UBC Faculty Research and Publications

Classifying health-related quality of life outcomes of total hip arthroplasty Xu, Min; Garbuz, Donald S; Kuramoto, Lisa; Sobolev, Boris Sep 6, 2005

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-12891_2005_Article_161.pdf [ 1.04MB ]
Metadata
JSON: 52383-1.0074641.json
JSON-LD: 52383-1.0074641-ld.json
RDF/XML (Pretty): 52383-1.0074641-rdf.xml
RDF/JSON: 52383-1.0074641-rdf.json
Turtle: 52383-1.0074641-turtle.txt
N-Triples: 52383-1.0074641-rdf-ntriples.txt
Original Record: 52383-1.0074641-source.json
Full Text
52383-1.0074641-fulltext.txt
Citation
52383-1.0074641.ris

Full Text

ralssBioMed CentBMC Musculoskeletal DisordersOpen AcceResearch articleClassifying health-related quality of life outcomes of total hip arthroplastyMin Xu*1, Donald S Garbuz2, Lisa Kuramoto3 and Boris Sobolev3Address: 1Arthritis Research Centre of Canada, Vancouver, BC, Canada, 2Department of Orthopaedics, University of British Columbia, Vancouver, BC, Canada and 3Centre for Clinical Epidemiology & Evaluation, Vancouver, BC, CanadaEmail: Min Xu* - mxu@arthritisresearch.ca; Donald S Garbuz - garbuz@shaw.ca; Lisa Kuramoto - lkuramot@vanhosp.bc.ca; Boris Sobolev - sobolev@interchange.ubc.ca* Corresponding author    AbstractBackground: Primary total hip arthroplasty (THA) is an effective treatment for hip osteoarthritis,assessed by whatever distribution-based measures of responsiveness. Yet, the group levelevaluation has provided very little evidence contributes to our understanding of the large variationof treatment outcome. The objective is to develop criteria that classify individual treatment healthrelated quality of life (HRQOL) outcome after primary THA, adjusted by preoperative scores.Methods: We prospectively measured 147 patients' disease specific HRQOL on the date ofconsultation and 12 months post operation by Western Ontario McMaster UniversitiesOsteoarthritis Index (WOMAC). Regression models were used to determine the "expected"outcome for a certain individual baseline score. The ceiling effect of WOMAC measurement isaddressed by implementing a left-censoring method.Results: The classification criteria are chosen to be the lower boundary of the 95% confidenceinterval (CI) of the estimated median from the regression. The robustness of the classificationcriteria was demonstrated using the Monte-Carlo simulation.Conclusion: The classification criteria are robust and can be applied in general orthopaedicresearch when the sample size is reasonable large (over 500).BackgroundStatistical tests are frequently called upon to assess treat-ments whose effect size is small or whose reduction of riskis modest, as is often the case with emerging treatments.But what of the evaluation of mature treatments whoseeffect is known to be substantial?Primary total hip arthroplasty (THA) is an effective treat-ment for patients with severe hip osteoarthritis (OA). TheYet, there are large variations reported in treatment out-come and very little good-quality evidence contributes toour understanding of this variation [7,8]. The evidence islimited mostly to patient- and implant-related factors [7].The role of variations in service delivery practices andother factors remains unclear. On the whole, patients' out-come is good. Nevertheless, the development of classifica-tion criteria to differentiate between overall good results isnecessary to achieve a better understanding of the vari-Published: 06 September 2005BMC Musculoskeletal Disorders 2005, 6:48 doi:10.1186/1471-2474-6-48Received: 15 February 2005Accepted: 06 September 2005This article is available from: http://www.biomedcentral.com/1471-2474/6/48© 2005 Xu et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.Page 1 of 9(page number not for citation purposes)improvement is large by any of the measures of respon-siveness commonly used in orthopaedic research [1-6].ance and ultimately to reduce the number of relativelypoor performers.BMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48Because THA aims to improve physical function andrelieve pain, and because it is broadly successful in itsgoal, health-related quality of life (HRQOL) is generallyacknowledged to be the primary outcome of interest[9,10]. Assessment of HRQOL is typically made at thegroup level, that is, by measures such as the t-test, effectsize (ES), and standard response mean (SRM), that arecharacteristic of a group. However, using such methodsmay not provide the best evidence to explain the associa-tion between postoperative outcomes and