Back Pain Doctor
SIJ Rehabilitation Protocols: A Clinician's Appendix
This page exists because the patient-facing programme is a synthesis, and anyone delivering it should be able to see what it was synthesised from. Where a parameter here differs from the version on the main rehabilitation page, this page is the source and that one is the translation.
Stuge 2004 — the reference protocol
| Parameter | As reported |
|---|---|
| Design | Randomised controlled trial, stratified block design, blinded assessor, no dropouts |
| Population | 81 women with pelvic girdle pain after pregnancy |
| Intervention | Physiotherapy with a focus on specific stabilising exercises |
| Comparator | Individualised physiotherapy without specific stabilising exercises |
| Duration | 20 weeks |
| Assessment points | Baseline, post-intervention, 1 year post partum, 2 years post partum |
| Primary outcomes | Pain, functional status, quality of life |
| Effect at post-intervention and 1 year | Statistically and clinically significant lower pain intensity, lower disability and higher quality of life in the specific exercise group. Median group difference in evening pain 30 mm on a 100 mm VAS. Disability reduced by more than 50 per cent in the exercise group; change negligible in controls. |
| Effect at 2 years | Minimal disability in 85 per cent of the exercise group versus 47 per cent of controls; minimal evening pain in 68 versus 23 per cent. |
| Author-stated limitation | Group differences disappeared for all measures when controlling for score level at 1 year by regression analysis. Those with the highest baseline disability improved most regardless of allocation. |
That final row is the one most often omitted when this trial is cited, and it should not be. It does not erase the result, but it does mean the trial establishes less about mechanism than the headline figures suggest.
The design comparison — Stuge 2006
Stuge and colleagues published a direct methodological comparison of their positive trial against Mens' negative trial of diagonal trunk muscle exercises. Both were judged high methodological quality with comparable populations, so the difference lies in the intervention design.
Design features associated with the positive result
- Local muscles targeted first, with global muscles added gradually — rather than global muscles alone.
- Supervised and corrected in person — rather than instructed by videotape.
- Choice of exercise, order and dosage individualised — rather than a fixed protocol.
- Pain explicitly avoided during exercise.
- Exercises addressed the entire spinal musculature rather than an isolated region.
The authors' own conclusion: effective treatment of postpartum pelvic girdle pain may be achieved when exercises for the entire spinal musculature are included, individually guided and adapted to each individual. Any protocol that drops supervision or individualisation is not delivering the tested intervention.
What the wider literature does and does not support
| Study | Design and n | Finding | How to weight it |
|---|---|---|---|
| Zitti 2025 | Narrative review; 6 eligible studies (2 SRs, 4 RCTs) | Five of six reported no statistically significant short-term pain change. Exercises judged unspecific and not designed per motor control theory. | The strongest published challenge to motor control exercise in this population. Note it is a narrative review with a small eligible set, and its criticism is of exercise design rather than of exercise per se. |
| Kordi 2013 | RCT, n=105, pregnancy-related PGP, 3 arms | Non-rigid lumbopelvic belt superior to home-based stabilising exercise plus information, and to information alone, at 3 and 6 weeks on VAS and ODI. | Directly relevant, but the exercise arm was unsupervised and home-based over 6 weeks — i.e. closer to the design associated with negative results. |
| Kluge 2011 | RCT, n=50, pregnancy, 10-week programme | Significant improvement in pain intensity in the exercise group (p<0.01); no significant change in controls. Functional ability change did not reach significance (p=0.06). | Supportive but small, and during rather than after pregnancy. |
| Li 2025 | Double-blind RCT, n=52, postpartum PGP, 4 weeks | Core stability plus whole-body vibration improved Timed Up and Go and 10-metre walk speed over core stability alone; no significant between-group difference in VAS (p=0.07) or Pelvic Girdle Questionnaire (p=0.28). | Function improved, pain did not differentiate. Useful mainly as a reminder that functional and pain outcomes diverge in this population. |
| Elden 2008 | Controlled single-blind trial, n=386, pregnancy | Safety study of acupuncture and stabilising exercises as adjuncts to standard treatment. No serious adverse events in any arm. | Cited for safety rather than efficacy. |
Assessment measures worth using
| Measure | Use | Note |
|---|---|---|
| Active straight leg raise | Load-transfer assessment and re-assessment | Used as an outcome measure in the SIJ fusion trials. Score the perceived effort and the change with manual pelvic compression. |
| Oswestry Disability Index | Primary disability outcome | The common currency across the injection, ablation and fusion literature, which makes cross-comparison possible. Minimum clinically important change is commonly taken as 15 points in the prolotherapy literature; the fusion trials used 15 for clinical improvement and 18.8 for substantial clinical benefit. |
| Pelvic Girdle Questionnaire | Condition-specific function | More responsive than ODI in the pregnancy-related population. |
| Numeric rating scale or 100 mm VAS | Pain | Record separately for morning, evening and provoked pain — the Stuge trial's clearest signal was in evening pain. |
| Long dorsal sacroiliac ligament palpation | Extra-articular screening | Standardised palpation caudal to the PSIS. Sensitivity 76 per cent in strictly defined peripartum pelvic pain, 86 per cent in those also ASLR- and P4-positive. A sensitivity figure only — it does not confirm the source. |
| Provocation cluster (distraction, thigh thrust, compression, Gaenslen's, sacral thrust) | Intra-articular screening | Three or more positive raises sensitivity and specificity significantly. Negative tests carry greater predictive value than positive ones. Does not assess extra-articular structures. |
Response thresholds from the interventional literature
Useful when deciding whether a procedure has done what it was meant to do, and when writing to a specialist.
