The association between psychosocial factors and mental health symptoms in cervical spine pain with or without radiculopathy on health outcomes: a systematic review

Table 3 Certainty of evidence. GRADE approach for health outcomes

Study Design	Study lead author	Number of studies/patients	Risk of bias	Imprecision	Inconsistency	Indirectness	Overall strength of evidence
Observational Neck pain without CSR
Depression	Bohman Caroll Elbinoune McClean Pico-Espinosa	5/1,718	High	Serious	Moderate	No seriousness	Low
Anxiety	Alipour Elbinoune McClean	1/12,415	High	Serious	High	No seriousness	Very low
Catastrophising	Meisingset	1/70	High	Serious	High	No seriousness	Very low
Stress	Grimby-Ekman	1/1200	High	Serious	High	No seriousness	Very low
Job strain	Van den Heuvel Hoe	2/1898	High	Serious	Moderate	No seriousness	Low
Distress	Lee Hill Hurwitz	3/802	High	Serious	Moderate	No seriousness	Very Low
Kinesiophobia	Beltran-Alacreu	1/128	High	Serious	Moderate	No seriousness	Low
Observation CSR
Distress	Diebo Divi Peoplsson	3/639	High	Serious	Moderate	No seriousness	Very Low
Depression	Kim; Peolsson Enquist MacDowell	4/471	High	Serious	Moderate	No seriousness	Very Low
Anxiety	Wilbault;	1/254	High	Serious	Moderate	No seriousness	Very Low

Through this, the certainty of the evidence was either increased (upgraded) or decreased (downgraded) against the following five criteria:
(1) Methodological limitations using the Cochrane Risk of Bias tool (downgraded where there was a high risk of bias for three or more items; upgraded where all items demonstrated a low risk of bias);
(2) Indirectness relating to similarity to clinical practice (downgraded where reviewers felt the study design was not generalisable to UK practice; upgraded where study design was generalisable to UK practice);
(3) Imprecision relating to the number of participants and events (downgraded where outcomes reported less than 300 participants or five events; upgraded where effects reported in excess of 450 participants or 20 events);
(4) Inconsistency in effect estimates across trials for a given analysis (downgraded where the CIs were four-times the magnitude of the effect estimate; upgraded where CIs were two-times the magnitude of the effect estimate)
(5) Likelihood of publication bias (downgraded when reviewers observed asymmetry in funnel plot shape; upgraded when reviewers observed symmetry in funnel plot shape)

ISSN: 1471-2474