Associations between serum electrolyte and short-term outcomes in patients with acute decompensated heart failure

Figures & data

Figure 1. Total Heb-ADHF population; exclusions and the final study population. Heb-ADHF: Hebei-acute decompensated heart failure; STC: serum total calcium; ADHF: acute decompensated heart failure.

Table 1. Pre-defined serum electrolytes concentration intervals.

Serum electrolyte concentration (mmol/L)
Potassium	Interval 1 (hypokalemia) <3.50	Interval 2 (Ref.) 3.50–4.00	Interval 3 4.01–4.50	Interval 4 4.51–5.00	Interval 5 (hyperkalemia) >5.00
Sodium	Interval 1 <130	Interval 2 (hyponatremia) 130–134	Interval 3 135–140	Interval 4 (Ref.) 141–145	Interval 5 (hypernatremia) >145
Chloride	Interval 1 (hypochloremia) <95	Interval 2 95–100	Interval 3 (Ref.) 101–105	Interval 4 (hyperchloremia) >105
STC	Interval 1 <2.00	Interval 2 (hypocalcemia) 2.00–2.24	Interval 3 (Ref.) 2.25–2.58

STC: serum total calcium.

Table 2. Baseline characteristics concerning four groups of participants classified by the number of electrolyte abnormalities.

