Author's response to reviews Title: Rotational strength, range of motion, and function in people with unaffected shoulders from various stages of life Authors: Jean-Sébastien Roy (jean-sebastien.roy.1@ulaval.ca) Joy C MacDermid (macderj@mcmaster.ca) Kirsty Usher Boyd (kuboyd@gmail.com) Kenneth J Faber (kjfaber@uwo.ca) Darren Drosdowech (ddros@mac.com) George S Athwal (gathwal@uwo.ca) Version: 2 Date: 3 February 2009 Author's response to reviews: see over
Hamilton, February 2 nd 2009 Dear Doctors Chan and Kurosaka: Please find the revised version of the manuscript entitled: ROTATIONAL STRENGTH, RANGE OF MOTION, AND FUNCTION IN PEOPLE WITH UNAFFECTED SHOULDERS FROM VARIOUS STAGES OF LIFE. We would like to thank the reviewers for their time and energy. We would also like to thank Sports Medicine, Arthroscopy, Rehabilitation, Therapy & Technology for giving us the opportunity to comment and act on the recommendations of the reviewers. The most important changes brought to the manuscript were made following the comments of Reviewer #1. In the comment number 6, under the Major Compulsory Revisions part of the report, Reviewer #1 was concerned on how the statistics were presented. It was proposed to outline the ANOVA results before the gender comparisons. We followed the recommendation. This modification implies changes in the results. All the changes are in red in the manuscript. The next pages respond, point-by-point, to the comments and concerns raised by the three reviewers. We thank you for the time and energy that will be put into the review process. Sincerely yours, Jean-Sébastien Roy, PT, PhD Postdoctoral fellow, School of Rehabilitation Sciences, McMaster University Corresponding author 1
REVIEWER # 1 MAJOR COMPULSORY REVISIONS 1. Page 4 - The authors present several arguments for how they designed their investigation that they did not support. For example, according to the authors, Otis et al discussed the value of strength assessment. What did Otis et al conclude? Similar questions can be generated regarding the Kuhlman et al citation. Details were added in the Introduction: Otis et al. discussed the value of assessing strength relative to a matched population, especially in the presence of a shoulder deficit, in order to characterize these deficits. Moreover, Hughes et al. noted the importance of having unaffected shoulders data for comparison in patients whose shoulder deficits manifest with bilateral involvement. Kuhlman et al. acknowledged that while many studies have attempted to define normal values for the strength of shoulder muscles, there have been varying degrees of success because of a lack of standardization (plane of motion, position of the shoulder or stabilization of the body). 2. Page 7 - Why are there more female than male subjects? Why is the distribution across groups so variable? As mentioned in the Methods, subjects were recruited for this study by a variety of methods (flyers advertising the study, people accompanying their spouses to the Hospital, and word of mouth, studies evaluating the psychometric qualities of the FIT-HaNSA and the concomitant validity of isometric strength devices). For the validation studies, the subjects recruited were mostly university students, therefore, they are more subjects in the younger category. Furthermore, as mentioned in the Discussion, older subjects were hard to recruit. For the number of male and female subjects, the goal was to recruit 100 subjects for each gender. This goal was achieved. However, other subjects volunteered to participate, therefore they were included regardless of the gender. More women volunteered, resulting with overall more female. This has been our experience in other studies as well- i.e. higher female volunteer rates 3. Page 7 - It is unclear why the tests sessions involved only 2 of 3 possible variables. When this study started our focus was typical impairments measured in clinical research: strength, range of motion and self-report measures of disability (SST, DASH and WORC). The FIT-HaNSA was not developed at that time. Once this new test was published and, validation studies were performed, we felt it was important to include and added a second wave of data collection. There were three different validation substudies performed to evaluate the FIT-HaNSA where we were able to acquire normative data. These substudies collected the following subsets of data: 1) FIT-HaNSA and DASH; 2) FIT-HaNSA, DASH, range of motion and strength; 3) FIT-HaNSA, DASH, SST, WORC and strength. Details were added in the methods, subjects section to make this clearer: Subjects were first specifically recruited for this study by a variety of methods, including flyers advertising the study, people accompanying their spouses to the Hospital, and word of mouth (n = 128; variables evaluated: strength, range of motion and self-reported function). Separate groups of subjects participated in studies evaluating the psychometric qualities of the FIT- HaNSA (n = 125; variables evaluated: upper limb functional performance and self- 2