risk factors [8].In contrast, the measurement of individual changes is anincreasingly attractive method of quantifying HRQOLoutcomes because it has the potential to document objec-tively the patient-perceived impact of treatment. Expecta-tion and satisfaction are highly individualized; theycontribute significantly to self-assessed quality of life. Butthese individualizing influences are lost in statistics suchas pre- and postoperative mean scores that only express agroup [11]. Raising the mean outcome is a worthwhileobjective, especially when the mean badly needs improve-ment. When, as with effective treatments, the mean is notan overriding concern, it is appropriate to turn our atten-tion to individuals within the mean [12]. Even groupswhose mean change due to treatment is equivalent arelikely to contain individuals who did substantially betterand worse than others [13]. Developing statistical meth-ods to assess these differences rather than the meansthemselves is a natural accompaniment to the refinementof treatments such as hip arthroplasty.Two recent studies have found an association betweenpreoperative health status and postoperative outcomes[14-16]. Fortin et al. examined the relationship betweenpreoperative functional status and postoperative out-comes in a prospective cohort study using the WesternOntario McMaster Universities Osteoarthritis Index(WOMAC) and Short Form 36 (SF-36). They found thatpoorer preoperative function was the strongest predictorof pain and functional outcomes at 6 and 24 months afterTHA [15,16]. The authors concluded that surgery per-formed later in the natural history of functional declineresults in worse postoperative functional status. They alsonoted that function and pain in patients with lower pre-operative function did not improve after the operation tothe level achieved by those with higher preoperativescores [15,16]. Thus, measures of postoperative HRQOLoutcome need to be adjusted by preoperative functionalstatus.MethodsThe objective of this study is to develop a tool for classify-ing the HRQOL outcome of THA based on the individ-ual's preoperative HRQOL score. The development of thestudy evaluating the postoperative outcome for THA. Inthe development of the instrument section, a left-cen-sored linear regression model is employed as a means ofunderstanding and communicating the relationshipbetween baseline and expected outcome. An expectedpostoperative HRQOL score for each individual preopera-tive score is estimated using this left-censored linearregression model. By using the expected HRQOL out-come, we identify patients whose benefit from THA is"better than expected." The performance of these classifi-cation criteria is evaluated in difference sample sizes bysimulation. In the development of these classification cri-teria we adjust the postoperative outcome by its preoper-ative score. The result of this simulation study shows thatthese classification criteria are robust.Study populationData from a prospective cohort study were used for a casestudy. This study included 201 patients registered on thewait list for THA between March, 2001 and May, 2003with 147 patients completed follow-up ending in March,2004. This study was conducted at the Vancouver Hospi-tal & Health Sciences Centre. Ethical approval was issuedby the University of British Columbia Clinical EthicsReview Board. Patients presenting during this period atthe Division of Reconstructive Orthopedics at VancouverHospital (VH) with a diagnosis of osteoarthritis (OA) andrequiring primary THA are included in the study. OA isdefined by the American College of Rheumatology's(ACR) clinical classification criteria for OA of the hip [17].Patients were excluded for the following reasons: previousThe distribution of baseline WOMAC functional scoresFigure 1The distribution of baseline WOMAC functional scores.Page 2 of 9(page number not for citation purposes)tool and the results of a simulation study are presented inthe methods section. We describe the design of a caseTHA to the index joint; inflammatory arthritis; bilateralTHA performed simultaneously; inability to respond to aBMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48questionnaire in English; and urgent surgery performedwithin 28 days after the decision for THA.Every patient requiring hip arthroplasty was requested tocomplete the WOMAC questionnaire on the date of con-sultation. The questionnaire is self-administered. Medicaloffice assistants handed each patient a WOMAC question-naire once the decision was reached