Five-year followup of a cognitivebehavioral intervention for patients with recently-diagnosed rheumatoid arthritisEffects on health care utilization.код для вставкиСкачать
Arthritis & Rheumatism (Arthritis Care & Research) Vol. 59, No. 3, March 15, 2008, pp 311–316 DOI 10.1002/art.23312 © 2008, American College of Rheumatology ORIGINAL ARTICLE Five-Year Followup of a Cognitive–Behavioral Intervention for Patients With Recently-Diagnosed Rheumatoid Arthritis: Effects on Health Care Utilization L. SHARPE,1 S. ALLARD,2 AND T. SENSKY3 Objective. To investigate whether cognitive– behavioral therapy (CBT) administered early in the course of rheumatoid arthritis (RA) has long-term effects on health care use. Methods. We reviewed the ﬁles of 47 of the original 53 patients with early RA who volunteered for a randomized controlled trial comparing CBT with no psychological intervention. Occasions of service provision associated with RA were documented and health care use was compared between groups. Results. The CBT group used fewer health care resources than the control group in the 5 years following intervention. Signiﬁcant differences were observed for the number of inpatient nights, physiotherapy referrals, injections, and for total health care use. There was a trend that closely approached signiﬁcance toward fewer episodes of surgery and orthopedic referrals in the CBT group. Conclusion. These results suggest that CBT administered early in the course of RA can reduce health care use for the ﬁrst 5 years after treatment. This is a stringent test of the efﬁcacy of a brief psychological intervention, and supports the fact that brief psychological treatments can have long-term effects. INTRODUCTION There is evidence from both meta-analysis and substantive literature review that cognitive– behavioral therapy (CBT) is an effective, adjuvant treatment for patients with rheumatoid arthritis (RA) (1,2). Despite this evidence, services for patients are rarely made available as part of routine care. Barriers to the provision of such services include the cost and lack of long-term followup studies. In the metaanalysis by Astin et al (1), followup was available for 76% of trials, but the average length of followup was ⬍6 months. Because RA is a chronic illness with a progressive ClinicalTrials.gov identiﬁer: NTC001214. Supported by a North Thames Regional Health Authority’s Research and Development grant. Dr. Sharpe’s work was supported by a National Health and Medical Research Council Senior Research fellowship. 1 L. Sharpe, PhD: The University of Sydney, Sydney, Australia; 2S. Allard, MD: West Middlesex University Hospital, Middlesex, UK; 3T. Sensky, MD: West Middlesex University Hospital, Middlesex, UK, and Imperial College, University of London, London, UK. Address correspondence to L. Sharpe, PhD, Clinical Psychology Unit A19, The University of Sydney, New South Wales, 2006, Australia. E-mail: firstname.lastname@example.org. Submitted for publication March 14, 2007; accepted in revised form September 4, 2007. and deteriorating course, it is the long-term efﬁcacy of CBT that is most important to service development. Because CBT is relatively expensive, cost is also an issue in service development. Most CBT programs involve 8 1-hour sessions administered by a clinical psychologist or allied health professional with appropriate training and experience. This raises the question of whether the costs of providing the service are outweighed by the reduction in burden of illness to patients, or are offset by a reduction in health care use due to better long-term physical function. Although studies have found beneﬁts for outcomes such as mood, joint function, pain, and disability (1), only 1 study has reported on the effect of CBT treatments on the health care use of patients with RA (3). In that study, Young et al conﬁrmed that for patients with chronic RA, CBT resulted in reductions in the number of RA-related health care visits and hospitalizations. However, this was based on a sample of patients with chronic illness who were treated in a group context, and health care use was assessed only over an 18-month period. Recent evidence suggests that early prognosis is particularly important to the long-term progression of RA. For example, radiographic damage has been found to be the most aggressive during the ﬁrst 2 years of illness (4). One study found that the majority of patients who had become unable to work due to RA after 8 years of illness had 311 312 stopped work within the ﬁrst year (5). Early functional ability has been found not only to predict outcome