Patients undergoing spinal surgery report high levels of physical complaints as well as insecurity, pain, low well-being, stress, and anxiety before and after surgery [1-3]. Postsurgical recovery often entails moderate to severe postoperative pain, and approximately 20% to 30% of patients undergoing spinal surgery do not experience (long-term) pain relief after surgery [3,4]. This results in a longer hospital stay, higher health care costs, longer physical and mental recovery, delayed return to work, and the potential development of chronic pain. The potential transition from postoperative pain to chronic pain is a major issue as chronic pain affects many aspects of a patient’s life, such as work; physical, emotional, and social well-being; and quality of life [5,6].

Mental factors such as cognition, emotions, and expectations play a significant role in the experience of pain [7]. The fear-avoidance model explains the trajectory from acute to chronic pain through fear and catastrophizing, which is the tendency to enlarge the threat of pain and a feeling of helplessness, leading to an increase in pain avoidance as a dominant coping strategy [8]. Pain avoidance then leads to a less active lifestyle, thereby worsening rather than relieving pain. High levels of catastrophizing and fear have been found to predict higher levels of (postoperative) pain and pain chronicity and a lower quality of life in patients who undergo surgery [4,9-11]. In addition, realistic expectation management is key as unrealistic or unfulfilled expectations about surgery and preoperative stress may lead to the experience of higher levels of postoperative pain [12-18]. This link between mental factors and expectation management with postoperative outcomes shows that psychological preparation before surgery is essential to improve recovery after surgery.

Although the potential benefits of psychological preparation before surgery have long been known [19], a more recent meta-analysis of psychological preparation techniques before surgery could not find strong evidence from high-quality research to verify these claims [20]. Potentially, interventions that focus on promoting adaptive coping skills and reducing maladaptive emotion regulation skills are more promising. Smith and Zautra [21] and Sturgeon and Zautra [22] found, for example, evidence that coping strategies such as (pain) acceptance, engaging in beneficial social interactions, and experiencing a value-based purpose in life have the potential to improve mental well-being and promote resilience in the face of (chronic) pain. Sustainable resilience to (chronic) pain requires skills promoting adaptation and mental health in the long term [22]. In other words, adaptation to different contexts and circumstances requires (psychological) flexibility and a long-term focus. In the context of pain, psychological flexibility implies that painful sensations, feelings, and thoughts are accepted as opposed to avoided and that attention shifts toward personally valued goals [23]. For patients undergoing spinal surgery, psychological flexibility enables them to cope with the fluctuating circumstances surrounding surgery (eg, insecurity, fear, and pain) in a flexible rather than rigid manner. Being able to accept negative emotions or sensations such as pain and insecurity in the face of surgery might prevent catastrophizing, fear, and avoidance behavior.

Psychological flexibility is the aim of acceptance and commitment therapy (ACT) [24]. ACT is based on the relational frame theory and focuses on performing value-based activities in life even in the face of insecurity and adversity [24]. In a similar way, positive psychology (PP) is the scientific study of well-being and optimal functioning, focusing on human flourishing instead of reducing the risk factors for psychopathology and malfunctioning. PP involves topics such as strengths, virtues, meaning, happiness, gratitude, compassion, resilience, and flourishing [25]. ACT and PP can help experience a value-based purpose in life, promote resilience, and improve mental well-being. Mental well-being comprises positive emotional, psychological, and social functioning [26]. The presence of higher levels of these 3 dimensions of well-being is an indicator of flourishing [26-28]. Promoting positive resources and skills that contribute to successful adaptation and (mental) health are the aims of PP interventions (PPIs) [25,29]. PPIs and ACT have been found to be effective in the treatment of chronic pain [30-35], improving affect and functional ability after knee surgery [21], and quicker cessation of pain and opioid use in veterans after orthopedic surgery [36]. In summary, PP and ACT could potentially benefit patients undergoing spinal surgery in terms of physical (eg, pain) and psychological (eg, well-being) factors.

