Model-based glycemic control in a Malaysian intensive care unit: performance and safety study

Figures & data

Figure 1 Summary of the STAR protocol. Grey blocks refer to actions taken at patient bedside.

Abbreviations: BG, blood glucose; STAR, Stochastic TARgeted.

Figure 2 Summary of the cohort data and its selection for analysis.

Abbreviations: STAR, Stochastic TARgeted.

Figure 3 Per-patient boxplots of BG measurements.

Abbreviation: BG, blood glucose.

Table 1 Cohort and per-patient glycemic control results

Performance and safety statistics	STAR: 6.0–10.0 mmol/L
	Normal	Interpolated
Number of episodes	142
Number of BG measurements	7,012
Total hours	9,587	9,654
Median hours on protocol	49 (34–80)
Median days on protocol	2.0 (1.4–3.3)
Average measurement interval	1/1.37 hours	1/1.38 hours
Insulin rate (U/h): Median (IQR)	1.6 (0.4–3.4)
Max Insulin rate (U/h): Median (IQR)	8.0 (6.9–8.9)
Dextrose rate (g/h): Median (IQR)]	5.2 (2.8–7.0)
Intervention of insulin in the control (%)	80.12
Median (IQR) BG (mmol/L)
Whole cohort	8.8	8.5
Per patient	8.2 (7.4–9.5)	8.15 (7.4–9.5])
Median time to BG <10 (hours)	2.8 (1.8–4.0)
% BG >10
Whole cohort	34.1	28.4
Per patient	25.8 (16.1–40.0)	21.0 (11.4–36.1)
% BG within 6.0–10.0 mmol/L
Whole cohort	52.6	60.7
Per patient	54.4 (44.4–62.2)	61.9 (51.2–71.9)
% BG within 4.4–8.0 mmol/L
Whole cohort	37.3	39.6
Per patient	41.8 (27.3–57.1)	44.2 (25.0–61.4)
% BG <4.0 mmol/L
Whole cohort	1.43	0.78
Per patient	0.0 (0.0–2.2)	0.0 (0.0–0.0)
# Incidences with BG <4.0 mmol/L	100
# Episodes with BG <4.0 mmol/L	42
% BG <2.2 mmol/L
Whole cohort	0.07	0.02
Per patient	0.0 (0.0–0.0)	0.0 (0.0–0.0)
# Incidences with BG <2.2 mmol/L	6
# Episodes with BG <2.2 mmol/L	5

Abbreviations: BG, blood glucose; IQR, interquartile range; STAR, Stochastic TARgeted.

Figure 4 Episode 38a.

Abbreviations: I, plasma insulin; Q, interstitial insulin; SI, insulin sensitivity; BG, blood glucose; EN, enteral nutrition; PN, perenteral nutrition.

Figure 5 Per-episode BG CDFs defined on whole cohort.

Abbreviations: BG, blood glucose; CDFs, cumulative distributive functions.

Table 2 1–3 hourly treatment chosen for each episode

Type of BG treatment	Number of available BG treatment
1-hourly	8,359
2-hourly	2,441
3-hourly	1,516
Number of chosen/available BG treatment
Context	1-hourly	2-hourly	3-hourly
Only 1-hourly available	186/186 (100%)
Only 1 and 2-hourly available	529/608 (87.0%)	79/115 (68.7%)
All three options available	5,442/7,565 (71.9%)	933/2,326 (40.1%)	1,190/1,516 (78.5%)
Total and overall %	6,157/8,359 (73.7%)	1,012/8,359 (12.1%)	1,190/8,359 (14.2%)

Abbreviation: BG, blood glucose.

Figure 6 Histogram for starting blood glucose level.

Figure 7 Percentage of insulin rate given over overall administration.

