Measuring muscle strength using isokinetic dynamometry allows evaluating and comparing normal and sick children, establishing recovery and rehabilitation goals, and quantitatively monitoring the course of a disease and the response to treatment. The purpose of this study was to carry out a systematic review and meta-analysis focusing on studies that examined the test-retest reliability of isokinetic knee strength measurements in children. This study is important because isokinetic dynamometry is the gold standard for evaluating muscle strength and it allows comparing muscle performance in children. The databases used were PubMed, Web of Science Scopus, and Embase (up to July 26, 2019). Only studies published in English were included in this review. All studies focused on the reliability of isokinetic knees in healthy children or those with cerebral palsy applied to dynamic contractions (concentric or eccentric) and provided measures of strength, reproducibility, ICC, peak torque, or SEM. We found a total of 143 abstracts and examined 94 articles to determine if they met the inclusion criteria. Finally, 10 articles were included in the systematic review and five studies (96 subjects) formed the meta-analysis sample, all of which focused on the reliability of isokinetic knees in the concentric mode. The CAT and QAREL scales were used to assess the quality of the included studies. The meta-analysis revealed high intra-class correlation coefficients (ICC) (0.84; p < 0.001; n = 96 subjects) in the flexion and excellent intra-class correlation coefficients (ICC 0.90; p < 0.001; n = 96 subjects) in the extension. Isokinetic dynamometry could be indicated as a method for measuring muscle strength training in children. However, the reviewed studies suggest some methodological issues in isokinetic tests, such as the rest days between testing and retesting, using the same speeds, protocols, and evaluators, and the performance of the subjects, so more research is required.
Citation: Muñoz-Bermejo L, Pérez-Gómez J, Manzano F, Collado-Mateo D, Villafaina S, Adsuar JC (2019) Reliability of isokinetic knee strength measurements in children: A systematic review and meta-analysis. PLoS ONE 14(12): e0226274. https://doi.org/10.1371/journal.pone.0226274
Editor: Wisit Cheungpasitporn, University of Mississippi Medical Center, UNITED STATES
Received: May 23, 2019; Accepted: November 22, 2019; Published: December 17, 2019
Copyright: © 2019 Muñoz-Bermejo et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
The term “muscle strength” refers to a muscle's or a group of muscles’ ability to exert maximum muscular force . Isokinetic dynamometry is the gold standard in muscle strength evaluation . It is an adequate system for assessment and diagnosis in the field of biomechanics. Measuring muscle strength allows evaluating and comparing normal and diseased children, establishing goals for recovery and rehabilitation, and quantitatively monitoring the course of a disease and the response to treatment . Therefore, following the safety standards and recommendations for children, isokinetic dynamometry is safe  and allows registering a range of angular velocities in both concentric and eccentric exercises .
Reliability is defined as the extent to which measurements can be replicated . Reliability can be presented in relative or absolute values. Relative reliability indicates the degree to which individuals maintain their position in a sample with repeated measurements ; the most common indicator of relative reliability is intra-class correlation coefficients (ICC) . In contrast, absolute reliability refers to the degree of conformity of the measures of a test from moment to moment . The most common indicators of absolute reliability are the standard error of measurements (SEM) and the smallest real difference (SRD).
The reliability of measuring isokinetic knee strength with a dynamometer in children has not been fully investigated. The majority of isokinetic strength reliability studies report the correlation coefficient as an indicator of the agreement between measurements . When reported, often only relative reliability is addressed, via ICC . The relative reliability of isokinetic knee strength measurements in this age group has been reported as moderate [2, 4] to high [10, 11], and this parameter has been given more importance in previous studies. However, absolute reliability, which refers to the degree of conformity in test measurements from one time-point to another, has not yet been determined in depth. This should be the next step for these studies to clarify clinically important changes for patients . Only two previous studies [4, 12] have given an approximation for the value of SRD. The present review presents more information on this variable. For the above reasons, it is important to know the reliability of isokinetic dynamometry.
Scattered information is available on the reliability of isokinetic measurements for children. Moreover, to our knowledge, no systematic reviews or meta-analyses on this topic have been published. Therefore, the purpose of this research was to perform a systematic review and meta-analysis of existing values of the test-retest reliability of isokinetic knee strength measurements in children, discuss potential limitations of the literature, and suggest recommendations for future research on statistical analyses for interpreting reliability. The current review and meta-analysis can provide valuable information for future guidelines and strategies for muscle strength reeducation in children.
Materials and methods
This systematic review meets the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (S1 File).
