Abstract
Background:
Previous studies on the survival of continuous ambulatory peritoneal dialysis (CAPD) patients have been done using the Cox model. This model requires establishing the basic assumption of proportional hazard (PH) and not establishing it leads to misleading inferences. Alternatively, if the time-to-event distribution is known, accelerated failure time models can be developed in such a way that no PH assumption is required. The purpose of this study is to compare the survival of CAPD patients in two hypertensive and non-hypertensive groups in the condition of not establishing the PH assumption.
Materials and Methods:
In this retrospective cohort study, 1,400 patients who were referred to dialysis centers from different parts of Iran between 1995 and 2010 were included. The loglogistic, lognormal, and generalized gamma parametric frailty models were compared using Stata 17 software.
Results:
The five-year survival rate was 54%. The median survival time for the hypertensive group was lower than for others. Age, body mass index, fast blood sugar, potassium, and creatinine were higher in hypertensive patients. In the early months of dialysis, the risk of death is high and decreases with time. Univariate results revealed that hypertension leads to acceleration of death (HR = 0.79; 95% CI (0.64,0.98)). Multivariate analysis showed that increasing age (HR = 0.97; 95% CI (0.96,0.98)), comorbidity (HR = 0.71; 95% CI (0.55,0.91)) and high blood sugar (HR = 0.997; 95% CI (0.999)) accelerated the risk of death. An increase in serum albumin (HR = 1.27; 95% CI (1.04,1.57)) leads to a delay in death event.
Conclusions:
Parametric models (especially the loglogistic model) are appropriate in evaluating factors affecting the survival of CAPD patients.
Keywords: Blood pressure, frailty, peritoneal dialysis, survival analysis
INTRODUCTION
Kidney failure is one of the non-communicable diseases whose prevalence has increased in recent years due to the epidemiological transmission of diseases.[1] According to the latest systematic review and meta-analysis study conducted on the prevalence of chronic kidney disease in Iran, its prevalence is 15.14%, which is higher than the global average.[2] In recent years, the development of chronic kidney failure has intensified in the final stages and has caused a sharp decrease in kidney function even over a short period of a few months. In general, the treatments used for this disease include kidney transplantation, hemodialysis, and peritoneal dialysis, and since transplantation is not possible for all patients, most patients have to undergo one of the dialysis methods. Dialysis is a treatment that performs some normal kidney functions and is needed when the kidneys are unable to meet the body’s needs. Peritoneal dialysis has been used more than hemodialysis due to its advantages such as lower cost and flexibility in lifestyle.[3] Peritoneal dialysis is performed in two ways, including continuous ambulatory peritoneal dialysis (CAPD) and automated peritoneal dialysis.
One of the benefits of dialysis is the blood pressure control of dialysis patients. High blood pressure and its fluctuations are common symptoms in kidney patients. This chronic disease leads to damage to vital organs, including eyes, heart, and kidneys. Among the factors affecting the increase in blood pressure are comorbidity, such as diabetes, obesity, high blood fat, and the body’s reaction to sodium, chlorine, and calcium ions.[4]
Today, due to the increasing use of survival studies in medical research, the need for more efficient statistical models to analyze these data seems necessary. In many studies of survival data of kidney patients, the Cox regression model is used. When using this model, the proportional hazard (PH) assumption needs to be established. In addition, parametric models are a more suitable alternative to the Cox model in the condition of knowing the shape of the hazard function. If the hazard function has a uniform shape, the Weibull parametric model is recommended, and if the shape is non-monotonic, it may be desirable to use lognormal, loglogistic, or gamma models.[5] For kidney patients undergoing dialysis, the risk of death may not be constant and may increase or decrease over time depending on some factors.
In most survival models, the population under study is considered homogeneous, but sometimes there are unknown factors that not only affect the survival of patients but also cause the population to become heterogeneous, and on the other hand, they may not be measurable. To compensate for the negative effects of this phenomenon, random effect models (i.e., frailty model) are used. Frailty is a random component to account for unobserved changes in survival time, which is entered as a multiplicative effect in the hazard function.[6] In the current study, the frailty model was used to control the effect of non-measurable factors, including genetic factors.
