To enhance guideline-based nonsurgical management of osteoarthritis (OA), a multidisciplinary stepped-care strategy has been implemented in clinical practice. This study aimed to describe health care use after implementation of this strategy and to identify factors related to such use at multiple levels.
For this 2-year observational prospective cohort, patients with symptomatic hip or knee OA were included by their general practitioner. Activities aligned with patients and health care providers were executed to implement the strategy. Health care use was described as the cumulative percentage of “users” for each modality recommended in the strategy. Determinants were identified at the level of the patient, general practitioner, and practice using backward stepwise logistic multilevel regression models.
Three hundred thirteen patients were included by 70 general practitioners of 38 practices. Their mean ± SD age was 64 ± 10 years and 120 (38%) were men. The most frequently used modalities were education, acetaminophen, lifestyle advice, and exercise therapy, which were used by 242 (82%), 250 (83%), 214 (73%), and 187 (63%) patients, respectively. Fourteen percent of the overweight patients reported being treated by a dietician. Being female, having an active coping style, using the booklet “Care for Osteoarthritis,” and having limitations in functioning were recurrently identified as determinants of health care use.
After implementation of the stepped-care strategy, most recommended nonsurgical modalities seem to be well used. Health care could be further improved by providing dietary therapy in overweight patients and making more efforts to encourage patients with a passive coping style to use nonsurgical modalities.
Despite the availability of several guidelines about the management of hip or knee osteoarthritis (OA) ([1-3]), the quality of OA care remains suboptimal in terms of effectiveness, timeliness and appropriateness, and efficiency (). In an initiative to improve the quality of care, a multidisciplinary patient-centered stepped-care strategy (SCS) has been developed (i.e., Beating osteoARThritis [BART]) (). This strategy presents the optimal sequence for care in 3 steps, is based on recommended modalities in guidelines, and considers advanced treatment modalities only if the options listed in the previous steps failed to produce satisfactory results. At each step, recommendations for diagnostic procedures, nonsurgical treatment modalities (both advised and optional), and length of treatment before evaluation are made (Table 1).
Table 1. Summary of the SCS recommendations in each step ()*
The stepped-care strategy (SCS) recommends all advised step 1 modalities to patients with symptomatic hip or knee osteoarthritis. Only if these modalities lead to unsatisfactory results, modalities of the subsequent steps should be advised. The SCS recommends explicitly to evaluate treatment goals in interaction with the patient after a preset period of time. NSAIDs = nonsteroidal antiinflammatory drugs; TENS = transcutaneous electrical nerve stimulation.
aIf there is a discrepancy between medical history and physical examination.
It is expected that implementation of the SCS in clinical practice will reduce underutilization of nonsurgical treatment modalities and inadequate use of medication or diagnostic procedures. In preparation of a nationwide implementation, assessment of the feasibility of the implementation on a limited scale is needed ([6, 7]). Therefore, first, an observational study was conducted in one region in The Netherlands, aiming to assess the health care use of patients with hip or knee OA after implementation of the SCS in clinical practice. This implementation was focused on all different disciplines involved in OA care, whereas the current study focused particularly on the health care use of patients visiting their general practitioner (GP), since the core treatment is mainly delivered in general practice.
Although health care use of patients with hip or knee OA has been studied before, no study has identified its determinants at different levels, i.e., the patient, physician, and organization levels, after active implementation of the SCS. A comprehensive insight into the factors related to health care use can be used to define the potentially most effective tailored implementation strategy. Previously identified factors related to health care use in hip or knee OA included coping style, health insurance, limitations in functioning, and number of comorbidities ([8-11]). So far, no studies are available describing determinants at the level of the GP or general practice. Therefore, we based the operationalization of these factors on studies describing determinants of physician adherence to guidelines and determinants that may facilitate or impede introduction of innovations in clinical practice ([12, 13]).
Given the lack of insight into barriers and facilitators for optimal health care use in hip or knee OA, the aim of this study was 2-fold: to describe health care use of patients with hip or knee OA after implementation of the SCS and to identify factors related to this health care use at the level of the patient, GP, and general practice.
Box 1. Significance & Innovations
This is the first study to describe health care use and its determinants at the level of the patient, physician, and organization after active implementation of a stepped-care strategy for the management of hip or knee osteoarthritis in clinical practice.
Nonsurgical evidence-based treatment modalities, which are recommended in the stepped-care strategy, seem to be well used after implementation of this strategy.