Table 3 Per-episode hourly interpolated glycemic control results for the three STAR cohorts in IIUMMC, Christchurch and Gyulá Hospital ICUs

Statistics STAR trial	Christchurch	Gyulá	IIUMMC
Starting criteria	2 successive BG measurements over 8 mmol/L within a 4-h period		2 successive BG measurements over 10 mmol/L within a 1-h period
Number of episodes	267	47	209
Measurements per day	13.6	11.7	18.1
APACHE ll score	21.0 (16.0–25.0)	32 (28.0–36.0)	16 (11.8–23.0)
% time 4.4–8.0 mmol/L	86.6 (75.0–94.1)	87.1 (79.3–91.1)	46.2 (25.5–63.6)
% time 6.0–10.0 mmol/L	No information		61.5 (50.0–73.5)
% time >10.0 mmol/L	4.4	3.0	28.4
% time <4.0 mmol/L	0.6	0.9	0.8
Median insulin rate	2.7 (1.9–3.5)	3.2 (2.4–4.6)	1.5 (1.0–2.0)

Notes: APACHE ll is one of the scoring system in the ICU to classify overall per-patient severity-of-disease. It is applied within 24 hours of admission of a patient to an intensive care unit (ICU). An integer score from 0 to 71 is computed based on several measurements; higher scores correspond to more severe disease and a higher risk of death.

Abbreviations: ICU, intensive care unit; IIUMMC, International Islamic University Malaysia Medical Centre; STAR, Stochastic TARgeted.

Figure S1 Per-episode insulin sensitivity (SI) profiles for the pilot trial cohort.

Table S1 Patient details for 5 episodes with severe hypoglycemia. Selected clinical results with potential diagnosis

ID	Specification	Diagnosis	Total hours	Number of BG measurements	Initial BG	BG median (IQR)	BG mean	BG SD	%BG <4.0	%BG <2.22	% BG between 4.0–10.0 mmol/L	Time to <10 mmol/L (h)	% of treatment with 1-hourly option	Median insulin rate (U/h)	Max insulin rate (U/h)	Median dextrose rate (g/h)
7a	Episode with highest performance and safety	Control was started early	41	31	9.7	6.4 (5.6–7.6)	6.44	1.26	0	0	96.77	0	9	0.0 (0.0–1.0)	3.4	6.3 (3.7–6.3)
38a	Episode with 1 severe hypoglycemia	Enteral nutrition was stopped	102	78	14.6	6.8 (5.0–8.4)	6.65	1.51	12.82	1.28	78.21	0.8	73.08	0.7 (0.0–1.1)	5	5.4 (5.4–5.4])
63a	Largest episode and 1 severe hypoglycemia	Septic shock with multi-organ failure 2‘ to nosocomial infection	805	618	15.3	10.2 (8.4–12.3)	10.08	1.36	0.16	0.16	48.54	9.1	81.23	1.9 (1.0–4.5)	8	5.2 (5.2–5.9)
69a	Episode with mild hypoglycemia >5% and 2 severe hypoglycemia	No suggestion at the moment	124	91	21.9	6.3 (4.4–7.7)	5.84	1.53	16.48	2.20	78.82	3.5	78.02	1.1 (0.6–2.4)	7.4	4.1 (4.1–4.8)
75c	Episode with 1 severe hyperglycemia	The first patient treated in this trial. Nurses did not follow the STAR recommendation properly due to lack of trust. Furthermore, the control was stopped directly after this incidence	28	28	11.6	9.9 (7.3–12.2)	8.92	1.59	3.57	3.57	50.00	1.58	96.42	2.9 (2.0–6.3)	8	6.3 (5.7–8.9)
78b	Episode with 1 severe hypoglycemia	A sudden spike of insulin sensitivity estimation	122	80	13.0	8.6 (6.8–11.1)	8.21	1.47	1.25	1.25	65.0	1.93	70.00	0.1 (0.0–1.5)	6.5	0.9 (0.9–0.9)

Note: *All BG measurements are in mmol/L, and all time measurement are in hours.

Abbreviations: BG, blood glucose; IQR, interquartile range; SD, standard deviation; STAR, Stochastic TARgeted.