Exploring the outcome “days with bothersome pain” and its association with pain intensity, disability, and quality of life

1 Background

Pain is a complex experience, essentially subjective, and influenced by biological, psychological, and social factors [1], in a complex interplay that may be difficult to untangle by both the patient and the assessor (clinician or researcher).

The most appropriate outcomes for patients with musculoskeletal pain have been debated within the clinical and research communities, including severity, function, and quality of life [2,3,4,5]. Simple and intuitive clinical questions (e.g., “How is pain impacting your life?”) have been suggested alongside extensive questionnaires to really capture patients’ experience of their pain [6]. Quantitative measures should feel relevant to the patient and provide useful information to the clinician regarding patient progress. In addition, researchers prefer a measure by which change can be quantified.

Pain severity is a construct that has been suggested to capture both the intensity and the consequences to the patient; i.e., how their life is affected by the pain experience [7]? Studies have shown that pain intensity is important to patients [8,9,10], but does not cause a patient to seek care. Rather, care-seeking is prompted when pain imposes functional limitations on their everyday life [11]. Therefore, a research group suggested that a measure of “bothersomeness” could be used as a relevant outcome both to the patient, the clinician, and the researcher [12]. This was recommended as part of a set of core outcomes for low back pain (LBP) and believed to capture pain that is relevant to the patient [13].

Since then, the concept of bothersomeness has been introduced to capture both the functional and intensity aspects of pain in the clinical context and has been used in several studies [14,15,16,17,18]. Therefore, we may conclude that patients seem to be able to answer questions concerning bothersomeness (“On a scale from 0 to 10, how bothersome is your pain?” or “How many days this week has your pain been bothersome?”). Thus, one may conclude that face validity has been explored to some degree. However, to our knowledge, bothersome pain has not been investigated against other measures, and it is not clear how the question is perceived or what dimensions of the pain experience it captures. Furthermore, it is not known if this measure could be used across pain conditions or populations. It was, therefore, necessary to explore this measure further to understand the correlation with other pain outcomes, so that it is known what constructs are covered and to what degree, to know if it is useful for clinicians and researchers.

The idea was that “bothersome pain” captures the pain that is relevant to the patient, i.e., a construct that captures both the intensity and the consequences of the pain. Even though this study investigated the number of days with such pain, a high rating was likely to reflect pain severity. The objective of this study was to investigate the criterion validity, specifically the concurrent validity of the measure of days with bothersome pain, by investigating its association with measures of pain intensity, disability, and quality of life.

2 Method

These were secondary analyses of data from two longitudinal cohort studies. In these studies, data were collected from patients with pain using frequent text messages in Norway and Sweden.

In study 1, LBP in patients consulting for chiropractic care in clinics all over Sweden, between 2012 and 2016, who were randomized in a clinical trial investigating preventive care, was examined [15]. A total of 321 patients were included and followed with 1 weekly SMS for 52 weeks, the duration of the follow-up. When the study ended at 52 weeks, 302 participants with a full dataset remained after dropouts.

In study 2, 503 pregnant women with pelvic girdle pain (PGP) were recruited in a prospective longitudinal cohort study at the routine ultrasound examination at around 18 weeks of pregnancy at Stavanger University Hospital, Norway, in 2010 [16]. Of these, 193 women reported PGP and were thus eligible for this study. After excluding dropouts and women with missing data, SMS responses from 170 women were included in the analysis. At 30 weeks of pregnancy, 80 women were included, due to missing data and further drop-out.

The SMS question for the studies was “How many days the previous week has your LBP (study 1)/pelvic pain (study 2) been bothersome (i.e., affected your daily activities or routines)? Please answer with a number between 0 and 7.”

In addition, both studies used questionnaires to collect data making it possible to examine the measure “number of days with bothersome pain.” Specifically, in study 1 (patients with LBP), information regarding pain intensity Numerical Pain Rating Scale, (NRS), disability Roland Morris Disability Index (RMDQ), and quality of life (EQ-5D) were collected before the study started and at follow-up. In study 2 (pregnant women), information regarding pain intensity (NRS), disability (Oswestry Disability Questionnaire), and quality of life (EQ-5D) were collected at baseline (18 weeks of pregnancy) and after 12 weeks (i.e., 30 weeks of pregnancy). The NRS is an 11-point numeric pain scale and measures pain intensity; it ranges from 0, representing no pain, to 10, the worst pain imaginable [19]. The RMDQ is a standardized questionnaire investigating self-rated physical disability caused by back pain [20]. The outcome score ranges from 0, indicating no disability, to 24, indicating maximum disability. The ODI also investigates the level of physical disability. The total percentage score ranges from 0, representing minimal disability, to 100, representing maximum physical disability. The EQ-5D is a standardized questionnaire examining health-related quality of life with an outcome score from −0.4 to 1, with negative values corresponding to health states worse than death and 1 representing the best health-related quality of life [21,22].