after 10 years (6), but even future mortality at 15 years (7). Therefore, although the results presented by Young et al (3) are encouraging, the greatest potential for the long-term beneﬁt of CBT in the management of RA is in the ﬁrst 2 years of illness. To our knowledge, there is 1 prior study examining the efﬁcacy of CBT in the treatment of RA during this crucial early period. Sharpe and colleagues (8) found that CBT was effective in improving mood and joint function at 6-month followup. Their 18-month followup indicated that over time, further improvements of physical and psychological morbidity became evident only in the CBT group (9). The aim of the present study was to follow up with these patients after 5 years and compare health care utilization among those who received CBT with those who did not. The earlier results showed that there were beneﬁts to patients in the CBT group in terms of physical outcomes (joint function) at 6 months (8), and in disability as measured by the Health Assessment Questionnaire (10), which has been found to predict long-term outcomes (6,7), at 18 months (9). We hypothesized that there would be signiﬁcantly less health care usage among those who had received CBT 5 years previously than those who had not. PARTICIPANTS AND METHODS Participants. The original cohort was recruited at 3 hospitals in or near London, UK, between July 1994 and July 1996. The cohorts consisted of consecutive patients attending the rheumatology departments who had been diagnosed with RA in the past 2 years and who were seropositive for rheumatoid factor. Of the patients approached, 89% agreed to take part in the study. The mean ⫾ SD age of the sample at recruitment was 55.06 ⫾ 14.07 years, and the mean ⫾ SD illness duration was 12.63 ⫾ 8.22 months. The majority of the sample were of Anglo-Saxon descent, with only 3 Asian participants. Nearly two-thirds of the sample was married or cohabiting. Despite the recent onset of RA, only 22% were working full-time, with a further 12% working part-time. The remainder of the sample was either retired (10%) or not working outside the home (55%). Only 14% were university-educated, 24% had completed school (i.e., 12 years), and the remainder had ⬍10 years of education (63%). Ninety-six percent of the sample was taking medication, with 10% taking diseasemodifying medication only, 6% taking nonsteroidal antiinﬂammatories only, 13% taking steroids only, and the remainder taking combination medications. The groups were relatively well-matched on most variables at baseline (8). The only difference between groups at pretreatment was in C-reactive protein (CRP) level, where the control groups had lower levels of inﬂammation. Groups did not differ on any other variables, including age, sex, medications taken, duration of illness, disability, joint function, erythrocyte sedimentation rate, or psychosocial functioning (8). Participants were randomly allocated to receive either standard clinical care alone (n ⫽ 27) or standard clinical Sharpe et al care plus CBT (n ⫽ 26). Four patients dropped out of both the treatment and control groups prior to the end of treatment, leaving 22 and 23 completers, respectively. Files were available for 24 participants in the CBT group and 23 participants in the control group (89% of the original sample). At the 5-year followup, 9 participants had been discharged from rheumatology services; therefore, data were available for those patients only until their point of discharge. For 5 patients (4 in the CBT group and 1 in the control group), this was due to improvements in clinical presentation. For all of the patients discharged due to improvements, the discharge occurred after the third year of treatment. Because all patients were recruited within the National Health Service, which refers patients to receive services at the local hospital in the area in which they reside, we are conﬁdent that had these patients been re-referred to a specialist for treatment of their RA, it would have been to the same services and we would have had access to the data. Therefore, we assumed that patients who were discharged due to clinical improvement did not receive any further rheumatology-related services. For the 4 participants who were not discharged due to improvement, the patients had moved away in each case. These participants were split across the groups (2 in CBT and 2 in control), and all moved after the third year of treatment; therefore, it is very unlikely that these low levels of attrition would affect our results. Further, because many of the more expensive medical procedures (e.g., surgery) were rare occurrences (i.e., if they occurred, they occurred only once in the 5-year period), there was not a reliable way in which to estimate health care usage for the participants who moved. Therefore, we used their data until the point at which they moved. Treatment. Detailed descriptions of the CBT program used in the current study are available elsewhere (8,9,11). The program was a broad-based approach that included psychoeducation, behavioral strategies, and cognitive strategies. Behavioral strategies aimed to help patients develop a balance between rest and exercise through the use of goal setting and pacing. We also included relaxation and imagery strategies. The cognitive strategies targeted the illness perceptions of the patients and aimed to facilitate a realistic yet optimistic approach to their illness. Strategies to help people communicate more effectively about their illness were also included. The session-by-session plan is presented in Table 1. Procedure. The ﬁrst author (LS) reviewed the ﬁles of patients for the 5 consecutive years following their recruitment into the trial. The ﬁle review took place in June and July of 2001. All rheumatology-related occasions of care or referrals to other services that took place in the 5 calendar years after recruitment were recorded (see Measures for details). Occasions of care that were associated with other medical conditions were excluded. The trial was approved by the ethics committees of the relevant health authorities. Cognitive–Behavioral Intervention for RA Patients Table 1. Session-by-session content of cognitive– behavioral treatment Session # Content 1 2 3 4 Education, treatment rationale Goal setting, relaxation training Relaxation, problem-solving skills Brief relaxation, attention diversion techniques, imaginal pain transformation Cognitive therapy: automatic thoughts, time management Challenging cognitions, identifying problematic attitudes, assertion skills Challenging pain beliefs, coping self-talk Summary, relapse prevention 5 6 7 8 Measures. The 3 main sources of medical care provision in rheumatology are medication, physician (or other health professional) visits, and hospitalizations and surgery (12). Unfortunately, the ﬁles were not sufﬁciently detailed to determine the precise medications that patients were taking at 1 of the sites. However, data in our previous work demonstrated that patients were well-matched on medication at baseline (8) and that no differences in medications prescribed were noted over the ﬁrst 18 months of the trial (9). The variables recorded were rheumatology consultations; inpatient nights; surgery; physiotherapy, orthopedic, and psychiatric referrals; pain management referrals and injections; magnetic resonance imaging (MRI) scans; discharges; and total occasions of care. The number of rheumatology consultations was recorded for each of the 5 years from the point at which each participant was recruited; therefore, the number of consultations is available for each consecutive year. The number of nights that each participant had been admitted due to a rheumatology-related problem was calculated and recorded as inpatient nights. Two types of surgery had been performed on participants: hip replacement operations (n ⫽ 4) and knee reconstructions (n ⫽ 2). Whether or not the participant had surgery was recorded. The number of physiotherapy, orthopedic, and psychiatric referrals made was calculated. We included psychiatry referrals because a large proportion of patients in the trial had clinically signiﬁcant levels of depression and anxiety, which improved with CBT but worsened over time without psychological intervention (8,9). A number of patients were referred to the pain management physician for cortisone injections, and those referrals were recorded for each patient. All patients in the sample had routine radiographs taken to assess radiographic progression. We recorded the number of those who were referred for more extensive investigations (i.e., MRIs). A number of patients were discharged from their rheumatology services. For some, this was due to practical issues (e.g., moving away). However, several patients were discharged due to improvement. The number of patients discharged as improved was recorded. We were interested in the total number of occasions of care that were saved by adding psychological intervention to routine care. Therefore, we derived a number for the total occasions of care from the data collected. The total number of occasions of care was determined by develop- 313 ing a weighting system based on the relative