Family practice nurses supporting self-management in older patients with mild osteoarthritis: a randomized trial

Background: Supporting self-management intends to improve life-style, which is beneficial for patients with mild osteoarthritis (OA). We evaluated a nurse-based intervention on older OA patients' self-management with the aim to assess its effects on mobility and functioning. Methods: Randomized controlled trial of patients (> 65 years) with mild hip o r knee OA from nine family practices in the Netherlands. Intervention consisted of supporting patients' self-management of OA symptoms using a practice-based nurse. Outcome measures were patients' mobility, using the Timed Up and Go test (TUG), and patient reported functioning, using an arthritis specific scale (Dutch AIMS2 SF). Results: Fifty-one patients were randomized to the intervention group and 53 to the control group. Patient-reported functioning improved on four scales in the intervention group compared to one scale in the control group. However, this result was not significant. Mobility improved in both groups, w ithout a significant difference between the two groups. There were no differences between the groups regarding consultations with family physicians or physiotherapists, or medication use. Conclusion: A nurse-based intervention on older OA patients' self-management did not improve self-reported functioning, mobility o r patients' use of health care resources. Background In our aging population osteoarthritis (OA) is a highly prevalent chronic disease, which has a high impact on burden of disease, quality of life, and use of healthcare. Worldwide estimates are that 10% of m en and 18% of women aged 60 years have symptomatic OA [1]. In early stages clinical management of OA is targeted at improving patients' self-management [2-5], losing weight [6], physi­ cal exercise [7-11] and adequate use of analgesics. But, medicalization of OA should be avoided. Patients' self­ management may improve their life-style and therefore health outcomes, analogue to diabetic patients [12]. Healthcare systems face the challenge to enhance self­ management in OA patients on a sufficiently large scale so that all patients are actually reached and helped. Barriers may be that improving patients' life-style often requires substantial investment of both patients' and health pro­ fessionals' time, as many education programmes require a large number of sessions [13]. And, the health behaviors in older patients tend to be reserved, as they attribute


Background
In our aging population osteoarthritis (OA) is a highly prevalent chronic disease, which has a high im pact on burden of disease, quality of life, and use of healthcare. Worldwide estimates are that 10% of m en and 18% of wom en aged 60 years have symptomatic OA [1]. In early stages clinical m anagem ent of OA is targeted at improving patients' self-management [2][3][4][5], losing weight [6], physi cal exercise [7][8][9][10][11] and adequate use of analgesics. But, m edicalization o f OA should be avoided. Patients' self m anagem ent may improve their life-style and therefore health outcomes, analogue to diabetic patients [12]. Healthcare systems face the challenge to enhance self m anagem ent in OA patients on a sufficiently large scale so that all patients are actually reached and helped. Barriers may be that improving patients' life-style often requires substantial investm ent of both patients' and health pro fessionals' time, as m any education programmes require a large num ber of sessions [13]. And, the health behaviors in older patients tend to be reserved, as they attribute many complaints towards getting older, and consequently arthritis symptoms are underreported [14]. Involving practice-based nurses in the m anagem ent of OA ensures that this care is delivered closely to the patient. A recent review showed that substituting physicians for appropri ately trained nurses could produce as highly quality care as primary care doctors and achieves as good outcomes for patients [15]. The availability of skilled nurses is limited and nursing time invested in any intervention needs to be examined critically. Therefore we wondered whether a sin gle individual session with a trained nurse, which was focused on supporting patients' self-management, would be effective in OA patients. O n the basis o f previous research on changing life style behavior by family physi cians (FPs), we expected a small b u t relevant change in patient behavior [16]. The aim of this proof o f principle study was to evaluate the clinical effectiveness o f a single session nurse-based intervention for enhancing self-man agement in older patients with m ild OA.

Study P o pu lation
This study was a patient random ized controlled trial, which was performed between April 2004 and January 2005. This trial has n o t been registered beforehand in a publicly accessible trial registry as it was performed before prospective registering o f these trials became obliged. The ethical committee of the Radboud University Medical Centre Nijmegen gave approval for the study.
The trial was based on the practice populations of seven teen FPs from nine urban non-academic practices in the Eastern region of the Netherlands. Patients were eligible if they were aged 65 or older and had been clinically diag nosed with OA of the hip or knee. The OA diagnosis needed to be registered in patient's practice medical his tory record as free text or as ICPC-code L89 (OA of the knee) or L90 (OA o f the hip). Patients were excluded if they had undergone a hip or knee replacement operation, or had been referred for it or when their GP thought they were n o t suitable for participating (for example because of severe psychosocial circumstances, or a terminal disease). No further classification of degree of OA was made. An inform ed consent letter was sent by the GP and patients were included after they had replied positively.

