Background
Annually, 15–20% of all adults attending general practice present with musculoskeletal conditions [
1]. Around 40% of these patients are prescribed analgesia during their first consultation for musculoskeletal pain, half of whom will receive a non-steroidal anti-inflammatory drug (NSAID), and 29% a moderate to strong opioid containing analgesic [
2,
3]. The general practitioner’s (GP’s) intention is to relieve their patient’s pain and the decision to use analgesia, for example in low back pain, is often based on the patient’s verbal report of their pain and their personal analgesic preferences [
4]. Any subsequent review of analgesic efficacy is similarly limited by the GP relying on the patient’s reports of their pain progress, often quantified by asking patients to rate their pain on a scale from 0 (no pain) to 10 (worst pain) [
5]. However, this existing approach to pain monitoring fails to account for the multiple dimensions of pain experience in relation to its onset (acute, or gradually developed over time), course over time (stable or highly variable), severity, and impact on everyday life [
6]. Obtaining more detailed pain experience information could be an important and useful tool for clinical practice. For example, symptom trajectories have been shown to help obtain an accurate diagnosis in headache [
7]; monitor the severity/impact of symptoms in asthma or abnormal vaginal bleeding [
8,
9]; or assess short-term responses to treatment [
10,
11]. Furthermore, information on short-term symptom trajectories has the potential to provide important prognostic information, and is likely to result in better long-term prediction of health outcomes [
12,
13].
Despite a wide variation in musculoskeletal pain trajectories over time, early changes in painful symptoms after prescribing an analgesic (such as an opioid) are not routinely collected as part of musculoskeletal follow-up assessments, which often take place several weeks or even months later [
14‐
16]. Monitoring of these changes in response to opioids is very important, and now underpins current guidelines [
17], especially since evidence for the long-term effectiveness of opioids is lacking [
18], and there is an increasing recognition of the harms related to opioid use [
19,
20]. Although many studies have reported on the long-term (6–12 months) outcomes of musculoskeletal pain [
12,
15], and investigated long-term trajectories using repeated pain assessment [
15,
21‐
23], little is known about short-term pain trajectories following primary care consultations, how they relate to long-term outcomes, and in what way they might be used to support the primary care management of musculoskeletal pain.
When pain trajectories are measured, data have often been collected using paper diaries, which are cumbersome, have low completion rates, and may be completed retrospectively resulting in inaccurate and potentially biased data [
24‐
26]. In recent years alternative methods for daily data collection have been proposed, including the use of text messaging [
27‐
29], palm top computers [
30], or Smartphone technology [
31]. These approaches are gaining popularity given the increasing use of Smartphones and particularly now with two thirds of British adults owning one [
32]. One systematic review identified 55 articles reporting the design, evaluation, or use of smartphone-based software for healthcare professionals, students, or patients. The authors highlight the increasing use of Smartphone technology in healthcare and their potential role in patient education, disease self-management, and remote monitoring of patients [
33]. Limitations of Smartphone technology have also been reported, most importantly the lack of personalised feedback, usability issues (e.g. ease of data entry), and poor integration of Smartphone data with electronic health records [
34,
35]. A systematic review of currently downloadable pain monitoring apps has highlighted the lack of scientific rigour used to ensure validity and reliability with respect to pain measurement [
36]. However, where attention to validation has been robust, moderate to high reliability and validity has been reported [
37,
38]. Therefore, in this feasibility study, our aim was to:
1.
develop a Smartphone Application (“Keele Pain Recorder”) for use by patients with painful musculoskeletal conditions to record daily information on their pain severity and the impact of pain on daily life.
2.
assess the acceptability and clinical utility of the Pain Recorder in terms of completion rates, feasibility of its use, and its influence on GP decision-making. Even if an app is found to have a high level of validity in collecting data, its clinical usefulness is only as good as its level of acceptability to the user in day to day use [
39]. It is therefore, as part of any app development, essential to examine how acceptable and useful to the user it is.
3.
assess face and content validity and explore construct validity of data collection using the Keele Pain Recorder in musculoskeletal patients presenting to primary care receiving new analgesic prescriptions. We hypothesised that single Pain Recorder items were highly correlated with validated questionnaires measuring the same domain of interest i.e. whether there was a strong correlation between the Keele Pain Recorder scores and questionnaire scores at baseline (day 1–3 for the Pain Recorder) and follow-up (last 3 days for the pain app), and between changes over time in scores from the app and questionnaires (longitudinal validity).
