Themes

Imputation of missing data in clinical trials

What is this about?

Missing data is always a limitation in the interpretation of clinical trial results. This missing data may seriously affect the inference from clinical trials. Therefore, it is necessary to establish the criteria to handle missing data prior to the clinical trial. Imputation is a potential tool to overcome the bias from missing data but it must be carefully used.

Why is this important?

Missing data are unavoidable in clinical trials. Frequently, complete cases analysis is used only including individuals with no missing data ^[1]. However, that can generate bias and can lead to exclude several individuals, causing loss of precision and power ^[2]. The risk of bias from missing data depends on the cause ^[3]:

Missing completely at random: There are no systematic differences between the missing values and the observed values.

Missing at random: Any systematic difference between the missing values and the observed values can be explained by differences in observed data.

Missing not at random: Systematic differences remain between the missing values and the observed values.

The determination of the type of missing values is difficult due to the nature of missing values ^[4]. Therefore, practical guidelines are needed to deal with missing data.

For whom is this important?

All stakeholders in researchClinical ethics consultantsClinical researchersComputer scientistsEthics committee membersResearch Ethics Committees

What are the best practices?

Planning stage of the clinical trial

To reduce the risks of missing data in the panning of the clinical trial, statistical analyses should be specified, key data items should be identified, and the procedures to prevent missing data ^[5].

Analysis stage

There are different strategies to deal with missing data that will depend on the specific clinical trial and type of missing data:

1. Complete cases analysis could be used when the proportions of missing data are below 5%and the potential impact of the missing data is negligible ^[6].

2. Single imputation replaces missing values by a value defined by a certain rule. However, this method ignores the data variation and can potentially introduce bias and should be used with great caution ^[7].

3. When the missing data accomplish certain characteristics, multiple imputation may be used to minimize bias . Missing values are replaced by a random sample of plausible values imputations. There are several multiple imputation methodologies that must be chosen according to the variable with missing values ^[5].

To conclude, handling missing data validly is an important, yet difficult and complex, task. This theme showed different strategies to handle missing data but always statistical expertise’s advice is needed.

↑ Bell ML, Fiero M, Horton NJ, Hsu CH. Handling missing data in RCTs; A review of the top medical journals. BMC Med Res Methodol [Internet]. 2014 Nov 19 [cited 2022 May 30];14(1):1–8. Available from: https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/1471-2288-14-118
↑ Sterne JAC, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ [Internet]. 2009 Jun 29 [cited 2022 May 30];338(7713):157–60. Available from: https://www.bmj.com/content/338/bmj.b2393
↑ Little RJA, Rubin DB. Statistical Analysis with Missing Data, 3rd Edition | Wiley [Internet]. 1991 [cited 2022 May 30]. Available from: https://www.wiley.com/en-us/Statistical+Analysis+with+Missing+Data%2C+3rd+Edition-p-9780470526798
↑ Little RJ, D’Agostino R, Cohen ML, Dickersin K, Emerson SS, Farrar JT, et al. The Prevention and Treatment of Missing Data in Clinical Trials. N Engl J Med [Internet]. 2012 Oct 4 [cited 2022 May 30];367(14):1355–60. Available from: https://www.nejm.org/doi/full/10.1056/nejmsr1203730
↑ ^5.0 ^5.1 Jakobsen JC, Gluud C, Wetterslev J, Winkel P. When and how should multiple imputation be used for handling missing data in randomised clinical trials - A practical guide with flowcharts. BMC Med Res Methodol [Internet]. 2017 Dec 6 [cited 2022 May 30];17(1):1–10. Available from: https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-017-0442-1
↑ Jakobsen JC, Gluud C, Winkel P, Lange T, Wetterslev J. The thresholds for statistical and clinical significance - A five-step procedure for evaluation of intervention effects in randomised clinical trials. BMC Med Res Methodol [Internet]. 2014 Mar 4 [cited 2022 May 30];14(1):1–12. Available from: https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/1471-2288-14-34
↑ Dziura JD, Post LA, Zhao Q, Fu Z, Peduzzi P. Strategies for Dealing with Missing Data in Clinical Trials: FromDesign to Analysis. Yale J Biol Med [Internet]. 2013 Sep [cited 2022 May 30];86(3):343. Available from: /pmc/articles/PMC3767219/

Víctor Urbiola-Salvador contributed to this theme. Latest contribution was May 30, 2022

Other information

Who

Victor Urbiola-Salvador

When

30/05/2022

Virtues & Values

Good Practices & Misconduct

Research Area

[1] Bell ML, Fiero M, Horton NJ, Hsu CH. Handling missing data in RCTs; A review of the top medical journals. BMC Med Res Methodol [Internet]. 2014 Nov 19 [cited 2022 May 30];14(1):1–8. Available from: https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/1471-2288-14-118

[2] Sterne JAC, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ [Internet]. 2009 Jun 29 [cited 2022 May 30];338(7713):157–60. Available from: https://www.bmj.com/content/338/bmj.b2393

[3] Little RJA, Rubin DB. Statistical Analysis with Missing Data, 3rd Edition | Wiley [Internet]. 1991 [cited 2022 May 30]. Available from: https://www.wiley.com/en-us/Statistical+Analysis+with+Missing+Data%2C+3rd+Edition-p-9780470526798

[4] Little RJ, D’Agostino R, Cohen ML, Dickersin K, Emerson SS, Farrar JT, et al. The Prevention and Treatment of Missing Data in Clinical Trials. N Engl J Med [Internet]. 2012 Oct 4 [cited 2022 May 30];367(14):1355–60. Available from: https://www.nejm.org/doi/full/10.1056/nejmsr1203730

[:0-5] 5.0 ^5.1 Jakobsen JC, Gluud C, Wetterslev J, Winkel P. When and how should multiple imputation be used for handling missing data in randomised clinical trials - A practical guide with flowcharts. BMC Med Res Methodol [Internet]. 2017 Dec 6 [cited 2022 May 30];17(1):1–10. Available from: https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-017-0442-1

[6] Jakobsen JC, Gluud C, Winkel P, Lange T, Wetterslev J. The thresholds for statistical and clinical significance - A five-step procedure for evaluation of intervention effects in randomised clinical trials. BMC Med Res Methodol [Internet]. 2014 Mar 4 [cited 2022 May 30];14(1):1–12. Available from: https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/1471-2288-14-34

[7] Dziura JD, Post LA, Zhao Q, Fu Z, Peduzzi P. Strategies for Dealing with Missing Data in Clinical Trials: FromDesign to Analysis. Yale J Biol Med [Internet]. 2013 Sep [cited 2022 May 30];86(3):343. Available from: /pmc/articles/PMC3767219/

[1]

[2]

[3]

[4]

[5]

[6]

[7]