Clinical implication of pulmonary hospitalization in heart failure with preserved ejection fraction: from the TOPCAT.

AIMS: The aim of the study was to explore the risk factors and evaluate the prognostic implication of pulmonary hospitalization on heart failure (HF) with preserved ejection fraction (HFpEF). METHODS AND
RESULTS: We performed a secondary analysis of the Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist Trial (TOPCAT). A total of 1714 patients with HFpEF were analysed in our study. In the multivariate Cox proportional hazards regression analysis, history of chronic obstructive pulmonary disease (COPD), smoking, bone fracture after the age of 45, and previous HF hospitalization were identified as independent risk factors for pulmonary hospitalization. To evaluate the prognostic significance of pulmonary hospitalization, patients were categorized into five groups according to the causes of their first hospitalization. The all-cause and cardiovascular (CV) mortality risks in these five groups were compared using time-varying Cox proportional hazards model. Compared with patients without hospitalization during follow-up, those with pulmonary hospitalization were associated with a 204% increase [hazard ratio (HR) 3.04, 95% confidence interval (CI) 2.07-4.47, P < 0.001] and 164% increase (HR 2.64, 95% CI 1.60-4.36, P < 0.001) in risks of all-cause and CV mortality, respectively, while the corresponding risk increases associated with HF hospitalization were 146% (HR 2.46, 95% CI 1.74-3.48, P < 0.001) for all-cause mortality and 186% (HR 2.86, 95% CI 1.87-4.36, P < 0.001) for CV mortality.
CONCLUSIONS: Pulmonary hospitalization was associated with a significant increase in risks of all-cause and CV mortality, which was comparable with that associated with HF hospitalization. The results suggested that pulmonary hospitalization could be another important clinical endpoint of HFpEF.

Introduction

Heart failure (HF) with preserved ejection fraction (HFpEF) composes approximately half of all HF, and hospitalizations due to HFpEF are still increasing these years. , , Steinberg et al. reported that the proportion of patients hospitalized with HFpEF increased from 33% in 2005 to 39% in 2010 in the USA. However, patient profiles, pathology, and treatment of HFpEF are still not fully established owing to its diverse aetiologies and presentations. ,

Studies have shown that patients with HFpEF mostly suffered from non‐cardiovascular (non‐CV) co‐morbidities. , Therefore, the management of non‐CV diseases is also of significance. The lung is one of the most closely related organs to the heart. Dysfunction of lung could significantly affect the heart and vice versa. Pulmonary dysfunction (manifested by airflow limitation and arterial hypoxaemia) and pneumonia were both associated with poor prognosis in HFpEF. , Coexisting HFpEF would also lead to worse long‐term prognosis in chronic obstructive pulmonary disease (COPD) patients. These findings suggested that the worsening of pulmonary function was closely related to HFpEF exacerbation. Therefore, pulmonary hospitalization might have significant implications on the prognosis of HFpEF, which, however, remained underexplored.

In the present analysis, we aim to determine the risk factors of pulmonary hospitalization and evaluate the impact of pulmonary hospitalization on mortality in HFpEF.

Methods

The current study was a secondary analysis of Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist Trial (TOPCAT). We obtained data from the National Institutes of Heart, Lung, and Blood Institute's Biologic Specimen and Data Repository Information Coordinating Center via an approved proposal.

Patients

TOPCAT was a multicentre, international, randomized, double‐blind, and placebo‐controlled trial, which aimed to determine whether aldosterone would improve clinical outcomes in patients with HFpEF. The design of TOPCAT has been published previously. Briefly, patients aged ≥ 50 years, with HF symptoms combined with left ventricular ejection fractions ≥ 45%, and with elevated natriuretic peptide levels or HF hospitalization in the last 12 months before randomization were recruited. The main exclusion criteria included severe illness with a life expectancy of <3 years, severe renal dysfunction, history of hyperkalaemia, and uncontrolled hypertension. In our study, we excluded patients enrolled in Russia and Georgia, for the uncertainty about the reliability of the diagnosis of HFpEF. The trial was approved by the ethics committee at each study centre, and an informed consent form was signed by each participant. Our investigation conforms with the principles outlined in the Declaration of Helsinki.