to enter the wait list.To assess postoperative outcomes, WOMAC question-naires were mailed at 12 months following surgery.WOMAC is recommended for OA-specific outcomes[18,19]. It contains dimensions for pain (5 items), stiff-ness (2 items), and function (17 items). Dimensions areequally weighted and reported as sums, where the highernumber indicates a greater burden of OA. At present it isthe most frequently used measure of pain and self-reported disability among arthroplasty patients [10]. TheWOMAC questionnaire has 24 questions, each question isgiven a Likert scale response from 0 (best health state) to4 (worst health state). The WOMAC score for each sub-scale is calculated as the sum of the scores of eachquestion included in the subscale. The range of eachsubscale is as follows: function: 0–68; pain: 0–20; stiff-ness: 0–8.Patients' names and provincial health numbers were usedto obtain age and gender through the medical officeadministrative database. Co-morbidity information wasobtained through medical chart review using the Charnleyclassification, which stratifies patients by the presence ofOA in one or both hips, or a co-morbid condition thatimpairs walking. This scale allows a meaningful compari-son between groups [20]. The Charnley classes we usedare:A: Single hip with osteoarthritisB1: Bilateral hips with arthritisB2: Previous THA on the contra-lateral hipC: Multiple joints affected with arthritis or a chronic diseasethat affects HRQOL (specifically walking)Statistical analysisLog-linear regression modelIn the following, we aim at building a linear model toexplore the relationship between follow-up score andbaseline score. Since the distribution is skewed (Fig. 1 &Fig. 2), one cannot use the follow-up score in a linearregression analysis as an outcome variable. We found thatthe logarithms of the follow-up WOMAC functionalscores follow a symmetrical distribution (Fig. 3). There-log(Follow-up) = α+β*Baseline+σ*ε,where Follow-up is the follow-up WOMAC score and Base-line is the baseline WOMAC score. The error term ε fol-lows a normal distribution with a mean of 0 and astandard deviation of 1, and σ is a fixed constant thatchanges the variability of the expected value.However the observed follow-up data has some WOMACThe distribution of follow-up WOMAC function scoresFigure 2The distribution of follow-up WOMAC function scores.The distribution of the logarithms of follow-up WOMAC functional scoresFigure 3The distribution of the logarithms of follow-up WOMAC functional scores.Page 3 of 9(page number not for citation purposes)fore we build a log-linear regression as following: function score equal to zero and the logarithm of 0 is infi-nite. So we can not censor the postoperative score at 0.BMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48Moreover, the WOMAC function 0 is corresponding tocomplete freedom from joint symptoms. It is unlikely thatpatients before and after THA would have no detectableimpairment in their hip. Therefore, as a measurementtool, the WOMAC questionnaire is limited in providingHRQOL information at the extreme low end of the scale(score of 0). Since a true score is unknown when the scoreis between 0 and 1, we regard measurements below 1 asleft-censored observations. In our model, we chose 0.9 tobe the censoring point so that 1 was preserved in themodel and 0 was censored. We transformed the observedfollow-up score as follows:Follow-up = 0.9, if Follow-up <= 0.9;Follow-up = Follow-up, if Follow-up > 0.9,The Tobit regression model is a well known instrumentfor measuring left-censored variables in economicresearch [21]. In order to incorporate the left-censoredobservations in the regression analysis, we built a Tobitmodel to incorporate the left-censored observations in theregression analysis. The maximum likelihood methodwas used to estimate the probabilities of log(Follow-up)given the baseline WOMAC score. The regression analysiswas conducted using the SAS 8.1 PROC LIFEREGprocedure.Instrument for classifying function outcomesThrough this regression analysis, an expected postopera-tive outcome for each baseline WOMAC functional scorewas obtained. Due to the skewed distribution of follow-up scores, we used the median of the follow-up scoreinstead of the mean as the classification criteria. The meanof the predicted logarithm of follow-up scores was esti-mated through the model, and the median of estimatedfollow-up scores is exp (mean of log(Follow-up)) accord-ing to the mathematical transformation.Since the model is derived from a