reported function and/or range of motion and/or strength) and the concomitant validity of isometric strength devices (n = 41; variables evaluated: strength and self-reported function). The test session involved the measurement of at least two of the following variables: selfreported shoulder and upper extremity function, range of motion, strength or upper limb functional performance (FIT-HaNSA). Separate subgroups of volunteers were enrolled sequentially for the strength/motion and functional performance subsets for the study. The subjects were assessed by research assistants following standardized tests procedures. 4. Page 8 - The reliability of goniometry is not high. There are much better tools available. We wanted to use clinical tools since this article was performed to give normative data to the clinicians. Goniometer is the most widely used instrument to measure ROM in clinics. That is why we decided to use this tool. Reliability is high when standardized methods are used and so we added: Intratester reliability of shoulder movements using a goniometer has been shown to be excellent with ICC of 0.88 to 0.93, and standard error of measurement around 3.0. 5. Page 10 - The data analysis is unclear. Multiple steps are necessary to correct this: a. Did the data meet the assumptions for parametric statistics? For example, there is no evidence of visual or statistical testing of normality. The majority of the skewness and normality tests indicated normality was present. Some of the self-report data was skewed to low scores as we discuss, but not sufficiently to warrant non-parametrics. b. How many levels of each factor were analyzed in each ANOVA? If all categories were included, they should have been a 2 gender by 3 age categories ANOVAs. The reviewer is right; it was a 2 X 3 ANOVA. Details were added in the Methods, Analyses section: Two-way analysis of variance (ANOVA) was used to determine, for each variable, the effects of age and gender. The factors in the model were gender (men or women) and age categories (18-39, 40-59, and over-60). c. The authors were not clear on their rationale for performing correlations. Additionally, Pearson product correlations are appropriate with ratio or interval data. Gender is not either. The correlations were used to establish the inter-relationships between the different outcomes. Details were added in the Methods, Analyses section: To establish the interrelationships between the different types of variables, correlations between age, strength, range of motion, functional performance and self-reported function were examined using Pearson product. There was an error in the manuscript; gender was not used in the correlation calculations. Therefore, it was taken out of the text: 3
The relationship between age, gender, strength, range of motion, functional performance and self-reported function was examined using Pearson product correlations. 6. Page 11 - The ANOVA results should be outlined before the gender comparisons. The first step would be to outline the interaction effects (if any) between gender and age category. If there are none, the authors could discuss main effects of gender without alpha adjustment or by age category after a Bonferroni adjustment to an alpha of.017. If there are interactions, each gender/age category would be evaluated after the appropriate alpha adjustment. The results were presented as suggested by the Reviewer. The alpha was adjusted according to Bonferroni adjustment. Details were first added in the Methods, analyses section: Twoway analysis of variance (ANOVA) was used to determine, for each variable, the effects of age and gender. The factors in the models were gender (men or women) and age categories (18-39, 40-59, and over-60). ANOVA and independent t-tests, with Bonferroni adjustment, were used for multiple comparisons. For the Results sections, changes have been made throughout the text. See changes in red. 7. Page 11 - How accurate is the LIDO? Can the authors demonstrate that the 1.0 or 2.2 Nm differences they cite between sides are true differences? The differences are true within the limits of statistical analysis parameters, but may not indicate clinically important differences in capability from side-to-side. The absolute measurement error of the LIDO is unknown for the evaluation of shoulder movement and thus group and individual level clinically important differences are not known. Therefore, we acknowledged that these differences may not be clinically relevant: However, the mean differences between dominant/non-dominant sides were small and may not be clinically relevant (mean differences: 2.2 ± 8.0 Nm or 6.9 % for internal rotation; 1.0 ± 5.2 Nm or 4.0 % for external rotation). 8. Page 12 - The intrarater goniometer reliability is between 3 and 5 degrees. Interrater is between 4 and 6 degrees. Is a 3 degree difference a true difference? Comments above apply here as well. Details were added in the Methods, test procedures section about the measurement error of the goniometer at the shoulder joint: Intratester reliability of shoulder movements using a goniometer has been shown to be excellent with ICC of 0.88 to 0.93, and standard error of measurement around 3.0. Then, in the results it was added that the difference was within the measurement error: Again, the differences between dominant/non-dominant sides were small and within the measurement error (mean differences for supine external rotation: 3.0 ± 10.6 or 3.4 %). 9. Page 13 - The discussion will likely change based on the authors addressing the Major and Minor concerns. Some changes were made following the comments throughout the Discussion and Conclusion. See changes in red. 4
