Background
Over the past 15 years there has been a steady increase in the rates of primary total knee arthroplasty (TKA) and total hip arthroplasty (THA) procedures undertaken annually in Australia [
1], reflecting trends also seen internationally [
2‐
5]. These increasing rates of surgery have not only led to an increase in the utilisation of acute-care services, but also an increase in the utilisation of inpatient rehabilitation and other modes of post-operative therapy provision [
5‐
7]. As the demand for these services has grown, the ongoing viability of the cost of inpatient rehabilitation in particular has been called into question [
8]. A pilot investigation in Germany demonstrated that inpatient rehabilitation was not cost-effective when compared to an outpatient alternative following hip arthroplasty from the perspective of the healthcare insurer [
9]. In order to restrain costs, policy changes in the US point towards tightening admission criteria restricting inpatient rehabilitation after arthroplasty surgery [
10,
11]. Implicit in these changes is the assumption that arthroplasty patients can be managed by outpatient services in a more affordable manner, without compromising their healthcare outcomes.
In the context of TKA and THA no high-level evidence supporting or contesting the benefits of inpatient rehabilitation over outpatient or home programmes was available at the time this study was instigated. One Canadian randomised controlled trial (RCT) that combined hip and knee patients compared 18 days of inpatient rehabilitation to eight domiciliary sessions (physiotherapy in the home). No significant differences were shown in the outcomes measured, which included the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), Short Form-36 and patient satisfaction [
12].
In Australia, inpatient rehabilitation following arthroplasty surgery is most commonly utilised by private consumers i.e. those who are privately insured or elect to cover some or all of the costs of their healthcare, in contrast to public patients who have their healthcare needs met within the public health system, without payment. Recent estimates indicate that a median 40% of privately insured patients per surgeon were transferred to inpatient rehabilitation in 2014 following TKA, though this figure ranges from 0% to 100% [
13]. This contrasts with the public sector utilization rate of 21% [
14], suggesting that factors other than need drive the high utilization rate in the private sector. In the absence of conclusive evidence, the journey from the post-operative acute care setting to inpatient rehabilitation services is presumably guided by the consumers and clinicians involved, with various opportunities for the expression of preferences as decisions. While taking into account variances in the preferences of clinicians regarding rehabilitation alternatives [
15], decisions around rehabilitation types utilised after arthroplasty are likely to be multi-dimensional. They may be guided by reasons related to a patient’s clinical status, but other factors, such as patient expectations, healthcare professionals’ personal preferences and related conveniences may also play a part [
8,
16].
Primary objectives
Given that there are serious questions about the long-term sustainability and cost effectiveness of broad and untargeted inpatient rehabilitation provision following arthroplasty surgery, a greater understanding of factors shaping consumer and clinician preferences, as well as the manner in which decisions relating to rehabilitation following surgery are made, will be invaluable when reviewing current services or designing new healthcare delivery systems [
17]. Utilising a mixed methods design, the primary objectives of this study were to identify the preferences of private consumers (patient and carer) and clinicians [orthopaedic surgeon (OS), physiotherapist (PT), and rehabilitation specialist (RS)] for different modes of rehabilitation utilised after knee or hip arthroplasty and the factors which influence decision making for rehabilitation following surgery.
Methods
Recruitment and consent
All participants were volunteers and provided written, informed consent. Data collection occurred in two phases. Phase 1 involved consumers and was nested within a larger multicentre, observational study (ClinicalTrials.gov NCT01899443). Consecutive eligible private patients about to undergo arthroplasty surgery and their carers were invited to participate while attending a pre-operative admission clinic at one of two private arthroplasty hospital providers in New South Wales, Australia. Patient eligibility included having a principal diagnosis of osteoarthritis and was about to undergo either a unilateral or bilateral TKA or THA. Carer eligibility included being identified as the primary carer for one of these individuals. Sites with different business models were chosen, to investigate whether these may have an impact on a patient’s treatment after surgery.
Phase 2 involved key clinicians from three disciplines involved in the care of knee or hip arthroplasty recipients. A separate computer-generated randomisation list was created for each of the individual groups: orthopaedic surgeons, rehabilitation specialists and physiotherapists. For orthopaedic surgeons and physiotherapists, the hospitals listed in the Australian Orthopaedic Association National Joint Replacement Registry were used to generate a random sample of sites and a surgeon and physiotherapist at each identified site were invited to participate. For rehabilitation specialists and physiotherapists working in the sub-acute rehabilitation services, a list provided by the Australasian Rehabilitation Outcomes Centre was used to generate a random sample of sites, with the same method of recruitment and random sample generation utilised. All clinician participants had been responsible for the care of knee or hip arthroplasty recipients in the private healthcare sector in New South Wales within the last 12 months.