rather small size sam-ple, the variation of the estimated median of the follow-up scores should be taken into consideration. Using thelower 95% confidence interval of the median as a cutoffpoint associated with the baseline score, the studypatients were divided into two groups. Group I: Patientsbelow the line were considered to have achieved a "better thanexpected" outcome. Group II: Patients above the line were con-sidered to have achieved a "not better than expected" outcome.Assessment of the classification instrumentWe implemented the Monte-Carlo simulation method toinvestigate the robustness of our classification criteria.Our intention was to assess the robustness across baselinesystematic relationship between the baseline scores andpostoperative WOMAC functional scores and adding arandom component. The systematic relationship and theparameters for the random component were specifiedfrom the Tobit model estimates in our case study. In eachdata set, postoperative scores were generated for baselineWOMAC functional scores fixed at 10, 17, 34, 51, and 68.We chose 10 since it is the lowest baseline score that is eli-gible for surgery and available in the case study. The func-tional subscale contains 17 questions; each question has aresponse on Likert scale from 0 to 4. Therefore, we chosethe folds of 17 as the baseline levels for simulation. Thepostoperative score was left-censored at 0.9. We looked atsample sizes increasing from 100 to 500 in increments of100. For each sample size, we generated 1000 data sets.Then, for each data set, the regression model was fit andthe median postoperative score and the cutoff points (ie.the lower bound of the 95% CI for the median score) wereestimated at each baseline score.We also tested the model using same method with differ-ent censoring points (0.9, 2, 3, 4, and 5) while the samplesize was fixed at 500. For each censoring point, we gener-ated 1000 datasets and the cutoff points were estimatedfor each data set.ResultsStudy populationThis study included 201 patients, among which there are147 patients completed follow-up ending in March, 2004.The average age is 64.8 years and there are 83 females(56%) and 66 males (44%) in the study. Seventy-twopatients (50%) have only one joint involved with OA; 34patients have bilateral disease. Of these 34 patients, thereare 18 with contra-lateral hip replacement prior to theindex surgery and 16 patients with moderate to severe OAin the contra-lateral hip. Thirty-nine patients (27%) havemultiple joints involved with OA or have a chronic sys-tematic disease. When compare the component of age,gender and disease statues, there are no statistical differ-ences of between the 147 patients and 54 patients whodid not complete follow-up. In the following analysis, allthe results are based on the 147 patients who completedTable 1: Parameter estimation for the log-linear regressionParameters Estimates 95% CIIntercept 0.98 0.29–1.67Coefficient 0.03 0.01–0.05Scale 1.30 1.14–1.48Page 4 of 9(page number not for citation purposes)scores and for different sample sizes. We generated ran-dom postoperative WOMAC functional scores assuming afollow-up. We found that the distribution of baselineWOMAC functional scores (scale 0–68) follows aBMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48symmetrical distribution and its' mean and standard devi-ation (SD) are 39 and 13 respectively (Fig. 1). Its mini-mum is 10 points and median is 41 points. While at theend of the follow-up, the distribution of WOMAC func-tional scores (scale 0–68) shows a truncated distributionbecause the follow-up outcome is nearly as good as a fullrecovery or normal function; that is, the follow-up out-comes have a limit as a score of 0 (best function). Themean follow-up WOMAC functional score is 14 (SD =14). Its minimum is 0 points and median is 8.5 points.Since the distribution is skewed, one cannot use the fol-low-up score in a linear regression analysis as an outcomevariable.Log-linear regression modelTable 1 shows the parameter estimates obtained throughthis regression analysis. For the sample population, theestimate of the expected value of the lognormal distribu-tion is given by:Log(Follow-up) = 0.98+0.03*Baseline.Based on the model, the baseline WOMAC functionalscore is a significant predictor of the follow-up WOMACfunctional score (p = 0.0005). Increasing the baselinescore by 10 points raises the estimated postoperative scoreby approximately 35%. The estimated median score line95% confidence interval for the median of expected