Physiotherapist as primary assessor for patients with suspected knee osteoarthritis in primary care—a randomised controlled pragmatic study

Osteoarthritis (OA) is one of the most common joint diseases and a major cause of chronic musculoskeletal pain and disability in working and older adults [1, 2]. In Sweden, 14% of those over 45 are estimated having knee OA (KOA) [3]. Common OA symptoms are pain, morning stiffness, reduced range of motion, joint instability, swelling, muscle weakness and fatigue [4]. This directly affects patients’ social interactions, mental functioning, and sleep quality [5], and patients with KOA report among the lowest health-related quality of life (HrQoL) compared with patients suffering other chronic diseases [6]. This patient group has a twofold risk for sick leave, and the diagnosis entails a 40–50% higher risk for disability pension. KOA accounts for 2% of all sick days in Sweden [2]. Patients with OA are less active and have more comorbidities than the overall population [7]. OA causes activity limitations, especially in walking [8]. Walking disability is related to a greater risk of mortality [9], which is largely explained by lack of physical activity [10‐12].

Over the last 40 years, the proportion of overweight and obesity in the Swedish population have increased from 35% to 56% among men, and 27% to 41% among women [13]. This will probably affect the incidence of OA since overweight is a strong risk factor [14]. Consultations to healthcare are expected to increase by 30–50% among patients with OA over the next 10 years [15] and primary care physicians will probably face this predicted escalation in OA consultations [16].

Early access to a physiotherapist (PT) has previously been shown safe and effective for patients with musculoskeletal disorders [17, 18]. A pilot study showed that most patients assessed by a rehabilitation professional first (PT, occupational therapist, psychologist or counsellor) did not need to see a physician later [19]. PTs as primary assessors reduce referrals, sick leave, and prescriptions of analgesics for most musculoskeletal conditions [20]. Previous studies of back and neck pain have shown that the most common expectation when consulting a clinician (PT or physician) is not recovery, but having their diagnosis confirmed [21, 22], which is similar to what has been seen in patients with OA [23]. At the same time, patients with OA seem reluctant to seek professional help, partly because they wait until their problems affect their lifestyle or safety (e.g. risk of falling) [23, 24]. Later in the healthcare process, patients with OA feel unsure when to see their physician, they believe that physicians were more for initial diagnosis rather than following treatment [25‐27].

Imaging is not required to diagnose a typical presentation of OA (i.e. usage-related pain, short duration morning stiffness, age > 40, symptoms affecting one or more joints) [28]. Detectable radiographic changes are not always present in early OA [29, 30]. When imaging is not required to diagnose typical symptomatic OA, both physicians and PTs can act as primary assessors. European League Against Rheumatism (EULAR) guidelines do not recommend any particular healthcare provider for the initial assessment since evidence of the effectiveness of various forms of assessment is lacking. The recommendation is that the initial assessment should use a biopsychosocial approach including physical status, activities of daily living, participation in work, leisure or education, mood and health education needs, health beliefs and motivation to self-manage [31]. Core treatment of OA should be individualized and include patient education, an exercise regimen, weight loss if overweight or obese, reduction of adverse mechanical factors, and consideration of walking aids [31, 32]. Muscle strengthening exercises and maintaining physical activities, give patients with OA a better chance to maintain their level of physical function [33]. Physical activity interventions should be delivered by healthcare providers competent in treating this patient group [34]. Advice on exercise and pain relief comprises the bulk of the PT assessment, in comparison with other medical staff members [35‐37], providing a key role in the acute and long-term management of OA. Common PT management in Sweden includes a nationwide program called “Better management of patients with OA” (BOA), consisting of patient education and supervised exercise to increase patients’ efficacy to self-manage the disease and increase their level of physical activity [38]. Participation in the Swedish BOA results in improvements in HrQoL, pain, and self-efficacy [39].

Management of expected increases in OA consultations by early referrals of patients with suspected OA to a PT could save time for primary care physicians and lead to fewer healthcare visits for patients. Early contact with a PT would also aid in assuring correct management through information about the disease and long-term guided strength training, physical performance and fitness. Today, in Swedish primary care, patients can access a PT without referral. Thus, patients with suspected KOA could have a first assessment by either a physician or a PT. However, it is unclear if there are differences between managements reflected in HrQoL, pain and physical function. The purpose of this study was to examine the differences in HrQoL, adjusted for pain and physical function, for patients with KOA when the healthcare process is initiated with PT assessment compared to a physician assessment in primary care. We hypothesise that all patients with suspected KOA could be assessed initially by a PT in primary care, and then referred to a physician if required.

