Research letter: Validity of Administrative data
1 Supplemental Table 1: Definitions using information available in administrative data
Definition of Outcomes
Stroke primary discharge diagnoses of
1. ischemic stroke, defined as an ICD-9 CM code of 433.x1, 434.x (excluding 434.x0), or 436;
2. intracerebral hemorrhage (ICH), defined as 431;
3. subarachnoid hemorrhage (SAH), defined as 430 exclude those with concurrent diagnosis codes of
1. 800-804, 850-854 (diagnoses of total brain injury/trauma) or 2. diagnosis code of 432 (epidural and subdural hematoma) or
3. diagnosis code 437 (other and ill-define cerebrovascular disease), or 4. diagnosis code 438 (late effects of cerebrovascular disease).
AMI primary discharge diagnoses of
1. 410.x Definition of Exposure
Metformin 1. fill in the 90 days prior to the cardiovascular hospitalization 2. fill in the 365 days prior to the cardiovascular hospitalization Sulfonylureas
Insulin
Definition of Covariates
Smoking 1. ICD 9-CM code for tobacco use (305.1, V15.82, 989.84) or 2. prescription for varenicline tartrate in the past 365 days
3. prescription for nicotine replacement therapy in the past 365 days Aspirin use 1. fill of aspirin in the 365 days prior to the cardiovascular hospitalization
Research letter: Validity of Administrative data
2 Supplemental Table 2: Validity of exposure classification using administrative pharmacy fill data compared to that recorded in medical chart among 400 patients with a cardiovascular
hospitalization.
Exposure status based on Administrative data*
Exposure Recorded in Medical Chart*
Metformin use No Metformin use Total PPV/ NPV (95% CI) Metformin fill
within 90 days
Yes 83 11 94 PPV= 88.3 (80.2, 93.3)
No 22 284 306 NPV=92.8 (89.4, 95.2)
Metformin fill within 365 days
Yes 101 42 143 PPV=70.6 (62.7, 77.5)
No 4 253 257 NPV=98.4 (96.0, 99.4)
Total 105 295 400
Sulfonylurea use No Sulfonylurea use
Sulfonylurea fill within 90 days
Yes 115 19 134 PPV=85.8 (78.9, 90.7)
No 25 241 266 NPV=90.6 (86.5, 93.6)
Sulfonylurea fill within 365 days
Yes 132 41 173 PPV=76.3 (69.4, 82.0)
No 8 219 227 NPV=96.5 (93.2, 98.2)
Total 140 260 400
Insulin use No Insulin use Insulin fill
within 90 days
Yes 88 12 100 PPV=88.0 (80.2, 93.0)
No 15 285 300 NPV=95.0 (91.9, 96.9)
Insulin fill within 365 days
Yes 97 25 122 PPV=79.5 (71.5, 85.7)
No 6 272 278 NPV=97.8 (95.4, 99.0)
Total 103 297 400
* The data in each column represents the number of patients using each exposure as abstracted from chart and is considered the referent standard. The row is the number of patients identified as using each exposure by each database definition (drug prescription fill within 90 days of event or 365 days of event)
Research letter: Validity of Administrative data
3 Supplemental Table 3: Validity of covariate information using administrative data indicator algorithm compared to that recorded in medical chart among 400 patients with a cardiovascular hospitalization
Administrative data indicator present within past 365 days
Chart Data N=400 cases identified Current Aspirin
use
No Aspirin use/ unknown
Total PPV/NPV (95% CI) Sensitivity/
Specificity (95% CI) Aspirin use
Yes 158 39 197 PPV= 80.2 (72.0, 86.7) Sn=79.4 (73.2, 84.4)
No 41 162 203 NPV= 79.8 (73.8, 84.7) Sp=80.6 (74.6, 85.5)
Total 199 201 400
Current Smoker
Former/ Non Smoker Tobacco use
Yes 83 32 115 PPV= 72.2 (63.4, 79.5) Sn=55.7 (47.7, 63.4)
No 66 219 285 NPV= 76.8 (71.6, 81.4) Sp=87.3 (82.6, 90.8)
Total 149 251 400
Research letter Validity of Administrative data
4 Supplemental Table 4: Age adjusted proportion of current smoking and aspirin use by chart review and administrative data indicator and PPV of administrative data indicator by antidiabetic class.
* Probabilities of outcome derived from an age adjusted logistic regression model with no drug fill in 90 days as the reference value.
The outcomes are either smoking or aspirin use as reported in patient chart or presence of the indicator as derived by the administrative algorithm.
† P=0.007
‡ P=0.006
§ P=0.002
No antidiabetic drug N=158
Single agent Metformin
N=34
Single agent Sulfonylurea
N=70
Metformin+
Sulfonylurea N=37
Single agent Insulin
N=57
2 and 3 drug Insulin combos
N=44
Median age 64.4
(54.9, 72.5)
58.1 (55.7, 63.5)
71.7 (59.6, 78.9)
60.3 (54.5, 69.3)
66.3 (61.4, 75.8)
64.8 (56.4, 69.4) Age adjusted proportion of current
smokers, by chart review*
44.9 (36.7, 53.4)
29.4 (16.7, 46.4)
34.7 (23.3, 48.1)
43.2 (27.4, 60.5)
22.3 (12.8, 35.9)†
27.4 (16.0, 42.8) Age adjusted proportion with current
smoker indicator*
35.3 (27.9, 43.6)
23.1 (12.3, 39.3)
25.3 (15.7, 38.0)
32.9 (19.6, 49.8)
20.5 (11.6, 33.7)
13.1 (6.1, 26.0)‡
PPV of current smoker indicator 75.9 (63.5, 85.0)
50.0 (23.7, 76.3)
80.0 (54.8, 92.9)
85.7 (60.1, 96.0)
45.5 (21.3, 71.9)
71.4 (35.9, 91.8) Age adjusted proportion of aspirin use, by
chart review *
43.9 (36.3, 51.8)
47.8 (31.7, 64.4)
53.3 (41.3, 64.9)
49.1 (33.6, 64.8)
60.9 (47.6, 72.7)
53.9 (39.0, 68.1) Age adjusted proportion of aspirin use
indicator*
45.2 (37.5, 53.0)
44.3 (28.7, 61.1)
54.1 (42.1, 65.6)
46.0 (30.9, 62.0)
68.3 (55.1, 79.1) §
41.9 (28.2, 57.0)
PPV of aspirin use indicator 71.4
(59.9, 80.7)
80.0 (54.8, 92.9)
84.2 (69.6, 92.6)
82.4 (58.9, 93.8)
84.6 (70.2, 92.8)
94.4 (74.2, 99.0)