Responsiveness of SF 36 Health Survey and Patient Generated Index in people with chronic knee pain commenced on oral analgesia: analysis of data from a randomised controlled clinical trial

(1)1. Alex Papou1, Salma Hussain2, Daniel McWilliams3, Weiya Zhang3 and Michael Doherty3. 2 3. Responsiveness of SF-36 Health Survey and Patient Generated Index in people with chronic knee pain. 4. commenced on oral analgesia: analysis of data from a randomised controlled clinical trial. 5 6. 1. Department of Rheumatology, Queen's Medical Centre, Derby Road, Nottingham, NG7 2UH, United Kingdom. 7. 2. Frimley Park Hospital, Portsmouth Road, Frimley, GU16 7UJ, United Kingdom. 8. 3. Division of Rheumatology, Orthopaedics and Dermatology, University of Nottingham, Academic Rheumatology,. 9. Clinical Sciences Building, Nottingham City Hospital, Hucknall Road, Nottingham, NG5 1PB, United Kingdom. 10 11. Corresponding author. 12. Weiya Zhang. 13. Academic Rheumatology, Clinical Sciences Building, City Hospital, Nottingham, NG5 1PB, United Kingdom. 14. Email Weiya.Zhang@nottingham.ac.uk. 15. Phone 0115 823 1756. 16. Fax 0115 823 1757. 1.

(2) 17. Introduction. 18 19. Health-related quality of life (HRQoL) instruments are important in the assessment of efficacy of various treatments and. 20. healthcare delivery decisions, and are of particular significance in people with chronic painful conditions such as. 21. osteoarthritis (OA) which is known to progress with age and significantly influence people's lives.. 22 23. Over the last decades a variety of HRQoL instruments have been developed, the majority being self-completed. 24. questionnaires. These instruments can be divided into three groups: generic, which are used for a broad range of. 25. conditions; disease-specific, which contain items particularly influenced by the specific problems associated with the. 26. disease of interest; and individualised patient-centred instruments. The SF-36 (Short Form Health Survey consisting of. 27. 36 items) is a widely used generic instrument, validated in many populations, which allows comparisons between. 28. different diseases and to the population 'norm' for scales [1]. The Patient Generated Index (PGI) is an individualised tool. 29. where people are able to choose and rate the most important areas where their HRQoL is affected [2]. Its validity,. 30. reliability and responsiveness have been examined in various conditions, including rheumatic diseases [3].. 31 32. Significant difficulties arise when various HRQoL instruments are being compared with each other and various. 33. instruments have been developed to ensure adequate methodology and reporting, such as COSMIN criteria and the. 34. CONSORT Patient-Reported Outcome (PRO) extension [6]. Responsiveness has been proposed as criterion for selecting. 35. HRQoL instruments for clinical trials [9]. There is a limited literature assessing responsiveness of different HRQoL. 36. instruments in musculoskeletal conditions. Only a few studies have compared responsiveness between instruments in. 37. patients with osteoarthritis [10] but none have included the PGI.. 38 39. The objectives of this study were: (1) to assess the responsiveness of the SF-36 and PGI in people with knee pain. 40. predominantly due to OA who were given oral paracetamol and/or ibuprofen, in the context of a community-based. 41. randomised controlled trial (RCT); and (2) to perform content analysis of the SF-36 and PGI aiming to identify. 42. differences between the instruments and causes of different responsiveness.. 43 2.

