Predictors of changes in gait performance over four years in persons with late effects of polio

Abstract

BACKGROUND:

Reduced gait performance is common in persons with late effects of polio.

OBJECTIVE:

To identify predictors of change in gait performance over four years in persons with late effects of polio.

METHODS:

Gait performance was assessed annually in 51 ambulatory persons (mean age 64 years, SD 6) by the Timed “Up & Go” (TUG), Comfortable and Fast Gait Speed (CGS, FGS), and 6-Minute Walk Test (6MWT). Isokinetic knee extensor and flexor muscle strength was measured with a Biodex dynamometer. Mixed Linear Models were used to analyze changes in gait performance and to identify any predictors of change among the covariates gender, age, body mass index, time with new symptoms, baseline reduction in gait performance and knee muscle strength.

RESULTS:

There were significant linear effects over time (reduction per year) for three gait performance tests; CGS (0.8%; p < 0.05), FGS (1.7%; p < 0.001), and 6MWT (0.7%; p < 0.05) with significant random effects for all tests. The strongest predictor of a change in gait performance was the individual variations in the knee flexor strength (p < 0.001).

CONCLUSION:

The small gradual reduction in gait performance over time in persons with late effects of polio is primarily determined by the individual variations in the knee flexor strength.

Keywords

Post poliomyelitis syndrome gait outcome assessment longitudinal rehabilitation

1 Introduction

Many persons with late effects of polio have reduced gait performance as a result of muscle weakness, muscle fatigue, musculoskeletal pain or decreased balance (Boyer et al., 2010; Gylfadottir, Dallimore, & Dean, 2006; Jensen et al., 2011; Lexell & Brogårdh, 2012). This can lead to difficulties to perform daily activities (Brogårdh, Flansbjer, Espelund, & Lexell, 2013; Horemans, Bussmann, Beelen, Stam, & Nollet, 2005; Willen, Sunnerhagen, Ekman, & Grimby, 2004) restrict perceived participation (Larsson Lund & Lexell, 2008) and increase the risk of falling (Brogårdh, Flansbjer, & Lexell, 2016; Brogårdh & Lexell, 2014).

Gait performance in persons with late effects of polio can decline over time but the rate varies (Bickerstaffe, Beelen, & Nollet, 2015; Klein, Braitman, Costello, Keenan, & Esquenazi, 2008; Sorenson, Daube, & Windebank, 2005; Willen, Thoren-Jonsson, Grimby, & Sunnerhagen, 2007). Klein et al. (2008) found no evidence of a change in maximum gait speed over three years. Willen et al. (2007) found a significant decrease in both spontaneous (– 9%; p < 0.001) and maximal gait speed (– 5%; p < 0.01) after 4 years. Bickerstaffe et al. (2015) showed a significant decline in comfortable gait speed after 10 years (– 6%; p = 0.024). Sörensen et al. (2005), who followed a sample over 15 years, reported that the participants walked somewhat faster after 5 years but had significantly decreased gait speed after 15 years (– 3%; p < 0.01). Thus, there seems to be small changes in gait performance over a few years, but if the follow up time is at least 4 years the annual decline in gait speed vary between – 0.2% and – 2.3% (Bickerstaffe, et al., 2015; Sorenson, et al., 2005; Willen, et al., 2007).

It is also shown that muscle strength in the lower limbs decline between 1% and 3% annually in persons with late effects of polio (Bickerstaffe, van Dijk, Beelen, Zwarts, & Nollet, 2013; Sorenson, et al., 2005; Stolwijk-Swuste, Beelen, Lankhorst, & Nollet, 2005; Willen, et al., 2007), and that the reduction over four years is greater in men compared to women (Flansbjer, Brogårdh, Horstmann, & Lexell, 2015). Knee muscle strength influence gait performance and the knee flexor strength seems to be more important than the knee extensor strength for both gait speed and distance (Flansbjer, Brogårdh, & Lexell, 2013). Thus, it is reasonable to assume that a change in muscle strength influences gait performance over time, but yet there is limited knowledge. To the best of our knowledge, only two studies have evaluated this relationship in persons with late effects of polio. Willén et al. (2007) concluded that reduced muscle strength reflected the change in walking speed, but only to a small extent. Similarly, Bickerstaffe et al. (2015) reported that loss of isometric knee extensor strength (on average 15% reduction over 10 years) explained only a small proportion of the decline in gait performance.

Overall, there is a lack of longitudinal studies identifying predictors of change in gait performance in persons with late effects of polio. Covariates like gender, reduced gait performance at baseline or decline in muscle strength, could potentially influence gait performance over time. However, few of these covariates have been taken into account in previous longitudinal studies. Moreover, when analyzing longitudinal data in order to identify predictors of change, the mixed-random and fixed effect regression models (the Mixed Linear Models; MLM) are currently recommended (Edwards, 2000; Hoffman & Stawski, 2009; Locascio & Atri, 2011). As maintained gait performance is an important goal in rehabilitation of persons with late effects of polio, the identification of predictors of change over time in gait performance may help the rehabilitation team to select appropriate interventions.

