Evaluation of intermediate care for knee and hip osteoarthritis: a mixed-methods study

All healthcare providers (GPs (n = 4), orthopaedists (n = 3) and healthcare managers (n = 1)) providing intermediate care in practices A and B were invited for semi-structured interviews. Non-responders received a reminder within 2 weeks of the invitation. These interviews focused on their perceived facilitators and barriers with intermediate care. In addition, GPs were asked to invite a convenience sample of patients with KHOA who had at least one intermediate care consultation to be interviewed. These interviews included pre-determined topics from the literature and based on the expert opinion of the research group. Based on these topics, interview guides with open-ended questions were composed and pilot tested (Supplementary Table 1). One researcher (IGA, physiotherapist and researcher) conducted the interviews face-to-face in the general practice or by telephone. The pre-determined topics needed to be covered during the conversations, although the interviewer was allowed to diverge from the interview guide to explore additional topics. Field notes were made by the interviewer during and after the interviews. The interviews were audiotaped, transcribed verbatim into written form and read by the interviewer to increase the validity. To guarantee transparency, all participants were offered to receive their transcript for comment and correction.

We included satisfaction questionnaires that were already designed and collected by GPs and orthopaedists from practices A and B. Immediately after the intermediate care consultation, patients were asked by the GP assistants to complete the questionnaire anonymously. These questionnaires included questions about satisfaction with: 1) the consultation; 2) the provision of information by healthcare providers; and 3) the patient-healthcare provider relationship. Satisfaction scales in the questionnaires varied between the two practices (1–10 scale vs. 4-point Likert scale). Patients were allowed to add free-text comments.

Data on patients’ characteristics (e.g. age and sex), healthcare consumption in terms of number of consultations (i.e. face-to-face consultation, visit and telephone contact) and GP referrals to orthopaedics in hospital care were collected retrospectively from the medical records of practices A-C. In the Dutch healthcare system, GPs can refer patients to primary care physiotherapists, which is recommended by the Dutch GP guidelines for non-traumatic knee complaints [19] as part of OA core treatment [15, 19]. GP referrals to primary care physiotherapy were also collected retrospectively from the medical records of practices A-C. Records of patients with KHOA and with at least one consultation, visit or telephone contact by their GP either before or after the implementation of intermediate care (pre-implementation and post-implementation period) were selected. A diagnosis of KHOA was defined following the International Classification of Primary Care [20], coding L89 (hip osteoarthritis) and/or L90 (knee osteoarthritis). The duration of the pre- and post-implementation periods varied between the practices, since the practice projects were running for different periods at the time of this study. In practice A data was collected six months before and after implementation (i.e. one year in total) and in practices B and C one year before and after implementation (i.e. two years in total). Figure 1 shows the time periods for the practices in the pre-and post-implementation periods.

Interviews were analysed following the steps of thematic analysis described by Braun and Clarke and with assistance of the software MAXqda Version 2018 [21, 22]. We used a semantic approach (i.e. analysis close to participants’ language, capturing explicit meaning) to describe the opinions of different participants in intermediate care. Deductive coding was done based on pre-determined themes, and inductive coding based on additional topics that we extracted from the open-ended questions. Two interviews were coded independently by two researchers (IGA and DS) and then compared. Any disagreements were discussed until consensus was reached. The resulting codes were further applied for analysis, and iteratively modified if necessary after each interview coding. Codes were structured hierarchically and analysed using the thematic framework. Relevant quotes were selected from the transcripts [22, 23].

Four GPs, two orthopaedists, and one healthcare manager were interviewed. Only one orthopaedists did not respond to the invitation letter and was therefore not interviewed. Furthermore, four patients invited by their GP were interviewed (see Supplementary Table 2 for their characteristics). The duration of the interviews varied from 7.5 to 45 min. Information about the content of the projects that we extracted from the interviews is presented in Supplementary Table 3.

In total, 100 patients from practices A and B completed the satisfaction questionnaires (data shown in Supplementary Figure 1). Results from practice A (n = 39) showed that most patients were ‘very satisfied’ with the consultation (63%), provision of information by healthcare providers (67%), and the patient-healthcare provider relationship (72%) (4-point Likert scale ‘very unsatisfied’ to ‘very satisfied’). The remaining patients were ‘satisfied’. Results from practice B (n = 61) showed a mean satisfaction score of 9 (range 0 to 10) for the patient-healthcare provider relationship and provision of information, and 8.9 for the consultation in general. Specific comments about the intermediate care consultation were positive, for example about the knowledge of the orthopaedist, short waiting times and consultation hours in the evening.

A total of 96 patients with KHOA were seen during the pre-implementation period and 208 during the post-implementation period. Of the patients in the post-implementation period, 26.4% received intermediate care and the remaining 73.6% received regular GP care. Patients in the pre-implementation period had a mean age of 71.3 years (SD = 10.8), 66.7% of them were female, 67.7% had knee OA, and the remaining 32.3% hip OA. Patients in the post-implementation period had a mean age of 69.3 years (SD = 9.8), 66.3% of them were female, 65.9% had knee OA, and the remaining 43.1% hip OA. Patients in the pre-implementation period received on average 2.40 consultations (SD = 1.59) and patients in the post-implementation period on average 2.52 consultations (SD = 1.78). These characteristics did not significantly differ between patients in the pre- and post-implementation period (Table 3).

