 2024-08-R-08). The study was conducted in accordance with the principles of the Declaration of Helsinki, and written informed consent was obtained from each patient or, when appropriate, from a family member after a full explanation of the study objectives. This study was conducted in compliance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.26) Data Collection Assessments were performed at two time points: baseline (within 1 week of hospital admission) and a preliminary follow-up (for participants still hospitalized 1 month later). All assessments were performed by physical therapists experienced in stroke and BL and familiar with the measurement procedures. Before data collection, standardized implementation manuals were used to harmonize the assessment methods across facilities. At baseline, we evaluated GoL,1) the Berg Balance Scale (BBS) for balance,27) SVV for vestibular function,28,29) the Scale for the Assessment and Rating of Ataxia (SARA) for ataxia,30) and the FAC for ambulatory independence.23) Grading of Lateropulsion Table 1 presents the criteria for determining GoL based on the report by Dieterich et al.1) GoL was assessed based on the degree of head and trunk tilt and the severity of imbalance during quiet standing, as well as the presence or absence of falls.1) Although the original description of the GoL provides only the criteria presented in Table 1,1) we implemented additional standardized procedures to enable a more quantitative assessment of its severity. Specifically, participants were instructed to stand with their arms crossed in front of their chest, without prior practice, and the assessment was performed once for a maximum duration of 30 s. Observations were made in the frontal plane, and both video and photographic recordings were obtained for documentation and analysis. GoL scoring was primarily determined by direct observation by the examiner, although both video and photographic recordings were reviewed in subsequent analysis as necessary. These standardized procedures were established by consensus among physical therapists who routinely manage BL in clinical practice. Using the same methodology as in the present study, we conducted a preliminary investigation of inter-rater reliability of GoL. Two physical therapists (with 15 and 7 years of clinical experience) who were well trained in the evaluation procedure independently assessed six patients with BL. The results, analyzed using the intraclass correlation coefficient (ICC), showed perfect agreement across all assessments, with ICC (3,1) = 1.00. GoL has been reported to be related to otolith function.1,2) Table 1. Grading of Lateropulsion: grades and diagnostic criteriaGradeDiagnostic criteriaGrade 1Moderate head and body tilt without considerable imbalanceGrade 2Head and body tilt with considerable imbalance, but without fallsGrade 3Head and body tilt and falls with eyes closedGrade 4Head and body tilt and falls with eyes openFrom Dieterich et al.1) Grading of Lateropulsion: grades and diagnostic criteria From Dieterich et al.1) Berg Balance Scale The BBS is a 14-item scale designed to assess both static and dynamic balance.27) It evaluates tasks such as standing and sitting without support, reaching forward, and placing an alternate foot on a stool. Each item is scored from 0 (unable to perform or requiring assistance) to 4 (normal performance), yielding a total score in the range of 0–56. Extensive psychometric evidence demonstrates that the BBS is a valid and reliable measure of balance.31) In addition, according to an international consensus, BBS is a standard tool for evaluating balance in patients with stroke.32) Subjective Visual Vertical The SVV test is performed in a dark room to measure the difference between an individual’s perceived vertical orientation and the true gravitational vertical.28) This assessment serves two main purposes: (1) to detect asymmetry in vestibular function involving the otolithic organs, and (2) evaluate the integrity of central pathways responsible for the perception of gravity. Because of site-specific constraints, both the conventional bucket test and the mobile virtual reality SVV measurement system (MVR-SVV) were used in this study; both methods have previously demonstrated high reliability and validity.28,33,34) We have also investigated the reliability of both measurement methods in 12 patients with acute stroke. To examine the level of agreement between methods and raters, two physical therapists (with 6 and 4 years of clinical experience), trained in the evaluation procedures, independently assessed each patient once using both the bucket test and the MVR-SVV system to measure SVV. The correlation between the two measurement methods exceeded r = 0.8 for both raters, and the inter-rater agreement, assessed using ICC (2,1), was greater than 0.7, indicating a satisfactory level of consistency. Before testing, the absence of ocular motor disorders was confirmed. During assessment, participants were seated in a chair with a back support that was adjusted to their height, and their neck was stabilized with a cervical collar to minimize compensatory head tilt.35) Each patient completed six trials, and the mean value was