Therefore, the aim of this study was to assess changes in gait performance over 4 years in persons with late effects of polio and to identify predictors of a change in gait performance among the covariates gender, age, body mass index (BMI), length of time with new symptoms, baseline reduction in gait performance and knee muscle strength measured annually over 4 years.

2 Methods

2.1 Participants

Community-dwelling ambulatory persons were recruited from a post-polio rehabilitation clinic at a university hospital in southern Sweden. Inclusion criteria were: i) 50 to 75 years of age; ii) a confirmed history of acute poliomyelitis affecting the lower limbs, with new weakness after a period of functional stability of at least 15 years; iii) ability to walk at least 300 m with or without an assistive and/or orthotic device. Exclusion criteria were: i) using wheelchair as the main mode of transportation; ii) ongoing systemic disease or any other disease that could affect their gait performance or muscle strength. We planned to include at least 50 participants to obtain a reasonably large sample to be assessed annually over four years. All participants were selected by the treating physician (JL, second author) in connection with their regular follow-up visits to the clinic. Those who met the criteria and were mildly to moderately affected in their lower limbs were consecutively invited to the study and after 18 months (2007 to 2009) 51 persons had agreed to participate.

For all participants an electromyogram (EMG) had been recorded in both lower limbs (vastus lateralis and tibialis anterior muscles) as part of the initial routine clinical examination and verification of prior polio. Following the National Rehabilitation Hospital (NRH) Post-Polio Limb Classification (Halstead, Gawne, & Pham, 1995), all participants had post-polio NRH class II to V (indicating clinically stable or unstable polio) in at least one of their lower limbs’ muscles. According to the persons’ own perception of their muscle weakness, one lower limb was defined as the “more affected lower limb” and the other as the “less affected lower limb”.

Prior to inclusion, information about the purpose of the study was provided, and each individual gave his or her written informed consent to participate. The study followed the principles of the Helsinki declaration.

2.2 Tests and measurements

The participants’ gait performance and knee muscle strength were tested annually over four years (a total of five times) as close as possible to one year between each test (mean 12.1 months, SD 0.8). All tests were performed by the same physiotherapist (UBF, first author).

2.2.1 Gait performance test

The following four gait performance tests were used: the Timed “Up & Go” test (TUG), the 10 meter Comfortable and the 10 meter Fast Gait Speed tests (CGS and FGS), and the 6-Minute Walk test (6MWT). These tests have been shown to be reliable in persons with late effects of polio (Flansbjer & Lexell, 2010a). All participants were allowed to use, if needed, their walking aid and/or orthotic device during the tests. A digital stopwatch with an accuracy of one decimal figure in units of 1 sec was used to measure time. All tests were performed in a 2.2 m wide corridor with a linoleum floor in a quiet part of the hospital.

For the TUG, the participants sat in an arm chair placed at the end of a marked 3-m walkway. They were instructed to sit with their back against the chair and on the word “go”, stand up, walk at a comfortable speed passed the 3-m mark, turn around, walk back and sit down in the chair. Each participant did one trial to become familiar with the test, and then performed the TUG twice with a one-min rest between each trial. The time from the start until they sat down again in the chair with back support was measured and the mean of the two tests was recorded. All participants were allowed, if needed, to use the armrests for support when raising and sitting down. The TUG test has been used as an outcome measure for mobility in persons with late effects of polio (Brogårdh, Flansbjer, & Lexell, 2010; Lehmann, Sunnerhagen, & Willen, 2006).

For the FGS and CGS, a 14-meter walkway was marked at the floor and the participants were timed over the middle 10 meter. For the CGS, the participants were told to walk at a self-selected comfortable pace, whereas for the FGS they were told to walk as fast and safely as possible without running. The test session started with the CGS three times in succession and with 30 s between each trial. After a further 30 s rest, the test session continued with the FGS, also three times in succession, with 30 s between each trial. The time taken to walk 10 meter was recorded (in seconds) for each trial and the mean value was calculated for CGS and FGS, respectively. These gait speed tests have been used to assess mobility in persons with late effects of polio (Brehm, Nollet, & Harlaar, 2006; Brogårdh, et al., 2010; Horemans, et al., 2005; Willen, et al., 2007).

For the 6MWT, the participants were instructed to walk 30 meter between two marks on the floor, and, after passing either mark, turn and walk back. They were told to walk as far as possible during six minutes, and were informed when three minutes of the test remained. The 6MWT was performed once and the numbers of 30 meter lengths were counted. In addition, every meter was marked on the wall, so the distance walked could be measured to the nearest meter. The 6MWT has previously been used as a sub-maximal test of aerobic capacity in persons with late effects of polio (Brogårdh, et al., 2010; Gylfadottir, et al., 2006).

