Evidence Review Methods

For the present guidelines the AHCPR review was updated to April 1996.

The target population for the review were adult patients presenting in primary care with acute or recurrent attacks of low back pain with or without radiating leg pain of less than three months duration. Note that the research literature does not consistently distinguish acute, sub-acute or recurrent low back pain or their durations. Patients with possible serious pathology (‘red flags’) were excluded.

For the present guidelines systematic reviews were carried out in four key areas:
bed rest and advice on activity (Waddell, Lewis & Feder 1996); exercise and manipulation (Lewis 1995). For these four key areas these reviews were restricted to randomised control trials (RCTs) relevant to primary care patients. The four main questions were:

1. What is the effect of advice about bed rest, and the duration of bed rest, on clinical outcomes?
2. What is the effect of advice about normal physical activities on clinical outcomes?
3. What is the effect of exercise on clinical outcomes?
4. What is the effect of spinal manipulation on clinical outcomes?

Waddell, Feder & Lewis (1996) carried out systematic reviews of bed rest and advice on staying active. Both reviews searched Medline from 1964 - April 1996 and EMBASE from 1980 - April 1996, supplemented by a personal bibliography (GW) and a citation search of all the acceptable articles.

The focus of the review of advice on staying active was normal physical activities and activities of daily living as distinct from specific back exercises. The interventions were advice to patients about maintaining or returning to activity. ‘Activity’ was not indexed, so multiple searches were carried out of these four databases under general headings of back pain or low back pain, acute or primary care, usual care/management and exercise. Finally, a complete title review was made of all RCTs in acute or not-chronic back pain from 1990. Abstracts were reviewed of all titles which appeared possibly relevant.

This search identified nine trials of bed rest and eight trials of advice on activity which fulfilled our criteria of acceptability. The evidence tables on bed rest and advice on activity are attached (Tables 3 & 4). The detailed methodology and results of the systematic reviews of bed rest and advice on activity have been published as Waddell, Feder and Lewis (1997).

For exercise and manipulation, Lewis (1995) carried out Medline searches of RCTs from January 1992 - September 1995. Strict selection criteria were used to match the original AHCPR methodology as closely as possible. This resulted in accepting three additional RCTs on spinal exercises and two additional RCTs on manipulation. Detailed methodology and results are available from the RCGP as a separate document.

The conclusions of Lewis’s reviews were confirmed by recent systematic reviews of spinal exercises by Faas et al (1996) and of manipulation and mobilisation by Koes et al (1996b). These used a much broader literature search and integrated the more recent RCTs with the pre-1992 evidence.

To update other areas of treatment for back pain, 14 other systematic reviews published since 1992 were identified (Table 1). The quality of these reviews was assessed according to the criteria described by Assendelft et al (1995) (Appendix 1). We also considered whether these new reviews significantly changed the AHCPR evidence base (Table 1).