Article Text
PDF
Abstract
Background The Knee injury and Osteoarthritis Outcome Score (KOOS) is a self-administered valid and reliable questionnaire for adults with joint injury or degenerative disease. Recent data indicate a lack of comprehensibility when this is used with children. Thus, a preliminary KOOS-Child was developed. This study aims to evaluate psychometric properties of the final KOOS-Child when used in children with knee disorders.
Methods 115 children (boys/girls 51/64, 7–16 years) with knee disorders were recruited. All children (n=115) completed the KOOS-Child, the Child-Health Assessment Questionnaire (CHAQ) and the EQ-5D-Youth version (EQ-5D-Y) at baseline to evaluate construct validity. Two additional administrations (1–3 weeks and 3 months) were performed for analyses of reliability (internal consistency and test–retest; n=72) and responsiveness (n=91). An anchor-based approach was used to evaluate responsiveness and interpretability.
Results After item reduction, the final KOOS-Child consists of 39 items divided into five subscales. No floor or ceiling effects (≤15%) were found. An exploratory factor analysis on subscale level demonstrated that items in all subscales except for Symptoms loaded on one factor (Eigenvalues 3.1–5.5, Symptom: 2 factors, Eigenvalue >1). Sufficient homogeneity was found for all subscales (Cronbach's α = 0.80–0.90) except for the Symptoms subscale (α = 0.59). Test–retest reliability was substantial to excellent for all subscales (Intraclass Correlation Coefficient 0.78–0.91, Smallest Detectable Change (SDC)ind 14.6–22.6, SDCgroup 1.7–2.7). Construct validity was confirmed, and greater effect sizes were seen in those reporting improved clinical status. Minimal important changes greater than the SDCs were found for patients reporting to be better and much better.
Conclusions The final KOOS-Child demonstrates good psychometric properties and supports the use of the KOOS-Child when evaluating children with knee disorders.
- Knee Injuries
- Physiotherapy
- Children
- Evaluation
- Orthopaedics
Statistics from Altmetric.com
Introduction
The knee is one of the most common sites for musculoskeletal injury in younger people.1–3 Serious knee injuries in children are frequently associated with long-term symptoms, functional impairment and disability.3 Musculoskeletal outcome measures developed for adults are often used for children. Recent data indicate a lack of comprehensibility when the adult Knee injury and Osteoarthritis Outcome Score (KOOS) is used with children.4 These data led to the development of the paediatric KOOS (KOOS-Child). Modifications related to comprehension, mapping of responses and jargon were made based on qualitative feedback from children.4 The KOOS-Child is an instrument that assesses paediatric patients’ opinion about their knee and associated problems. The KOOS-Child is designed to assess and monitor groups and individuals over short-time as well as long-time intervals. The comprehensibility and content validity of the KOOS-Child were ensured through cognitive interviews with 34 Swedish children aged 10–16 years with a symptomatic knee disorder.4 The KOOS-Child is available in Swedish, Danish and English (http://www.koos.nu). Until now the psychometric properties of the KOOS-Child have not been determined in a clinical population. The overall aims of this study were to refine the preliminary KOOS-Child through the process of deleting redundant items and to test the final KOOS-Child for construct validity, reliability, responsiveness and interpretability in children with knee disorders.
Methods
All definitions and psychometric properties evaluated in this study follow the recommendations of the COSMIN (COnsensus-based Standards for the Selection of health Measurement Instruments) checklist, an international consensus protocol.5 ,6
Subjects
Children seeking medical care due to knee symptoms were recruited. A flowchart of the patients included for the different analyses is presented in figure 1. Patients were recruited from primary and secondary care clinics in Stockholm, Sweden, by an orthopaedic surgeon or a physiotherapist and screened for study eligibility following a protocol. Inclusion criteria were: (1) seeking medical care and/or receiving treatment because of acute or chronic knee-related problems, (2) restricted in physical activities, (3) able to independently read and complete the forms by themselves, (4) Swedish as native language and (5) no comorbidities over-riding knee symptoms. Parents or legal guardians provided written and verbal consent to participate.
