Freezing of Gait in Parkinson's Disease: A Heterogeneous Phenomenon Across Assessment Tools

ABSTRACT

Freezing of gait (FOG) is common in Parkinson's disease (PD), yet uncertainty remains regarding the optimal criterion for identifying freezers. Self‐reported FOG and performance‐based FOG assessments are widely used, but their relative value is unclear. This study compared four different tools for identifying freezing and how these classifications relate to simple functional tests feasible in home or clinical environments. The participants with PD completed four FOG‐related assessments (two movement disorder society‐unified Parkinson's disease rating scale [MDS‐UPDRS] items and two observations during Parkinson's disease functional movement battery [PD‐FUNC] items). Based on the outcomes, individuals were classified as freezers (positive on at least two tools), occasional freezers (nonzero score on one indicator), or nonfreezers (zero on all tools). Disease severity was recorded using Hoehn and Yahr (H&Y) staging. Groups were compared using the PD‐FUNC test battery, which includes rotation, dexterity, balance, and locomotion tasks. Analyses were adjusted for disease severity. Additionally, a 5‐year longitudinal case study illustrated fluctuations in FOG‐related and functional performance over time. As expected, freezing was the most prevalent in advanced disease (H&Y Stage 3). However, up to 40% of those classified as freezers were in earlier stages (H&Y Stages 1 and 2). After adjustment for disease severity, only a small subset of PD‐FUNC items differentiated groups. Post hoc analysis of these items showed that significant differences were limited to comparisons between freezers and nonfreezers. Freezers required more time to complete a 360° rotation. Unexpectedly, typing speed, a dexterity measure, also differed significantly between freezers and nonfreezers. Occasional freezers showed higher MDS‐UPDRS Part III scores compared with nonfreezers, whereas Part II scores displayed a graded increase across groups. The longitudinal case study demonstrated recurrent periods of reported freezing and severe falls that coincided with prolonged rotation times, followed by recovery periods linked with return to physical activity. Combining multiple indicators improves the identification of FOG and reveals that freezing is present even in earlier PD stages. Simple PD‐FUNC items, particularly rotation time and typing speed, distinguish freezers from nonfreezers after adjusting for disease severity. The case study highlights that reported freezing may occur in discrete periods, often associated with falls, but recovery is possible and appears related to resumption of physical activity. These findings support the clinical utility of PD‐FUNC for monitoring FOG‐related functional changes over time.

Freezing of gait in Parkinson's disease is a heterogeneous, context‐ and tool‐dependent phenomenon that may emerge in early disease. Capturing freezing requires multimethod assessment, and interpretation of freezing‐related outcomes should take into account disease severity.

Abbreviations

ANCOVA

Analysis of covariance

BMI

Body mass index

FOG

Freezing of gait

H&Y

Hoehn & Yahr

LEDD

Levodopa equivalent daily dose

MDS‐UPDRS

Movement disorder society‐unified Parkinson's disease rating scale

NFOG‐Q

New freezing of gait questionnaire

NOFOG

No freezing of gait

PD‐FUNC

Parkinson's disease functional movement battery

PWP

Persons with Parkinson's disease

SD

Standard deviation

1. Introduction

Freezing of gait (FOG) is one of the most disabling symptoms of Parkinson's disease (PD), often leading to falls and reduced independence (Nutt et al. 2011). Most studies classify individuals as either “freezers” or “nonfreezers,” assuming these groups are homogeneous. However, this assumption can be questioned. Emerging evidence suggests that FOG may reflect a continuum rather than a binary state. Mancini et al. (2023) introduced a freezing index describing a spectrum of FOG severity—from absent to severe—in people with PD and even among healthy controls, and the possibility of temporary freezers further challenges the traditional dichotomy. Recently, it was suggested to classify into three groups: those without FOG, those with occasional FOG, and those with frequent FOG (Gilat et al. 2026, Tosserams et al. 2025; see also Mancini et al. 2023). If FOG is truly homogenous, individuals classified as freezers should score consistently across questionnaires, clinical observations, and quantitative tests. Heterogeneity or tool‐dependency would instead appear as discrepancies between these measures. In this study, we reexamined the homogeneity of FOG using four different freezing indicators (two movement disorder society‐unified Parkinson's disease rating scale [MDS‐UPDRS] items and two performance tasks), expecting consistent scores only if FOG is truly uniform across tools.

Many factors have been proposed to explain conversion from nonfreezing to freezing but it is often unclear whether these reflect FOG itself or the more advanced disease stage associated with it. To explore this, we used items from the Parkinson's disease functional movement battery (PD‐FUNC), a recently proposed set of simple motor test items suitable for home and clinical settings (Smits‐Engelsman and Duysens 2025), for examining differences between PWP with and without FOG and determining which differences remain after controlling for disease severity.

