Outcome analysis of surgery for disorders of the rotator cuff

Transcription

1 Outcome analysis of surgery for disorders of the rotator cuff A COMPARISON OF SUBJECTIVE AND OBJECTIVE SCORING TOOLS R. Allom, T. Colegate-Stone, M. Gee, M. Ismail, J. Sinha From King s College Hospital, London, England A consecutive series of 72 patients who underwent surgery for disorders of the rotator cuff involving arthroscopic subacromial decompression and open or arthroscopic repairs of the cuff were prospectively investigated as to the comparability of subjective and objective assessment scores of shoulder function. Assessments were made before operation and at,,, and months after surgery using the Disabilities of the Arm, Shoulder, and Hand score, the shoulder score and the -Murley score, which was used as a reference. All scores were standardised to a scale of to for comparison. Statistical analysis compared the post-operative course and the mean score for the subjective Disabilities to the Arm, Shoulder and Hand score and shoulder score, with the objective score at each interval. A strong correlation was evident between both subjective scores and the score. We concluded that both the subjective scores would be useful substitutes for the score, obviating the need for a trained investigator and the specialist equipment required to perform the score. R. Allom, MRCS, Specialist Registrar The Lewisham Hospital NHS Trust, Lewisham High Street, London SE1 LH, UK. T. Colegate-Stone, MRCS, Clinical Research Fellow J. Sinha, FRCS(Orth), Consultant Shoulder & Upper Limb Surgeon Upper Limb Unit, King s College Hospital, Denmark Hill, London SE5 9RS, UK. M. Gee, MRCS, Specialist Registrar Medway Maritime Hospital, Windmill Road, Gillingham, Kent ME7 5N, UK. M. Ismail, MRCS, Specialist Registrar The Newcastle upon Tyne Hospitals NHS Foundations Trust, Freeman Hospital, Freeman Road, High Heaton, Newcastle upon Tyne NE7 7DN, UK. Correspondence should be sent to Mr J. Sinha; joydeep.sinha@btinternet.com 9 British Editorial Society of Bone and Joint Surgery doi:.12/1-x.91b. 714 $2. J Bone Joint Surg [Br] 9;91-B:7-7. Received 8 January 8; Accepted after revision October 8 In 191, Codman et al 1 proposed the systematic analysis of outcomes of treatment to improve practice. Scoring systems have become widely used in orthopaedics to give a tangible measure of the outcome of a given procedure for an individual patient, as well as providing a common language for the comparison of different methods treatment, surgeons and operative centres. As a research tool, they have implications for the practice of evidencebased medicine. 2, Outcome measures may be categorised as subjective or objective, depending on whether they are based on the perception of the patient or upon clinical assessment. Tools such as the Disabilities of the Arm, Shoulder, and Hand () questionnaire 4 and the shoulder score 5 rely on the patient s assessment of their condition, whereas the -Murley score contains an objective assessment of function and strength as well as subjective elements. Frequently, objective scoring systems which reflect clinical and radiological assessment may be at variance with the subjective perceptions of the patient. 5 However, from our own experience, variations in cultural norms as well as attitudes to illness and disability lead to marked differences in the subjective interpretation of symptoms and their significance. Objective measures may be difficult to perform, especially with limited personnel. However, it would be valuable for all surgeons to have the facility to measure treatment outcomes. Simpler tools would enable this. The aim of this study was to evaluate the comparability of two subjective patient-based scoring tools with a single system which combined both objective and subjective assessments. The objective scoring tool is the score, which is composed of an objective assessment of the range of movement and power, together with subjective questions pertaining to pain and activities of daily living. The subjective scoring tools are the questionnaire and the shoulder score. For the purpose of this study, the score is referred to as the objective scoring tool. The comparisons drawn are made in the context of surgery for pathology of the rotator cuff. Patients and Methods We performed a prospective cohort study on all patients undergoing surgery for disorders of the rotator cuff between January and July 7, with prospective follow-up. This involved 72 consecutive patients (5 women, 17 men). They were was further subclassified into the following groups: subacromial decompression without rotator cuff repair in 8 patients (7%) and all repairs of the rotator cuff in 1 patients (%). The latter group was further divided into the 9 patients for whom VOL. 91-B, No., MARCH 9 7

