Max van der Velden
Find out if resistance training could be effective for neck- and shoulder pain
Learn to ask the right questions in an RCT
Neck and shoulder pain is the second most common musculoskeletal disorder. Different interventions have been published to reduce pain and improve function in mainly office workers. Interventions such as resistance training and aerobic training seem to be effective. However, mechanisms remain elusive. The aim of this study was to determine a dose-response relationship between resistance training frequency and pain relief in office workers.
The authors planned to detect an effect size of 0.3 with 80% power. Fourteen participants were needed. Before any intervention started, participants were required to undergo an eight-week control period after baseline testing of outcome measures. When finished, the subjects were allocated to either a ten-minute training group (TG10) or a 20-minute training group (20TG). Four exercises were to be performed for four sets, every day, or twice daily, respectively. Participants were instructed to drive up the intensity (adjusting the tube) to a 12-15RM in the first four weeks. After that, the intensity was scaled up to an 8-10RM.
The primary outcome measure was general pain intensity on a visual analog scale (VAS). Secondary measures were worst pain, health-related quality of life (HRQL), and two strength measures (shrug, seated row).
Outcome measures did not change after the eight-week control period. Post-intervention, no differences were detected between the training groups. This led the authors to merge the groups for further analysis. General and worst pain reduced 25% and 43%, respectively, from baseline. HRQL improved 10,6% and no differences were seen for the strength measures.
More and more studies investigating strengthening exercises for neck- and shoulder pain are emerging. Different researchers investigated dose-responsiveness with mixed results. We could argue a few different explanations for the lack of an effect for these patients. First things first, we do not know if there’s such a thing as dose responsiveness for this subgroup of patients.
Let’s say there is; was the study sufficiently large to detect these possibly small differences? Twenty-seven people were included with a mean general pain VAS of 2/10. That’s not much. You will probably need more people to tease out dose responses.
How about the exercises? They seem alright. Although one could argue that they might not be ‘specific’ to the neck. Sure, there are exercises in it that work the traps to quite an extent — but that’s it. No cervical rotation, flexion or extension, just ‘neutral’. I’m playing the devil’s advocate here (as we all should with reading a paper). You see where I’m getting at. Also, the curve of resistance is different for tubes compared to free weights. We could argue that sufficient intensity and/or volume is tough to quantify with a tube.
Talking about intensity, were the exercises sufficiently intense? The researchers aimed for a 12-15RM in the first four weeks, to be increased to an intensity of 8-10RM. You probably noticed in the clinic that patients are notoriously bad at estimating their reps in reserve. Every physio knows a patient that says ‘I cannot do any more reps’, and you ask them to do five more, and they succeed.
So, were they intense enough? We don’t know. Do we need such intensity? Maybe.
Also, the researchers tested for improvements in strength, none were founded. Were the tests invalid for the exercises performed? Was the intensity insufficient? Was it a combination? More questions than answers, as usual.
Was this a suitable group? As you can see in the paper, the average VAS score was 2/10 for these patients. You and I can both agree that this is not a lot. Maybe this exercise program would’ve been proven more effective in a higher pain subgroup? Maybe there we’d be able to find a dose-response. We don’t know, but it could be the case.
This study has several limitations apart from the ones listed above, mostly from a statistics and methodology standpoint. First, the small sample size. The study was powered to detect a 0.3 effect size, coming from a study investigating clinically important differences on the VAS in children. But, being powered for this outcome measure does not mean they can justify multiple measure points and different outcomes. For some — but not all — tests, they corrected for type 1 errors using the Bonferroni correction.
The study was able to detect a 25% reduction in pain, this sounds like a lot. However, the average VAS score for general pain was 20/100; meaning 25% is just about 5/100 of a difference. Would you be happy for your pain to be just 0,5/10 less after 8 weeks of hard work? The same counts for ‘worst pain’, a 43% reduction, 35/100 to 20/100. The improvement seen in this paper over an eight-week course could very well be contextual or regression to the mean.
The bottom line is, as usual, more research with larger sample sizes is needed.