The percentage of referrals to physiotherapy increased significantly in the post-implementation period compared to pre-implementation (absolute difference = 15%; 95% CI = 7.19 to 22.8). In contrast, the percentage of referrals to orthopaedics increased slightly, but not statistically significant (absolute difference = 5.9%; 95% CI = -6.18 to 17.9) (Table 4). Supplementary Table 4 shows the referrals and number of consultations stratified by patients who received regular GP care and patients who received intermediate care during the post-implementation period.

This evaluation study showed that GPs and orthopaedists experienced more intensive collaboration due to the implementation of intermediate care in general practice. This led to a perceived increase in their knowledge, for the GP enabling a better selection of referrals to orthopaedics and physiotherapy. Patients were satisfied and experienced better access to healthcare, and the benefits of a trusted environment and specialists’ knowledge. The percentage of referrals to physiotherapy increased significantly after the implementation of intermediate care. The observed increase in referrals to physiotherapy contributes to the quality of care, since offering patients with KHOA physiotherapy is an indicator for high quality of care [24]. Healthcare providers experienced better selection of referrals to orthopaedics and less healthcare consumption. However, the actual observed percentage of orthopaedic referrals and the mean number of consultations in the general practice did not decrease after the implementation.

A strength of the study is the mixed methodology that enabled a comprehensive evaluation of intermediate care with regard to the experiences of patients and other stakeholders, patient satisfaction, and referral trends. However, the findings of this study are subject to several limitations. First of all, the retrospective design of this study led to a lack of proper baseline measurements. This limited our information on for example the severity of KHOA and conclusions about the appropriateness of referrals to orthopaedics and physiotherapy are therefore not possible. It should be noted that the differences in referrals might partly be due to confounding by indication for a referral (e.g. more severe patients may be more likely to be referred to orthopaedics) and not only the effect of intermediate care. We were not able to draw conclusions about the effect of differences in patients’ characteristics between the pre- and post-implementation period on referrals. A regression model which would be appropriate for this kind of analysis requires independent samples, which might not be the case in our study. Nevertheless, explorative analysis showed no effect of age, sex or affected joint on referrals (data not shown). Furthermore, the current study only captured GP referrals to physiotherapy. Since 2006 patients in the Netherlands can also access physiotherapy care without a GP referral [25]. The number of physiotherapy uptake might therefore be underestimated in this study. Also, GPs invited a convenience sample of patients for the interviews and the experiences of those patients were generally positive. However, this may be the result of selection as GPs may have been more inclined to invite patients who are more positive about the provided care. Furthermore, all patients preferred a telephone interview instead of face-to-face interview. This, in addition to the low number of patients included, might have influenced the limited data saturation. As a consequence, findings from the interviews with patients might not be reflective of the full range of patient experience. Lastly, the findings of this study are restricted to intermediate care, a ‘shifted-outpatient’ model specifically in the Netherlands. Therefore, applicability to other countries may be limited.

Previous studies have shown that GPs have little confidence in their ability to diagnose and manage musculoskeletal conditions [26, 27]. The present study showed that GPs and orthopaedists providing intermediate care felt that they learned from each other and that their knowledge increased. Therefore, intermediate care might be a solution to increase the confidence of GPs.

Furthermore, this study showed that orthopaedists experienced a higher workload due to intermediate care. Previous research has shown that a substantial proportion of patients referred to secondary care could instead be seen by a GP with special interest in this area [28]. This may therefore be helpful in managing the high workload for orthopaedists and is worth exploring in future research. Orthopaedists also felt they had limited access to diagnostic facilities in the general practices, which is in line with a previous study [29] that evaluated barriers and facilitators in substituting hospital care with primary care. This barrier may lead to an increase in healthcare costs, as GPs in one project started requesting X-rays routinely before referring patients to intermediate care, while current clinical practice guidelines [19] do not recommend routine X-rays in primary care settings. Previous studies reported a decrease in referrals to orthopaedics [11, 13, 29, 30]. However, the current study shows that while healthcare providers experienced a better selection of referrals to orthopaedics, the actual observed percentage of referrals did not decrease. This might be due to the short follow-up time of the intermediate care projects. A longer follow-up time is probably needed to observe more reliable effects of intermediate care on referrals and healthcare consumption.

Healthcare providers who were interviewed in the present study felt that the longer consultation in intermediate care is a benefit for patients. However, a recent study [31] showed that patients did not find the duration of the consultation very important, while healthcare providers did. Our study shows that healthcare providers experience longer consultations as a facilitator for providing better medical advice to patients, which might reduce the patient’s need for further consultations. This finding is in line with results from a previous observational study [32]. Even though the observed mean number of consultations in the current study did not yet decrease, future research with a longer follow-up time may show a reduction.

We strongly recommend replication of this study with more rigorous data collection methods and a prospective study design (e.g. cluster or stepped wedged randomized controlled trial that decrease potential bias) to increase the reliability of the findings. In addition, a longer follow-up time in future research would be justified to show the long-term effects of intermediate care on referrals and healthcare consumption. Further research including different forms of intermediate care is also needed to provide more extensive recommendations on how to implement intermediate care most effectively, such as electronic consultations between GPs and specialists [33]. Lastly, our research indicated that intermediate care reduces healthcare costs based on the experiences of healthcare providers, as expressed in the interviews. Future research into the cost-effectiveness of intermediate care is recommended to strengthen the evidence for this result.