2 8 R. ALLOM, T. COLEGATE-STONE, M. GEE, M. ISMAIL, J. SINHA Table I. Demographics of the surgical population studied and its respective subgroups Surgical group Number of patients Male in group (%) Female (%) Mean age in yrs (range) All operations (44.9) 5 (55.1) 5. (19 to 87) Subacromial decompression without rotator cuff 8 7 (4.) 141 (57.) 52.5 (2 to 87) repair All rotator cuff repairs 1 58 (4.8) (5.2) 59.5 (19 to 82) Arthroscopic rotator cuff repair 9 42 (45.2) 51 (54.8) 58. (19 to 8) Open rotator cuff repair 1 15 (48.4) 1 (51.). (22 to 82) arthroscopic repair was performed and 1 who required an open repair. Open surgery was reserved for those patients whose defect in the rotator cuff exceeded mm (Table I). The repair was performed using bioabsorbable anchors and non-absorbable sutures, regardless of the approach employed. In some patients requiring open repair, side-toside margin convergence of the tear was necessary. This was achieved with Ethibond mattress sutures prior to reanchoring the cuff to its footprint using anchors. Scoring systems. The patients were assessed before operation and at,,, and months after operation using the questionnaire, the shoulder score and the score. Clinical fellows (RA, TC-S, MG, MI) made all the measurements in each patient using a standardised method. The European Society for Shoulder and Elbow Surgery recommendations for the assessment of range of movement in the score were adopted. 7 These require the patient to be sitting on a chair or bed, with the weight evenly distributed between the ischial tuberosities. Active movements of the shoulder are performed until pain develops. Strength was measured by a Nottingham Mecmesin Myometer (Arthocare UK Ltd, North Yorkshire, United Kingdom) with the patient standing, the arm elevated to 9 in the scapular plane and the forearm pronated. The strap of the myometer was placed over the dorsal aspect of the distal forearm. The maximum pull was measured five times and the mean determined. The score allocates a maximum of 15 points and points to the subjective measures of pain and disability, respectively, and a maximum of 4 points and 25 points to the objective assessment of range of movement and power, respectively. The total score is between and, where reflects great disability and a normal shoulder. The 4 is a -item questionnaire about the disability and symptoms that occurred in the previous week. It measures the difficulty of performing various activities, the severity of symptoms, including pain, weakness, and stiffness, as well as the impact of the pathology on work, sleep and social activities. Each question has five possible responses, and an equation converts the total score into a value on the to scale, within which represents a normal shoulder and a shoulder with the maximum disability. The shoulder score 5 uses questions, each with five potential answers, covering pain and functional disability, and yielding a score ranging between and, where would be the score obtained with a normal shoulder. Standardisation technique. In view of the different scales used with each score, all were converted into a scale of to, where is representative of a normal shoulder. The standardised scores for each of the respective scoring systems were calculated using the following conversions: : x = y : x = - y : x = -(((y-)/48) ) where x = corrected score and y = original score obtained. Outcome measures. The outcome measures assessed were the standardised scores at each time for each of the subgroups using the scoring systems described (, and ). The post-operative course was assessed by comparing the mean scores obtained at each interval. Statistics. Paired Student s t-tests at the 5% level were used to individually compare the scores obtained with either subjective tool at each interval with the score obtained from the objective tool at the same time. For each point of follow-up, statistical differences were sought between the questionnaire and the score, and between the and the scores. The purpose of this analysis was to assess whether the or the