Test-retest studies published in English (up to July 26, 2019) that measured knee flexion and extension in children using an isokinetic dynamometer were included.
For this research, we applied further eligibility criteria. Eligible studies were those that assessed the reliability of isokinetic knees in healthy children or those with cerebral palsy applied to dynamic contractions (concentric or eccentric) and that provided measures of force, reproducibility, ICC, peak torque, or SEM.
We did not include studies that reported duplicate results.
The meta-analysis included studies on the reliability of knee movement in the concentric mode, both for the extension and for the flexion of the joint at a speed of 60°/s, that always included the reliability values of the dominant leg of the subjects.
Two of the authors (FM and JCA) independently examined and screened the titles and abstracts of the retrieved articles to assess study eligibility. Any disagreement or uncertainty was resolved through discussion. All reviewers reviewed the full-text articles that met the inclusion criteria or had uncertain eligibility. Any disagreement was resolved by consensus.
Electronic literature search
Several databases were searched, including PubMed, Web of Science, Scopus, and Embase.
A combination of MeSH terms was used for the PubMed search: child, knee, muscle strength, reproducibility of results, and validation studies.
The specific keywords used for the Web of Science, Scopus, and Embase were combinations of “isokinetic”, “knee”, “reliability”, “reproducibility”, “children”, “kids”, “boys”, “girls”, and “dynamometer”. The exact search was: ((children [Title/Abstract] OR kids [Title/Abstract] OR boys [Title/Abstract] OR girls [title/abstract]) AND knee [Title/Abstract] AND (reliability [Title/Abstract] OR retest [Title/Abstract] OR test-retest [Title/Abstract] OR reproducibility [Title/Abstract]) AND isokinetic [Title/Abstract]).
After examining the records, additional searches were conducted in health improvement sources, meta-search engines (Google/Google scholar), and on the Retraction Watch website to identify additional publications and gray literature.
Isokinetic movement is defined by maintaining an angular velocity of constant movement throughout the joint path. The current isokinetic dynamometry system allows evaluating in both concentric and eccentric modes of exercise. The dynamometer shows the value of the moment of force developed at each instant. The most important data recorded by the isokinetic dynamometer is the peak torque or maximum moment of force, which indicates the highest value of force recorded during the test. Another variable an isokinetic dynamometer provides is “work”, which expresses the product of the moment of force and angular distance. Peak torque was used in this review.
Evaluation of the quality of the included studies
The CAT scale is an instrument developed specifically to evaluate the methodological quality of studies, considering the validity and reliability of the objective clinical tests. The CAT scale contains 13 evaluation items. Four of the 13 items refer to validity issues, but the other nine refer to reliability; therefore, only these nine were used for this review. Each article was classified as "yes" when information was described in sufficient detail or "no" when there was not enough information for clarification . A final percentage (%) evaluation column was added based on the items that each study achieved. Thus, the maximum possible score was 90%, which represents the highest methodological quality. Studies were considered high quality if they scored above 45%.
The Quality Appraisal for Reliability Studies (QAREL) scale is a quality assessment tool for diagnostic reliability studies. It consists of 11 questions (meets, does not meet, doubtful or not applicable) grouped into three categories of internal validity (items 3–9), external validity (items 1, 2, and 10), and the relevance of the statistical analyses (item 11). The maximum score is 110%.
The meta-analysis included only five articles [2, 4, 10–12] of the ten incorporated in this review. This meta-analysis focused on the reliability of the concentric mode, both for extension and knee flexion. The choice of these items was because the studies all used a similar speed, 60°/s, always with the values of the dominant leg of the subjects. In addition, we considered the sample size and the type of participants; all were healthy except for those in Moreau’s study, whose subjects had cerebral palsy (CP) .
The random-effects model and the fixed-effects model were used to combine standardized effect sizes with a 95% confidence interval.
The funnel plot and Egger´s weighted regression tests were used to evaluate possible publication bias (p ˂ 0.1 was considered statistically significant publication bias).
Regarding the reliability indicators used in this review, we started with the ICC, which is usually the main indicator of reliability. All ICCs collected in our review are shown at a 95% confidence interval (CI) .
The SEM was calculated from the square root of the mean error term derived from the analysis of variance (ANOVA)  and was used to determine the minimum difference that is considered important for a single subject . SEM was calculated as , where SD is the standard deviation of day 1 and day 2 .
The percentages of SEM and SRD are shown to represent the error of measurement in relative terms and thus allow comparing the different variables . This was calculated with: [SEM or SRD / mean of all values] .