4.
assess the daily changes in pain and other symptoms (short-term pain trajectories) for patients presenting with a new episode of MSK pain during the first 4 weeks use of the Keele Pain Recorder. This is an area of data collection that is important since information on short-term symptom trajectories might help in establishing the possible cause of symptoms (diagnosis) [
7]; estimate the future course of a condition (prognosis) [
11,
40]; monitor the severity of symptoms in patients with chronic conditions [
9]; and assess early response to treatment (intervention) [
10,
11]. However, early changes in pain and other symptoms are often not assessed until several weeks or even months after the first consultation or start of treatment [
14‐
16], and development of the Keele Pain Recorder (KPR) offers a clear opportunity to examine these short-term trajectories in relation to these areas of clinical assessment.
Discussion
In close collaboration with patient representatives and clinical stakeholders, we developed a smartphone/tablet app, which can be used to help clinicians and patients monitor painful musculoskeletal conditions in response to analgesic prescribing. Early testing in a small sample of people consulting with musculoskeletal pain in general practice showed promising results in terms of face and content validity, acceptability, and clinical usefulness.
Despite there being more than 270 pain apps available to download [
50], there have been few studies that have validated pain monitoring apps. Often existing apps are designed by software engineers and there is little input from patients and clinicians in the design and evaluation of these apps [
50,
51]. A recent review and Editorial examining pain app design and use highlighted this fact, particularly the finding that apps sometimes appear to offer solutions to pain management with little thought given to the content of the app being validly related to clinical factors [
31,
52]. This same review concluded that the development of apps needs to be evidence-based with rigorous evaluation of outcomes being important in enhancing the understanding of the potential of these apps [
52]. Our study directly addresses this issue in that the design of the Keele Pain Recorder was driven by both patient and clinician experience and advice, whilst its clinical usefulness, acceptability and validity was assessed by several primary care clinicians in conjunction with the app users. One recent study has offered a more vigorous examination of a pain monitoring App. Suso-Ribera found that weekly monitoring of pain over a 30-day period found similarly high levels of compliance, acceptability, ease of use and construct validity to the KPR [
53]. Jamison also found that a pain monitoring app was acceptable to patients and easily utilised [
54]. This is encouraging as it suggests that pain monitoring apps in varying forms are devices that patients will be willing to engage with when managing their pain.
A series of iterative workshops and interviews with patients, clinicians and musculoskeletal researchers were used to (i) discuss content and functionality of the app, (ii) further improve the design of the app, and (iii) explore opinions regarding acceptability, validity, and usefulness. Patients confirmed that they felt discussing recordings from the app helped with their GP’s understanding of their pain condition, whilst GPs considered it useful in helping their patients make choices about medication. Patients found the Keele Pain Recorder easy to use, and the GP found the graphical output easy to interpret. As in a study from the USA, both groups found using a pain recorder app acceptable in clinical practice [
51].
The individual pain trajectories of patients varied widely, reflecting wide differences in the impact and experience of pain, even over the course of only 1 month after consulting in primary care. However, the trajectories were determined through a visual analysis and a consensus exercise amongst experienced clinicians rather than using statistical methods. The number of trajectories available to examine were too few and precluded this. Therefore, it is possible that the trajectories determined were subject to individual bias, however, agreement in each case was by majority, and each participant had extensive experience in the management of musculoskeletal pain. There may be value in assessing early symptom trajectories in people with musculoskeletal conditions, though further research is needed before any potential clinical usefulness can be established. Many prognostic models developed in studies of low back pain have been shown to have limited predictive performance [
55], although some tools, such as the Start Back Screening Tool, have been extensively tested and are now also available as a smartphone app. Although its predictive performance has been confirmed in several populations, one study showed that it was no better than clinical acumen in predicting low back pain outcomes [
21]. One reason for limited predictive performance of prognostic tools may be that most are based on only a single assessment of pain. In our study, the Keele Pain Recorder app demonstrated three main short-term pain patterns: improving, fluctuating or worsening, which reflect those previously reported in patients with low back pain [
6,
22,
23]. Potentially the use of a smartphone/tablet pain app might allow for more frequent and detailed characterisation of pain trajectories shortly after healthcare consultation. This, therefore, might be used in the future to help develop more accurate predictive models and early identification of patients likely to do well (preventing unnecessary treatment) versus those who may benefit from early, more intensive treatment [
15].