(3) 44. Methods. 45 46. Participants and study design. 47 48. This study was undertaken in people who participated in an RCT evaluating the comparative efficacy and safety of. 49. widely available over-the-counter analgesics: paracetamol, ibuprofen or the combination of the two which took place in. 50. the UK from June 2007 to July 2008 (ISRCTN registry, www.isrctn.com, trial registration number ISRCTN77199439). 51. [13]. Participants in this trial were randomly allocated to one of 4 oral treatments which were taken three times daily for. 52. 13 weeks: paracetamol 1000 mg tds, ibuprofen 400 mg tds, paracetamol 500 mg plus ibuprofen 200 mg tds, or. 53. paracetamol 1000 mg plus ibuprofen 400 mg tds. The majority were recruited by postal questionnaire in. 54. Nottinghamshire and the remainder from general practice lists or from local advertisement. The subjects' written consent. 55. was obtained according to the Declaration of Helsinki and the study has been approved by the Southampton & South. 56. West Hampshire Research Ethics Committee A in March 2007 (ref: 07/Q1702/19). Inclusion criteria were: age 40 years. 57. and older; knee pain for most of the past 3 months and for 4 of the preceding 7 days; Steinbrocker functional. 58. classification of I-III [14]; and pain affecting the index knee (after a washout period if currently taking analgesics) of 30-. 59. 80mm on a 100mm visual analogue scale over the previous 48 hours for at least one of number of specified daily. 60. activities (walking on a flat surface, going up/down stairs, at night, sitting, lying, standing upright). For further details of. 61. the original study including randomisation and blinding methods, sample size calculation, statistical analyses and trial. 62. outcomes, etc., see Doherty et al 2011 [13].. 63 64. HRQoL assessments. 65 66. In addition to pain and functional assessments which have been reported in the study [13], each participant was asked to. 67. complete the SF-36 version 2 and the PGI questionnaires at baseline, day 10, week 7 and week 13. The participants had. 68. access to the research nurse for clarification of how to complete the questionnaires if required. The data were collected. 69. and transferred to SPSS for Windows version 20 where statistical calculations were performed. Both overall results and. 70. the results according to treatments were analysed.. 3.

(4) 71 72. The SF-36 Each of the 8 scale scores were derived using standard methodology [15] and were then norm-transformed. 73. using the weightings and transformations recommended by Jenkinson 1999 [16]. Missing data were imputed with the. 74. average value for the remaining items on the subscale, unless more than 50% of items were missing (in which case the. 75. subscale was not calculated) [15]. This produces values between 0 and 100 with the average population in the UK. 76. having an average score of 50 and a standard deviation of 10 in each domain. The lower the score the worse is the. 77. condition reported. The two summary scores - Mental Component Score (MCS) and Physical Component Score (PCS). 78. were calculated from the corresponding scale scores.. 79 80. The PGI This is an individualised self-reported 18-item questionnaire that comprises 3 parts (Online Resource Table. 81. S1). In part 1 responders are asked to list any 5 areas of their life affected by the knee pain and its treatment, together. 82. with a sixth predefined area headed 'All other areas of your life affected by your knee pain and its treatment'. In part 2. 83. participants are asked to score the impact that each of the six components has on their life, ranging from 0 ('as bad as it. 84. could possibly be') to 6 ('as good as it could possibly be'). In the final part they are asked to weight the importance of. 85. each area using 10 points to distribute as they choose. When the participants completed the PGI at day 10, week 7 and. 86. week 13 they were reminded of the areas they entered at baseline but not the weightings they afforded to each. The PGI. 87. score was calculated using a standard approach which gives a value ranging from 0 being the worst quality of life. 88. possible to 6 being the best [2].. 89 90. Responsiveness. 91 92. The responsiveness and 95% confidence intervals (CI) were measured as a standardised response mean (SRM), which is. 93. calculated as:. 94. 𝑆𝑅𝑀 =. 𝑋̅0 − 𝑋̅1 𝑆𝐷01. 95. ̅ 0 is the mean at baseline and X ̅1 is the mean at follow-up and SD01 is the standard deviation of the change [17]. where X. 96. This will allow for comparison between different scoring systems. The data were interpreted using the usual consensus. 4.