A related question is whether FOG represents an irreversible state from which there is no return. Recovery seems plausible, as freezing has been observed in less advanced disease stages (Okuma and Yanagisawa 2008; Amboni et al. 2015; Sharafkhah et al. 2024; Zhang et al. 2021). Addressing this requires longitudinal evaluation. Previous studies have examined short interictal periods (Plotnik and Hausdorff 2008; Zhao et al. 2025) or long‐term changes, though most focus on conversion to freezing (D'Cruz et al. 2020; Giladi et al. 2001; Forsaa et al. 2015; Zhang et al. 2016; Ou et al. 2018; Herman et al. 2018; Ehgoetz Martens et al. 2018; Virmani et al. 2023), while the reverse trajectory (FOG to NOFOG) is rarely documented. Furthermore, the studies mentioned typically rely on laboratory tools not widely available. The present case study therefore examines whether the evolution of freezing can be estimated using turning 360° on a mat in the home setting.

2. Methods

2.1. Procedures

Testing was conducted either at physical therapy institutions or at the patients' homes by 14 physical therapists as part of their master's project in geriatric physical therapy. Most therapists were affiliated with a healthcare network specializing in care for PWP. Prior to data collection, they received 2–3 h of additional training on the tools. Each therapist piloted the test administration with at least two individuals before starting the study, and any questions about instructions or scoring were clarified with the project leader (BSE).

PWP referred for intervention were invited to participate. Participants were excluded if they had another neurological disorder, or a score of 24 or lower on the Mini‐Mental State Examination. Additional exclusion criteria included the required use of a walking aid, recent surgery, recent illness (COVID‐19 or influenza), and the presence of comorbidities as identified by the Physical Activity Readiness Questionnaire (PAR‐Q). (Warburton et al. 2011). All PWP were tested in their standardized “ON” state. Written informed consent was obtained from all the participants before enrollment. Approval for this study was obtained from the CMO Radboud University Medical Center (file number 2022–15731).

2.2. Participants

In total 102 PWP participated in this study with a mean age of 72 (SD 8.1 range 52–88). There were more males in the sample than females (72/30). With regard to disease severity, 38 were in Hoehn and Yahr (H&Y) Stage 1, 40 in Stage 2, and 24 in Stage 3. Ten PWP did not use any form of Levodopa medication. The equivalent daily dose of levodopa was calculated using Parkinson's Measurement Levodopa Equivalent Dose Calculator. (https://www.parkinsonsmeasurement.org/toolBox/levodopaEquivalentDose.htm) Medication dose for two PWP was missing (Table 1).

TABLE 1.

Demographic and clinical characteristics of participants with Parkinson's disease (PWP), presented for the total sample and stratified by freezing status (nonfreezers, occasional, and freezers).

	Total	Nonfreezers	Occasional	Freezers	Statistics
Number	102	65	12	25
Age	73 (67–78)	73 (67–78.0)	72 (67.3–75.8)	73 (67.5–80)	Z = 1.204, p = 0.548
Sex, male/female (%)	72/30 (70.7/29.4%)	45/20 (69.2/30.8%)	10/2 (83.3/16.7%)	17/8 (68/32%)	Chi2 1.077, p = 0.584
Years since diagnosis	3.0 (2–7)	2.0 (1.0–7)	4.5 (3.5–9.5)	4.0 (2.5–7)	Z = 5.041, p = 0.08
Height (cm)	176 (169–182)	174 (167–182)	178 (173–182)	175 (169–184)	Z = 1.047, p = 0.592
Weight (kg)	79 (70–88)	80 (70–88)	78 (70–83)	75 (71–90)	Z = 0.532, p = 0.766
BMI	26.0 (23.1–27.6)	26.2 (24.2–27.8)	24.2 (22.7–26.1)	25.9 (22.9–27.6)	Z = 2.176, p = 0.337
LEDD a	694 (450–750)	680 (400–750)	750 (565–945)	675 (450–950)	Z = 2.387, p = 0.302
Hoehn & Yahr 1 (%)	38 (37.3%)	30 (46.2%)	3 (25.0%)	5 (20.0%)	Chi2 17.213, p = 0.002
Hoehn & Yahr 2 (%)	40 (39.2%)	27 (41.5%)	6 (50%)	7 (28.0%)
Hoehn & Yahr 3 (%)	24 (23.4%)	8 (12.3%)	3 (25%)	13 (52.0%)

Note: Values are medians (IQR) unless otherwise indicated. Statistics describe the results for the group comparison based Kruskal–Wallis and cross tabulation. Ten people were not using medication (nonfreezers n = 6, occasional freezers n = 1, and freezers n = 3).