Thresholds as the 2025 consensus guidelines state them
- Diagnostic or prognostic block called positive at 50 per cent relief. Higher thresholds have not been shown to improve outcomes from subsequent definitive procedures — although three participating societies dissented and would require 75 per cent relief from two blocks before radiofrequency ablation.
- Therapeutic response threshold is lower: at least 30 per cent pain relief, or meaningful benefit on a non-pain outcome such as opioid cessation.
- Where a prolotherapy series is being run, Hoffman's retrospective cohort found that a 15-point ODI improvement before the second injection predicted eventual responder status with 92 per cent sensitivity and 80 per cent specificity — a defensible basis for stopping a series that is not working.
The three-month rule from the Sydney protocol
Cusi and colleagues' prolotherapy protocol at the dorsal interosseous ligament required patients to complete a three-month stability programme before injection. That sequencing is worth preserving whatever the injectate, for two reasons: a proportion of patients no longer need the procedure by the time the programme is complete, and those who do proceed have the loading capacity to consolidate any window the injection opens.
It is also the defensible position medico-legally. Offering a regenerative injection at a joint where the guideline evidence is graded weak, before an adequately delivered rehabilitation trial, is difficult to justify on the record.
Related on this site
- The patient-facing programme
The staged translation of everything on this page.
- Referral and surgery
Where the ablation and fusion literature sits, and what a referral should contain.
Common questions
Is there a validated SIJ-specific rehabilitation protocol?
How long should a rehabilitation trial run before considering a procedure?
What outcome measures should I send with a referral?
Evidence reviewed
Every clinical statement on this page traces to one of the sources below, including the ones that point the other way. Where a study is small, unblinded or authored by people with a commercial interest in the result, that is noted alongside it rather than left out.
- Specific stabilising exercises for pelvic girdle pain after pregnancy: RCT (Stuge et al, Spine 2004;29(4):351-9) ↗
Source of every parameter in the reference protocol table.
- Two-year follow-up (Stuge et al, Spine 2004;29(10):E197-203) ↗
Including the regression-adjustment caveat.
- To treat or not to treat postpartum pelvic girdle pain with stabilising exercises? (Stuge et al, Man Ther 2006;11(4):337-43) ↗
The design comparison against the negative trial.
- Motor control exercises for short-term pain in pelvic girdle pain (Zitti et al, Healthcare 2025;13(5):572) ↗
The counter-case, in full.
- Whole-body vibration for postpartum pelvic girdle pain (Li et al, J Back Musculoskelet Rehabil 2025;39(2):500-510) ↗
Function improved, pain did not differentiate between groups.
- Adverse effects of standard treatment, acupuncture and stabilising exercises in pregnancy (Elden et al, BMC Complement Altern Med 2008;8:34) ↗
Safety data across 386 pregnant women; no serious adverse events in any arm.
- The use of prolotherapy in the sacroiliac joint (Cusi et al, Br J Sports Med 2010;44(2):100-4) ↗
Source of the three-month stability programme prerequisite.
- Functional outcome from sacroiliac joint prolotherapy in patients with SI joint instability (Hoffman & Agnish, Complement Ther Med 2018;37:64-68) ↗
Source of the 15-point ODI stopping rule: sensitivity 92 per cent, specificity 80 per cent for predicting eventual responder status.
- Consensus practice guidelines on sacroiliac joint complex pain (McCormick & Cohen et al, Pain Medicine 2025;26(12):817-917) ↗
Multispecialty international working group convened by the American Academy of Pain Medicine and ASRA. Twenty-one questions, complete committee consensus on all of them, 21 endorsing organisations. This is the reference point for almost everything in this section.
Last reviewed 2026-09-04. This is general information about a condition and its treatments, not personal medical advice, and it is no substitute for assessment by a clinician who has examined you.
More in this section
Ready for a clearer plan for your back or musculoskeletal pain?
Book an assessment with Dr Joshua Hatch.
Your assessment focuses on understanding the likely source of your pain and the most appropriate non-surgical options for your diagnosis, with the aim of reducing pain and improving function.
Book an appointment