used for analysis.36) Two SVV parameters were calculated: (1) Absolute SVV, defined as the absolute deviation from the true vertical in each trial,1) and (2) Standardized SVV, where deviations toward the BL tilt side were assigned positive values and deviations toward the nontilt side were assigned negative values.37) Scale for Assessment and Rating of Ataxia SARA is a clinical tool for quantifying the severity of cerebellar ataxia.30) It comprises eight subtests and can be completed within 15 min, making it practical for routine clinical use. Scores ranged from 0 (no ataxia) to 40 (most severe).30) In patients with stroke, SARA has demonstrated excellent reliability and validity as a measure of cerebellar ataxia severity. SARA was used because patients with BL frequently present with lesion-specific ataxia, which may substantially influence BL severity and functional outcomes. Functional Ambulation Category The FAC was used to evaluate the degree of independent walking ability.23) This six-point ordinal scale (0–5) rates the level of assistance required for tasks such as walking and stair negotiation, with higher scores reflecting greater independence. Its reliability and validity have been well established in individuals after stroke.38) The FAC was assessed at baseline and during the preliminary follow-up evaluation. The following demographic and clinical variables were collected from patient medical records: age, sex, height, weight, type of stroke, lesion side, lesion location, and presence of spontaneous nystagmus. Lesion locations involving multiple sites were counted in duplicate when applicable. Statistical Analysis Descriptive statistics were used to report the frequency and median with interquartile range (IQR) for each variable. The proportion of participants in each GoL grade at baseline was calculated. In addition, because this study included patients from different clinical settings—medical institutions with acute and convalescent wards—we conducted a sensitivity analysis to assess potential heterogeneity across samples. Specifically, GoL scores in patients admitted to acute and convalescent wards were compared using the Mann–Whitney U test. The percentages of participants with GoL scores of 1 and with the maximum score were also determined; a proportion exceeding 15% was considered to be indicative of floor or ceiling effects.39) In addition, skewness was computed to assess the asymmetry of the distribution, where a value of 0 indicates a symmetric distribution, negative values indicate skewness to the right, and positive values indicate skewness to the left.40) Associations between GoL and other functional measures at baseline—including BBS, SARA, Absolute SVV, Standardized SVV, FAC, and age—were examined using Spearman’s rank correlation. These analyses were conducted to determine whether the assessment of BL severity using GoL demonstrates convergent or discriminant validity with respect to balance function, ataxia, SVV, and gait ability. To assess robustness, a sensitivity analysis was performed in the subgroup of patients with lesions confined to the medulla oblongata. Additional analyses explored correlations between GoL and the SARA subitems. All correlations were assessed using Spearman’s rank method, and the 95% confidence interval (CI) for each coefficient was obtained by internal validation with bootstrap resampling (1000 repetitions). The strength of correlations was interpreted as follows: 0.00–0.25, negligible; 0.26–0.49, weak; 0.50–0.69, moderate; 0.70–0.89, strong; and 0.90–1.00, very strong.41) For participants who completed the follow-up assessment, changes in GoL were visualized using alluvial maps. The Wilcoxon signed-rank test was used to compare FAC scores between baseline and follow-up. In addition, we calculated the standardized response mean (SRM) to evaluate the responsiveness of the GoL. The SRM was computed by dividing the mean change in GoL scores between baseline and follow-up by the standard deviation of that change. Values of 0.5 or greater were interpreted as indicating moderate responsiveness, and those of 0.8 or greater as indicating high responsiveness. The significance level was set at 5%. All statistical analyses were performed using R version 4.4.1. RESULTS Figure 1 presents a flowchart of patient recruitment and monitoring. Among the 21 patients who completed the baseline assessment, 14 were discharged before the follow-up evaluation; therefore, 7 participants were included in the preliminary longitudinal analysis. Flowchart of participant recruitment and monitoring. Baseline assessments were conducted within 1 week of admission, and follow-up assessments were conducted approximately 1 month later. Table 2 summarizes the demographic and clinical characteristics. Most patients had cerebral infarction, and 11 had medullary lesions. BL was directed toward the side of the lesion in 13 patients and toward the contralateral side in 8 patients. Overall balance impairment was of moderate severity. Table 2. Patient demographics and clinical characteristicsCharacteristicAge, years62.0 (53.0–75.0)Female7 (33.3)Height, m1.6 (1.6–1.7)Weight, kg65.8 (56.0–70.2)Cerebral infarction19 (90.5)Onset to baseline