costs of each type of service. Relative costs were assessed using the Australian Medicare Beneﬁts Schedules (13), which list the costs for different medical procedures and have been used previously in health economics research in RA (14). Each variable was multiplied by the relative cost of that procedure, and the weighted scores were added together to give a weighted total of occasions of care. Statistical analyses. With the exception of the number of rheumatology consultations, data were not normally distributed and the majority of analyses were conducted using nonparametric statistics. However, because there was a signiﬁcant difference between the groups at baseline in CRP, we performed a series of correlations (Pearson’s product-moment for the rheumatology consultations and Spearman’s rank correlations for the other variables) to determine whether the differences in CRP were associated with long-term health care use. A 2 (treatment: CBT versus no psychological intervention) ⫻ 5 (years) mixed-model analysis of variance was conducted for the number of rheumatology consultations. Mann-Whitney U tests for independent samples were conducted to compare the number of referrals for physiotherapy, orthopedic surgeons, and psychiatric services; number of injections; number of inpatient nights; number of MRIs; and total occasions of care for the CBT and control groups. For dichotomous variables, chi-square analyses were performed for whether the patient had surgery or not and whether the patient was discharged as improved or not. Finally, in order to ensure that the baseline differences in CRP could not account for the differences in overall health care use, which we deﬁne as the primary outcome measure, we also performed an analysis of covariance (ANCOVA) with total occasions of care, controlling for CRP. RESULTS Correlations with CRP level. CRP level was only significantly correlated with 1 of the 15 variables, the number of physiotherapy referrals (r ⫽ 0.296, P ⫽ 0.048). That is, those with higher levels of baseline CRP were more likely to have a referral to physiotherapy services at baseline than those with lower levels of baseline CRP. However, there were also 2 trends in the analysis, such that those who had higher levels of CRP at baseline also tended to be referred for MRIs more often (r ⫽ 0.273, P ⫽ 0.07) and to have more rheumatology appointments in the ﬁrst year (r ⫽ 0.266, P ⫽ 0.08). None of the other correlations approached signiﬁcance (r ⬍ 0.22, P ⬎ 0.16). As a result, we decided to include CRP as a covariate in all parametric analyses. Health care utilization. As seen in Figure 1, there was a main effect of year on the number of rheumatology consultations, with both groups having fewer consultations as the years progressed (F[1,42] ⫽ 7.838, P ⫽ 0.008). However, there was no effect of treatment group (F[1,42] ⫽ 0.187, P ⫽ 0.668) and no interaction effect (F[1,42] ⫽ 0.104, P ⫽ 0.981). 314 Sharpe et al health care utilization among those participants who had received CBT 5 years previously. Speciﬁcally, signiﬁcant reductions in health care use were seen in the number of inpatient nights, physiotherapy referrals, and injections. Moreover, trends were identiﬁed that suggested that the CBT group tended to be referred to orthopedic surgeons or to have surgery less often. Even when controlling for acute inﬂammation at baseline, there were signiﬁcant differences for total health care use, which reﬂected an average Figure 1. Number of rheumatology consultations by year. CBT ⫽ cognitive– behavioral therapy. Signiﬁcant differences emerged using Mann-Whitney U tests between the groups for the number of inpatient nights (t ⫽ ⫺1.768, P ⫽ 0.039), number of physiotherapy referrals (t ⫽ ⫺1.895, P ⫽ 0.029), number of injections (t ⫽ ⫺2.507, P ⫽ 0.007), and total occasions of care (t ⫽ ⫺2.144, P ⫽ 0.032). In each instance, the CBT group used fewer health care resources (see Figure 2). No differences were evident in the number of psychiatric referrals (t ⫽ ⫺1.177, P ⫽ 0.115) or the number of patients discharged as improved (2 ⫽ 1.607, P ⫽ 0.217). However, a trend was evident for fewer orthopedic referrals (t ⫽ ⫺1.567, P ⫽ 0.059) and surgeries (2 ⫽ 3.740, P ⫽ 0.067) in the CBT group. In order to ensure that the effect of CBT on total weighted health care use could not be attributed to baseline differences in CRP, we excluded physiotherapy referrals from the total health care use variable because 1) it was correlated signiﬁcantly with CRP, and 2) there were signiﬁcant differences between the groups, favoring CBT. Although the correlations with CRP and