The baseline characteristics of the study participants are shown in Table 1. A total of 402 eligible participants with a mean age of 69.05 years were enrolled at the start of the study. Of this number, 52 participants were lost between the first and second follow-up. Of the 350 participants who were fully enrolled in the study and followed up 7.6 years later, 173 were alive (49.43%) while 177 had died (50.57%). We compared the baseline characteristics of the patients with these two outcomes at the time of enrollment and found the following results: The age of the death group was significantly higher than that of the survival group (64.8 vs. 73.3, P < 0.001), and the proportion of patients with coronary heart disease, stroke, and diabetes in the death group was higher than that in the survival group (P = 0.041, P < 0.001, P = 0.004, respectively). These outcomes indicate that a history of cardiovascular and cerebrovascular disease and diabetes has a substantial impact on the long-term prognosis of patients. Owing to the non-normal distribution of the overall data, we calculated the eGFR of the two groups using the interquartile distance (92.0 vs. 65.0), and the result was significantly higher in the survival group than in the death group (P < 0.001). We used the same method to calculate the SBP and DBP of the two groups and found that the survival group was significantly lower than the death group (152 vs. 159, P < 0.001; 91 vs. 95, P = 0.003).
We divided the patients into three subgroups based on their eGFR values and age (Fig. 1). The eGFRc=1 subgroup was the population without kidney disease, the eGFRc=3 subgroup comprised those with kidney damage, and the eGFRc=2 subgroup consistent of the subclinical population with decreased eGFR. Among the three subgroups, eGFRc=1 had the highest survival rate (85.5%), followed by eGFRc=2 (12.1%), and lastly, eGFRc=3 (2.3%). These findings indicate that kidney function is associated with long-term all-cause death in patients with hypertension.
Table 1: Baseline characteristics of study participants
Characteristics
|
Survivors
n = 173
|
Death
n = 177
|
P value
|
Demographic
|
|
|
|
Age (SD*)
|
64.8 (9.3)
|
73.3 (10.6)
|
<0.001
|
Male, n (%)
|
77 (44.5%)
|
76 (42.9%)
|
0.770
|
Medical history, n (%)
|
|
|
|
CHD*
|
74 (42.8%)
|
95 (53.7%)
|
0.041
|
antihypertensive drugs
|
157 (90.8%)
|
161 (91.0%)
|
0.951
|
stroke
|
13 (7.5%)
|
43 (24.3%)
|
<0.001
|
DM*
|
46 (26.6%)
|
73 (41.2%)
|
0.004
|
Laboratory tests (IQR*)
|
|
|
|
eGFR, mL/min/1.73 m2
|
92.0 (77.0, 102.0)
|
65.0 (43.0, 85.0)
|
<0.001
|
TC*, mg/dL
|
4.5 (4.0, 5.4)
|
4.5 (3.9 5.3)
|
0.45
|
TG*, mg/dL
|
1.3 (1.0, 1.9)
|
1.2 (0.9, 1.8)
|
0.060
|
LDL-C, mg/dL
|
2.8 (2.3, 3.3)
|
2.7 (2.2, 3.3)
|
0.18
|
HDL-C, mg/dL
|
1.1 (0.9, 1.3)
|
1.1 (1.0, 1.3)
|
0.87
|
Vital signs
|
|
|
|
SBP*, mmHg
|
152.0 (140.0, 164.0)
|
159.0 (146.0, 172.0)
|
<0.001
|
DBP*, mmHg
|
91.0 (83.0, 99.0)
|
95.0 (87.0, 106.0)
|
0.003
|
eGFR, mL/min/1.73 m2
|
|
|
<0.001
|
1
|
148 (85.5%)
|
86 (48.6%)
|
|
2
|
21 (12.1%)
|
42 (23.7%)
|
|
3
|
4 (2.3%)
|
49 (27.7%)
|
|
Abbreviations:
*SD, standard deviation; CHD, coronary atherosclerotic heart disease; DM, diabetes mellitus; IQR, interquartile range; TC, total cholesterol; TG, triglyceride; SBP, systolic blood pressure; DBP, diastolic blood pressure.