assessment, days5.0 (1.5–26.0)Lesion side Right6 Left14 Bilateral1Lesion location Midbrain2 Pons7 Medulla oblongata11 Cerebellum5Spontaneous nystagmus4 (19.0)Ipsilateral body lateropulsion13 (61.9)FAC 02 12 23 34 410 50BBS39.0 (30.5–53.5)SARA6.0 (2.8–12.5)Standardized SVV, degrees3.3 (−0.9 to 7.0)Absolute SVV, degrees5.6 (3.3–7.5)Data presented as median (IQR), number (percentage), or number.If a lesion involved more than one anatomical location, each affected region is counted. Descriptive statistics for SVV are calculated from six measurements. Standardized SVV is normalized with the ipsilateral side of the body lateropulsion as positive and the contralateral side as negative (in cases with bilateral lesions, the positive direction is the same side as the body lateropulsion). Absolute SVV is calculated from the absolute value of each SVV. Patient demographics and clinical characteristics Data presented as median (IQR), number (percentage), or number. If a lesion involved more than one anatomical location, each affected region is counted. Descriptive statistics for SVV are calculated from six measurements. Standardized SVV is normalized with the ipsilateral side of the body lateropulsion as positive and the contralateral side as negative (in cases with bilateral lesions, the positive direction is the same side as the body lateropulsion). Absolute SVV is calculated from the absolute value of each SVV. In a sensitivity analysis conducted to assess sample homogeneity, the median GoL score was 1.0 (IQR: 1.0–3.0) in the acute-care setting (n = 11) and 1.5 (1.0–4.0) in the convalescent setting (n = 10), with no significant difference observed between the groups (P = 0.452). Furthermore, when the acute-care and convalescent settings were analyzed independently, the respective proportions of Grade 1 and Grade 4 exceeded 15% in each setting, indicating the presence of floor and ceiling effects. The distribution of baseline GoL grades is shown in Fig. 2. Grade 1 accounted for 57.1% and Grade 4 accounted for 28.6%, indicating the presence of both floor and ceiling effects. The skewness of the distribution was 2.075, indicating a negative-direction skew. Regarding lesion location, all six patients classified as Grade 4 at baseline had medullary lesions. Distribution of grading of lateropulsion at baseline (within 1 week of admission). SE, standard error. Figure 3 shows the associations between GoL and other functional measures. Strong correlations were observed with balance, SARA, and FAC, whereas no significant correlations were found with the other variables. Table 3 presents the results of the sensitivity analysis limited to patients with medullary lesions. Similar to the full cohort, strong correlations of GoL grading were found with balance, SARA, and FAC, and no significant associations were observed with the other measures. Table 4 presents the descriptive statistics for the SARA subitems, and Table 5 summarizes their correlations with GoL. Performance-based subitems such as gait, standing, and sitting correlated strongly with GoL, whereas items assessing limb coordination showed no correlation. Associations of lateropulsion grading with balance (BBS), age, SVV, gait (FAC), and ataxia (SARA) at baseline. Each plot shows Spearman’s rank correlation coefficient (ρ), P value, and 95% CI. 95% confidence intervals are estimated by the bootstrap method based on 1,000 resamplings. Table 3. Sensitivity analysis using correlation analysis that included only cases with lesions in medulla oblongata (n = 11)Assessmentρ95% CIP valueBBS−0.85−0.96 to −0.520.001SARA0.750.20 to 0.970.008Age0.33−0.53 to 0.860.326Absolute SVV0.21−0.52 to 0.800.534Standardized SVV0.08−0.64 to 0.780.815FAC−0.73−0.90 to −0.160.012Spearman's rank correlation coefficient (ρ) is given for grading of lateropulsion. The significance level is 5%. The 95% confidence intervals are estimated by the bootstrap method based on 1,000 resamplings. Sensitivity analysis using correlation analysis that included only cases with lesions in medulla oblongata (n = 11) Spearman's rank correlation coefficient (ρ) is given for grading of lateropulsion. The significance level is 5%. The 95% confidence intervals are estimated by the bootstrap method based on 1,000 resamplings. Table 4. Descriptive statistics of SARA subitemsSubitem of SARAMedian (IQR)1Gait3.0 (2.0–6.0)2Stance2.0 (0.0–3.0)3Sitting0.0 (0.0–0.0)4Speech disturbance0.0 (0.0–0.5)5Finger chase0.0 (0.0–0.8)6Nose–finger test0.5 (0.0–1.0)7Fast alternating hand movement0.0 (0.0–0.5)8Heel–shin slide0.5 (0.0–1.0) Descriptive statistics of SARA subitems Table 5. Correlation analysis between grading of lateropulsion and subitems of SARASubitem of SARAρ95% CIP value1Gait0.500.07 to 0.810.0492Stance0.720.44 to 0.890.0023Sitting0.750.51 to 1.000.0014Speech disturbance0.160.28 to 0.720.5495Finger chase0.27−0.23 to 0.740.3166Nose–finger test0.30−0.23 to 0.760.2637Fast alternating hand movement0.09−0.36 to 0.660.7548Heel–shin slide0.13−0.42 to 0.660.619The Spearman's rank correlation coefficient (ρ) is given for grading of lateropulsion.The significance level is 5%. The 95% confidence intervals are estimated using the bootstrap method based on 1,000 resamplings. Correlation