2.2.2 Knee muscle strength measurements

Isokinetic concentric knee extension and knee flexion muscle strength were measured with a Biodex^® Multi-Joint System 3 PRO dynamometer using a standardized test protocol (Flansbjer & Lexell, 2010b).

Before each strength measurement the range of motion (ROM) was set and the Biodex software applied the gravity correction. The participants were seated in an adjustable chair of the dynamometer, without shoes or any orthotic device, and were stabilized with straps across the shoulders, waist and thigh. After a structured warm-up they performed three maximal isokinetic extensor and flexor contractions at 60°/s and the highest peak torques were recorded (in Nm). More details can be found in our previous study of changes in knee muscle strength over time (Flansbjer, Brogårdh, Horstmann, & Lexell, 2015).

2.2.3 Testing procedure

All test sessions started with the knee muscle strength measurements, which took about 30 minutes to complete, and after a 30-minutes rest the different gait performance tests were assessed in the following order; TUG, CGS, FGS, and 6MWT. A standardized test protocol was carefully followed, with the same time interval between each test, using the same commands and performed in the same environment. The total time for the gait performance tests at each test session was approximately 30 minutes. After completion of the annual test sessions a summary of the test results were given to the participants.

2.3 Statistical analyzes

Data from the gait performance tests provided 4 quantitative variables to be analyzed. Four participants missed between one and three test session during the four years, but the missing values were compensated for in the statistical model.

The characteristics of the sample were described by frequencies, means and standard deviations. To obtain a value of the reduction in gait performance at baseline the estimated gait performance for 6MWT was calculated for each participant based on gender, age, height, and body mass specific values for a healthy population (Enright & Sherrill, 1998). The difference between the measured 6MWT at baseline compared to the estimated values were then calculated (i.e., the performed 6MWT in meter minus the estimated length in meter); positive values indicate better performance than predicted.

Mixed Linear Models (MLM) (Edwards, 2000) were used for the longitudinal data analyzes. Estimated means and standard deviations of all gait performance tests and knee muscle strength measurements were obtained based on the covariance matrix. The fixed linear effects (mean values for the group), the random effects (individual variations around the mean) and the deviations from the linear slopes over the four years were calculated.

To identify predictors for a change in gait performance, the baseline time-invariant covariates gender, age, BMI, length of time with new symptoms and baseline reduction in gait performance were analyzed in the mixed linear regression models. To be able to analyze knee extensor and flexor muscle strength in the model (time-varying covariates, measured annually) we first created two new time-invariant covariates: the person’s mean strength and the within-person changes in strength (each person’s raw values of strength minus their mean strength). All interval scaled covariates used in the analyzes were centred to the mean, giving positive values for the participants above the mean values of age, BMI, time with new symptoms, baseline reduction in gait performance, person’s mean strength and within-person changes in strength. The covariate gender was coded 0 for women and 1 for men. All baseline covariates and their interaction with time were inserted into the models together with the new strength covariates. The non-significant covariates were then removed in a backward manner, starting with the least significant interaction term; no covariate was removed when the interaction term was statistically significant. The final models comprised only covariates where the covariate itself or its interaction with time was significant.

All calculations were performed using the IBM SPSS Statistics Software version 22.0. Significance of random effects was tested with the deviance test (two degrees of freedom, df) based on Restricted Maximum Likelihood (REML). Significance levels below 0.05 represent statistical significance.

3 Results

In Table 1, the characteristics of the 51 participants (28 men and 23 women) are presented. Their mean age at the study enrolment was 64 years (SD 6, range 52– 74) and mean time since onset of new symptoms was 13 years (SD 7, range 2– 28). Fifty-two percent were most affected in their right lower limb and 71% had verified prior polio in both lower limbs. Walking aids were used by 8% of the participants and ankle foot orthoses by 18%. On average the participants walked 80 meters less than the estimated value for 6MWT at baseline.

Table 1
Characteristics of the 51 participants with late effects of polio at baseline

Gender

Men; number (%) 28 (55)

Women; number (%) 23 (45)

Age (years): mean (SD); range 64 (6); 52 to 74

Age at acute polio infection (years): mean (SD); range 5 (4); 0 to 15

Time with new symptoms (years): mean (SD); range 13 (7); 2 to 28

Body mass index: mean (SD); range 26 (3); 21 to 37

Most affected leg: number (%)

Right 27 (53)

Left 24 (47)

Both lower limbs affected; number (%) 36 (71)

Walking aid: number (%)

None 47 (92)

Stick or crutch 3 (6)

Rollator 1 (2)

Orthotic device: number (%)

None 42 (82)

Ankle Foot Orthosis (AFO) 9 (18)

The performed 6 MWT in meter minus the predicted – 80 (89); – 302 to 102

length in meter^a: mean (SD); range

^a For each participant the predicted gait performance for 6MWT was calculated based on gender, age, height, and body mass specific values for a healthy population (Enright & Sherrill, 1998). These values were used to calculate the difference in gait performance compared to the predicted at baseline (i.e., the performed 6 MWT in meter minus the predicted length in meter).

