Skip to main content

Table 3 Distribution of erroneous outcomes of the detection algorithm of cardiovascular disease or stroke through the records of an electronic medical record

From: A rule-based electronic phenotyping algorithm for detecting clinically relevant cardiovascular disease cases

False negative results (FN) n (%)

Proportion

Total errors (FP + FN) (n = 98)

Total FN (n = 32)

CaVD FN (n = 21)

CeVD FN (n = 11)

Codes not included in the algorithm

20 (20.41)

20 (62.5)

15 (71.43)

5 (45.45)

Event included in the free-text clinical notes, but not coded

9 (9.18)

9 (28.13)

5 (23.81)

4 (36.36)

Coded terms not detected

2 (2.04)

2 (6.25)

1 (4.76)

1 (9.09)

Erroneous date

1 (1.02)

1 (3.13)

0 (0)

1 (9.09)

False positive results (FP) n (%)

Proportion

Total errors (FP + FN) (n = 98)

Total FP (n = 66)

CaVD FP (n = 39)

CeVD FP (n = 27)

Dismissed diagnosesa

43 (43.88)

43 (65.15)

33 (84.61)

10 (37.04)

Incidental finding without clinical relevance

13 (13.27)

13 (19.7)

0 (0)

13 (48.15)

Coded terms without correlation in the clinical notes

8 (8.16)

8 (12.12)

4 (10.26)

4 (14.81)

Incorrect abbreviations

2 (2.04)

2 (3.03)

2 (5.13)

0 (0)

  1. FP false positive result, FN false negative result, CaVD cardiovascular events, CeVD cerebrovascular events
  2. aFor example, patients in whom an acute myocardial infarction was suspected but later on the diagnosis was rejected due to new information