Condition: Low Back Pain

In new episodes, average pain fell from 52 out of 100 at onset to 23 at six weeks, 12 at six months and 6 at one year. In already-persistent pain the curve is flatter: 51 at baseline, 33 at six weeks, 26 at six months, 23 at one year. Measured in: 33 inception cohorts, 11,166 people with acute or persistent low back pain, pooled by mixed models. What could explain it instead: Regression to the mean. People enroll in these cohorts at the moment their pain is bad enough to seek care, which is by definition near their personal peak, so some of the fall would occur with no intervention and no natural healing at all.. These are cohorts, not treated-versus-untreated comparisons, so the improvement includes whatever care people happened to receive. Almost all cohorts were recruited in primary care in high-income countries, where most people do get some treatment.

Against no treatment, usual care or placebo, exercise reduced pain by 15.2 points on a 0 to 100 scale (95% CI -18.3 to -12.2), which clears the threshold for a change a patient notices. Function improved by only 6.8 points (-8.3 to -5.3), which does not. Against other conservative treatments, pain fell a further 9.1 points and function 4.1, neither clinically important, and there was no difference against manual therapy (MD 1.0). Measured in: 249 randomized trials in adults with low back pain lasting more than 12 weeks, worldwide. Moderate certainty, downgraded for risk of bias: exercise cannot be blinded, so expectation is present in every trial. The comparison that matters most to a reader, exercise against another active treatment, is the one where the benefit stops being clinically important. Adverse events ran 33% in exercise arms against 29% in comparison arms, mostly increased soreness.

Across 39 trials and 20,827 patients with chronic pain, acupuncture beat no-acupuncture control by about 0.5 standard deviations and sham by about 0.2, with roughly a 15% decay at one year. For back pain specifically, the analysis reports 0.17 (95% CI 0.07 to 0.26) against sham and 0.46 (0.41 to 0.50) against no acupuncture. Measured in: 39 randomized trials, 20,827 patients, individual patient data, not published summaries. The back-pain stratum rests on only seven trials against sham. A 0.2 standard deviation advantage over sham is small and it is consistently present, not absent, which is a different claim from clinically decisive. These three acupuncture rows are not three independent bodies of evidence. The Vickers individual-patient analysis and the Cochrane review share at least six back-pain trials, including GERAC itself, and Cochrane reports that seven German studies account for more than 67% of its participants. Read them as three views of one largely German trial set.

A difference of 0.5 points on a 0 to 100 pain scale against placebo, with 0.4 points on disability and 0.4 on quality of life. Graded high quality. Patients on paracetamol were 3.8 times more likely to have abnormal liver function tests. Measured in: Randomized placebo-controlled trials of paracetamol in spinal pain and hip or knee osteoarthritis; the back pain analysis rests on three trials. Measured on a 0 to 100 pain scale, where the difference from placebo was about half a point on high-quality evidence. GlaxoSmithKline funded the main trial behind this and four of the review’s authors, and the result still went against paracetamol, which makes it harder, not easier to dismiss.

Six people need treating for one additional person to reach a clinically important pain reduction (95% CI 4 to 10). In only 3 of 14 analyzes by pain type, outcome and timepoint did the pooled effect even marginally clear a 10-point threshold on a 0 to 100 scale. Gastrointestinal adverse events rose 2.5 fold (95% CI 1.2 to 5.2). Measured in: 35 randomized placebo-controlled trials in spinal pain, median trial duration 7 days. A median of seven days means the gastrointestinal risk figure is for short courses. Longer use raises it, and adds cardiovascular and renal risk that these trials were far too short to measure. The authors' own conclusion is that NSAIDs are effective but that the difference from placebo is not clinically important.

For acute low back pain, advice to stay active beat advice to rest in bed on pain (SMD 0.22, 95% CI 0.02 to 0.41) and function (SMD 0.29, 0.09 to 0.49). For sciatica there was little or no difference on either (pain SMD -0.03; function SMD 0.19). Measured in: 10 randomized trials; the acute back pain comparison rests on 2 trials and 401 people, the sciatica comparison on more. The acute low back pain result is graded moderate quality. Bed rest is the comparator, so this establishes that resting in bed is worse, not that any particular activity is best.

Both CFT arms beat usual care by 4.6 points on the 0 to 24 Roland-Morris activity limitation scale at 13 weeks (95% CI -5.9 to -3.4), and effect sizes were similar at 52 weeks. Societal costs were AU$5,276 to AU$8,211 lower per person. Adding movement sensor biofeedback added nothing. Measured in: 492 adults with chronic disabling low back pain across Australian primary care physiotherapy. The effect held at 52 weeks, where the table shows differences slightly larger than at 13 weeks, which is unusual in this literature. The developers of the therapy declare speaker fees for teaching it and two are clinical directors of a clinic that uses it, which belongs alongside a result this favorable.