In the context of surgery, digital health interventions can be used to aid patients during their recovery process [37]. These interventions can be used to improve postoperative outcomes by supporting healthy lifestyle behavior change before and after surgery [38,39] and to improve medication adherence [40,41]. Digital health interventions can also better prepare patients for surgery or shorten postoperative recovery through behavioral modification, patient monitoring, or protocol adherence [42-45]. Behavioral modification can also be applied to promote a healthy lifestyle. This is important as healthy lifestyle behavior changes before and after orthopedic surgery, such as increased preoperative physical activity or smoking cessation, have been associated with improved postoperative bone healing [46] and wound healing [47], quicker recovery times, and reduced pain scores [48]. Moreover, patients undergoing surgery who engage with digital health interventions show better medication adherence, better adherence to discharge instructions, greater patient satisfaction, improved clinic attendance, lower readmission, and less emergency department visits after surgery [49]. van der Meij et al [50] found in their review that, in most studies, perioperative digital health interventions improved clinical patient-related outcomes compared with face-to-face perioperative care alone for patients who had undergone various forms of surgery. This shows the potential of digital health interventions to improve perioperative care in addition to face-to-face meetings with professionals.

Despite the clear potential of ACT and PPIs for mental well-being and long-term resilience, they have not yet been used to develop interventions targeting psychological flexibility in patients undergoing spinal surgery. In addition, seeing the potential of digital health interventions in the context of surgery, this form also seems promising for patients undergoing spinal surgery. For this reason, a digital health intervention called Strength Back [51] was developed. Strength Back aims to increase psychological flexibility and well-being and improve postoperative recovery in patients undergoing spinal surgery. This intervention, developed through cocreation with different stakeholders, contains procedural information, pain education, and PPI and ACT exercises.

Although the intervention is based on a scientific, theoretical framework, the effects and impact of the intervention need to be proven to be able to implement it in an evidence-based health care setting. This requires a collection of evidence, such as through a randomized controlled trial (RCT). As an RCT is complex and expensive, preparation and insights in advance into the required parameters, such as measures, recruitment, and dropout numbers, are essential to perform a thorough RCT. In light of the aforementioned research on PP and ACT in other target populations, several outcome measures covering physical and psychological factors need to be explored as potential benefits of this intervention. In addition, in an RCT, the best possible version of the intervention needs to be tested. This is in line with the recommendation of the Medical Research Council of first testing and refining an intervention to ensure it is acceptable to successfully evaluate whether it is effective [52]. In addition, to reduce psychological and practical barriers and maximize acceptability, it is important to understand and accommodate patients’ views [53]. As of yet, it is unknown how patients undergoing spinal surgery value the intervention Strength Back in terms of acceptability, potential benefits, and feasibility in a real-life setting and what the required parameters for a future RCT would be. Therefore, the objective of this study was 3-fold: to explore the potential benefits for patients undergoing spinal surgery of the digital ACT and PPI Strength Back (research question [RQ] 1), explore the feasibility of a future RCT in terms of recruitment and dropout (RQ 2), and assess the acceptability of Strength Back by patients undergoing spinal surgery (RQ 3).

This was a nonrandomized pilot study, a subset of feasibility studies as described by Eldridge et al [54]. Our study design included an intervention group and a historic control group (Figure 1). We used a historic control group as the development of the intervention was still ongoing and the target population was small. The participants in the control group received care as usual. In addition to receiving care as usual, the participants in the intervention group were given access to the digital health intervention Strength Back. To test the acceptability of the intervention, individual interviews were conducted with the participants in the intervention group. Both groups filled out a questionnaire before (pretest time point) and after (posttest time point) surgery.

As suggested by Orsmond and Cohn [55], we used a combination of qualitative and quantitative measures in this feasibility study. We aimed to include 20 participants in our control group and 20 participants in our intervention group, which is in line with the study by Billingham et al [56], who found a median sample size of 36 in their review of feasibility studies.

Patients undergoing decompression or spinal fusion surgery were eligible for inclusion in either the control or intervention group of this study. Patients undergoing spinal surgery other than decompression or spinal fusion surgery (eg, patients with hernias or patients of oncology) were excluded. Other inclusion criteria were age of >18 years, proficiency in Dutch, and an email address. Participants in the intervention group needed an Android or iOS smartphone or tablet at home and had to be willing to use a digital health intervention both before and after surgery.

