Physical activity with person-centered guidance supported by a digital platform or with telephone follow-up for persons with chronic widespread pain: Health economic considerations along a randomized controlled trial

1 Introduction

Chronic pain causes direct costs to the health system, reduced work productivity, and out-of-pocket expenditures to individual patients [1]. However, chronic pain is a broad concept, including diagnoses such as low back pain, neck pain, spinal pain, fibromyalgia, noncancer pain, or any chronic pain. Chronic widespread pain (CWP) is reported by one in ten individuals within the general population [2]. The American College of Rheumatology defines CWP as pain in both sides of the body, both above and below the waist, and axial skeletal pain for at least 3 months [3].

Few studies have investigated the economic impact of CWP specifically, but it has been reported that costs for these patients triple the costs among matched patients without CWP who were attending the same primary care centers [4], with productivity loss due to sick leave being a main contributor to the overall cost.

The evidence regarding cost-effectiveness of, e.g., cognitive-behavioral therapy in this patient group, has varied, indicating that some fairly low-cost interventions conducted over the telephone can improve the quality of life in this patient population [5], but others have not demonstrated cost-effectiveness compared to established willingness-to-pay thresholds [6]. eHealth interventions are becoming more common and it has been suggested it could support self-management among patients with CWP [7]. eHealth has also been used as a tool for communication in a person-centered care intervention [8]. However, in previous studies of person-centered eHealth support [8,9], person-centered care was provided as an add-on to standard care, and the digital tool was used as one of several tools used to conduct person-centered care. These studies thus say nothing about the effects of introducing eHealth specifically, in a healthcare setting already practicing person-centered care.

Person-centered care has been promoted as cost containing with cost-savings being one reason for its implementation in health systems [10]. Thus, a randomized controlled trial was set up to determine the effectiveness of a person-centered intervention, including advice on physical activity, followed by a digital support (intervention group) compared with telephone follow-up (active control) for patients with CWP [11]. The effectiveness evaluation of the trial reported no effect on the primary outcome pain intensity, or on the secondary outcomes except for global fatigue, which was improved in the active control group compared to the intervention group (p = 0.004) [11]. However, for interventions changing how we use our societal resources, another important aspect of the evaluation is the economic outcomes [12]. Several studies have reported that digital support can be cost-effective [13] and potentially even dominating [14,15] a waitlist control alternative. Moreover, among digital interventions, self-guided interventions, or those with on-demand or low clinician activity, appear to be more cost-effective than both clinician-driven digital interventions [16] and the waitlist alternative [15,13], but no studies have previously compared the resource use associated with alternative methods to provide and follow-up person-centered interventions: through digital support or telephone.

The aim of this evaluation was to investigate the resource use and costs associated with co-creation of a person-centered health-enhancing physical activity plan for persons with CWP followed by support through a digital platform, compared to telephone follow-up. This study thus focuses on the excess costs for service delivery as it changes and the opportunity costs of healthcare providers; resources that could be spent on other services had said resources not been used for the intervention under study.

2 Methods

The methods have been reported in detail previously [11], and the single-blinded multicenter randomized controlled trial is reported in ClinicalTrials (identifier: NCT03434899). This section starts with a description of the RCT in accordance with the CONSORT guidelines for information to include when reporting a randomized trial [17], and the TIDieR-telehealth checklist for precision in reporting interventions delivered remotely [18]. The health economic analysis was reported according to the Consolidated health economic evaluation reporting standards [19]. Costs were analyzed from the healthcare perspective and disregarded discounting due to the short time horizon.