Characteristics	Level	0 (n = 3465)	1 (n = 1135)	2 (n = 283)	3 (n = 83)	p Value
Age (years)	Median (IQR)	68 (60–76)	70 (61–78)	73 (63–80)	68 (58–76)	<0.001
Older people (≥65 years)	n (%)	2147 (62.0)	756 (66.6)	202 (71.4)	53 (63.9)	0.001
Male	n (%)	2401 (58.9)	651 (57.4)	151 (53.4)	37 (44.6)	0.019
BMI (kg/m^2)	Median (IQR)	24.5 (22.2–27.0)	24.2 (21.6–26.4)	23.8 (20.8–26.6)	23.4 (20.6–25.5)	<0.001
Missing	n (%)	792 (22.9)	312 (27.5)	99 (35.0)	29 (34.9)
First admission	n (%)	1841 (53.1)	542 (47.8)	101 (35.7)	23 (27.7)	<0.001
Smoking	n (%)	875 (25.3)	255 (22.5)	51 (18.0)	13 (15.7)	0.004
Drinking	n (%)	738 (21.3)	230 (20.3)	46 (16.3)	9 (10.8)	0.027
NYHA class	n (%)					<0.001
II		398 (11.5)	78 (6.9)	8 (2.8)	3 (3.6)
III		1106 (31.9)	288 (25.4)	76 (26.9)	12 (14.5)
IV		1417 (40.9)	553 (48.7)	153 (54.1)	61 (73.5)
Killip class
II		364 (10.5)	146 (12.9)	29 (10.2)	4 (4.8)
III		125 (3.6)	38 (3.3)	12 (4.2)	3 (3.6)
IV		55 (1.6)	32 (2.8)	5 (1.8)	0 (0)
Comorbidity	n (%)
CAD		1854 (53.5)	668 (58.9)	160 (56.5)	30 (36.1)	<0.001
AMI		652 (18.8)	261 (23.0)	56 (19.8)	10 (12.0)	0.006
Hypertension		2023 (58.4)	681 (60.0)	142 (50.2)	39 (47.0)	0.004
VHD		500 (14.4)	156 (13.7)	51 (18.0)	21 (25.3)	0.012
DCM		484 (14.0)	143 (12.6)	43 (15.2)	17 (20.5)	0.170
DM		866 (25.0)	338 (29.8)	76 (26.9)	25 (30.1)	0.013
Stroke/TIA		590 (17.0)	224 (19.7)	55 (19.4)	10 (12.0)	0.082
A fib/A flutter		996 (28.7)	317 (27.9)	89 (31.4)	33 (39.8)	0.101
CKD		190 (5.5)	93 (8.2)	20 (7.1)	15 (18.1)	<0.001
Physical examinations	Median (IQR)
Heart rate (bpm)		84 (70–99)	84 (72–100)	86 (72–102)	78 (67–102)	0.079
SBP (mmHg)		130 (117–147)	130 (114–150)	122 (109–140)	113 (102–134)	<0.001
DBP (mmHg)		80 (70–90)	79 (70–90)	73 (65–86)	70 (62–84)	<0.001
LVEF (%)		45 (37–57)	44 (35–56)	44 (34–56)	44 (33–57)	0.002
Missing	n (%)	83 (2.4)	24 (2.1)	12 (4.2)	2 (2.4)
Laboratory tests	Median (IQR)
BNP (pg/mL)		649 (287–1360)	917 (393–1742)	971 (423–2018)	1180 (698–2180)	<0.001
Missing	n (%)	1244 (35.9)	439 (38.7)	117 (41.3)	36 (43.4)
NT-proBNP (pg/mL)		3902 (1895–7521)	5057 (2350–9867)	6597 (2442–12,960)	7315 (3464–15,039)	<0.001
Missing	n (%)	2340 (67.5)	736 (64.8)	171 (60.4)	53 (63.9)
eGFR (mL/min/1.73 m^2)		81 (63–101)	75 (53–96)	65 (47–89)	64 (38–98)	<0.001
Missing	n (%)	40 (1.2)	20 (1.8)	5 (1.8)	1 (1.2)
FBG (mmol/L)		5.6 (4.8–7.0)	5.9 (5.0–7.9)	6.2 (5.2–8.7)	5.9 (5.0–8.4)	<0.001
Missing	n (%)	261 (7.5)	104 (9.2)	32 (11.3)	10 (12.0)
Serum albumin (g/L)		39 (36–42)	38 (34–41)	38 (35–41)	36 (33–40)	<0.001
Missing	n (%)	96 (2.8)	42 (37)	17 (6.0)	4 (4.8)
Potassium (mmol/L)		4.1 (3.8–4.4)	3.7 (3.4–4.5)	4.1 (3.6–4.7)	3.4 (3.3–5.4)	<0.001
Sodium (mmol/L)		140 (138–142)	138 (134–142)	132 (128–134)	130 (127–133)	<0.001
Chloride (mmol/L)		103 (101–106)	101 (98–104)	93 (90–95)	90 (86–93)	<0.001
Baseline medications	n (%)
Loop diuretics		3002 (86.6)	1030 (90.7)	261 (92.2)	73 (88.0)	<0.001
Thiazide diuretics		327 (9.4)	91 (8.0)	26 (9.2)	7 (8.4)	0.545
ACEI/ARBs		1889 (54.5)	583 (51.4)	131 (46.3)	35 (42.2)	0.004
β-blockers		2416 (69.7)	737 (64.9)	177 (62.5)	52 (62.7)	0.002
MRA (spironolactone)		2833 (81.8)	952 (83.9)	232 (82.0)	76 (91.6)	0.057
Digitalis		789 (22.8)	274 (24.1)	77 (27.2)	26 (31.3)	0.099
Nitrates		1710 (49.4)	600 (52.9)	127 (44.9)	32 (38.6)	0.009
Levosimendan		158 (4.6)	48 (4.2)	14 (4.9)	7 (8.4)	0.337
Nesiritide		262 (7.6)	114 (10.0)	35 (12.4)	11 (13.3)	0.002
Statin		2358 (68.1)	775 (68.3)	165 (58.3)	40 (48.2)	<0.001

BMI: body mass index; NYHA: New York Heart Association; CAD: coronary artery disease; AMI: acute myocardial infarction; VHD: valvular heart disease; DCM: dilated cardiomyopathy; DM: diabetes mellitus; TIA: transient ischemic attack; A fib/A flutter: atrial fibrillation/atrial flutter; CKD: chronic kidney disease; SBP: systolic blood pressure; DBP: diastolic blood pressure; LVEF: left ventricular ejection fraction; BNP: brain natriuretic peptide; NT-pro BNP: N-terminal pro brain natriuretic peptide; eGFR: estimated glomerular filtration rate; FBG: fasting blood glucose; ACEI/ARBs: angiotensin-converting enzyme inhibitors/ angiotensin receptor blockers; MRA: mineralocorticoid receptor antagonist.