Many studies have been conducted on renal failure disease, but the investigation of factors affecting the survival of peritoneal dialysis patients, especially with the continuous ambulatory method, has received less attention. Also, despite the fact that blood pressure control is important in the management of renal failure disease, the role of high blood pressure on the survival of Iranian CAPD patients has not been investigated so far. In addition, the possible functional form of survival time is still unclear for these patients. Therefore, to solve these defects, in the present study, the effect of blood pressure on the survival of CAPD patients has been investigated. Also, a suitable parametric model has been proposed to evaluate the survival of these patients.
MATERIALS AND METHODS
The present study is a retrospective cohort study and its data are related to patients with kidney failure who are treated with CAPD. The data were provided from the Iranian peritoneal dialysis registry project.[7] The Department of Organ Transplantation and Special Diseases of the Ministry of Health of Iran has approved the executive and operational duties of this project.
In this project, patients were referred to dialysis centers across the country (48 centers) between 1995 and 2010, and a medical file was prepared for each of them in the respective center. Information is recorded electronically and compiled by an integrated program under a similar format. All steps of peritoneal dialysis have been performed under the same protocol and all its operators have been trained to perform this process. The patients were given the necessary information to participate in the study, and those with full consent were included. Each person was identified by only one ID; other personal information was kept confidential and could not be extracted.
In these data, basic characteristics such as sex (male/female), age (year), weight (kg), height (m), body mass index/BMI (kg/m2), comorbidity (yes/no), triglycerides/Trg (mg/dL), fast blood sugar/FBS (mg/dL), protein albumin (g/dL), serum sodium/Na (mg/L), potassium/K (mol/L), creatinine/Cr (mg/dL), and blood urea nitrogen/BUN (mg/dL) have been collected from each patient upon entering the plan. To determine the blood pressure of patients, the guidelines provided by the American College of Cardiology/American Heart Association were used.[8] Blood pressure levels were determined according to this guide for each person when going for dialysis, and in the analysis of understudy data, patients with degrees I and II of hypertension were included in the group of hypertensive and other patients were included in the non-hypertensive group.
The inclusion criteria for the study included the prescription of CAPD treatment and being over 18 years old. Exclusion criteria included kidney transplant or hemodialysis. Also, people for whom the desired information was not complete were excluded.
Statistical analysis
In the data analysis, the number and percentage indices were used to describe the qualitative variables, and the mean and standard deviation indicators were used to describe the quantitative variables. The student’s T-test was used to compare the mean of the groups and to compare the percentage of categories of qualitative variables between groups, Chi-square test was performed. The survival time of patients was calculated in months. The minimum follow-up time was 5 days and the maximum was 15 years. The survival time for patients who survived until the end of the study or whose information was not available was considered as right censored. Wilcoxon test and Kaplan–Meier chart were used to compare survival probability between groups. Survival regression models were used to investigate the factors affecting the survival of patients. For this purpose, univariate analysis was used first, and then significant variables (P value < 0.1) in univariate analysis were entered in multivariate approach. For survival regression analysis, parametric survival models (loglogistic, lognormal, and also generalized gamma) were used under the assumption of accelerated failure time (AFT). Also, in order to include random effects, the frailty parameter with gamma distribution was also added to these models. Finally, the best parametric model was selected by Akaike information criteria (AIC) and Bayesian information criteria (BIC).[9] Statistical analysis of the data was done by Stata 17 software, and the value of 0.05 was considered as the significance level.
RESULTS
In the present study, the data of 1,400 peritoneal dialysis patients who entered the study with hypertension or without hypertension were used (700 patients equally from each group). Of all the patients under study, 715 (51.10%) were women and the rest were men. The youngest person was 20 years old and the oldest was 89 years old. The average BMI of all subjects was 23.59 ± 4.65. Also, 1,152 (82.3%) patients had at least one comorbidity. Other basic characteristics of the patients are given in Table 1. Also, in Table 1, patients are compared based on their blood pressure status at the time of entering the study. According to the results of Table 1, patients with high blood pressure were older than others (P value = 0.044). There was a significant difference between women and men in terms of initial blood pressure status (P value < 0.001) and men had the highest percentage (55.3%) of the hypertensive group, while the highest percentage (57.4%) of the non-hypertensive group belonged to women. Also, the mean of factors such as BMI, FBS, K, and Cr in patients with high blood pressure was higher than others (P value < 0.001), while the amount of albumin in these patients was lower than in patients without hypertension (P value < 0.001). No significant difference was found regarding other factors.
Table 1.
