A systematic survey shows that reporting and handling of missing outcome data in networks of interventions is poor

Spineli, Loukia M.; Yepes-Nuñez, Juan J.; Schünemann, Holger J.

doi:10.1186/s12874-018-0576-9

Table 2 Characteristics of 273 selected network meta-analyses

From: A systematic survey shows that reporting and handling of missing outcome data in networks of interventions is poor

Characteristic	Levels		Total		2009 (n = 10)		2010 (n = 14)		2011 (n = 35)		2012 (n = 29)		2013 (n = 44)		2014 (n = 76)		2015 (n = 44)		2016 (n = 13)		2017 (n = 8)
Characteristic	Levels		n	%	2009 (n = 10)		2010 (n = 14)		2011 (n = 35)		2012 (n = 29)		2013 (n = 44)		2014 (n = 76)		2015 (n = 44)		2016 (n = 13)		2017 (n = 8)
Addressing and handling missing outcome data
Reported choice of primary analysis in the context of MOD	Intention-to-treat analysis		105	38.5	3	2.9	2	1.9	13	12.4	12	11.4	24	22.9	29	27.6	16	15.2	3	2.9	3	2.9
	Modified intention-to-treat analysis		4	1.5	0	0.0	0	0.0	0	0.0	1	25.0	0	0.0	2	50.0	1	25.0	0	0.0	0	0.0
	Per-protocol analysis		4	1.5	0	0.0	0	0.0	1	25.0	1	25.0	0	0.0	1	25.0	0	0.0	0	0.0	1	25.0
	Not mentioned		160	58.6	7	4.4	12	7.5	21	13.1	15	9.4	20	12.5	44	27.5	27	16.9	10	6.2	4	2.5
We judged the actual method of primary analysis rather than the reported method to be possibly	Available case analysis with LOCF		6	2.2	0	0.0	2	33.3	2	33.3	1	16.7	0	0.0	0	0.0	1	16.7	0	0.0	0	0.0
	Available case analysis without LOCF		15	5.5	0	0.0	3	20.0	3	20.0	1	6.7	2	13.3	3	20.0	2	13.3	1	6.7	0	0.0
	Imputation method	with LOCF	2	0.7	0	0.0	0	0.0	1	50.0	0	0.0	1	50.0	0	0.0	0	0.0	0	0.0	0	0.0
	Imputation method	without LOCF	14	5.1	1	7.1	0	0.0	3	21.4	3	21.4	2	14.3	2	14.3	3	21.4	0	0.0	0	0.0
	Modified intention-to-treat analysis		1	0.4	0	0.0	0	0.0	0	0.0	1	100	0	0.0	0	0.0	0	0.0	0	0.0	0	0.0
	Outcome analyzed as reported in the included trials		57	20.9	3	5.3	5	8.8	10	17.5	7	12.3	11	19.3	12	21.1	6	10.5	3	5.3	0	0.0
	Unclear		178	65.2	6	3.4	4	2.2	16	9.0	16	9.0	28	15.7	59	33.1	32	18.0	9	5.1	8	4.5
If applicable, which imputation method was judged to be used (16 NMAs)	All MOD as non-events		8	50.0	1	12.5	0	0.0	2	25.0	1	12.5	1	12.5	1	12.5	2	25.0	0	0.0	0	0.0
	All MOD as events		1	6.2	0	0.0	0	0.0	0	0.0	0	0.0	1	100	0	0.0	0	0.0	0	0.0	0	0.0
	Not mentioned		7	43.8	0	0.0	0	0.0	2	28.6	2	28.6	1	14.3	1	14.3	1	14.3	0	0.0	0	0.0
If imputation of MOD was employed (16 NMAs), the scenarios considered in the network are	Common across trials and interventions		9	56.3	1	11.1	0	0.0	2	22.2	1	11.1	2	22.2	1	11.1	2	22.2	0	0.0	0	0.0
	Unclear		7	43.7	0	0.0	0	0.0	2	28.6	2	28.6	1	14.3	1	14.3	1	14.3	0	0.0	0	0.0
For the scenarios selected to impute MOD (16 NMAs), the reviewer considered relevant trial information on the reasons for dropout	No		9	56.3	1	11.1	0	0.0	2	22.2	1	11.1	2	22.2	1	11.1	2	22.2	0	0.0	0	0.0
	Unclear		7	43.7	0	0.0	0	0.0	2	28.6	2	28.6	1	14.3	1	14.3	1	14.3	0	0.0	0	0.0
Sensitivity analysis on missing outcome data
Strategy reported to handle MOD in a sensitivity analysis^a	Trial exclusion		7	2.6	2	28.6	0	0.0	1	14.3	0	0.0	1	14.3	2	28.6	1	14.3	0	0.0	0	0.0
	Available case analysis		5	1.8	0	0.0	1	20.0	1	20.0	0	0.0	0	0.0	2	40.0	1	20.0	0	0.0	0	0.0