3.1 Fixed linear and random effects of gait performance

The results of the estimated means (SD) of the gait performance tests for the 51 participants from baseline and at each test occasion over the four years, as well as the fixed linear effects for the group and the random effects for the individuals, are presented in Table 2.

Table 2
The results of the estimated means of the gait performance tests for the 51 participants with late effects of polio

Test assessment Estimated mean of fixed effects Fixed linear Random

At baseline After 1 year After 2 years After 3 years After 4 years effect effect^†

TUG (s) 11.0 (0.3) 10.8 (0.3) 10.8 (0.3) 10.8 (0.3) 10.8 (0.3) – 0.04^ns 17.2^***

Gait speed over 10 m

CGS (s) 9.0 (0.2) 9.0 (0.2) 9.0 (0.2) 9.2 (0.2) 9.2 (0.2) 0.07^* 34.4^*

FGS (s) 6.4 (0.2) 6.7 (0.2) 6.6 (0.2) 6.8 (0.2) 6.9 (0.2) 0.11^* 48.0^***

6MWT (m) 454 (13) 451 (13) 449 (13) 447 (13) 440 (13) – 3.0^* 45.9^***

Test assessment	Estimated mean of fixed effects	Fixed linear	Random
TUG (s)	11.0 (0.3)	10.8 (0.3)	10.8 (0.3)	10.8 (0.3)	10.8 (0.3)	– 0.04^ns	17.2^***
Gait speed over 10 m
CGS (s)	9.0 (0.2)	9.0 (0.2)	9.0 (0.2)	9.2 (0.2)	9.2 (0.2)	0.07^*	34.4^***
FGS (s)	6.4 (0.2)	6.7 (0.2)	6.6 (0.2)	6.8 (0.2)	6.9 (0.2)	0.11^***	48.0^***
6MWT (m)	454 (13)	451 (13)	449 (13)	447 (13)	440 (13)	– 3.0^*	45.9^***

Estimated values are presented as mean (standard deviation; SD). TUG = Timed Up & Go; CGS = Comfortable gait speed; FGS = Fast gait speed; 6MWT = 6 minute walk test; s = seconds, m = meter. ^***p < 0.001, ^*p < 0.05, ns not significant, ^†Deviance (df = 2).

There were significant fixed linear effects over time (reduction per year) for three gait performance tests; CGS (0.8%; p < 0.05), FGS (1.7%; p < 0.001), and 6MWT (0.7%; p < 0.05). There were significant random effects (individual variations) for all tests (p < 0.001).

Over the four years, there was no significant deviation from the linear slopes for any gait performance test, indicating that declines in gait performance did not accelerate over time.

3.2 Fixed linear and random effects of knee muscle strength

The results of the estimated means (SD) of knee muscle strength for the 51 participants from baseline and at each test occasion over the four years, as well as the fixed linear effects for the group and the random effects for the individuals, are presented in Table 3. There were significant fixed linear effects over time (reduction per year) for two of the knee muscle strength measurements; for the extensors in both the more affected (– 2.8%; p < 0.001) and the less affected lower limb (– 1.2%; p < 0.05) with significant random effects for all tests.

Table 3
The results of the estimated means of knee strength measurements for the 51 participants with late effects of polio

Test assessment Estimated mean of fixed effects Fixed linear Random

At baseline After 1 year After 2 years After 3 years After 4 years effect effect^†

Less affected

Knee extensor 109.8 (6.3) 109.7 (6.3) 104.9 (6.3) 105.7 (6.3) 105.2 (6.3) – 1.3^* 16.0^*

Knee flexors 57.8 (3.5) 60.4 (3.5) 58.7 (3.5) 56.7 (3.5) 58.7 (3.5) – 0.2^ns 11.2^

More affected

Knee extensor 77.7 (5.9) 79.1 (5.9) 73.8 (5.9) 72.5 (5.9) 69.8 (5.9) – 2.2^* 15.5^*

Knee flexors 39.9 (3.3) 40.9 (3.3) 39.6 (3.3) 38.0 (3.3) 39.7 (3.3) – 0.3^ns 14.0^**

Test assessment	Estimated mean of fixed effects	Fixed linear	Random
Less affected
Knee extensor	109.8 (6.3)	109.7 (6.3)	104.9 (6.3)	105.7 (6.3)	105.2 (6.3)	– 1.3^*	16.0^***
Knee flexors	57.8 (3.5)	60.4 (3.5)	58.7 (3.5)	56.7 (3.5)	58.7 (3.5)	– 0.2^ns	11.2^**
More affected
Knee extensor	77.7 (5.9)	79.1 (5.9)	73.8 (5.9)	72.5 (5.9)	69.8 (5.9)	– 2.2^***	15.5^***
Knee flexors	39.9 (3.3)	40.9 (3.3)	39.6 (3.3)	38.0 (3.3)	39.7 (3.3)	– 0.3^ns	14.0^**

Estimated values are presented as mean (standard deviation; SD). ^*** p < 0.001, ^*p < 0.05, ^ns not significant, ^†Deviance (df = 2).

