Impact of intraindividual pain variability on functional pain outcomes among adults with chronic pain: an ecological momentary assessment study

Chronic pain, defined as pain persisting for at least 3 months, is a major public health problem impacting over 20% of adults (Rikard et al., 2023), and leads to significant economic burden (i.e. medical expenditures, missed work days) and functional impairment (Cohen et al., 2021; Lohan et al., 2023; Turk & Patel, 2022; Yong et al., 2022). Further, chronic pain has been linked to the initiation and persistence of the opioid epidemic (Pielech et al., 2020; Witkiewitz & Vowles, 2023). Chronic pain treatments, including pharmacological and non-pharmacological treatments, yield small to medium sized effects for pain complaints, and interestingly, often show stronger effects on factors comorbid with pain, such as depression and anxiety (Hoffman et al., 2007; Malone & Strube, 1988; Niknejad et al., 2018; Shetty et al., 2024; Veehof et al., 2011). Importantly, the majority of clinical trials for chronic pain utilize single-item and/or retrospective reports of average pain level to assess efficacy and effectiveness, which provides a limited and potentially biased understanding of pain experience during treatment (Stone et al., 2010; Winger et al., 2019). There is a need to pursue a more nuanced understanding of pain experience among adults with chronic pain to improve clinical interventions for this population.

Within- and between-day changes in pain intensity (intraindividual pain variability) represent an important avenue to improve our understanding of pain experience (Mun et al., 2019). Theoretical work has highlighted the importance of utilizing intensive longitudinal assessment (e.g., ecological momentary assessment (EMA)) to calculate unique metrics that capture the momentary nature of pain experience above and beyond average pain level (Stone et al., 2021a, 2021b); these include the percentage of time spent in high and low pain, and how much pain tends to fluctuate for a given person, differentiating between dynamic aspects of pain and pain instability (Schneider et al., 2021; Stone et al., 2021a, 2021b). Thus far, several important findings have emerged in empirical work examining these pain variability metrics. For example, in one study, patients with moderate to high intra-individual pain variability (compared to low variability) reported poorer coping and maladaptive responses to pain among adults with chronic low back pain (Wesolowicz et al., 2021). In another investigation among youth with sickle cell disease, higher pain variability was associated with poorer mental health symptoms, more emergency department visits, and a greater number of healthcare visits (Pascale et al., 2023). Additionally, among adults with chronic pain, variability metrics were associated with psychosocial and functional outcomes over and above the variance accounted for by average pain level (Schneider et al., 2021). Further, some work suggests that higher baseline variability is associated with a greater pain treatment response (Harris et al., 2005; Pinto et al., 2021), whereas other work shows that variability is not related to treatment outcome (Gillving et al., 2022; Tiwari et al., 2022).

Despite the varied findings noted in past work, which may be a product of distinct methodological features of those investigations, the corpus of past work highlights that conceptualizing pain as a ‘dynamic experience’ represents an important avenue for further scientific inquiry (Mun et al., 2019). There are several gaps in the current literature. First, limited research has examined whether the range of pain variability metrics impacts pain-related functional outcomes (e.g., disability, activity avoidance). This information is important because it may suggest that conceptualizing pain as a dynamic experience is a clinically relevant factor for functional behavior and pain-related quality of life.

The current study utilized EMA to capture intraindividual variation in pain over 7 days among adults with chronic pain. We explored the impact of established pain variability metrics on pain-related disability, pain-related negative affect, and pain-related activity avoidance. It was hypothesized that greater pain variability would be associated with greater pain-related disability, negative affect, and activity avoidance.

Considering pain as a dynamic process is an important, yet relatively under-explored, aspect of pain that can provide insight into nature of pain and its comorbidities (de Koning et al., 2018; Han et al., 2024; Mun et al., 2019; Tupper et al., 2013). The current study examined the impact of intraindividual pain variation metrics on pain-related disability, pain-related negative affect, and pain-related activity avoidance.