Main characteristics of peritoneal dialysis patients by initial blood pressure status
| Variable | Total | Non-hypertensive patients n=700 | Hypertensive patients n=700 | P |
|---|---|---|---|---|
| Age | 51.44±16.160 | 50.64±16.38 | 52.25±15.91 | 0.044 |
| Gender (%) | <0.001 | |||
| Male | 685 (48.9) | 298 (42.6) | 387 (55.3) | |
| Female | 715 (51.1) | 402 (57.4) | 313 (44.7) | |
| BMI | 23.59±4.65 | 23.00±4.96 | 24.17±4.23 | <0.001 |
| Comorbidity (%) | 0.624 | |||
| Yes | 1152 (82.3) | 580 (82.9) | 572 (81.7) | |
| No | 248 (17.7) | 120 (17.1) | 128 (18.3) | |
| FBS | 130.11±67.149 | 124.65±63.20 | 164.86±63.03 | <0.001 |
| Tg | 170.01±77.130 | 175.15±88.77 | 165.57±70.50 | 0.122 |
| K | 4.71±0.73 | 4.64±0.76 | 4.78±0.70 | <0.001 |
| Na | 139.20±3.88 | 139.32±3.91 | 139.07±3.85 | 0.332 |
| Cr | 6.69±2.69 | 6.49±2.71 | 6.90±2.67 | 0.001 |
| Albumin | 3.76±0.40 | 3.79±0.43 | 3.74±0.37 | <0.001 |
| BUN | 76.60±38.55 | 77.86±40.68 | 75.33±36.28 | 0.934 |
Of the total patients, 542 (38.71%) died (women: 260 (48%) and men: 282 (52%)) and the rest (61%) were censored. Table 2 shows the comparison of patients who died or survived until the end of the study. Based on the results of this comparison, factors such as age and FBS were higher in dead patients than in alive patients, while Na, Cr, and albumin levels were lower in dead patients (P value < 0.05). Also, in Table 2, a more detailed comparison has been made by the initial blood pressure status. From the non-hypertensive group 297 (42%) and from the hypertensive group 245 (35%) were deceased. In both groups, the age and FBS in dead patients were higher than those who survived, and Cr and albumin in dead patients were lower than those who survived (P value < 0.05). Although there was no significant difference between women and men in terms of the number of deaths, a more detailed analysis revealed that among patients without hypertension, women had the highest percentage of deaths (50.2%) (P value = 0.001) and in the hypertensive group, death percentage was almost similar between men and women and there was no significant difference (P value = 0.817).
Table 2.
Main characteristics of peritoneal dialysis patients by death and initial blood pressure status
| Variable | Total |
Non-hypertensive patients |
Hypertensive patients |
||||||
|---|---|---|---|---|---|---|---|---|---|
| Alive n=858 | Dead n=542 | P | Alive n=403 | Dead n=297 | P | Alive n=455 | Dead n=245 | P | |
| Age | 47.41±15.66 | 57.82±14.84 | <0.001 | 46.31±15.68 | 56.51±15.47 | <0.001 | 48.39±15.59 | 59.41±13.90 | <0.001 |
| Gender (%) | 0.070 | 0.001 | 0.817 | ||||||
| Male | 403 (47.0) | 282 (52.0) | 150 (37.2) | 148 (49.8) | 253 (55.6) | 134 (54.7) | |||
| Female | 455 (53.0) | 260 (48.0) | 253 (62.8) | 149 (50.2) | 202 (44.4) | 111 (45.3) | |||
| BMI | 23.62±4.55 | 23.53±4.82 | 0.778 | 23.10±5.03 | 22.86±4.88 | 0.799 | 24.08±4.00 | 24.34±4.63 | 0.408 |
| Comorbidity (%) | 0.886 | 0.919 | 0.682 | ||||||
| Yes | 707 (82.4) | 445 (82.1) | 333 (82.6) | 247 (83.2) | 374 (82.2) | 198 (80.8) | |||
| No | 151 (17.6) | 97 (17.9) | 70 (17.4) | 50 (16.8) | 81 (17.8) | 47 (19.2) | |||
| FBS | 122.74±59.74 | 141.79±76.05 | <0.001 | 117.96±81.55 | 133.74±71.12 | 0.003 | 126.70±62.77 | 151.59±80.87 | <0.001 |
| Tg | 170.00±75.26 | 170.02±80.06 | 0.868 | 174.40±82.47 | 176.17±96.80 | 0.411 | 166.74±88.38 | 162.51±52.48 | 0.428 |
| K | 4.74±0.74 | 4.66±0.72 | 0.102 | 4.68±0.76 | 4.58±0.75 | 0.159 | 4.80±0.72 | 4.75±0.66 | 0.537 |
| Na | 139.29±3.78 | 139.05±4.03 | 0.043 | 139.44±3.77 | 139.16±4.09 | 0.120 | 139.15±3.80 | 138.91±3.95 | 0.166 |
| Cr | 6.90±2.72 | 6.37±2.64 | <0.001 | 6.65±2.67 | 6.26±2.75 | 0.012 | 7.11±2.74 | 6.50±2.49 | 0.003 |
