Table 1 Quality assessment of prognostic models

STUDY

Loss to follow-up

   < 10%

10 (15%)

5 (16%)

5 (14%)

   >10%

19 (29%)

7 (23%)

12 (34%)

   Not reported

37 (56%)

19(61%)

18 (52%)

VARIABLES

Discussion about predictors

   Yes

21(32%)

11 (35%)

10(29%)

   No

45(68%)

20 (65%)

25 (71%)

Description of measurement of predictors

   Yes

12 (18%)

8 (26%)

3 (9%)

   No

54 (82%)

23(74%)

32 (91%)

Validity of outcome reported

   Yes

31 (47%)

14 (45%)

17(49%)

   No

20 (30%)

7 (23%)

13 (37%)

   Not applicable

15(23%)

10 (32%)

5 (14%)

Handling of missing data

   Estimated statistically

4 (6%)

4 (13%)

0

   Excluded

36(55%)

16(52%)

20 (57%)

   Not reported

26(39%)

11(35%)

15 (43%)

ANALYSIS

Multivariable analysis Stepwise

   Backwards

-

12 (39%)

N/A

   Forwards

-

3 (10%)

   Not specified

-

10 (32%)

Not reported

-

5 (16%)

Other

-

1 (3%)

Interactions examined

   Yes

-

4 (13%)

   Not reported

-

27 (87%)

Handling of predictors variables

   Continuous

-

6 (19%)

   Categorical

-

16 (52%)

   Not clear

-

9 (29%)

More than 10 events per variable

   Yes

-

9 (29%)

   No

-

16 (52%)

   Not reported

-

6 (19%)

EXTERNAL VALIDITY

Description of the sample

   Yes

55 (83%)

28 (90%)

27 (77%)

   No

11(17%)

3 (10%)

8 (23%)

Presentation of the prognostic model

   Normogram

1 (1%)

1 (3%)

0

   Simplified score

8 (12%)

4 (13%)

4 (11%)

   Figure

13 (20%)

1 (3%)

12 (34%)

   Regression formula

15 (23%)

12(39%)

3 (9%)

   Not explained

29 (44%)

13(42%)

16 (46%)

EXTERNAL VALIDITY

Performance reported A.U.C (Discrimination)

   Yes

-

18 (58%)

NA

   No

-

13(42%)

   C.I. presented

-

8 out of 18 (44%)

H-L (Calibration)

   Yes

-

7 (23%)

NA

   No

-

23(74%)

   Other

-

1 (3%)

Overall accuracy

   Yes

37 (56%)

15 (48%)

22 (63%)

   No

29 (44%)

16 (52%)

13 (17%)

Validation

   Yes

25 (38%)

17 (55%)

8 (23%)

   External

7 (11%)

7 (23%)

0

   No

41 (62%)

14 (45%)

27 (77%)