Variance in health care use is mainly located at the patient level and hardly at the level of the general practitioner or general practice.
Health care use could be further improved by providing dietary therapy in overweight patients and encouraging patients with a passive coping style to use nonsurgical modalities.
MATERIALS AND METHODS
This 2-year observational prospective cohort study was executed from August 2010 to March 2013 in the region Nijmegen, The Netherlands. GPs of the Nijmegen University Network of General Practitioners (NUHP) were invited to implement the SCS and to participate in the study. Patients were recruited by their GP. GPs received 2 questionnaires (at baseline and at 2 months after the inclusion period) and patients received 5 biannual questionnaires by mail. GPs were reminded by e-mail and by phone, if required. Patients were reminded by letter and by phone. The study was approved by the Medical Ethics Committee on Research Involving Human Subjects (CMO) Regio Arnhem-Nijmegen (approval: CMO 2009/246).
The members of the NUHP were informed by their board before they were invited by phone by one of the researchers. This network consists of 157 GPs working in 70 general practices and is associated with the Department of Primary and Community Care of the Radboud University Nijmegen. In addition, 6 practices outside of this network were approached. Practices with GPs expressing willingness to participate were visited by the corresponding author (AJS) to receive additional information.
Individuals were eligible to participate in the study if they visited their GP with a new episode of hip or knee symptoms due to symptomatic hip or knee OA, but only if they had not visited their GP for the same condition during the preceding 3 months, and if they were age ≥18 years. Exclusion criteria were a joint replacement procedure for the hip or knee in the patient's history or on the waiting list for it, and inability to complete the questionnaire because of language barriers or terminal illnesses.
Patients were recruited in 2 different ways. First, eligible, consecutive patients visiting their GP received information about the study from their GP. After the consultation, patients were contacted by one of the researchers by telephone and decided whether or not to participate. Second, patients were recruited after an extraction from GP records based on the coded diagnoses of hip or knee OA, i.e., code L89 or L90 according to the International Classification of Primary Care, in the preceding 6 months. These patients received an invitation letter from their GP and, if interested, were contacted by one of the researchers.
A regional implementation advisory board, consisting of a patient representative and 9 experts representing the main disciplines involved in OA care, i.e., 2 GPs, 1 practice nurse, 1 physical therapist, 1 dietician, 1 rheumatologist, 1 orthopedic surgeon, and 2 researchers, was set up to agree on implementation activities aligned to patients as well as different health care providers. The format and content of these activities were based on previous implementation studies in related research fields describing the effectiveness of different types of interventions ([7, 14-19]). Grol and Grimshaw presented an overview of effective implementation strategies that can target knowledge, attitude, and motivation in different phases of the change process (). The expert panel selected those activities that were also considered suitable to integrate in the existing routines of the Dutch health care system. For example, we used patient education in the form of a booklet, “Care for Osteoarthritis” (in Dutch: “Zorgwijzer Artrose”) (), and educational outreach visits at general practices. The interventions described in Table 2 were executed from May 2010 until July 2012.
Table 2. Executed multifaceted implementation activities to implement the SCS in clinical practice*
SCS = stepped-care strategy; GP = general practitioner; OA = osteoarthritis.
All 313 participating patients received the booklet “Care for Osteoarthritis” () from their GP or from one of the researchers. This booklet was developed for patients with hip or knee OA to provide information about OA and the nonsurgical treatment modalities (according to the SCS) and provides tools to enhance patients' active role and the communication with health care providers (for more detailed information about this booklet, see ref.).
All 313 participating patients were sent information about the booklet “Care for Osteoarthritis” and SCS in the form of newsletters and postcards.
Educational outreach visits
One of the researchers visited the 38 participating general practices to inform the GPs about the SCS, the booklet “Care for Osteoarthritis,” and the study.
The 70 participating GPs received information material, a flyer, and a pocket card about the SCS on several occasions during the study period.
The 70 participating GPs and another >1,000 GPs in the city region Nijmegen were personally invited for a seminar about OA with interactive workshops. Twenty physicians attended the seminar. Six of the physicians were participating GPs.
The 70 participating GPs were reminded about the SCS and the study in the form of equivalent data and postcards.
Rheumatologists and orthopedic surgeons
Educational outreach visits
The heads of 26 departments of orthopedics and rheumatology in the city region Nijmegen received an invitation for an educational outreach visit. Sixteen of the 26 heads of these departments responded to the invitation for an educational outreach visit. Of those, 6 heads of the departments of orthopedics and 4 heads of the departments of rheumatology were interested and therefore were visited by one of the researchers for an educational outreach visit.