Sampling and data collection
The determination of an appropriate sample size in qualitative research is a key component of the legitimacy of analysis and conclusions drawn [
18]. Heterogenous sampling was used for both consumer and clinician components to capture a wide range of experiences relating to modes of rehabilitation [
19]. The quality of the information collected was assessed after each interview, with consideration of newly emerging themes [
20]. The sample ceased once it was determined that a sufficient number of information-rich cases had been drawn, coupled with saturation, i.e. no new information was being revealed from the interviews [
21]. For patients, this occurred after 38 interviews, while the threshold was 19 for carers. It also took 19 interviews to reach this point with OS, while only 10 PT and eight RS interviews were required to reach this point due to the relative homogeneity of their responses.
Data for both consumers and clinicians were collected from one-to-one semi-structured interviews conducted between January 2014 and February 2015. For consumers, the semi-structured interviews were developed by clinicians familiar with the pre-admission and post-surgery settings, while for clinicians they were developed in consultation with an expert panel comprising an OS, RS and PT. Both were piloted before use. For patients, eight open-ended questions were posed which covered the following areas: the rehabilitation they had received, the process of decision-making they had undertaken, who had been involved in this decision, influencing factors, and options they had available (Additional file
1: Appendix 1). Four open-ended questions were asked of carers about their relationship with the patient and what they had done to support them post-surgery (Additional file
1: Appendix 2). There were seven open-ended questions posed to clinicians that canvassed their thoughts on inpatient rehabilitation and other options, as well as their current practices and options in this regard (Additional file
1: Appendix 3).
For consumers, demographic and other contextual data obtained at the time of consent included age, gender, and working status. In addition, Oxford Knee or Hip Scores [
22] and EuroQol health related quality of life scores [
23] were also obtained from patients. Consumer interviews took place approximately 6 weeks after the patient participants’ had their surgery. Data about clinicians included age, gender, years practised in the related field and, for OS, the number of lower limb arthroplasty’s performed annually.
In addition to the open-ended questions, five alternative rehabilitation types were presented in the interviews for participants to rate. These were based on modes provided in other countries and on those composed by the investigators based on current knowledge of patient preference for rehabilitation [
24]. The options were: outpatient group therapy; outpatient one-to-one therapy, domiciliary therapy, hotel-based rehabilitation and inpatient rehabilitation. Consumers and carers were asked to rank the five options in order of preference. Clinicians were asked to rate the acceptability of each using a five point Likert scale, which was anchored with highly unacceptable and highly acceptable.
Data analysis and management
The qualitative data were examined using principles of thematic analysis, a method utilised to identify, analyse and report patterns within data [
25]. All interviews in this study were digitally recorded and transcribed verbatim and, as recommended by Miles et al. [
26], were reviewed against audio recordings to maximise integrity and trustworthiness of data. This allowed for coding of ideas and understandings that may otherwise have been missed, and the ability to return to and recode old data. It also preserved the tone and tempo, silences and statements of participants [
27].
Initial discursive codes were generated by one of the researchers (MB). The elucidation of these codes was assisted by a process of listening to and reading transcriptions of the audiotapes, as well as consulting colleagues and perusing relevant literature. QSR’s NVivo qualitative analysis software [
28] was utilised to electronically manage data. An initial set of themes were categorised, examined for variability and consistency, re-checked against audio recordings and transcriptions, and then combined to outline the primary overarching factors and their components. These were then discussed with two researchers (JN and GS) to collate and refine the themes, so that clear, identifiable distinctions were developed between them [
25]. In the final stages of analysis, earlier transcripts were re-read to ensure faithfulness of final results drawn, with emphasis placed on explaining the meaning and implications of each of the factors identified. Finally, categories were merged into larger groups, culminating in the finalisation of major and sub groups, and the drawing and verification of conclusions [
26].
Descriptive statistics were generated for the demographic, clinical and quantitative components of the study.
Discussion
Considering the concerns around the long-term sustainability and cost effectiveness of broad and untargeted inpatient rehabilitation provision following arthroplasty surgery, the findings of this study are timely, and provide a clear description of factors shaping consumer and clinician preferences, as well as how decisions relating to rehabilitation following surgery are decided upon.”
This study clearly exposes the many factors which influence private consumer and clinician preferences for particular modes of rehabilitation and associated decision-making following arthroplasty. These factors generally relate to clinical and social matters, but also include a sense of entitlement and extrinsic influences. The pattern of consumer preference tending towards the mode of rehabilitation they received mirrors those outlined by Naylor et al. [
24] in a previous study investigating patient preferences after knee arthroplasty in the public sector. That preference is linked to past experience suggests a general satisfaction with therapy received. It is unclear whether this is because the mode of rehabilitation delivery undertaken by each patient was the most appropriate for their unique situation, or that the perceived quality of care across modes was high. If the latter, this does not discount the importance of the mode itself, but suggests that alternative modes of rehabilitation are likely to be acceptable to patients if they are of high quality. As outlined by Perkovic et al. [
29], if it can be demonstrated that alternative models of care are as effective as existing services, but cheaper, an efficient system would encourage the uptake of these services.