| Albumin | 3.80±0.42 | 3.70±0.37 | <0.001 | 3.83±0.47 | 3.73±0.36 | <0.001 | 3.78±0.37 | 3.67±0.37 | <0.001 |
| BUN | 77.38±38.58 | 75.37±38.50 | 0.340 | 80.21± | 74.67±39.30 | 0.074 | 74.87±35.60 | 76.20±37.57 | 0.636 |
For all patients, the 1-, 5-, and 10-year survival rates were 89%, 54%, and 38%, respectively. The median survival time of the patients was estimated by the Kaplan–Meier method to be 72.43 with a standard error of 4.16. Also, for patients in the non-hypertensive group, the median survival time (74.96 with a standard error of 5) was higher than for the hypertensive group (72.43 with a standard error of 8) (P value using the Wilcoxon test is equal to 0.006). Figure 1 compares the probability of survival time for two groups. According to Figure 1, the survival of patients in the non-hypertensive group is higher than other group, but with the passage of time (up to about 80 months after the start of the study), the survival of patients with high blood pressure was increased.
Figure 1.
Kaplan–Meier survival probability diagram for patients treated with continuous ambulatory peritoneal dialysis by blood pressure status
In order to investigate the effective factors in the survival of patients and to compare the survival of the two groups under study, loglogistic, lognormal, and gamma survival models were used under assumption of AFT. Table 3 shows the results of univariate analysis for the mentioned models. According to Table 3, the increase in age and FBS had a decreasing effect on the survival of patients. Increased Cr and albumin increased survival time. Survival of patients in the hypertensive group was lower than that of patients in the non-hypertensive group (all P values < 0.05).
Table 3.
Results of univariate analysis of parametric AFT1 models for peritoneal dialysis patients
| Variable | Models |
||||||||
|---|---|---|---|---|---|---|---|---|---|
| Loglogistic |
Lognormal |
Generalized gamma |
|||||||
| HR2 | 95% CI3 | P | HR | 95% CI | P | HR | 95% CI | P | |
| Age | 0.97* | (0.96,0.98) | <0.001 | 0.97* | (0.96,0.97) | <0.001 | 0.97* | (0.96,0.97) | <0.001 |
| Sex/male | 0.94 | (0.77,1.14) | 0.514 | 0.94 | (0.76,1.16) | 0.572 | 0.94 | (0.77,1.13) | 0.511 |
| Comorbidity/yes | 0.78 | (0.54,1.65) | 0.055 | 0.85 | (0.64,1.13) | 0.258 | 0.78 | (0.60,1.001) | 0.051 |
| BMI | 0.99 | (0.98,1.02) | 0.917 | 0.99 | (0.97,1.02) | 0.663 | 0.99 | (0.98,1.02) | 0.918 |
| Cr | 1.07* | (1.03,1.12) | <0.001 | 1.07* | (1.03,1.17) | <0.001 | 1.07* | (1.03,1.11) | <0.001 |
| Tg | 1.001 | (0.99,1.002) | 0.409 | 1.001 | (0.99,1.002) | 0.467 | 1.001 | (0.99,1.002) | 0.425 |
| K | 1.105 | (0.97,1.26) | 0.129 | 1.09 | (0.95,1.26) | 0.207 | 1.09 | (0.97,1.26) | 0.127 |
| FBS | 0.99* | (0.996,0.998) | <0.001 | 0.99* | (0.995,0.998) | <0.001 | 0.99* | (0.996,0.998) | <0.001 |
| Na | 1.02 | (0.991,1.04) | 0.221 | 1.02 | (0.99,1.05) | 0.098 | 1.01 | (0.99,1.05) | 0.230 |
| Blood pressure condition/high | 0.81 | (0.66,1.001) | 0.051 | 0.79* | (0.64,0.98) | 0.028 | 0.83 | (0.68,1.01) | 0.065 |
| Albumin | 1.39* | (1.13,1.72) | 0.002 | 1.42* | (1.13,1.79) | 0.003 | 1.39* | (1.13,1.72) | 0.002 |
| Bun | 1.001 | (0.99,1.004) | 0.336 | 1.00 | (0.99,1.003) | 0.766 | 1.001 | (0.99,1.004) | 0.323 |
1Accelerated failure time; 2Hazard ratio; 3Confidence interval; *significant at 0.05 level