the number of rheumatology consultations and MRI scans approached signiﬁcance, because neither reached signiﬁcance and neither were different between the groups we decided to retain them in the weighted total health care use variable. We then analyzed the data parametrically, controlling for CRP at baseline. We performed a log-linear transformation of the weighted average score, which resulted in normal distribution of the data. We then conducted the ANCOVA, which conﬁrmed that there were signiﬁcant differences in total weighted health care use between the groups (F[1,42] ⫽ 5.745, P ⫽ 0.02). These results showed that the patients receiving CBT used fewer health care resources overall. The mean ⫾ SD per person cost of the health care resources monitored in this study was $1,454.58 ⫾ $1,357.58 for the CBT group compared with $3,156.00 ⫾ $3,038.11 for the control group. This represents a savings of $1,701.42 per patient. In this small sample, this represents a total savings of $40,834.08. DISCUSSION Given that we had found both psychological and physical beneﬁts of CBT at posttreatment, 6-month followup, and 18-month followup (8,9), we hypothesized that CBT administered early in the course of RA would result in longterm beneﬁts in terms of reduced health care usage. These hypotheses were largely supported. Overall, there was less Figure 2. Number of occasions of care for the cognitive– behavioral therapy group (striped bars) and the control group (solid bars) over 5 years, recorded by category. A, Drs appt ⫽ appointments with the treating rheumatologist; In-pt ⫽ number of nights inpatient admissions; MRI ⫽ number of magnetic resonance imaging scans; Orthopod ⫽ number of referrals to orthopedic surgeons, and B, Physio ⫽ percent of patients referred to physiotherapy services; Psychol ⫽ percent of referrals to psychological or psychiatric services; Disch-Imp ⫽ percent patients who were discharged as improved. For both A and B, ⴱⴱ ⫽ P ⬍ 0.01; ⴱ ⫽ P ⬍ 0.05; # ⫽ P ⬍ 0.07. Cognitive–Behavioral Intervention for RA Patients savings on the indices measured of approximately $1,700 per patient. These results strongly support the use of CBT for patients who are diagnosed with RA. Although these ﬁndings are very encouraging, there are several limitations that need to be considered in interpreting them. First, the sample size was small and therefore a number of ﬁndings, such as those requiring surgery or referral to orthopedic surgeons, approached but failed to reach signiﬁcance. The participants were still relatively early in the trajectory of their illness and only small proportions needed such interventions; this study is likely to be underpowered in detecting these differences. As a result, it is important to interpret the data cautiously. Further, we did not adopt a stringent health economics model, and we assessed only selected health care outcomes. Nonetheless, the fact that signiﬁcant differences were observed in a number of variables indicates that large effect sizes were evident between the groups, supporting the clinical signiﬁcance of the present ﬁndings. We did not include general health care utilization measures because in an aging population such as this, we expected that there would be other illnesses and medical issues that one would not expect to be affected by psychological therapy speciﬁcally targeting RA. Second, the ﬁle reviews for the present study were conducted by the ﬁrst author (LS). Although she was not aware of the group to which all participants were allocated, because she provided treatment to some of the participants (n ⫽ 11) 5 years earlier, she was not entirely blind to condition. Nonetheless, some of the reported indices are clearly objective, such as the number of operations, inpatient nights, and referrals to orthopedic surgeons, physiotherapists, and injections. Differences were found on a number of these variables and it is difﬁcult to imagine that these could be inﬂuenced by judgment bias. Third, we were unable to collect accurate information on medication from the ﬁles in this study. Clearly, this is a major limitation. However, we assessed for differences in medication usage in our earlier publications (8,9) and none were evident up until the 18-month point. It is therefore unlikely that differences in medication can account for these ﬁndings. Finally, although the groups were relatively wellmatched on most of the outcome indices, there was a difference at baseline in CRP levels. In our previous work, however, we were able to show that this did not affect outcome at posttreatment, 6-month followup (8), or 18month