2.1 Study population and randomization

As previously described [11], invitation letters to participate in the trial were sent out, between March 2018 and April 2019, to 701 persons who previously had participated in a pain management program at five primary healthcare centers in western Sweden. The sample size calculations based on the primary outcome of the trial indicated 70 participants per group [11]. Screening for eligibility (CWP and aged 20–65 years) was conducted over the phone 1–2 weeks after the postal invitations. Exclusion criteria included other severe somatic or psychiatric disorders, other dominating causes of pain, other severe diseases interfering with the ability to be physically active, or pregnancy. Also, participants were required to have access to a smartphone or a computer and be able to speak and understand Swedish. People with ongoing physiotherapy treatment or exercising regularly were excluded. The final randomization included 139 persons (intervention group: 69; control group: 70, Figure 1); including 15 persons who spontaneously contacted the study coordinator for screening after advertisement in local newspapers, but excluding persons not reachable by phone or declining screening, not meeting inclusion/exclusion criteria, declining participation or not attending baseline examination [11]. Randomization was conducted by site, in blocks, by a person not involved in the evaluation and using numbered envelopes prepared by an independent statistician.

Figure 1

Flow diagram.

2.3 Data collection and cost analysis

Physiotherapists registered the time used for the two visits and for tasks related to the digital platform used by the intervention group, per primary healthcare center. The time used for the control group was approximated assuming that the time for the initial visit and telephone follow-up, combined, equaled the time used for the two visits in the intervention group (Figure 2): thus, the phone call was assumed to be equally time-consuming as the follow-up contact in which participants in the intervention group were provided with access to the digital platform.

Figure 2

Overview of the resource use, by study group.

Resource use was translated to costs (2021-year value) using a cost per hour of SEK 338 based on the mean monthly wage of SEK 33,609 reported for staff involved in primary care rehabilitation services in Region Västra Götaland, social security contributions (46.15%) and holiday payments (13.45% addition to annual salary), divided by working hours (165 h per month). Resource use and associated costs were reported by cost categories and their included components as mean hours and SEK per participant, respectively, with the range by health centers used to indicate variation in a sensitivity analysis. Cost categories were further organized according to distinguishable types of costs relevant in program evaluation [26].

The digital platform used in this project, MinHälsa, was provided without charge by the University of Gothenburg GPCC. The cost for developing and running the digital platform was approximated using a 10-year depreciation (with sensitivity analysis of 5–20 years) of the initial development cost, approximately SEK 1 million, and assuming annual costs (depreciation and rent for server space of 10,000 per year) was distributed over 5,000 or 500 participants, respectively. The contribution of running costs to the overall costs was also plotted and analyzed visually.

All analyses were conducted in Excel.

3 Results

Physiotherapists registered a total of 366 h of resource use (63–129 h per unit, Table S1, Supplementary material), during the 12 months, related to 69 participants included in the intervention group (i.e., 5.3 h per included person). Of these, two participants did not attend the follow-up visit (2.9%), while 60 participants (87.0%) used the platform, and 47 participants (68.1%) continued using the digital platform after the follow-up visit.

Time for the physiotherapists to organize and conduct the initial visit and follow-up was assumed to be the same for the intervention group and the control group (i.e., initial physical visit and either follow-up visit or phone call, respectively); 2.8 h per participant (range: 2.2–4.5) (Table 1). This time covered administration, preparatory work, direct patient contacts, and later documentation of the initial visit and follow-up encounter combined, to create the physical activity plan, as well as work conducted to accommodate rebooking of visits and similar tasks. Excluding time for rebooking encounters, the time corresponded to approximately 2.4 h per participant. No additional time was expected for providing access to the digital support.

For participants in the intervention group, an additional 2.5 h per participant (range: 1.9–3.7) of physiotherapists time was used for the platform, including patient interactions and getting the platform to work correctly. Thus, person-centered support through phone calls was less time-consuming than support through the digital platform; 2.8 h for support through phone calls vs 2.8 + 2.5 h (equals 5.6 h) per patient for support through the digital platform.

Calculating the corresponding costs; the time used by physiotherapists for the initial visits and follow-up (both study groups) corresponded to costs of SEK 958 (range: 746–1,517), with an additional SEK 833 (range: 636–1,257) for time spent in the digital platform to support the intervention group (Table 1). Thus, the total cost in the intervention group was SEK 1791 per participant, compared to SEK 958 in the active control group. Adding the development and running costs for the digital platform resulted in a total cost per participant in the intervention group of SEK 1813 (difference: SEK 855 compared to the active control group). This was based on a 10-year depreciation and 5000 participants, resulting in costs for the digital platform of SEK 22 per participant per year. The sensitivity analysis of depreciation time resulted in platform costs of SEK 12-42 per participant, respectively. Correspondingly, if covering only 500 participants, the cost for the platform increases to SEK 220 per participant (range: SEK 120-SEK 420 based on depreciation time). The contribution of running costs to the overall cost at different rates of inclusion is provided in Figure 3.