scoring systems were producing scores that were systematically significantly different from those from the score. These calculations were applied to the entire dataset collectively, as well as to each individual type of surgery detailed previously. Paired t-tests were chosen as it was assumed that the mean of the differences in the pairs would follow a normal distribution. Repeated analysis of variance (ANOVA) testing at the 5% level was also performed with the patients grouped in the same surgery subtypes, with the testing repeated at each interval to establish whether there was a statistical difference between scores obtained using the different outcome measures. The level of significance was set at p =.5. Lin s concordance correlation coefficients 8 were calculated to analyse the comparability of the post-operative course as measured with each subjective tool against that measured with the objective score. The maximum value of Lin s coefficient is one that represents perfect agreement, whereas values below THE JOURNAL OF BONE AND JOINT SURGERY

3 OUTCOME ANALYSIS OF SURGERY FOR DISORDERS OF THE ROTATOR CUFF 9 Table II. The p-values for the comparison of, Disabilities of the Arm, Shoulder and Hand and shoulder scores using analysis of variance (ANOVA). The ANOVA result given is for each surgical subtype on each occasion when scoring was undertaken Time All operations Subacromial decompression with rotator cuff repair All rotator cuff repairs Arthroscopic rotator cuff repair Open rotator cuff repair Pre-operatively.1 (n = 72).2 (n = 8).27 (n = 1).41 (n = 9).7 (n = 1) months.4 (n = 158). (n = 82).49 (n = 7).45 (n = 2).52 (n = 14) months.2 (n = 79).25 (n = 49).1 (n = ).5 (n = 21). (n = 9) months.5 (n = 8).98 (n = 1).44 (n = 7).51 (n = ).28 (n = 1) months.19 (n = 47).41 (n = 22).21 (n = 25).21 (n = 2) + months.1 (n = 17).48 (n = 8). (n = 9) +.52 (n = 7) +, there were insufficient data to allow statistical analysis of the arthroscopic rotator cuff repair at months and open rotator cuff repair at months Table III. Comparison of, Disabilities of the Arm, Shoulder and Hand () and shoulder scores (OSS) for each group. 95% confidence intervals (CI) are given at each interval for the mean of each score. The British Standard s reproducibility coefficient (BSRC) for vs and vs also given All operations Subacromial decompression without rotator cuff repair All rotator cuff repairs Arthroscopic rotator cuff repair Open rotator cuff repair Time 95% CI BSRC 95% CI BSRC 95% CI BSRC 95% CI BSRC 95% CI BSRC Pre-operative (41.2 to 45.8) (41.59 to 4.21) (8.5 to 45.4) (4.88 to 48.) (27.9 to 9.) (4.5 to 47.5) 5.2 (4.91 to 49.9) 4. (4.7 to 47.27).2 (4.1 to 5.4) 2.92 (2.2 to 41.7). OSS (45.4 to 5.4) n = (4.25 to 51.95) n = 8 8. (42. to 49.48) n = (44.17 to 52.2) n = (29.51 to 49.49) n = months (5.7 to 59.) (55.88 to.72) (4.7 to 52.87) (48.25 to 55.9) (9.57 to 52.) (5.8 to.).4 (58.17 to 8.) 9.8 (4. to 51.2). (42.7 to 51.4).4 (. to 59.) 19. OSS (5. to.4) n = 158. (1.74 to 71.) n = (4.1 to 55.44) n = (5.9 to.1) n = (.7 to 5.27) n = 14 months (59.81 to 8.59) (4.7 to 74.7) (49.4 to.4) (51.4 to 8) (5.77 to 58.8) (59.11 to 7.49) 1.78 (59.8 to 7.17) 1.8 (5. to 8.2) 1.7 (5.4 to 7) (4.19 to 58.81).8 OSS (7. to 75.) n = (5.1 to 7.19) n = (.97 to 79.2) n =.52 (8.54 to 81.2) n = (55. to 7.) n = 9 months (5. to 74.5) (4. to 77.7) (2.7 to 7.) (.5 to 8.2) (45.5 to 2.79) (2.81 to 7.9) 5.7 (2.1 to 78.79) 7.4 (58.41 to 71.99) 2.82 (58.5 to 75.7) 5.9 (4. to 9.7) 9.22 OSS (.81 to 75.79) n = 8.9 (4.85 to 78.5) n = 1.5 (4.9 to 77.) n = 7 1. (5. to 8.2) n =. (55. to 7.54) n = months (.49 to 72.51) (8.84 to 8.5) (5.1 to.5) (48.19 to.81) (.5 to 7.55) 5.1 (7.51 to 85.9).84 (47.9 to.47) 1.8 (45.4 to 4.54) 8.48 OSS (.77 to 72.) n = (57.8 to 7.22) n = (58. to 77.8) n = (55.81 to 75.9) n = months (71. to 88.8) (.89 to 9.71) (72.5 to 91.55) (2.42 to 8.98) (5.84 to 74.9) 8.2 (48.9 to 78.1) 2.45 (2. to 89.27).28 (47. to 8.2) 1.4 OSS (9.81 to 87.79) n = (5.8 to 89.94) n = 8 + there were insufficient data available to allow statistical analysis 8.48 (72.2 to 94.7) n = (2.8 to 9.) n = represent poorer agreement. Further, the British Standard s reproducibility coefficient (BSRC), which is the maximum likely difference between a pair of measurements for a given comparison, for versus scores and versus were calculated as a means of establishing whether the two scoring systems were comparable. VOL. 91-B, No., MARCH 9