(5) 97. that absolute responsiveness values between 0 and 0.2 represent no response, values of 0.2 to 0.5 indicate a weak. 98. response, values of 0.5 to 0.8 indicate a moderate response and values above 0.8 represent a strong response [18].. 99 100. Content analysis. 101 102. An attempt was made to identify prevailing domains in the PGI, their frequency, and areas of life which are of most. 103. concern for the patients. Initially we attempted to stratify PGI responses using the 8 domain scores from the SF-36. 104. questionnaire but this was impossible since the different HRQoL areas in the SF-36 were intersecting or too broad.. 105. Therefore all individual PGI responses were aggregated into other, more discrete areas, and their frequencies were. 106. calculated. Items asked about in the SF-36 survey were compared to those in the PGI domains and individual responses. 107. in order to detect items present in the PGI but not in SF-36.. 108 109. Results. 110 111. Characteristics of the study population. 112 113. Baseline demographic of the participants appear in Table 1. Of 884 people completed both SF-36 and PGI. 114. questionnaires at baseline, 783 (11.4% missing) and 776 (12.2% missing) at day 10, 647 (26.8% missing) and 642. 115. (27.4% missing) at week 7, and 807 (8.7% missing) and 597 (32.5% missing) at week 13 completed the SF-36 and PGI. 116. respectively. Having analysed four treatment groups separately we found no statistically significant difference in. 117. responsiveness between them, therefore a combined analysis was undertaken for this report. Of the 884 participants,. 118. 85% satisfied American College of Rheumatology criteria for knee OA (1986) and 63% had radiographic evidence of. 119. knee OA (definite osteophyte and definite narrowing in at least one compartment).. 120 121. Responsiveness. 122. 5.

(6) 123. In general, the SF-36 has shown no or weak responsiveness in the majority of the domains (Table 2). Some moderate. 124. responses were observed in the Bodily Pain Score for which the SRM values ranged from 0.49 to 0.63 upon different. 125. time points. In contrast, PGI had weak responsiveness at day 10 but moderate responsiveness throughout the rest of the. 126. study period (SRM 0.41 to 0.61) (Table 2). Apart from the Bodily Pain Score, none of the other SF-36 scores were. 127. comparable to PGI in terms of the responsiveness to the analgesics used in this trial.. 128 129. Content analysis of the SF-36 and the PGI. 130 131. A total of 3356 items were recorded within the PGI at baseline. It was found that all responses could be mapped into 13. 132. main domains and their frequencies were then calculated (Figure 1). The most frequent areas for the participants with. 133. knee pain appeared to be: mobility (31.0%); sports and activities (14.9%); and house work and gardening (12.9%). It. 134. was apparent that many responses in the PGI were highly individualised and would not be captured within the specific. 135. topics listed within the SF-36. There were multiple individual responses in each of the 13 PGI domains which are not. 136. covered by SF-36 questions, such as many individual sports activities, gardening, family contact, driving, hobbies, etc.. 137. (Online Resource Table S2). Several domains appeared not to be represented in the SF-36 survey at all (family and. 138. relationships, driving, shopping, and hobbies/leisure/holidays).. 139 140. Discussion. 141 142. This study focused on middle-aged and older community-derived people with chronic knee pain, the majority of whom. 143. had clinical and radiographic features of knee OA. The results show that following commencement of paracetamol. 144. and/or ibuprofen in an RCT setting the PGI is more responsive than the SF-36. Only the SF-36 Bodily Pain scale had. 145. comparable responsiveness values. Five out of eight SF-36 domains and one of two summary scores showed no response. 146. to oral analgesics at any time point and only 14 out of the 36 questions asked in the SF-36 appeared relevant in. 147. calculations of scale scores which responded to oral analgesics.. 148. 6.