Search strategy and quality of studies
A total of 143 studies were identified through searches in electronic databases. Among these studies, we identified and eliminated 49 duplicates. Only 14 studies met the inclusion criteria, but four of these studies were excluded. One study was excluded because it presented different ages, another because it did not evaluate the knee joint, and another two because they did not perform test-retest. A manual search of the bibliographies of the relevant articles revealed no additional studies. Therefore, ten prospective studies were included in this systematic review on isokinetic knee movement in children (Fig 1).
The quality of the articles according to the CAT quality score varied between 23% and 78%, with the maximum possible being 90%. According to the analysis, eight articles were evaluated as high quality (Table 1). The quality of the articles varied according to the designs of the studies.
According to the QAREL scale, the quality of the articles varied between 30% and 100%, with the maximum possible being 110% (Table 2). Evaluation data indicate that eight articles scored between 60 and 100%.
Characteristics of the studies
Table 3 shows the characteristics of the participants. The sample sizes (n) of the final ten studies ranged from 12 to 39 participants, with participants aged 5–15 years. Some of the studies compared the dominant side with the non-dominant side. The joint evaluated in all studies was the knee. All studies evaluated the segments using computerized isokinetic dynamometers: Lido—Active, Cybex II, Cybex Norm, Biodex 6000, Biodex System 3, and Biodex System 4.
The contraction speeds ranged from 15°/s to 247.5°/s (4.32 rad/s) for the evaluation of the concentric mode and 60 to 180°/s for the eccentric mode.
To classify the strength of reliability, we used the scale proposed by Landis and Koch : 0 represents poor reliability, 0.01–0.20 slight reliability, 0.21–0.40 regular reliability, 0.41–0.60 moderate reliability, 0.61–0.80 substantial reliability, and 0.81–1.00 very high (almost perfect) reliability.
All selected articles presented strength measurements for both knee flexion and extension in the concentric mode. Of these, only three [4, 5, 11] analyzed knee flexion and extension in the eccentric mode as well. Four studies [2, 5, 11, 12] included tests on both dominant and non-dominant legs. Studies found substantial and very high (almost perfect) reliability for all analyses, regardless of movement, speed, type of contraction, evaluation of the dominant or non-dominant leg, and whether the participants had CP or not.
The Pierce’s study identified regular and moderate reliability where the knee flexion was assessed at 15°/s and 90°/s with ICCs of 0.31 and 0.38, respectively. This study also identified moderate reliability, because it measured knee extension assessed at 15°/s and at 90°/s with ICCs of 0.51 and 0.50, respectively. Both movements were in the concentric mode . Fagher also measured knee flexion at 60°/s and 180°/s with ICCs of 0.62 and 0.49, respectively. This study, although it showed substantial reliability when measuring knee extension at 180°/s with an ICC of 0.68 in the eccentric mode, showed moderate reliability (0.60) for knee flexion  (Table 4).
There were many differences in the values of mean and standard deviation of peak torque (Tables 4 and 5). For example, in Fagher et al. , in the concentric mode, the values of the second measure were better than the first one for knee extension and for knee flexion. However, in the eccentric mode, the values of the second measure were worse than the first one for both knee extension and knee flexion.
We performed a meta-analysis of the five studies for the reliability of knee movement in the concentric mode for both the extension and flexion of the joint of the subjects’ dominant leg at a speed of 60°/s.
In this aspect, the aforementioned reliability was observed in more detail. For the knee extension, the correlation coefficient was 0.89 for fixed effects and 0.90 for random effects (Table 6).
For knee flexion, this correlation coefficient reached values of 0.82 for fixed effects and 0.84 for random effects.
Fig 2 shows the values mentioned in the above table in a more schematic way. This figure allows quickly comparing each of the correlation coefficients, both for fixed effects and for random effects, between the concentric extension and flexion of the knee.
Extension of knee (first figure) and knee flexion (second figure) in the concentric mode.
Regarding the heterogeneity of each action of the knee, we emphasize:
- For knee extension, a value of Q = 12.59; I2 = 68.22; Significance (p) = 0.0135; 95% of the IC for I2 = 18.05–87.68.
- For knee flexion, a value of Q = 10.29; I2 = 61.14%; Significance (p) = 0.0358; 95% of the IC for I2 = 0.00–85.41.
Fig 3 shows funnel plots corresponding to knee extension and flexion in the studies.