The 26 heads of these departments also received similar reminder materials as the participating GPs.
Physical and exercise therapists and dieticians
Regional societies for physical therapists of the Royal Dutch Society for Physical Therapy, exercise therapists of the Dutch Association of Cesar and Mensendieck Exercise Therapists, and dieticians of the Dutch Dietetic Association were approached to inform their members about the SCS, the booklet “Care for Osteoarthritis,” and the study on their web sites and by mail.
∼400 physical and exercise therapists in the city region Nijmegen were also invited to the seminar about OA. Seventy-three physical therapists attended the seminar.
GPs received 2 questionnaires: a short baseline questionnaire to assess demographics and practice characteristics and a second questionnaire that was based on a cross-sectional study of GPs' agreement with the SCS (). In this study, we collected data regarding their organization of OA care and attitudes about OA management and the SCS.
Demographics and practice characteristics concerned age, sex, and how long they had been working in their practice. Moreover, one of the GPs of each practice was asked to answer the following additional questions regarding their general practice: practice type (group/duo/solo) and practice location (rural/suburban/urban).
Organization of OA care was assessed on the involvement of practice nurses in the OA care in their setting (yes/no) and the type of collaboration with other health care providers, i.e., physical and exercise therapists, dieticians, rheumatologists, or orthopedic surgeons. We considered the collaboration “structural” if the GP reported at least 1 of the 2 following types: participation in periodic meetings concerning individual OA patients or following protocols or agreements concerning specific working procedures to treat OA patients.
Attitude regarding OA management included GPs' special interest in musculoskeletal disorders (yes/no), their attendance at the educational outreach visit (yes/no), and their attitude concerning the booklet “Care for Osteoarthritis” with the question: “Would you recommend the booklet” (yes, certainly/yes, probably/no, or certainly).
Attitude regarding the SCS was assessed using 3 indices. One index concerned GPs' attitudes regarding the effectiveness of all advised and optional modalities of the SCS. The second index concerned GPs' attitudes regarding nonrecommended modalities (such as massage, manual therapy, and laser therapy). Both indexes were scored on a 4-point Likert scale (ranging from 0–3, where 3 = effective). Finally, we calculated an index regarding GPs' agreement with 7 recommendations of the SCS (ranging from 0–4, where 4 = totally agreed).
Patients received a questionnaire every 6 months, i.e., at baseline (T0), after 6 months (T6), after 12 months (T12), after 18 months (T18), and after 24 months (T24), that included sociodemographics and health characteristics (at T0), disease-related factors and health care use (at T0, T6, T12, T18, and T24), and psychosocial factors (at T0, T12, and T24).
Sociodemographics and health characteristics included age, sex, weight and length, level of education (low/medium/high), household composition (alone/with partner/with partner and children/with children with others), employment (paid work/no paid work), health insurance (basic/basic with additional coverage), and residence (rural/suburban/urban). The number of important comorbidities (ranging from 0–15) according to the Dutch Arthritis Impact Measurement Scales (), i.e., diabetes mellitus, stroke, myocardial infarction, cancer, severe heart insufficiency, migraine, high blood pressure, peripheral or abdominal arterial disease, asthma or chronic obstructive pulmonary disease, psoriasis, chronic eczema, dizziness with falling, severe intestinal problems for >3 months, incontinence, and chronic inflammation of the joint, was calculated.
Disease-related factors included the location (i.e., hip, knee, and other joints such as the feet, ankles, hands, elbows, shoulders, wrists, back, and neck) and number of painful joints and the duration of hip or knee symptoms (<1 year/1–5 years/5–10 years/>10 years ago). Pain and limitations in activities were assessed with the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) (). Standardized scores ranging from 0–100 were used, where higher scores reflect better health status.
Health care use was assessed by asking the patients at baseline and after each 6-month period which treatment modalities they had used in the preceding 6-month period related to their hip or knee symptoms. The list of modalities was based on the recommended modalities of the SCS. These nonvalidated questions were pilot tested regarding the readability and comprehensiveness in 10 patients with OA by one of the researchers (AJS). The use by the patient of the booklet “Care for Osteoarthritis” was assessed at T6 using the question, “Do you use the care booklet?” (yes, regularly/yes, occasionally/yes, but not in the past 6 months/no, or never).
