Concordance between administrative health data and medical records for diabetes status in coronary heart disease patients: a retrospective linked data study

Table 5 Characteristics associated with false negatives in administrative data on index admission

	ICD-9		ICD-10
Variable	False negatives, n (%)^*(n = 34)	Multivariable model, odds ratio (95% CI)^*	False negatives, n (%)^*n = 146	Multivariable model, odds ratio (95% CI)^*
Admission Type
Booked	12 (25.5)^†	3.35 (1.36, 8.28)	22 (29.8) ^†	1.96 (1.12, 3.44)
Emergency	22 (6.6)	-	124 (17.3)	-
In-hospital death
Yes	8 (19.5)^†	2.61 (0.97, 7.02)	15 (34.9) ^†	3.70 (1.74, 7.85)
No	26 (7.7)	-	131 (17.6)	-
Length of stay (days)
1-2	4 (7.3)^†	0.47 (0.14, 1.56)	57 (27.3)^†	1.66 (0.99, 2.77)
3-5	11 (5.3)	0.50 (0.21, 1.19)	62 (14.9)	0.91 (0.55, 1.49)
≥6	19 (16.4)	-	27 (16.6)	-
Principal diagnosis
MI	16 (8.3)^‡	0.29 (0.09, 0.86)	61 (16.4) ^†	0.32 (0.16, 0.61)
UA	9 (5.6)	0.24 (0.07, 0.76)	66 (18.0)	0.34 (0.18, 0.66)
Other CHD	9 (33.3)	-	19 (36.5)	-
Comorbidities
0-3	4 (5.3)	-	71 (23.4)^‡	2.41 (1.20, 4.81)
4-7	15 (7.5)		58 (16.5)	1.78 (0.92, 3.45)
≥8	15 (14.3)		17 (12.7)	-

*Percentage is the proportion of false negatives within each category for each variable (FN / FN + TP in each category). Multivariable models are adjusted for all variables which are significant univariable predictors of false negatives in each period, and all are included in the models as categorical variables. Comorbidities was not included in the multivariable model for the ICD-9 sample. †p < 0.05, ‡p < 0.0001.
ICD, International Classification of Diseases; CI, confidence interval; MI, myocardial infarction; UA, unstable angina; CHD, coronary heart disease; FN, false negative; TP, true positive.

ISSN: 1471-2288