To establish if HrQoL differed between the effects of being assessed by a PT or a physician for suspected KOA, 69 patients were randomised to either a PT or physician as primary assessor. Most of the patients, 79%, participated in the 6-month follow up and 64% completed the 12-month follow-up.

Demographic data and clinical characteristics are presented in Table 1. All 69 patients were included in the mixed effect model analyses. We found that HrQoL improved significantly 12 months after assessment for all patients regardless of assessor (variable “Time”: EQ-5D-3L index, p < 0.001; EQ-5D-3L VAS, p = 0.0049). There were no significant differences in HrQoL between PTs and physicians as primary assessors (variable “Time*Group”: EQ-5D-3 L index, p = 0.18; EQ-5D-3 L VAS, p = 0.49). See Table 2 for EQ-5D-3L index results and Table 3 for EQ-5D-3L VAS results. The final model of the EQ-5D-3L index was adjusted for the confounder’s sex, educational level, pain intensity and physical function. The final model of EQ-5D-3 L VAS was adjusted for the confounder’s sex, pain intensity and physical function.

The model means for the EQ-5D-3L index increased for both groups. The physician group had a larger increase from baseline to the 12-month follow-up (PT = + 0.084, physician = + 0.181). For the EQ-5D-3L VAS, the total increase from baseline to 12-month follow-up in model means was similar between groups (PT = + 9, physician = + 8). The changes in model means are illustrated in Figs. 2 and 3 for adjusted models.

The model means showed an increase in EQ-5D-3L index for the physician group only for the period between baseline and the 3-month follow-up. Additional mixed effect model analyses were made for the first three months only. The final model showed no significant change in HrQoL for both groups (variable “Time”: EQ-5D-3L index, p = 0.42; EQ-5D-3L VAS, p = 0.99), or significant differences between PTs and physicians as primary assessors (variable “Time*Group”: EQ-5D-3L index p = 0.24; EQ-5D-3L VAS, p = 0.55). See Additional file 1 a and b.

The findings of this study imply that PTs can be the first assessor when patients with suspected KOA seek primary care for the first time. Our results are similar to previous findings showing that PTs are appropriate primary assessors for patients with musculoskeletal disorders [19, 20, 51]. PT as primary assessor is suggested as a model that uses healthcare resources more efficiently where the most appropriate healthcare professional assesses the patient’s needs [52]. Several studies [18, 53‐57] have reported that patients experienced as much or even greater patient satisfaction with a PT assessment than with physician assessment. Thus, using this model of care could require recruitment of more PTs, which could mean increased costs. Future research is needed to explore how patients with KOA experience PTs as primary assessors and the cost efficiency of this task-shifting model.

In this study, we found no statistically significant differences in HrQoL between groups 12 months after assessment. Even though the physician group had a 0.097 larger increase in model means for the EQ-5D-3L index 12 months after assessment, this does not exceed the minimal clinical difference of 0.121. The reason for the large improvement in HrQoL for the physician group over the first three months could be related to the baseline mean value for the EQ-5D-3L index, which was much lower for the physician group when compared with the PT group. Patients allocated to a PT as primary assessor were somewhat older and had a slightly higher BMI at baseline. These patients also had longer pain duration, but graded lower pain intensity and had better physical function, which could also explain why patients in the PT group rated a higher HrQoL. The variation in the baseline values of EQ-5D-3L index were adjusted for in the mixed effect model analyses where patterns of change for each patient were used in the analysis [58]. The pragmatic study design might have contributed to a larger variation in patient characteristics within groups [59]. A larger sample size and/or use of a questionnaire with more levels such as EQ-5D-5L [60], could probably provide a better distribution of the EQ-5D index values at baseline.

Both groups improved in HrQoL 12 months after assessment. PT treatment in this study (i.e. the BOA program) were individualized, comprising patient education and exercise regime with the purpose of increasing patients’ abilities to self-manage. This program has been developed in accordance with national and international guidelines [31, 32]. Participation in the BOA program decreases pain, and increases HrQoL and self-efficacy [39]. Exercise therapy, with or without being combined with other treatments, is an effective intervention to improve HrQoL in patients with KOA [61, 62]. It is possible that most patients in the present study received PT treatment, either by randomisation to a PT as primary assessor or were referred to a PT by a physician. This could explain why the improvements in HrQoL were seen in both groups. If there were no significant differences in HrQoL regardless of the primary assessor, and most patients probably ended up with PT treatment, one way to make the healthcare process more efficient for patients with KOA could involve PTs as the sole primary assessor and use physicians when required.