Over the four years, there was no significant deviation from the linear slopes for the muscle strength measurements, indicating that decline in muscle strength did not accelerate over time.

3.3 Predictors of change in gait performance

In Table 4, the final model of predictors of change in gait performance are presented including the significant covariates gender, age, BMI, length of time with new symptoms, baseline reduction in gait performance together with any interaction with time, and the person’s mean strength and within-person changes in strength for the knee extensor and flexors in the less affected and more affected lower limb.

Table 4
The final model of predictors of change in gait performance with significant covariates and interaction with time for the 51 participants with late effects of polio

Intercept Trend Gender Age BMI E6MWT PPS PPS Pmstr Pmstr Pmstr Wpstr

^*time moreex morefl lessfl lessfl

TUG (s) 10.9 – 0.04^ns – 0.01^** – 0.03^ns 0.02^* – 0.02^ – 0.04^*

Gait speed over 10 m

CGS (s) 9.3 0.06^ns – 0.68^** 0.10^* – 0.01^ – 0.01^ – 0.03^*

FGS (s) 6.3 0.12^ – 0.01^** – 0.04^ns 0.01^* – 0.03^*** – 0.02^* – 0.02^

6MWT (m) 433 – 3.6^ns 40.4^ – 3.7^ – 5.6^ 0.7^* 1.4^* 0.9^***

	Intercept	Trend	Gender	Age	BMI	E6MWT	PPS	PPS	Pmstr	Pmstr	Pmstr	Wpstr
TUG (s)	10.9	– 0.04^ns				– 0.01^**	– 0.03^ns	0.02^*	– 0.02^**			– 0.04^***
Gait speed over 10 m
CGS (s)	9.3	0.06^ns	– 0.68^**		0.10^*	– 0.01^**			– 0.01^**			– 0.03^***
FGS (s)	6.3	0.12^**				– 0.01^**	– 0.04^ns	0.01^*		– 0.03^***	– 0.02^*	– 0.02^**
6MWT (m)	433	– 3.6^ns	40.4^**	– 3.7^**	– 5.6^**	0.7^***				1.4^***		0.9^***

Intercept and trend for the significant factors in the final model for each gait performance test. TUG = Timed Up & Go; CGS = Comfortable gait speed; FGS = Fast gait speed; 6MWT = 6 minute walk test; Gender = woman is reference category; Age = The age at baseline; BMI = body mass index at baseline; E6MWT = the difference in meter between 6MWT and the estimated length at baseline; PPS = time with late effects of polio; Pmstr moreex = person mean strength for knee extension in the more affected lower limb; Pmstr morefl = person mean strength for knee flexion in the more affected lower limb; Pmstr lessfl = person mean strength for knee flexion in the less affected lower limb Wpstr lessfl = within person changes for knee flexion in the less affected lower limb. ^***p < 0.001, ^**p < 0.01, ^*p < 0.05; ^ns = not significant.

There were only significant trends for FGS (p < 0.01). Several baseline covariates as well as the person’s mean strength had a significant influence on the gait performance tests, with the same effect on the linear slope at any time-point. Gender had an effect on CGS and 6MWT with better gait performance for men. Older persons and persons with higher BMI walked shorter distance in 6MWT. Lower mean strength in the more affected knee extensors reduced gait performance for the TUG and the CGS, and lower mean strength in the knee flexors reduced the gait performance for the FGS and the 6MWT. Only two covariates interacted with gait performance over time: time with late effects of polio (for TUG and FGS; p < 0.05) and within-person changes in strength for the less affected knee flexors (all gait performance tests; p < 0.001).

4 Discussion

In this longitudinal study we assessed changes in gait performance annually over 4 years in persons with late effects of polio, and identified predictors of change in gait performance among the covariates gender, age, BMI, length of time with new symptoms, baseline reduction in gait performance and changes in the knee strength measurements over 4 years. We found a significant fixed linear effect over time for three gait performance tests (measuring gait speed and gait endurance) and a random effects for all gait performance tests, but no deviations from the linear slopes over the 4 years. The only predictors of change in gait performance over time were the individual variations in strength for the less affected knee flexors and time with new symptoms.