All participants in this study were recruited through purposive sampling at an orthopedic center in the Netherlands. The participants in the historic control group were recruited between February 2020 and July 2020. The participants in the intervention group were recruited between September 2020 and February 2021.

For both conditions, patients received a leaflet from their personal health care professional (ie, orthopedic surgeon, specialized physical therapist, or advanced nurse practitioner) explaining this study during their visit to the orthopedic center and were subsequently contacted by phone by the researcher. When patients agreed to take part in the study, their email address was collected, after which they received an information email with a link to the first web-based questionnaire (through Qualtrics [Qualtrics International Inc]; pretest assessment). We chose this specific moment for the pretest assessment to clarify the current and most up-to-date picture of the preoperative situation. In the questionnaire, the participants first had to provide active informed consent to continue. Participants in both the control and intervention groups filled out this questionnaire (pretest assessment; Figure 1).

A total of 3 months after surgery, participants in both groups received an invitation by email to fill in the posttest questionnaire. At this time, it was expected that (nearly) all participants were in the final phase of their recovery process. We chose this specific moment to collect the data, when the recovery situation had stabilized. At this point, the participants in the intervention group were contacted by the principal investigator (AH) to make an appointment for an interview to assess how they valued the intervention. These semistructured interviews were conducted between January 2021 and June 2021 by 2 researchers (AH and LM). Owing to the COVID-19 pandemic, interviews were conducted via telephone. The interviews were audio recorded and lasted between 30 and 60 minutes.

The recruitment and dropout numbers in the intervention group were explored in this study to provide insights into what is needed for the recruitment process of a future RCT to reach the desired power.

Regarding RQ 1 (potential benefits), the questionnaires contained several outcome measures (ie, pain intensity and interference, anxiety, depression, psychological inflexibility, valued living, and mental well-being). These potential benefits were measured at the pre- and posttest time points for both the control and intervention groups.

Pain intensity was measured using the Numeric Pain Rating Scale (NRS). For the NRS, participants were asked to circle the number (0-10) that best described their experience of pain in the last week in general and in the last week at worst. The questions asked were as follows: “Please indicate the average intensity of your pain in the last week” and “Please indicate the intensity of your pain at the worst moments in the last week.” A higher score indicates a higher intensity of pain. The NRS is considered a valid, reliable, and appropriate pain rating scale for use in clinical practice, with good sensitivity [57].

Pain interference was measured using the pain interference subscale of the Multidimensional Pain Inventory (MPI) [58]. This subscale consists of 11 items measuring to what degree pain influences the daily life of a respondent, with total mean scores ranging from 0 to 6. A higher score indicates a higher degree of pain interference. The subscale showed good reliability in our study population, with a Cronbach α of .874.

Anxiety and depression were measured using the Hospital Anxiety and Depression Scale (HADS) [59]. Both the anxiety and depression subscales consist of 7 items with scores ranging from 0 to 21. A higher score indicates a higher degree of anxiety or depression. Both subscales showed good reliability, with a Cronbach α of .874 for the anxiety subscale and .789 for the depression subscale.

Psychological inflexibility was measured using the Psychological Inflexibility in Pain Scale (PIPS) [60]. The 2 subscales measure avoidance and cognitive fusion, concepts derived from the theoretical framework of ACT [24]. As the cognitive fusion subscale showed low reliability in our study population with a Cronbach α of .524, only the total score of the PIPS was used in this study. This total score had good reliability, with a Cronbach α of .893. A higher score indicates more inflexibility with pain (ie, lower acceptance and higher cognitive fusion).

Value-based living was measured using the “Valued living” subscale of the Engaged Living Scale (ELS; Trompetter et al [61]), which covers the concept of “valued living”—recognizing personal values and undertaking committed actions that are congruent with these values. The “Valued living” subscale consists of 10 items and uses a 5-point scale, with scores ranging from 10 to 50. A higher score indicates a higher degree of valued living. This subscale of the ELS showed high reliability, with a Cronbach α of .917.