Table 4 contains the results of multivariate analysis for all three candidate models. In Table 4, the estimation of ancillary parameters for checking the suitability of candidate models is also given. For the loglogistic model, the assumption of Ln gamma = 0 was rejected (P value < 0.001), so this distribution can be considered as a candidate. Also, in this model, the existence of heterogeneity between individuals was not confirmed (P value for Theta = 0 was 1.00). For the lognormal model, the assumption of Ln sigma = 0 was also rejected (P value < 0.001), so the functional form of this model can also be kept as a candidate. Also, the existence of heterogeneity between individuals was not confirmed (P value = 1.00). For the generalized gamma model, although there was no reason to reject the hypothesis of Ln sigma = 0, the hypothesis K = 0 was rejected, so this model can be reduced to the lognormal model. In generalized gamma model, the existence of heterogeneity between individuals was confirmed (P value = 0.006). The results of comparing models showed that according to both criteria of AIC and BIC, the loglogistic model was proposed as the most appropriate model. In the final loglogistic model, the value of the shape parameter was estimated to be 0.85, so the functional form of hazard distribution is a curve that increases up to a maximum point and then decreases over time. According to the results of loglogistic model, the survival of patients decreased by 3% with one-year increase in age (HR = 0.97). Survival of patients with comorbidities was 29% lower than that of others (HR = 0.71). With a one-unit increase in FBS, patient survival decreased by 1% (HR = 0.99). The increase in albumin led to a 27% increase in patient survival (HR = 1.27). Survival of patients with high blood pressure was 10% lower than that of other patients, although this variable did not show a significant effect (HR = 0.90).
Table 4.
Results of multivariate analysis of parametric AFT1 models for peritoneal dialysis patients
| Variable | Models |
||||||||
|---|---|---|---|---|---|---|---|---|---|
| Loglogistic |
Lognormal |
Generalized gamma |
|||||||
| HR2 | 95% CI3 | P | HR | 95% CI | P | HR | 95% CI | P | |
| Age | 0.97* | (0.96,0.98) | <0.001 | 0.97* | (0.96,0.98) | <0.001 | 0.97* | (0.96,0.98) | <0.001 |
| Sex/male | - | - | - | - | - | - | - | - | - |
| Comorbidity/yes | 0.71* | (0.55,0.91) | 0.006 | - | - | - | 0.71* | (0.55,0.91) | 0.007 |
| BMI | - | - | - | - | - | - | - | - | - |
| Cr | 1.02 | (0.99,1.06) | 0.234 | 0.01 | (0.98,1.06) | 0.369 | 1.02 | (0.98,1.05) | 0.293 |
| Tg | - | - | - | - | - | - | - | - | - |
| K | - | - | - | - | - | - | 0.07 | (-0.05,0.19) | 0.279 |
| FBS | 0.997* | (0.995,0.999) | 0.018 | 0.998* | (0.997,0.999) | 0.030 | 0.998* | (0.996,0.999) | 0.019 |
| Na | - | - | - | 1.02 | (0.99,1.04) | 0.218 | - | - | - |
| Blood pressure condition/high | 0.90 | (0.75,1.09) | 0.309 | 0.87 | (0.71,1.07) | 0.194 | 0.89 | (0.73,1.09) | 0.287 |
| Albumin | 1.27* | (1.04,1.57) | 0.020 | 1.27* | (1.01,1.58) | 0.040 | 1.27* | (1.03,1.55) | 0.023 |
| Bun | - | - | - | - | - | - | - | - | - |
|
| |||||||||
| Beta | 95% CI | P | Beta | 95% CI | P | Beta | 95% CI | P | |
|
| |||||||||
| Ancillary parameter | |||||||||
| Ln sigma | - | - | - | 0.46* | (0.40,0.52) | <0.001 | -0.31 | (-0.74,0.12) | 0.166 |
| Ln gamma | -0.16* | (-0.23,-0.1) | <0.001 | - | - | - | - | - | - |
| Kappa | - | - | - | - | - | - | 1.17* | (0.67,1.68) | <0.001 |
| Heterogeneity parameter | |||||||||
| Theta | 0.00008 | - | 1.000 | 0.00 | - | 1.000 | 1.37* | (0.48,3.90) | 0.006 |
| Goodness of fit criteria | |||||||||
| AIC | 2766.84 | 2802.82 | 2769.18 | ||||||
| BIC | 2772.89 | 2808.87 | 2777.53 | ||||||
1Accelerated failure time; 2Hazard ratio; 3Confidence interval; *significant at 0.05 level
DISCUSSION
In the present study, common laboratory test indicators were analyzed to predict the prognosis of CAPD patients. This is useful for improving the therapeutic intervention of patients and improving their prognosis. Also, in a more detailed analysis, dead patients were compared with others. In addition, the diagnostic factors for two groups of hypertensive and non-hypertensive patients were investigated. It should be noted that the survival of dialysis patients may not be constant over time and may increase, decrease, or even be nonmonotonic form under the influence of factors. Therefore, in the current research, parametric survival models were used (with the AFT assumption), which handle this phenomenon and are more accurate than the common COX model.
