Trial ID	Intervention(s) and comparator(s)	Randomised / safety (N)	Deaths (n/N)	All adverse events (n/N (%))	Severe, serious adverse events (n/N (%))	Attrition due to adverse events (n/N (%))	All hypoglycaemic episodes (n/N (%))	Severe hypoglycaemic episodes (n/N (%))
Altuntas 2003	I: insulin lispro	20	—	—	—	0/20	—	—
	C: RHI	20	—	—	—	0/20	—	—
Bastyr 2000	I: insulin lispro	186	2/186	—	2/186 (1)a	3/186 (2)	—b	—c
	C: RHI	189	0/189	—	0/189 (0)a	4/189 (2)	—b	—c
Dailey 2004	I: insulin glulisine	‐/435	1/435	358/435 (82)	55/435 (13)	5/435 (1)	317/435 (73)d	6/416 (1)d,e
	C: RHI	‐/441	2/441	351/441 (80)	51/441 (12)	6/441 (1)	322/441 (73)d	5/420 (1)d,e
	all:	878f/876	3/876	709/876 (81)	106/876 (12)	11/876 (1)	639/876 (73)d	11/836d,e
Hermann 2013	I: insulin aspart	18	—g	—	—	—h	—i	—
	C: RHI	11	—g	—	—	—h	—i	—
NCT01650129	I: biphasic insulin aspart 50	58	0/58	53/58 (91)	5/58 (9)	2/58 (3)	40/58 (69)	2/58 (3)j
	C: biphasic human insulin 50/50	26f/25	0/25	22/25 (88)	1/25 (4)	1/25 (4)	16/25 (64)	1/25 (4)
Pfützner 2013	I1: insulin aspart	4	—g	—	—	—	—	—
	I2: insulin glulisine	4	—g	—	—	—	—	—
	C: RHI	4	—g	—	—	—	—	—
Rayman 2007	I: insulin glulisine	448	2/448	260/448 (58)	43/448 (10)k	9/448 (2)	140l/‐m (33)	6/‐n (1)
	C: RHI	444o/442	1/442	260/442 (59)	52/442 (12)k	3/442 (1)	144l/‐m (33)	14/‐n (3)
	All	892/890	3/890	520/890 (58)	95/890 (11)k	12/890 (1)	284/‐ (33)	20/‐ (‐)
Ross 2001	I: insulin lispro	70	—	—	—	1/148 (1)	—	—
	C: RHI	78	—	—	—		—	—
Z012 1997	I: insulin lispro	72	0/72	—	0/72 (0)	0/72 (0)	—p	—q
	C: RHI	73	0/73	—	0/73 (0)	1/73 (1)	—p	—q
Z014 1997	I: insulin lispro	73	0/73	—	3/73 (4)	3/73 (4)	—r	—s
	C: RHI	77	0/77	—	0/77 (0)	3/77 (4)	—r	—s
aExcluding hyper‐ and hypoglycaemic events bEvents/participant/30 days (mean ± standard deviation) at study end: 0.9 ± 2.1 (lispro) vs 0.8 ± 1.9 (RHI); P = 0.39 cInconsistent with other numbers presented in the same table in NCT01650129 dBased on author's response in IQWIG 2005 eOnly for the period month 4 to study end fOne participant did not receive treatment gNot explicitly reported, but likely zero hNot explicitly reported, but the results sections leads us to assume that there were no dropouts iNumber of participants with up to three episodes per year: aspart: 5/18, RHI: 3/11 jRHI: coma: 2 participants; IV glucose: 1 participant, glucagon: 1 participant; lispro: coma: 0 participants, IV glucose: 1 participant, glucagon: 1 participant kInconsistent with information from IQWIG 2005: serious adverse events (other than severe hypo‐ and hyperglycaemia): 38 (8.5%) and 40 (9.0%); adding the number of serious hypoglycaemia (6 and 14) to these numbers results in a higher number than what was reported in Rayman 2007 lMonths 4 to 6 only, according to IQWIG 2005, 226 (glulisine) and 240 (RHI) participants with at least one episode over a period of 6 months mThe number of participants presented in table 4 of the publication cannot be correct (Rayman 2007) nAccording to IQWIG 2005, the total number of participants should be 448 and 442 (deducted from percentage numbers presented in the original study report), but in the same table, it is also reported that there are 21 and 8 missing values oTwo participants did not receive treatment pEvents/participant/30 days (mean ± standard deviation) at study end: 2.1 ± 3.2 (lispro) vs 2.5 ± 4.6 (RHI); P = 0.51 qRHI: coma: 1 participant; IV glucose: 2 participants, glucagon: 0 participants; lispro: coma: 1 participant, IV glucose: 2 participants, glucagon: 0 participants rEvents/participant/30 days (mean ± standard deviation) at study end: 0.8 ± 2.3 (lispro) vs 0.8 ± 2.1 (RHI); P = 0.65 sRHI: coma: 2 participants; IV glucose: 1 participant, glucagon: 0 participants; lispro: coma: 1 participant, IV glucose: 1 participant, glucagon: 0 participants —: denotes not reported C: comparator; I: intervention; RHI: regular human insulin; IV: intravenous