Mental well-being was measured using the Mental Health Continuum–Short Form [62] on three dimensions: (1) emotional well-being, defined as positive feelings (3 items); (2) psychological well-being, described as individual functioning (eg, environmental mastery and purpose in life; 6 items); and (3) social well-being, measuring functioning in community life (eg, social contribution and social acceptance; 5 items). The mean scores per subscale range from 0 to 5, with a higher score indicating a higher degree of emotional, psychological, or social well-being. The total score (Cronbach α=.930) as well as that of the different subscales (emotional: Cronbach α=.930; psychological: Cronbach α=.874; social: Cronbach α=.799) showed good reliability, with a high Cronbach α in our study population.

Regarding RQ 2 (recruitment and dropout), we kept an overview in Microsoft Excel (Microsoft Corp) of the recruitment and dropout numbers in the intervention group in this study.

Regarding RQ 3 (acceptability), the scheme used for the interviews consisted of 2 parts. The first part of the interview started with questions on the intervention in general (eg, topics on layout and user-friendliness based on the Mobile App Rating Scale [63]). The second part continued by discussing the content of the modules (Multimedia Appendix 1) followed by the exercises in the weekly modules (Multimedia Appendix 2) as topics. Engagement with the intervention was measured using the Twente Engagement With eHealth Technologies Scale (TWEETS) [64]. This scale of 9 items comprises 3 subscales: behavioral, cognitive, and affective engagement. In this study, the total score, with a Cronbach α of .883, was used. In addition, the automated logs from the system were analyzed to see when participants opened which weekly module. The information modules were continuously available (before and after surgery), and activity in these modules was not recorded in the automated logs.

All statistical analyses were conducted using SPSS (version 26.0; IBM Corp). For the exploratory aim of this study, full cases provide the most valuable information. Therefore, only full-case analyses were conducted. In addition, as the quantitative data were used to generate a general picture of between-group differences, no imputation techniques were used. When participants filled out the pre- or posttest assessment more than once, the first completed test was used for the analysis.

For RQ 1 (potential benefits), between-group differences were assessed using nonparametric Mann-Whitney U tests because of the small sample size and nonnormal distribution of the data. First, between-group differences were analyzed at baseline using the Mann-Whitney U test. Second, within-group differences were analyzed for both groups between the pre- and posttest assessment using a Wilcoxon signed rank test. The effect sizes were calculated using the Cohen d: (mean at the posttest assessment – mean at the pretest assessment) / (SD). When the SD of both moments differed, the mean SD of the pre- and posttest assessments was calculated and used. Values for the Cohen d of 0.2 were considered a small effect size, 0.5 was considered a medium effect size, and values of ≥0.8 were considered a large effect size.

Third, between-group differences were analyzed by comparing the pre- and posttest difference scores of both groups (Mann-Whitney U test). All statistical tests were 2 sided. P values of <.05 were considered statistically significant. As this was a feasibility pilot study, the results were considered exploratory, and no correction for multiple testing was performed.

For RQ 2 (recruitment and dropout), we analyzed the recruitment and dropout numbers of the intervention group in this study. For the analysis, we looked at the number of approached patients, the number of patients who consented to participate in the study, and the time span in which all these patients were approached. Dropout was defined as agreeing to participate in the study but not downloading the intervention, not filling out the pretest questionnaire, or not filling out the posttest questionnaire.

For RQ 3 (acceptability), interview data were collected, and the audio recordings were transcribed verbatim. For the analysis, the transcripts of all interviews were read and reread by 2 researchers (AH and LM) to familiarize themselves with the data. Subsequently, the transcripts were open coded by both researchers independently, followed by a discussion to reach a consensus regarding the coding. As a next step, both researchers sorted the codes into several groups. A deductive coding approach was used based on the topics on the interview scheme combined with an inductive approach for important content within these topics. For the coding and analysis process, the ATLAS.ti software (version 9.0; ATLAS.ti Scientific Software Development GmbH) was used.

Ethics approval was granted by the Medical Ethical Committee of the Radboud University Medical Centre in Nijmegen (2019-5608) and the Ethics Committee of the Faculty of Behavioural, Management, and Social Sciences of the University of Twente (191080). The study is registered at the repository of the University of Twente for processing of personal data (AVG Register; WBP18ME0048).

Participants in both the control group and the intervention group filled out a questionnaire before and after surgery. When patients agreed to take part in the study, they received a link to the first questionnaire, in which they had to provide active informed consent to continue.

All human participant data were deidentified before analysis. Study participants received no compensation for their inclusion in the study.