(7) 149. The SF-36 is probably one of the most widely used and best studied generic HRQoL tools. Many of the SF-36 questions. 150. are relevant for lower limb OA. For example, the SF-36 Physical Function score is calculated based on ten items, three. 151. of which assess walking, two assess climbing, with other items assessing kneeling, bending, running, etc. The SF-36. 152. Bodily Pain scale asks about pain severity and functional limitations due to pain. Although it is generally considered a. 153. reliable tool for the assessment of HRQoL in arthritis [20], debate continues as to whether the SF-36 is sufficiently. 154. responsive for OA and whether it captures all relevant information. For example, Rannou et al (2007) have shown that. 155. the two summary SF-36 scores (the MCS and PCS) are not optimal for use in OA due to multiple intrinsic issues,. 156. especially with respect to mixing of data [21]. Salaffi et al (2005) concluded that a disease-specific instrument WOMAC. 157. (Western Ontario and McMaster Universities Index of Osteoarthritis) is preferred to SF-36 for evaluating patients with. 158. lower limb OA [12]. Similarly in a study by Angst et al (2001) the WOMAC was significantly more responsive than the. 159. SF-36 in assessing function in patients with osteoarthritis of the legs undergoing a comprehensive rehabilitation. 160. intervention, and for both instruments, the pain scales were more responsive than the function scales [10].. 161 162. Although the PGI was developed in the 1990s it has been relatively little studied or used. It was shown to have good. 163. validity and responsiveness in graded structured reviews [4] but there is only a very limited literature comparing the PGI. 164. and other HRQoL instruments. We consider that the main merit of this instrument is that it can exclude the 'noise' of. 165. questions which are not of direct concern to the individual person, and capture those areas of life which are important for. 166. people but often not represented in generic HRQoL tools. More than half of the items that patients were entering into the. 167. PGI did not have corresponding questions in the SF-36 survey, although we understand some of these items could have. 168. possibly been captured indirectly by the SF-36 general questions. Interestingly, pain per se was not the predominant. 169. complaint mentioned in the PGI in our group, and issues of mobility and various activities were of far more concern for. 170. the patients (Figure 1). We believe this reflects the flexibility and ability of the PGI to prioritise items, and this not only. 171. can improve responsiveness but also comes closer to what we conceptualise as 'HRQoL' which is always specific to an. 172. individual [22]. Further study of the PGI and other individualised instruments in OA, and other conditions, would seem. 173. warranted.. 174. 7.

(8) 175. The study has several caveats. Firstly, a disease (OA) specific questionnaire was not included so assessment of. 176. performance between the three types of HRQoL instruments (generic, disease-specific, individualised) could not be. 177. undertaken. Such an assessment in OA would be of great interest. Secondly, we used only one widely recognised. 178. quantitative method (SRM) to assess the responsiveness to one treatment modality, specifically over-the-counter doses. 179. of established oral analgesics. Other methods for assessing responsiveness and other treatments with different efficacy. 180. potential and different associated contextual responses may result in different outcomes in HRQoL instruments. The. 181. same applies to content analysis as evaluating content validity still remains a challenge and various methods exist [23].. 182. Thirdly, although OA is a chronic condition the length of our study was only 13 weeks so the relative performance of the. 183. two HRQoL instruments over a longer time period remains unknown. Finally, the PGI is a relatively complex instrument. 184. which participants can have difficulties with if there is no access to appropriate guidance and support.. 185 186. Conclusions. 187 188. The PGI has been shown to be responsive to analgesic treatment in community-derived people with knee pain. In. 189. contrast only one domain in the SF-36 (Bodily Pain) is responsive to this treatment. The PGI is able to elicit HRQoL. 190. areas which are not captured by generic instruments and appears to be a sensitive tool in the assessment of HRQoL in. 191. knee pain and OA. It appears to merit further study by inclusion in future OA clinical trials.. 192 193. Compliance with Ethical Standards. 194 195. Funding source: There is no specific grant associated with this study. Reckitt Benckiser Healthcare International Ltd. 196. funded the original randomised controlled trial which provided the data for analysis but the conception and design of the. 197. current nested study was undertaken independently to the funder with their approval.. 198 199. Disclosure of potential conflicts of interest: The authors declare no conflicts of interest.. 200. 8.

(9) 201. Ethical approval: All procedures performed in studies involving human participants were in accordance with the ethical. 202. standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later. 203. amendments or comparable ethical standards.. 204 205. Informed consent: Informed consent was obtained from all individual participants included in the study.. 206 207. Acknowledgments. 208 209. The authors are grateful to Reckitt Benckiser Healthcare International Ltd for financial support for the initial RCT and to. 210. the other investigators and research nurses involved in the trial which generated the data analysed in the present study.. 211. 9.