4 7 R. ALLOM, T. COLEGATE-STONE, M. GEE, M. ISMAIL, J. SINHA Table IV. Lin s concordance correlation coefficients between and Disabilities of the Arm, Shoulder and Hand for each operation, with reduced periods of follow-up Up to months Up to months Up to months All operations Arthroscopic subacromial decompression without rotator cuff repair All rotator cuff repairs Arthroscopic rotator cuff repairs Open rotator cuff repairs Table V. Lin s concordance correlation coefficients between and scores for each operation, with reduced periods of follow-up Up to months Up to months Up to months All operations Arthroscopic subacromial decompression without rotator cuff repair All rotator cuff repairs Arthroscopic rotator cuff repairs Open rotator cuff repairs Pre -operative (n = 158) (n = 79) (n = 8) (n = 47) (n = 17) (n = 72) Fig. 1 Post-operative course all operations (mean values). Results The differences between the mean standardised score and mean standardised questionnaire did not reach statistical significance (p >.5) at any interval regardless of the intervention studied (Tables II and III). However, comparison of the mean standardised shoulder and mean standardised scores, with the exception of those managed by open rotator cuff repair, revealed a statistically significant difference for at least one interval in each treatment group (Table II). However, 2 of the intervals, when scoring was undertaken, did not demonstrate significant differences between the mean and mean scores. Insufficient data were available at two intervals (Table II). Repeated ANOVA testing also demonstrated significant differences for at least one interval in each treatment group studied, apart from those in the open repair group (Table II and III). Post-operative course. A strong correlation was demonstrable using Lin s concordance correlation coefficients between the and both the score (.77 to.95) and the shoulder score (.72 to.9) when following the longitudinal course of each of the treatment groups (Tables IV and V), although a degree of variability was noted with the BSRC analysis (Table III). The analysis of the post-operative course of all operations grouped together showed that up to months a considerable agreement exists between the mean standardised and scores, whereas the mean standardised shoulder score followed a course which was marginally separated (Fig. 1). Beyond months, the and scores matched one another, whereas the produced lower scores. However, it should be noted that the statistical analysis of the pre-operative mean standardised values of this group by both ANOVA and t-test did demonstrate a statistically significant difference between the scoring systems used at that interval (ANOVA, p =.1 t-test versus, p =.1) and hence it is less easy to appreciate the change from the pre- to the post-operative scores from the mean scores alone. When the subgroup which underwent subacromial decompression without repair of the cuff was studied, the relationship between the three scoring systems was more variable. However, correlations remain high (Fig. 2). In those patients who underwent repair of the cuff either as an open procedure or arthroscopically, a close relationship is seen between the and scores. The questionnaire yields higher scores suggestive of better shoulder function, but despite this, correlations remain high (Fig. ). This pattern is repeated when considering arthroscopic repair (Fig. 4), and is exaggerated when assessing open repair (Fig. 5). THE JOURNAL OF BONE AND JOINT SURGERY

5 OUTCOME ANALYSIS OF SURGERY FOR DISORDERS OF THE ROTATOR CUFF (n = 82) (n = 49)( n = 1) (n = 22) (n = 8) (n = 8) (n = 1) (n = 7) (n = ) (n = 7) (n = 25) (n = 9) Fig. 2 Fig. Post-operative course isolated subacromial decompression (mean values). Post-operative course all rotator cuff repairs (mean values) (n = 9) (n = 2) (n = 21) (n = ) (n = 2) (n = 2) (n = 1) (n = 14) (n = 9) (n = 1) (n = ) (n = 7) Fig. 4 Fig. 5 Post-operative course arthroscopic rotator cuff repair (mean values). Post-operative course open rotator cuff repair (mean values). Indeed, although correlation remains good between the and scores, the association between the and scores is marginally reduced. The mean of the difference between the and the subjective score (i.e. against and against ) at each interval in the all operations group was calculated (Fig. ). The score was seen to be more positive than the, whereas the was more positive than the. Both subjective