Figure 3 shows an overview of the inclusion of participants in the intervention and control groups. A total of 25 patients agreed to participate in the intervention group, of whom 3 (12%) participants did not fill out the pretest questionnaire; the reason for this is unknown. Therefore, a total of 22 participants filled out the pretest questionnaire. After filling out the pretest questionnaire, 9% (2/22) of the participants failed to download the intervention to their phones. Therefore, 20 participants started using the intervention (20/35, 57% of the approached patients). Before the posttest questionnaire, 15% (3/20) of the patients discontinued their participation. This resulted in a total of 17 participants filling out the pre- and posttest questionnaires. All these participants (17/17, 100%) also took part in a telephone interview.

A total of 31 patients were approached by phone to participate in the control group. This resulted in 71% (22/31) of these participants completing the pretest questionnaire. A total of 9% (2/22) of these participants did not complete the posttest questionnaire without giving a reason. In total, 9% (2/22) of the participants started the posttest questionnaire but only partially completed it. This resulted in a control group of 20 participants, of whom 18 (90%) completed the entire pre- and posttest questionnaires.

The participant characteristics at baseline are presented in Table 1. The Mann-Whitney U test revealed no substantial differences between the intervention and control groups in terms of these characteristics.

First, between-group differences were analyzed at baseline (Mann-Whitney U test). The total score on the PIPS was significantly higher in the control group (U=262.5; P=.004). Other measures did not differ significantly between both groups at baseline.

Second, within-group differences were analyzed for both groups between the pre- and posttest assessments (Wilcoxon singed rank test; see Table 2 for the results). When looking at the differences between the pre- and posttest assessment, it should be taken into account that both groups had undergone surgery, which is an intervention by itself. Both groups showed significant improvement in the NRS (pain intensity week average and week worst), HADS anxiety subscale, HADS depression subscale, pain interference subscale of the MPI, and total score on the PIPS between the pre- and posttest time points. The score on the valued living scale of the ELS did not significantly change over time in either of the 2 groups. In the control group, the subscores as well as the total score on the Mental Health Continuum–Short Form did not differ significantly between the pre- and posttest time points. In the intervention group, the total score (V=99; P=.03 as well as the score on emotional well-being (V=55; P=.004) increased significantly over time.

Third, between-group differences were analyzed by comparing the pre- and posttest difference scores of both groups (Mann-Whitney U test). A significant difference was found in the weekly average NRS score between the intervention and control groups (U=75; P=.003), showing a significantly larger decline in pain intensity in the intervention group.

To determine the feasibility of a future RCT, recruitment, dropout rates, and dropout characteristics were examined. Over a period of 5 months, a total of 35 patients were approached by phone to participate in the intervention group (Figure 3). In total, 51% (18/35) of the approached patients did not want to participate and did not start with or dropped out of the intervention. Of these 18 patients, 14 (78%) were female and 14 (78%) underwent decompression surgery. Most of the dropouts did so before the intervention started as they did not want to participate in the study (10/18, 56%, of whom 9/10, 90% were female; age range 19-79 years). Most of the additional 44% (8/18) of dropouts did so at the pretest (3/8, 38%) or posttest (3/8, 38%) time points, and only 25% (2/8) of these participants dropped out because of nonadherence as they did fill out the pretest questionnaire but did not download the intervention. Of these 2 nonadherers, 1 (50%) was male and 1 (50%) was female; both underwent decompression surgery, and both were aged >65 years.

The objective of this study was 3-fold: to explore the potential benefits for patients undergoing spinal surgery of the digital ACT and PPI Strength Back (RQ 1), explore the feasibility of a future RCT in terms of recruitment and dropout (RQ 2), and assess the acceptability of Strength Back by patients undergoing spinal surgery (RQ 3).

The focus of this study was to explore the potential benefits of the intervention and not to determine its effectiveness. The latter should be the focus of a future RCT including more participants and allowing for more robust statements. Nonetheless, this study does show that Strength Back seems promising as pain intensity decreased more in the intervention group than in the control group and emotional well-being as well as overall well-being improved in the intervention group but not in the control group (RQ 1). This is in line with research in patients with chronic pain showing beneficial effects of ACT and PPIs compared with a control group on pain intensity and emotional functioning [65,66], in the treatment of chronic pain [30-35], in improving affect and functional ability after knee surgery [21], and in quicker cessation of pain and opioid use in veterans after orthopedic surgery [36].