^a Equivalent daily dose of levodopa.

2.3. Materials

In addition to demographic and PD history, the following assessments were administered: MDS‐UPDRS, PD‐FUNC, and NFOG‐Q.

2.3.1. MDS‐UPDRS

The movement disorder society–unified Parkinson's disease rating scale (MDS‐UPDRS) consists of four parts (Goetz et al. 2008). Part I, addressing nonmotor experiences of daily living, was not deemed relevant for this motor‐focused evaluation of freezing. Part II assesses motor experiences of daily living through 13 patient‐reported items, while Part III, the motor examination, comprises 18 items evaluated by a physical therapist. Part IV focuses on motor complications, specifically dyskinesia and fluctuations, using six items and is not reported in this study. Each item is scored on a 0–4 scale, where 0 indicates normal function and 4 indicates severe impairment.

2.3.2. PD‐FUNC

Quantitative assessment of functional movement is essential for guiding treatment and tracking progression in persons with PD (Smits‐Engelsman and Duysens 2025). While numerous motor function tests exist, few provide a holistic, practical, and sensitive measure across different domains relevant to PD. To meet this clinical need, we developed the Parkinson's disease functional movement battery (PD‐FUNC), a concise and comprehensive test battery based on movement disorder society‐unified Parkinson's disease rating scale part III (MDS‐UPDRS‐III) domains (Goetz et al. 2008). The entire testing process takes approximately 30–40 min, depending on participants' motor ability. Task instructions are detailed in the PD‐FUNC manual (Smits‐Engelsman and Duysens 2025). Reliability of the PD‐FUNC was examined using Cronbach's alpha and was 0.867. The inter‐rater reliability was not statistically evaluated.

2.3.3. New Freezing of Gait Questionnaire

The new freezing of gait questionnaire (NFOG‐Q) is a self‐reported questionnaire consisting of nine items that measure FOG, using a 5‐point answering scale for the severity of the freezing episodes experienced during the last months. This questionnaire has its limitations when used as primary outcome but is useful in defining freezers (Nieuwboer et al. 2009; Hulzinga et al. 2020).

2.3.4. Freezing Assessment and Classification

FOG was assessed using multiple sources: patient‐reported: MDS‐UPDRS item 2.13, clinician‐observed: MDS‐UPDRS item 3.11 during gait assessment, performance‐based: observed freezing during turning 360° on a circle (Figure 1), and performance‐based freezing observed during agility ladder task with turning around a pylon. All nonzero values for 2.13 FREEZING or 3.11 FREEZING OF GAIT were considered a positive score on these items. To minimize misclassification, we adopted a composite definition of freezing based on the number of positive indicators (self‐report, clinical observation, and observation in two performance‐based tasks). Individuals were classified as occasional freezers if they had a nonzero score on a single indicator, as freezers if positive on at least two indicators, and as nonfreezers if all indicators were zero. This definition aligns with recommendations to avoid reliance on a single tool, such as a questionnaire when identifying FOG (Hulzinga et al. 2020). This approach improves specificity and ensures that participants classified as “Freezers” demonstrate freezing more consistently rather than incidentally. The patient‐reported NFOG‐Q (only available for 81 participants) was not used to classify the participants into groups. MDS‐UPDRS questions were scored independently by patients (Part II) and therapists who treated the PWP (Part III). The PD‐FUNC was scored independently by the examiners (trained master students).

FIGURE 1.

Mat used for the turning test. The instruction was to remain in the red area while turning 360°.

2.4. Statistics

Data distribution was assessed, and test outcomes were found to be non‐normally distributed. Frequencies (sex and H&Y stage) between freezers, occasional freezers, and nonfreezers were compared using cross tabulation (chi‐square test). Agreement between the tools was calculated using Cohen's Kappa. Kappa result should be interpreted as follows: values ≤ 0 as indicating no agreement and 0.01–0.20 as none to slight, 0.21–0.40 as fair, 0.41–0.60 as moderate, 0.61–0.80 as substantial, and 0.81–1.00 as almost perfect agreement (Landis and Koch 1977).

A three‐group comparison (Kruskal–Wallis test) was conducted to identify items, which were significantly different between PWP with different levels of measured freezing. Pairwise post hoc comparison was performed, and p‐values are reported. These can be interpreted without (p‐value 0.05) and with Bonferroni correction (p‐value 0.017). Subsequently, a nonparametric ANCOVA (Quade) was conducted to compare freezer groups while controlling for H&Y stage. Hedges' g effect sizes were calculated because this reduces the overestimation of Cohen's d in small samples (Cohen 1988).