assessment tools had similar trends in the post-operative course when compared with the scores, except at six and months, although no significant difference was present at any interval when individual t-testing was performed between each of the subjective and the scores (p >.5). Discussion The questionnaire and the score returned very similar results over the period of follow-up, regardless of the type of surgery considered. This is reflected by both the absolute values at a given stage, and also the Lin s concordance correlation coefficient returned between the rates of recovery, as determined by each of these outcome measures. The relationship between the shoulder and the scores was also very good, but some variation was noted. However, it is important to note that the interpretation of the correlation coefficients depends to a degree upon the range of values of the variable being assessed, and this will tend to change as the sample size increases. With protracted follow-up, there is often a degree of loss to follow-up which VOL. 91-B, No., MARCH 9

6 72 R. ALLOM, T. COLEGATE-STONE, M. GEE, M. ISMAIL, J. SINHA Mean of the difference of standardised scores between scoring systems n = 72 (n = 158) Mean difference - Mean difference -OSS (n = 79) (n = 8) Fig. (n = 47) (n = 17) Post-operative course of the mean of the difference between the and subjective ( or shoulder) scores in the all operations group. has a potential negative impact on both the ability to compare the correlations of outcome measures in the populations studied and on the power of the result. Subgroup analysis increases the risk of erroneous results through the influence of individual, atypical patients. In such circumstances, outliers can lead to large skews of data trends. Our study was affected by changes in the sample size, statistically assessed both within the surgical subgroups and at the intervals studied. The correlation between scoring tools was assessed once follow-up had been limited to either months or one year. Analysis demonstrated the correlation to have improved for both and scores when all operations were considered. However, correlations in the all-cuff repairs and arthroscopic cuff repair groups demonstrated a tendency to deteriorate when the period of follow-up analysed was reduced. A less obvious trend of change in the correlation coefficients was evident when a reduced period of follow-up was considered in the open repair group. Here an improvement in correlation between the and the scores was evident, whereas a deterioration was found between the and scores. Despite these observations, the overall correlations remained good for all operations. Although the score remains a valuable outcome measure, criticisms have been levelled at its use by several groups, particularly regarding the accuracy of the strength assessment. The capacity to lift a 25 lb mass with the shoulder abducted to 9 is reduced in some individuals without any disorder of the shoulder, most notably in the elderly and in women.,9, The difficulty with the assessment of power was also commented on by Othman and Taylor 11 in a study of patients following manipulation under anaesthesia for adhesive capsulitis. In their study a greater correlation was found between the and scores by omitting the power assessment in what had been termed previously an abbreviated score, or by assessing power with the myometer strap positioned over the mid-humerus as opposed to the distal forearm. Studies of healthy shoulders have highlighted the need for normalisation of data for age and gender through the creation of reference tables and the calculation of a relative score. 9, In our study an individual relative score was used. This previously validated method 1 allows the contralateral shoulder to act as a comparator, leading to the calculation of a relative strength component, thereby avoiding these recognised difficulties with the score. Dawson et al 14 highlighted the difficulty of assessing the long-term outcome of shoulder surgery, citing three key problems: the cost of additional clinics; the reluctance of patients to attend for assessment a long time after surgery, and a lack of appropriate means of assessing outcome. In a study in which outcome measures of shoulder surgery were compared, they noted that patient-based tools better reflected a patient s own perception of their quality of life after surgery than did clinical tools; an observation which has been supported elsewhere. 