The Egger test was used to evaluate the asymmetry of the funnel plot. The results of the Egger test for both knee extension and flexion were not significant (p = 0.109 and p = 0.136), suggesting that there is no publication bias.
The knee is one of the most studied joints in isokinetic dynamometry research, probably because it is easy to evaluate using a dynamometer . Knee extensor and flexor strength are recognized as important for daily tasks, e.g., moving, standing or sitting, weight lifting, and climbing stairs . In addition, many studies of knee extensors indicate that these could be representative of the total strength of the lower extremities. It is not only important to analyze the absolute values of flexors or extensors, but also the balance of both at the same time if the contralateral deficit is a compensated deficit; i.e., there is an adequate agonist/antagonist relationship . Isokinetic dynamometry offers clear advantages over simple measurements to assess muscle strength, such as the continuous and accurate measurement of force production across a range of velocities of movement .
As we have indicated in the results, there were different heterogeneities for the test-retest data in each of the articles: they showed high and moderate values. Most authors thought that the results for the second tests would be better due to a learning effect [8, 24]. Some studies indicated a need for additional practice and familiarization with test procedures in this age group . Other studies emphasized that the isokinetic method has the advantage of making it possible to detect large differences in the same muscles when analyzing at peak torque .
We analyzed the studies that reported reliability for the assessment of muscle strength in healthy children with CP. Our review suggests that the studies had good-to-excellent levels of reliability of muscle strength assessment in dominant and non-dominant knees using isokinetic dynamometry (0.7–1 correlation), except for Pierce et at. , which had low and moderate intra-class correlations at the velocities 15°/s and 90°/s, and Fagher et al. , where most results were moderate. Our results also support the contention that the muscle strengths of healthy children and children with CP are quantifiable and measurement is reliable, even in the presence of spasticity.
Otherwise, there were four studies whose participants were children with CP [10, 20, 21, 27] and one study whose participants were children without CP  where the values for the second test decreased compared to the first test in the concentric mode, but only for knee extension. The ICCs of these articles, except for Pierce et al. , were high for knee extension (Table 4). These results demonstrate that voluntary muscle fatigue of the knee flexors and extensors can be reliably and feasibly assessed in children. It is believed that the number of repetitions and the type of contraction may influence the assessment of peak voluntary torque in subjects with CP. A possible explanation for this could be that the coordination of the agonist and antagonist muscles is more impaired at higher velocities than at lower velocities in children with CP . It is suggested that there is higher heterogeneity for children with CP than for their healthy peers. The tests may be less suitable for children with CP than for healthy children because of attention deficits, which are known to occur with CP . Future research should establish normative values regarding spasticity of the knee flexion and knee extension in children with CP to allow for the clearest interpretations of clinically meaningful change in these subjects.
This review emphasizes the need to find a consensus around the minimum change necessary to indicate a clinically important change for an individual or a particular population. Analyzing the only two articles that included SRD [4, 12] led us to suggest that: 1) for the knee extension in the concentric mode, the minimum change is around 15–15.5 Nm in absolute value and 31% in relative value with a speed of 60°/s and 34.5% with a speed of 180°/s; in the eccentric mode, the SRD should be 49.8%. 2) For knee flexion in the concentric mode, the minimum change is around 9.8–13.0 Nm in absolute value and 36.5–38.5% in relative value; in the eccentric mode, the SRD should be 59.6%.
Most of the studies had a time interval of one week between tests. One study had a three-days interval  and another two studies had time intervals lower than two hours [20, 21]. It is reasonable to establish a one-week interval because intervals of approximately 1 week between tests could maximize the effects of learning while still managing any effects of muscle fatigue .
A meta-analysis evaluates the replicability and generalizability of results, which are the hallmarks of good science . This review presents a meta-analysis for the reliability of each of the included studies. The results of our analyses are shown in Tables 3 and 4. We hope to provide more accurate and definitive values that will guide future research on the population treated here. Table 6 shows that the reliability varies. However, reliability is high for all the studies analyzed: for knee extension, reliability reaches a high—good level (0.89) for fixed effects and an excellent—very high level (0.90) for random effects; for knee flexion, reliability reaches a high—good level (0.82–0.84) for both fixed and random effects, which are lower values than for extension. The study of Kellis et al.  is the most reliable for extension, with an intra-class correlation coefficient of 0.98 and a range of 0.933–0.994 for 95% of the CI. For flexion movement, Moreau et al.’s study  meets the high expectations seen in Tables 4 and 5 for articles dealing with subjects with CP: this article has a correlation coefficient of 0.96 and a range of 0.860 to 0.989 for 95% of the CI. According to the meta-analysis, the relative reliability is high.