R and o m ization
An independent statistician m ade random ization lists in advance for each practice. To ensure similar num bers of patients from different practices in each group, block-random ization (blocks of two) was used. These random iza tion lists were represented in nine different spreadsheets. Every patient who entered the study was given a num ber that represented the order of entrance in the study for that practice. Subsequently, the num ber of entrance per prac tice in the spreadsheet was used to random ly assign the patient to intervention or control group. This was proce dure was performed by a research assistant who was blinded for patients' characteristics.

Interven tio n
The intervention consisted of education and self-manage m ent o f OA symptoms. It was performed by a nurse and aimed to change life style behavior, by improving m obil ity and physical functioning. O n a time-scale the interven tion consisted of three parts. Firstly, patients had to prepare for the hom e visit of the nurse, using an educa tional leaflet about osteoarthritis (developed by the Dutch College of General Practitioners) and a booklet with health-status charts. The health-status charts were based on the Wonca COOP-charts [17]. The patients needed to fill out their level of exercise, pain-level and their im pair m ents prior to the nurse hom e visit. The charts were dis cussed during a 30-m inute nurse hom e visit; this is the second part of the intervention. In this hom e visit patients got insight in their own OA symptoms. Subsequently, they agreed to try to change one of four life style items (physical exercise, weight loss, use of a walking aid and how to use over the counter (pain) medication). The third part of the intervention was a follow-up phone call after approximately 3 m onths. In this phone call the nurse eval uated to w hat extent the patient had been able to adapt his life style change and subsequently w hat possibly was nec essary to m aintain this change.
The nurse had undergone a certified education in rheum a tology. Patients in the control group received only the educational leaflet about osteoarthritis.

O u tc o m e measures
Primary outcome measurements were 4 subscales of the Dutch version AIMS2 SF [18] and the Timed Up and Go test (TUG) [19]. The Dutch-AIMS2 SF is an arthritis spe cific health status scale and we used the following sub scales: physical functioning, pain, social functioning and m ood symptoms, all scored on a 5-point scale. The AIMS2 SF has been validated for OA in the USA [20] and Ger m any [21]. The TUG is an objective outcome measure for m obility in older patients: the patient is observed and tim ed while (s)he rises from a chair, walks 3 meters, turns, walks back, and sits down again. Secondary outcome measures were patient-reported num ber of contacts with the GP and physiotherapist and whether they used pain m edication (over the counter (OTC) or prescribed).
All outcome measures were collected at baseline and after 6 m onths. Baseline and post-intervention data were obtained in two ways. A patient questionnaire was used to collect all patient reported outcomes. The TUG was per formed by the nurse in the intervention group and by a BMC Fam ily Practice 2008, 9:7 http://www.biomedcentral.com/1471-2296/9/7 research assistant in the control group for the baseline data. A research assistant measured in all patients the post-intervention TUG, at this stage he was blinded for intervention-control condition.

P o w er calculation
To estimate sample size, a power calculation was per formed using the subscale lower body limitations of the Dutch AIMS2 SF (Arthritis Impact Measurement Scales Short Form) [18,20] and the Timed Up and Go test [22]. We wanted to detect a small to m edium effect (Mean Standardized Difference of 0.4), with alpha 0.05 and beta 0.20. We needed to include 49 patients per group [23]. Anticipating on refusal rates and loss to follow-up we approached 158 patients.

Analysis
In the analysis, follow-up scores of patients were adjusted for baseline scores [24]. Independent variables were there fore random ization (intervention or control group) and the baseline scores of the respective dependent variables. Data from dropouts and lost to follow-up cases was not available, therefore only cases with data from baseline and after 6 m onths were included. The analyses were per formed using SPSS (version 12) software. Data were checked for norm ality of residuals. For the primary out come measure Timed Up and Go test we used a logistic regression technique. TUG times were divided into two clinically relevant groups (=<12 and >12 seconds) on the basis of literature [25]. Dutch AIMS2 SF scales were ana lyzed with a linear regression technique. The secondary outcome measures (GP visits, physiotherapist visits and use of pain medication) were analyzed using a chi-square test. We did not substitute missing values in any of the scales.

