Skip to main content

Table 2 Comparison of different improvement methods for smoking categories

From: Methods to improve the quality of smoking records in a primary care EMR database: exploring multiple imputation and pattern-matching algorithms

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)

  1. CI confidence interval, CPCSSN Canadian Primary Care Sentinel Surveillance Networkm, CTADS Canadian Tobacco, Alcohol and Drug Survey
  2. aMAR Missing at Random; assumes missing values are random for all smoking statuses (current, past, never)
  3. bMNAR Missing not at Random; assumes all current smokers have been documented and missing values are random for never or past smokers