Skip to main content

Table 2 Sensitivity (sens), specificity (spec), discriminative performance (AUC) and calibration (Brier score, Hosmer-Lemeshow [H-L] goodness of fit p-value) for models based on two years of administrative data

From: The validity of using ICD-9 codes and pharmacy records to identify patients with chronic obstructive pulmonary disease

   

Standard for diagnosis of COPD

   

GOLD criterion

 

LLN criterion

Model*

  

(FEV1/FVC < 0.70)

 

(FEV1/FVC < LLN)

   

AUC

 

H-L

 

AUC

 

H-L

#

Input variables

Sens. (%)/Spec. (%)

(95% CI)

Brier

p-value

Sens. (%)/Spec. (%)

(95% CI)

Brier

p-value

1

≥1 outpatient ICD-9 code

76/67

0.75 (0.74-0.76)

0.198

 

79/62

0.71 (0.70-0.72)

0.198

 

2

≥2 outpatient ICD-9 code

70/73

0.75 (0.74-0.76)

0.199

 

73/69

0.72 (0.71-0.73)

0.196

 

3

≥3 outpatient ICD-9 codes

58/83

0.74 (0.73-0.75)

0.201

 

62/80

0.71 (0.70-0.72)

0.195

 

4

≥1 inpatient ICD-9 codes

55/67

0.62 (0.61-0.63)

0.235

 

6.5/99

0.57 (0.56-0.58)

0.230

 

5

≥1 outpatient ICD-9 codes

        
 

≥1 Ipratropium Bromide MDI

76/67

0.77 (0.76-0.78)

0.192

 

53/83

0.75 (0.74-0.76)

0.191

 

6

≥1 outpatient ICD-9 codes

        
 

≥1 albuterol MDI

76/67

0.76 (0.75-0.77)

0.196

 

68/70

0.74 (0.73-0.75)

0.195

 

7

≥1 outpatient ICD-9 code

        
 

≥1 inpatient ICD-9 code

        
 

≥3 ipratropium bromide MDI

        
 

≥6 albuterol MDI

68/76

0.78 (0.77-0.79)

0.190

 

60/79

0.76 (0.75-0.77)

0.188

 

8

Model 7 and age

72/74

0.79 (0.78-0.80)

0.187

0.86

60/80

0.77 (0.76-0.78)

0.187

0.50

9

Model 8 and smoking

73/74

0.79 (0.78-0.80)

0.185

0.65

60/81

0.77 (0.76-0.78)

0.185

0.34

  1. * All models are stratified by age ≥65 years. Input variables represent independent variables in the logistic regression equation for that particular model. All variables modeled as dichotomous (present/absent) except for age (continuous).
  2. AUC, area under the receiver operating characteristic curve; H-L, Hosmer-Lemeshow test; LLN, lower limit of normal; MDI, metered dose inhaler