Flowchart over included participants in the different steps of evaluation of the KOOS-Child.
Study design
At baseline, when visiting their doctor/physiotherapist, all respondents completed the 48-item preliminary version of KOOS-Child,4 the Child-Health Assessment Questionnaire (CHAQ),7 the EQ-5D-Youth version (EQ-5D-Y)8 ,9 and five specific Visual Analogue Scale (VAS) formatted questions to determine construct validity. The VAS questions ranging from 0mm to 100mm (no problems to worst possible problems) were specifically designed for this study to represent the different subscales of the KOOS-Child. These were designed as follows: ‘How much knee problems (swelling, moving your knee, etc) do you have?’, ‘How much pain do you have in your knee?’, ‘How do you think your knee works in activities in daily life (climbing stairs, rising, etc)?’, ‘How do you think your knee works when you are playing or doing sports?’ and ‘How has your knee injury affected your life?’.
Second (after 1–3 weeks) and third (after 3 months) administrations of the preliminary KOOS-Child were mailed to evaluate reliability and responsiveness (figure 1). At these administrations all patients were also asked to rate change in their knee condition since last administration using one generic and five subscale specific Global Perceived Effect (GPE) scores, as external criteria for clinically relevant change.10 The seven-item GPE-scoring scale ranged from ‘much worse’ to ‘much better’. If needed, patients were reminded once by telephone or email to complete the forms.
The KOOS-Child
The preliminary KOOS-Child4 consisted of 48 items divided into the same five subscales as the original KOOS: Pain (11 items), other Symptoms (7 items), Activities in daily living (ADL; 17 items), Function in sports and play (Sports/Play; 7 items) and knee-related quality of life (QOL; 6 items).4 The time frame considered when answering the questionnaire is over the prior week. Scoring is based on Likert scales and each question receives a score from 0 (no symptoms) to 4 (extreme symptoms). More than 50% of the items for a subscale must be completed to calculate a subscale score. The scores for each subscale are normalised to a score ranging from 0 (extreme symptoms) to 100 (no symptoms).11 ,12 Detailed scoring information is available at http://www.koos.nu. The final KOOS-Child takes about 10–20 min to complete depending on age and reading level.
Psychometric properties—definitions and statistical considerations
Confirmation of subscales and process for item deletion
First and second administrations of the 48-item preliminary KOOS-Child were used for the item reduction process.4 Predefined criteria for possible deletion of an item from the preliminary version were: a ceiling effect of 70% or above (at baseline; ie, ≥70% of the respondents answered that they had no problem on item level) and/or an Intraclass Correlation Coefficient (ICC) of less than 0.41 (test–retest)13 and/or >5% missing responses (at baseline).13 The research team met to agree on the deletion of items based on the above criteria.
To evaluate whether subscale scores were an adequate reflection of the construct to be measured, we performed an exploratory factor analysis using principal component analysis with varimax rotation.14 This procedure was conducted to evaluate subscale unidimensionality. Eigenvalues above 1, an index of the amount of variance accounted for by each factor, was set as the limit.15
Evaluation of the final version of the KOOS-Child
Aspects of reliability (test–retest, internal consistency, measurement error)
A test–retest limit of 3 weeks was established to ensure a stable condition and prevent recall. Children who rated their condition on the generic GPE score as ‘somewhat better’, ‘no change’ or ‘somewhat worse’ were included in analyses. An ICC2.1 was calculated for all subscales of the final KOOS-Child. ICCs were interpreted according to recommendations by Landis and Koch13 where a value of 0.81–1.0 represents almost perfect agreement, 0.61–0.80 substantial agreement, 0.41–0.60 moderate agreement, 0.21–0.40 fair agreement and 0–0.20 poor agreement.