Contrary to our expectations, no differences between the groups were observed in improvement in pain interference and psychological flexibility. Possibly, the effect of the ACT and PPI content shows through in the significantly larger decline in pain intensity in the intervention group. It is reasonable to suppose that a decline in pain intensity also helps prevent pain interference with daily activities and improve psychological flexibility. However, the sample size may have been too small, or the effect of surgery may have been so massive that these effects did not appear in this study. Moreover, in the intervention group as well as the control group, valued living and psychological and social well-being were almost the same at the pre- and posttest time points. The amount of ACT and PPI content may have been too small to influence valued living or psychological and social well-being. It is also possible that these effects only occur in the longer term. Living more accordingly to one’s values, resulting in enhanced psychological and social well-being, may only start later in the recovery process and not immediately after surgery.

In terms of recruitment and dropout rates, a future RCT seems feasible (RQ 2). A large proportion of the approached patients wanted to participate in the study, and once they started with the intervention, almost all participants used it until the posttest assessment and final interview. The willingness to use a digital health intervention seemed quite high in our study compared with in previous studies [67,68]. It seems that, once participants started the intervention in this study, they saw its value and kept using it over a longer period. In line with other research [69], patients stated that receiving notifications improved their use of the intervention. The dropout numbers found in this study were low and correspond to those of a recent RCT on PP exercises for patients with chronic pain [66]. However, to recruit a larger number of participants for an RCT, multiple hospitals or orthopedic centers should be included as there are a limited number of potential participants in a single facility. Interestingly, almost all the patients who were approached and did not want to participate in our study were female. Perhaps this group needs more attention during recruitment to explore their reasons for not wanting to participate.

The vast majority of participants in the intervention group were positive about the digital health intervention and would recommend it to future patients (RQ 3). The scores on the TWEETS indicated a moderate to above-moderate engagement with the intervention, which was in line with log data showing that participants completed on average approximately 75% of the available weekly modules. The information modules containing videos of the surgery room and nursing ward, practical tips from previous patients, physical activity guidelines, and pain medication were the most appreciated by participants. The PP and ACT content evoked the strongest opinions in the participants, with a small minority indicating that they preferred no psychological content at all, whereas most of the participants saw added value in a number of specific exercises. Although the development of the intervention was based on a participatory design process and most participants appreciated the content, others felt that the focus was too much on mental health, whereas they experienced their issues as physical. At the same time, to increase the effectiveness of the intervention on the ACT- and PP-related outcomes, even more psychoeducation and exercises might be needed. In addition, the current version of Strength Back focuses on certain elements of ACT, whereas it may be necessary to address all ACT processes to achieve optimal benefits for patients undergoing spinal surgery. Indeed, Carr et al [29] found in their meta-analysis of PPIs that interventions were more effective when they contained multiple PPIs, were of longer duration, and contained more sessions. This poses a dilemma for the further development of this intervention. Information on the psychological content of the intervention before inclusion and tailoring the content and dose of the intervention to individual patients might further improve acceptance of the PP exercises, as suggested by previous research [70-72]. In a future version of Strength Back, patients could be introduced with a few of the most appreciated ACT and PP exercises and from there on be given the possibility to determine the amount of ACT and PP content themselves. This might reduce the potential effect on some patients who opt for a lower amount of ACT and PP content but might increase overall adherence to and acceptance of the intervention.

Our results question what the most appropriate primary outcome is for a subsequent RCT. As the intervention is based on ACT and PP, it makes sense to choose a measure corresponding to this content, such as well-being or pain interference. However, our feasibility study shows that the intervention and control groups particularly differed in improvement in pain intensity, which is an outcome that is important to patients but not directly targeted within the intervention. Perhaps targeting both pain intensity and increasing well-being might be the best option for sustainable resilience in patients undergoing spinal surgery. This is in line with previous research proposing a balanced, complaint and strength–oriented approach to reach sustainable mental health [73].