An important limitation is the small sample size of the feasibility study, which limits generalisability and precision of our findings regarding construct validity. Though the sample was small, we found a good correlation between established and validated pain measures in the baseline and follow-up questionnaires. However, there was no significant correlation with interference, mood and sleep. It is possible that in all these 3 domains (mood, interface and sleep), this might have occurred because we averaged results from the app over the first and last 3 days of the study. Therefore, due to the potential variability in these factors recorded in the app over this time, they may not reflect those recordings in the baseline and follow-up questionnaire. Small numbers in the study will also have limited our power to detect any correlation. Further testing of the Pain Recorder in larger groups of patients is needed to more formally and quantitatively investigate construct validity of the app, and to establish its clinical utility for monitoring pain by investigating impact of its use on clinical decision making and patient outcomes. Additionally, we only tested the app amongst patients with musculoskeletal pain, so the generalisability of its use in other conditions such as headache or pelvic pain cannot be assured. Similarly, our study focused on monitoring pain following prescription of a stronger class of analgesics, but the app could also be used to study pain trajectories following other types of treatments for pain.
Assessing acceptability and usability of the KPR presents problems in common with other smartphone apps [
56]. Both factors are interdependent, and it is likely that testing these elements 1 week after starting using the app would provide different results to our assessment which was at 1 month of use. There may be issues for users when they first use the software due them being unfamiliar with it, and therefore limit ease of use. Consequently, acceptance of the app might be diminished. However, after a month’s use, familiarity with the mechanics of the software might improve usability for the patients. Equally though, with loss of novelty, the user might lose interest in it and so its acceptance as a daily activity might be lost. Future research might overcome these issues through testing at both points in time to give a more comprehensive view of an apps acceptability and usability.
Low completion rates were a limiting factor in this study. This might have been compounded by the intrusion of the technology into daily life with it being perceived as an interference in the user’s normal routine. There are ways in which this might be overcome, for example gamification has been shown to improve engagement and retention in app use [
57]. Equally if the app had been used on the patient’s own mobile smartphone, the more immediate access to this device (rather than a tablet kept elsewhere) might have improved completion rates. An additional limitation relates to the comparison of the baseline and follow-up questionnaire pain scores with those recorded in the app to determine how valid these were. The most valid figure would have been to equate the single baseline/follow-up figure with the first and last day score in the app giving a direct contemporaneous comparison. However, we chose to use the initial and final 3-day average of the study. This might have led to errors in the comparison due to the potential variability of the patient’s pain during that period when compared to the single recording at baseline and follow-up. However, due the possibility that we might recruit low total numbers to this novel research, we chose to use the 3-day average which would potentially give at least 1 record during the 3 days. If we had only used the 1st or last day alone, there might have been more missing values.
Two recordings could not be classified due to a lack of consecutive data (> 7 days). This limitation was not overcome by using reminder on the tablet and is likely to be due to external circumstances, or interference of the technology in the participants’ everyday life, which was indicated to occur sometimes according to 6 of 18 participants in the feasibility study. This may have been compounded by the fact that the users did not have direct access to their pain graphs, which might have acted in a positive way to reinforce use of the app. Concerns have been expressed regarding the potential negative impact of frequent pain reporting on physical health and work productivity [
58]. When asked specifically, participants reported that they felt using the app had not directly influenced their thoughts, feelings or actions related to mood, pain interference, or medication usage. However, further research should investigate to what extent the use of the Keele Pain Recorder is associated with consultation rates, healthcare resource use, and changes in physical or mental health.
We developed secure methods for archiving, downloading and emailing pain trajectories from the Pain Recorder to the GP and patient to be used in their consultation. These methods will now be extended to allow open-access to the Pain Recorder and support use of the app on both Android and Apple phones or tablets [
59,
60]. Future research, however, needs to examine how such data may be accessed in a ‘live’ format such that GPs or other health care professionals may use information regarding pain trajectories to manage a patient’s condition when it deteriorates, for example during an acute attack of gout or a flare of knee osteoarthritis. Research may also focus on the potential of using the Keele Pain Recorder in self-management, such that software might independently recognise when a patient is at risk of developing disabling pain, offering feedback and advice to the patient without the input of a third party such as the GP. However, these devices will require rigorous assessment to ensure the advice is safe, relevant, and does not miss the possibility of ‘red-flag’ conditions such as cancer pain or other conditions (e.g. inflammation) that need medical attention.