For each subscale of the final version, a Chronbach’s α of the baseline values was calculated to evaluate internal consistency6 and a score above 0.70 was set as an indication of sufficient item homogeneity.16
The measurement error was evaluated by the Standard Error of Measurement agreement (SEM) calculated as (√mean square—within targets).6 To determine the clinical change apart from the measurement error the Smallest Detectable Change (SDC) was calculated at individual (SEM×1.96×√2) and group levels (SEM×1.96×√2×√n).17 ,18
Construct validity
Construct validity is evaluated by testing a predefined hypothesis between the present and other valid questionnaires that are assumed to measure similar or dissimilar constructs.6 Since there are no other questionnaires available in Swedish measuring knee outcomes in children, other scores measuring similar and dissimilar constructs were used. Construct validity was determined by correlating the final KOOS-Child subscale scores with the subscales and items of the CHAQ, the EQ-5D-Y and with the five subscale-specific VAS questions constructed specifically for this study.
The CHAQ is an instrument measuring self-assessed disabilities in children with inflammatory joint diseases.7 It contains eight domains that evaluate the degree of difficulty in performing activities of daily living. The eight domains are summarised in one disability index and complemented by two VAS scores. A shorter CHAQ was used including the domains ‘pain’, ‘walking’, ‘arising’, ‘grip’ and one item from the ‘activity’ domain. These domains are scored separately; ‘pain’ on a VAS score and the rest on an ordinal scale ranging from 0 (best) to 3 (worst).7
The EQ-5D-Y measures health-related QOL in children.8 The EQ-5D-Y consists of five items: mobility, self-care, usual activities, pain and discomfort and anxiety and depression. Each item has three response categories, ‘no problem’, ‘some problem’ and ‘a lot of problems’. It is not recommended yet to calculate a total score of the EQ-5D-Y. Therefore, data are presented as the percentage of individuals reporting level of problem for each item. In this study, the EQ-5D-Y domain ‘doing usual activities’ was used (for correlation with the KOOS-Child QOL subscale). The EQ-5D-Y also includes a VAS score for rating overall health status; however, it was not included in this study.9 The CHAQ and the EQ-5D-Y are reliable and valid instruments when used in children.7 ,9
The five previously described specific VAS questions were used for determination of construct validity. These were developed due to a lack of existing appropriate measures.
Correlations between the final KOOS-Child subscales and the other questionnaires were assessed at baseline using the Spearman Correlation Coefficients.15 We considered correlation coefficients >0.5 as strong, 0.35–0.5 moderate and <0.35 weak.19 In accordance with the COSMIN guidelines, we defined construct validity of the KOOS-Child to be good if 75% of our hypotheses were confirmed.16 The a priori hypotheses are outlined in table 1.
Construct validity
Responsiveness
Responsiveness was evaluated using a questionnaire mailed after 3 months, a time point chosen when a clinical improvement in the study sample could be expected. Changes in the final KOOS-Child subscale scores between baseline and 3 months administration were assumed to correlate with the subscale-specific GPE scores according to our a priori hypotheses.6 A Spearman's correlation coefficient was used and correlation was set to be ≥0.3.20 Standardised response means (baseline score—follow-up score/SD change in score) and effect sizes (baseline score—follow-up score/SD of baseline) were also calculated15 and were hypothesised to be higher for patients who reported improvement than patients reporting their condition to be stable according to the generic GPE scores. Effect sizes of 0.2, 0.5 and 0.8 are interpreted as small, medium or large, respectively.21
Interpretability
Interpretability was evaluated by examining the distribution of total and change KOOS-Child subscale mean scores, floor/ceiling effects and evaluation of minimal important change (MIC). A floor or ceiling effect was considered present if >15% of the respondents reported the lowest/worst (0) or the highest/best score (100) for a subscale.16 MIC has been defined as the smallest score change, which patients perceive as important in the construct to be measured.5 MIC was determined with an anchor-based method employing the subscale-specific GPE-scores for each subscale of the KOOS-child.22 The mean change method described by Jaeschke et al10 was used. Children reporting to be ‘better’ or ‘much better’ were considered to experience a clinically meaningful change. MIC was defined separately for these two groups as the mean change in score between baseline and 3-month follow-up and complemented by a 95% CI (mean change±1.96 (SDchange /√n)).10 ,23
All statistical analyses were performed using Statistica software V.10 and SPSS V.19.