According to the findings of the present study, it was observed that in the dialysis patients under investigation, men are more susceptible to hypertension than women. Among non-hypertensive patients, women are more likely to die, but in patients with high blood pressure, there was no significant difference between women and men in terms of mortality. These findings are consistent with the results of the study by Qiu et al.[10] Also, in their studies, it has been shown that factors such as BMI, gender, serum albumin, sodium, and urine volume may also be related to changes in blood pressure in peritoneal dialysis patients. In the current study, the effect of some of these factors along with high blood pressure on the survival of patients has been investigated, and such a relationship can affect the results. In the patients studied in our article, for the hypertensive group, factors such as K, Cr, BMI, and FBS were significantly higher than the non-hypertensive group. However the amount of albumin in this group was lower than in the group with non-hypertensive pressure. This finding, in the results of the study by Dai et al.[11] has also been confirmed.
In many studies, the effect of gender on the death of dialysis patients has been investigated, and in this context, the results of some of them are different. Some studies have not found a significant effect of the relationship between gender and death, and in some studies, women are more likely to die and in others, men are more likely to die. In the present study, the examination of dead patients revealed that, in general, the number of deaths in men is higher than in women. But no significant relationship between gender and death risk was found.
Also, in accordance with other findings of our study, the amount of Cr and albumin factors in the dead people was lower than that of the survivors, and the blood sugar level of the deceased was higher than the others. The mentioned results were in line with the results of Shi et al.’s study.[12] In their study, such results have been obtained comparing the people who died and those who survived undergoing peritoneal dialysis.
The five-year survival of the patients under investigation in the present study was estimated at 54% and the median survival time was 72.43 months. In general, to check the survival of dialysis patients, several studies have been conducted under different treatment conditions (including treatment methods or patients’ characteristics) and therefore the survival rates may be different in these studies. For example, in the study of Guzman-Ventura et al.,[13] who compared patients treated with peritoneal dialysis and hemodialysis, they concluded that the survival of patients treated with both methods was similar. The five-year survival rate of peritoneal dialysis patients was 37% and their average survival was 32.5 months. In their study for peritoneal dialysis patients, the distinction between the two automatic and continuous outpatient methods was not considered, and the survival of patients under both methods was examined together. But our study was on patients with continuous outpatient method and the survival rates are higher than the recent study. Also, in the study of Shi et al.,[12] the average survival time was less (892.36 days) than our study, which could be due to the older patients under investigation or the difference in the follow-up period.
