The Summit Score Stratifies Mortality and Morbidity in Chronic Obstructive Pulmonary Disease

Figures & data

Table 1 Baseline Characteristics of the SUMMIT Derivation Population and the Four Risk Score Validation Populations

Characteristics	SUMMIT	SUMMIT	Intermountain	Intermountain	Intermountain
			Cardiovascular	Non-Cardiovasc
	Derivation	Validation	Outpatients	Outpatients	Inpatients
Sample size	N=8181	N=8304	N=9251	N=8551	N=26,170
Demographics
Age (years)	65.2 ± 7.9	65.2 ± 7.9	67.4 ± 11.0	59.9 ± 11.5	66.5 ± 14.6
Sex (female)	24.8%	26.1%	47.4%	53.6%	46.6%
Race (Non-White)	18.8%	19.1%	2.5%	2.4%	3.5%
Ethnicity (Hispanic)	6.8%	7.5%	0.2%	0.2%	0.5%
Body mass index (kg/m^2)	28.0 ± 5.9	28.0 ± 6.0	30.9 ± 8.6	29.2 ± 7.9	29.2 ± 8.8
Randomized SUMMIT Trial Treatment
Placebo	25.0%	24.8%	–	–	–
VI 25	25.0%	25.0%	–	–	–
FF 100	24.9%	25.3%	–	–	–
FF 100/VI 25	25.1%	24.9%	–	–	–
Clinical History (Selected Variables)
Prior COPD exacerbations (≥1)	13.4%	13.3%	3.6%	1.3%	5.3%
Prior COPD treatments (≥1)	32.4%	32.4%	NA	NA	NA
Number of Prior COPD Hospitalizations
0	86.6%	86.6%	97.3%	99.0%	82.1%
1	11.6%	11.7%	2.7%	1.0%	13.6%
2 or more	1.8%	1.6%	0%	0%	4.3%
Heart rate (beats per minute)	75.8 ± 9.9	76.0 ± 9.9	80.7 ± 15.6	83.6 ± 15.4	87.8 ± 19.7
Systolic blood pressure (mmHg)	133 ± 15	133 ± 15	130 ± 19	130 ± 19	133 ± 26
FEV1 (mL)	1686 ± 425	1679 ± 422	NA	NA	NA
Mean corpuscular volume (fL)	94.7 ± 5.9†	94.8 ± 6.4†	91.8 ± 5.8	92.1 ± 5.9	92.0 ± 6.7
Red cell distribution width (%)	15.1 ± 1.5†	15.3 ± 1.6†‡	14.5 ± 1.8	14.1 ± 1.7	14.8 ± 2.2
Risk Factors and Comorbidities
Smoking (current)	46.6%	46.5%	NA	NA	NA
Smoking: Pack Years	41.2 ± 24.4	40.3 ± 24.2	NA	NA	NA
Hypertension	89.9%	90.3%	85.8%	48.0%	73.6%
Hypercholesterolemia	70.0%	69.7%	80.4%	40.6%	55.4%
Diabetes mellitus	30.2%	30.5%	42.5%	0%	25.6%
Myocardial infarction	15.4%	14.8%	18.9%	0%	14.0%
Heart failure	21.2%	20.7%	30.7%	5.6%	23.7%
Stroke	7.2%	6.9%	24.8%	0%	15.5%
Coronary artery disease	50.8%	50.8%	63.1%	0%	42.2%
Pneumonia	16.5%	15.9%	42.0%	25.0%	42.4%
Baseline Medications
Lipid lowering agent	65.1%	64.6%	6.3%	1.5%	25.9%
Anti-thrombotics	52.8%	52.5%	2.5%	0.5%	25.8%
Beta-blockers	31.6%	31.0%	4.9%	1.0%	26.2%
Diuretics	32.9%	33.2%	5.6%	1.6%	33.2%
Calcium channel blockers	34.5%	34.5%	2.8%	0.8%	14.7%
Anti-arrhythmic agents	5.4%	5.3%	2.7%	0.9%	19.8%
Renin-angiotensin-aldosterone inhib.	66.9%	66.3%	5.9%	1.8%	24.0%
Anti-hypertensives	4.5%	4.7%	5.8%	1.8%	24.4%
Xanthine	14.6%	15.3%	0.1%	0%	0.7%
Short-acting anticholinergic	10.5%	12.1%§	2.6%	1.9%	23.2%
Mucolytics	3.0%	3.0%	0.6%	0.2%	2.3%
Corticosteroids	0.6%	0.5%	4.2%	5.5%	22.1%
Long-acting beta-2 agonist	0.4%	0.4%	3.8%	3.2%	14.2%
Long-acting anticholinergic	0.1%	0.2%	2.5%	2.2%	9.3%
Mortality Outcomes
Mortality, total	6.5%	6.1%	27.1%	12.6%	36.8%
Follow-up, mean±SD (years)	1.96 ± 0.78	1.95 ± 0.79	6.44 ± 2.12	6.36 ± 2.01	4.38 ± 1.97
Follow-up, range (years)	0.88–3.91	0.21–3.91	2.23–9.29	2.25–9.29	0.50–7.50
Mortality, annual rate	3.40%/year	3.01%/year	3.96%/year	1.71%/year	8.40%/year