Results
Subjects
In total, 115 children (boys/girls 51/64) aged 7–16 years (mean 13.01, SD 1.85) seeking medical care due to knee symptoms were recruited. Patient characteristics are presented in table 2. Forty-three children (37%) were not included in the test–retest analyses due to missing/late responses, missing GPE or a report of an unstable knee condition. In total, 24 (21%) were not included in the responsiveness analyses due to missing responses, missing GPE or incomplete forms (figure 1). The percentage of missing responses at item level as well as at subscale level are presented for baseline and follow-up in tables 3 and 4, respectively. In total, 4–7% and 4–6% missing responses were found for the CHAQ subscales and items and the EQ-5D-Y items, respectively. Four per cent missing responses were found for the subscale-specific VAS questions.
Patient characteristics at baseline
Preliminary KOOS-Child item characteristics at baseline distribution (n=115)
Subscale characteristics of the final KOOS-Child at baseline
Confirmation of subscales
Descriptive statistics of the preliminary KOOS-Child baseline data for the 115 children are presented in table 3. No items had a missing response rate >5% and all items had an ICC value >0.53, which was above the prespecified criterion. Ceiling effects ≥70% were found for nine items (P5, P7, P8b, A4, A6, A8, A9, A11, A15) in the preliminary version (table 3). Based on these results and following discussions the research group decided that all nine items should be deleted. Thus, 39 items were kept in the final version.
The five subscale structure was maintained and the distribution was as follows; Pain (8 items), other Symptoms (7 items), ADL (11 items), Function in sports and play (Sport/Play; 7 items) and knee-related QOL (6 items). An exploratory factor analysis of the final version showed that the items in Pain, ADL, Sports and QOL subscales loaded on one factor with Eigenvalue of 4.0 (50% of the variance), 5.5 (50% of the variance), 4.3 (61% of the variance) and 3.1 (52% of the variance), respectively. The Symptoms subscale loaded on two factors with Eigenvalues >1 (factor 1 representing more chronic problems such as noise, mechanical problems and stiffness (S2, S3, S6, S7, Eigenvalue 2.1, 30% of the variance)) and factor 2 relating more to acute symptoms such as swelling and lack of joint mobility (items S1, S4, S5, Eigenvalue 1.7, 24% of the variance)).
Descriptive statistics and the distribution of KOOS-Child scores are presented as median (minimum–maximum) and mean (SD) scores at item and subscale levels in tables 3 and 4, respectively. The final KOOS-Child is in the public domain and can be downloaded from http://www.koos.nu
Evaluation of the final KOOS-Child
Aspects of reliability (test–retest, internal consistency, measurement error)
Data from 72 children were analysed for test–retest reliability (mean follow-up time 11 days, SD 4.2; figure 1). Sufficient homogeneity of the final version was found for all subscales according to Cronbach’s α; pain α = 0.85, ADL α = 0.90, Sports/Play α = 0.89 and QOL α = 0.80 except for the Symptoms subscale α = 0.59. Test–retest reliability was excellent for all subscales, except for the Symptoms subscale showing a slightly lower reliability; ICC 0.78–0.91, SEM 5.3–8.1, SDCind 14.6–22.6 and SDCgroup 1.7–2.7 (table 5).
Aspects of test–retest reliability of the final KOOS-Child
Construct validity
All seven predefined hypotheses were confirmed indicating excellent construct validity of the final KOOS-Child (table 1).