analysis between grading of lateropulsion and subitems of SARA The Spearman's rank correlation coefficient (ρ) is given for grading of lateropulsion. The significance level is 5%. The 95% confidence intervals are estimated using the bootstrap method based on 1,000 resamplings. Figure 4 shows changes in GoL among participants who completed the follow-up assessment in the preliminary investigation. Overall, 57.1% showed no change, 42.9% improved, and no patients showed deterioration. When changes in GoL grades were examined according to lesion site, among the four patients whose grades remained unchanged, two had pontine lesions and two had medullary lesions. In contrast, among the three patients who demonstrated improvement, one had a midbrain lesion and two had medullary lesions. Figure 5 compares FAC scores between baseline and follow-up. The median FAC score increased from 4 at baseline to 5 at follow-up, showing a significant improvement (P = 0.026). The SRM for GoL was 0.64, indicating moderate reactivity. Changes in lateropulsion grading between baseline and follow-up shown by alluvial maps. Only participants who completed follow-up are included (n = 7). Baseline assessments were conducted within 1 week of admission, and follow-up assessments were conducted approximately 1 month later. Comparison of walking independence between baseline and follow-up. A Wilcoxon signed-rank test is used to compare FAC scores at baseline and follow-up, with the significance level set at 5%. DISCUSSION This study investigated the clinical utility of the GoL for assessing BL severity in patients with infratentorial stroke by validating the GoL score against multiple functional indices and examining GoL distribution. The results demonstrated that GoL was strongly correlated with balance function, cerebellar ataxia, and walking independence, suggesting that it reflects key aspects of the pathophysiology of BL. Sensitivity analysis confirmed the robustness of this association, and preliminary longitudinal assessment highlighted the close relationship between BL and walking independence. However, the GoL score distribution exhibited both floor and ceiling effects with a negatively skewed pattern. Over time, the floor effect became more pronounced, indicating limited sensitivity of GoL to subtle changes in BL severity. Although the current GoL demonstrates a certain degree of responsiveness for capturing changes in BL severity, the presence of pronounced floor and ceiling effects suggests that the grading of patients with mild or severe BL remains relatively coarse, thereby highlighting inherent limitations of the scale. Infratentorial strokes account for approximately 20% of all strokes,42) yet they require diagnostic and management approaches that differ from those for supratentorial strokes and are frequently associated with poorer functional outcomes.43) Impairments such as cerebellar ataxia, vestibular dysfunction, and balance disorders commonly necessitate rehabilitation. BL is likewise a serious manifestation of infratentorial stroke and has been shown to exert a profound negative impact on functional recovery.8) However, BL is an atypical neurological disorder often overlooked unless clinicians specifically recognize it as a neurological sign.44) Given this background, research and discussion on the clinical course and detailed pathogenesis of BL remains insufficient.8) These findings suggest that patients with medullary lesions tend to exhibit more severe BL, whereas no consistent pattern is identified between BL improvement and lesion characteristics, highlighting the need for further investigation. The results underscore the necessity of future studies incorporating more detailed anatomical analyses, such as advanced neuroimaging and tractography, to better elucidate these relationships. This study revealed that GoL exhibited pronounced floor and ceiling effects. Such distributional bias may limit clinicians’ ability to monitor subtle recovery trajectories or evaluate the efficacy of targeted rehabilitation interventions. Specifically, the pronounced clustering at the extremes of severity constrains the scale’s responsiveness and diminishes its ability to detect subtle changes, particularly among patients in the mild and severe categories. GoL primarily relies on observing head and trunk tilt in the frontal plane; however, case-level reports show that parameters such as center of pressure,9,10,11,12,13) muscle activity,11) and motion analysis19) can detect finer changes in BL. Although BL was initially reported mainly in lateral medullary lesions, recent studies have shown that it can also occur in other infratentorial lesions.8) Because the contributions of the LVST, DSCT, and vestibulo-ocular pathways differ by lesion location, a more precise level of observation and interpretation is required to capture the complex pathophysiology. Therefore, these results highlight the limitations of GoL as an observational assessment and emphasize the need for a more granular tool or for complementing GoL with additional biomechanical assessments. In future refinements of the GoL, increasing the number of categories and further elaborating on the item structure will be necessary