Notes: †Laboratory data were only available for n=1745 SUMMIT subjects; ‡p=0.007 or §p<0.001 for the comparison of SUMMIT validation vs derivation populations.

Abbreviation: NA, not available.

Table 2 The Association of Quartiles of the Summit Risk Score with All-Cause Mortality in the Internal SUMMIT Validation Population Subset and in the Intermountain Outpatients with COPD and Cardiovascular Risks. As per Table 1, Length of Follow-Up for Mortality Differed Between Populations

Summit Score Category	Mortality	Hazard Ratio (95% CI)	p-value	Sample Size
SUMMIT Trial Internal Validation (N=8304)
Quartile: 1 (Score: 1–12)	2.6%	1.0 (referent)	—	n=1977
2 (Score: 13–14)	4.5%	1.73 (1.22, 2.45)	0.002	n=1813
3 (Score: 15–17)	6.3%	2.50 (1.83, 3.42)	<0.001	n=2664
4 (Score: 18–32)	11.0%	4.43 (3.27, 6.01)	<0.001	n=1850
Intermountain COPD Outpatients with Cardiovascular Risks (External Validation) (N=9251)
Quartile: 1 (Score: 0–9)	8.5%	1.0 (referent)	—	n=2207
2 (Score: 10–11)	17.9%	2.17 (1.82, 2.59)	<0.001	n=1932
3 (Score: 12–14)	28.4%	3.59 (3.06, 4.21)	<0.001	n=2795
4 (Score: 15–26)	50.9%	7.62 (6.53, 8.89)	<0.001	n=2317

Figure 1 Kaplan–Meier survival curves displaying the association of the Summit Risk Score with all-cause mortality among: (A) the validation half of the SUMMIT trial population, N=8304 (for quartiles 4, 3, and 2 compared with quartile 1: Log rank p<0.001, p<0.001, and p=0.002, respectively; c-statistic: c=0.662), (B) an Intermountain Healthcare validation population of outpatients with COPD and cardiovascular risks, N=9251 (for quartiles 4, 3, and 2 compared with quartile 1: Log rank p<0.001, p<0.001, and p<0.001, respectively; c-statistic: c=0.736). Time intervals on the x-axes were designated in 365-day intervals for SUMMIT populations and years for Intermountain populations.

Figure 2 Hazard curves for the effects of the four SUMMIT trial randomized treatments on all-cause mortality among all SUMMIT trial subjects (ie, both the Summit Score derivation and validation groups). Trial randomizations included: once daily inhaled placebo, vilanterol (VI) 25 μg, fluticasone furoate (FF) 100 μg, or the combination of FF 100 μg and VI 25 μg. Here the combination of FF 100 μg/VI 25 μg (n=2347) had a statistically significantly lower mortality compared with placebo (n=2305) (p=0.0158, HR= 0.76, CI=0.61, 0.95) among subjects with a Summit Score ≥14 and ≤19 (n=9243, which is 56.1% of the SUMMIT trial subjects). The hazard curves for subject with Summit Score ≤13 and ≥20 are provided in Supplemental Figures S2B and S2C.