Reply to Taito et al.: Improved Active Mobilization Program for Critically Ill Patients with Diabetes

From the Authors:

We thank Dr. Taito and colleagues for their interest in our recent paper (1). Regarding the first concern raised, we want to emphasize that all patients randomized in the main trial were included in this analysis, contrary to the concern that the high-dose early mobilization group did not include all eligible patients.

We also understand that some patients in the usual-care mobilization group received prolonged sessions of mobilization, which was permitted by the main trial’s protocol, allowing usual-care mobilization to be administered according to the participating unit’s standard practices, including prolonged sessions. We did not assess the impact of duration itself to avoid violating the intention-to-treat principle, which could result in selection bias, biased treatment effect estimates, loss of randomization benefits, and reduced generalizability of results.

Regarding the second concern, the mean mobilization time in patients with diabetes was 7 minutes in the high-dose early mobilization group and 0 minutes in the usual-care mobilization group. In patients without diabetes, the mean mobilization time was 8 minutes in the high-dose early mobilization group and 4 minutes in the usual-care mobilization group. Additional information is provided in Table 1.

Table 1.

Mobilization and Metabolic Variables in Patients Who Died at Day 180

	Diabetes		No Diabetes
	High-Dose Early Mobilization (n = 36)	Usual-Care Mobilization (n = 12)	High-Dose Early Mobilization (n = 47)	Usual-Care Mobilization (n = 59)	P Value
Mobilization in ICU
Minutes of active mobilization per day, mean	7.0 (0.0–15.1)	0.2 (0.0–2.8)	7.6 (1.1–19.0)	3.8 (1.0–8.7)	0.072
Glucose concentration, mmol/L
Highest concentration	15.9 (13.7–17.6)	17.6 (15.9–19.4)	—	—	—
>14	14/19 (73.7)	3/3 (100.0)	—	—	—
Lowest concentration	5.7 (4.8–6.9)	4.7 (4.2–5.2)	—	—	—
<4	2/19 (10.5)	1/3 (33.3)	—	—	—
<2.2	0/19 (0.0)	0/3 (0.0)	—	—	—
Heart rate, beats/min
Highest rate	120.0 (104.0–139.5)	142.0 (137.0–153.5)	—	—	—
>130	6/19 (31.6)	3/3 (100.0)	—	—	—
Lowest rate	58.0 (51.5–63.5)	73.0 (66.5–78.5)	—	—	—
<50	4/19 (21.1)	0/3 (0.0)	—	—	—
Mean arterial pressure, mm Hg
Highest pressure	100.0 (83.5–114.5)	99.0 (92.0–117.5)	—	—	—
>100	9/19 (47.4)	1/3 (33.3)	—	—	—
Lowest pressure	57.0 (45.0–61.0)	62.0 (48.5–62.0)	—	—	—
<50	5/19 (26.3)	1/3 (33.3)	—	—	—
Use of noradrenaline	17/19 (89.5)	2/3 (66.7)	—	—	—
Days with noradrenaline use	4.0 (1.5–5.0)	4.0 (2.0–4.5)	—	—	—
Mean highest dose per day, μg/kg/min	0.12 (0.08–0.20)	0.18 (0.11–0.25)	—	—	—
Highest dose, μg/kg/min	0.17 (0.12–0.30)	0.31 (0.19–0.42)	—	—	—
Use of noradrenaline >0.20 μg/kg/min	7/17 (41.2)	1/2 (50.0)	—	—	—

Data are expressed as median (interquartile range) or frequency (percentage). Denominators are the numbers of patients with data available.

Finally, we agree that a dose–response analysis would help generate appropriate hypotheses, and this is planned and currently in progress. The TEAM (Treatment of Mechanically Ventilated Adults with Early Activity and Mobilization) trial implemented a dynamic treatment regime in both arms, in which the “dose” of mobilization depended on various factors, leading to varying adherence over time. As mentioned before, an analysis of dose would necessarily violate the intention-to-treat principle, and adjustment only for baseline variables might miss important prognostic factors affecting adherence during follow-up. To obtain reliable estimates, appropriate adjustments for measured prerandomization and time-varying postrandomization confounding and selection bias are needed, requiring more complex analyses (2, 3). Such analyses are under way.