Effect of Cognitively Stimulating Activities on Symptom Management of Delirium Superimposed on Dementia

A Randomized Controlled Trial

Ann Kolanowski, PhD; Donna Fick, PhD; Mark Litaker, PhD; Paula Mulhall, RN; Linda Clare, PhD; Nikki Hill, PhD; Jacqueline Mogle, PhD; Malaz Boustani, MD; David Gill, MD; Andrea Yevchak-Sillner, PhD


J Am Geriatr Soc. 2016;64(12):2424-2432. 

In This Article


Participant flow through the study is depicted in Figure 1. Seventeen (6%) participants withdrew from the study. Reasons for withdrawal were similar in both groups. Table 1 lists participant demographic and baseline clinical characteristics and usual care that the total sample received and according to group. Baseline values for cognitive and physical function outcomes are shown in Appendix Table S1. The majority of participants were elderly, white, and female, with mild to moderate dementia. The adjudication panel reached agreement on 98.9% of the dementia diagnoses and 100% of the delirium diagnoses. There were no differences between the groups on any demographic or clinical characteristic at baseline or usual care received with the exception of CLOX 2 (constructional praxis) score, which was significantly greater in the intervention group (P = .02). No adverse events were reported in either group.

Figure 1.

Flow of primary participants (N = 283) through study.

Very little information was missing; 92.6% of all possible CAM assessments, 95.3% of all possible DRS assessments, 86.3% of all possible MoCA assessments, 79.1% of all possible CLOX assessments, and 95.8% of all possible Barthel Index assessments were obtained. There were no differences in missing data between the groups, and no evidence was found that missing data were due to any substantive correlations with demographic or clinical characteristics.

There was very good adherence to the intervention, and the intervention itself did not cause any withdrawals. Intervention group participants participated in an average of 70.5% of total sessions possible during their PAC stays. On average, 13.0 ± 8.2 intervention sessions that lasted for 19.1 ± 13.4 minutes were delivered to participants. Participation during these sessions was primarily active (mean 2.7 ± 0.7 on 3-point scale; 3 = active, 0 = dozing). There was no association between participant clinical and demographic characteristics and number of intervention sessions they participated in. Interventionists achieved between 95%, and 100% adherence on specified intervention delivery elements.


For participants with 2 or more PAC days, 86.3% (120/139) of intervention and 87.1% (122/140) of control participants experienced a remission of delirium at some point over the PAC stay (CAM = 0 or 1 for ≥2 consecutive days). Figure 2 is a graph of the Kaplan-Meier survival analysis depicting time to first remission of delirium according to group. It took intervention participants an average of 6.88 days (95% CI = 6.14–7.61 days) and usual care participants 7.39 days (95% CI = 6.47–8.31 days) to reach two consecutive days without delirium (P = .89, log-rank test). Despite a relatively early first remission, only 37.7% of intervention participants and 34.8% of usual care participants did not have symptoms of delirium at discharge or completion of the intervention period. Mean percentage of delirium-free days was also examined, and it was found that 64.8% (95% CI = 59.6–70.1) of intervention days and 68.7% (95% CI = 63.9–73.6) of control days were delirium free (P = .37, Wilcoxon rank sum test).

Figure 2.

Time to first remission.

Figure 3 is a graph of mean DRS over time according to group. Participants exhibited primarily subsyndromal delirium, defined as a score between 8 and 15 on the DRS (mean score: intervention: 10.77, 95% CI = 10.10–11.45; control: 11.15, 95% CI = 10.50–11.80; difference 0.37, 95% CI = 0.56–1.31, P = .43). Full delirium was observed on 14.5% of intervention days and 14.8% of control days.

Figure 3.

Mean delirium severity score.

Table 2 shows the cognitive and physical function outcomes. Statistically significant differences between the groups found for executive function and constructional praxis (CLOX 1 and 2) favored the intervention. Cohen effect size value suggested a small effect for CLOX 1 (d = 0.19) and a small to moderate effect for CLOX 2 (d = 0.32). After adjusting for baseline differences in CLOX 2, the difference between the groups for constructional praxis was no longer significant. There were no differences between the groups on other cognitive measures or physical function.

Length of stay was nonnormally distributed and depended on the facility (χ 2(1) = 71.77, P < .001). Excluding withdrawals (n = 17) and accounting for facility, the model estimated average length of stay was 36.1 days for intervention and 53.1 days for usual care (b = 0.39, SE = 0.15, P = .01). To avoid overinterpretation of these results, group assignments were examined according to site and it was found that the groups were distributed across facilities equivalently χ 2(7) = 4.35, P = .74). No differences were found in mortality (intervention: n = 20 deaths (15.4%); control: n = 17 deaths (12.5%) (P = .61, χ 2). In addition to withdrawals, 15 participants were lost at 3-month follow-up. Discharge location of the 214 participants who were alive and remained in the study were examined at that point. More intervention participants returned to the community (n = 46, 32.6% vs n = 39, 27.5%), and fewer were institutionalized (n = 60, 42.5% vs n = 69, 48.6%) than usual care participants, but the difference was not statistically significance (P = .54, χ 2).