Psychosocial factors, i.e., self-efficacy and active pain coping, were assessed with validated questionnaires: the Dutch Generalized Self-Efficacy Scale (DGSS) () and the Pain Coping Inventory list (), respectively. Higher scores on the DGSS, ranging from 10–40, reflect higher self-efficacy. A higher score on the subscale active coping, ranging from 12–48, indicates more use of an active coping style.
Differences between groups were analyzed using the chi-square test and the t-test or Mann-Whitney U test, when appropriate. Nine variables with more than 5% missing cases were found. These variables mainly concerned GP characteristics.
Health care use
For each modality, the cumulative percentage of “users” was calculated. We considered the patient as a user if the patients reported having used that particular modality in one of the preceding time periods. If more than 30% of the items (i.e., values on 2 or more time periods) were missing, the scores were treated as missing. Furthermore, we considered referral to a dietician only applicable in overweight patients (body mass index >25 kg/m2).
Determinants of health care use
Previously identified determinants of health care use at the patient level were selected and categorized according to Andersen and Newman's Behavioral Model of Health Care (). This commonly used model divides factors into predisposing, enabling, and disease-related factors. Since there are no previously identified determinants at the GP or practice level, we selected determinants that were identified in related research fields, i.e., physician adherence to guidelines and influencing factors of a successful introduction of innovations in clinical practice. GP-related and practice-related factors were categorized into individual, social, and organizational factors ([12, 13]). Potential determinants are shown in Table 3.
Table 3. Baseline characteristics of the patients, their GPs, and the general practices*
Missing values, no.
Values are the number (percentage) unless indicated otherwise. GP = general practitioner; BMI = body mass index; IQR = interquartile range; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; MSD = musculoskeletal disorder; SCS = stepped-care strategy; OA = osteoarthritis.
Higher scores reflect higher self-efficacy.
aHigher scores indicate more use of an active coping style.
bStandardized scores were used, where higher scores reflect better health status.
Effectiveness recommended modalities (range 0–3), mean ± SD
1.8 ± 0.3
Effectiveness nonrecommended modalities (range 0–3), mean ± SD
1.0 ± 0.5
Agreement with SCS statements (range 0–4), mean ± SD
3.0 ± 0.3
Social factors (at 2 months)
Practice nurse involved in OA management
Structural collaboration with other disciplines
Practice-related factors (n = 38)
We assessed the determinants of the most frequently used recommended treatment modalities. Given the hierarchical structure of the data set, i.e., patients (level 1) are nested in the sample of GPs (level 2) who are nested in general practices (level 3), logistic multilevel regression models were built. Missing data were imputed using switching regression, an iterative multivariable regression technique, to preserve power and obtain less biased results ().
Considering that the number of determinants in the models would likely lead to overfitting (), we selected the most important determinants before fitting the final model. For this selection procedure, we divided the variables in 3 blocks. One block included the predisposing and enabling factors, another block included the disease-related factors, and a third block included the GP-related and practice-related variables. In the case that 2 variables were highly correlated (correlation >0.8), only 1 was entered in the model. Subsequently, the most important variables within each block were selected using backward stepwise regression models based on 5 imputed data sets (in each step, a variable was removed if there was no deterioration in fit using the −2 log-likelihood). The overall final model then consisted of entering the selected variables from each of the 3 blocks simultaneously and was based on 20 imputed data sets combined using Rubin's rules ([29, 30]). If necessary, the final model was reduced to the maximum allowable number of variables also using backward stepwise regression. The relative contribution of each determinant selected in the previous steps, corrected for all other selected variables, was expressed in an odds ratio with a 95% confidential interval. Statistical analyses were executed using STATA/IC 10.1 software (StataCorp).
Of the 76 approached practices, 40 (53%) expressed their willingness to participate. The main reasons not to participate were lack of time (n = 30), not interested (n = 3), no response (n = 2), and retired (n = 1). Seventy GPs in 38 different practices included patients for this study.
The GPs identified 528 patients as eligible. Eighty-three patients (16%) were excluded by one of the researchers because they did not meet the eligibility criteria (Figure 1). Another 132 patients (25%) did not participate, of which 76 patients were not interested, 48 patients gave no reason, and 8 patients reported another reason not to participate. In total, 313 patients (59%) were included in the study, of which 29 patients (9%) were lost to followup: 13 due to a (terminal) illness, 9 were not interested in the study anymore, 4 due to an incorrect diagnosis of OA, and 3 due to other reasons.
