Licciardone 2013.

Study characteristics
Methods	RCT with 2 x 2 factorial design
Participants	455 patients aged 21 to 69 years with low back pain for at least 3 months recruited through newspaper advertisements, community agencies, and medical clinics. Exclusion criteria included pregnancy, red‐flag conditions (cancer, spinal osteomyelitis, spinal fracture, herniated disc, ankylosing spondylitis, or cauda equina syndrome); low back surgery in the past year; workers' compensation benefits in the past 3 months; ongoing litigation involving back problems; angina or congestive heart failure symptoms with minimal activity, history of a stroke, or transient ischaemic attack in the past year; implanted biomedical devices (such as cardiac pacemakers or artificial joints); active bleeding or infection in the lower back, or other conditions impeding protocol implementation; use of corticosteroids in the past month; or use of manual treatment (osteopathic or manual therapies delivered by chiropractors or physical therapists) or therapeutic ultrasound in the past 3 months or more than 3 times in the past year. Candidates whose screening was successful by telephone received a clinical screening to exclude those with a high probability of lumbar radiculopathy, a relative contraindication to osteopathic manual therapy (OMT).
Interventions	Participants were allocated to 4 groups: OMT plus ultrasound, OMT plus placebo ultrasound, placebo OMT plus ultrasound, or placebo OMT plus placebo ultrasound. Treatments were scheduled at weeks 0, 1, 2, 4, 6, and 8 using 15 different physicians. Participants could self‐initiate low back pain co‐interventions, such as non‐prescription drugs, complementary and alternative medicine therapies, or usual care.
Outcomes	The current level of low back pain was measured before each treatment and at week 12 using a 100‐millimetre VAS. Secondary outcomes were measured at baseline and at weeks 4, 8, and 12 using the Roland‐Morris Disability Questionnaire, Medical Outcomes Study Short Form‐36 Health Survey general health scale, number of lost workdays in the past 4 weeks because of low back pain, and satisfaction with back care on a 5‐point Likert scale.
Notes	Data are not published separately for the 4 randomised groups. We contacted the authors, who provided stratified outcome data (unpublished).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Central randomisation site based on a computer program that generated pseudorandom numbers.
Allocation concealment (selection bias)	Low risk	Assignments were then conveyed directly to the physicians using numbered, opaque, sealed envelopes, which were subsequently placed in secured, segregated treatment files.
Blinding of participants and personnel (performance bias) Participants	Low risk	Participants blinded as placebo ultrasound and placebo OMT were used.
Blinding of participants and personnel (performance bias) Care providers	High risk	Providers unable to be blinded.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Self‐reported measure, participants blinded to intervention
Incomplete outcome data (attrition bias) Dropout rate described	Low risk	Low rate of attrition, reasons provided in supplemental figure
Incomplete outcome data (attrition bias) Intention‐to‐treat	Low risk	Intention‐to‐treat analysis presented.
Selective reporting (reporting bias)	Low risk	Trial was pre‐registered, and all relevant outcomes were presented. No published protocol available.
Similar groups	Low risk	Baseline measures similar across groups (Table 1).
Co‐interventions	Low risk	Measured co‐treatments reported by participants
Compliance	High risk	Relatively high contraindication and adverse event rate (4% serious adverse events)
Timing of outcome measures	Low risk	All groups measured at same time postrandomisation.