function outcomesFigure 495% confidence interval for the median of expected function outcomes.Page 5 of 9(page number not for citation purposes)and its 95% confidence interval are shown in Fig. 4.BMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48Age, gender, co-morbidity, and waiting time were alsotested as covariates in the log-linear regression model.None of these variables were significant predictors of thefollow-up WOMAC functional score and there was no dif-ference in the regression coefficient for baseline WOMACfunctional scores with or without these covariates. Agoodness of fit test showed that the Tobit model is well fit-ted. In the model, the outliers are detected by the studen-tized residual; those observation having an absolutestudentized residual over 3.5 were removed.Instrument for classifying function outcomesThe simulation results are summarized in Fig. 5, 6, 7, 8and Table 2. Fig. 5 summarizes that the median estima-tion is very consistent despite the increase in sample size.Table 2 represents the same information as Fig. 5, but pro-vides the actual values for classification criteria that can beused as a reference table for future researchers.Fig. 6 summarizes the results of classification criteria insimulated data sets with different sample size. While thesample size increases, the mean of the cutoff pointsapproaches the mean of the median estimation. That is,when the sample size is reasonable large (n = 500), the cutoff points are almost equivalent to the estimated median.Fig. 7 summarizes the coefficient of variation (CV) for thedistribution of classification criteria, in simulated datasets with different sample size. The lower the CV is, thehigher the precision of the estimation is. This plot showstwo trends. First, the CV is lowest at the median baselinelevel and increases toward the extreme values in bothdirections, as expected. For example, in a sample of 100patients, the CV of the estimated cutoff point is 12.5% atbaseline 34, 20.2% at baseline 10 and 22.8% at baseline68. That is, the precision of this estimation is the highestwhen the baseline is around 34 and reduced toward bothextremes. Second, we also found that the CV decreaseswith the sample size. For example, at baseline 34, the CVindicates that the precision of the estimation increaseswith a larger sample size.Fig. 8 summarizes the CV for the distribution of classifica-tion criteria, in simulated data sets with differentcensoring points. While sample sizes being fixed at 500,we found that the CV increases with higher censoringlevel. This indicates that the precision of the estimationincreases with a lower censoring point. Therefore censor-ing postoperative scores at 0.9 is preferred over censoringat a higher level.DiscussionIn the past decade the orthopaedic community has shiftedtoward the inclusion of patient-based measures of out-come assessments [22]. It was typical of earlier orthopedicpractice that the patient's perspective received less atten-tion than did clinician's measures of disease andimpairment [23,24]. Clinicians used complication rates,mortality, most frequently revision rates and clinical judg-ment to assess the degree of improvement [25]. SinceTHA, in most cases, aims explicitly to improve HRQOL,using HRQOL measures as endpoint in orthopaedicresearch on evaluation of treatment outcome is now seenas a necessity to fully understand the effects of this inter-vention [26].THA is an effective treatment by any of the distribution-based measures of responsiveness [6]. Yet, there are largevariations reported in treatment outcome. Why somepatients do better than others post-operation? Group levelperspective on evaluation of treatment outcome has pro-vided very little evidence contributes to our understand-ing of this question.Most reports of the HRQOL outcome of THA use distribu-tion-based approaches that test the significance of thechange due to treatment [22]. However, distribution-based approaches are based on the statistical characteris-Table 2: Average of the classification criteriaAverage of the classification criteriaBaseline N = 100 N = 200 N = 300 N = 400 N = 50010 3.5 3.5 3.4 3.4 3.517 4.2 4.2 4.2 4.2 4.234 6.8 6.8 6.8 6.8 6.851 11.1 11.1 11 11 1168 18.2 18.1 17.9 17.8 17.8Page 6 of 9(page number not for citation purposes)is 12.5% for 100 scores and 5.3% for 500 scores. This tics of the sample. For example, paired t-statistics arefrequently used to estimate the statistical significance ofBMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48the change [27]. The problem with using the t-test as ameasure of change is that it focuses exclusively on thesignificance which will inevitably increase with samplesize [28]. A different problem exists in determining theminimal clinically significant difference for THA. Inimprovement in arthritis symptoms [29]. But the 9.3-point change is too small to be applied to the outcomes ofTHA which typically show a 60–100% improvement overbaseline [15,16]. The expected change in WOMAC func-tional scores after THA is four times larger than the mini-mal clinically important difference derived from drugtrials in OA. Effect size (ES) and standard response mean(SRM) are also common measures for responsiveness atthe group level. Cohen's criteria can be used to classifyresponsiveness as mild, moderate, and large [30]. Butthese statistics may be influenced by the heterogeneity ofthe sample. Moreover, Cohen's magnitude of effect doesnot suit the nature of orthopaedic surgery. An effect sizelarger than 0.8 is considered a "large effect". However, bythat criterion, the majority of patients in our case studywould be considered to have experienced a "large effect"both by ES and SRM statistics. Such criteria are inadequatefor documenting the positive impacts of treatment. Char-acteristics of the baseline distribution will strongly influ-ence the effect size, while variability of the change in thesample may influence the standard response mean.We developed a method to classify the HRQOL outcomeon an individual level. Group distributions can have anegligible mean difference with large variance. Therefore,the large differences that are important to individuals arenot measured by group level, whereas the individual leveltakes them into account. This makes the individual per-spective important for clinical treatment decisions[13,28]. We have shown that improvement after hiparthroplasty is not as big when the patients have a betterpreoperative score; therefore, postoperative outcomes arenot evaluated at the group level but rather at eachindividual baseline level, so that for each individualpatient an expected outcome can be generated.We addressed the ceiling effect of the WOMAC instrumentin the measurement of postoperative outcome of THA, as10% of patients in our case study recorded a postoperativeWOMAC score of 0. A ceiling effect occurs when a patientcan improve only minimally or not at all. In the presenceof a ceiling effect, the paper by Austin et al. suggests thatthe coefficient estimates from the left-censored regressionmodel are better than the estimates from a least squareregression [31]. We address the ceiling effect by imple-menting a linear regression of log-transformed WOMACfunction score while treating postoperative scores as left-censored at 0.9. The regression model represents the rela-tionship between baseline and postoperative outcome.The estimated median of postoperative scores was chosento distinguish between those who are able to benefit fullyfrom treatment and those who are not. Due to the smallThe average of the median estimates from simulationFigure 5The average of the median estimates from simulation.Average of the estimated median and classification criteriaFigure 6Average of the estimated median and classification criteria.Page 7 of 9(page number not for citation purposes)clinical drug trials, a 9.3 point change in WOMAC func-tional score was accepted as a minimally significantsample size, the classification criteria in this case study isthe lower boundary of the 95% confidence interval (CI) ofBMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48the estimated median. Our study results agrees with theprevious literature in that postoperative HRQOL scoreswere found to be strongly associated with their baselinepost operation. The effects of age, gender, and co-morbid-ity on follow-up WOMAC scores were not statistically sig-nificant, so these are excluded from the regression model.The performance of the classification criteria was demon-strated using the Monte-Carlo simulation. The variationof the classification criteria will decrease with increasingsample size; likewise, the classification criteria becomecloser to the estimated median with increasing samplesize. Thus, with a small sample set, researchers could usethe lower boundary of 95% CI of the estimated median asthe classification criteria. When there is a reasonablelarger sample (bigger than 500), one could use the esti-mated median itself as the classification criteria.The limitation of this research is that the estimated