Second, attention should be paid to what the appropriate process measurements in a subsequent RCT would have to be. On the basis of the content and assumed working mechanism of this intervention, we propose psychological flexibility as a process measure in a future RCT. Nonetheless, we would not advise using the PIPS to measure this concept as we found low reliability in our study. Other studies have found similar issues with this scale and have recommended only interpreting it as a whole [74] or only using the avoidance subscale [75]. In addition, the PIPS measures inflexibility and not flexibility, which does not fit as well with the focus of ACT and PP. Perhaps the Personalized Psychological Flexibility Index [76] is a suitable alternative measure of psychological flexibility for a future RCT. In addition, we propose pain interference as a process measure in a future RCT, measured using the MPI subscale of pain interference. In addition, the ELS might be used in a future RCT to measure engaged living, including valued living.

The final discussion point relates to the length and timing of the intervention. The current intervention starts before surgery and lasts 6 to 12 weeks after surgery (for decompression and spinal fusion surgery, respectively). This was regarded positively by most participants, and from both the interviews and log data, we learned that most participants were already active in the intervention before surgery. This focus on pre- and postoperative timing is supported by a recent systematic review on perioperative psychological interventions that found that interventions (also) delivered after surgery tended to be more effective for postsurgical pain and disability than interventions delivered (only) before surgery [77]. It seems that, while waiting for surgery, patients are quite eager to learn more about the surgery and rehabilitation afterward. An intervention such as Strength Back can take advantage of these “teachable moments” [39] by providing content that participants may not have found or chosen themselves but that is known to help them in the recovery process. The duration of our intervention is in line with a recent review on perioperative psychological interventions for patients undergoing spinal fusion showing a reduction in pain and disability starting from immediately after surgery up to 3 months [78].

The high recruitment and relatively low dropout rates were a strength of this study. In addition, qualitative data were collected through individual interviews with almost all the users of the intervention. This yielded crucial information on the value of the intervention and the reasons behind participants’ use of the intervention. Using multiple methods combining qualitative data with log data and quantitative data retrieved through the questionnaires provided a complete picture of the value of the intervention. Therefore, this research design is close to a convergent parallel mixed methods design [79]. Another strength of this study is the underlying theoretical framework of the intervention. As this is lacking for many digital health interventions, there are calls for further research to enhance the scope and use of this technology [80,81]. The design of digital interventions should be anchored in behavior change theories to optimally engage patients in the intervention and behavior change [82,83]. In a review of 85 studies using the Behavior Change Support Systems by Oinas-Kukkonen and Harjumaa [82], less than half referred to theories of behavior change, but those that did were uniformly successful [84]. Clearly, a sound theoretical base for digital health interventions is warranted for optimal behavior change and effectiveness.

As all participants underwent surgery, which is an intervention by itself, we wanted to see in this study what the digital health intervention might add to the effect of the surgery. An exploratory design regarding outcome measures was used to see which potential benefits could be achieved for patients undergoing spinal surgery. Owing to this exploratory design and the small sample size, more and larger effects of the intervention were not expected to be found in this study. In addition, no correction for multiple testing was performed in this study, and thus, conclusions can only be drawn with caution. Therefore, the preliminary findings of this study need to be replicated in a future larger study.

This study used a historic control group. A future RCT should use a design in which the intervention group and the control group run parallel in time and participants are assigned to them randomly.

In this study, log data were gathered on modules in which participants answered questions. For a future study, we would recommend registering every log-in and activity of users instead of only activities in which participants provide answers. We now see certain patterns in the use of participants, but more data are needed to further investigate use of and engagement with the intervention.

Previous hospital experience, preexisting comorbidities, patient technology readiness, health literacy, and ward factors such as staffing were not measured in this study. As these factors may influence how prepared patients are and what their capability is to engage with the intervention, they should be considered in a future RCT.

This study shows that combining ACT and PP in a digital health intervention is promising for patients undergoing spinal surgery as the content was accepted by most of the participants and (larger) improvements in pain intensity and well-being were found in the intervention group. A digital intervention for patients undergoing (spinal) surgery can use teachable moments, when patients are open to learning more about the surgery and rehabilitation afterward, to also provide patients with content that they may not have found or chosen themselves but that is known to help them in the recovery process.