If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence and reprint requests to: E. Parry, Primary Care Centre Versus Arthritis, Research Institute for Primary Care & Health Sciences, Keele University, ST5 5BG, UK.
Primary Care Centre Versus Arthritis, Research Institute for Primary Care & Health Sciences, Keele University, UKMedical Statistics and Clinical Trials, Nottingham Clinical Trials Unit, School of Medicine, University of Nottingham, D Floor, South Block, QMC, Nottingham, NG7 2UH, UK
To determine the natural history of flare-ups in knee osteoarthritis and their relation to physical exposures.
Design
Adults aged ≥45 years with a recent primary care consultation for knee OA/arthralgia completed a daily pen-and-paper diary for up to three months, including questions on average knee pain intensity, pain descriptors, other symptoms, activity interference, and selected physical exposures (prolonged kneeling, squatting, climbing stairs, ladders, and moving/lifting heavy objects). Informed by a systematic review, flare-ups were defined a priori. We calculated the rate of flare-ups in the sample, described their nature and duration, and estimated their association with physical exposures in the prior 48 h.
Results
67 participants completed at least one month of diaries, 37 (55%) were female, mean age 62 years (SD 10.6) with a mean body mass index of 24.6 kg/m2 (SD 5.1). 30 participants experienced a total of 54 flare-ups (incidence density 1.12 (95%CI 0.80, 1.57) flare-ups/person-days). The median duration of flare-ups was eight days (range: 2–30). During a flare-up participants were more likely to report sharp, throbbing, stabbing, burning pain, swelling, limping, stiffness, being woken by pain, taking more analgesia, and stopping usual activities. Exposure to one or more physical exposure increased the risk of a flare-up in the subsequent 48 h (odds ratio 2.19 (95%CI: 1.22, 4.05)).
Conclusions
Our study with intensive longitudinal data collection suggests acute flare-ups may be experienced by a substantial number of patients. These episodes often last a week or longer, are disruptive, prompt changes in self-management, and may be triggered by high-loading physical activities.
Longitudinal studies following patients with symptomatic knee osteoarthritis over several years suggest that non-progressive symptom trajectories are relatively common
Pain trajectory groups in persons with, or at high risk of, knee osteoarthritis: findings from the Knee Clinical Assessment Study and the Osteoarthritis Initiative.
. However, within these average long-term trajectories, it is well-recognised that patients can experience substantial variability in the presence, nature, and severity of symptoms over time, including episodes of increased pain that may be experienced as acute events
. These events, particularly when they have an unpredictable, sudden onset, can be distressing for patients and can impact on quality of life, normal activities (including productivity losses
) and health service use. Yet unlike other long-term conditions, there is still considerable uncertainty around the nature, definitions, and terminology of these phenomena in osteoarthritis. Classification criteria for flare-ups in knee OA have been proposed but have not been widely adopted
, and other proximal and more distal determinants.
We sought to describe the natural history of ‘flare-ups’ in knee OA in a sample of community-dwelling symptomatic adults, using an observational daily diary study. We chose the term ‘flare-up’ over terms such as ‘exacerbation’ following workshops with patients and members of the public on their preferred terminology and from more frequently used terminology in the medical literature as found in a recent systematic review
Adults aged 45 years and over registered at one of two General Practices in the West Midlands, England and with a recorded consultation for knee osteoarthritis or knee pain/arthralgia in the previous two years were mailed a short questionnaire containing items used to describe the sample, judge eligibility for the daily diary study, and provide baseline values of ‘what is normal for me’ with respect to knee symptoms, activity, and analgesia (adapted from Trappenburg et al.
Effect of an action plan with ongoing support by a case manager on exacerbation-related outcome in patients with COPD: a multicentre randomised controlled trial.
). A standard three-stage mailing procedure was used with non-respondents sent a reminder postcard at two weeks and repeat questionnaire at four weeks. Respondents were eligible for inclusion in the daily diary study if they reported knee symptoms on at least one day in the previous 12 months, provided written informed consent to further contact and indicated their willingness to complete daily diaries for up to three months. Respondents were excluded if they self-reported a diagnosis of inflammatory disease, previous bilateral total knee replacement (TKR) or TKR in the index knee (the worst affected knee) or did not complete baseline knee symptom questions.
