Non-adherence to Haemodialysis, Interdialytic weight gain and cardiovascular mortality: a cohort study

A prospective cohort study was conducted that included patients on HD in a non-hospital service in Salvador, the northeastern region of Brazil. The sample consisted of adults without cognitive or psychiatric impairment who had been on HD for at least 3 months, to avoid inclusion of patients with acute renal injury and patients without knowledge of the nutritional guidelines, and still in initial clinical stabilization in the RRT. Of the 269 patients undergoing haemodialysis at the clinic, 8 were excluded due to the impossibility of obtaining consent because of physical limitations and the absence of authorized representatives, and 6 refused to participate in the study. Between November 2011 and January 2012, 255 patients joined the cohort and were followed until November 2017, at which point they were continuing with the HD in the service, the outcome had transpired or they were lost to follow-up. All had been given prescriptions for three sessions of haemodialysis per week with a duration of 4 h per session, using dialyzers with biocompatible membranes and variable dialysate sodium of 136 to 142 mEq/L. The initial data collection was part of the master’s project of the author started in 2011 and the baseline data were collected in the same month for all patients [9, 17]. The study was in compliance with the Helsinki declaration, and the protocol was approved by the Research Ethics Committee of the Spanish Hospital and Bahia School of Medicine and Public Health.

Sociodemographic data of interest (age, race, sex, civil status, schooling level), RRT data (period of dialysis, vascular access, residual renal function, history of parathyroidectomy and kidney transplant), clinical data, aetiology of CKD and diagnosis of diabetes mellitus (DM), systemic arterial hypertension (SAH) and cardiovascular disease or cerebrovascular were collected at the beginning of the study. The diagnoses of coronary arterial disease (history of acute myocardial infarction, myocardial revascularization or coronary angioplasty), congestive cardiac failure with low ejection fraction, cerebrovascular disease (history of transient ischaemic stroke or cerebrovascular ischaemic or haemorrhagic accident—CVA) and peripheral arterial occlusive disease (history of non-traumatic amputation of extremities or symptomatic ischaemic disease confirmed by Doppler or arteriography) were grouped and considered to represent clinical evidence of cardiovascular or cerebrovascular diseases for analysis in the study.

The primary outcome of interest was all-cause mortality being cardiovascular mortality considered as a secondary outcome. IDWG, hyperphosphatemia and regular attendance at treatment were assessed as parameters of NA to HD. IDWG was calculated as the difference between the pre-HD weight and the weight registered after the previous session; the average of the sessions in a month were registered and assessed as absolute IDWG and relative IDWG (absolute IDWG divided by dry weight [DW]). IDWG as an absolute value was analysed as a continuous variable. Relative IDWG was classified into three groups: < 3%, 3 to 3.99% and ≥ 4% of DW. IDWG > 4% of DW was the range considered to be indicative of excessive IDWG and corresponded to the threshold value used in a prior study of Mi Jung Lee [15]. DW was determined and adjusted by the assistant nephrologist in the context of the clinical condition, intradialytic and post-HD symptoms. The average phosphorus (P) level for 3 consecutive months was calculated and considered as hyperphosphatemia when greater than 5.5 mg/dL. For the analysis, reductions in regular attendance at treatment and absences from dialysis sessions over 3 consecutive months from the beginning of the study were computed, and the relationship between the period of HD completed and the period prescribed was also computed. NA was considered when the patient did not perform 100% of the prescribed HD period, independent of the reasons for absence from or reductions in the sessions.

The UF rate was considered to be the volume removed during the HD session divided by the DW and the duration of the dialysis; 12 consecutive sessions were assessed, and the average rate of UF was calculated. All of the samples were collected before the first HD session of the week for the laboratory dosages of interest and after dialysis to calculate the fractional clearance of the urea (spKt/V) using the Daurgirdas 2nd generation formula [18]. Missing laboratory results were repeated and we had no lack of information from the NA variables to haemodialysis. The urinary volume for the 24 h was quantified during the longest interdialytic period for evaluation of the residual renal function (RRF) and classified as anuria when diuresis was < 100 ml per day.

Patient deaths during follow-up were recorded according to death certificates or hospital records. These included deaths that occurred on or outside haemodialysis. If the reason for death was not clarified, it was recorded as death of undetermined cause. Cardiovascular mortality was defined as sudden cardiac death, death from acute myocardial infarction, heart failure, acute lung edema, arrhythmias, stroke or other fatal ischemic events.

To calculate the sample size for our study we considered a 24% reduction in the relative risk of overall mortality related to improvement in the proportion of adherence time for haemodialysis, which was identified in the Kimmel study, with 5% accuracy and 80% power [8]. Sociodemographic, clinical and laboratory characteristics at the beginning of the study were described as absolute and relative frequencies (percentages) when qualitative and as the mean ± standard deviation (SD) when continuous variables with a normal distribution or median and interquartile range (IQR) for a non-normal distribution. For these analyses, we had no missing covariate values. The prevalence of NA was expressed in percentages with the respective confidence intervals of 95% (CI 95%). The groups were compared using the chi-square test, t-test for independent samples or analysis of variance (ANOVA) with the post hoc Bonferroni test.