Results
A total of 158 patients were sent an inform ed consent let ter and a questionnaire. After one rem inder 125 patients (79.1%) responded. O f these 104 patients were included and random ly assigned (Figure 1). Fifty-one patients were allocated to the intervention group and 53 were allocated to the control group. Fifty-four patients (of the initial 158) could n o t be included: 33 did no t respond to the study invitation, 7 forgot to fill in their names, 12 did not give inform ed consent, 1 moved to another region and 1 died. Those excluded were not significantly different in age and gender compared to participants. At baseline no differ ences in self-reported characteristics between intervention and control group patients were detected (Table 1). Due to several reasons seven patients withdrew their participa tion during the study (m otivation problems, moved else where, hip/knee surgery, too severe problem s of co m orbidity and treatm ent by a geriatric specialist) and nine patients did not respond to the final patient self-assess m ent questionnaire (Figure 1). No differences in self reported characteristics were found com pared to post intervention responders. Main results are described below and schematically presented in Table 2. P rim ary outcom es W hen considering patients' self-reported functioning, intervention patients' m ean scores changed towards better functioning. In the control group three out of four sub scales in the before-after measurements went in the differ ent direction, thus a worsening in function. However, none of the subscales in the intervention group had a sig nificant im provem ent compared to the control group (table 2).
With respect to the Timed Up and Go test the shift towards the group 12 seconds in the intervention group was m ore or less equal to the shift in the control group. One third of the intervention patients (35%) performed the TUG below 12 seconds at baseline and half o f the patients after the intervention (50%). For the control group this was 41% and 55% respectively.

Secondary outcom es
Intervention patients did not visit their GP or physiother apist more often compared to the control group. In the intervention group 6/40 (15%) patients had 3 or more visits in the past half-year to their GP, com pared to 7/48 (14.6%) patients in the control group (p = 0.81). 8/40 (20%) patients in the intervention group received physio therapist treatm ent for their osteoarthritis complaints, com pared to 6/48 (12.5%) patients in the control group (p = 0.28). Pain m edication use did n o t significantly differ between the two groups (p = 0.49). However, there was an increase in m edication use in the intervention group. In the intervention group at baseline 17/40 (42.5%) patients used m edication for osteoarthritis pain, whereas post intervention this was 22/40 (55%) patients. In the control group the num bers were respectively, 24/48 (50%) and 23/48 (47.9%).

Discussion
This nurse-based intervention did not improve an older OA patient's m obility and functional status, although a non-significant trend towards better functional status was observed. In both study groups patients showed an im provem ent in functional status. There were no signs of negative side effects, such as more pain am ong interven tion patients, and no signs of increased num bers of visits to the GP or physiotherapist. Numbers were small and only powered to identify a m edium difference, and so there is a possibility o f a type 2 error. If the trend in effects observed were confirmed in a larger trial, and if such small effects in a com m on problem such as OA proved to be worthwhile in the long-term, then the intervention might  still eventually prove to be effective. However, it is clear from our results that the intervention did not achieve clear or substantial effects. Several considerations for these findings may appear, such as the time between interven tion and final m easurem ent may have been to short to detect differences. Also, the intervention itself m ight have been too simple to detect differences in these outcome measures. However, our results are consistent with a study similar to ours in the same time period [26]. The interven tion in this study was slightly more extensive, and their follow-up was 6 m onths longer; but their findings were that a nurse-led education programme for patients with osteoarthritis (40 years or older) did not benefit these patients. O n the other hand, another study showed that a nurse-led intervention aimed at improving non-pharm acologic treatm ent modalities instead of NSAIDs was effec tive for OA patients (aged 60 years or older) in primary care [27]. In this study a structured algorithm was used and patients were individually and regularly followed up. A recent trial of self-management o f arthritis in patients 50 years and older showed reduced anxiety and improved patients' perceived self efficacy in managing symptoms, bu t also no significant effects on pain, or physical func tioning [28].

Conclusion
Non-extensive interventions to improve self-management and life style in OA are, on average, no t effective. The counseling may need to be targeted m ore explicitly to individual problems in order to be successful. Perhaps counseling is only useful for a subgroup of OA patients, such as those with insufficient physical exercise who have a m inim um of m otivation to increase their physical activ ities. Furthermore, regular follow up could contribute substantially to the effectiveness of a short intervention. Finally, if a non-extensive intervention is n o t effective in a patient, m ore intensive interventions should be available as part of a larger care programme for osteoarthritis. Other health professionals may need to become involved in the delivery o f m ore intensive interventions, such as special ized nurses (rather than generalistic primary care nurses as in this study) and physiotherapists. It is crucial that the effectiveness and feasibility of such interventions and care programmes are tested, before wide-scale im plem entation is promoted.

C o m p e tin g interests
The author(s) declare that they have no competing inter ests.

A u th o rs' contributions
RW carried out the study and drafted the manuscript. RG and CvW participated in the concept and design of the study and were im portant in revising the manuscript. MW participated in the design of the study, helped in coordi nating the study and drafting the manuscript. All authors read and approved the final manuscript.

A ckn o w led g em en ts
W e gratefully acknowledge Janine Liefers and Karin de W e s t fo r assisting w ith the data-analysis and fo r visiting the elderly at home. The Netherlands Organisation fo r Health Research and D evelopm ent funded this study (Z o n M W , num ber 920-03-252).