Outcomes of interest

The outcomes of this study were all‐cause and CV mortality. These outcomes were adjudicated by a clinical endpoint committee at Brigham and Women's Hospital.

Statistical analysis

Patients were grouped according to the cause of their first hospitalization during follow‐up: patients without hospitalization, patients first hospitalized for HF, patients first hospitalized for pulmonary diseases, patients first hospitalized for non‐HF CV diseases, and patients first hospitalized for other causes. Specifically, pulmonary hospitalization was adjudicated by the investigators without predefined criteria. Baseline characteristics were presented as mean ± standard deviation (SD) for normally distributed continuous variables and as median with inter‐quartile range (IQR) for continuous variables, which are not normally distributed. Numbers with percentages were used to summarize categorical variables. Differences in baseline characteristics between groups were assessed by Kruskal–Wallis test for continuous variables. Categorical variables were compared by χ 2 test.

The univariate Cox proportional hazards model was used to recognize potential risk factors of pulmonary hospitalization. Baseline characteristics with clinical implications that were significant different among groups were analysed. The variables with a P value < 0.05 were then included in a multivariate Cox proportional hazards model to explore the independent predictors of pulmonary hospitalization.

To evaluate the association between pulmonary hospitalization and all‐cause/CV mortality, patients were grouped according to the cause of their first hospitalization, and patients without any hospitalization served as the reference group. The crude all‐cause/CV death rates among groups were compared using χ 2 test. Kaplan–Meier survival curve with log‐rank test was used to compare the survival of the five patient groups. Cox regression model was used to assess the association of different causes of first hospitalization and prognosis. Hospitalization for HF, pulmonary diseases, non‐HF CV diseases, and other reasons were treated as time‐dependent covariates, for the status of hospitalization changed over time. The following variables were adjusted in this model: gender, age, race, previous HF hospitalizations, New York Heart Association functional status, COPD, coronary heart disease, diabetes, insulin use, stroke, atrial fibrillation (AF), hypothyroidism, smoking status, bone fracture after the age of 45, body mass index, diastolic blood pressure, heart rate, haemoglobin, and glomerular filtration rate.

All P values are two‐sided, and P < 0.05 was considered significant. All analyses were performed using Stata version 14 (Stata Corp., College Station, TX, USA).

Results

A total of 3445 patients were enrolled from 10 August 2006 to 31 January 2012 in TOPCAT. A total of 1678 patients enrolled from Russia and Georgia were excluded as mentioned in Methods. Patients with missing information on variables in the multivariate Cox analysis of mortality risk were also excluded. Finally, a total of 1714 patients were included in the present study.

Prevalence of pulmonary hospitalization among heart failure with preserved ejection fraction patients

Figure 1 shows the distribution of hospitalizations during a median follow‐up time of 3.3 years. There were 3050 hospitalizations during follow‐up, and 1024 of 1714 (59.74%) patients had hospitalized at least once. There were 1529 (50.13%) hospitalizations due to CV causes and 1521 (49.87%) due to non‐CV causes. HF was the most common cause of hospitalizations, accounting for 27.57% of the total admissions. Among non‐CV hospitalizations, pulmonary disease was the most common cause and accounted for 10.26% of total admissions. Among 1024 hospitalized patients, 405 had at least one HF hospitalization, 227 had at least one pulmonary hospitalization, and 826 had at least one hospitalization for other reasons. Pulmonary hospitalization was significantly correlated to HF hospitalization (P < 0.001). Among those who were hospitalized due to pulmonary disease, 46.7% of them also experienced at least one HF hospitalization.

Figure 1

Prevalence of hospitalizations among heart failure with preserved ejection fraction (HFpEF) patients. (A) Number of hospitalizations for different causes. (B) Number of patients hospitalized for different causes.

Risk factors of pulmonary hospitalization among heart failure with preserved ejection fraction patients

Table 1 shows the baseline characteristics of patients according to their first hospitalization. Compared with those without hospitalization or first hospitalized for non‐pulmonary reasons, patients who were first hospitalized for pulmonary diseases had a higher burden of baseline COPD, asthma, and steroid therapy. Interestingly, they also have a higher prevalence of previous HF hospitalization, bone fracture after the age of 45, and AF. Patients who were first hospitalized for HF have more concomitant diseases like diabetes, chronic kidney diseases, and anaemia, along with higher value of NT‐pro‐BNP.