3. Results

3.1. Participants

Demographic information about the 102 participants is summarized in Table 1. Our classification of occasional FOG as those having a single indicator, while the frequent freezers (“freezers” in short) were defined as having at least two positive indicators (from the 4 used) across different sources led to a division in a group of 65 nonfreezers, 12 occasional freezers, and 25 freezers, which were used for the group comparisons. No significant differences were found for the demographic characteristics or in LEDD across groups. However, the frequency of the disease severity was different between freezer groups (chi‐square 17.213, p < 0.002) (for details, see Table 1).

3.2. Freezing

Of the total group, 63.7% (n = 65) of the PWP experienced no signs of freezing on any of the tools (Figure 2). The rest experienced freezing; however, the frequencies of reported or measured freezing varied depending on the tools used (Table 2).

FIGURE 2.

Distribution of PWP according to the number of positive freezing tests (0, 1, 2, 3, or all 4), based on MDS‐UPDRS items and PD‐FUNC tasks.

TABLE 2.

First column gives the description of the assessment tools used in the freezing assessment.

Assessment tools	Freezing negative (%)	Freezing positive (%)
MDS‐UPDRS 2.13 FREEZING: Over the past week, on your usual day when walking, do you suddenly stop or freeze as if your feet are stuck to the floor.	No freezing reported (0): 80 (78.4%)	1: Slight 16 (15.7%) 2: Mild: 3 (2.9%) 3: Moderate 3 (2.9%) 4: Severe 0
MDS‐UPDRS 3.11 FREEZING OF GAIT Instructions to examiner: While assessing gait, also assess for the presence of any gait freezing episodes. Look for start hesitation and stuttering movements especially when turning and reaching the end of the task. To the extent that safety permits, patients may NOT use sensory tricks during the assessment.	No freezing observed (0): 85 (83%)	1: Slight: 11 (10.8%) 2: Mild: 6 (5.9%) 3: Moderate 0 4: Severe 0
PD‐FUNC Agility ladder task: The individual runs or walks, as fast as safely possible, placing one foot in each square of the agility ladder, then goes around a pilon and back through the ladder to the end. Time ends when both feet are back over the starting line and on the floor. Time is recorded and if freezing occurs during the run or at the turning point.	No freezing observed: 87 (85.3%)	Freezing observed 15 (14.7%)
PD‐FUNC Turning on the mat 360°: Turn in the circle Clockwise and Anticlockwise. The patient starts with two feet outside the mat; steps into the circle and rotates until both feet are back on each side of the circle line and then steps out again. The task is done twice (clockwise and anticlockwise). Time is recorded and if episodes of freezing occur.	No freezing observed 78 (76.5%)	Freezing observed 24 (23.5%)
NFOG‐Q (n = 81) a self‐reported questionnaire consisting of nine items that measure FOG. Completion of these nine questions is limited to individuals who have reported freezing of gait in the past month.	No freezing reported 60 (74.1%)	Score 1–10: 14 (17.3%) Score 11–23: 7 (8.5%)

Note: Prevalence of the outcomes across different tools are shown as number (percentage) of participants scoring zero (Freezing negative) or nonzero on the tool (Freezing positive) is displayed in the second and third column.

To study whether freezing was tool‐dependent, the frequency of the number of positive test items was evaluated (Figure 2). Only four people froze on all four tools. Interestingly, these were three PWP in H&Y3 but one in H&Y1. The person in H&Y1, however, was diagnosed 20 years earlier.

The mean agreement in no signs of freezing over the four instruments was 73%. Agreement between the four tools is shown in Table 3. Agreement between the tools was fair (0.21–0.40) to moderate (0.41–0.60). The percentage of PWP without signs of freezing on any of the tools decreased with increasing disease severity and was 79%, 68%, and 33% in H&Y 1, 2, and 3, respectively (see Figure 3).

TABLE 3.

Agreement between two assessment tools in the number (percentage) of participants scoring zero (negative for freezing) and Kappa values for the agreement between yes or no freezing score between two tools.

Assessment tools		Agreement number (%) of PWP without signs of freezing out of the total (n = 102)	Kappa and p‐values
MDS‐UPDRS 2.13 FREEZING	MDS‐UPDRS 3.11 FREEZING	75 (73.5)	0.526, p = 0.001
MDS‐UPDRS 2.13 FREEZING	Agility ladder task	72 (70.6)	0.247, p = 0.01
MDS‐UPDRS 2.13 FREEZING	Rotation on the circle	69 (67.6)	0.439, p = 0.001
MDS‐UPDRS 3.11 FREEZING	Agility ladder task	76 (74.5)	0.259, p = 0.009
MDS‐UPDRS 3.11 FREEZING	Rotation on the circle	73 (71.6)	0.485, p = 0.001
Agility ladder task	Rotation on the circle	74 (72.5)	0.468, p = 0.001

FIGURE 3.

