Measuring depression in nursing home residents with the MDS and GDS: an observational psychometric study

BMC Geriatrics

Table 1 Sample statistics and psychometric properties of Geriatric Depression Scale and MDS Depression Rating Scale.

Sample statistic or psychometric property	Cognitively Impaired Group (n = 495)	Cognitively Intact Group (n = 209)	All Participants (N = 704)
Number (%) with missing data on all items
GDS-15	81 (16%)	35 (17%)	116 (16%)
GDS-5	110 (22%)	5 (2%)	115 (16%)*
E1SUM	3 (1%)	2 (1%)	5 (1%)
DRS	0 (0%)	0 (0%)	0 (0%)
Number (%) with complete data on all items
GDS-15	139 (28%)	111 (53%)	250 (36%)*
GDS-5	223 (45%)	157 (75%)	380 (54%)*
E1SUM	492 (99%)	207 (99%)	699 (99%)
DRS	494 (100%)	209 (100%)	703 (100%)
Number (%) with majority of scale items not missing
GDS-15	357 (72%)	203 (97%)	560 (80%)*
GDS-5	352 (71%)	200 (96%)	552 (78%)*
E1SUM	495 (100%)	209 (100%)	704 (100%)
DRS	495 (100%)	209 (100%)	704 (100%)
Mean* (SD)
GDS-15	4.7 (3.5)	4.9 (3.4)	4.8 (3.5)
GDS-5	2.0 (1.5)	2.1 (1.6)	1.5 (1.6)
E1SUM	3.9 (3.8)	2.7 (3.4)	3.6 (3.7)*
DRS	1.9 (2.1)	1.8 (2.3)	1.9 (2.2)
Proportion at floor
GDS-15	0.081	0.053	0.072
GDS-5	0.166	0.167	0.166
E1SUM	0.206	0.349	0.249*
DRS	0.356	0.397	0.368
Alpha – internal consistency reliability – row-wise complete cases only
GDS-15	0.799	0.781	0.791
GDS-5	0.609	0.597	0.602
E1SUM	0.695	0.738	0.708
DRS	0.542	0.672	0.583*
Alpha – internal consistency reliability – row-wise mean substitution for missing data^†
GDS-15	0.825	0.798	0.814
GDS-5	0.663	0.634	0.651
E1SUM	0.695	0.738	0.708
DRS	0.542	0.672	0.583*
Correlation coefficients
(GDS-5, GDS-15)	0.858	0.852	0.856
(DRS, GDS-15)	0.073	0.098	0.080
(DRS, GDS-5)	0.065	0.064	0.062
(E1SUM, GDS-15)	0.068	0.096	0.072
(E1SUM, GDS-5)	0.055	0.058	0.049
(E1SUM, DRS)	0.850	0.940	0.865*

Abbreviations: GDS-15, 15-item Geriatric Depression Scale; GDS-5, 5-item Geriatric Depression Scale; DRS, Minimum Data Set (MDS) Depression Rating Scale; E1SUM denotes the sum of all mood and anxiety items in section E1 of the MDS; SD, standard deviation
* For each respondent, if a majority of the items were not missing, any missing item was replaced with the mean of the non-missing items.
† Missing item responses replaced with mean score for all respondents 'non-missing' on that item.

ISSN: 1471-2318