The baseline characteristics of the patients (n = 313), their GPs (n = 70), and their general practices (n = 38) are shown in Table 3. Eighty-eight patients (28%) were recruited by their GP during their consultation and 225 patients (72%) were recruited after extraction from GP records. Patients who were recruited during their consultation reported fewer comorbidities than the patients who were recruited from GP records (median 1 [interquartile range (IQR) 1–2] versus median 2 [IQR 1–3]; P < 0.05 by Mann-Whitney U test). Additionally, more patients who were recruited during their consultation reported having symptoms of hip or knee OA for at least 1 year compared with the patients who were recruited from GP records (62 [71%] versus 185 [82%]; P < 0.05 by Mann-Whitney U test).
Health care use
The most frequently used modalities were education, acetaminophen, lifestyle advice, exercise therapy, and nonsteroidal antiinflammatory drugs (NSAIDs) (Figure 2). The cumulative percentage of users of all modalities gradually increased during the study period. After 2 years, step 1 modalities, i.e., education, acetaminophen, lifestyle advice, and glucosamine, were used by 242 (82%), 250 (83%), 214 (73%), and 95 (34%) patients, respectively. Step 2 modalities, i.e., exercise therapy, NSAIDs, consultation with a dietician (if overweight), and tramadol, were reported by 187 (63%), 155 (54%), 27 (14%), and 43 (15%) patients, respectively. Step 3 modalities, i.e., intraarticular injections and multidisciplinary care, were reported by 65 (23%) and 23 (8%) patients, respectively. Consultation in secondary care in the first time period (T0–T6), i.e., with an orthopedic surgeon or rheumatologist, was reported by 67 (21%) and 25 (8%) patients, respectively. After 2 years, these numbers increased to a total of 129 (45%) and 45 (16%) patients, respectively. After 6, 12, 18, and 24 months, the cumulative percentages of surgical procedures were 16 (5%), 28 (10%), 39 (14%), and 49 (18%) patients, respectively.
Determinants of health care use
We assessed the determinants of health care use in those modalities that were used by more than 50% of the patients, i.e., acetaminophen, education, lifestyle advice, exercise therapy, and NSAIDs. The determinants WOMAC pain, age of the GP, and location of the practice were not included because they were highly correlated with WOMAC functioning, working years of the GP, and the patient's residence, respectively.
The variances in use of all assessed treatment modalities located at the level of the GP ranged from <0.01% (for use of NSAIDs) to 7.32% (for use of acetaminophen). The variances at the general practice level were <0.01% for all 5 modalities. Therefore, the variances in health care use were mainly located at the patient level (ranging from 93–100%). Sixteen variables were selected as important determinants for at least 1 of the treatment modalities, of which 11 determinants (69%) were found at the patient level (Table 4). Having an active coping style, using the booklet “Care for Osteoarthritis,” and having limitations in functioning were identified as determinants for more than 1 modality. Four GP characteristics and 1 practice characteristic were identified as determinants of health care use for one of the modalities (either for education, lifestyle advice, or acetaminophen), i.e., not being present at the educational outreach visit, not recommending the booklet, having a positive attitude regarding the effectiveness of recommended modalities, having a positive attitude regarding the effectiveness of nonrecommended modalities, and not working in a solo practice.
Table 4. Logistic multilevel regression analysis of predictors of the use of different treatment modalities in patients with symptomatic hip or knee osteoarthritis*
Education, yes (n = 242)/no (n = 52)
Lifestyle advice, yes (n = 214)/no (n = 81)
Acetaminophen, yes (n = 250)/no (n = 52)
Exercise therapy, yes (n = 187)/no (n = 108)
NSAIDs, yes (n = 155)/no (n = 134)
Sample sizes (n) reflect values at 24 months. NSAIDs = nonsteroidal antiinflammatory drugs; OR = odds ratio; 95% CI = 95% confidence interval; BMI = body mass index; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; GP = general practitioner.
aHigher scores reflect higher self-efficacy.
bHigher scores indicate more use of an active coping style.
cStandardized scores were used, where higher scores reflect better health status.