Proportion of PWP without freezing signs, with one positive test, and with two of more across Hoehn & Yahr stages (1, 2, and 3). The percentage of participants without freezing decreased with increasing disease severity. More than half the PWP (54%) in Hoehn & Yahr 3 were freezers.

Next, we evaluated differences on the items of the PD‐FUNC between the three freezing groups (Tables 4 and 5). Many items showed significant differences, which could be partly related to later stages of the disease. When differences between groups were controlled for the severity of the PD (H&Y as covariate), only two PD‐FUNC items remained significant and with large effect sizes based on uncorrected means (see Table 4). Interestingly, one of these was a fine motor item (tapping between two letters on a keyboard; “typing”), and the second one was turning 360° on a circle. For MDS‐UPDRS Part III, scores were higher in occasional freezers compared with nonfreezers, whereas scores were similar between occasional freezers and freezers (total scores 15.7, 25.3, and 25.5, respectively). For Part II, adjusted group differences were attenuated after controlling for H&Y stage (p = 0.06), although mean scores increased across groups (6.7, 10.0, and 12.6, respectively).

TABLE 4.

Comparison of PD‐FUNC items and MDS‐UPDRS scores between nonfreezers, occasional freezers, and freezers.

	Kruskal–Wallis		Post hoc nonfreezers (0), occasional freezers (1), and freezers (2)			Quade with H&Y as covariate
Item PD‐FUNC	H‐value	p	p‐value 0–1	p‐value 1–2	p‐value 0–2	F‐value	p
Pegboard	9.010	0.011	0.028	0.636	0.015	2.764	0.068
Threading	12.505	0.002	0.042	0.772	0.001	2.606	0.079
Tracing time	4.378	0.112				1.830	0.166
Tracing mistakes	6.835	0.033	0.240	0.520	0.011	1.524	0.223
Writing	8.225	0.016	0.067	0.914	0.011	2.451	0.091
Typing	9.712	0.008	0.137	0.499	0.003	3.169	0.046 a
Running	8.310	0.016	0.604	0.142	0.004	2.225	0.113
Stepping	4.936	0.085	0.692	0.257	0.026	1.080	0.343
Step sideways	7.028	0.030	0.184	0.616	0.012	1.562	0.215
Foot on foot	6.855	0.032	0.565	0.035	0.017	1.890	0.156
Tandem stance	7.510	0.023	0.386	0.018 b	0.018 b	2.154	0.121
Hug knee	6.624	0.036	0.614	0.203	0.010	1.085	0.342
Rotation standing	15.890	0.001	0.086	0.288	0.001	4.631	0.012 c
Rotation lying	4.589	0.101				1.533	0.221
Grip force	4.029	0.133				1.259	0.288
Sit to stand	4.794	0.091				0.918	0.403
Lift box	0.914	0.633				0.066	0.936
MDS‐UPDRS
Part II experiences of daily living	14.517	0.001	0.071	0.403	0.001	2.887	0.06
Part III motor examination	14.997	0.001	0.200	0.786	0.001	3.172	0.046 d

Note: Kruskal–Wallis results are shown for unadjusted group differences with post hoc analysis; ANCOVA (Quade) results are shown for analyses adjusted for disease severity (Hoehn & Yahr stage as covariate). Post hoc results and Hedges' g effect sizes are shown for the three variables that remained significant in the covariance analysis. These effect sizes reflect between‐freezer group differences unadjusted for disease severity.

^a Post hoc analysis nonfreezers (0), freezers (2). t = 2.38, p = 0.019, Hedges' g = −0.736.

^b Not significant after Bonferroni correction.

^c Post hoc analysis nonfreezers (0), freezers (2). t = −2.915, p = 0.004, Hedges' g = 0.922.

^d Post hoc analysis nonfreezers (0), occasional freezers (1). t = −2.107, p = 0.038, Hedges' g = 0.886.

TABLE 5.

Test outcomes PD‐FUNC items and MDS‐UPDRS for the three groups based on freezing classification (mean, medians, SD, and IQR range).