Four different tests were used to evaluate gait performance and for three of them we found a reduction over time: CGS, FGS and 6MWT (– 0.7% to – 1.7% per year; p < 0.05). In the study by Willén et al. (2007) the percentage decline in gait performance over four years was somewhat higher (– 1.3% to – 2.3% per year) than in our study. In other studies with less than 4 years follow-up time no change in gait performance over time has been reported (Klein, et al., 2008; Nollet, Beelen, Twisk, Lankhorst, & De Visser, 2003). However, with longer follow-up time the decline in gait speed has ranged between – 0.2% to – 0.6% per year (Bickerstaffe, et al., 2015; Sorenson, et al., 2005). This indicates that changes in gait performance over time for persons with late effect of polio are small and almost comparable with healthy persons. In a meta-analysis by Bohannon and coworkers (Bohannon & Williams Andrews, 2011) it was reported that the annual change in normal walking speed for persons between 50 to 79 years of age is around – 0.5%. However, even if the reduction in gait performance in the present study was low, we found significant random effects for all gait performance tests.

To be able to identify any predictors of change in gait performance over 4 years we formed a model with covariates. The predictors were chosen from different baseline time-invariant covariates and from the covariates of muscle strength measured each year. The final model revealed that four of the baseline covariates and three of the mean strength covariates were significantly related to some of the gait performance at baseline (see Table 4), with the same influence on the linear trend at any time point. We only found one baseline time-invariant covariate, length of time with new symptoms, which had an interaction with time (TUG and FGS); those persons who had had their late effects of polio for longer time declined more in gait performance over time. However, there was no significant relation between time with new symptoms and the gait performance at baseline. Another interesting finding was that the baseline reduction of gait performance did not have an interaction with time, so the effect was similar over time.

The mean knee muscle strength had different effects on the gait performance. For the TUG, that measures mobility including rising and sitting down in a chair, the mean strength in the more affected extensors had a significant and similar effect on gait performance at any time-point, but for the FGS and the 6MWT the mean strength in the knee flexors had a significant and similar effect on gait performance at any time-point. Thus, despite the significant decline in knee extensors strength over time (cf Table 3) the strongest predictor of change in gait performance over time was the within-person changes in knee flexor strength in the less affected limb. A person who loses more knee flexor muscle strength at any time compared to that person’s mean strength, the reduction in gait performance is significantly greater at that time point. To the best of our knowledge, only one previous study (Bickerstaffe, et al., 2015) has analyzed potential prognostic factors for a change in gait performance. These authors reported that different baseline factors had little influence on walking capacity and loss of isometric knee extensor strength (in average 15% reduction over 10 years) only explained a small proportion of the decline in walking. However, the results in that study are not quite comparable to our results, as they did not measure changes in knee flexor strength over time.

It has previously been shown that characteristics like gender, age, and BMI only marginally influence gait performance and that knee muscle strength, in particular the knee flexors, is related to gait speed and gait endurance, as assessed by FGS and 6MWT (Flansbjer, Brogårdh, & Lexell, 2013). In our longitudinal study of changes in muscle strength (Flansbjer, et al., 2015), gender had a significant interaction with time and was the strongest prognostic determinant for the decline in muscle strength over 4 years. However, as gender did not interact with time for gait performance, this indicates that the individual variations in knee flexor muscle strength are important for both men and women to explain the variance in gait performance over time. The importance of the knee flexor strength for gait performance in this population implies that strength training of the less affected knee flexors may improve gait speed and gait endurance, which could be beneficial for the performance of daily activities over time among persons with late effects of polio.

4.1 Methodological considerations, strengths and limitations

Several factors might have influenced the results in the present study. At baseline, the age of the participants varied from 52 to 74 years and their mean age was about 10 years older than in previous studies of persons with late effects of polio (Bickerstaffe, et al., 2015; Nollet, et al., 2003; Sorenson, et al., 2005). However, even if our participants were older at baseline a person’s age did not interact with time over four years, implying that the effects of being older were the same at any time point. Other factors that could have influenced the results are the frequency of test occasions and the length of follow-ups. In our study, all participants were measured annually over 4 years, the minimum follow-up time suggested for persons with late effects of polio (Stolwijk-Swuste, Beelen, Lankhorst, & Nollet, 2005). In previous studies, the follow-up times have varied as much as 2 to 15 years (Bickerstaffe, et al., 2015; Klein, et al., 2008; Nollet, et al., 2003; Sorenson, et al., 2005; Willen, et al., 2007). If the follow-up time is short, long-term changes are difficult to detect; but on the other hand, if annual measurements are performed, any progressive decline may be more easily detected.