3 Ethics

The two studies were conducted in accordance with the Helsinki Declaration and received local ethical permissions. Study 1 (patients with LBP) from the regional ethics committee at Karolinska Institutet, Stockholm: 2007/1458-31/4 and study 2 from the Regional Ethics Committee of Western Norway: no. 2010/174. Participants in both studies signed an informed consent to participate.

4 Statistical analysis

Data were extracted from the two datasets, and each dataset was analyzed separately. For descriptive purposes, gender, age, and socioeconomic data were used to get an overview of the population. These data were presented as mean values and standard deviations, and numbers and percentages. For clinical data (days with bothersome pain, pain intensity, disability, and quality of life) additionally, due to skewed distributions, medians and interquartile ranges were presented.

To explore concurrent validity, Pearson’s correlation coefficients were used to investigate the correlation between the number of days with bothersome pain per week and pain intensity, disability, and quality of life. Spearman correlations were also estimated to ensure results were not unduly influenced by outliers.

In the second step, we performed univariable and multivariable linear regression analyses with the number of days with bothersome pain as the dependent variable and pain intensity, disability, and quality of life as independent variables, from which we present regression coefficients with 95% confidence intervals (CI), p-values from Wald tests, R ² and change in R ² from including each independent variable in the models.

We explored non-linear association by applying restricted cubic splines (three knots; default locations) to the independent variables in regression analyses. Unadjusted and adjusted associations were assessed in prediction plots, and the extent of non-linearity was tested using Wald tests. For statistically significant non-linear associations, we further tested if the association was statistically significant and estimated R ² for unadjusted associations and change in R ² for adjusted associations. For adjusted associations, the predictions were performed assuming median values for the adjustment variables. The normality of residuals was checked by plotting, and the assumption of homoscedasticity was tested using the Breusch–Pagan test. Robust standard errors were applied in analyses where assumptions failed (for the LBP study and pregnant women at 30 weeks of pregnancy). Multicollinearity was investigated using the variance inflation factor (VIF). The construct pain, disability, and health-related quality of life are related, but not so much as to cause estimation problems or severely inflated standard errors (LBP: all VIFs <2.1, PGP W18: all VIFs <2, and PGP W30: all VIFs <2.3).

Statistical analyses were performed in IBM SPSS Statistics Version 26 and STATA v. 17. P values <0.05 were considered statistically significant.

5 Results

6 Discussion

In this study, we used two different cohorts to explore the concurrent validity of the variable “days with bothersome pain” by testing the associations with pain intensity, disability, and quality of life.

For people with LBP seeking chiropractic care, days with bothersome pain were moderately correlated [23] with pain intensity and disability. However, for the population of women with PGP, these correlations were weaker. In both populations, days with bothersome pain were inversely correlated with quality of life, again stronger among people with LBP.

The regression analysis showed that the number of days with bothersome pain was best explained by pain intensity for the LBP population, and by pain intensity, disability, and quality of life for women with PGP.

In study 2, women at 18 weeks of pregnancy, pain intensity, disability, and quality of life showed a leveling off in the number of days with bothersome pain at high values of pain and disability and low quality of life, possibly reflecting a ceiling effect, and that women with severe PGP (reporting 7 days a week with bothersome pain) is still a heterogenous group regarding pain, disability, and quality of life. At 30 weeks of pregnancy, this leveling-off effect was only evident for quality of life. For study 1 (patients with LBP), no non-linear associations were found, possibly due to the lower levels of pain, disability, and higher levels of QoL found in this study.