	Imputation		3	1.1	0	0.0	1	33.3	0	0.0	1	33.3	0	0.0	0	0.0	1	33.3	0	0.0	0	0.0
	Modified intention-to-treat analysis with other definitions		2	0.7	0	0.0	0	0.0	0	0.0	1	50.0	0	0.0	0	0.0	1	50.0	0	0.0	0	0.0
	No sensitivity analysis		257	94.1	8	3.1	13	5.1	33	12.8	27	10.5	43	16.7	72	28.0	40	15.6	13	5.1	8	3.1
If sensitivity analysis applied (16 NMAs), the authors reported any changes in the inferences after sensitivity analysis	Yes, it was reported that no changes were detected		13	81.2	1	7.7	0	0.0	2	15.4	2	15.4	1	7.7	3	23.1	4	30.8	0	0.0	0	0.0
	Yes, it was reported that changes were detected		2	12.5	0	0.0	1	50.0	0	0.0	0	0.0	0	0.0	1	50.0	0	0.0	0	0.0	0	0.0
	Not mentioned		1	6.3	1	100	0	0.0	0	0.0	0	0.0	0	0.0	0	0.0	0	0.0	0	0.0	0	0.0
Explanation given by the reviewers to support their strategy to handle MOD in primary and sensitivity analysis for the selected outcome	Yes, they provided an explanation		12	4.4	2	16.7	1	8.3	0	0.0	2	16.7	1	8.3	4	33.3	1	8.3	0	0.0	1	8.3
	No explanation is given		112	41.0	2	1.8	1	0.9	16	14.3	12	10.7	24	21.4	32	28.6	18	16.1	4	3.6	3	2.7
	Not applicable		149^b	54.6	6	4.0	12	8.1	19	12.8	15	10.1	19	12.8	40	26.8	25	16.8	9	6.0	4	2.7
Availability of primary and missing outcome data for extraction
Provided numerical information on MOD for post-hoc analysis	Extractable (i.e., at arm- or trial-level of every trial)		55	20.1	4	7.3	5	9.1	13	23.6	4	7.3	8	14.5	12	21.8	7	12.7	2	3.6	0	0.0
	Not usable information (e.g., at intervention level)		27	9.9	0	0.0	1	3.7	4	14.8	4	14.8	5	18.5	8	29.6	2	7.4	1	3.7	2	7.4
	No numerical information on MOD		191	70.0	6	3.1	8	4.2	18	9.4	21	11.0	31	16.2	56	29.3	35	18.3	10	5.2	6	3.1
Data extraction for the selected primary outcome is judged to be overall	Extractable^c		39	14.3	3	7.7	3	7.7	8	20.5	3	7.7	8	20.5	8	20.5	5	12.8	1	2.6	0	0.0
	Contact authors for raw outcome data^d		8	2.9	1	12.5	2	25.0	2	25.0	0	0.0	0	0.0	1	12.5	2	25.0	0	0.0	0	0.0
	Contact authors for raw MOD^e		4	1.5	0	0.0	0	0.0	1	25.0	0	0.0	0	0.0	3	75.0	0	0.0	0	0.0	0	0.0
	Not extractable^f		222	81.3	6	2.7	9	4.1	24	10.8	26	11.7	36	16.2	64	28.8	37	16.7	12	5.4	8	3.6

Abbreviations: LOCF last observation carried forward, MOD missing outcome data, NMA network meta-analysis
^aMultiple selections have been applied
^b120 systematic reviews did not mention any strategy to address missing outcome data and 29 systematic reviews addressed missing outcome data only as a secondary outcome
^cBoth primary and missing outcome data are provided for each arm in every trial
^dNo raw primary outcome data are provided but missing outcome data are provided for each arm of every (or some) trial(s)
^ePrimary outcome data are provided at arm-level for every trial but missing outcome data are provided at trial-level
^fPrimary outcome data are provided at contrast-level in each trial with trial- or intervention-level or no information at all on missing outcome data; primary outcome data are provided at arm-level for every trial with intervention-level or no information at all on missing outcome data; no raw primary outcome data are provided as well as no information on missing outcome data

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com

BMC Medical Research Methodology

Contact us