At 26 weeks, clinically meaningful improvement in function reached 60.5% on mindfulness-based stress reduction and 57.7% on CBT, against 44.1% on usual care. Pain bothersomeness improved in 43.6% and 44.9% respectively, against 26.6%. The two active arms did not differ. Measured in: 342 adults aged 20 to 70 with chronic low back pain, Seattle, eight weekly two-hour groups. Usual care participants received no group contact at all, so attention and group membership are not controlled for. Roughly half of those assigned to the group programs attended six or more of the eight sessions, so this is closer to an offer-of-treatment effect than a full-dose effect.

In subacute pain of four weeks to three months, high Fear Avoidance Beliefs Questionnaire scores predicted failure to return to work, with odds ratios from 1.05 to 4.64 across four cohorts of 258 to 1,068 patients. The signal was weak or absent in very acute and in long-established chronic pain. Measured in: 21 studies from 2,031 screened references, adults with nonspecific low back pain. What could explain it instead: Pain severity. People whose pain is worse are both more afraid of movement and less likely to return to work, so part of the association is severity showing up twice. Most included cohorts adjusted for baseline pain, and adjustment cannot fully separate the two.. Prognostic, not causal. The odds ratio range is wide enough that the practical size of the effect is uncertain, and it holds only in the subacute window. Whether treating the belief changes the outcome is a separate question, answered better by the CFT trial than by this review.

Moderate-certainty evidence of greater pain relief and better back-specific function immediately after a course of acupuncture compared with no treatment. Against usual care the results were mixed. Measured in: 33 trials, 8,270 participants with chronic nonspecific low back pain, in Europe, Asia, North and South America. Compared against no treatment or a waiting list, so it measures the whole encounter, not the needles, and the trials cannot be blinded on that comparison. These three acupuncture rows are not three independent bodies of evidence. The Vickers individual-patient analysis and the Cochrane review share at least six back-pain trials, including GERAC itself, and Cochrane reports that seven German studies account for more than 67% of its participants. Read them as three views of one largely German trial set.

Cochrane concluded acupuncture may not be more clinically meaningful than sham for pain immediately after treatment. In the largest single trial, response at 6 months reached 47.6% on real acupuncture against 44.2% on sham, a 3.4 point difference that did not reach significance (P = .39), while guideline drug and physiotherapy care reached 27.4%. Measured in: 33 trials and 8,270 participants for the pooled result; 1,162 German outpatients with a mean of 8 years of back pain for the single trial. Sham acupuncture is not an inert control: it usually means a practitioner, a private room, half an hour of attention and something touching the skin. GERAC found 47.6% responding to real needling against 44.2% to sham, a gap of 3.4 points at P=0.39, while conventional drug and physiotherapy care managed 27.4% (P<0.001 against needling). These three acupuncture rows are not three independent bodies of evidence. The Vickers individual-patient analysis and the Cochrane review share at least six back-pain trials, including GERAC itself, and Cochrane reports that seven German studies account for more than 67% of its participants. Read them as three views of one largely German trial set.

Pain-related function was the same at 12 months (BPI interference 3.4 opioid against 3.3 non-opioid). Pain intensity was significantly better on non-opioids (BPI severity 4.0 against 3.5, P = .03). Medication side effects were twice as common on opioids (1.8 against 0.9 symptoms). Measured in: 240 US Veterans Affairs primary care patients with moderate to severe chronic back pain or hip or knee osteoarthritis despite analgesic use; 97.5% completed. A VA population with a much higher burden of psychiatric comorbidity and substance use history than general primary care. It tests a treat-to-target opioid strategy, not any single drug, and it excluded people already on long-term opioids, which is the group clinicians most often ask about.

Heat wrap therapy gave a small short-term reduction in pain, significant at five days against placebo in two trials of 258 people, and adding exercise to the heat wrap reduced pain further at seven days. Evidence on cold was insufficient, with only poor-quality studies found. Measured in: Adults with acute, sub-acute and chronic low back pain across the review’s included trials. This review is from 2006 and the heat wrap trials were largely manufacturer-funded. The effect is small and short-lived, and nothing here establishes a benefit for chronic pain. The absence of good cold trials is a gap in the literature, not a finding that cold does nothing.

Against other recommended therapies, manipulation gave no clinically meaningful short-term pain difference (MD -3.17, 95% CI -7.85 to 1.51) and a small functional advantage (SMD -0.25, -0.41 to -0.09). Against non-recommended therapies the pain difference was -7.48 points, still below the clinical threshold. One adequately powered trial found no excess adverse events against sham manipulation (RR 1.24, 0.85 to 1.81). Measured in: 47 randomized trials, 9,211 adults with chronic low back pain, mean ages 35 to 60. The review records one serious adverse event judged possibly related to spinal manipulation, alongside the transient soreness that is common. Two conflicts belong alongside this result: the lead author’s post was funded by European, Belgian and Netherlands chiropractic bodies, and two of the authors practice as chiropractors.