Association of eGFR with survival outcomes
Based on the latest literature,9–11 patients were divided into three subgroups according to age and eGFR value: eGFRc=1, eGFRc=2, and eGFRc=3 (Fig. 1). K–M analyses showed that the low eGFR groups had seriously inferior rates of survival over time. In addition, the survival curve of the eGFRc=3 population decreased the fastest, followed by the eGFRc=2 population, and lastly by the eGFRc=1 population, which had the lowest mortality. A log-rank test with p = 0.000 indicates a close association between the mentioned groups and long-term mortality in patients with hypertension in the long-term follow-up (Fig. 2). The COX proportional risk model was established with eGFRc=1 as reference (Fig. 3). The result showed that lower eGFR groups had significantly lower rates of survival (hazard ratio [HR] eGFRc=2 = 2.407, 95% confidence interval [CI]: 1.663–3.484, P = 0.000; HR eGFRc=3 = 7.081, 95% CI: 4.925–10.179, P = 0.000). From the results, we can see that the combination of age and eGFR grouping can effectively correlate eGFR and outcome. Interestingly, the average risk of death of patients in the eGFRc=3 group is 7.081 times higher than that in the eGFRc=1 group.
Relationship between covariates and outcomes
We analyze the potential prognostic clinicopathological factors by univariate and multivariate Cox regression and calculate the independent prognostic parameters (age, ACR, CHD, anti-htn drugs, DM, stroke, SBP, DBP) and risk score using univariate Cox regression (Table 2). The results show the following: Age (HR = 1.067, 95% CI: 1.051–1.083), ACR (HR = 1.006, 95% CI: 1.005–1.009), CHD (HR = 1.380, 95% CI: 1.027–1.856), anti-HTN drugs (HR = 0.964, 95% CI: 0.577–1.612), DM (HR = 1.640, 95% CI: 1.215–2.213), stroke (HR = 2.376, 95% CI (1.682–3.357)], SBP (HR = 1.026, 95% CI (1.006–1.028), and DBP (HR = 1.087, 95% CI (1.022–1.035)]. We then adjusted the confounding factors according to the above results. Finally, DM, stroke, age, SBP, DBP, ACR, and eGFRc were included in the prediction model (Table 2) (Figure 4). Since the eGFRc grouping was related to age, age could not be included in the model. The results showed that after adjusting for confounding factors, ACR (HR = 1.006, 95% CI: 1.004–1.008), DM (HR = 1.476, 95% CI: 1.076–2.026), stroke (HR = 1.949, 95% CI: 1.362–2.789), SBP (HR = 1.013, 95% CI: 1.006–1.021), and DBP (HR = 1.011, 95% CI: 1.000–1.022) were still associated with all-cause death outcomes (P < 0.05). eGFRc=2 and eGFRc=3 still had strong predictive value for outcome, especially eGFRc=2, which exhibited a higher HR after adjusting for confounding factors. This result indicates a very close relationship between subclinical eGFR decline and survival.
Table 2: Results of analysis of factors associated with all-cause mortality
Time-dependent ROC analysis
The results of time-dependent ROC analysis for model 1 revealed that the discriminatory ability of the all-cause mortality was robust. DM, stroke, age, SBP, DBP, and Acr were integrated into nomogram model 1, whose AUC = 0.798; the other model added age-adapted eGFR criteria based on the relevant factors of model 1. The AUC of this new model is 0.827, and its predictive value is significantly higher than that of model 1, P = 0.0105 (Table 3).
Table 3: Predictive ability of the age-adapted eGFR criteria for death.
Model
|
AUC
|
P value
|
*Model 1
|
0.798
|
Ref.
|
Model 1+ age-adapted eGFR criteria
|
0.827
|
0.0105
|
Abbreviations: AUC, area under curve.
*Model 1: DM, Stroke, Age, SBP, DBP, ACR.