Percentages may not total 100 because of rounding.

Table 3. Short-term outcomes in participants according to different study populations.

K^+-related study population	All (5145), n (%)	Interval 1 (n = 631)	Interval 2 (n = 1784)	Interval 3 (n = 1726)	Interval 4 (n = 720)	Interval 5 (n = 284)	p Value
All-cause mortality	287 (5.6)	39 (6.2)	74 (4.1)	96 (5.5)	49 (6.7)	29 (10.2)	<0.001
In-hospital	211 (4.1)	30 (4.8)	57 (3.2)	71 (4.1)	34 (4.7)	19 (6.7)	0.041
30-day post-discharge	76 (1.5)	9 (1.4)	17 (0.9)	25 (1.4)	15 (2.0)	10 (3.5)	0.014
All-cause 30-day readmission	397 (7.8)	51 (8.1)	120 (6.8)	136 (7.9)	61 (8.5)	29 (10.3)	0.223
HF	275 (5.3)	38 (6.0)	76 (4.3)	93 (5.4)	44 (6.1)	24 (8.5)	0.026
Non-HF	122 (2.4)	13 (2.1)	44 (2.5)	43 (2.5)	17 (2.4)	5 (1.8)	0.925
Na^+-related study population	All (5135), n (%)	Interval 1 (n = 170)	Interval 2 (n = 481)	Interval 3 (n = 2378)	Interval 4 (n = 1831)	Interval 5 (n = 275)	p Value
All-cause mortality	286 (5.6)	21 (12.9)	52 (10.8)	137 (5.8)	66 (3.6)	9 (3.3)	<0.001
In-hospital	210 (4.1)	15 (8.8)	37 (7.7)	99 (4.2)	53 (2.9)	6 (2.2)	<0.001
30-day post-discharge	76 (1.5)	7 (4.1)	15 (3.1)	38 (1.6)	13 (0.7)	3 (1.1)	<0.001
All-cause 30-day readmission	396 (7.8)	22 (13.0)	47 (9.8)	203 (8.6)	112 (6.1)	12 (4.4)	<0.001
HF	275 (5.4)	19 (11.2)	34 (7.1)	144 (6.1)	70 (3.8)	8 (2.9)	<0.001
Non-HF	121 (2.4)	3 (1.8)	13 (2.7)	59 (2.5)	42 (2.3)	4 (1.5)	0.851
Cl^–-related study population	All (4966), n (%)	Interval 1 (n = 381)	Interval 2 (n = 1297)	Interval 3 (n = 1964)	Interval 4 (n = 1324)	p Value
All-cause mortality	280 (5.6)	49 (12.9)	93 (7.2)	77 (3.9)	61 (4.6)	<0.001
In-hospital	207 (4.2)	33 (8.7)	66 (5.1)	62 (3.2)	46 (3.5)	<0.001
30-day post-discharge	73 (1.4)	16 (4.2)	27 (2.1)	15 (0.7)	15 (1.1)	<0.001
All-cause 30-day readmission	388 (7.8)	49 (12.8)	111 (8.5)	139 (8.0)	89 (6.7)	<0.001
HF	269 (5.4)	42 (11.0)	73 (5.6)	93 (4.7)	61 (4.6)	<0.001
Non-HF	119 (2.4)	7 (1.8)	38 (2.9)	46 (2.3)	28 (2.1)	0.464
STC-related study population	All (4143), n (%)	Interval 1 (n = 500)	Interval 2 (n = 1872)	Interval 3 (n = 1771)	p Value
All-cause mortality	234 (5.7)	29 (5.8)	122 (6.5)	83 (4.7)	0.057
In-hospital	173 (4.2)	21 (4.2)	90 (4.8)	62 (3.5)	0.143
30-day post-discharge	61 (1.5)	8 (1.6)	32 (1.7)	21 (1.2)	0.410
All-cause 30-day readmission	306 (7.4)	50 (10.0)	141 (7.5)	115 (6.5)	0.150
HF	210 (5.1)	29 (5.8)	99 (5.3)	82 (4.6)	0.484
Non-HF	96 (2.3)	21 (4.2)	42 (2.2)	33 (1.9)	0.009

STC: serum total calcium.