Median time to a recurrence that limited activity was 208 days in the walking group against 112 days in the control group. The program was cost-effective at a willingness-to-pay threshold of AU$28,000 per quality-adjusted life year. Measured in: 701 Australian adults who had recently recovered from an episode of low back pain; 81% women, mean age 54. The control group received nothing at all, not an alternative program, so attention and monitoring are part of the effect. Lower limb adverse events were more common in the walking group. Six physiotherapist sessions over six months is more support than most people get, and the trial cannot say how much of the benefit came from the walking, not the coaching.

Against inactive controls, massage improved function in sub-acute and chronic low back pain (SMD -0.72, 95% CI -1.05 to -0.39, 725 participants). Short-term pain relief was found in acute pain, but the largest of those figures rests on a single trial of 51 people. Measured in: 25 randomized trials, 3,096 participants, mostly subacute or chronic low back pain. The evidence is graded low to very low throughout, and the benefits are short-term. The strongest-looking acute pain figure comes from one trial of 51 people, so it should not be read as the expected result.

Topical Capsicum frutescens (cayenne) reduced pain more than placebo. Devil's claw, white willow bark, comfrey, Brazilian arnica and lavender oil appeared to reduce pain more than placebo on weaker evidence. No significant adverse events were recorded in the included trials. Measured in: 14 randomized trials, 2,050 adults with acute, subacute or chronic nonspecific low back pain. The herbs assessed are cayenne, devil’s claw, willow bark, comfrey, arnica, lavender and Solidago chilensis, which come from American, southern African, European and Brazilian traditions. No Chinese herb or formula appears anywhere in the review, so nothing here transfers to Du Huo Ji Sheng Tang or the blood-stasis formulas. One included trial delivered lavender oil by acupressure, not topically, which makes it a poor test of the herb alone.

Among people with no back pain, disc degeneration ran 37% at age 20 rising to 96% at age 80; disc bulge 30% to 84%; disc protrusion 29% to 43%; annular fissure 19% to 29%. Prevalence rises steadily with each decade. Measured in: 3,110 asymptomatic individuals across 33 studies using CT or MRI. The pooled age-specific figures are modeled estimates fitted across studies that used different scanners, sequences and reporting definitions, so a single decade figure carries more uncertainty than the smooth curve suggests. It does not follow that a disc finding is never the cause of a given person's pain; it follows that the finding on its own cannot establish that it is.

No difference in pain at up to 3 months (SMD 0.19, 95% CI -0.01 to 0.39) or at 6 to 12 months (SMD -0.04, -0.15 to 0.07), and none in function, quality of life, mental health, overall improvement or satisfaction. Measured in: 6 randomized trials, 1,804 people with low back pain and no features suggesting serious underlying disease. This applies only where red flags are absent, which is the entire premise of the trials. Imaging modality varied between radiography, CT and MRI, and the trials predate current MRI protocols.

Workers scanned within 30 days of onset came off disability at much lower rates and incurred roughly $12,948 to $13,816 more in medical costs, whether or not radiculopathy was present. Even among those with minimal disability the excess cost ran $7,643 to $8,584. Measured in: 555 workers with acute, work-related, disabling low back pain in a US workers' compensation cohort. What could explain it instead: Confounding by indication. Clinicians order early MRI for the patients who look worse or who are more distressed, so some of the longer disability belongs to the reason for the scan, not to the scan. The authors matched on measured severity, which does not remove unmeasured severity.. Four of the five authors work at an insurer's research institute analyzing that insurer's own claims data, which bears on a claim about time off work.

Past malignancy was the single item that meaningfully raised the probability of a spinal tumor, to around 33% in the settings studied. For fracture, long-term corticosteroids raised it to around 33% and visible bruising or grazing over the spine to around 62%. One combination rule reached about 90%, on a confidence interval running from 34% to 99%. Measured in: 14 diagnostic accuracy studies across primary, secondary and tertiary care. Scoped to fracture and malignancy only. It says nothing about cauda equina or spinal infection, whose red flags rest on guideline consensus, not on diagnostic accuracy data, which is why the emergency box above does not cite it. Only 5 of the 14 studies examined combinations of flags at all, so the combination figure is one rule from one study, not a general property.

Four studies measuring intralumbar angles found no difference in peak lumbar flexion during lifting between people with and without back pain. Seven cross-sectional studies using thoracopelvic angles found people with back pain lifted with 6.0 degrees less lumbar flexion. Nine of eleven studies reported no significant between-group difference. Measured in: 11 studies of lifting biomechanics in adults with and without low back pain. Low-quality evidence, so this loosens a widely held belief, not reversing it. This review shares four authors with the cognitive functional therapy trial cited above, so two claims on this page come from the same group, not from independent lines of work.