The Briefing Issue 12

My bias first. I use electrical stimulation through the needle and I teach it. That is the reason I read this review carefully.

A new systematic review compared percutaneous electrical stimulation through a needle, with a surface electrode as the return, against dry needling alone. Four randomized trials, 150 patients, every one of them upper trapezius trigger points in myofascial neck pain. Pain improved in both arms. The needle-plus-current arm had some short-term advantages, but not consistently across outcomes or follow-up points. Disability did not separate. Certainty of evidence: low. The current may help. The evidence does not show it helps more than the needle.

A new clinical practice guideline on acupuncture for neck pain, built on 88 trials and ten thousand patients, points the same direction. All five recommendations are weak. The useful ones for us: a reference course of three to four weeks at three to five sessions a week, and electroacupuncture preferred over manual needling with low certainty. Acupuncture, not trigger point work, so I read it as a dosing document. In Issue 11 I reported that percutaneous electrolysis had not beaten the plain needle in a 13-trial meta-analysis. This week stimulation got the same question and a softer version of the same answer.

Here is how I reconcile the two. The consensus decision tree lists stimulation as an option, and an option is a decision with a reason. "It works better" is not the reason, because the best review we have cannot show it. The clinical reasons survive: a muscle that will not twitch with the needle alone, a patient who tolerates the current better than repeated passes, a motor point you want to hold for minutes rather than seconds. Why stimulation helps in some trials and not others, I cannot say for sure. My read is that it depends on whether the needle is at the root of the problem or at the symptom. Current through a guarding upper trapezius does nothing for the cervical segment that is making it guard.

One anatomy note that belongs under technique. A 39-year-old with an L4-L5 herniation had an ultrasound-guided erector spinae plane block, 20 millilitres of anaesthetic at L4, and had weakness in both legs within fifteen minutes. It cleared by eighteen hours. That is an injectate story, not a needle story, and we put no volume into the tissue. The point is the map: the plane where we needle the multifidus is next to the roots. Treat that depth with the respect you give the thorax.

The model says: find the segment, treat the segment, then re-test the periphery. This week's papers show why the first step is harder than it sounds.

Back to the case. Three months of severe thoracolumbar pain, hyperalgesia over the area, painful myofascial tenderness in the latissimus. Everyone who treated her before the neck was imaged treated the place that hurt. The cervical MRI showed right-sided foraminal stenosis at C6-C7 and C7-T1. Peri-radicular steroid at C6, C7 and C8 brought the pain down substantially in 48 hours, followed by two sessions of percutaneous electrical nerve stimulation along the thoracodorsal pathway. One patient, Level V, an injection rather than a needle, and the authors propose the mechanism rather than prove it. It still belongs in your reasoning. The latissimus was doing what a muscle downstream of an irritated root does. It guarded, it got tender, and it was never the driver.

The harder part is confirming the segment. In Issue 11 the straight leg raise meta-analysis showed the test could neither confirm nor rule out a root. This week a neurosurgical group pooled the other end of the chain: the selective nerve root block, the injection used to prove which root is symptomatic before surgery. Twenty-five studies reviewed, seven pooled. Sensitivity 0.88. Specificity 0.57, with a confidence interval from 0.27 to 0.83. The block catches most true roots and says yes to many that were not the problem. The authors' own conclusion is that the block has to be read alongside the clinical picture. So: the exam under-calls the segment, the block over-calls it, and neither one decides alone. The clinician decides, by reasoning across both and then treating the segment and re-testing the periphery.

Two supporting papers. Forty-three asymptomatic elite soccer players had lumbar MRIs at team entry; 43 percent had a disc herniation and herniation burden had no relationship to missed training. The picture does not tell you who is in trouble. And in 126 college pitchers followed through a season, side-to-side difference in upper cervical rotation had a small association with time-loss arm injury, small enough that the authors say so. I am not building a model on it. I am noting that the neck appears in the arm-injury data at all, because almost no one looks.

How it presents: anterior or medial knee pain, worse on stairs, a quadriceps that looks fine and tests weak, an MRI that says chondral change, and a referral that says "knee OA, strengthen." Somewhere in the first visit the patient mentions their back, and no one writes it down.

Ask the RMO question out loud. What is driving this, and is the driver central or peripheral? The quadriceps is L2 to L4. In Issue 10 I wrote about the chondromalacia trial where needling plus exercise took the pain and left the strength behind, and said the assessment no one ran was the one that would have told us whether the knee was also an L3-L4 problem. The honest evidence on that link: a 26-year cohort of eleven hundred women in which severe lumbar disc degeneration doubled the risk of a later hip replacement, while the knee link lost significance after adjusting for body mass. The knee is a question, not an answer. That is why you treat L3 and L4 before you needle the quad.

Four papers this week, and together they say the knee is a strength and timing problem with a spinal question attached.

A 29-trial meta-analysis of manual therapy added to exercise for knee osteoarthritis: pain down immediately, at short term by about 1.3 points, at medium term by about one point with moderate certainty, function improved short and medium term. Long term, nothing. Not one trial measured pressure pain threshold. The reset works and then fades; what remains is what the patient loaded.

An active-controlled trial of sixty knee OA patients: warm needle acupuncture against mild moxibustion for four weeks, with isokinetic quadriceps torque as the primary outcome. The needle group gained 18 newton-metres more peak torque and 10 watts more power, with better pain and function and no serious adverse events. No sham, retrospectively registered, and the authors cannot separate heat from needle from expectation. I cite it for one reason. The quadriceps got measurably stronger in four weeks with no strengthening program. The weakness was inhibition, and inhibition reverses when you change the input. In the model, the input that matters most is the segment.

Thirty patients with patellofemoral pain, eight weeks of home squats with or without a ball squeezed between the knees. Both improved; the squeeze group improved more, and on EMG the VMO went from firing after the vastus lateralis to firing before it, the pattern of the healthy reference group. Timing, not size, tracked the outcome. Small trial. The muscle was present; it was late.

The reload. Seventy-five knee OA patients, eight weeks, three arms: passive blood flow restriction at 50 percent of limb occlusion pressure, passive BFR at 70 percent, or moderate-intensity resistance training. Moderate load built the most torque and cost the most exertion. Seventy percent occlusion gave the largest functional gain with no rise in exertion. Fifty percent was the weakest arm on almost everything. The number on the cuff is the dose. Measure it; do not estimate it.