Figure 3

Running costs for the digital platform compared to overall per patient costs, by number of patients the platform can include.

4 Discussion

This study indicates that the time spent by physiotherapists in the co-creation of a person-centered health-enhancing physical activity plan for patients with CWP, and consecutive visit to provide access to a digital platform or follow-up through a phone call was approximately 2.8 h and salary costs of SEK 958 per patient. In addition, physiotherapist and patient interaction via the platform corresponded to approximately 2.5 h per patient (corresponding to SEK 833) in the intervention group.

Self-management and eHealth have been recommended among the available tools for adjusting health systems to the future, including the demographic changes and larger demands for individualized treatments and self-care. It has been suggested as a potentially useful tool for documentation and follow-up of health plans. Our previous randomized controlled trial [11] did not demonstrate statistically significant differences between the intervention and active control groups, in either the primary outcome (pain intensity) or most of the secondary outcomes during the 6-month follow-up period. Both groups received person-centered rehabilitation, which may also contribute to the lack of difference in the between-groups analysis in primary outcome in the randomized controlled trial [11]. Several studies include digital platforms similar or equal to the one used in this project [8,9,23]. However, in those studies, the digital platform was part of introducing person-centered care in the first place. Such an approach is in line with the reported preference among many clinicians to view digital interventions as complementary tools rather than replacement of clinical interactions [27]. In the present study, both study groups were self-managing their pain, and regardless of the study group, the total time spent interacting face-to-face or over telephone with physiotherapists was small and provided within a limited time frame. A meta-analysis showed that while potentially more effective than wait list or usual care, there is large variability in the literature as to if self-management has been more effective in reducing pain than studied active control interventions [28].

This study provides a new addition to the available literature on person-centered eHealth interventions, in that it compares support using a digital platform to support using phone calls only. The study indicates that telephone follow-up was less time-consuming and costly but with similar health benefits to the participants. Moreover, a previous qualitative study from the same randomized controlled trial shows that the digital platform enabled participation and interaction for some participants, while the person-centered health-enhancing physical activity plan, offered to both the intervention and control groups, was perceived as the most important component of the intervention [25]. This illustrates how the here reported interventions differ from what has been seen in previous studies of digital interventions directed to patients with pain, reporting beneficial cost-effectiveness of self-guided or low clinician involvement, compared to both clinician-intense interventions [15,16] and inactive waitlist controls [14,13]. Although the time spent per patient was fairly low irrespective of study arm, in this study physiotherapists and participants were actively involved in developing the physical activity plan, with the intent for patients to be encouraged and empowered to self-management of their condition. However, a large proportion of the reported resource use was time spent by physiotherapists using the digital platform, which included handling problems with the platform. Two patients could be provided services using the person-centered guidance and telephone follow-up in the corresponding time used for the services if supported by a digital platform.

Economic evaluation of eHealth interventions has previously been identified as challenging due to, e.g., complexity in its provision and outcomes [29], and this can be amplified by platforms that are not designed to register data on use. Similarly to a previously published study of a person-centered eHealth intervention [30], not all patients were using the platform actively, with approximately 87% of participants using the platform at least once. Low use of eHealth tools can result from several reasons, including lack of relevance to participants and technical difficulties [31,32]. However, low use can risk reducing the potential health benefits of therapy. At the same time, hours spent by physiotherapists using the digital platform may not be as affected by the exact number of patients being active users, as technical challenges and checking up on users’ activities would still to some extent remain. The variation in resource use between healthcare units indicates that the time spent in preparation before patient visits and in handling problems with the digital platform varied most between units, while for many other components, the different units reported more similar resource use. It is, thus, recommended for future studies to report time spent by healthcare practitioners and patients, respectively, on a more disaggregate level to enhance understanding of how the digital platform contributed to patient outcomes for each participant.