In this study we used a statistical method (the MLM models) recommended for longitudinal clinical studies (Edwards, 2000; Hoffman & Stawski, 2009; Locascio & Atri, 2011). The MLM technique takes into account the correlation between the measures obtained in the same persons and compensates for missing observations. The MLM combines the effect of average time with any individual variation over time (fixed linear trends, random effects and changes in the slope). However, to be able to calculate deviations from the linear slopes over time at least 3 measurement points are required (Locascio & Atri, 2011). Furthermore, in the MLM predictors of change can be analyzed in the same model by creating time-invariant covariates from time-varying covariates. In previous longitudinal studies in persons with late effects of polio, older and more traditional statistical methods have been used and most participants have only been measured twice. The difference in statistical models may explain some of the differences in our results compared to other studies. To our knowledge, no study has used the MLM in a longitudinal study of decline in gait performance in persons with late effect of polio.

Four different gait performance tests were used in the present study (Flansbjer & Lexell, 2010a) to cover various aspects of gait performance, such as mobility, velocity and endurance. Overall, we found small changes in all gait performance tests over time, which could be due to both the study sample and the length of follow-up. The sample size of this study could be regarded as small but to collect data annual is time-consuming for both the participants and the assessors. During the 4 years of this study only 4 persons missed 8 out of a total of 260 test-sessions and this must be considered a low dropout rate, as the missing values are compensated for in the statistical model. Taking into account the very low dropout rate in this study, we believe the number of participants to be satisfactory for this type of analyzes.

All participants were ambulatory and had mild to moderate late effects of polio, so the results cannot be generalized to the entire population of persons with late effects of polio. To include persons with a wider range of disability, as well as a longer follow-up period would have been desirable. Moreover, it cannot be excluded that other factors such as pain, fatigue, physical activity or the size of motor units at baseline could have influenced gait performance over time. In the future, larger studies over longer periods are needed, which would enable more potential predictors to be included in the model to control for a wider spectrum of factors that might explain any changes in gait performance.

5 Conclusion

The small gradual reduction in gait performance in persons with late effects of polio is primarily determined by the individual variations in the knee flexor strength. This implies that maintained muscle strength in the knee flexors, primarily in the less affected lower limb, positively influences gait performance.

Conflict of interest

None to report.

Footnotes

Acknowledgments

We thank all persons who volunteered to participate in the study. The authors are also grateful to Vibeke Horstmann for statistical consultation. The study was accomplished within the context of the Centre of Ageing and Supportive Environments (CASE), Lund University, funded by the Swedish Council on Social Science and Working Life. Financial support was also received from Stiftelsen för bistånd åt rörelsehindrade i Skåne, the Alfred Österlunds Stiftelse, and Skane county council’s research and development foundation.

References

Bickerstaffe

, Beelen

, & Nollet

(2015). Change in physical mobility over 10 years in post-polio syndrome. Neuromuscul Disord, 25, 225–230.

Bohannon

R. W.

, & Williams Andrews

(2011). Normal walking speed: A descriptive meta-analysis. Physiotherapy, 97, 182–189.

Boyer

F. C.

, Tiffreau

, Rapin

, Laffont

, Percebois-Macadre

, Supper

, et al. Yelnik

A. P.

(2010). Post-polio syndrome: Pathophysiological hypotheses, diagnosis criteria, drug therapy. Ann Phys Rehabil Med, 53, 34–41.

Brehm

M. A.

, Nollet

, & Harlaar

(2006). Energy demands of walking in persons with postpoliomyelitis syndrome: Relationship with muscle strength and reproducibility. Arch Phys Med Rehabil, 87, 136–140.

Brogårdh

, Flansbjer

U. B.

, Espelund

, & Lexell

(2013). Relationship between self-reported walking ability and objectively assessed gait performance in persons with late effects of polio. NeuroRehabilitation, 33, 127–132.

Brogårdh

, Flansbjer

U. B.

, & Lexell

(2010). No effects of whole-body vibration training on muscle strength and gait performance in persons with late effects of polio: A pilot study. Arch Phys Med Rehabil, 91, 1474–1477.

Brogårdh

, Flansbjer

U. B.

, & Lexell

(2017). Determinants of falls and fear of falling in ambulatory persons with late effects of polio. PM R, 9(5), 455–463.

Brogårdh

, & Lexell

(2014). Falls, fear of falling, self-reported impairments, and walking limitations in persons with late effects of polio. PM R, 6, 900–907.

Edwards

L. J.

(2000). Modern statistical techniques for the analysis of longitudinal data in biomedical research. Pediatr Pulmonol, 30, 330–344.

10.

Enright

P. L.

, & Sherrill

D. L.

(1998). Reference equations for the six-minute walk in healthy adults. Am J Respir Crit Care Med, 158, 1384–1387.

11.

Flansbjer

U. B.

, Brogårdh

, Horstmann

, & Lexell

(2015). Men with late effects of polio decline more than women in lower limb muscle strength: A 4-year longitudinal study. PM R, 7, 1127–1136.

12.