	Pegboard (s)	Threading (s)	Tracing time (s)	Tracing mistakes (#)	Writing (#)	Typing (#)	Running (s)	Stepping	Step sideways (s)	Foot on Foot (s)	Tandem stance (s)	Hug knee (s)	Rotation standing (s)	Rotation lying (s)	Grip force (kg)	Sit to stand (#)	Lift box (#)	MDS_UPDRS part II experiences of daily living (points)	MDS_UPDRS part III motor examination (points)
No freezing (n = 65)
Mean	37.1	27.5	40.0	2.2	342.9	52.3	12.9	19.7	8.5	10.8	13.6	9.3	7.0	7.7	33.4	15.1	16.2	6.7	15.7
Median	33.6	24.7	37.1	2.0	322.0	53.0	10.5	17.7	6.8	11.6	15.0	9.3	6.0	5.7	33.0	15.0	15.0	6.0	14.0
SD	18.5	9.0	15.9	2.3	109.0	14.2	7.0	8.1	4.4	4.6	2.7	5.0	3.4	6.6	11.5	4.9	6.2	4.7	9.8
Lower IQR	26.1	22.1	29.0	0.0	253.8	42.6	7.9	13.7	5.6	7.3	13.4	5.1	4.6	3.5	23.0	12.0	11.5	3.0	9.0
Upper IQR	40.8	30.1	47.3	3.0	399.5	59.5	15.1	25.5	10.2	15.0	15.0	15.0	8.1	9.0	41.3	18.0	18.0	10.0	19.5
Occasional freezing (n = 12)
Mean	52.4	33.3	57.3	3.6	275.3	45.7	13.2	21.2	10.7	11.7	13.9	8.5	9.0	6.7	35.5	13.6	15.3	10.0	25.3
Median	48.6	29.7	55.3	3.0	287.5	47.0	13.5	19.4	10.4	13.2	15.0	8.7	8.1	5.5	36.3	12.5	15.0	10.0	22.0
SD	24.3	11.0	27.4	3.4	75.1	14.8	5.5	10.0	6.9	4.1	2.9	5.1	4.2	5.3	9.7	4.3	4.4	5.9	15.2
Lower IQR	32.4	25.9	34.0	0.0	220.3	37.5	8.5	13.1	6.3	9.0	15.0	4.9	6.3	3.8	29.1	10.5	12.3	5.3	13.8
Upper IQR	66.4	41.0	74.0	6.0	330.8	51.5	15.6	28.6	12.9	15.0	15.0	13.5	10.7	7.7	43.6	17.3	19.5	13.5	30.8
Freezing (n = 25)
Mean	46.1	36.6	48.0	4.5	275.7	41.5	16.2	24.2	11.1	8.3	11.8	6.3	11.3	9.5	28.7	13.6	15.0	12.6	25.5
Median	36.1	31.7	42.7	3.0	262.0	41.5	15.7	22.5	10.0	10.0	11.4	6.7	9.7	8.2	29.5	12.0	13.0	12.0	22.0
SD	23.9	16.3	27.1	4.1	96.1	16.0	6.6	9.8	5.9	5.1	3.8	3.9	7.0	5.7	9.6	5.1	6.2	7.0	12.8
Lower IQR	33.6	26.0	26.4	1.0	192.0	28.0	11.6	19.8	7.9	4.4	10.2	3.0	6.8	5.0	20.3	10.0	11.0	6.0	16.5
Upper IQR	48.6	43.3	62.8	8.0	330.0	53.8	18.0	26.2	12.4	12.8	15.0	9.1	13.1	13.4	37.8	15.0	19.0	18.5	32.5

We also analyzed the available NFOG‐Q data (N = 81); 74.1% had a zero score, and group comparison showed significant differences (H = 21.552, p < 0.001); after correction for disease severity, differences remained significant (0–1, p = 0.021; 1–2, p = 0.018; 0–2, p = 0.001).

3.3. A Case Study

To highlight the elements involved in the conversion to FOG, it is worthwhile to consider detailed longitudinal studies. An example is provided here, based on the experience of one of the authors (JD), who was tested at irregular intervals with the PD‐FUNC over several years. An increment of rotation time is known to be associated with freezing (Bengevoord et al. 2016). To get a reliable picture of the evolution of the disease, it is valuable to use timed turning performance instead of relying on self‐reporting. Indeed, Goris et al. (2025) found that many PWP were unaware of their FOG yet experienced FOG during this task. On the other hand, it has to be recognized that other factors (e.g., rheumatologic disease) could also affect turning, but such factors are unlikely to have the episodic characteristic, typical for FOG. In the presented case, self‐reported freezing occurred in the two periods with increased turn duration (22‐10‐2022 and 27‐09‐2025) (Figure 4). Importantly, between these two periods the measured durations returned to normal, and no freezing was detected.

FIGURE 4.