The methodological quality of this review, shown in Tables 1 and 2, means it is a very useful tool for comparing the studies and allows drawing conclusions about the reliability of the studies. First, the percentages of items met (Tables 1 and 2) indicate the limitations of the articles depending on the requested variables. Only one article reaches 78% quality , in CAT scale, and three are below 50%. The final mean of the percentages is approximately 53%, meaning that only half of the variables are fulfilled. In QAREL scale only one article reaches 100% quality and eight articles are above 55%.
The meta-analysis shows that one study has a 78% (CAT scale) and 100% (QAREL scale) score for methodological quality , another had 67% and 70% , and three studies had 56% and 60% [4, 10, 11]. The meta-analysis (5 studies) has a small sample (n = 96), but these results seem very interesting, since the reported reliability coefficients are high and excellent. High reliability coefficients are usually reported by studies with small sample sizes, which are associated with large standard errors and higher than normal CIs .
The main limitation is the heterogeneity (isokinetic devices, populations, protocols…) of the identified studies. Because of the relatively small numbers of eligible studies, and the variability in the statistics used between studies, there were insufficient numbers to allow a meta-analysis for all velocities and the eccentric mode. For the same reason, a sub-analysis of the possible subgroups could not be performed. Of the five articles selected, four treated healthy children and only one study evaluated children with cerebral palsy. As for another possible subgroup, two articles evaluated soccer-playing children and three evaluated children who were not soccer players. As stated above, one evaluated children with cerebral palsy.
Publication bias exists in any literature review and should be considered when interpreting the results. Measures were taken to minimize bias as much as possible, including conducting a comprehensive search of multiple databases with gray literature databases. Only articles published in English and in peer-reviewed journals were included, which may have reduced the number of results. In addition, this meta-analysis was not registered online.
The present review and meta-analysis explores existing data about different results in the test-retest reliability of isokinetic knee strength measurements in healthy and with CP children. This study provides a reliable analysis of isokinetic knee force measurements in the concentric mode. These findings suggest that isokinetic dynamometry can be used not only for measuring muscle strength for training, but also for muscle strength reeducation programs in children with or without CP.
More studies are needed in children with different pathologies to confirm our results. In addition, future research should evaluate knee force in the eccentric mode and show the reliability of test-retest measurements.
- 1. De Ste Croix M, Deighan M, Armstrong N. Assessment and interpretation of isokinetic muscle strength during growth and maturation. Sports medicine (Auckland, NZ). 2003;33(10):727–43. Epub 2003/08/05. pmid:12895130.
- 2. Santos AN, Pavao SL, Avila MA, Salvini TF, Rocha NA. Reliability of isokinetic evaluation in passive mode for knee flexors and extensors in healthy children. Braz J Phys Ther. 2013;17(2):112–20. Epub 2013/06/20. pmid:23778769.
- 3. Merlini L, Dell'Accio D, Granata C. Reliability of dynamic strength knee muscle testing in children. J Orthop Sports Phys Ther. 1995;22(2):73–6. Epub 1995/08/01. pmid:7581434.
- 4. Fagher K, Fritzson A, Drake AM. Test-retest reliability of isokinetic knee strength measurements in children aged 8 to 10 years. Sports health. 2016;8(3):255–9. pmid:26895853
- 5. Iga J, George K, Lees A, Reilly T. Reliability of assessing indices of isokinetic leg strength in pubertal soccer players. Pediatric exercise science. 2006;18:436–45.
- 6. Daly L, Bourke GJ. Interpretation and uses of medical statistics: John Wiley & Sons; 2008.
- 7. Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports medicine (Auckland, NZ). 1998;26(4):217–38. Epub 1998/11/20. pmid:9820922.
- 8. Dvir Z. Isokinetics: muscle testing, interpretation, and clinical applications. Sciences EH, editor2004.
- 9. Parsons JL, Porter MM. Reliability of measuring hip and knee power and movement velocity in active youth. Pediatric physical therapy: the official publication of the Section on Pediatrics of the American Physical Therapy Association. 2015;27(1):82–9. Epub 2014/12/19. pmid:25521270.
- 10. Moreau N, Li L, Damiano DL. A feasible and reliable muscle fatigue assessment protocol for individuals with cerebral palsy. Pediatric physical therapy: the official publication of the Section on Pediatrics of the American Physical Therapy Association. 2008;20(1):59–65. Epub 2008/02/28. pmid:18300935.
