There were 89 subjects in the protocol group and 85 in the control group with both groups well matched for age, sex, weight, Parsonnet cardiac risk score, APACHE II score, and type of surgery.
Groups were well matched for subject numbers, gender and age, with the dementia groups well matched for age at onset and duration of dementia, Mini-Mental State Examination and total Cambridge Cognitive Examination scores.
Whereas the typical Alzheimer's disease and healthy control groups were well matched for age, the PCA group was on average significantly younger than both the control group [beta = 6.18, 95% confidence interval (CI) 4.5 to 7.8, P < 0.001] and the typical Alzheimer's disease group [beta = 5.67, CI 3.9 to 7.4, P < 0.001].
Both groups were well matched for demographics: intervention and control group respectively, mean age of 68.9 (±14.7) vs. 70.9 (±13.2), p = 0.08; admission after cardiac surgery 21 (7.1%) vs. 26 (8.3%), p = 0.40; admission after percutaneous interventions 60 (20.3%) vs. 59 (18.8%), p = 0.55.
When all study subjects were divided into 3 groups as the AA, AC, and CC group in rs2283228 and rs2237895, or the CC, CT, and TT group in rs2237897 and rs2237892, we found that 3 groups were well matched for mean age and gender composition.
We recognise that there were significant differences between the pre- and post-protocol groups with respect to the indications for warfarin, however, patients in both groups were well matched for age, gender and ASA grade.
The case and control groups were well matched for most variables, but there were frequency differences between the groups in gender, year of recruitment, and history of having malaria previously (Table 1), so these variables were incorporated in a logistic regression analysis with any polymorphisms that emerged as different between the groups in a univariate analysis (Tables 2 and 3).
Results The two groups were well matched for baseline characteristics, and withdrawal rates were similar in both groups (5%).
The experimental groups were well matched for age; males were significantly overrepresented in the byFTD group.
As the goal of stratification is to remove differences in covariates between the two groups, covariates were examined within each quintile for statin initiators and non-initiators to ensure that the two groups were well matched for each covariate; any variables that were not balanced within a quintile were included in the final multivariate models as first order interaction terms.
Overall, the groups were well matched for age, sex, duration of symptoms, pain severity, and pain attack frequency, and there were no significant differences at baseline (Table 1).