Evolution of measured durations in PWP subject (JD) over several years. The tests were taken at irregular intervals; hence the equal distances of the data points should not be seen as constant intervals. Falls occurred between October 2022 till Januari 2023 and in the period of March till September 2025.

4. Discussion

This study provides a detailed examination of freezing phenomena in PD across multiple assessment tools, highlighting the variability in detection and its relationship with disease severity. While most participants exhibited no freezing on any measure, results differed between tools, underscoring the complexity of assessing this symptom. Importantly, only four PWP demonstrated freezing consistently across all tools, suggesting that freezing may not represent a uniform construct but rather depend on the context of the different tools. These findings emphasize the need for multimethod approaches and careful consideration of disease stage when interpreting freezing‐related outcomes. In particular, caution is needed when self‐reporting is used as the primary outcome (Gilat et al. 2026, Goris, M. et al. 2024, Hulzinga et al. 2020). Furthermore, it should be noted that our population missed very severe cases of FOG (see Table 2). In more severe freezing, the certainty of correct classification will be higher (Gilat et al. 2026). Importantly, freezing was not related to the dosage of the dopamine medication.

The first key finding was the marked heterogeneity of FOG. Only four PWP (3.9%) scored positive on all indicators tested, whereas up to 64% showed no freezing on any of the indicators. This left approximately a third of the PWP (32.3%) as scoring positive on 3, 2, or 1 indicator. This finding aligns with the literature: In some studies, self‐reported FOG is more frequent than the clinically observed FOG (Sawada et al. 2019), whereas in others, the inverse pattern was found (Goris et al. 2025). In the present study, the clinical rotation test item yielded a slightly higher percentage of freezers than the self‐reported percentage freezers (based on MSD‐UPDRS).

A major source for differences between freezer and nonfreezers appears to be the disease stage. Up to 31% of freezing occurred in PWP with relatively low H&Y stages (1 and 2) confirming previous reports of mild FOG in the early stages (Okuma and Yanagisawa 2008; Amboni et al. 2015; Sharafkhah et al. 2024; Zhang et al. 2021). Conversely, only one‐third of PWP in H&Y Stage 3 showed no signs of freezing, underscoring the importance of considering disease severity when interpreting FOG‐related outcomes. In the present analysis of differences between freezers, occasional freezers and nonfreezers, H&Y was included as a covariate to control for severity effects. Furthermore, the freezers group was strictly defined as those positive on at least two indicators.

4.1. Clinical Factors Distinguishing Freezers From Occasional Freezers and Nonfreezers

The rotation test of the PD‐FUNC was used to differentiate freezers, occasional freezers, and nonfreezers. Comparable versions of rotation tests are used widely to elicit freezing (Snijders et al. 2012; Conde et al. 2023). In agreement with Goris et al. 2025, rotation while standing (turning 360°) emerged as an effective clinical task to evoke FOG, despite differences in test protocols. In Goris et al. 2025, the subjects were required to turn rapidly in both directions for 1 min, sometimes with an added cognitive dual‐task. In the present study, the duration required to complete the full 360 turn within the circle on the mat served as an outcome measure. Controlling for H&Y ensured that this result truly reflected the tendency to be classified as a freezer (Hedges g 0.92).

Another strong predictor of freezing was the MDS‐UPDRS section III (Motor examination), thereby supporting the value of this scale and the impact freezing has on motor performance. Others have emphasized the value of the motor part of the MDS‐UPDRS in predicting the conversion to freezing (D'Cruz et al. 2020). The higher Part III scores observed in occasional freezers compared with nonfreezers, together with the consistent increase in scores across groups in Part II scores, may suggest progressive involvement of motor and motor‐experienced limitations across freezing severity stages. Additionally, the NFOG‐Q (available for a subgroup of N = 81) showed significant differences among all groups.

Gait deficits have also frequently been mentioned as predictors of FOG (Giladi et al. 2001; Perez‐Lloret et al. 2014; Ehgoetz Martens et al. 2014; D'Cruz et al. 2020; Aktürk et al. 2021). In the PD‐FUNC, the gait velocity was measured as the time needed to run back and forth across an agility ladder, including a turn. Herman et al. (2018), using a 10‐m walking test, similarly found a significant but minor contribution of gait speed to predict FOG. Results in the present study revealed that there were no longer differences in walking or running speed after correction for H&Y stage.

A more surprising finding was the association between FOG and the deficit in fast finger movements. The PD‐FUNC typing test item involves alternating between two letters using one finger. Our typing results are consistent with D'Cruz et al. (2020), who identified finger tapping abnormalities to be predictive for FOG. Specifically, alternating flexion and extension of the index finger strongly predicted conversion to FOG.