to develop a scale capable of capturing the severity of BL with greater granularity. Despite its inherent limitations as a clinical scale, GoL showed significant associations with balance, cerebellar ataxia, and walking independence. This finding supports the notion that BL is not merely a manifestation of isolated vestibular dysfunction but rather reflects a more complex pathophysiology.8,19) Naoi et al.19) further supported that BL may emerge under dynamic conditions such as walking, rather than under purely static conditions. Although SARA is designed to evaluate cerebellar ataxia, it assesses both limb coordination (e.g., heel–shin slide) and antigravity performance tasks (e.g., standing and sitting).30) In our study, correlation analysis between SARA subitems and BL severity revealed no association with joint coordination subitems, whereas all antigravity performance subitems showed significant correlations. These findings suggest that BL extends beyond postural orientation or limb positioning and is more closely linked to dynamic postural control. Preliminary longitudinal analysis further highlighted the relationship between BL and walking performance, underscoring the need to evaluate BL from a dynamic perspective. Another key finding of this study was the absence of a significant association between BL severity and SVV. SVV is regarded as an indicator of asymmetry in ascending vestibular sensory input.28) Dieterich et al. demonstrated a clear relationship between GoL and SVV in patients with lateral medullary infarction,1,2) yet in studies of patients with midbrain infarction, they found that the direction of SVV tilt did not necessarily correspond to the direction of body tilt.45,46) Moreover, a case of BL following pontine infarction has been described in which severe BL was present despite SVV remaining within the normal range.11) In addition, reports on pontine and cerebellar infarction have demonstrated that SVV tilt can occur not only toward the ipsilesional side but also toward the contralesional side.47,48) BL can manifest either ipsilateral or contralateral to the lesion. In addition, nystagmus, ocular movement disorders, and vertiginous symptoms that characteristically arise from lesions in the infratentorial regions may serve as lesion-specific confounding factors. The severity of these disturbances could compromise the validity of GoL. Therefore, it will be important for future research to clarify these relationships according to lesion site and severity, as well as to examine changes over time that may result from vestibular compensation and adaptive processes. However, the characteristics of the study participants may have influenced the results. Specifically, only 9.5% of the participants were classified as GoL Grade 2, and 4.8% as Grade 3, and more than half of the participants had lesions in the cerebellum or pons. Such an uneven grade distribution and lesion-site predominance could have limited the ability to detect potential associations between BL severity and SVV and should be carefully considered when interpreting these findings. This study has some limitations. First, the SVV measurement method was not entirely uniform, because both the conventional bucket test and the MVR-SVV system were used across institutions. Nevertheless, both approaches have demonstrated high reliability and validity,28,33,34) and all examiners—experienced physical therapists—used a standardized manual to ensure consistent procedures. Second, the findings were derived from a relatively small sample size. The markedly low representation of patients with GoL Grades 2 and 3 limited our ability to evaluate the scale’s performance across the full spectrum of severity. This underrepresentation may have resulted in a reduced statistical power in subgroup analyses and attenuation of the observed associations. In the multicenter study reported by Naoi et al.,8) only 2.0% of consecutively enrolled stroke patients met the criteria for BL over an approximate 1-year period. The variation in days from stroke onset to baseline assessment likely reflected data collection at both acute-care and convalescent hospitals. Although clinical research on BL remains limited and most previous reports are single cases or very small case series,9,10,11,12,13,14,15,19,20,44) the present investigation represents a prospectively assembled cohort. Moreover, internal validation with bootstrap resampling and sensitivity analysis restricted to defined lesion sites confirmed the robustness of the results. These findings provide new insights into BL pathophysiology and clarify the limitations of the GoL as an assessment scale. Third, this study introduced additional standardized procedures to facilitate a more quantitative evaluation of the GoL criteria. While preliminary investigations indicated robust reliability of this approach, its validity and reproducibility remain to be fully established. Further validation of the assessment method is required to achieve a clearer distinction of BL. Finally, although we examined the responsiveness of the GoL, it should be noted that the results may vary depending on the timing of the assessments and the interval between measurement points.