Table 2

Combination of terms with the best sensitivity (keeping specificity ≥50%), best specificity (keeping sensitivity ≥50%), and best optimization of sensitivity and specificity (based on abs [sensitivity-specificity] < 1%) for detecting studies of prognosis in MEDLINE in 2000

Search Strategy OVID search*

Sensitivity (%) Development Validation Diff (95% CI)

Specificity (%) Development Validation Diff (95% CI)

Precision (%) Development Validation Diff (95% CI)

Accuracy (%) Development Validation Diff (95% CI)


Best Sensitivity

incidence.sh.

90.1

79.7

1.7

79.7

OR exp mortality

82.3

79.7

1.6

79.7

OR follow-up studies.sh.

-7.8 (-17.9 to 2.3)

0

-0.1 (-0.5 to 0.5)

0

OR prognos:.tw.

OR predict:.tw.

OR course:.tw.

Best Specificity

prognos:.tw.

52.3

94.1

3.3

93.9

OR first episode.tw.

48.1

94.2

3.2

94.0

OR cohort.tw.

-4.2 (-18.6 to 10.3)

0.1 (-0.3 to 0.5)

-0.1 (-1.3 to 1.3)

0.1 (-0.4 to 0.5)

Best Optimization of Sensitivity & Specificity

prognosis.sh.

82.9

83.7

1.9

83.7

OR diagnosed.tw.

73.4

84.1

1.8

84.0

OR cohort:.mp.

-9.5 (-21.5 to 2.5)

-0.4 (-0.2 to 1.1)

-0.1 (-0.7 to 0.5)

0.3 (-0.2 to 1.0)

OR predictor:.tw.

OR death.tw.

OR exp models, statistical


*Search strategies are reported using Ovid's search engine syntax for MEDLINE. The PubMed syntax is embedded in PubMed's Clinical Queries (see Discussion). Diff = Difference, comparing the development and validation data sets using the iterative method of Miettinen and Nurminen for two independent binomial proportions. None of the differences were statistically significant. sh = subject heading; exp = explode, a search term that automatically includes closely related indexing terms; : = truncation; tw = textword (word or phrase appears in title or abstract); mp = multiple posting (term appears in title, abstract, or MeSH heading).

Wilczynski et al. BMC Medicine 2004 2:23   doi:10.1186/1741-7015-2-23

Open Data