Data collection
All eligible, consenting questionnaire respondents were invited to complete three consecutive one-month pen-and-paper diaries. The diaries contained nine items for each day (see Supplementary Data 1) which participants were asked to fill out at the end of the day. Average pain intensity in the previous 24 h was assessed using a 0–10 numerical rating scale (NRS)
to be associated with OA flare-ups: stiffness lasting >20 min, swelling, night pain and limping. Participants were also asked about pain quality using a short list of descriptors which included continuous pain descriptors (dull, aching, throbbing), intermittent pain descriptors (sharp, stabbing), and neuropathic-type pain descriptors (burning, numbness, pins and needles)
. Participants were also asked about any changes in usual medication (more than normal, less than normal, the same) and whether symptoms had stopped them taking part in usual activities.
The study was approved by the North of Scotland Research Ethics Committee (Reference: 13/NS/0049).
Statistical analysis
Scatterplots were used to visually inspect daily pain intensity scores for each participant. To estimate within-person variability we calculated a Variability Index
for each participant based on the average standard deviation of their daily pain intensity scores within half-monthly periods. The periods ranged from 14 to 16 days due to the varying length of the month over the 3 month study period. The standard deviation was chosen as it is the most common measure of variability which averages the absolute deviation of each day's pain intensity from the mean pain over the 14–16 days period thus capturing any pain fluctuations. This method has also been used in a previous study investigating pain variability of patients with fibromyalgia
. The 14–16 day periods were chosen based on the number of available data points and to allow for reliable estimation of SD due to the distribution assumptions. The possible values were positive, with zero indicating no variability and a higher number indicating greater variability. Informed by a systematic review of flare-up definitions in the medical literature
we defined a flare-up a priori as: an increase of at least two points from baseline (‘normal for me’) in average pain intensity in the past 24 h (0–10 NRS) which was sustained for at least two consecutive days. A flare-up was judged to be resolved when pain intensity returned to baseline level for five consecutive days.
We then estimated the proportion of respondents experiencing at least one flare-up during the period of observation, the incidence density with 95%CI of flare-ups for the sample as a whole (expressed as the number of flare-ups per 100 person-days at risk, i.e., denominator excluded days in flare-up and the five days needed for it to be judged ‘resolved’) using Poisson regression taking into account recurrent events, and the duration of flare-ups (median, interquartile range (IQR)).
The time course of an ‘average’ flare-up was illustrated by plotting combined group-mean daily pain intensity NRS scores across all first flare-ups, anchored to a common timescale with zero representing the first day of flare-up and extending to seven days prior and 30 days after the flare-up. To describe the nature of flare-ups and their impact on individuals, descriptive statistics (means, SD or proportions) were used to summarise symptoms, change in medication and whether pain had stopped usual activities across all flare-up days and all non-flare-up days, among participants experiencing at least one flare-up. We then used mixed-effect models to estimate the relative frequency (expressed as odds ratios) and severity (expressed as regression coefficients) of symptoms on flare-up vs non-flare-up days accounting for the clustered nature of the observations.
To determine whether exposure to selected physical activities was associated with flare-up onset, we conducted a case-crossover analysis with each individual acting as their own control
. Case windows for exposures were defined as the 48 h prior to the first day of a flare-up. For each case we selected up to four matched ambidirectional control windows which were 48-h periods on at-risk days which corresponded with the same days of the week as the case window to remove confounding by variation in physical exposures across days of the week (e.g., weekdays vs weekends). We calculated unadjusted exposure odds ratios (OR) based on the conditional maximum likelihood estimate with 95% mid-P exact confidence intervals using OpenEpi (www.OpenEpi.com).
Analyses were performed using Stata, Version 13 (StataCorp 2013).
Results
Of the 220 out of 330 responders to the baseline questionnaire, 106 (48%) were eligible and were invited to take part in the diary study, consented to further contact, and were mailed the first diary. Reasons for non-eligibility included; inflammatory disease (41), TKR (27), missing/blank Q (22), no recent knee pain (14), withdrew/died/moved (10). Of the 67 (63%) participants who completed at least one monthly diary, 37 (55%) were female, mean age 62 years (SD 10.6) with a mean body mass index of 24.6 kg/m2 (SD 5.1). Of a possible 5,491 diary days, 4,328 (79%) were fully completed, 1,163 (21%) were partially completed, and 111 (2%) were missed completely. Comparing responders to non-responders ages were similar (mean age 62.2 (SD 10.6) and 61.7 (SD 11.0) respectively) and there were slightly more females amongst the non-responders (25 (64.1%) non-responders vs 37 (55.2%)).
