Skip to main content

Table 3 Methods used in studies for the handling of missing data

From: How are missing data in covariates handled in observational time-to-event studies in oncology? A systematic review

Missing data methods

Count

(%)*

Complete-case

79

(53)

Removed individuals with incomplete data for a subset of covariates

67

(45)

Multiple Imputation

33

(22)

Missing indicator

10

(7)

Worst or best case scenario1

2

(1)

Stochastic imputation

1

(1)

Mean value imputation

1

(1)

Mode value imputation

1

(1)

Growth models

1

(1)

Bayesian model incorporating handling of missing data

1

(1)

Full-information maximum likelihood estimation 2

1

(1)

Selection procedure3

1

(1)

Unclear

33

(22)

  1. *Percentages do not sum to 100 as there is overlap with some studies using more than one method.
  2. 1[11,12]
  3. 2[11]
  4. 3A selection model to account for missing data and time-varying covariates [13]
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us