Table 1

Baseline characteristics

Characteristic	No hospitalization	HF hospitalization	Pulmonary hospitalization	Non‐HF CV hospitalization	Hospitalization for other causes	P value
Age (years), median (IQR)	72 (64, 78)	71 (65, 79)	75 (66, 80)	73 (64, 79.5)	74 (64, 80)	0.055
Male, n (%)	336 (48.70)	116 (51.79)	67 (54.47)	121 (49.59)	220 (50.81)	0.764
Race, n (%) b						0.019
White	554 (80.29)	161 (71.88)	100 (81.30)	188 (77.05)	348 (80.37)
Black	97 (14.06)	54 (24.11)	18 (14.63)	49 (20.08)	67 (15.47)
Other	39 (5.65)	9 (4.02)	5 (4.07)	7 (2.87)	18 (4.16)
Previous HF hospitalization, n (%) b	373 (54.06)	164 (73.21)	76 (61.79)	156 (63.93)	238 (54.97)	<0.001
EF (%), median (IQR)	59 (53, 65)	57 (51, 63.5)	60 (55, 64)	58.5 (50, 64)	58 (54, 64)	0.662
NYHA class, n (%) b						<0.001
I–II	491 (71.16)	110 (49.11)	82 (66.67)	163 (66.80)	271 (62.59)
III–IV	199 (28.84)	114 (50.89)	41 (33.33)	81 (33.20)	162 (37.41)
COPD, n (%) b	88 (12.75)	38 (16.96)	39 (31.71)	43 (17.62)	72 (16.63)	<0.001
Asthma, n (%) b	58 (8.41)	23 (10.27)	24 (19.51)	28 (11.48)	53 (12.24)	0.005
CHD, n (%) b						0.001
No CHD	412 (59.71)	113 (50.45)	67 (54.47)	107 (43.85)	222 (51.27)
CHD without MI	163 (23.62)	63 (28.13)	26 (21.14)	67 (27.46)	120 (27.71)
CHD with MI	115 (16.67)	48 (21.43)	30 (24.39)	70 (28.69)	91 (21.02)
DM, n (%) b						<0.001
No DM	424 (61.45)	91 (40.63)	73 (59.35)	126 (51.64)	237 (54.73)
DM without microvascular complication	196 (28.41)	72 (32.14)	38 (30.89)	78 (31.97)	113 (26.10)
DM with microvascular complication	70 (10.14)	61 (27.23)	12 (9.76)	40 (16.39)	83 (19.17)
Insulin usage, n (%) b	111 (16.09)	69 (30.80)	23 (18.70)	61 (25.00)	102 (23.56)	<0.001
Stroke, n (%)	61 (8.84)	17 (7.59)	13 (10.57)	24 (9.84)	38 (8.78)	0.883
AF, n (%) b						0.005
No AF	418 (60.58)	135 (60.27)	58 (47.15)	141 (57.79)	234 (54.04)
Paroxysmal AF	84 (12.17)	22 (9.82)	24 (19.51)	46 (18.85)	76 (17.55)
Chronic AF	188 (27.25)	67 (29.91)	41 (33.33)	57 (23.36)	123 (28.41)
Hypothyroid, n (%)	107 (15.51)	39 (17.41)	26 (21.14)	38 (15.57)	80 (18.48)	0.458
Depression, n (%)	131 (25.49)	44 (24.31)	32 (29.63)	53 (25.73)	106 (27.82)	0.800
Bone fracture after the age of 45, n (%) b	78 (11.30)	36 (16.07)	29 (23.58)	36 (14.75)	82 (18.94)	0.001
Physical activity (mets per week), n (%)						0.467
0	228 (33.24)	85 (37.95)	37 (30.58)	72 (29.51)	134 (31.09)