4.2. Bradykinesia

The question arises whether these various correlated factors share an underlying mechanism. Many of the elements listed above are related to slowness of movement. Previous studies have shown bradykinesia to be a risk factor for developing FOG (Giladi et al. 2001), although it has been argued that these two symptoms (bradykinesia and FOG) may have a different origin (Bartels et al. 2003; Aktürk et al. 2021). Bradykinesia may result from a deficit in motor motivation, consistent with the finding that subthalamic nucleus stimulation reduces bradykinesia (Baraduc et al. 2013). According to Berardelli et al. (2001), bradykinesia is due to a lack of basal ganglia output to reinforce the cortical mechanisms needed to prepare and execute the commands to move. In contrast, FOG is seen as an alarm reaction, involving amygdala and related circuitry (Duysens and Smits‐Engelsman 2023).

In PWP, low motor energy may induce overly cautious and conscious movement. Moreover, this slowness is exacerbated by events such as falls, triggering a panic reaction, termed freezing, that bypasses conscious control.

To throw light on which elements are involved in the conversion to FOG, it is worthwhile to consider detailed longitudinal studies. The present paper includes an example based on the experience of one of the authors (JD). The data show that there are periods of increased rotation duration, which coincided with self‐reported FOG. The present findings are in line with previous work showing an increase in turn duration in freezers (Son et al. 2022; Mancini et al. 2018). Importantly, there was an intervening period of normal rotation, accompanied by an absence of self‐reported FOG, indicating possible recovery. This is in line with the report of Hulzinga et al. (2020), which mentioned that seven of the 57 participants of their study changed from being self‐reported freezers to nonfreezers within a 6‐week period. In other words, there is recovery potential: to freeze is not to cease.

4.3. Limitations

The present study has several limitations. First, most participants were relatively newly diagnosed with PD (median of 2 years since diagnosis). Second, the main section of this paper uses a cross‐sectional design, preventing conclusions about the conversion between nonfreezer, occasional freezer, and freezer status and vice versa. Third, although multiple tools were used, variability in freezing assessment protocols and the mix of self‐report versus observational measures may have introduced classification inconsistencies. Finally, participants were evaluated on their normal medication regime, potentially underestimating freezing during OFF periods.

Future research should use larger samples and incorporate longitudinal designs and real‐life assessments using wearable technology to improve ecological validity. Wearable devices offer significant advantages for monitoring PD by providing continuous, objective, and real‐world data on motor symptoms (e.g., bradykinesia, tremor, and dyskinesia) and activity levels, which may help explain and capture heterogeneity not detected by traditional in‐clinic assessments (Reichmann et al. 2023; Moreau et al. 2023). Additionally, more sensitive measures of executive function and affect (e.g., anxiety and fear of falling) are advised because these factors may have contributed to the observed heterogeneity.

5. Conclusions

FOG in PD is a heterogeneous and context‐dependent phenomenon that can occur even in early disease stages. Our findings demonstrate substantial variability in detection across assessment tools, with only a small proportion of individuals consistently classified as freezers. This underscores the importance of using multimethod approaches and considering disease severity when interpreting FOG‐related outcomes. Importantly, adjustment for disease severity (H&Y) explains a substantial proportion of the observed group differences, indicating that disease severity plays a major role, rather than freezing status per se. By controlling for H&Y, we ensured that our results reflected a true association with freezing. Clinical tasks such as 360° rotation and fine motor performance contributed to differentiation.

Author Contributions

Jacques Duysens: conceptualization, writing – original draft, investigation, methodology, writing – review and editing. Bouwien Smits‐Engelsman: conceptualization, investigation, methodology, supervision, data curation, formal analysis, writing – review and editing, project administration, validation.

Funding

The authors have nothing to report.

Ethics Statement

Researchers adhered to the principles outlined in the Declaration of Helsinki. Written informed consent was obtained from all the participants before their enrollment. Approval for this study was obtained from the CMO Radboud University Medical Center (file number 2022‐15731).

Conflicts of Interest

The authors declare no conflicts of interest.

Duysens, J. , and Smits‐Engelsman B.. 2026. “Freezing of Gait in Parkinson's Disease: A Heterogeneous Phenomenon Across Assessment Tools.” European Journal of Neuroscience 63, no. 11: e70572. 10.1111/ejn.70572.

Data Availability Statement

All relevant data are within the manuscript. Availability of data and materials. The full datasets presented in this article are not readily available because of Human Subjects' protections. Requests to access the data sets or use the PD‐FUNC manual should be directed to bouwiensmits@hotmail.com