Figure 2. Final adjusted odds ratios with 95% confidence interval of a primary composite outcome in the four serum electrolyte-related study populations.

Figure 3. Final adjusted odds ratios with 95% confidence interval comparing a composite outcome among different number of electrolyte abnormalities in the chloride-related study population.

Table 4. Odds ratios with the 95% confidence interval for different logistic regression analyses models for a composite outcome (chloride-related study population).

Number of electrolyte abnormalities
Model	N = 0 n = 3465	N = 1 n = 1135	N = 2 n = 283	N = 3 n = 83	p for trend
Crude model p Value	1 (reference)	1.59 (1.29–1.95) <0.001	3.24 (2.41–4.36) <0.001	3.47 (2.09–5.77) <0.001	<0.001
Model 1 p Value	1 (reference)	1.57 (1.28–1.94) <0.001	3.21 (2.38–4.32) <0.001	3.54 (2.13–5.89) <0.001	<0.001
Model 2 p Value	1 (reference)	1.42 (1.15–1.76) 0.001	2.74 (2.02–3.72) <0.001	2.83 (1.68–4.78) <0.001	<0.001
Model 3 p Value	1 (reference)	1.40 (1.13–1.73) 0.002	2.51 (1.85–3.42) <0.001	2.47 (1.45–4.19) 0.001	<0.001

Model 1: adjusted for age and sex. Model 2: further adjusted for first admission, smoking, NYHA or Killip functional class, CAD, AMI, hypertension, VHD, DM, Stroke/TIA, A fib/A flutter and CKD. Model 3: further adjusted for heart rate, SBP, DBP, loop diuretics, ACEI/ARBs, β-blockers, MRA, digitalis, nitrates, nesiritide and statin.

Table 5. Sensitivity analyses: logistic regression models of a composite outcome stratified by number of electrolyte abnormalities.

Number of electrolyte abnormalities	A total number of patients (n = 4966)				After exclusion of patients with NYHA II (n = 4479)
	n	Adjusted^a OR (95% CI)	Adjusted^b OR (95% CI)	Adjusted^c OR (95% CI)	n	Adjusted^a OR (95% CI)	Adjusted^b OR (95% CI)	Adjusted^c OR (95% CI)
N = 0	3465	1 (reference)	1 (reference)	1 (reference)	3067	1 (reference)	1 (reference)	1 (reference)
N = 1	1135	1.40 (1.13–1.73)	1.23 (0.98–1.56)	1.32 (1.07–1.64)	1057	1.47 (1.18–1.83)	1.31 (1.03–1.67)	1.39 (1.12–1.73)
N = 2	283	2.51 (1.85–3.42)	2.27 (1.61–3.19)	2.34 (1.71–3.20)	275	2.52 (1.85–3.45)	2.27 (1.60–3.21)	2.35 (1.71–3.22)
N = 3	83	2.47 (1.45–4.19)	2.36 (1.32–4.20)	2.24 (1.31–3.82)	80	2.36 (1.37–4.07)	2.41 (1.34–4.33)	2.14 (1.24–3.70)
p for trend		<0.001	<0.001	<0.001		<0.001	<0.001	<0.001

OR: odds ratio.

^aLogistic regression model adjusted for age, sex, first admission, smoking, NYHA or Killip functional class, CAD, AMI, hypertension, VHD, DM, stroke/TIA, A fib/A flutter, CKD, heart rate, SBP, DBP, loop diuretics, ACEI/ARBs, β-blockers, MRA, digitalis, nitrates, nesiritide and statin.

^bLogistic regression model further adjusted for not imputed LVEF, eGFR, serum albumin and log (BNP or NT-pro BNP).

^cLogistic regression model further adjusted for multiple imputed LVEF, eGFR, serum albumin and log (BNP or NT-pro BNP).

Data availability statement

The data used to support the findings of this study are available from the corresponding author upon request.