Data for the primary analysis was largely in line with expectations. Specifically, mean pain, highest pain, lowest pain, and percent of time in high and low pain were associated with pain-related disability and pain-related negative affect (dynamic metrics), whereas intraindividual standard deviation, mean square of successive differences were associated with activity avoidance (instability metrics); these findings suggest that there may be an important conceptual difference between pain instability and variability generally. The Fear-Avoidance model of chronic pain (Vlaeyen & Linton, 2012) suggests that activity avoidance is associated with greater chronic pain symptoms and pain-related disability. The current data contribute to this model by suggesting variability in pain experience, or what clinically could be referred to as pain unpredictability (Pavy et al., 2024a), is related to constructs often associated with pain avoidance (i.e., pain-related disability and negative affect; Meulders, 2019; Wideman et al., 2013; Zale & Ditre, 2015). Future research could build on our findings by exploring pain variability in terms of pain-related avoidance. Instability, as measured by multiple metrics, was consistent in its association with pain-related activity avoidance, highlighting that the unpredictable nature of pain may be the most clinically salient when examining activity avoidance (Yoshida et al., 2013).

The results of this study should be considered preliminary in need of replication but, when coupled with recent theoretical and empirical work (Mun et al., 2019; Schneider et al., 2021; Stone et al., 2021a, 2021b), fill an important gap in the literature. Relying on a retrospective, average pain score alone does not provide an adequate characterization of pain and the way it may impact those with chronic pain (Ballantyne & Sullivan, 2015). The present findings are consistent with previous work (Gillving et al., 2022; Harris et al., 2005; Tupper et al., 2013) in suggesting that intraindividual pain variability is important to numerous aspects of pain experience. Measuring and addressing pain variability may be important to treatment success, by specifically targeting pain predictability and expectancies that have clinically meaningful associations with functional outcomes (Pavy et al., 2024b). Here, it may be useful for adults with chronic pain to complete a daily diary of their pain symptoms prior to initiating treatment to understand their intraindividual pain variation and its impact on pain and functioning (work, social), and leverage this to provide treatments more tailored to the individual. For example, persons with higher levels of pain instability (and thus at risk for pain-related activity avoidance) may benefit directly from exposure (Macedo et al., 2010), whereas those who spend a large percentage of time in high pain may benefit from a more cognitive approach (Cheng & Cheng, 2019; Otis, 2007).

The current study is not without limitations. First, while everyone in the sample identified experiencing chronic pain, the sample was not treatment-seeking, did not undergo a comprehensive pain diagnostic evaluation, and are limited in variability in demographic and cultural factors, which may not be reflective of all persons with chronic pain. Similarly, demographic assessment of functional impairment (e.g., work performance) was not specifically anchored to pain, and therefore it was unclear the degree to which pain impacted work/school functioning. Therefore, replication of the current findings on different samples of individuals with chronic pain, with more specific assessment of functional impairment across multiple domains is warranted. Second, the measure of pain-related negative affect assessed multiple emotions together, in one item, limiting understanding of how specific emotions may be related to variability metrics. It may be possible that, if parsed out into individual emotions in a future study, the impact of variability may have been different and is worth exploring in future research given the documented associated with negative emotions and pain (Lumley et al., 2011).

Third, it is important to discuss potential limitations of variability calculations. Previous research indicates that estimates of pain variability are influenced, in part, by mean pain level (Mun et al., 2019; Schneider et al., 2021), but in the current study, the correlations between mean pain level and variability metrics would have likely resulted in issues of multi-collinearity. Future research, replicating the current findings and utilizing alternative statistical approaches to account for both mean level and variability metrics (e.g., mixed effects location scale models; Dzubur et al., 2020) would be an important next step. Additionally, while the current study utilized established cut-points for “high” and “low” pain, there are inherent limitations to this approach that does not account for individual differences in pain experience (e.g., ceiling effects, reporting bias) that may influence results. Furthermore, while not testable in the current study, it is possible that a sequential model exists that incorporates aspects of pain variability as well as pain-related activity avoidance, disability, and negative affect, which requires significant within-day variability and a study designed to capture time-lagged relationships The pain processing model (Riley et al., 2000) is one empirical model that could be tested in future work to disentangle the sequential order of these relationships. Finally, the current study assessed individuals over 7 days and focused on variability metrics in pain intensity. While there is a large amount of data that was captured, it is possible that patterns of variability (and their associations with pain outcomes, both within- and between-person and day) may differ when assessing pain over a longer period, or when assessing variability in functional impairment over time. Future research should examine the optimal number and type (e.g., intensity vs. interference) of assessments of pain needed to both balance patient burden as well as providing the necessary information for clinical care.

Overall, the current study provides empirical support for the importance of pain variability in conceptualizing pain experience and factors that lead to pain-related functional impairment. Moving towards understanding pain as a dynamic experience has the potential to improve our treatments and reduce the negative impact of chronic pain.