Jousimaa 2002.

Methods	Study design: CRCT Data collection: Review of anonymized patient records (case notes from attending physician). This review was conducted by 1 study author; goal was to assess use of a series of clinical practices. Patient records from physician‐reported information‐searching consultations were collected. Patient records for the consultation preceding the information searching consultation were also collected Self report via questionnaire: for each consultation during which they searched for information, physicians completed a questionnaire to describe why and where they searched, the type of information sought, whether the search identified relevant information, and whether or not the physician complied with the information Computer logs: this objective measure of database/electronic information resource use was applied as a means to assess concordance, or lack thereof, between reported searching activities and actual searching activities Unit of analysis issue: "A retrospective power calculation was done, adjusting for clustering using an intra cluster correlation coefficient (ICC) of 0.015 and an average cluster size of 27. With 3,484 patients in total, we had 80% power to detect a 3% difference between the computer and textbook groups for the common elements of the consultation at the 5% significance level" Targeted behavior: use of and compliance with EBMG in patient care
Participants	Participants: newly qualified physicians entering requisite postgraduate patient practice for a period of at least 2 months Total number randomized: 139; IG: 72; CG: 67 Lost‐to follow‐up: 0 Baseline characteristics of participants: Age, mean: IG: 27.3 years; CG 26.9 years Gender: IG: 69.4% women; CG: 73.1% women Setting: 96 primary health centers Country: Finland
Interventions	Description: an electronic version of the EBMG (a CD‐ROM version of EBMG to install on a DC in a consultation room; or a laptop with EBMG installed) Comparison: a textbook version of the EBMG Type of intervention: Organizational: provision of health information to practitioners Duration of intervention: 1 month or 50 patient consultations, whichever occurred first
Outcomes	Primary outcome Physicians' compliance with guidelines based on 9 measurable elements of care ‐ laboratory tests, radiologic tests, physical exam, other exams, procedures, physical therapy, non‐pharmacologic treatment, referrals ‐ for 99 common diagnoses Secondary outcome Amount of evidence sought based on self reported data (via questionnaire) and computer logs
Notes	Non‐compliance with the guidelines was classified into 4 categories (none, minor, major, serious) according to their clinical significance (p. 589, para 2)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: participants "were randomised centrally using computer‐generated numbers to receive either computerised or textbook guidelines" (p. 588, para 4)
Allocation concealment (selection bias)	Unclear risk	For the purposes of this study, we identified newly qualified physicians who would work in a Finnish health center for at least 2 months during the study period from February 1998 until September 1999 (p. 589, para 2)
Blinding (performance bias and detection bias) All outcomes	High risk	Detection bias: low. Assessors evaluating the processes of patient care were "blinded to the study group (computer or textbook, information searching or non‐information searching" "Assessment of patient records/case files for compliance with clinical practices was conducted by one study author who was blinded to whether or not the attending physician had searched for information to support his/her decisions (p. 589, para 2) Performance bias: high. "Prior to the study, the participating physicians agreed not to use the other version of the guidelines if it was available in the health centre" (p. 588, para 2)
Incomplete outcome data (attrition bias) All outcomes	Low risk	1149/4633 (24.8%) participants (> 20% missing but the number missing is approximately the same for each group so that there was only a difference of 6% between the number of eligible records for CG and IG). Figure 1: 4633 participant consultations; 3484 participant records eligible for evaluation; 1149 records excluded from evaluation. Final numbers 1793 records for computer group; 1691 records for textbook group ‐ a difference between groups of < 6%
Selective reporting (reporting bias)	Low risk

CG: control group; col: column; CPG: clinical practice guidelines; CRCT: cluster randomized controlled trial; DC: desktop computer; EAS: Evidence Alert System; EBM: evidence‐based medicine; EBMG: evidence‐based medicine guideline; EBP: evidence‐based practice; EHI: electronic health information; ER: emergency room; GAS: Goal Attainment Scale; IG: intervention group; IQR: interquartile range; IP: Internet protocol; MC: mobile computer; p.: page; para: paragraph; RCT: randomized controlled trial; RHL: Reproductive Health Library; WBCG: well‐built clinical questions.