Skip to main content

Table 1 Number of records and indicators of bibliographic search strategies applied in PubMed and EMBASE

From: A comparison of metrics and performance characteristics of different search strategies for article retrieval for a systematic review of the global epidemiology of kidney and urinary diseases

 

PubMed

EMBASEb

Manual search and unpublished data

Total by merging SuHeSS and FreeWoSS

By SuHeSS

By FreeWoSS

Intercept of FreeWoSS and SuHeSSa

Indicators of bibliographic search obtained by search strategy

 Number of bibliographic records

  Initial search

29,460

13,147

23,352

7039

4247

  Potentially useful abstracts

2056

1444

1806

1194

105

  Selected for full-text extraction

450

384

424

358

21

28

  Obtained as a full text and data extracted

381

325

360

304

14

28

 % from the total number of records on given stepc

  Initial search

87.4

39.0

69.3

20.9

12.6

  Potentially useful abstracts

95.1

66.8

83.6

55.3

4.9

  Selected for full-text extraction

95.5

81.5

90.0

76.0

4.5

  Obtained as a full text and data extracted

96.5

82.3

91.1

77.0

3.5

 Number of journals covered

  Initial search

2512

1685

2154

1134

935

6d

  Potentially useful abstracts

473

374

426

317

79

6d

  Selected for full-text extraction

168

154

160

146

21

6d

  Obtained as a full text and data extracted

127

116

121

111

14

6d

Core indicators of the search strategy based on extraction from full-text articles

 Number of countries

117

115

116

114

12

54

 Number of extracted data rows

7729

6507

7612

6390

163

12,696e

 Number of extracted data rows per one full-text article

20.3

20.0

21.1

21.0

11.6

453.4e

Number needed to retrieve (NNR) to obtainf:

 1 potentially useful article

14 (14–15)

9 (9–10)

13 (12–14)

6 (6–6)

40 (33–49)

1

 1 article intended for full-text extraction

65 (60–72)

34 (31–38)

55 (50–61)

20 (18–22)

202 (130–318)

1

 1 obtained full-text article with data extracted

77 (70–85)

40 (36–45)

65 (58–72)

23 (21–26)

303 (189–603)

1

  1. aintercept means the set of records captured both by SuHeSS and FreeWoSS
  2. bEMBASE results exclude records indexed by MEDLINE
  3. c100% refers to the number obtained by both PubMed and EMBASE
  4. dconcern only published data
  5. ehigh number of extracted rows and mean number of data rows per article for unpublished data resulted from extraction of information by fine grain age and sex categories based on a request from IHME for data modelling
  6. fnumbers in brackets indicate 95% confidence interval
  7. SuHeSS Subject headings based search strategy, FreeWoSS Free word search strategy
  8. Correspondence between description in the table and step number (see Methods): Initial search – step 1 and 2; Potentially useful abstracts – step 3; Selected for full-text extraction – step 5; Obtained as a full-text and data extracted – step 6 and 7