The median Variability Index across the sample was 0.68 (IQR 0.41, 1.05), and this was higher during non-flare-up days in participants classed as having experienced a flare-up than in those participants who did not experience a flare-up ((median (IQR)): 0.83 (0.51, 1.13) vs 0.52 (0.27, 0.97).
Over the period of observation 30 participants (45%) were classed as having experienced a total of 54 flare-ups (one flare-up (n = 16), two flare-ups (n = 6), three flare-ups (n = 6), four flare-ups (n = 2)) giving an estimated incidence density of 1.12 (95%CI 0.80, 1.57) flare-ups per 100 person-days. Illustrative examples of participants with contrasting variability in daily pain scores are provided in Supplementary Data 2.
Those experiencing a flare-up were slightly more likely to be male (50% vs 40%; proportion difference (95% CI): 10% (−14%, 33%)), have a higher BMI (mean (SD) 25.5 (5) vs 23.9 (6) kg/m2; mean difference (95% CI): 1.6 (−0.9, 4.1)), and report previous injury (50% vs 42%; proportion difference (95% CI): 8% (−15%, 33%)), although none of these differences were statistically significant given the relatively small numbers of participants in each group.
The median duration for a flare-up was eight days (IQR 3, 23; range 2–30). The ‘average’ time course shows the sudden onset and relatively quick reduction in pain intensity within 48 h followed by a longish plateau (Fig. 1). Flare-up days compared to non-flare-up days were accompanied by a higher occurrence of knee stiffness (OR 10.9; 95% CI: 7.0, 17.1), limping (12.4; 7.4, 20.8), swelling (14.5; 8.3, 25.4), being woken by pain (7.0; 4.3, 11.2), use of most pain descriptors (but particularly ‘throbbing’, ‘sharp’, ‘stabbing’ pain, and ‘numbness’), interference with usual activities (26.5; 3.7, 51.5) and taking more medication than normal (23.9; 13.8, 41.4) (Table I). The above features tended to be at their highest levels during the first two days of a flare-up (data not shown).
Fig. 1Time course of an ‘average’ flare from 7 days prior to 30 days after the onset of a flare (day 0). Points are group-mean pain intensity for all participants' first flares combined.
Table ISeverity and occurrence of symptoms and impact experienced during flare-up days vs at-risk days among participants experiencing one or more flare-ups (n = 30)
From mixed-effect model (logistic for binary and linear for continuous outcome). Results are expressed as odds ratios (95% CI) except average knee pain intensity which is expressed as a regression coefficient (i.e., mean difference) and 95% CI.
Average knee pain intensity (0-10NRS): mean (SD)
5.4 (1.9)
3.1 (2.0)
2.5 (2.3, 2.6)
Pain descriptors
Dull
52 (17)
673 (35)
0.4 (0.2, 0.7)
Aching
218 (73)
1122 (59)
6.9 (4.1, 11.6)
Throbbing
95 (32)
301 (16)
18.1 (9.8, 33.3)
Sharp
146 (49)
206 (11)
11.2 (6.8, 18.6)
Stabbing
108 (36)
269 (14)
11.8 (7.2, 19.5)
Burning
73 (24)
171 (9)
6.7 (4.0, 11.1)
Numbness
72 (24)
15 (1)
6.7 (1.9, 23.1)
Pins and needles
1 (<1)
1 (<1)
–
Other
9 (3)
45 (2)
1.3 (0.6, 3.1)
Knee swelling
149 (50)
668 (35)
14.5 (8.3, 25.4)
Limping
191 (64)
804 (42)
12.4 (7.4, 20.8)
Knee stiffness lasting >20 min
178 (60)
505 (26)
10.9 (7.0, 17.1)
Woken at night by knee pain
103 (35)
189 (10)
7.0 (4.3, 11.2)
Taking more medication than usual
94 (34)
181 (12)
23.9 (13.8, 41.4)
Pain stopped usual activities
44 (15)
76 (4)
26.5 (13.7, 51.5)
Figures are n (%) unless otherwise stated.
NRS Numerical rating scale; SD Standard deviation.
∗ Days during a flare-up (i.e., from start of flare-up to first day when pain returned to ‘normal’ levels for five consecutive days.