Other findings of the present study showed that the survival time for patients with high blood pressure was shorter than for patients with normal blood pressure. In other words, death may be accelerated for dialysis patients suffering from high blood pressure. The latter result was confirmed by using both non-parametric analysis of comparison of survival rates and univariate regression analysis under AFT assumption. Although the result of multivariate analysis was not significant, it showed that controlling for other variables, the survival of patients with high blood pressure is 11% less than that of other patients. It is necessary to note that there have been limited studies on the effect of blood pressure on the survival of peritoneal dialysis patients. Among those studies is the study of Dai et al.,[11] in which peritoneal dialysis patients were examined in general, and no detailed analysis was done for the CAPD method. In this study, 24-ambulatory blood pressure has been introduced as an independent predictor of death, the increase of which leads to an increase in the risk of all-causes mortality as well as the risk of heart disease. In another study by Qiu et al.[10] on continuous ambulatory method, which specifically investigated the effect of serum sodium, the existence of an interaction between systolic blood pressure (SBP) and serum sodium was pointed out. According to the results of this study, if sodium is high (above 140 mmol/l), SBP > 130 significantly increases the risk of death in CAPD patients. The results of the mentioned studies are also in line with the results of our study, although there are differences in the details of the studies.
Another result of univariate and multivariate regression analysis of the current study is that increasing FBS increases the risk of death in the patients under study. These results were also obtained by Ventura’s study. In his study, the role of diabetes mellitus in reducing the survival of dialysis patients has been confirmed.[13] In addition, previous studies have shown that diabetes is a significant issue in dialysis patients. Chang et al.’s study[14] have investigated the effect of diabetes on the time of catheter infection and its removal. Catheter infection is one of the most common problems of peritoneal dialysis patients. In this article, it has been proven that for diabetic patients with HbA1C > 7, the probability of catheter removal due to infection is higher than others.
The findings of the present study showed that in univariate analysis, the increase in Albumin and Cr delays the occurrence of death. This result was also confirmed for albumin in multivariate analysis. This finding was also observed in Shi et al.’s study.[12] In their study, the authors showed that low albumin levels increase the risk of death in peritoneal dialysis patients. Other studies showed that serum albumin level in peritoneal dialysis patients is lower in elderly people than in young people, and we know that the risk of death in elderly people is higher than in young people. In our study, the average age of the subjects was 51 years. Regarding Cr in Shi’s study, increasing Cr significantly decreased the risk of death. In this study, in line with the results of our study, no significant effect was found for age and gender. In another study to investigate the survival of peritoneal dialysis patients, the effect of different levels of serum creatinine has been studied. The authors showed that the increase of this factor above 10 mg/dl leads to a decrease in the risk of death. Also, this effect is more prominent for patients who have been undergoing dialysis for more than 12 months compared to those with less dialysis time.[15]
One of the strengths of this study was the use of frailty models. Frailty models, by including a random component for uncontrollable factors, lead to the completion of response variability and thus help to increase the accuracy of parameter estimation. In this study, the significance of the frailty parameter confirmed the existence of inter-individual differences (due to uncontrollable factors). In addition, in the comparison of the three parametric models used in this research, Weibel’s model was confirmed, which can be a confirmation of the monotonic and non-constant shape of the hazard function of CAPD patients. Also, considering the assumption of AFT instead of PH also helped to interpret the hazard ratio more easily through the understanding of the survival rate instead of the hazard.
The limitation of the present study was not investigating the simultaneous effects of drugs taken by patients because, as mentioned in the results section, a high percentage of patients had comorbidities and were taking drugs. If drug information is available and their effects are considered, more complete results could be obtained.
CONCLUSION
In CAPD patients, the risk of death is not constant over time, and some factors such as age, blood sugar, and blood pressure lead to an increase in this risk. The probability of death in patients with high blood pressure is higher than in patients with normal blood pressure. It seems necessary to pay more attention to factors such as K, Cr, BMI, and FBS, especially in dialysis patients with high blood pressure.
Ethics approval and consent to participate
Ethical approval was obtained from the Research Ethics Committee of the Faculty of Public Health, Tehran University of Medical Sciences (Ethical approval number: IR.TUMS.SPH.REC.1401.250).
Conflicts of interest
There are no conflicts of interest.
Acknowledgments
The authors are grateful to the staff of the peritoneal dialysis data registry project.
Funding Statement
This work was supported by Tehran University of Medical Science (TUMS).