- 11. Kellis E, Kellis S, Gerodimos V, Manou V. Reliability of isokinetic concentric and eccentric strength in circumpubertal soccer players. Pediatric Exercise Science. 1999;11:218–28.
- 12. Johnsen MB, Eitzen I, Moksnes H, Risberg MA. Inter- and intrarater reliability of four single-legged hop tests and isokinetic muscle torque measurements in children. Knee Surg Sports Traumatol Arthrosc. 2015;23(7):1907–16. Epub 2015/08/19. pmid:26284271.
- 13. Brink Y, Louw QA. Clinical instruments: reliability and validity critical appraisal. Journal of evaluation in clinical practice. 2012;18(6):1126–32. Epub 2011/06/22. pmid:21689217.
- 14. Lucas NP, Macaskill P, Irwig L, Bogduk N. The development of a quality appraisal tool for studies of diagnostic reliability (QAREL). J Clin Epidemiol. 2010;63(8):854–61. pmid:20056381.
- 15. Cochran WG. The combination of estimates from different experiments. Biometrics. 1954;10(1):101–29.
- 16. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557–60. pmid:12958120.
- 17. Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res. 2005;19(1):231–40. Epub 2005/02/12. pmid:15705040.
- 18. Schuck P, Zwingmann C. The 'smallest real difference' as a measure of sensitivity to change: a critical analysis. Int J Rehabil Res. 2003;26(2):85–91. Epub 2003/06/12. pmid:12799601.
- 19. Kramer MS, Feinstein AR. Clinical biostatistics. LIV. The biostatistics of concordance. Clin Pharmacol Ther. 1981;29(1):111–23. pmid:7460469.
- 20. Pierce SR, Lauer RT, Shewokis PA, Rubertone JA, Orlin MN. Test-retest reliability of isokinetic dynamometry for the assessment of spasticity of the knee flexors and knee extensors in children with cerebral palsy. Archives of physical medicine and rehabilitation. 2006;87(5):697–702. Epub 2006/04/26. pmid:16635633.
- 21. van den Berg-Emons RJ, van Baak MA, de Barbanson DC, Speth L, Saris WH. Reliability of tests to determine peak aerobic power, anaerobic power and isokinetic muscle strength in children with spastic cerebral palsy. Dev Med Child Neurol. 1996;38(12):1117–25. Epub 1996/12/01. pmid:8973297.
- 22. Huesa JF., García D J., Montes J. Dinamometría isocinética. Rehabilitación. 2005;39:288–96.
- 23. Rabelo M, Nunes GS, da Costa Amante NM, de Noronha M, Fachin-Martins E. Reliability of muscle strength assessment in chronic post-stroke hemiparesis: a systematic review and meta-analysis. Topics in stroke rehabilitation. 2016;23(1):26–36. Epub 2015/08/06. pmid:26243549.
- 24. Eng JJ, Kim CM, Macintyre DL. Reliability of lower extremity strength measures in persons with chronic stroke. Archives of physical medicine and rehabilitation. 2002;83(3):322–8. Epub 2002/03/12. pmid:11887111.
- 25. Jones , Stratton G. Muscle function assessment in children. Acta Paediatr. 2000;89(7):753–61. Epub 2000/08/16. pmid:10943949.
- 26. Perrin DH. Isokinetic exercise and assessment: Human Kinetics; 1993.
- 27. Ayalon M, Ben-Sira D, Hutzler Y, Gilad T. Reliability of isokinetic strength measurements of the knee in children with cerebral palsy. Dev Med Child Neurol. 2000;42(6):398–402. Epub 2000/06/30. pmid:10875525.
- 28. Binder H, Eng GD. Rehabilitation management of children with spastic diplegic cerebral palsy. Archives of physical medicine and rehabilitation. 1989;70(6):482–9. Epub 1989/06/01. pmid:2658915.
- 29. Woodson C, Bandy WD, Curis D, Baldwin D. Relationship of isokinetic peak torque with work and power for ankle plantar flexion and dorsiflexion. J Orthop Sports Phys Ther. 1995;22(3):113–5. Epub 1995/09/01. pmid:8535468.
- 30. Bornmann L, Mutz R, Daniel HD. A reliability-generalization study of journal peer reviews: a multilevel meta-analysis of inter-rater reliability and its determinants. PLoS One. 2010;5(12):e14331. Epub 2010/12/24. pmid:21179459.