Flansbjer

U. B.

, Brogårdh

, & Lexell

(2013). Muscle strength is only a weak to moderate predictor of gait performance in persons with late effects of polio. NeuroRehabilitation, 33, 457–464.

13.

Flansbjer

U. B.

& Lexell

(2010a). Reliability of gait performance tests in individuals with late effects of polio. PM R, 2, 125–131.

14.

Flansbjer

U. B.

& Lexell

(2010b). Reliability of knee extensor and flexor muscle strength measurements in persons with late effects of polio. J Rehabil Med, 42, 588–592.

15.

Gylfadottir

, Dallimore

, & Dean

(2006). The relation between walking capacity and clinical correlates in survivors of chronic spinal poliomyelitis. Arch Phys Med Rehabil, 87, 944–952.

16.

Halstead

L. S.

, Gawne

A. C.

, & Pham

B. T.

(1995). National rehabilitation hospital limb classification for exercise, research, and clinical trials in post-polio patients. Ann N Y Acad Sci, 753, 343–353.

17.

Hoffman

, & Stawski

(2009). Persons as contextx: Evaluating between-person and with-inperson effects in longitudinal analysis Research in Human Development. (Vol.6, pp.97-120): Taylor & Francis Group.

18.

Horemans

H. L.

, Bussmann

J. B.

, Beelen

, Stam

H. J.

, & Nollet

(2005). Walking in postpoliomyelitis syndrome: The relationships between time-scored tests, walking in daily life and perceived mobility problems. J Rehabil Med, 37, 142–146.

19.

Jensen

M. P.

, Alschuler

K. N.

, Smith

A. E.

, Verrall

A. M.

, Goetz

M. C.

, & Molton

I. R.

(2011). Pain and fatigue in persons with postpolio syndrome: Independent effects on functioning. Arch Phys Med Rehabil, 92, 1796–1801.

20.

Klein

M. G.

, Braitman

L. E.

, Costello

, Keenan

M. A.

, & Esquenazi

(2008). Actual and perceived activity levels in polio survivors and older controls: A longitudinal study. Arch Phys Med Rehabil, 89, 297–303.

21.

Larsson Lund

, & Lexell

(2008). Perceived participation in life situations in persons with late effects of polio. J Rehabil Med, 40, 659–664.

22.

Lehmann

, Sunnerhagen

K. S.

, & Willen

(2006). Postural control in persons with late effects of polio. Acta Neurol Scand, 113, 55–61.

23.

Lexell

, & Brogårdh

(2012). Life satisfaction and self-reported impairments in persons with late effects of polio. Ann Phys Rehabil Med, 55, 577–589.

24.

Locascio

J. J.

, & Atri

(2011). An overview of longitudinal data analysis methods for neurological research. Dement Geriatr Cogn Dis Extra, 1, 330–357.

25.

Nollet

, Beelen

, Twisk

J. W.

, Lankhorst

G. J.

, & De Visser

(2003). Perceived health and physical functioning in postpoliomyelitis syndrome: A 6-year prospective follow-up study. Arch Phys Med Rehabil, 84, 1048–1056.

26.

Sorenson

E. J.

, Daube

J. R.

, & Windebank

A. J.

(2005). A 15-year follow-up of neuromuscular function in patients with prior poliomyelitis. Neurology, 64, 1070–1072.

27.

Stolwijk-Swuste

J. M.

, Beelen

, Lankhorst

G. J.

, & Nollet

(2005). The course of functional status and muscle strength in patients with late-onset sequelae of poliomyelitis: A systematic review. Arch Phys Med Rehabil, 86, 1693–1701.

28.

Willen

, Sunnerhagen

K. S.

, Ekman

, & Grimby

(2004). How is walking speed related to muscle strength? A study of healthy persons and persons with late effects of polio. Arch Phys Med Rehabil, 85, 1923–1928.

29.

Willen

, Thoren-Jonsson

A. L.

, Grimby

, & Sunnerhagen

K. S.

(2007). Disability in a 4-year follow-up study of people with post-polio syndrome. J Rehabil Med, 39, 175–180.

Gender
Men; number (%)	28 (55)
Women; number (%)	23 (45)
Age (years): mean (SD); range	64 (6); 52 to 74
Age at acute polio infection (years): mean (SD); range	5 (4); 0 to 15
Time with new symptoms (years): mean (SD); range	13 (7); 2 to 28
Body mass index: mean (SD); range	26 (3); 21 to 37
Most affected leg: number (%)
Right	27 (53)
Left	24 (47)
Both lower limbs affected; number (%)	36 (71)
Walking aid: number (%)
None	47 (92)
Stick or crutch	3 (6)
Rollator	1 (2)
Orthotic device: number (%)
None	42 (82)
Ankle Foot Orthosis (AFO)	9 (18)
The performed 6 MWT in meter minus the predicted	– 80 (89); – 302 to 102
length in meter^a: mean (SD); range