Skip to main content

Table 1 Estimated regression coefficients and standard errors (SE) of data with no values missing; when values are missing completely at random; when outcome blood pressure (BP) is missing at random; when covariate (baseline BP) is missing at random; and when outcome BP is missing not at random

From: When and how should multiple imputation be used for handling missing data in randomised clinical trials – a practical guide with flowcharts

Type of missingness Randomised groups Systolic blood pressure
(mmHg) at baseline
mean (SE) N 		Systolic blood pressure
(mmHg) at end of trial
mean (SE) N 		Parameters
Intercept
estimate
(SE)
P 		Baseline blood pressure
estimate
(SE)
P 		Intervention
estimate
(SE)
P
None Experimental
N = 103 		181.6 (2.90) N = 103 130.8 (3.17) N = 103 −2.48
(4.69)
P = 0.60 		1.013
(0.025)
P < 0.0001 		−50.8
(1.48)
P < 0.0001
Control
N = 97 		180.9 (2.98)
N = 97 		180.9 (3.13) N = 97
MCAR Experimental
N = 103 		181.6 (3.64) N = 74 131.0 (4.02) N = 74 −6.85
(5.22)
P = 0.19 		1.041
(0.028)
P < 0.0001 		−51.2
(1.69)
P < 0.0001
Control
N = 97 		181.8 (3.43) N = 75 182.3 (3.71) N = 75
MAR
(outcome missing) 		Experimental
N = 103 		181.6 (2.90) N = 103 129.7 (3.97) N = 72 −2.75
(5.13)
P = 0.59 		1.015
(0.028)
P < 0.0001 		−51.2
(1.66)
P < 0.0001
Control
N = 97 		181.0 (2.98) N = 97 180.9 (3.13) N = 97
MAR
(baseline missing) 		Experimental
N = 103 		181.6 (2.90) N = 103 130.8 (3.17) N = 103 −5.32
(5.56)
P = 0.34 		1.004
(0.034)
P < 0.0001 		−46.2
(2.22)
P < 0.0001
Control
N = 97 		156.2 (2.82) N = 36 180.9 (3.13) N = 97
MNAR
(outcome missing) 		Experimental
N = 103 		181.6 (2.90) N = 103 127.8 (3.24) N = 95 −8.13
(5.67)
P = 0.15 		1.026
(0.032)
P < 0.0001 		−47.6
(2.12)
P < 0.0001
Control
N = 97 		181.0 (2.98) N = 97 163.4 (5.37) N = 38
  1. The analyses used in all scenarios were a complete case analysis
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us

\