Responsiveness
Data from 91 children were included in the responsiveness analyses (mean follow-up time 96 days, SD 7.6; figure 1). The predefined hypotheses of a correlation of ≥0.3 between the final KOOS-Child subscale scores and the respective GPE items were confirmed (Pain 0.43, Symptoms 0.38, ADL 0.48, Sports/Play 0.44 and QOL 0.57, p<0.05). As hypothesised, effect sizes and standardised response means were higher for patients reporting an improvement in their knee condition compared with those who reported themselves as stable (figure 2).
Effect size and standardised response mean values for patients reporting an improvement in their knee condition compared to those who reported themselves as stable.
Interpretability
No floor or ceiling effects were found (defined as ≤15% reporting a score of 0 or 100, respectively; table 4). MIC values complemented with 95% CI for the KOOS-Child subscales are presented in figure 3. MIC scores were greater than the SDC for patients reporting to be ‘better’ and ‘much better’.
Mean change KOOS-Child score values—minimal important change.
Discussion
This study evaluated the psychometric properties of the KOOS-Child when used in children with knee disorders. The preliminary KOOS-Child4 was developed using cognitive interviews to evaluate and establish the scale content validity. Based on these results, the original KOOS was modified in the areas of instructions, terminology and grammar, item format, response format, mapping and layout to better accommodate a paediatric population.4 The KOOS-Child is the first questionnaire for children with knee disorders designed to evaluate self-reported knee function that includes separate sports-related and knee-related QOL subscales. However, until now the psychometric properties of the KOOS-Child have not been evaluated.
Subscale homogeneity, that is, internal consistency was found to be acceptable for all final KOOS-Child subscales; however, a somewhat lower item homogeneity was found for the Symptoms subscale. This result is consistent with data from the adult KOOS.11 The lower homogeneity of the Symptoms subscale can be explained by the variety of symptoms experienced by patients with different and often concomitant knee pathologies. The subscale structure suggested splitting the items into one subscale consisting of more acute symptoms and one of more chronic symptoms. The Symptoms subscale was not modified in the KOOS-Child for two reasons: (1) different knee pathologies often appear concomitantly and (2) more acute and more chronic symptoms both relate to the construct ‘clinical symptoms’.
According to the ICC values, our results showed excellent test–retest reliability (ICC 0.85–0.91) for all final KOOS-Child subscales, with the exception of the Symptoms subscale that demonstrated slightly lower reliability (ICC 0.78). The ICC was complemented by different measures of measurement error, the SEM and the SDC. Data demonstrated that small changes (2–3 points) can be measured over time at the group level; however, a considerable larger change (15–23 points) is needed to detect a true change in an individual over time. This is a previously known phenomenon consistent with other patient-reported outcome measures.17
For assessment of test–retest reliability the children answered the KOOS-Child at the clinic the first time and then at home during the follow-up session. Despite the difference in administration mode, high reliability was achieved indicating this common way of administrating questionnaires in clinical practice (ie, first administration in the clinic and the second at home) is of little importance.
Since no gold standard patient-reported outcome measures exist in Swedish with which to compare the KOOS-Child, criterion validity could not be assessed. However, construct validity was evaluated. Construct validity refers to the degree to which an instrument measures the construct it is supposed to measure based on its relationship to another accepted instrument. In accordance with the COSMIN statement, hypotheses for the determination of construct validity were formulated a priori. The COSMIN statement does not include a statement specifying the cut-off value necessary for a strong correlation. Thus, we used a cut-off of 0.5 to remain consistent with previous studies and recommendations.9 ,19 ,21 ,24 ,25 We recognise that there is currently a debate regarding the appropriate cut-off value for a strong correlation, with values ranging from 0.5 to 0.7.9 ,19 ,21 ,24–27 Our subscale correlations were 0.6, in the mid-range of current recommendations. Several instruments were used to cover different components of the KOOS-Child, the EQ-5D-Y, the CHAQ and subscale-specific VAS-scales. The EQ-5D-Y is a generic well established, valid patient-reported outcome measure used to evaluate QOL in paediatric populations.8 ,9 It is not possible to calculate a total score of the EQ-5D-Y as a measure of the construct QOL. In an attempt to find a domain to correlate with the KOOS-Child QOL subscale, we hypothesised that the KOOS-Child subscale QOL would moderately correlate with the EQ-5D-Y domain ‘doing usual activities’. This hypothesis was confirmed. The CHAQ is a patient-reported outcome measure valid for evaluating self-assessed disabilities in children with inflammatory joint diseases.7 The CHAQ items—‘pain’, ‘run and play’ and ‘walking’—were highly correlated to the KOOS-Child subscales Pain, Sports/Play and ADL. Strong correlations between these scales were hypothesised as they assess joint function and higher level of activity/sports and activities in daily living. As expected, the CHAQ domain ‘grip’ designed to assess upper extremity function had a weak correlation with the KOOS-Child subscales, which focus on functions related to the lower extremity.
