Table 1

Biomarker-stratified data and rate ratios derived from included trials

Blood Eos (×10⁹/L)	FeNO (ppb)	Novel START4			CAPTAIN5			Pooled AZ trials: Benralizumab 2b, PATHWAY, STRATOS 1–27			QUEST6			DREAM3			Aggregate data for the prototype risk scale
		Step 1 asthma; low risk; 9% with attack in past 12 months			Step 4 asthma; high risk; 62% with attack in past 12 months			1% step 3 asthma, 50% step 4 asthma, 49% step 5 asthma; high risk; with attack in past 12 months			47% step 4 asthma, 53% step 5 asthma; high risk; with attack in past 12 months			Step 5 asthma; high risk; with attack in past 12 months			Aggregate data for the prototype risk scale
		N†	Attack rate‡	Rate ratio	N	Attack rate‡	Rate ratio	N†	Attack rate	Rate ratio	N	Attack rate	Rate ratio	N	Attack rate	Rate ratio	N	Rate ratio
<0.15	<25	18	0.05	0.98	228	0.85	0.54	199	0.58	0.81	106	0.56	0.52	23	1.98	0.76	574	0.65
	25–<50	23	0.00	0.00	40	0.10	1.11	82	0.46	0.64	35	0.62	0.61	(9)	(1.78)	(0.71)	180	0.66
	≥50	8	0.00	0.00	17	0.15	1.74	23	0.57	0.81	21	0.53	0.53	(9)	(1.78)	(0.71)	69	0.86
0.15–<0.30	<25	19	0.07	1.50	240	0.07	0.82	191	0.56	0.76	96	0.82	0.80	12	1.54	0.59	558	0.81
	25–<50	42	0.02	0.36	87	0.07	0.79	173	0.67	0.96	53	1.14	1.17	(23)	(2.70)	(1.07)	355	0.88
	≥50	32	0.01	0.24	24	0.12	1.43	52	1.29	1.93	25	0.48	0.47	(23)	(2.70)	(1.07)	133	1.16
≥0.30	<25	4	0.30	6.35	248	0.11	1.29	102	0.58	0.82	89	0.84	0.84	18	1.95	0.75	461	1.12
	25–<50	22	0.00	0.00	147	0.09	1.00	133	0.87	1.30	97	1.24	1.31	(66)	(3.08)	(1.22)	399	1.12
	≥50	51	0.13	4.40	66	0.18	2.14	107	1.01	1.53	98	1.78	2.12	(66)	(3.08)	(1.22)	322	2.29
Analysed		219	0.05	1.00	1097	0.09	1.00	1062	0.70	1.00	620	0.99	1.00	151	2.52	1.00	3051	1.00
Missing*		4			121			120			14			4			262
Total		223			1218			1182			634			155			3313

Aggregate ratios in bold (rightmost column) are those included to derive the prototype risk scale: in effect, an aggregate rate ratio is a mean fold change in the asthma attack rate for patients with that biomarker combination. Numbers between brackets were extracted to calculate frequency-weighted rate ratios but were not used to derive the scale, as this analysis was stratified using only two cut points for fractional exhaled nitric oxide (FeNO <25 or ≥25 ppb).
Blood Eos, peripheral blood eosinophil count; n, number of patients.
*Missing data were excluded from analyses.
†For Novel START and the pooled AstraZeneca (AZ) trials, we regrouped the data of patients with a baseline FeNO of 20–<50 ppb into our 25–<50 ppb group, as the difference of 5 ppb in FeNO is not clinically relevant.
‡For both the Novel START and CAPTAIN, only the percentage of patients with one or more severe attack(s) in the 52 weeks of follow-up was reported, so we imputed the annualised rate as ^–log₁₀(1 − %incidence).