A recent randomised, controlled trial, published after this study concluded, indicated that inpatient rehabilitation was not more effective than a monitored home-based program when measured across a range of outcomes up to 52 weeks post-surgery among adults undergoing uncomplicated total knee arthroplasty [
30]. These findings do not support inpatient rehabilitation for this group of patients. With these findings in mind, one point for consideration for healthcare providers, policy makers and insurers is a re-examination of the pathway to inpatient rehabilitation, to look for ways to reserve this intensive, and costly, mode for patients who have a specific clinical or social need for it. Our interviews showed that healthcare staff involved in these decisions, particularly OS, are aware of alternatives, and often have a preference for them. However, they are reluctant to modify consumer expectations and sense of entitlement, or simply do not see the need for this course of action. This principle of entitlement alluded to by both consumers and clinicians during the interviews can be seen to be somewhat at odds with the attribute of efficiency, which has long been a pillar of the Australian healthcare system [
31]. This is particularly pertinent when a sense of entitlement overrides clinical judgement, as inappropriate care is inefficient.
When exploring possible reasons for this situation, a number of potential dilemmas may be responsible. One reason may be the additional time it would take within an orthopaedic consultation to provide information and talk through other options available; a review of 38 studies looking at the most frequently reported barrier to implementing shared decision making in clinical practice was time constraints [
32]. Another may be the financial implications for surgeons who did not refer patients with a preference for inpatient rehabilitation to that mode. If a patient presented with a strong preference for inpatient rehabilitation that the clinician opposed, the patient could always go to another surgeon who supported or allowed their preferred treatment. The formulation of guidelines or standards for orthopaedic surgeons or other healthcare providers regarding the reservation of inpatient rehabilitation for patients with a specific clinical or social need might be considered to counter this. OS and healthcare providers need to also consider these needs prior to surgery, and play a more active role in promoting modes of rehabilitation that would address these. Private health funds could do the same for their members by looking to address competing financial tensions, and develop supplemental services and clearer pathways to alternative rehabilitation modes and settings. One option may be the removal of financial barriers in the form of ‘out of pocket expenses’ for members who elect to go directly home from the acute facility after arthroplasty. Also, although familial relationships cannot be changed, what could be offered to facilitate an early discharge home after surgery are services such as transport, cleaning and meal preparation, which were identified as determining factors for participants who chose not go to inpatient rehabilitation, and from those to whom alternative modes were preferred. Another avenue which could be considered to at least partially overcome the issue of transport is domiciliary physiotherapy.
A separate factor influencing the treatment pathway after arthroplasty may be the business model of the site in question. The site at which inpatient rehabilitation was generally presented as ‘part of the package’ was run by a healthcare organisation which owned both the acute and rehabilitation facilities. The second, where patients went through surgery and associated treatment without knowledge of inpatient rehabilitation as an option, was owned by an organisation which did not have a business interest in any local rehabilitation facilities. Further study in this area may establish whether this pattern exists on a wider scale.
The clear preference for inpatient rehabilitation by RS is largely unsurprising and revolved primarily around the perceived clinical benefits of this mode. The fact that these same RS benefit personally from the provision of this service was acknowledged during the interviews. Their preferences in this regard, however, seemed to have limited influence on the decision to send patients to inpatient rehabilitation, as they were generally not a part of the decision pathway leading to that setting.
Conclusion
This study has provided a unique opportunity to articulate stakeholder attitudes towards various rehabilitation types, and the factors which guide consumer and clinician decisions relating to the utilisation of different rehabilitation modes following arthroplasty. An understanding of consumer and clinician preferences for rehabilitation, particularly inpatient rehabilitation, should help to inform ongoing and future models of care delivery, hand-in-hand with new evidence of effectiveness as it emerges. This will be particularly crucial to consider if alternative, less costly models of care are to be developed for, and acceptable to, the private sector.
This study indicates that no one mode of rehabilitation provided following knee or hip arthroplasty is singularly preferred by stakeholders. Consumers may have a preference for the more expensive rehabilitation approaches, but if such modes are shown not to be more effective, then there is a need to explore less costly modes of rehabilitation following arthroplasty. If change is to be enacted, clinicians involved in the decision-making process will need to consider the comparative effectiveness of rehabilitation types, as well as the clinical and social needs of their patients, and play a more active role in promoting appropriate options to their patients
Acknowledgements
The authors wish to thank: Helen Badge, as the project manager of the larger study from which our consumers were recruited; Deanne Jenkins, who helped with consumer recruitment; Violette Benjamin, who assisted with the interview transcription, and Jason Li and Minh Nguyen, who assisted with administrative tasks related to the study.