followup (9). Nonetheless, in order to ensure that the effects could not be accounted for by CRP level differences, we excluded physiotherapy referrals from the total health care use variable and included it as a covariate in all parametric analyses. The results remained signiﬁcant, making it very unlikely that they are attributable to this difference in a group that was otherwise very wellmatched at pretreatment. Despite these limitations, our ﬁndings are important and mark the ﬁrst long-term demonstration of the reduction in health care use among RA patients treated with CBT in the crucial ﬁrst 2 years of illness. However, in terms of service planning, these reductions in health care use need to be weighed against the cost of service provision. The treat- 315 ment in this study was administered by 2 psychologists: an experienced clinical psychologist and a psychology graduate. The cost of this service per patient, using the same medical beneﬁts costs (12) as we used to weight health care costs, is $1,000 ($125 per session ⫻ 8 sessions per patient). Therefore, these data suggest that, on the limited range of variables that we were able to assess, the cost of the service is offset by the savings in health care use within the ﬁrst 5 years following diagnosis. There are a number of strengths of the present study that should be highlighted. The randomized controlled trial was methodologically rigorous. We recruited consecutive patients and the study beneﬁted from high recruitment rates and low attrition rates. Treatment was manualized and could be replicated in clinical practice relatively easily. To our knowledge, our followup over 5 years is the longest followup of a psychological intervention for RA. This study was also the ﬁrst to investigate the effect of CBT administered in the crucial ﬁrst 2 years after diagnosis of RA on health care usage. Investigating the effect of CBT on long-term health care use is a particularly stringent test of a psychological invention administered over a brief, 2-month period without any followup psychological care. The results support the long-term efﬁcacy of CBT administered early in the course of RA. If these results are extrapolated from the small sample reported here, the savings to the health service industry are likely to be substantial. While it is possible that not all of the differences we found are directly attributable to CBT, there are a number of reasons to believe that CBT contributed substantially to our results. First, the earlier trial results found treatmentrelated differences in physical parameters, including CRP level, joint function (8), and disability (9) that one might expect to affect health care use. Second, the treatment was administered within the ﬁrst 2 years of illness, which is known to be important to the long-term prognosis (4 – 7,15). The groups were relatively well-matched at baseline (8). Further, other authors have reported changes in physical parameters of illness following CBT (16 –18). Although some other studies have failed to ﬁnd physical effects of CBT, these have typically involved patients with severe and chronic RA (19,20). Therefore, our ﬁnding of long-term health beneﬁts of CBT use could be predicted on the basis of the literature, including the only other study to investigate health care use (3). The net savings in terms of health care use are large for a small sample with earlystage disease. The net savings included 1 knee reconstruction and 3 hip replacement operations, 20 inpatient nights, 3 courses of physiotherapy, 3 referrals to orthopedic surgeons, 3 psychiatric referrals, 5 cortisone injections, and 1.45 rheumatology consultations per patient (37.7 total for this sample). When the costs were used to weight the data, the net savings from each patient to the health service industry was $1,700 (over $40,000 total). The cost of 8 sessions of CBT ($1,000) is offset by these savings. These results provide further support for recent arguments that providing evidence-based treatments for RA is likely to be cost-effective (14). In summary, the present study indicates that CBT not only produces improvements in psychological and physi- 316 Sharpe et al cal morbidity (8,9), but also produces long-term reductions in health care usage over 5 years. This provides further strong support for the efﬁcacy of CBT for patients with RA, particularly when administered early in the course of the illness. Importantly, in addition to reducing the emotional burden to patients, these results suggest that CBT is likely to be cost-effective. Over as short a period as 5 years, CBT was found in this small sample to reduce the economic burden of RA to patients and health care systems alike. These results suggest that CBT within the ﬁrst or second year of the illness has long-term beneﬁts, and that health service systems should be encouraged to provide such services to patients. ACKNOWLEDGMENTS 6. 