Retrieval of diagnostic and treatment studies for clinical use through PubMed and PubMed's Clinical Queries filters

Cynthia Lokker; R Brian Haynes; Nancy L Wilczynski; K Ann McKibbon; Stephen D Walter

doi:10.1136/amiajnl-2011-000233

Table 2

Diagnosis question unadjusted mean values for satisfaction, number of relevant articles, proportion of relevant articles, number of articles retrieved, and placement order of the first relevant article

Outcome Measure	Search mode	Mean	SE	95% CI
Satisfaction^†	CQ	4.9	0.263	4.34 to 5.39
	PM	4.3	0.290	3.75 to 4.91
	y-intercept	0.7	0.362	−0.024 to 1.42
No relevant (standard)^‡	CQ	4.65	0.552	3.53 to 5.77
	PM	3.77	0.515	2.73 to 4.82
	y-intercept	1.71	0.665	0.366 to 3.03*
No relevant (own)^§	CQ	3.28	1.05	1.06 to 5.49
	PM	2.83	0.678	1.40 to 4.26
	Difference	0.444	0.781	−1.20 to 2.09
Proportion relevant^†	CQ	0.38	0.041	0.294 to 0.460
	PM	0.25	0.032	0.191 to 0.318
	y-intercept	0.106	0.048	0.010 to 0.202*
No of retrievals (standard)^‡	CQ	15.4	1.13	13.1 to 17.7
	PM	17.1	0.903	15.2 to 18.9
	y-intercept	−1.36	1.37	−4.63 to 0.914
No of retrievals (own)^§	CQ	11.9	2.03	7.60 to 16.2
	PM	16.7	1.53	13.4 to 19.9
	Difference	−4.78	1.91	−8.81 to −0.743*
Order of first relevant article^†	CQ	3.0	0.533	1.93 to 4.07
	PM	3.94	0.570	2.79 to 5.08
	y-intercept	−1.76	0.526	−2.81 to −0.695*
No of retrievals in Core Journal subset (standard)^‡	CQ	0.375	0.128	0.117 to 0.633
	PM	0.450	0.124	0.200 to 0.700
	y-intercept	−0.075	0.184	−0.447 to 0.297
^†No of retrievals in Core Journal subset (own)^§	CQ	0.333	0.162	−0.008 to 0.674
	PM	0.389	0.183	0.002 to 0.776
	Difference	−0.056	0.151	0.373 to 0.262
No meeting methods criteria^†	CQ	0.474	0.060	0.353 to 0.596
	PM	0.280	0.049	0.181 to 0.379
	y-intercept	0.058	0.102	−0.146 to 0.262

Analysis was performed on the difference between scores for Clinical Queries (CQ) and PubMed (PM) to account for participant effects and adjusted for question origin (standardized or participant's own). For variables analyzed by linear regression and ANOVA, the y-intercept represents the difference between clinical queries and PubMed; for those tested with paired t tests, the difference is the mean difference.

*p<0.05.

^†Analyzed with linear regression with pooled standardized and own questions.

^‡Analyzed with an ANOVA on standardized questions only, with question as an independent variable.

^§Analyzed with a paired t test, since standard and participant questions could not be pooled as a result of a significant ‘question’ effect for standard questions.