Skip to main content
. 2018 Aug 2;37(28):4142–4154. doi: 10.1002/sim.7913

Table 3.

Parameter estimates from the Clinical Practice Research Datalink example estimated from the treatment‐naïve model and the marginal structural model (MSM). All of the variables are those extracted at baseline (ie, the index date). Note that both models include a year‐specific intercept fitted as a restricted cubic spline

Variable Treatment Naïve Model (SE) MSM (SE)
Statin Initiation N/A −0.1002 (0.0103)
Female −0.5566 (0.0066) −0.5331 (0.0062)
Age  0.0711 (0.0003)  0.0723 (0.0002)
Atrial Fibrillation  0.4827 (0.0156)  0.4353 (0.0148)
Chronic Kidney Disease (stage 4/5)  0.3646 (0.0352)  0.3612 (0.0318)
Type I diabetes  0.7307 (0.0585)  0.5859 (0.0564)
Type II diabetes  0.5967 (0.0138)  0.5414 (0.0128)
Ethnicity
      White or not stated Reference Reference
      Asian −0.0840 (0.0700) −0.1936 (0.0679)
      Bangladesh  0.1343 (0.1557)  0.1191 (0.1504)
      Black −0.6454 (0.0568) −0.7092 (0.0536)
      Chinese −0.7970 (0.1629) −0.7965 (0.1553)
      Indian −0.0250 (0.0498)  0.0787 (0.0440)
      Mixed −0.6138 (0.1287) −0.6547 (0.1229)
      Other −0.3416 (0.0758) −0.2489 (0.0674)
      Pakistani  0.3351 (0.0794)  0.3109 (0.0766)
Family history of coronary heart disease  0.1391 (0.0088)  0.1172 (0.0082)
Hypertension  0.1390 (0.0078)  0.2553 (0.0072)
Rheumatoid arthritis  0.4200 (0.0237)  0.4072 (0.0220)
Systolic blood pressure  0.0109 (0.0002)  0.0104 (0.0002)

SE = standard error.