classi-fication criteria were not validated in a different clinicalsetting. Instead, they were evaluated through simulation.Therefore, we are recommending that clinicians use onlythe methods rather than the actual values of the classifica-tion criteria until further research is done in this area.ConclusionThe contribution of this paper is two-fold. First, the crite-ria for classify individual treatment outcome adjusted bybaseline score was proposed. The development of theseclassification criteria also addresses the ceiling effect of theHRQOL measurement. Second, the performance of theclassification criteria was found to be more precise with areasonable larger sample size (n > 500). Vancouver Hos-pital (VH) is a tertiary referral centre and teaching hospitalfor the University of British Columbia (UBC). The demo-graphics of arthroplasty patients, however, are not differ-ent from elsewhere. The study result is expected to begeneralizable to a similar clinical setting.This paper provides intuitive criteria for classifyingHRQOL outcomes based on individual scores before sur-gery. The result of this method is an individual outcomewhich can serve as a standard advice for patient coun-seling based on HRQOL status at consultation. It givesorthopaedic researchers a means of defining "success" ofeffective surgery. In the future, we will evaluate thismethod in different populations and with other HRQOLinstruments such as Oxford Hip Score and the Short Form12 questionnaire.Competing interestsThe author(s) declare that they have no competinginterests.Authors' contributionsAnalysis of data, interpretation and the original draft wereThe coefficient of variation of the classification criteria from simulation (Different sample size)Fig re 7The coefficient of variation of the classification criteria from simulation (Different sample size).The coefficient of variation of the classification criteria from simulation (Different censor point)Fig re 8The coefficient of variation of the classification criteria from simulation (Different censor point).Page 8 of 9(page number not for citation purposes)values. We evaluated the changes in the WOMACdimensions of pain, stiffness, and function from pre- tocompleted by Min Xu. Donald Garbuz conceived thestudy, participated in the design and contributed to clini-Publish with BioMed Central   and  every scientist can read your work free of charge"BioMed Central will be the most significant development for disseminating the results of biomedical research in our lifetime."Sir Paul Nurse, Cancer Research UKYour research papers will be:available free of charge to the entire biomedical communitypeer reviewed and published immediately upon acceptancecited in PubMed and archived on PubMed Central BMC Musculoskeletal Disorders 2005, 6:48 http://www.biomedcentral.com/1471-2474/6/48cal conception and interpretation. Lisa Kuramotoperformed the Monte-Carlo simulation in the statisticalanalysis. Boris Sobolev also participated in the design,provided critical evaluation of methodological contentand revision of the manuscript. All authors read andapproved the final manuscript.AcknowledgementsWe thank Mr. James Latteier and Mr. Francisco Luna for data collection. Findings previously presented at annual meeting of American College of Rheumatology, San Antonio, Texas, 2004. Abstract Title: Effect of Delays on Individual Quality of Life Outcome after Primary Total Hip Arthroplasty.References1. Cabanela ME, Campbell DC 2nd, Henderson ED: Total jointarthroplasty. The hip.  Mayo Clin Proc 1979, 54(9):559-63.2. NIH consensus conference: Total hip replacement. NIH Con-sensus Development Panel on Total Hip Replacement.  JAMA1995, 273(24):1950-6.3. Harris WH, Sledge CB: Total hip and total knee replacement(1).  N Engl J Med 1990, 323(11):725-31.4. Harris WH, Sledge CB: Total hip and total knee replacement(2).  N Engl J Med 1990, 323(12):801-7.5. Jones CA, Voaklander DC, Johnston DW: Health related qualityof life outcomes after total hip and knee arthroplasties in acommunity based population.  J Rheumatol 2000, 27(7):1745-52.6. Laupacis A, Bourne R, Rorabeck C: The effect of elective total hipreplacement on health-related quality of life.  J Bone Joint SurgAm 1993, 75(11):1619-26.7. Liang MH, Cullen KE, Poss R: Primary total hip or knee replace-ment: evaluation of patients.  Ann Intern Med 1982, 97:735.8. Fitzpatrick R, Norquist JM, Dawson J, Jenkinson C: Rasch scoring ofoutcomes of total hip replacement.  