REFERENCES
- 1.Hill NR, Fatoba ST, Oke JL, Hirst JA, O’Callaghan CA, Lasserson DS, et al. Global prevalence of chronic kidney disease–A systematic review and meta-analysis. PLoS One. 2016;11:e0158765. doi: 10.1371/journal.pone.0158765. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bouya S, Balouchi A, Rafiemanesh H, Hesaraki M. Prevalence of chronic kidney disease in Iranian general population: a meta‐analysis and systematic review. Ther Apher Dial. 2018;22:594–599. doi: 10.1111/1744-9987.12716. [DOI] [PubMed] [Google Scholar]
- 3.Liu S, Tang X, Li X, Cao F, Luo J, Zou F. Efficacy of continuous ambulatory peritoneal dialysis combined with hemodialysis versus single hemodialysis in patients with end-stage renal disease. Am J Transl Res. 2021;13:10485. [PMC free article] [PubMed] [Google Scholar]
- 4.Murthy S, Rajwar E, Johnson R, Shrivastava K, Kamath P, Rahman R, et al. Validation of blood pressure devices as per 2020 World Health Organization technical specifications: A scoping review of global literature. Hypertension. 2023;80:1110–6. doi: 10.1161/HYPERTENSIONAHA.122.20425. [DOI] [PubMed] [Google Scholar]
- 5.Benítez-Parejo N, del Águila MR, Pérez-Vicente S. Survival analysis and Cox regression. Allergol Immunopathol. 2011;39:362–73. doi: 10.1016/j.aller.2011.07.007. [DOI] [PubMed] [Google Scholar]
- 6.Austin PC. A tutorial on multilevel survival analysis: Methods, models and applications. Int Stat Rev. 2017;85:185–203. doi: 10.1111/insr.12214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Najafi I, Alatab S, Atabak S, Majelan NN, Sanadgol H, Makhdoomi K, et al. Seventeen years’ experience of peritoneal dialysis in Iran: First official report of the Iranian peritoneal dialysis registry. Perit Dial Int. 2014;34:636–42. doi: 10.3747/pdi.2012.00054. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Whelton PK, Carey RM, Aronow WS, Casey DE, Collins KJ, Dennison Himmelfarb C, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: A report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. J Am Coll Cardiol. 2018;71:e127–248. doi: 10.1016/j.jacc.2017.11.006. [DOI] [PubMed] [Google Scholar]
- 9.Liu X-R, Pawitan Y, Clements M. Parametric and penalized generalized survival models. Stat Methods Med Res. 2018;27:1531–46. doi: 10.1177/0962280216664760. [DOI] [PubMed] [Google Scholar]
- 10.Qiu Y, Ye H, Fan L, Zheng X, Li W, Huang F, et al. Serum sodium modifies the association of systolic blood pressure with mortality in peritoneal dialysis patients. Kidney Blood Press Res. 2020;45:916–25. doi: 10.1159/000510478. [DOI] [PubMed] [Google Scholar]
- 11.Dai S, Chen Y, Shang D, Ge X, Xie Q, Hao C-M, et al. Association of ambulatory blood pressure with all-cause mortality and cardiovascular outcomes in peritoneal dialysis patients. Kidney Blood Press Res. 2020;45:890–9. doi: 10.1159/000510298. [DOI] [PubMed] [Google Scholar]
- 12.Shi Y, Cai J, Shi C, Liu C, Zhou J, Li Z. Low serum albumin is associated with poor prognosis in patients receiving peritoneal dialysis treatment. J Healthc Eng. 2022;2022:7660806. doi: 10.1155/2022/7660806. [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
- 13.Guzman-Ventura W, Caballero-Alvarado J. Survival of patients on chronic hemodialysis versus chronic peritoneal dialysis. Rev Peru Med Exp Salud Publica. 2022;39:161–9. doi: 10.17843/rpmesp.2022.392.10853. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Chang H-H, Chang C-H, Hsiao C-Y, Kao S-Y, Chen J-Y, Chen T-H, et al. Diabetes is the most critical risk factor of adverse complications after peritoneal dialysis catheter placement. Front Med. 2021;8:719345. doi: 10.3389/fmed.2021.719345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Park J, Mehrotra R, Rhee CM, Molnar MZ, Lukowsky LR, Patel SS, et al. Serum creatinine level, a surrogate of muscle mass, predicts mortality in peritoneal dialysis patients. Nephrol Dial Transplant. 2013;28:2146–55. doi: 10.1093/ndt/gft213. [DOI] [PMC free article] [PubMed] [Google Scholar]