‘Putting Life in Years’ (PLINY) telephone friendship groups research study: pilot randomised controlled trial

Table 4 Mean observed and imputed 6-month post-randomisation follow-up SF-36 MH outcomes by randomised group (N = 70)

	Group
	Intervention			Control
6-month SF-36 MH outcome	N	Mean	SD*	N	Mean	SD*	Unadjusted difference	Lower	Upper	Adjusted difference**	Lower	Upper
Observed data (N = 56)	26	77.5	18.4	30	70.7	21.2	6.5	-3.0	16.0	9.5	4.5	14.5
LOCF imputed data (N = 70))	35	78	16.6	35	69.6	22.5	8.3	-0.5	17.2	7.7	3.7	11.8
Regression imputed data (N = 70)	35	77.8	16.2	35	69.8	21.4	7.8	-0.6	16.2	7.6	3.6	11.6
Multiple imputation PMM (N = 70)	35	78.9	3.7	35	70.6	3.7	8.3	-0.6	17.2	8.0	2.8	13.3
Multiple imputation regression (N = 70)	35	77.3	3.2	35	69.7	3.9	7.6	-1.8	16.9	7.4	1.8	13.0
Per protocol data (N = 39)	9	73.9	17.5	30	70.7	21.2	3.2	-5.2	11.6	8.0	3.3	12.7

The Short Form (36) Health Instrument Mental Health Dimension (SF-36 MH) is scored on a 0 (poor) to 100 (good) health scale. All analyses use a marginal general linear model, with regression coefficients estimated using generalised estimating equations, with robust standard errors. Regression imputation based on a model with age sex and baseline Mental Health score. Multiple imputation based on 20 imputed data sets, with age, sex and baseline score as covariates, using predictive mean matching (PMM) or linear regression. *For the multiple imputation methods the SD is the standard error of the mean. **Adjusted for randomised group, age, sex and baseline score. LOCF, last observation carried forward.

ISSN: 1745-6215