4. Characteristics of observational studies.

Demir 2008	Methods	Case‐controlled study
	Participants	n = 65 Intervention n = 34, control n = 31 Inclusion: all patients admitted to a rehabilitation unit over a 5‐year period who fulfilled the standard diagnostic criteria for GBS Control participants were selected from among hospital workers and patients’ relatives who did not have any major illness and were not taking any medication. Exclusion: Age < 18 years and atypical variants of GBS Turkey
	Interventions	Treatment: inpatient rehabilitation program ‐ no further description provided Control: healthy controls ‐ no intervention
	Outcomes	Primary outcome: nil Secondary outcomes: Limitation of activity < 12 months ‐ Functional Independence Measure (FIM) Quality of life < 12 months ‐ Nottingham Health Profile (NHP)
	Assessment timepoints	FIM ‐ Admission, discharge and 6 months in intervention group only. NHP ‐ 6 months only in all participants.
	Risk of Bias	Adequate sequence generation: No Adequate allocation concealment: No Blinding: No. Whilst blinding of patients and therapists would not have been possible due to the nature of the study, outcome assessors could have been blinded. Incomplete outcome data addressed: No.  Three patients did not complete participation (1 death and 2 unable to complete rehabilitation) but none were included in the analysis. Free of selective reporting: Yes. Other bias: Study design ‐ prospective cohort study. Intervention was not clearly described other than “inpatient rehabilitation”.  Neither duration of intervention nor type of rehabilitation stated.  Also unclear if all the subjects in the intervention group received a similar program. Intervention and control groups were not comparable as control group consisted of healthy individuals not receiving any interventions. No rationale was given for choice of controls. FIM was not measured in control group, hence no comparison of functional measures between intervention and control groups. Baseline NHP was not collected for any of the participants. No sample size calculation performed.
Meythaler 1997	Methods	Retrospective cohort study
	Participants	n = 39 Inclusion: all patients admitted to a rehabilitation unit over a 3 year period who fulfilled the standard diagnostic criteria for GBS. Exclusion: patients not admitted directly from acute care; patients who had undergone previous inpatient rehabilitation. USA
	Interventions	Treatment: Individualised inpatient multidisciplinary rehabilitation program with 3 to 4 hours daily of physical and occupational therapy, and psychology and speech therapy if indicated.
	Outcomes	Primary outcome: nil Secondary outcome: Limitation of activity < 12 months ‐ Functional Independence Measure (FIM)
	Assessment timepoints	Baseline (admission to rehabilitation) and discharge from inpatient rehabilitation (26 SD ± 18.6 days from baseline).
	Risk of Bias	Adequate sequence generation: No Adequate allocation concealment: No Blinding: No. Incomplete outcome data addressed: Yes. All subjects fulfilling the inclusion criteria were included in the final analysis.  Free of selective reporting: Yes. Other bias: Study design: retrospective case review therefore subject to documentation bias No control group
Nicholas 2000	Methods	Retrospective cohort study
	Participants	N = 24 Inclusion criteria: all patients admitted to a rehabilitation unit over a 3 year period who fulfilled the standard diagnostic criteria for GBS Exclusion criteria: not provided U.K.
	Interventions	Intervention: Individualised inpatient multidisciplinary rehabilitation (consultant neurologist, clinical nurse specialist, two sessions of physiotherapy and one session of occupational therapy daily, and if required, speech pathology and psychology).
	Outcomes	Primary outcome: nil Secondary outcome: Activity limitation < 12 months ‐ modified Barthel Index (BI), Functional Independence Measure (FIM) Quality of life < 12 months ‐ Environmental Status Scale (ESS), Handicap Assessment Scale (HAS)
	Assessment timepoints	Admission to and discharge from inpatient rehabilitation (48 ± 32 days)
	Risk of Bias	Adequate allocation concealment: No Blinding: No. Incomplete outcome data addressed: Yes. All subjects fulfilling the inclusion criteria were included in the final analysis.  Free of selective reporting: Yes. Other bias: Study design: retrospective case review therefore subject to documentation bias No control group