This study comprehensively described the health care use of patients with hip or knee OA after implementation of the SCS in clinical practice and its determinants at multiple levels, i.e., the level of the patient, GP, and general practice. Our results indicate that the use of health care in the first 2 years after a consultation for a new episode of hip or knee symptoms was in line with that recommended in the SCS; education, lifestyle advice, acetaminophen, and exercise therapy were used by the majority of the patients. However, dietary therapy was found to be relatively underutilized, as only 1 of 6 overweight patients reported being treated by a dietician. A notable high number of patients were referred to secondary care. Variance in health care use was mainly located at the patient level. Also, most determinants of the use of education, lifestyle advice, acetaminophen, exercise therapy, and NSAIDs were found at the patient level. In particular, being female, having an active coping style, using the booklet “Care for Osteoarthritis,” and having limitations in functioning were recurrently identified as determinants.
Although most recommended modalities seem to be frequently used, our results provide starting points to further optimize the use of health care. The determinants that were found in this study provide possible explanations for the nonuse of these modalities: for example, an active coping style and using the booklet “Care for Osteoarthritis” were positively associated with utilization of recommended treatment modalities. This is in line with previous findings that show a positive association between an active coping style and health care use (). Possibly, GPs could make more efforts to encourage patients with a passive coping style to use nonsurgical treatment modalities and use behavioral approaches such as motivational interviewing to encourage patients to self-manage their disease. The booklet “Care for Osteoarthritis” could be a helpful tool.
Another possible target for improvement is to offer patients guidance in reducing weight by a dietician, since only 14% of the overweight patients reported having dietary therapy within 2 years after a consultation for a new episode of hip or knee symptoms. The impact of weight reduction on patients' symptoms has been well demonstrated in a meta-analysis that shows a significant reduction in disability if patients achieve 5% weight loss (effect size 0.34) (). Providing GPs and patients with the knowledge that modest changes in weight improve symptoms in OA could possibly enhance implementation of weight reduction efforts. Previous research findings show that only a minority of overweight patients recall receiving lifestyle advice from their GP (). Since weight reduction is very difficult to achieve and to maintain, we recommend that patients should be encouraged to get structured guidance in achieving weight loss, which is in line with the Osteoarthritis Research Society International recommendation that patients with hip OA should be encouraged to lose weight and maintain their weight at a lower level (level of evidence IV) ().
To our knowledge, this is the first study on determinants of health care use in OA that takes different levels into account. Interestingly, the variance in health care use was almost entirely located at the patient level. GP-related determinants of use of nonsurgical treatment modalities included a positive outcome expectancy on SCS-recommended treatment modalities, not being present at the educational outreach visit, and not recommending the booklet. The outcomes of the latter 2 determinants were not as expected. It is possible that those GPs who were confident in providing the optimal care for OA patients were less likely to have attended the visits and use materials and that, therefore, these factors reflect self-confidence regarding OA management. However, those factors were not consistently found across the treatment modalities, and therefore might have been found by chance.
This study is not without its limitations. First, it is an observational prospective study and not a randomized controlled trial in which we could give statements about the efficacy of implementation activities. However, our design allowed us to identify useful and practical targets to improve health care use in OA patients. Second, the uptake of several implementation activities seems to be poor, since 9% of the participating GPs and 18% of the invited physical/exercise therapists were present at the seminar and 38% of the heads of the rheumatology and orthopedic departments were visited for an educational outreach visit. When implementing the SCS nationwide, these GP-oriented activities need to be reconsidered. However, the simultaneous use of different implementation activities might have contributed to a greater reach of the target population. Third, GPs in our study sample may have been more interested in and dedicated to OA and/or guideline-consistent clinical practice (after the implementation of the SCS) than the average GP in The Netherlands. This might affect the generalizability of our study findings. Fourth, we do not have data concerning health care use before the implementation. As a result, our study design does not allow inferences about changes in provision of health care after implementation of the SCS. Further research is necessary to study the impact of implementation of the SCS on health care use and its outcomes. Furthermore, characteristics of health care providers other than GPs were not taken into account. Finally, the power of this study was not sufficient to assess all determinants. However, we found a good alternative to preselect the potential determinants first within content matter–motivated blocks using backward regression models.
In conclusion, after implementation of the SCS for the management of hip or knee OA, most recommended evidence-based nonsurgical treatment modalities seem to be well used. Health care use could be further optimized in patients with a relatively passive coping style and by providing dietary therapy in overweight patients.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Ms Smink 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 conception and design. Smink, Dekker, Vliet Vlieland, Swierstra, Kortland, Bijlsma, Voorn, Bierma-Zeinstra, van den Ende.
Acquisition of data. Smink.
Analysis and interpretation of data. Smink, Dekker, Vliet Vlieland, Kortland, Teerenstra, Bierma-Zeinstra, Schers, van den Ende.