<4.7	228 (33.24)	78 (34.82)	44 (36.36)	92 (37.70)	144 (33.41)
≥4.7	230 (33.53)	61 (27.23)	40 (33.06)	80 (32.79)	153 (35.50)
BMI (kg/m2), n (%)						0.207
<30	240 (34.78)	64 (28.57)	49 (39.84)	86 (35.25)	169 (39.03)
30–35	185 (26.81)	67 (29.91)	32 (26.02)	73 (29.92)	101 (23.33)
>35–39	128 (18.55)	35 (15.63)	23 (18.70)	41 (16.80)	78 (18.01)
≥40	137 (19.86)	58 (25.89)	19 (15.45)	44 (18.03)	85 (19.63)
Smoking status, n (%) b						0.001
Non‐smoker	330 (47.83)	83 (37.05)	32 (26.02)	105 (43.03)	175 (40.42)
Former smoker	315 (45.65)	130 (58.04)	80 (65.04)	124 (50.82)	229 (52.89)
Current smoker	45 (6.52)	11 (4.91)	11 (8.94)	15 (6.15)	29 (6.70)
Haemoglobin < 120 g/dL, n (%) b	160 (23.19)	99 (44.20)	32 (26.02)	70 (28.69)	147 (33.95)	<0.001
eGFR, n (%) b						<0.001
≥90	99 (14.35)	26 (11.61)	14 (11.38)	24 (9.84)	33 (7.62)
≥60–89	296 (42.90)	69 (30.80)	54 (43.90)	96 (39.34)	170 (39.26)
≥30–59	295 (42.75)	129 (57.59)	55 (44.72)	124 (50.82)	227 (52.42)
<30					3 (0.69)
Systolic blood pressure (mmHg), median (IQR)	129 (118, 139)	129 (118, 139)	125 (114, 137)	130 (115, 139)	128 (118, 138)	0.612
Diastolic blood pressure (mmHg), median (IQR) b	72 (64, 80)	70 (60, 80)	67 (60, 78)	71 (62, 80)	70 (62, 80)	<0.001
Heart rate (b.p.m.), median (IQR) b	69 (60, 76)	70 (63, 79)	65 (60, 74)	68 (60, 74)	68 (61, 76)	0.005
BNP (ng/mL), median (IQR) (n = 671)	239.5 (148, 427)	291 (152, 556)	240 (171.5, 394.5)	276 (141, 491)	276 (152, 440)	0.724
NT pro‐BNP (ng/mL), median (IQR) b (n = 354)	917 (554, 1720)	2239 (777, 3890)	1195.5 (573, 2941)	798 (503, 1373)	966 (533, 2018)	0.011
Beta‐blocker, n (%)	526 (76.23)	188 (83.93)	97 (78.86)	188 (77.05)	347 (80.32)	0.126
ACEI/ARB, n (%)	547 (79.28)	179 (79.91)	94 (76.42)	199 (81.56)	332 (76.85)	0.597
Diuretic, n (%) b	593 (85.94)	209 (93.30)	112 (91.06)	219 (89.75)	396 (91.67)	0.005
CCB, n (%)	250 (36.23)	96 (42.86)	47 (38.21)	90 (36.89)	176 (40.74)	0.344
Aspirin, n (%)	389 (56.38)	133 (59.38)	73 (59.35)	147 (60.25)	256 (59.26)	0.781
Warfarin, n (%) b	208 (30.14)	74 (33.04)	48 (39.02)	84 (34.43)	167 (38.66)	0.036
Steroid, n (%) b	86 (12.46)	22 (9.82)	28 (22.76)	29 (11.89)	76 (17.55)	0.001
Vitamin D, n (%)	90 (13.04)	27 (12.05)	18 (14.63)	36 (14.75)	81 (18.71)	0.081
Death (all‐cause) b	65 (9.42)	74 (33.04)	47 (38.21)	57 (23.36)	130 (30.02)	<0.001
Death (CV) b	43 (6.23)	53 (23.66)	26 (21.14)	34 (13.93)	60 (13.86)	<0.001