VASs are often used for different purposes and have been proven valid and reliable to measure knee function in adults11 but not in children. The VAS questions were created to match the KOOS-Child subscales and all VAS items except for the Symptoms subscale were hypothesised to strongly correlate with KOOS-Child subscales. Since the Symptoms subscale was known to be more variable and less homogenous than the other KOOS subscales12 we hypothesised a moderate correlation with that subscale. Our predefined hypotheses were confirmed, indicating excellent construct validity of the final KOOS-Child when used in this specific setting and population.5 ,16
The ability of a patient-reported outcome measure to detect change over time is very important. Responsiveness was evaluated by comparing the global index measure of health status (GPE score) to the change in the KOOS-Child subscale scores. Correlations between the GPE scores and the final KOOS-Child subscale change scores ranged from 0.38 to 0.57, with the lowest and highest correlations corresponding to the Symptoms and QOL subscales, respectively. As expected, the effect sizes and the standardised response means were higher for the group who reported their condition to be improved, compared with those not improving. Since the children took part in a variety of interventions, or in no intervention at all during the 3-month follow-up period, effect sizes determined in this study cannot be related to any clinical intervention effect or used to determine future sample sizes needed in clinical studies.
Another important aspect of questionnaire interpretation is the presence of floor and ceiling effects. We found levels of floor and ceiling effects that were below the criteria set (≤15%) for all final KOOS-Child subscales indicating that it is possible to measure improvement and deterioration over time in groups of patients. The lowest floor and ceiling effects were found for the sports and QOL subscales, consistent with the children's clinical impression of their condition.
The degree to which qualitative meaning can be ascribed to quantitative scores is important in score interpretation. MIC was determined using an anchor-based method to incorporate the patient's perspective of change.20 ,23 The MIC is dependent on contextual factors and a single MIC value cannot necessarily be applied to the KOOS-Child across different patient groups or settings.23 A commonly used anchor-based approach, employing a global index of change score, is the mean change method.10 ,23 We defined the MIC for the groups of patients reporting to be ‘better’ and ‘much better’ during a 3-month follow-up. Results show incongruence between the groups in the degree of improvement needed in different KOOS-Child subscale scores to detect a clinically important difference. One reason might be the dependence on retrospective self-reported GPE scores, which have been shown to have a questionable reliability in adults due to recall bias and reliance on one single index.23 We suspect this problem might be even larger in children. Small group sizes in our MIC analyses influence the reliability of the results as well, and larger sample sizes are needed to more accurately evaluate the MIC for the KOOS-Child. Receiver operating curves were considered as a potential alternative to calculate MIC. However, this method was considered even more unreliable due to the small number of patients in each treatment response subgroup (according to the GPE scores). In our study 34–35 patients reported being unchanged, and sample size requirements for determination of receiver operating curve MIC values have been suggested to be 50–100 patients for each treatment response subgroup.16
Another methodological consideration was the choice to perform separate exploratory factor analysis for each KOOS-child subscale to evaluate subscale dimensionality. The reasons for this approach were multiple. First, we aimed to develop a paediatric version of the KOOS that was similar in structure to the adult version to enable a smooth transition between the versions in long-term follow-ups. Second, an exploratory factor analysis requires a large amount of data to produce reliable results. The recommended sample size and item volume was ≥100 participants and seven times the number of items in the questionnaire.16 Thus, a sample size of approximately 273 children would be needed to perform one exploratory factor analysis including all 39 items. Given the extended time to meet this requirement, this recommendation was not considered realistic.