15 Dawson et al 14 also noted correlations between the shoulder score and the score to be high at each stage of their study (r >.5). In this study, using our statistical analysis, we demonstrated that both the and the scores as well as the and scores are similar in assessing post-operative outcome. Previously the questionnaire has been shown to be sufficiently sensitive to differentiate between large and small changes in disability following subacromial decompression. 1 It remains valid as an outcome measure when translated into other languages, 17 and can be abbreviated to an 11-question format called the Quick while maintaining its precision. In conclusion, our results indicate that both the subjective scores analysed and the score provided good indicators of post-operative outcome for disorders of the rotator cuff. It can be seen that the sole use of either the questionnaire or the shoulder score for follow-up of surgery to the cuff is possible, as both are effective scoring tools. We suggest that either of the subjective scoring systems would be useful substitutes for the score. This obviates the need for a trained investigator and specialist equipment required to perform the score, and also facilitates telephone or postal follow-up. No benefits in any form have been received or wil be received from a commercial party related directly or indirectly to the subject of this article. References 1. Codman EA, Chipman WW, Clark JG, Kanavel AB, Mayo WJ. Standardisation of hospitals: report of the committee appointed by the Clinical Congress of Surgeons of North America. Trans Clin Con Surg North Am 191;4: Harvie P, Pollard TCB, Chennagiri RJ, Carr AJ. The use of outcome scores in surgery of the shoulder. J Bone Joint Surg [Br] 5;87-B: THE JOURNAL OF BONE AND JOINT SURGERY

7 OUTCOME ANALYSIS OF SURGERY FOR DISORDERS OF THE ROTATOR CUFF 7. Conboy VB, Morris RW, Kiss J, Carr AJ. An evaluation of the -Murley Shoulder Assessment. J Bone Joint Surg [Br] 199;78-B: Hudak PL, Amadio PC, Bombardier C. The Upper Extremity Collaborative Group (UECG): development of an upper extremity outcome measure: the (disabilities of the arm, shoulder, and hand). Am J Ind Med 199;29: Dawson J, Fitzpatrick R, Carr A. Questionnaire on the perceptions of patients about shoulder surgery. J Bone Joint Surg [Br] 199;78-B:59-.. CR, Murley AHG. A clinical method of functional assessment of the shoulder. Clin Orthop 1987;214: CR. scoring technique for shoulder function. SECEC information 1991 Nr. 8. Lin L I-K. A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989;45: Grassi FA, Tajana MS. The normalization of data in the -Murley score for the shoulder: a study conducted on 5 healthy subjects. Chir Organi Mov ;88:5-7.. Katolik LI, Romeo AR, Cole BJ, et al. Normalization of the score. J Shoulder Elbow Surg 5;14: Othman A, Taylor G. Is the score reliable in assessing patients with frozen shoulder?: shoulders scored years after manipulation under anaesthesia. Acta Orthop Scand 4;75: Patel VR, Singh D, Calvert PT, Bayley JI. Arthroscopic subacromial decompression: results and factors affecting outcome. J Shoulder Elbow Surg 1999;8: Fialka C, Oberleitner G, Stampfl P, et al. Modification of the -Murley shoulder score: introduction of the individual relative score individual shoulder assessment. Injury 5;: Dawson J, Hill G, Fitzpatrick R, Carr A. The benefits of using patient-based methods of assessment: medium-term results of an observational study of shoulder surgery. J Bone Joint Surg [Br] 1;8-B: Tingart M, Bathis H, Lefering R, Bouillon B, Tiling T. score and Neer score: a comparison of score results and subjective patient satisfaction. Unfallchirurg 1;4:48-54 (in German). 1. Gummesson C, Atroshi I, Ekdahl C. The disabilities of the arm, shoulder, and hand () outcome questionnaire: longitudinal construct validing and measuring selfrated health change after surgery. BMC Musculoskeletal Disord ;4: Atroshi I, Gummesson C, Anderson B, Dahlgren E, Johansson A. The disabilities of the arm, shoulder, and hand () outcome questionnaire: reliability and validity of the Swedish version evaluated in 17 patients. Acta Orthop Scand 4;71:1-.. Gummesson C, Ward MM, Atroshi I. The shortened disabilities of the arm, shoulder, and hand questionnaire (Quick): validity and reliability based on responses within the full-length. BMC Musculoskeletal Disord ;7:44. VOL. 91-B, No., MARCH 9