(10) 212. References. 213 214. [1]. 215. framework and item selection. Medical Care, 30(6), 473-483.. 216. [2]. 217. measurement of quality of life. The Patient-Generated Index. Medical Care, 32(11), 1109-1126.. 218. [3]. 219. spondylitis-specific health related quality of life: evaluation of the Patient Generated Index. The Journal of. 220. Rheumatology, 30(4), 764-773.. 221. [4]. 222. Generated Index (PGI) of quality of life: a graded structured review. Quality of Life Research, 16(4), 705-715.. 223. [5]. 224. four common conditions. Quality in Health Care, 8(1), 22-29.. 225. [6]. 226. H.C. (2010). The COSMIN checklist for assessing the methodological quality of studies on measurement properties. 227. of health status measurement instruments: An international delphi study. Quality of Life Research, 19(4), 539–549.. 228. [7]. 229. guidelines for reporting parallel group randomized trials. Annals of Internal Medicine, 152(11), 726–732.. 230. [8]. 231. (2013). Reporting of Patient-Reported outcomes in Randomized Trials: The CONSORT PRO Extension. JAMA, Feb. 232. 27; 309(8), 814-822.. 233. [9]. 234. as a criterion for selecting health status measures. Quality in Health Care, 1(2), 89-93.. 235. [10]. 236. index as compared with the SF-36 in patients with osteoarthritis of the legs undergoing a comprehensive. 237. rehabilitation intervention. Annals of the Rheumatic Diseases, 60(9), 834-840.. Ware J.E., Jr., Sherbourne C.D. (1992). The MOS 36-item short-form health survey (SF-36). I. Conceptual. Ruta D.A., Garratt A.M., Leng M., Russell I.T., MacDonald L.M. (1994). A new approach to the. Haywood K.L., Garratt A.M., Dziedzic K., Dawes P.T. (2003). Patient centered assessment of ankylosing. Martin F., Camfield L., Rodham K., Kliempt P., Ruta D. (2007). Twelve years' experience with the Patient. Ruta D.A., Garratt A.M., Russell I.T. (1999). Patient centred assessment of quality of life for patients with. Mokkink, L.B., Terwee, C.B., Patrick, D.L., Alonso, J., Stratford, P.W., Knol, D.L., Bouter L.M., de Vet. Schulz, K.F., Altman, D.G., Moher, D., & Group, C. (2010). CONSORT 2010 statement: Updated. Calvert M., Blazeby J., Altman D.G., Revicki D.A., Moher D., Brundage M.D.; CONSORT PRO Group.. Fitzpatrick R., Ziebland S., Jenkinson C., Mowat A., Mowat A. (1992). Importance of sensitivity to change. Angst F., Aeschlimann A., Steiner W., Stucki G. (2001). Responsiveness of the WOMAC osteoarthritis. 10.

(11) 238. [11]. 239. for meta-analysis of knee osteoarthritis trials: empirical evidence from a survey of high impact journals. Arthritis,. 240. 2012, 136245.. 241. [12]. 242. osteoarthritis: comparison of generic and disease-specific instruments. Clinical Rheumatology, 24(1), 29-37.. 243. [13]. 244. controlled trial of ibuprofen, paracetamol or a combination tablet of ibuprofen/paracetamol in community-derived. 245. people with knee pain. Annals of the Rheumatic Diseases, 70(9), 1534-1541.. 246. [14]. 247. Journal of the American Medical Association, 140(8), 659-662.. 248. [15]. 249. Manual for the SF-36v2(TM) Health Survey (2nd Ed). Lincoln, RI, USA: QualityMetric Incorporated.. 250. [16]. 251. United Kingdom. Journal of Epidemiology & Community Health, 53(1), 46-50.. 252. [17]. 253. coefficient. Health and Quality of Life Outcomes, 3, 23.. 254. [18]. 255. Associates.. 256. [19]. 257. sensitivity of short and longer health status instruments. Medical Care, 30(10), 917-925.. 258. [20]. 259. generic outcome measure in clinical trials of patients with osteoarthritis and rheumatoid arthritis: tests of data. 260. quality, scaling assumptions and score reliability. Medical Care, 37(5 Suppl), MS10-22.. 261. [21]. 262. Should aggregate scores of the Medical Outcomes Study 36-item Short Form Health Survey be used to assess quality. 263. of life in knee and hip osteoarthritis? A national survey in primary care. Osteoarthritis Cartilage, 15(9), 1013-1018.. Juhl C., Lund H., Roos E.M., Zhang W., Christensen R. (2012). A hierarchy of patient-reported outcomes. Salaffi F., Carotti M., Grassi W. (2005). Health-related quality of life in patients with hip or knee. Doherty M., Hawkey C., Goulder M., Gibb I., Hill N., Aspley S., Reader S. (2011). A randomised. Steinbrocker O., Traeger C.H., Batterman R.C. (1949). Therapeutic criteria in rheumatoid arthritis. The. Ware J.E., Jr., Kosinski M., Bjorner J.B., Turner-Bowker D.M., Gandek B., Maruish M.E. (2007). User's. Jenkinson C., Stewart-Brown S., Petersen S., Paice C. (1999). Assessment of the SF-36 version 2 in the. Stratford P.W., Riddle D.L. (2005). Assessing sensitivity to change: Choosing the appropriate change. Cohen J . (1988). Statistical power analysis for behaviour sciences. (2nd Ed). Hillsdale, N.J. : L. Erlbaum. Katz J.N., Larson M.G., Phillips C.B., Fossel A.H., Liang M.H. (1992). Comparative measurement. Kosinski M., Keller S.D., Hatoum H.T., Kong S.X., Ware J.E., Jr. (1999). The SF-36 Health Survey as a. Rannou F., Boutron I., Jardinaud-Lopez M., Meric G., Revel M., Fermanian J., Poiraudeau S. (2007).. 11.