Missing data were excluded.

P < 0.05.

ACEI, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; AF, atrial fibrillation; BMI, body mass index; BNP, B‐type natriuretic peptide; CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; CV, cardiovascular; DM, diabetes mellitus; EF, ejection fraction; eGFR, estimated glomerular filtration rate; HF, heart failure; IQR, inter‐quartile range; NT‐pro‐BNP, N‐terminal pro‐B‐type natriuretic peptide; NYHA, New York Heart Association.

Table 2 presents the results of potential risk factors of pulmonary hospitalizations using Cox proportional hazards model. In the univariate analysis, prior HF hospitalizations, history of COPD, asthma, bone fracture after the age of 45, smoking, steroid therapy, and diastolic blood pressure were significantly associated with pulmonary hospitalizations. In the multivariate analysis, except for asthma and steroid therapy, the above variables remained to be significantly associated with risk of pulmonary hospitalizations: previous HF hospitalizations [hazard ratio (HR) 1.41, 95% confidence interval (CI) 1.07–1.85, P = 0.014], COPD (HR 1.76, 95% CI 1.28–2.42, P = 0.001), bone fracture after the age of 45 (HR 1.62, 95% CI 1.18–2.22, P = 0.003), former smoker (HR 1.56, 95% CI 1.16–2.09, P = 0.003), current smoker (HR 2.08, 95% CI 1.25–3.48, P = 0.005), and diastolic blood pressure (HR 0.98, 95% CI 0.97–0.99, P = 0.006).

Table 2

Risk factors of pulmonary hospitalization

Risk factors	Univariate		Multivariate
	HR	P value	HR	P value
Previous HF hospitalization	1.44 (1.10–1.89)	0.009	1.41 (1.07–1.85)	0.014
COPD	2.22 (1.66–2.97)	<0.001	1.76 (1.28–2.42)	0.001
Asthma	1.72 (1.21–2.45)	0.003
Bone fracture after age of 45	1.69 (1.24–2.30)	0.001	1.62 (1.18–2.22)	0.003
Former smoker	1.76 (1.31–2.35)	<0.001	1.56 (1.16–2.09)	0.003
Current smoker	2.35 (1.42–3.87)	0.001	2.08 (1.25–3.48)	0.005
Diastolic blood pressure	0.98 (0.97–0.99)	0.003	0.98 (0.97–0.99)	0.006
Steroid	1.52 (1.09–2.11)	0.014

COPD, chronic obstructive pulmonary disease; HF, heart failure; HR, hazard ratio.

The multivariate model was adjusted for previous HF hospitalization, COPD, asthma, bone fracture after age of 45, former smoker, current smoker, diastolic blood pressure, and steroid.

Prognostic significance of pulmonary hospitalization among heart failure with preserved ejection fraction patients

We then evaluated, for HFpEF patients, the association of pulmonary hospitalization and all‐cause mortality risk. Patients were categorized according to the cause of their first hospitalization, if they had any. Those without hospitalization were regarded as the reference group. Time to first hospitalization among the four groups was not statistically significant, which was analysed by log‐rank test (Figure S1). During follow‐up, 373 (21.76%) patients died. Crude mortality rate was numerically the highest in patients who were first hospitalized for pulmonary diseases (38.21%, 47 patients), followed by patients who were first hospitalized for HF (33.04%, 74 patients), other reasons (30.02%, 130 patients), and non‐HF CV diseases (23.36%, 57 patients), while patients without any hospitalization owned the lowest mortality (9.42%, 65 patients) (Table 1). The Kaplan–Meier survival curve showed the same trend (P for log‐rank test < 0.001, Figure 2A). Table 3 presents HRs for mortality in different groups. In the univariate Cox regression model, both pulmonary hospitalization and HF hospitalization were associated with a similar increase in mortality risk (pulmonary hospitalization HR 3.46, 95% CI 2.38–5.04, P < 0.001; HF hospitalization HR 3.35, 95% CI 2.40–4.67, P < 0.001), while the HR of that first hospitalization for non‐HF CV diseases and other causes was 2.00 (95% CI 1.40–2.85, P < 0.001) and 2.67 (95% CI 1.98–3.60, P < 0.001). In the multivariate analysis, hospitalization for pulmonary diseases, HF, non‐HF CV diseases, and other causes was associated with 204%, 146%, 76%, and 126% increase in mortality risk, respectively (pulmonary hospitalization HR 3.04, 95% CI 2.07–4.47, P < 0.001; HF hospitalization HR 2.46, 95% CI 1.73–3.48, P < 0.001; non‐HF CV hospitalization HR 1.76, 95% CI 1.22–2.53, P = 0.002; other reason hospitalization HR 2.26, 95% CI 1.67–3.06, P < 0.001).

Figure 2

Kaplan–Meier curves in five groups according to the causes of the first hospitalization for all‐cause mortality (A) and cardiovascular (CV) mortality (B).