10. Page 24 - The 121 men and 173 women are not distributed equally across the age categories. Was this intentional? What implications do these differences have for the statistical analyses? The distribution was not intentional. As mentioned in the manuscript, the data are sparse for the over-60 group since it was difficult to find people in this age group who could fit the criteria. In the answer to the comment #2, we explained that in the validation studies, the subjects recruited were mostly university students, leading to younger subjects. The fact that the subjects are not distributed equally across age categories have probably been detrimental in detecting differences across age categories. Details were added in the Discussion: The small number of older subjects could have also been detrimental in detecting differences across age categories. MINOR ESSENTIAL REVISIONS 1. Page 5 - The authors switch their orientation in the first 2 sentences. They begin by discussing absolute strength and end with discussing age and weight controlled strength. Absolute should not be written in the first sentence, therefore it was taken off: Gender related differences in absolute strength have been reported. More specifically, Hughes et al. have shown that men are stronger than women when controlling for age and weight. 2. Page 5 - How well does the FIT-HaNSA discriminate between persons with and without shoulder disorders? In their validation study of the FIT-HaNSA, MacDermid et al have shown that it significantly discriminate between with mild-moderate shoulder pathologies and a control group (p = 0.008). Details were added in the Methods, test procedures section: The FIT-HaNSA has been shown to significantly discriminate between subjects with and without shoulder pathologies. There is more recent data in larger numbers also confirming this which we do not reference since it is a master s thesis and not yet published in an accessible format. 3. Page 6 - The authors list their purposes. They do not cite their hypotheses. Hypotheses were added at the end of the introduction: The hypotheses were that men will be stronger than women in all age categories. Furthermore, rotational strength, range of motion, functional performance and upper extremity function will be decreased in the over-60 age category compared to the 18-39, 40-59 age categories. 4. Page 8 - It is unclear why the authors only tested the strength and ROM of the internal and external rotators. We decided to focus on muscles of internal rotation and of external rotation since they are the muscles that are usually involves in subjects with shoulder disorders such as impingement, tendinosis or rotator cuff tears. Details were added in the Methods, test procedures section: These muscles groups were chosen since they are the ones usually involve in patients with shoulder disorders such as impingement or rotator cuff tendinosis or tear. 5. Page 10 - What are the psychometrics for the DASH, SST, and WORC? 5
The psychometric properties of the DASH, SST and WORC were added in the Methods, test procedures section: These three scales have been shown to be valid, reliable (test-retest reliability > 0.90) and responsive (effect size/standardized response mean > 0.80). 6. Page 32 - The number of subjects involved in the correlations is unclear. The number of subjects for each correlation was added in the Tables. See table 5 and 6. DISCRETIONARY REVISIONS 1. Although the number of subjects would decrease, the authors may consider choosing 1 functional outcome tool. They could then perform a multiple regression to determine which of the factors they have measured explain a significant amount of the variation in the tool. Although the suggestion is interesting, we prefer to keep all the tools in order and to keep the number of subjects that we have recruited. The main objective of this study was to present the unaffected comparison data. If we have more subjects, we have a better representation of the normal population. 6
REVIEWER # 2 SPECIFIC COMMENTS: 1. Page 2 Abstract, line 12 - add - "isometric" before rotational strength to be sure the reader knows that this is how the strength was measured. Done 2. Page 7, methods, line 19 - why were the test session limited to only (at least 2 of the following variables) - why did the subjects not complete all of the test measures? This should be explained in detail as if the researchers had gathered all the data sets from all the subjects this would obviously provided a larger population of responses. Please see answer to comment #3 from Reviewer #1. 3. Page 8, line 1 - strength was measured - add the d to the measure Done 4. Page 8, line 18 - with your supine ROM measures. Greater detail has to be given here in my opinion. Was the scapula stabilized? Was any overpressure used? i.e. - passive implied you moved the arm, did you press into end ROM - if you did how did you standardize the force you pushed with? Was it measured in the coronal plane at 90 degrees or did you put something under the elbow to put the shoulder in the scapular plane. How did you limit scapular motion? Details were added: Supine external and internal rotations were measured passively with the humerus abducted in frontal plane to 90 and the elbow flexed to 90. The scapula was stabilized during internal rotation by the research assistant in order to avoid protraction of the shoulder girdle. The scapula was not stabilized in external rotation. Passive external/internal rotation of the shoulder was performed to subject tolerance with the olecranon serving as the axis of rotation as per the American Academy of Orthopaedic Surgeons. The movement was stopped when the first resistance was felt. Intratester reliability of shoulder movements using a goniometer has been shown to be excellent with ICC of 0.88 to 0.93, and standard error of measurement around 3.0. 5. Page 9, line 12 - How did you standardize the identification of the substitute movement? I assume you are talking about the shoulder shrug type movement? Was this just with visual observation or was it filmed? Also, were the three tasks randomly ordered? If not - why? Here are the FIT-HaNSA Stopping Criteria: Each task can be continued for up to 5 minutes, but is terminated based on the following stopping rules: 1. The subject stops or states it is too painful to continue. 2. The subject is severely off pacing to the extent that they are unable to complete one repetition of the movement within 2 beats of the metronome. 3. The subject substitutes using trunk/whole body movement and cannot correct with feedback for 5 successive repetitions of the task. 7