(12) 264. [22]. 265. towards a comprehensive model. Health and Quality of Life Outcomes, 2, 32.. 266. [23]. 267. practices based on science and experience. Quality of Life Research, 18, 1263-1278.. 268. [24]. 269. validity - establishing and reporting the evidence in newly developed patient-reported outcomes (PRO) instruments. 270. for medical product evaluation: ISPOR PRO good research practices task force report: part 1 - eliciting concepts for. 271. a new PRO instrument. Value in Health, 14(8), 967-977.. 272. [25]. 273. validity - establishing and reporting the evidence in newly developed patient-reported outcomes (PRO) instruments. 274. for medical product evaluation: ISPOR PRO Good Research Practices Task Force report: part 2 - assessing. 275. respondent understanding. Value in Health, 14(8), 978-988.. Asadi-Lari M., Tamburini M., Gray D. (2004). Patients' needs, satisfaction, and health related quality of life:. Brod M., Tesler L.E., Christensen T.L. (2009). Qualitative research and content validity: developing best. Patrick D.L., Burke L.B., Gwaltney C.J., Leidy N.K., Martin M.L., Molsen E., Ring L. (2011). Content. Patrick D.L., Burke L.B., Gwaltney C.J., Leidy N.K., Martin M.L., Molsen E., Ring L. (2011). Content. 276 277 278. Figure legend. 279. Figure 1. 280. Areas of life affected by knee pain as self-reported by the patients in baseline Patient Generated Index (PGI).. 12.

(13)

New documents

Abstract: Background: Multikinase inhibitors in combination with chemotherapy have recently been evaluated in patients with advanced breast cancer ABC in the adjuvant treatment, but

To implement cross-lingual delexicalized parsing, we used the Romance language corpora from the UD project as training material, we created a manually annotated sample of Occitan to be

Our objective is therefore twofold: first we want to see how different properties of syntactic dependency trees correlate, in particular properties that are related to syntactic

- The patients were classified according to the GINA asthma guidance of asthma manage- ment into moderate or severe persistent bron- chial asthma... - Full clinical examination and

Hence, when the antecedent is a subject, both obliqueness and topicality as manifested in word order favor the choice of a reflexive possessive, leading to a very high proportion of

Conclusions: Significant hypomethylation and high expression of TNFAIP8 in the placenta and peripheral blood cells were observed in EOPE, suggesting TNFAIP8 may be associated with the

Non-projective sentences in the Czech treebank we used are significantly longer with the 95% confi-.. Length of projective and non-projective sentences in the Czech treebank with 95%

The size of the flux in an inter-word position is the number of dependencies that cross this position.1 The flux size of a sentence is the sum of the sizes of the inter-word fluxes...

We established a mammalian Fra-1 overexpressing plasmid pcDNA3.1/Fra-1 in MCF-7 and MDA-MB-231 cells and the results showed that transfection of the plasmid can substantially upregulate

Apart from the quantitative distribution of dependents, we study two types of word order constraints conditioned on the dependency pattern, namely head direction and sib- ling