What Does a Hand Know?

The January 2026 Cochrane plain-language summary covers chronic low-back pain and reports 76 trials/11,866 participants, searched through 18 October 2024. At one month, average pain differences on a 0–100 scale favor manipulation/mobilisation by 7.0 points versus sham, 14 versus no treatment, and 4.7 versus other conservative care. These are comparison-specific averages, not responder percentages.

Cochrane describes low/very-low-certainty evidence and limited harms reporting. Its public page’s embedded Main results says 26 trials, unlike the updated summary. We retain section-specific figures, leave the mismatch unresolved, and do not claim to have inspected the full review.

Goertz studied 750 active-duty adults aged 18–50, allocated 375/arm to usual care with or without a chiropractic package containing manipulation and other care. Patients/clinicians were unblinded.

At six weeks, adjusted between-group differences favored added care: pain −1.1/10 (95% CI −1.4 to −0.7); disability −2.2/24 (−3.1 to −1.2).

Responder meant ≥30% improvement from baseline. Table 3 estimates six-week pain responders at 57.5% (95% CI 51.2–64.6) with added care versus 32.5% (27.3–38.8) with usual care: a calculated 25.0-percentage-point difference. These are adjusted model estimates, not raw counts divided by 375; a raw responder denominator is not supplied there.

Table 3 gives six-week pain RR 1.77; Results says 1.43. We do not resolve this discrepancy by guessing.

Six-week adverse effects: 43 with added care, 19 with usual care; no serious related events. Events are not patient risks.

PACBACK randomized 1,000 adults with acute/subacute low-back pain at increased chronicity risk. Over one year, manipulation alone versus medical care showed a disability difference of −0.4/24 (95% CI −1.2 to 0.4), without a significant pain difference. Its reported ≥50% disability-response estimates were 54% in both groups, derived from adjusted models; allocated groups were 201 and 301. These differ in threshold and follow-up from Goertz.

Nolet’s review found low inter-examiner reliability for static palpation of segmental mobility and inconsistent evidence for motion palpation; some soft-tissue tests performed differently. Its search covered January 2000–July 2019 and synthesized 14 low-risk-of-bias articles. This is not one verdict about everything a hand can feel.

Cooperstein/Young examined 34 minimally symptomatic students. Two chiropractors with 30/10 years’ experience, masked to each other’s findings, alternated examination order and independently located the most restricted lumbar site, at least five minutes apart. Additional study-specific training was not reported in Methods. Table 1: ICC 0.39 (95% CI 0.06–0.64); mean absolute location difference 2.6 cm. Table 2: limits of agreement −7.05 to 6.05 cm.

The authors favor continuous distance measures and conclude agreement was useful; they acknowledge small subgroups and a young, healthy sample. Their study did not measure repeatability by the same examiner or establish diagnostic truth.

Bracht’s rotational-asymmetry study assessed 51 asymptomatic people in two sessions with three observers masked to participants and earlier findings. Within-examiner kappa/weighted-kappa estimates ranged 0.07 (95% CI −0.10–0.245) to 0.37 (0.11–0.63). This is repeatability on another palpation task, not repeated clinical improvement.

Bracht’s between-examiner estimates ranged 0.12 (95% CI −0.06–0.29) to 0.30 (0.08–0.52). Within- and between-examiner estimates answer separate questions; neither supplies a reference-standard truth.

Koppenhaver compared manual ratings with mechanized indentation at L3 in 50 people with low-back pain. The manual examiner had eight years’ experience; examiners were blinded to the other assessment. The device measured force/displacement stiffness, averaged over three repetitions. With stiffness above the sample mean as reference, sensitivity was 38% (95% CI 21–59), specificity 45% (28–62).

Indentation measured linear stiffness; thresholds came from this sample’s distribution. The authors discuss other possible constructs, including rotation and end feel. Poor correspondence challenges interchangeability with this reference, without deciding every meaning of touch or whether a treatment-selection rule helps.

Chiradejnant randomized 140 low-back-pain patients to therapist-chosen or randomly chosen mobilisation technique. Nominated spinal level and grade stayed the same. Immediate outcomes did not differ by technique choice. The abstract does not establish long-term benefit from a palpation-guided choice of level.

Degenhardt’s abstract describes 64 participants, six sessions over four months, two blinded examiners after consensus training, and quality-control checks with recalibration when findings differed. Inter-examiner agreement generally improved. This structured research setting tests a shorter feedback process, not years of ordinary practice or accuracy against an independent truth.

Keating’s abstract describes 12 postgraduate physiotherapists, six receiving added force-production training and six controls. Error improved, with a significant between-group result at one-month retention. The measured skill was producing specified forces; diagnostic accuracy and patient benefit were not tested.

Our bounded searches did not establish whether years of ordinary chiropractic practice improve lumbar mobility judgments through outcome feedback. The feedback study cannot answer that career-long causal question. Search terms, related leads, access failures and exclusion reasons are retained; this is a gap in this investigation, not proof that no study exists.

The authors favor continuous distance measures and conclude agreement was useful; they acknowledge small subgroups and a young, healthy sample. Their study did not measure repeatability by the same examiner or establish diagnostic truth.

Indentation measured linear stiffness; thresholds came from this sample’s distribution. The authors discuss other possible constructs, including rotation and end feel. Poor correspondence challenges interchangeability with this reference, without deciding every meaning of touch or whether a treatment-selection rule helps.

Chiradejnant randomized 140 low-back-pain patients to therapist-chosen or randomly chosen mobilisation technique. Nominated spinal level and grade stayed the same. Immediate outcomes did not differ by technique choice. The abstract does not establish long-term benefit from a palpation-guided choice of level.

Goertz studied 750 active-duty adults aged 18–50, allocated 375/arm to usual care with or without a chiropractic package containing manipulation and other care. Patients/clinicians were unblinded.

Indentation measured linear stiffness; thresholds came from this sample’s distribution. The authors discuss other possible constructs, including rotation and end feel. Poor correspondence challenges interchangeability with this reference, without deciding every meaning of touch or whether a treatment-selection rule helps.

Our bounded searches did not establish whether years of ordinary chiropractic practice improve lumbar mobility judgments through outcome feedback. The feedback study cannot answer that career-long causal question. Search terms, related leads, access failures and exclusion reasons are retained; this is a gap in this investigation, not proof that no study exists.