Table 3

Association between patients first hospitalized for different causes and all‐cause mortality risk

Causes for first hospitalization	Univariate		Multivariate
	HR	P value	HR	P value
HF hospitalization	3.35 (2.40–4.67)	<0.001	2.46 (1.73–3.48)	<0.001
Pulmonary hospitalization	3.46 (2.38–5.04)	<0.001	3.04 (2.07–4.47)	<0.001
Non‐HF CV hospitalization	2.00 (1.40–2.85)	<0.001	1.76 (1.22–2.53)	0.002
Hospitalization for other causes	2.67 (1.98–3.60)	<0.001	2.26 (1.67–3.06)	<0.001

CV, cardiovascular; HF, heart failure.

The multivariate model was adjusted for gender, age, race, previous HF hospitalizations, New York Heart Association functional status, COPD, coronary heart disease, diabetes, insulin use, stroke, atrial fibrillation, hypothyroid, bone fracture after the age of 45, smoking status, body mass index, diastolic blood pressure, heart rate, haemoglobin, and estimated glomerular filtration rate.

The crude CV mortality rates were 6.23% (43 patients), 13.86% (60 patients), 13.93% (34 patients), 21.14% (26 patients), and 23.66% (53 patients) in the group of patients without hospitalization, patients first hospitalized for other reason, non‐HF CV diseases, pulmonary diseases, and HF, respectively (Table 1). Kaplan–Meier survival curve yielded similar results (P for log‐rank test < 0.001, Figure 2B). Compared with patients without hospitalization, patients first hospitalized for pulmonary diseases had an increased CV mortality risk by 190% (HR 2.90, 95% CI 1.78–4.73, P < 0.001) in unadjusted Cox regression analysis, while patients first hospitalized for HF had a 263% increase (HR 3.63, 95% CI 2.42–5.42, P < 0.001), patients first hospitalized for non‐HF CV diseases had a 80% increase (HR 1.80, 95% CI 1.15–2.83, P = 0.01), and patients first hospitalized for other causes had a 87% increase (HR 1.87, 95% CI 1.26–2.76, P = 0.002) in CV mortality risk. In the multivariate analysis, HR was 2.86 (95% CI 1.87–4.36, P < 0.001), 2.64 (95% CI 1.60–4.36, P < 0.001), 1.60 (95% CI 1.01–2.54, P = 0.044), and 1.60 (95% CI 1.07–2.39, P = 0.021) in patients first hospitalized for HF, pulmonary diseases, non‐HF CV diseases, and other reason, respectively (Table 4).

Table 4

Association between patients first hospitalized for different causes and CV mortality risk

Causes for first hospitalization	Univariate		Multivariate
	HR	P value	HR	P value
HF hospitalization	3.63 (2.42–5.42)	<0.001	2.86 (1.87–4.36)	<0.001
Pulmonary hospitalization	2.90 (1.78–4.73)	<0.001	2.64 (1.60–4.36)	<0.001
Non‐HF CV hospitalization	1.80 (1.15–2.83)	0.01	1.60 (1.01–2.54)	0.044
Hospitalization for other causes	1.87 (1.26–2.76)	0.002	1.60 (1.07–2.39)	0.021

CV, cardiovascular; HF, heart failure.

The multivariate model was adjusted for gender, age, race, previous HF hospitalizations, New York Heart Association functional status, COPD, coronary heart disease, diabetes, insulin use, stroke, atrial fibrillation, hypothyroid, bone fracture after the age of 45, smoking status, body mass index, diastolic blood pressure, heart rate, haemoglobin, and estimated glomerular filtration rate.

Discussion

In this relatively large sample of patients with HFpEF from the TOPCAT study, pulmonary hospitalization was the most prevalent among non‐CV hospitalizations. Patients with pulmonary hospitalization were more likely to experience HF hospitalization during follow‐up. Except for COPD and smoking, we also identified bone fracture after the age of 45 and previous HF hospitalization as independent risk factors of pulmonary hospitalization. All‐cause and CV mortality risks elevated markedly after first pulmonary hospitalization in HFpEF patients. The all‐cause mortality risk associated with pulmonary hospitalization was comparable with that associated with HF hospitalization. These results highlighted that pulmonary hospitalization was a prevalent clinical endpoint with significant prognostic implication.

