Table 2.
Comparison of different improvement methods for smoking categories
| Category | 2017 CTADS general population survey (Alberta) N = 1444 |
CPCSSN hypertensive patients (Alberta) N = 48,377 |
|||
|---|---|---|---|---|---|
| Survey prevalence % (95% CI) |
Complete cases n (%) |
Method 1: MAR imputationa n (%) |
Method 2: MNAR imputationb n (%) |
Method 3: CPCSSN pattern matching n (%) |
|
| Current Smoker | 18.9 (14.8–22.9) | 6061 (21.2) | 12,255 (25.3) | 6061 (12.5) | 6729 (18.2) |
| Past/Former Smoker | 21.1 (17.0–25.1) | 6494 (22.7) | 9628 (19.9) | 11,305 (23.4) | 8868 (24.0) |
| Never Smoker | 60.1 (55.2–65.0) | 16,079 (56.2) | 26,494 (54.8) | 31,011 (64.1) | 20,846 (56.4) |
| Not Current | – | – | – | – | 502 (1.4) |
|
Missing, n/N (%) |
– | 19,743/48,377 (40.8) | 0/48,377 (0) | 0/48,377 (0) | 11,432/48,377 (23.6) |
CI confidence interval, CPCSSN Canadian Primary Care Sentinel Surveillance Networkm, CTADS Canadian Tobacco, Alcohol and Drug Survey
aMAR Missing at Random; assumes missing values are random for all smoking statuses (current, past, never)
bMNAR Missing not at Random; assumes all current smokers have been documented and missing values are random for never or past smokers