7. 8. 9. We would like to thank Dr. Natalie Timberlake and Mrs. Bridget Ryan for their work on the project. We would also like to thank all the participants who took part in the study. 10. AUTHOR CONTRIBUTIONS 12. Dr. Sharpe had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study design. Sharpe, Allard, Sensky. Acquisition of data. Sharpe. Analysis and interpretation of data. Sharpe, Sensky. Manuscript preparation. Sharpe, Sensky. Statistical analysis. Sharpe. 11. 13. 14. 15. REFERENCES 1. Astin JA, Beckner W, Soeken K, Hochberg MC, Berman B. Psychological interventions for rheumatoid arthritis: a metaanalysis of randomized controlled trials. Arthritis Rheum 2002;47:291–302. 2. Keefe FJ, Smith SJ, Bufﬁngton AL, Gibson J, Studts JL, Caldwell DS. Recent advances and future directions in the biopsychosocial assessment and treatment of arthritis. J Consult Clin Psychol 2002;70:640 –55. 3. Young LD, Bradley LA, Turner RA. Decreases in health care resource utilization in patients with rheumatoid arthritis following a cognitive behavioral intervention. Biofeedback Self Regul 1995;20:259 – 68. 4. Fex E, Jonsson K, Johnson U, Eberhardt K. Development of radiographic damage during the ﬁrst 5– 6 years of rheumatoid arthritis: a prospective followup of a Swedish Cohort. Br J Rheumatol 1996;35:1106 –15. 5. Fex E, Larsson BM, Nived K, Eberhardt K. Effect of rheumatoid arthritis on work status and social and leisure time ac- 16. 17. 18. 19. 20. tivities in patients followed 8 years from onset. J Rheumatol 1998;25:44 –50. Sherrer YS, Bloch DA, Mitchell DM, Young DY, Fries JF. The development of disability in rheumatoid arthritis. Arthritis Rheum 1986;29:494 –500. Corbett M, Dalton S, Young A, Silman A, Shipley M. Factors predicting death, survival and functional outcome in a prospective study of early rheumatoid arthritis over ﬁfteen years. Br J Rheumatol 1993;32:717–23. Sharpe L, Sensky T, Timberlake N, Ryan B, Brewin CR, Allard S. A blind, randomized, controlled trial of cognitive-behavioral intervention for patients with recent onset rheumatoid arthritis: preventing psychological and physical morbidity. Pain 2001;89:275– 83. Sharpe L, Sensky T, Ryan B, Timberlake N, Allard S. Longterm efﬁcacy of a cognitive behavioural treatment from a randomized controlled trial for patients recently diagnosed with rheumatoid arthritis. Rheumatology (Oxford) 2003;42: 435– 41. Fries JF, Spitz P, Young DY. The dimensions of health outcomes: the Health Assessment Questionnaire, disability and pain scales. J Rheumatol 1982;9:789 –93. Sharpe L, Sensky T, Timberlake N, Allard S, Brewin CR. The role of cognitive behavioural therapy in facilitating adaptation to illness in rheumatoid arthritis: a case series. Behav Cogn Psychother 2001;29:303–9. Lubeck DP. A review of the direct costs of rheumatoid arthritis: managed care versus fee-for-service settings. Pharmacoeconomics 2001;19:811– 8. Australian Government: Department of Health and Ageing. Medical beneﬁts schedule. Canberra, Australia: Department of Health and Ageing; 2006. Andrews G, Simonella L, Lapsley H, Sanderson K, March L. Evidence-based medicine is affordable: the cost-effectiveness of current compared with optimal treatment in rheumatoid and osteoarthritis. J Rheumatol 2006;33:671– 80. Emery P. Therapeutic approaches for early rheumatoid arthritis: how early? How aggressive? Br J Rheumatol 1995;34 Suppl 2:87–90. Bradley LA, Young LD, Anderson KO, Turner RA, Agudelo CA, McDaniel LK, et al. Effects of psychological therapy on pain behavior of rheumatoid arthritis patients: treatment outcome and six-month followup. Arthritis Rheum 1987;30: 1105–14. O’Leary A, Shoor S, Lorig K, Holman HR. A cognitive-behavioral treatment for rheumatoid arthritis. Health Psychol 1988; 7:527– 44. Parker JC, Smarr KL, Buckelew SP, Stucky-Ropp RC, Hewett JE, Johnson JC, et al. Effects of stress management on clinical outcomes in rheumatoid arthritis. Arthritis Rheum 1995;38: 1807–18. Parker JC, Frank RG, Beck NC, Smarr KL, Buescher KL, Phillips LR, et al. Pain management in rheumatoid arthritis patients: a cognitive-behavioral approach. Arthritis Rheum 1988;31:593– 601. Kraaimaat FW, Brons MR, Geenen R, Bijlsma JW. The effect of cognitive behavior therapy in patients with rheumatoid arthritis. Behav Res Ther 1995;33:487–95.