Tuesday, September 11, 2012
Writing in an advance edition of BMC Research Notes, researchers from Norway’s National Institute of Occupational Health report on 48 volunteers who agreed to spend 90 minutes performing computer tasks that were known to cause pain in the shoulders, neck, forearms, and wrists [Strøm et al. 2012]. Nearly half (N=22) of the subjects had chronic neck and shoulder pain and the researchers wanted to assess how those persons tolerated the tasks as compared with persons who were pain-free at the outset (N=26).
As a matter of convenience, the researchers allowed all subjects to have a light breakfast before performing the tasks, including coffee to avoid unpleasant effects of caffeine deprivation, such as decreased vigor and alertness, sleepiness, or fatigue. Nineteen (40%) of the subjects drank from 1/2 to 1 cup of coffee (median 1 cup); 8 subjects in the chronic pain group, 11 in the pain-free reference group.
Unexpectedly, the 19 coffee drinkers all reported a lower intensity of pain during the task than the 29 people who did not have coffee. At the end of the work task the average pain intensity was rated as 41mm (on a 0-to-100mm Visual Analog Scale, or VAS) among the coffee drinkers and 55mm for coffee abstainers — a statistically significant difference. Similar gaps were found for all pain sites measured — ie, neck, shoulders, forearms, wrists — and coffee’s apparent pain-mitigating effects held up regardless of whether or not the subjects had preexisting chronic pain.
The researchers conclude that their results may have interesting implications for the modulating effects of caffeine on musculoskeletal pain in everyday settings. However, double-blinded, randomized, placebo-controlled trials (RCTs) are needed to confirm these results.
COMMENTARY: Based on this research, news media published enthusiastic stories ballyhooing the pain-relieving effects of a “cuppa joe.” But, the findings relating to the benefits of coffee in this study were serendipitous, in that it was designed for another purpose; ie, to investigate mechanisms of pain development during computer office work. So, the coffee outcomes are speculative and Strøm et al. duly note that RCTs to specifically test this effect are required.
Cause-effect relationships cannot be inferred in this study, since other factors besides caffeine intake might have been important. In fact, the authors note that a plausible biological mechanism of caffeine in affecting pain response is uncertain. While other studies have reported attenuated muscle pain after coffee administration during rigorous exercise, the caffeine doses were >5 mg/kg. In this present study, caffeine doses were unmeasured, but a single cup of coffee typically provides only 0.4 to 2.5 mg/kg of caffeine.
Group sizes in the present study, and particularly the subgroups of coffee drinkers, were too small to assure sufficient statistical power for robust and reliable results. Furthermore, the pain-attenuating outcome associated with caffeine — a statistically significant 14mm difference in VAS scores — exerted only a moderate 0.54 effect size. While the authors claim that the 14mm difference can be considered as clinically significant, this might be questioned.
Technical Side Note…
How much of a difference in pain score on a 100mm VAS is clinically important?
Strøm et al. observe that 14mm is sufficient, and this was based on a study by Kelly  designed to establish the Minimum Clinically Significant Difference (MCSD) in a VAS score that is perceived by patients as being of consequence. Overall, it was found that a change of approximately 12mm (95% Confidence Interval, 9mm to 15mm) is sufficient to denote that patients are distinguishing between pain that becomes “a little better” or “a little worse” over time. This MCSD was slightly lower (11mm) if the initial pain was in the mild range (0-30mm), somewhat higher (14mm) if pain was initially in the moderate range (31-69mm), and 10mm if pain was initially in the severe range (70-100mm). Differences between MCSDs were not statistically significant; overlapping 95% CIs ranged from 4 to 18.
Kelly emphasized that small changes in pain score may be statistically significant, but those less than 12mm would not be detectable by patients as having clinical importance. Furthermore, a 12mm MCSD improvement characterizes pain that is only a “little better” and does not signify a level of pain relief that is optimal or desired by patients as a result of some therapy or intervention.
In the study by Strøm and colleagues the moderate effect size (Cohen’s d=0.54) and the 14mm improvement on VAS score was likely of minor clinical importance.
As much as anything, this study demonstrates that even seemingly simple and straightforward research, suggesting an interesting and compelling conclusion, must be considered circumspectly. While news reporters seldom examine the nuances that may challenge the validity of research outcomes, healthcare providers and their patients should insist upon valid and reliable evidence as proof for clinical decision making purposes.
A morning jolt of caffeine may offer certain benefits to persons accustomed to it, but significant relief of pain is probably expecting too much. At least until better research confirms otherwise.
> Kelly A-M. The minimum clinically significant difference in visual analogue scale pain score does not differ with severity of pain. Emerg Med J. 2001;18:205-207 [available here].
> Strøm V, Røe C, Knardahl S. Coffee intake and development of pain during computer work. BMC Research Notes. 2012(Sep 3);5:480 [available here].