Most patients were recruited from primary care centres, implying that they sought a physician consultation for their knee pain. It might be possible that the patients in the physician group were positively affected by the fact that they met a physician, which they expected from the beginning. Patients expect investigations such as x-rays or magnetic resonance imaging (MRI) to provide evidence of their experienced problem. Lack of these investigations could be experienced as a possible barrier for being understood and helped [63]. In this study, it could have led to higher satisfaction in the physician group due to receiving expected examinations and treatments, which could have affected the results with a higher HrQoL, less pain and improved physical function because of decreased anxiety for their problems. It would have been interesting if the patients did not know what profession the primary assessor had in order to rule out a potential placebo effect. Unfortunately, this was not possible in this primary care setting because most primary care centres and rehabilitation centres in primary care have different locations.

Changing the inclusion criteria is one of the methodological limitations of this study. With two criteria removed, only age > 38 years and knee pain without traumatic onset was used to detect suspected KOA. Wesseling et al. have been using knee pain and age over 45 as the only inclusion criteria when screening for early OA. None of the participants in their study had radiographic signs for OA, and 76% fulfilled the clinical ACR criteria for KOA [64]. Another study [30] using only knee pain of less than three months’ duration, age 35–54 years, and no history of previous knee injury or inflammatory joint disease found that 70% had clinically classified OA according to ACR [40]. In the same study, they found that 86% of the patients with normal radiographs at baseline, later developed signs for fulfilling the criteria of radiographic OA (according to Kellgren/Lawrence grade 1) in a 12-year period [30]. The motive for changing the inclusion criteria in this study was due to low patient inflow. The diagnostic accuracy of the ACR [40] clinical criteria for patients with early mild OA could have been too low. ACR clinical criteria seem to reflect later signs in advanced disease [65]. Future research would certainly benefit from specific criteria to detect early KOA to enhance the knowledge of early diagnosis and treatment in this stage of the disease.

One reason for the low patient flow could have been organisational, which involved both primary care centres and rehabilitation centres in primary care during the study period. The recruitment process was closed when no new participant was recruited for an entire year. Ongoing reorganisation was probably given priority at the clinics instead of recruiting study participants. Despite reorganisation, similar Swedish clinical trials have had similar problems with low patient flow when recruiting participants to their studies [66, 67].

Due to the low patient flow, the required sample size of 100 patients was not reached, which is a limitation of this study. Sixty-nine participants were included at baseline, which increases the risk of a type II error in the study. The dropout rate was 40% for the PT group and 32% for the physician group at the 12-month follow-up. The benefits of an analysis with mixed effect models is that participants with missing data can be used in the analysis as long as the missing data is missing-at-random. The missing value analysis for this study showed that the gender distribution, age range for the dropouts, and the reason for missing at follow-ups were similar in both groups (see Fig. 1). Mixed effect models handle the imbalanced data in available observations. Using mixed effect models is a strength of this study. The analysis is specifically designed for analysing data characterised by repeated measurements on the same individual [58]. The mixed effect models gave us a result showing how the primary assessment for patients with KOA affected their HrQoL 12 months after first assessment with consideration to possible confounders, which we cannot obtain from analyses with traditional statistical methods such as the Mann-Whitney U test.

A traditional randomised controlled trial with highly controlled treatments aim to test the true effect of a treatment, by ruling out placebo effects and extraneous effects (patient’s or healthcare provider’s knowledge or expectations of the treatment that could affect their behavior), and assuming that a patient group is homogenous. Compared to pragmatic trials, the generalizability for traditional randomised controlled trials are lower, since treatment outcomes would likely be affected by heterogeneous patient groups, placebo and extraneous effects which are present in the real clinic [59]. The pragmatic design is a strength of this study and the results have good external validity since the interventions have already been tested in a real clinical setting.

With this study, we showed that PTs and physicians did not differ as primary assessors for patients with suspected KOA, regarding HrQoL up to 12 months after patients consulted primary care. These results support previous findings showing that PTs could be used as the primary assessor for patients with musculoskeletal disorders. Our results imply a task shift in primary care, which would probably enhance access for patients with KOA to a better OA management including core treatment of patient education and exercise.

In this study, we found no differences in HrQoL, when adjusted for pain and physical function, for patients with KOA when the healthcare process was initiated with a PT assessment compared to a physician assessment in primary care. Both assessments resulted in significantly higher HrQoL at the 12-month follow-up. The results imply that PTs and physicians in primary care are equally qualified as primary assessors.