4. The examiner believes the subject is at risk of injury or adverse complication if tests were to continue. Therefore, the research assistant was supervising the subject during the whole test. The three tasks are not randomly ordered, because this is how the test was described and validated. Also, there is an increasing level of difficulty from task 1 to task 3. No details were added following this comment. The clinicians can refer to the original publication of the FIT-HaNSA to get specific details on the test. 6. Page 9; line 18 - Why did subjects only complete one of the three instruments and how did you select which one - random? Please see answer to comment #3 from Reviewer #1. 8
REVIEWER # 3 MINOR ESSENTIAL REVISIONS: 1. The statement of purpose sentence beginning The purpose of this investigation is not clearly stated. After the and is the portion of the sentence that is not clear. Possibly stating to add information on functional performance and self-reported function would be clearer. As suggested by the reviewer, changes were made in the statement of purpose: The purpose of this investigation is to enhance existing comparative data on normal shoulder range of motion and strength, and by adding to add information on functional performance and self-reported function. 2. In the Methods section, there was no indication of the speed at which the internal and external rotation strength was measured. This should be stated for comparison purposes. It was stated that the testing was conduction per Leroux et al. That group tested at two different speeds. Isometric strength was measured. The reference to Leroux et al. refers to the position of the upper limb. Details were added in the first sentence (Methods section; test procedures): Strength: Shoulder internal and external rotation isometric strength was measure in Newton metres (Nm) using a computerized dynamometer (LIDO WorkSET TM ; Loredan Biomedical, West Sacramento, California) as per Leroux et al.. Testing was performed with subjects seated in a heavy straight-backed chair with both feet flat on the floor. The arm to be tested was positioned with 30 forward flexion and abducted to 45 (in the plane of the scapula) as per Leroux et al. 3. The heading Analysis should be Analyses. Done 4. The Results section is poorly written. The section jumps from topic to topic too frequently making it confusing for the reader. Better organization of the tables and the presentation of data would make the section easier to follow. One example would be that table 5 should be table 4 based on the order that the information is presented in the text. Changes were made throughout the Results section following comment # 6 by reviewer #1. We think that these changes improve the comprehension of the results. The tables were reorganized following the comments. Subheadings were also added in the Results section. 5. It is also very confusing when the data presented in the text is not clearly stated as data examined by sex or if the information is referring to the data examined with men and women combined. This should be stated more clearly. In one section it is stated but at the end of the presentation of the data and would be more clear if it was stated at the beginning of the presentation of data. (i.e. Page 11 beginning with The dominant side was significantly and ending with Gender and dominance had ) As mentioned in response to comment #4, the changes made following the comments from Reviewer #1 clarify all this. 9
6. Many variables that were stated as significantly different in the text were not indicated as different on the tables. An example is the difference in strength between men and women. The text states that the men were significantly stronger than the women in all age categories. This is not shown on the table. There are many examples of this in the results section. Changes were made throughout the Tables. 7. For Table 1 and 2, the differences appear to exist only with the data combined. Were there any differences in the data by sex? Again, the text states on page 12 that women had significantly more range of motion than men for seated external rotation.. For range of motion, there were differences in the data by sex. Details were added in Table 2 (now Table 4). As for the SST, the significant differences with age were only observed when to all subjects were combined. A sentence was added in the Results section: Therefore, age categories were compared across all subjects using ANOVA. The results show that SST scores were significantly decreased in the over-60 age category when compared to the 18-39 and 40-59 age categories (P < 0.011) (Table 1). 8. For Table 3, again the text states differences between men and women but the table only appears to show differences between the age categories for men. Details were added in the Table. 9. The text on page 12 discussing the FIT-HaNSA data suggests significant differences between the tasks while the table indicates no differences. Details were added in the Table. 10. The text on page 12, the last paragraph refers to table 4 for SST scores while table 4 only shows the data for the FIT-HaNSA test. The change was made 11. Spelling (line is counted as lines of text): All the spelling errors were corrected. 10