Deterioration of nutritional status has been one of the reasons for dialysis initiation. Certainly, most of guidelines (Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group 2013; National Kidney Foundation (NKF-DOQI) 1997; Watanabe et al. 2015; National Kidney Foundation 2015; European Best Practice Guidelines Expert Group on Hemodialysis ERA 2002; Tattersall et al. 2011; The CARI guidelines 2005a; Warwick et al. 2014; Churchill et al. 1999) in their statements recommended dialysis initiation when the patients experience the deterioration of nutritional status that can be attributable to uremia . However, the actual descriptions about malnutrition to make nephrologists consider dialysis initiation are diverse. The early guidelines , NKF KDOQI 1997 and CSN 1999, recommended normalized protein equivalent of nitrogen appearance (nPNA) to use and to initiate dialysis if nPNA falls below 0.8 g/kg/day spontaneously (National Kidney Foundation (NKF-DOQI) 1997; Churchill et al. 1999). CNS 1999 guideline mentioned subjective global assessment as an index of malnutrition (Churchill et al. 1999). The latest NKF KDOQI 2015 guideline mentioned protein-energy wasting as one of the signs to be monitored closely (National Kidney Foundation 2015). On the other hand, other guidelines did not specifically tell about the indices to monitor in their statements (Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group 2013; Watanabe et al. 2015; European Best Practice Guidelines Expert Group on Hemodialysis ERA 2002; Tattersall et al. 2011; The CARI guidelines 2005a; Warwick et al. 2014). Although the guidelines told that malnutrition is one of the signs to initiate dialysis, the detailed criteria for dialysis initiation remain to be investigated. Therefore, the nutritional status of CKD patients should be assessed globally from such indices as SGA, lean body mass, serum albumin, PNA, or other indices (Kalantar-Zadeh et al. 2001).

Nonetheless, the proportion of the patients who have PEW are quite high, and it can be related to subsequent worse survival (de Mutsert et al. 2008). Therefore, the nutritional management on the advanced CKD, especially those preventing PEW, is required to attain better clinical outcomes. Once the patients develop decline in nutritional status that is resistant for dietary therapy, we should start dialysis therapy properly.

Obviously, the compelling reasons for dialysis initiations such as overhydration or congestive heart failure independent of patients’ GFR worsen the prognoses afterward. Several studies have elucidated this point. A study from France investigated the association of eGFR at the initiation of dialysis with subsequent survival. The higher mortality was observed among those with higher eGFR by crude model (HR 1.40, 95%CI 1.36–1.45). However, the association was attenuated (HR 1.08, 95%CI 1.04–1.12) by adjustment for comorbidities, mobility, and nutritional status as well as age and gender. The fact indicated that age or comorbidities affected the association of higher eGFR and mortality partly but not entirely (Lassalle et al. 2010). Moreover, the frailty among CKD population comes to draw attentions. Among the incident dialysis population, the proportion of the patients with frailty was reportedly as high as 73%. The frail population significantly related to higher eGFR at dialysis initiation and poorer survival (Bao et al. 2012). These evidences remind us that the patients who started dialysis early might be forced to start early by compelling reasons due to comorbidities. To elucidate this association, Rosansky et al. investigate the effects of eGFR at initiation only among the “healthiest” population with serum albumin ≥3.5 g/dl. But they found that the association of higher eGFR and worse outcome was not changed even among such healthiest population (Rosansky et al. 2011). Another study on older population was also performed on the United States Renal Data System (USRDS) database. They investigated the association of eGFR values at initiation and subsequent survival for up to 3 years. The results indicated that higher eGFR (≥ 10 mL/min/1.73 m²) was associated with poorer prognosis even after rigorous adjustment for patients’ health status and it was consistent across subgroups (Crews et al. 2014). Therefore, comorbidities can affect the outcome but not entirely.

Recently, the rate of decline of renal function reportedly associates with the incidence of ESRD among pre-dialysis CKD population (Kovesdy et al. 2016). The rate of decline in eGFR before dialysis initiation has been reported to relate with the survival after start of dialysis therapy. O’Hare et al. investigated the trajectories of eGFR decline and subsequent clinical outcome among incident dialysis patients. They reported that steeper eGFR decline was associated with higher mortalities during the first year of dialysis and higher probabilities of hospitalization or diagnoses of AKI (O’Hare et al. 2012). Similarly, several studies demonstrated that abrupt (Hsu et al. 2016) or even faster (Browne et al. 2014; Ramspek et al. 2016) decline of eGFR was associated with higher mortality during short (Hsu et al. 2016) or longer periods (Browne et al. 2014; Ramspek et al. 2016). Interestingly, a study from NECOSAD demonstrated such relationship could only be observed with mGFR but not with eGFR (Ramspek et al. 2016). CKD is one of the major risk factors for AKI requiring dialysis (Hsu et al. 2008). Thus we can imagine that such abrupt or steeper renal function decline is related to the acute on chronic renal failure. We should pay special attentions for patients who experience faster decline of renal function to prevent vicious cycles worsening clinical outcomes.

Clinical practice patterns of facilities might be associated with the timing of initiation. Margaret et al. investigated eGFR values for veterans who initiated within versus outside the Veteran Affairs (VA) medical centers. They found that the patients less likely started dialysis at eGFR ≥10.5 mL/min/1.73 m² within the VA medical centers compared to outside of the VA facilities (Yu et al. 2015). This observation was confirmed by another study which investigated the average eGFR at initiation of dialysis within 804 health service areas in the United States. The study found that there was wide variety in mean eGFR values and only 11% of variation was explained by the patient characteristics (Scialla et al. 2014). On the other hand, a Canadian study investigates the proportion of the patients with eGFR of 10.5 mL/min/1.73 m² or more at the initiation of dialysis across the regions. They found that a large heterogeneity existed for the proportion by regions investigated. However, only 3.1% of variabilities could be attributed to the facility, while remaining 96.9% was attributed to patient factors (Sood et al. 2014).

Many researchers have developed clinical scores that predict mortality after initiation of dialysis . For this purpose, Charlson’s comorbidity index (CCI) (Charlson et al. 1987) has historically been used. However, this index was developed for the patients on admission to predict 1-year survival and was not developed for the incident dialysis population. Park et al. developed modified CCI from the Korean incident dialysis population of 24,738 patients, and they found that the index had improved predictive power of 6-month, 1-year, and 2-year survival compared to the original CCI (Park et al. 2015). Doi et al. also developed an equation to predict 1-year mortality among incident dialysis population. They found that eGFR, albumin , calcium , modified CCI, performance status, and ESA use were associated with the survival (Doi et al. 2015). These indices clearly indicate that we should pay attentions not only to the patients’ laboratory data or clinical symptoms but also the comorbidities of the patients. Thereby, they provide the opportunity to improve the outcomes of the incident patients.

From the evidences above described, many guidelines have been published regarding the timing of initiation of dialysis (Table 1.3).