In our study, non‐CV admissions accounted for nearly 50%, while HF admission accounted for about 24% of total admissions in HFpEF patients, which were in line with other analyses. One study enrolled 445 HFpEF patients from Emory Healthcare (Atlanta, GA) demonstrated 57.3% of total admissions for non‐CV causes and 28.4% for HF during their follow‐up. In the Irbesartan in Heart Failure and Preserved Ejection Fraction (I‐Preserve) trial, hospitalizations for non‐CV causes constituted 43.7% and those for HF constituted 21.1% of total hospitalizations during the 49 months of follow‐up. However, little was known about the constitution and details of non‐CV admissions in HFpEF patients. We found that in TOPCAT, pulmonary diseases were the primary cause in non‐CV admissions (20.58%) and the third cause in total admissions (10.26%). This indicates that pulmonary hospitalization was a considerable health care burden among HFpEF patients.

COPD, smoking, bone fracture after the age of 45, and previous HF admission were identified as independent risk factors for pulmonary hospitalization among HFpEF patients in the current study. COPD is one of the most common airway diseases. Cigarette smoking has been proved to be related to various pulmonary diseases. , Therefore, it is reasonable that they were associated with pulmonary hospitalization. Interestingly, bone fracture after 45 years of age was also found to be an independent predictor for pulmonary hospitalization. Bone fracture could be caused by steroid therapy in COPD patients. Therefore, it might be speculated that bone fracture is only a confounding factor of the relation between COPD and pulmonary hospitalization. However, we found no significant association between steroid therapy and bone fracture, or between COPD and bone fracture (data not shown). Steroid was not associated with pulmonary hospitalization in multivariate Cox analysis, while bone fracture remained to be an independent risk factor. Therefore, bone fracture in this population was not likely the result of steroid therapy in COPD and was not likely to be a confounding factor in predicting pulmonary hospitalization. Bone fracture may elevate the risk of pneumonia in patients with old age, co‐morbidities, and a high Injury Severity Score. , , Studies have shown the rate of pneumonia in patients with rib fractures varied between 11% and 30%. In TOPCAT, the age of patients was relatively old (mean age 71.5 years old); hence, bone fracture might be associated with pulmonary hospitalization by increasing the risk of infectious lung diseases in our study. Pulmonary congestion is a common finding in HF patients. Dwyer et al. used lung ultrasound to identify lung congestion and found that ambulatory patients with HFpEF demonstrated a similar spectrum of pulmonary congestion as those with HF with reduced ejection fraction (HFrEF). They also discovered that chronic HF patients with lung congestion had a higher risk of 12 months' HF hospitalization and all‐cause mortality. Therefore, patients with HF hospitalization may present more severe lung congestion and be compulsive for bed rest, which would consequently increase the risk of pneumonia and pulmonary hospitalization as well.

We discovered that pulmonary hospitalization was significantly correlated with HF hospitalization. As mentioned above, HF hospitalization increased the risk of pulmonary hospitalization, and vice versa, exacerbation of lung diseases could also aggravate HF. Lung infection, impaired lung function, and hypoxaemia might explain the phenomenon. Pneumonia has been proved to be one of the most important factors that leads to the exacerbation of HF. Silvestre et al. analysed 10 351 patients in Atherosclerosis Risk in Communities (ARIC) Study and demonstrated that the rapid decline of lung function was associated with the incidence of HF. Another study enrolling patients from Gutenburg Health Study discovered that per cent forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC), and FEV1/FVC ratio were significantly related to both HFrEF and HFpEF. Besides, hypoxaemia was also found to be significantly associated with CV hospitalization. Inflammation might be one of the most important mechanisms that account for the association between pulmonary and HF hospitalization. Inflammation was present in most patients with lung diseases that derived from the combination of genetic susceptibility and environmental influences. It also participated in the pathophysiologic process of the development of HFpEF. The co‐morbidities in HFpEF patients created an inflammatory status and damaged the myocardium, which led to cardiac remodelling. Besides, some medications for lung diseases such as long‐acting beta2‐agonists (LABA) would induce tachycardia and shortened diastolic filling period, which might worsen HF.