The concept of bothersome pain was introduced as a comprehensive measure to capture pain that mattered to the patient [12,24]. We therefore did not expect the concurrent validity to be perfect, bothersome pain was not thought to replace any one of the other constructs. A previous study found good validity between pain intensity and bothersomeness for patients with severe pain [25]. The question concerning “the number of days” has not been validated, but was thought to be easy to relate to, also in terms of answering the number of days the past week (Monday was fine, Tuesday I could not play tennis as usual, Wednesday I canceled a visit to friends, etc.). Using “the number of days” could possibly complicate matters, as participants were not rating the degree of bothersomeness directly. We thought it likely that, if a participant reported many days with bothersome pain, this would be reflected in their pain intensity, disability, and quality of life. Effectively, even many days with little pain would affect well-being. The results showed that days with bothersome pain did encompass these suggested constructs. However, none of these constructs could explain, alone or together, the full concept of days with bothersome pain; thus, we also concluded that concurrent validity was moderate. In light of previous consensus regarding appropriate pain outcomes, it seems that bothersome pain days may cover suggested severity, function, and quality of life outcomes, and could, therefore, be used as an indicator of progress. This result suggested that other constructs should be tested to fully understand the concept of days with bothersome pain. In light of recent evidence of the mental aspects of the pain experience [26,27,28], it would be interesting to explore correlations with anxiety, depression, and pain perception.

We also concluded that patients with LBP and patients with PGP seemed to interpret the concept of “days with bothersome pain” differently. Patients with LBP typically related days with bothersome pain with pain intensity, which was also found in previous studies of similar populations [24]. Pregnant women seemed quite similar up until 18 weeks of pregnancy, relating days with bothersome pain with pain intensity and disability, but later in pregnancy, quality of life seemed to reflect the concept to a higher degree. One previous study suggests that PGP tends to get worse as pregnancy progresses [29]. Women might experience symptoms other than PGP that are reducing the quality of life as the pregnancy proceeds, perhaps explaining why the quality of life had a statistically significant non-linear effect.

In our opinion, moderate concurrent validity lends credibility to the intention of the concept of “days with bothersome pain,” as patients were likely valuing different things, different factors were influencing their pain, factors that may not even be conscious. Thus, the concept stands strengthened by this analysis. We have tested the concept in two different cohorts and found some similarities but also some differences in what participants with pain conditions relate to this term. Therefore, the concept may be considered valid for use in pain populations, but it is possible that patients with other pain conditions (such as headache, arthritic pain, and neck pain) may assign yet different constructs to the bothersome concept. Therefore, it should be tested in similar ways for other pain populations.

The major strength of this study was the data quality. We had data from two prospective studies with validated measures for testing associations. Furthermore, the different populations made it possible to test the proposed association across two different conditions, even though LBP and PGP may share some common features. We did not merge the two datasets as we wanted to explore the differences between the populations.

The major limitation was the sparseness of other explanatory variables. The two studies were not designed to test the present associations, and thus, such variables were not included. Furthermore, we also had few variables with which to compare the two study populations, leading to construct underrepresentation.

Another limitation was the translation of the term bothersomeness to Swedish and Norwegian, respectively. In both studies, the research teams had extensive experience in the clinical use of pain questionnaires, but it cannot be ruled out that participants interpreted the term in ways we did not intend and that they indeed did it differently in the two countries.

In pain research, recall is a common source of bias, as fluctuating pain is known to influence pain intensity ratings [30]. However, 1 week seems to be a valid recall period for pain intensity [31]. It is unknown whether remembering days with bothersome pain will be influenced in a similar manner.

A potential weakness is the use of different questionnaires for measuring disability (ODI and RMDQ); however, a previous systematic review and meta-analysis suggested that these two measures may be used interchangeably [32].

One may consider the fact that dropouts could exhibit different pain characteristics, and therefore the result may be over- or underestimated. In a previous study using the SMS technology, it was found that dropouts tended to be younger males, with less severe pain of shorter duration [33]. On the other hand, dropouts might not affect estimates of associations (the primary focus of the present study) equally severely as they would do estimates of means, prevalences, etc.

The CI in study 2 (pregnant women) are noticeably wider than those of study 1 (patients with LBP), which calls for a more cautious interpretation of the results of study 2. The populations were purposefully chosen as their differences were meant to deepen the understanding of our analysis. Both populations were in pain, but it should be noted that the included patients did not have severe pain intensity, but rather low (LBP patients) and moderate (pregnant women with PGP) pain, respectively. The influence on the level of bothersomeness is unknown, but our results suggest that days with bothersome pain may be best suited for use in populations with low pain intensity (as in the study with LBP patients), and this should be explored in other studies where participants have a wider range of pain intensities.

7 Conclusions

The concept “days with bothersome pain” had moderate concurrent validity as it was associated with pain intensity, disability, and quality of life. However, patients with LBP and pregnant women with PGP seemed to conceptualize the term differently.