The survival time of each patient was considered as the interval between the beginning of the study and death or the end of the observation period. The survival univariate curve was researched using the Kaplan-Meier analysis with the log-rank test and the Cox proportional hazard model to evaluate mortality predictors. The variables with biological plausibility and p <  0.20 in the univariate analysis were included in the multivariate logistic regression model, and the results were expressed as the hazard ratio (HR) with the respective 95% CI. The existence of colinearity between the variables that were included in the multivariate regression model was investigated. The level of significance was set at two-tailed p value of < 0.05. All analyses were performed using SPSS version 20.0 (SPSS Inc., Chicago, Ill., USA).

The sociodemographic and clinical characteristics of the patients at the beginning of the cohort are provided in Table 1. The mean age was 50 ± 13.1 years, and the proportion of men was higher (62.7%) than that of women; other characteristics included non-white (85.5%), married (62%) and on haemodialysis with arteriovenous fistula (81.7%). The median period of haemodialysis treatment was 39 months (17–76 months). SAH was the most frequent aetiology of nephropathy (28.6% of the cases), followed by glomerulonephritis (16.5%) and DM (14.5%). A total of 39.2% of the patients were smokers, and of these, 91% abandoned the habit after beginning the RRT. A total of 77.3% of the patients underwent their treatment through the public health service (PHS). Table 2 demonstrates the baseline characteristics of the patients in accordance with the assessed NA parameters. The patients with IDWG ≥4% of DW underwent HD for longer periods, were younger and had higher Kt/V than patients with IDWG < 3% of DW, as well as a higher prevalence of anuria and SAH, and the highest average of phosphate than the other groups. Of the patients with excessive IDWG, 5.2% had a catheter with vascular access for HD versus 14.7% of patients with IDWG < 3% of DW; nevertheless, this difference was not significant (p = 0.05). Sex, race, body mass index (BMI), diagnosis of cardiovascular or cerebrovascular disease and DM, haemoglobin and albumin results did not differ among the three groups. The average UF rate was significantly different among the three investigated relative IDWG ranges, with an average of 12.5 ± 2.95 ml/h/kg in the group with excessive IDWG versus 6.18 ± 2.60 in the group with IDWG < 3% of DW. Those patients with phosphorus levels > 5.5 mg/dL were undergoing dialysis for a longer period, had lower average spKt/V, higher average BMI, higher frequency of anuria and of parathyroid hormone (PTH) ≥ 600 pg/mL and a higher UF average in HD than patients with phosphorus levels ≤5.5 mg/dL. Patients undergoing dialysis for less time than the amount prescribed per month were predominantly non-diabetic, non-white and had a higher frequency of PTH ≥ 600 pg/mL.

There were 87 deaths during 347,636 person-years of follow-up, with a mortality rate of 9.1 per 100 people-years. The average follow-up duration was 1493 days (535–2191), and 54% of the deaths occurred due to cardiovascular causes, of which 14.9% were sudden deaths, 13.8% were acute myocardial infarctions, and 13.8% were due to stroke; the remaining deaths were due to acute oedema of the lung, arrhythmias or mesenteric ischaemia.

Figure 1 demonstrates the survival curve for cardiovascular mortality by relative IDWG range. Cardiovascular mortality was significantly higher in the group of patients with IDWG ≥4% of DW (p = 0.041) than in patients with IDWG < 3% of DW. The same analysis was performed to assess all-cause mortality, but no significant difference (p = 0.15) was observed in the univariate analysis (the Kaplan-Meier graph is attached as Additional file 1: Figure S1).

Table 3 demonstrates the Cox regression analysis to identify the independent predictors of all-cause mortality; IDWG ≥4% of DW was associated with an increase in the risk of mortality of 102% compared with IDWG under 3% of DW. Age, clinically evident cardiovascular or cerebrovascular diseases and malnutrition were independent predictors for the outcome. The remaining parameters of NA assessed were not identified as independent predictors of all-cause mortality.

IDWG ≥4% of DW presented borderline result in multivariate analysis for cardiovascular mortality with hazard ratios of 2.09 (CI 95% 1.01–4.35, p = 0.047). Clinical predictors for cardiovascular mortality were malnutrition, dialysis using a catheter, DM and a clinically evident background of cardiovascular or cerebrovascular diseases (Table 4). IDWG analysed as an absolute value was not associated with mortality, and there was no colinearity between the variables that were included in the multiple regression model for all-cause mortality and cardiovascular mortality.

Table 5 demonstrates the Cox regression analysis for all-cause mortality in the subgroup of 97 patients with IDWG ≥4% of DW. It was found that malnutrition represented a significant increase in the risk of death (306%) in relation to the eutrophic state. Age and diagnosis of cardiovascular and cerebrovascular diseases were also independent predictors of mortality, with HR values of 1.05 and 2.74, respectively.