Pulmonary admission was found to be associated with all‐cause and CV mortality in our study. A more challenging observation is that the risk of all‐cause mortality in patients with first pulmonary hospitalization was comparable with that in patients with first HF hospitalization. Exacerbation of most lung diseases, like COPD, asthma, interstitial diseases, and cancer, would bring about hypoxaemia, impaired lung function, and diffusion capacity, which elevated mortality risk in HFpEF patients. One study analysed data from the ARIC Study Heart Failure Community Surveillance and found that in patients admitted with acute decompensated HFpEF and are >55 years old, hypoxia predicts 28 day and 1 year mortality. In another study that enrolled 71 newly diagnosed HFpEF patients, airflow limitation independently predicted all‐cause mortality. A retrospective study showed that in patients with HFpEF and pulmonary hypertension, low diffusion capacity of the lung increased death risk by over six times. Therefore, hypoxaemia, impaired lung function, and diffusion capacity induced by the aggravation of lung diseases may be a reason for high mortality risk in HFpEF patients with first pulmonary hospitalization.

Our study demonstrated that pulmonary hospitalization was a considerable health issue and significantly increased death risk among HFpEF patients. These results implicated that exacerbation of lung diseases should be paid more attention to and carefully managed. However, some medications for the aggravation of COPD or asthma, like LABA, may not be suitable for decompensated HF, and the treatment should be carefully chosen. Whether the reduction of pulmonary admission would indeed reduce the mortality risk among HFpEF patients was not certain. Future studies are needed to illustrate this issue.

The strengths of our study included that it was the first study to identify the association between pulmonary hospitalization and adverse outcome in HFpEF patients. We studied patients in the TOPCAT trial, which had valid follow‐up data. The outcome was adjudicated by a professional committee. Our study also had some limitations. First, it was a secondary analysis of data from the TOPCAT trial. Second, patients enrolled in Russia and Georgia were excluded, and the patients in our analysis were mainly from the USA. Thus, we are not sure whether our results are suitable for other races. Third, in the analysis of the association between hospitalization and mortality, only the cause of the first hospitalization was taken into consideration, which could introduce bias. Fourth, hospitalization due to right heart dysfunction could be mistakenly recognized as pulmonary hospitalization, because of lack of pulmonary ‐edema. However, among 473 patients with data of right ventricular fractional area change, only 25 had right heart dysfunction. Given the relatively low prevalence, the major conclusion was not likely to be significantly affected by this limitation. Fifth, our study lacks strict definition of pulmonary hospitalization. However, in a HF trial like TOPCAT, this is almost inevitable, as pulmonary hospitalization is seldom an outcome of interest. This is one of the reasons why we performed this analysis, to highlight the clinical significance of pulmonary hospitalization in HFpEF patients.

Conclusions

In conclusion, we analysed 1714 patients in the TOPCAT trial and found that COPD, bone fracture after the age of 45, previous HF hospitalization, and smoking were independent risk factors of pulmonary hospitalization in HFpEF patients. Patients with pulmonary hospitalization were more likely to experience HF hospitalization during follow‐up. Mortality risk increases associated with pulmonary hospitalization were comparable with those associated with HF hospitalization. Because of its correlation with HF hospitalization and significant prognostic implication, pulmonary hospitalization might partially reflect the worsening of HF and should be regarded as an important clinical endpoint in HFpEF.

Conflict of interest

Bin Dong, Xin He, Ruicong Xue, Yili Chen, Jingjing Zhao, Wengen Zhu, Weihao Liang, Zexuan Wu, Dexi Wu, Huiling Huang, Yuanyuan Zhou, Yugang Dong, and Chen Liu declare that they have no conflict of interest.

Funding

This work was supported by National Natural Science Foundation of China (81500279, 81570354, 81770392, 81770394, 81700344, 81800344, 81800345, and 81970340), Guangdong Natural Science Foundation (2016A030310180 and 2017A030310311), Science and Technology Program Foundation of Guangzhou (201610010125), Science and Technology Program Foundation of Guangdong (2017A020215156), Medical Research Foundation of Guangdong Province (A2018107 and A2018082), and China Postdoctoral Science Foundation (2019M663312, 2019TQ0380, and 2019M660229).

Figure S1. Kaplan–Meier curves for any hospitalization in 4 groups according to the causes of the first hospitalization.

Click here for additional data file.