† Excludes flare-up days as well as the five consecutive days after last flare-up day.
‡ From mixed-effect model (logistic for binary and linear for continuous outcome). Results are expressed as odds ratios (95% CI) except average knee pain intensity which is expressed as a regression coefficient (i.e., mean difference) and 95% CI.
For the case-crossover analysis using physical exposures there were 88 cases periods in total and 328 control periods. The number of control periods per case ranged from 1 to 4 (median:2, mean: 2.43 (SD1.12)). Exposure to one or more of the selected physical exposures (prolonged kneeling, lifting/moving heavy objects, climbing several flights of stairs, prolonged squatting, climbing ladders) was associated with increased risk of acute flare-up in the subsequent 48 h (odds ratio 2.19; 95%CI: 1.22, 4.05) (Supplementary Data 3). Exposure to each individual physical activity was positively associated with flare-up onset but estimates lacked precision due to small numbers of discordant pairs.
Discussion
Our study supports the notion that knee OA, for some people, is characterised by intermittent acute or sudden increases in pain with an associated change in pain quality and knee symptoms.
Pain intensity in our study was highly variable for some and stable for others. Although not explored in this study, Schneider et al. found a link between pain variability in OA and depression
Individual differences in the day-to-day variability of pain, fatigue, and well-being in patients with rheumatic disease: associations with psychological variables.
Predictability of pain is important for episode management and patient understanding. Prior to flare-up onset we saw a marked step up in symptoms rather than a gradual onset. This may be partly due to once daily measurement, however, we found that certain activities reported in the 48-h period prior to a flare-up were associated with flare-up onset. Zobel et al. identified that knee buckling and knee injury were triggers for acute events
. The link we have found may give an insight into short term events and their intrinsic part in how OA develops and progresses. It is possible that these acute events lead to a cumulative insult on the knee joint that eventually leads to disability.
Identifying potential triggers are important in the management of flare-ups in terms of activity avoidance and for earlier and preventative management strategies. Other management strategies that may lead to early termination of a flare-up include recognising the early changes in symptoms, for example knee stiffness, swelling, limping and night pain which have previously been used as part of the criteria for flare-up identification
In response to flare-ups a third increased usual medication and a small number reported stopping usual activities. Our study did not capture lesser but still problematic interference with activities which may affect quality of life if sustained. We recognise that there may be potential time-varying confounders that could not be controlled in the study, for example, those causing psychosocial stress or a perceived improvement in symptoms, leading to a reduction in medication and increase in activity.
Limitations of the study include the potential for inaccurate recall of average pain scores over the 24-h period and the extent to which participants may have retrospectively filled in diary entries. We also acknowledge the bias that may be introduced by missing data. Our case-crossover analysis was unable to explore individual physical exposures due to small numbers and the use of ambidirectional control windows assumes no strong effect of flare-ups on subsequent exposure levels.
In providing an operational definition of a flare-up we recognise the need for continued empirical work. Sensitivity analyses in larger datasets could usefully explore whether absolute or relative (to baseline) increases in pain are best for defining flare-ups as well as the minimum duration of these.
Conclusion
Our study with intensive longitudinal data collection suggests acute flare-ups may be experienced by a substantial number of patients. These episodes often last a week or longer, are disruptive, prompt changes in self-management, and may be triggered by high-loading physical activities.
Author contributions
All authors were involved in conception and design of the study, analysis and interpretation of data, drafting the article, critical revision of the article for important intellectual content, final approval of the article. GP takes responsibility for the integrity of the work as a whole from inception to finished article.
Competing interest
GP received consultancy fees from InFirst plc and Good Relations plc.
Role of the funding source
The paper presents research funded by Arthritis Research UK [Centre of Excellence award: 18139]. ELP received funding from a National Institute for Health Research (NIHR) In-Practice Fellowship, NIHR Academic Clinical Fellowship and SPCR GP Progression Fellowship.
The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Pain trajectory groups in persons with, or at high risk of, knee osteoarthritis: findings from the Knee Clinical Assessment Study and the Osteoarthritis Initiative.
Effect of an action plan with ongoing support by a case manager on exacerbation-related outcome in patients with COPD: a multicentre randomised controlled trial.
Individual differences in the day-to-day variability of pain, fatigue, and well-being in patients with rheumatic disease: associations with psychological variables.
30924-0&id=gr1.jpg){kind=link}