There are some limitations to this study that to some extent limit the use of the here presented findings to predict costs of implementing such an intervention in usual care. The main limitation is that resource use was only measured through self-reporting by staff for time spent to care for patients included in the intervention, and it is thus possible that differences between the groups were not accounted for. The resources used by the control group were estimated based on the time used for contacts in the intervention group, assuming that the time used was similar for a face-to-face meeting and a phone call. However, it is possible that the time and resources used for telephone follow-up were smaller than the time spent for follow-up visits with the intervention group, thus resulting in even lower costs for the active control group. This is also indicated in the national statistics for costs of healthcare visits in Sweden, where an indirect encounter, such as a phone call, is weighted as 1/3 of the cost of a face-to-face visit [33]. This assumption is to some extent strengthened when comparing the estimated 2.8 h spent to the time observed in a previous study of person-centered care telephone support (80 min per patient [34]), thus indicating that person-centered care can be effectively implemented over the telephone. Also, quite a few participants were lost to follow-up in both study arms (24.6% in the intervention group and 18.6% in the active control group [11]); thus, the time used may have resulted from fewer participants than included, thus further increasing the time used by staff for handling the platform, per participant. The limitation to self-reports from staff also hindered any assessment of impact on other healthcare resource use or societal costs of either alternative. It is thus possible that we are underestimating the cost-effectiveness of the intervention, if, e.g., access to the digital platform made participants less prone to contact other healthcare professionals, like physicians and nurses. However, the lack of improvement in the primary and secondary outcomes previously reported [11] indicates that such changes, if occurring, were not noticeably beneficial. Another limitation relates to patients being recruited among those who previously had participated in a pain management program, they might have had some pre-understanding that cannot be expected among future participants from a general patient population. Moreover, both study groups were provided person-centered care according to the Gothenburg model, but it can be argued that physiotherapists probably do not practice similar person-centered care in all settings and countries, thus making the comparison to usual care in different settings challenging. Due to limitations the calculations in this study are not directly generalizable to all patients, but it gives an indication of the resource use associated with the use of the digital platform, that can guide future research and implementation. The results should be viewed as a rough estimate indicating how much time can be used by staff to support patients through such a digital tool and how time is divided between tasks, while the time spent on phone calls and visits helps put the time spent in the platform in perspective to how much time is otherwise spent in the care of this patient population.

This health economic study was designed as part of the process evaluation from the trial, thus not providing in-depth knowledge on the time consumption used by or for individual patients, but rather expected to provide an overview of how time was divided between face-to-face interactions and the platform. Still, it is preferable that conducted studies are complemented with evaluations of the associated costs and cost-effectiveness since a lack of cost information has been identified as a barrier to implementation [35]. However, our findings can be used as one source of information as to what can be expected in terms of time use by physiotherapists, in this case for follow-up through alternative channels of communication. Future studies are needed to understand the full economic impact of introducing eHealth tools in person-centered care interventions, including all the aspects relevant to evaluations of eHealth interventions [29]. Assuming the digital platform is further developed and becomes part of standard care, the difference could potentially be reduced, if some of the time spent by physiotherapists with the platform were avoided through standardization and experience. Moreover, further work needs to explore how changes in platform components, as well as preferences and computer skills among potential users, would influence the resources used by both healthcare professionals and patients. With similar outcomes for the patients, and similar, or lower costs to the healthcare provider, our findings can be interpreted to suggest that in the provision of person-centered physiotherapy services, both methods of follow-up could provide useful additions to the tools available, and thus, method of follow-up could be based on the needs and wishes of the individual patient.

In conclusion, after the co-creation of a person-centered health-enhancing physical activity plan for patients with CWP, it was more costly to support patients through a digital platform, compared to telephone follow-up. However, the resource use and costs used for the co-creation of physical activity plan and follow-up were fairly low for both study groups. Future research is needed to explore the full economic impact of introducing this or similar digital platforms to support person-centeredness, including platform functionality, patient preferences, and long-term consequences.