Skip to main content
. Author manuscript; available in PMC: 2015 May 1.
Published in final edited form as: Med Care. 2014 May;52(5):e30–e38. doi: 10.1097/MLR.0b013e31825a8c22

Table 5.

Performance of ICD-9 coding algorithms for incident heart failure or cardiomyopathy among those patients exposed to anthracycline and/or trastuzumab therapy

ONLY EXPOSED PATIENTS ONLY UNEXPOSED PATIENTS
INCIDENT by Algorithm #2
(with first ICD-9 code occurring
AFTER date of breast cancer
diagnosis)
N=29 (weighted N=7)
INCIDENT by Algorithm #2
(with first ICD-9 code occurring AFTER
date of breast cancer diagnosis)

N=27 (weighted N=4)
INCIDENT by gold standard
(with first chart mention of HF/CM
or LVEF<50% occurring BEFORE
date of breast cancer diagnosis)
N=32 exposed, 14 unexposed
(weighted N= 5 exposed, 4 unexposed)
Sens = 57.8 (19.5–88.6)
Spec = 98.1 (95.2–99.3)
PPV = 42.1 (14.2–76.1)
NPV =99.0 (96.3–99.7)
Sens = 19.9 (2.0–75.0)
Spec = 98.0 (94.3–99.3)
PPV = 18.3 (1.9–72.6)
NPV = 98.2 (94.6–99.4)
INCIDENT by LVEF criteria
(with first documented
LVEF<50% occurring AFTER
date of breast cancer diagnosis)
N=28 exposed, 8 unexposed
(weighted N= 4 exposed, 2 unexposed)
Sens = 55.3 (15.6–89.2)
Spec = 97.8 (94.8–99.1)
PPV = 33.0 (9.5–69.9)
NPV =99.1 (96.5–99.8)
Sens = 4.7 (0.0–97.8)
Spec = 97.6 (93.9–99.1)
PPV = 2.1 (0.0–94.6)
NPV = 99.0 (95.6–99.8)