Additional limitations in psychometric testing of the KOOS-Child include: difficulties finding appropriate instruments to compare with the outcome measure being tested and the fact that the number of hypotheses influences the percentage of confirmed hypotheses. To the best of our knowledge there are no recommendations or standards for how to construct hypotheses and regarding the number of hypotheses being constructed.
Our results suggest comparable psychometric properties of the final version of KOOS-Child to the original adult version of KOOS.11 Future studies will help resolve possible issues with longer term follow-up of children with knee disorders (eg, into adolescence and adulthood). Validation is an ongoing process. Other investigators should evaluate the psychometric properties and performance of the KOOS-Child in other groups of children with knee disorders. The study strengths include: good psychometric properties, use of an instrument measuring knee function and knee-related QOL in children and inclusion of participants with a broad spectrum of knee injuries. Thus, our results can be generalised to a broad population of children with knee disorders.
Conclusion
The KOOS-Child was developed as a patient-reported outcome measure to evaluate knee function and knee-related QOL in a paediatric population with knee disorders. Our results demonstrate good psychometric properties supporting the use of the KOOS-Child when evaluating children with knee disorders.
What are the new findings?
-
Knee injury and Osteoarthritis Outcome Score for Children (KOOS-Child) is the first questionnaire designed to evaluate self-reported knee function that includes separate sports-related and knee-related quality of life (QOL) subscales for children with knee disorders.
-
Good psychometric properties were demonstrated for the KOOS-Child, which supports its use when evaluating knee function and knee-related QOL in children with knee disorders.
-
The KOOS-Child can be used in a broad population of children with knee disorders.
How might it impact on clinical practice in the near future?
-
The KOOS-Child is recommended for use in clinical practice and research to evaluate knee function and knee-related QOL in a paediatric population with knee disorders.
-
A change of 2–3 and 15–23 KOOS-Child points is needed at group and individual levels, respectively, to detect a true change over time.
-
By using the KOOS-Child in monitoring children with knee disorders, a more valid assessment of knee function and knee-related QOL is possible.
-
Using validated questionnaires for children with knee disorders may help improve the quality of clinical care they receive.
Acknowledgments
The authors would like to thank the children and their families for participating in this study. They would also like to thank the orthopaedic clinics (Astrid Lindgren children's hospital, Artro Clinic and Järva Närsjukhus) as well as the physiotherapy clinics (Sport och Rehabkliniken Vasastan, Astrid Lindgren children's Hospital and OS-kliniken) for their help with patient recruitment.
References
Footnotes
-
Contributors MÖ, EMR, EWB and MDI participated in the study design. MÖ and P-MJ coordinated and recruited study participants. MÖ and EMR carried out the statistical analyses and interpretation of results. MÖ drafted the manuscript. All authors reviewed, commented on and approved the final manuscript.
-
Funding Funding was provided by Stiftelsen Promobilia, Karolinska Institutet, Sällskapet Barnavård and Trygg-Hansa. Grant numbers: Stiftelsen Promobilia (K62922043), Karolinska Institutet (K62976012), Sällskapet Barnavård (K62922073), Trygg-Hansa (K62976013).
-
Competing interests None.
-
Ethics approval The study was approved by the regional ethical committee in Stockholm (2007/1076-31/4) and was conducted according to the Declaration of Helsinki.
-
Provenance and peer review Not commissioned; externally peer reviewed.