J Clin Epidemiol 2003,56(1):68-74.9. Gartland JJ: Orthopaedic clinical research. Deficiencies inexperimental design and determinations of outcome.  J BoneJoint Surg Am 1988, 70(9):1357-64.10. Ethgen O, Bruyere O, Richy F, Dardennes C, Reginster JY: Health-related quality of life in total hip and total knee arthroplasty.A qualitative and systematic review of the literature.  J BoneJoint Surg Am 2004, 86-A(5):963-74.11. Bowling A: Measuring health. A review of quality of life meas-urement scales.  Buckingham, England: Open University Press;1992. 12. Hageman WJ, Arrindell WA: Establishing clinically significantchange: increment of precision and the distinction betweenindividual and group level of analysis.  Behav Res Ther 1999,37(12):1169-93.13. Samsa G, Edelman D, Rothman ML, Williams GR, Lipscomb J, MatcharD: Determining clinically important differences in health sta-tus measures: a general approach with illustration to theHealth Utilities Index Mark II.  Pharmacoeconomics 1999,15(2):141-55.14. MacWilliam CH, Yood UM, Verner JJ, McCarthy BD, Ward RE:Patient-related risk factors that predict poor outcome aftertotal hip replacement.  Health Serv Res 1996, 31:623-638.15. Fortin PR, Clarke AE, Joseph L, Liang MH, Tanzer M, Ferland D, et al.:Outcomes of total hip and knee replacement: Preoperativefunctional status predicts outcomes at six months aftersurgery.  Arthritis Rheum 1999, 42:1722-1728.16. Fortin PR, Penrod JR, Clarke AE, St-Pierre Y, Joseph L, Belisle P, et al.:Timing of total joint replacement affects clinical outcomesamong patients with osteoarthritis of the hip or knee.  ArthritisRheum 2002, 46(12):3327-30.17. Altman R, Alarcon G, Appelrouth D, Bloch D, Borenstein D, BrandtK, et al.: The American College of Rheumatology criteria forthe classification of reporting of osteoarthritis of the hip.Arthritis Rheum 1991, 34:505-514.18. Bellamy N, Kirwan J, Boers M, Brooks P, Strand V, Tugwell P, et al.:thritis. Consensus development at OMERACT III.  J Rheumatol1997, 24:799-802.19. Bellamy N, Buchanan W, Goldsmith CH, Campbell J, Stitt LW: Vali-dation study of WOMAC: A health status instrument formeasuring clinically important patient relevant outcomes toantirheumatic drug therapy in patients with osteoarthritis ofthe hip or the knee.  J Rheumatol 1988, 15:1833-40.20. Charnley J: The long-term results of low-friction arthroplastyof the hip performed as a primary intervention.  J Bone JointSurg Br 1972, 54(1):61-76.21. Tobin J: Estimation of Relationships for Limited DependentVariables.  Econometrica 1958, 26:24-36.22. Swiontkowski MF: Outcomes measurement in orthopaedictrauma surgery.  Injury 2003, 26:653-657.23. Cole B, Finch E, Gowland C: Physical Rehabilitation OutcomeMeasures.  Toronto, Ontario, Canadian Physiotherapy Association;1994. 24. Ponzer S, Nasell H, Bergman B: Functional outcome and qualityof life in patients with Type B ankle fractures: A two-year fol-low-up study.  J Orthop Trauma 1999, 13:363-368.25. Beaton DE, Bombardier C, Katz JN, Wright JG: A taxonomy forresponsiveness.  Journal of Clinical Epidemiology 2001, 54:1204-1217.26. Rejeski WJ, Shumaker S: Knee osteoarthritis and health-relatedquality of life.  Med Sci Sports Exerc 1994, 26(12):1441-5.27. Liang M, Larson MG, Cullen KE: Comparative measurement effi-ciency and sensitivity of five health status instruments forarthritis research.  Arthritis Rheum 1985, 28:542-547.28. Crosby Ross D, Kolotkin Ronette L, Williams G Rhys: Defining clin-ically meaningful change in health-related quality of life.  Jour-nal of Clinical Epidemiology 2003, 56(5):395-407.29. Ehrich EW, Davies GM, Watson DJ: Minimal perceptible clinicalimprovement with the Western Ontario and McMaster Uni-versities Osteoarthritis Index questionnaire and globalassessments in patients with osteoarthritis.  J Rheumatol 2000,27:2635-41.30. Cohen J: Statistical power analysis for the behavioral sciences.2nd edition. New York: Academic Press; 1988. 31. Austin PC, Escobar M, Kopec JA: The use of the Tobit model foranalyzing measures of health status.  Qual Life Res 2000,9(8):901-10.Pre-publication historyThe pre-publication history for this paper can be accessedhere:http://www.biomedcentral.com/1471-2474/6/48/prepubyours — you keep the copyrightSubmit your manuscript here:http://www.biomedcentral.com/info/publishing_adv.aspBioMedcentralPage 9 of 9(page number not for citation purposes)Recommendations for a core set of outcome measures forfuture phase III clinical trials in knee, hip, and hand osteoar-

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.52383.1-0074641/manifest

Comment

Related Items