ORIGINAL RESEARCH

Association of Nutritional Risk Index With

Infection-Related Hospitalization and Death
After Hospitalization in Patients
Undergoing Maintenance Hemodialysis
Katsuhito Mori, MD, PhD,* Yosuke Yamamoto, MD, PhD,† Norio Hanafusa, MD, PhD,‡
Suguru Yamamoto, MD, PhD,§ Shingo Fukuma, MD, PhD,k,{ Yoshihiro Onishi, PhD,**
Masanori Emoto, MD, PhD,*,†† and Masaaki Inaba, MD, PhD‡‡

Objective: Patients undergoing dialysis frequently experience hospitalization due to cardiovascular disease (CVD) and infection. This
population is also at high risk of rehospitalization and subsequent death. In addition to serious outcomes, hospitalization incurs substan-
tial medical cost. Prevention of hospitalization is accordingly an urgent matter. Here, we examined whether nutritional disorder was
associated with hospitalization and subsequent death.
Methods: The study was conducted under a prospective design using data from the Japanese Dialysis Outcomes and Practice
Pattern Study. The exposure was the Nutritional Risk Index for Japanese Hemodialysis (NRI-JH), through which patients were divided
into low-, medium-, and high-risk groups, with the low-risk group as referent. The primary outcome was CVD-related or infection-related
hospitalization. Secondary outcome was all-cause mortality. For exploratory analyses, the associations of baseline or latest NRI-JH just
before hospitalization, with death after hospitalizations, were examined.
Results: Of 4021 patients, 566 patients had CVD-related hospitalization and 375 had infection-related hospitalization during a median
follow-up of 2.6 years. NRI-JH at baseline was significantly associated with infection-related hospitalization but not with CVD-related
hospitalization, in multivariable Cox models (hazard ratio [HR] 1.46, 95% confidential interval [CI]: 1.09 to 1.97, P 5 .012 for medium-
risk vs. low-risk group) (HR 2.46, 95% CI: 1.81 to 3.35, P , .001 for high-risk vs. low-risk group). NRI-JH was also associated with
all-cause mortality. In addition, the baseline and latest high-risk NRI-JH groups were significantly associated with death after both
CVD-related and infection-related hospitalizations.
Conclusions: A higher nutritional risk as evaluated by NRI-JH was associated with infection-related hospitalization but not with CVD-
related hospitalization. However, NRI-JH was significantly associated with death after both CVD-related and infection-related hospital-
izations, suggesting that nutritional risk may be separately involved in hospitalization or subsequent death. NRI-JH may be useful in the
planning of individual care to improve outcomes.
Keywords: cardiovascular disease; hemodialysis; hospitalization; infection; NRI-JH
Ó 2024 The Authors. Published by Elsevier Inc. on behalf of the National Kidney Foundation, Inc. This is an open access article under the
CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

*
Department of Nephrology, Osaka Metropolitan University Graduate School Financial Disclosures: Yosuke Yamamoto has received personal fees from Sun
of Medicine, Osaka, Japan. Pharmaceutical Industries Ltd., Asahi Kasei Pharma Corporation, Toray Med-
†
Department of Healthcare Epidemiology, School of Public Health in the ical Co., Ltd., Nippon Shinyaku Co., Ltd., and Ono Pharmaceutical Co., Ltd.,
Graduate School of Medicine, Kyoto University, Kyoto, Japan. outside the submitted work. Suguru Yamamoto has received research funding from
‡
Department of Blood Purification, Tokyo Women’s Medical University, To- Toray Medical Co., Ltd and Kaneka Medix Co., Ltd., and honoraria from
kyo, Japan. Kyowa Kirin Co.
§
Division of Clinical Nephrology and Rheumatology, Niigata University Data were obtained from the Japanese Dialysis Outcomes and Practice Pattern
Graduate School of Medical and Dental Sciences, Niigata, Japan. Study phases 4, 5, and 6, the protocols of which were approved by the Ethics
k
Human Health Science, Graduate School of Medicine, Kyoto University, Committee of Tokyo Women’s Medical University (Approval Numbers 1527,
Kyoto, Japan. 2388 and 2388-R4, respectively).
{
Department of Epidemiology Infectious Disease Control and Prevention, Address correspondence to Katsuhito Mori, MD, PhD, Nephrology, Osaka
Hiroshima University Graduate School of Biomedical and Health Sciences, Hir- Metropolitan University Graduate School of Medicine, 1-4-3, Asahi-machi,
oshima, Japan. Abeno-ku, Osaka, Japan, 545-8585. E-mail: ktmori@omu.ac.jp
**
Institute for Health Outcomes and Process Evaluation research (iHope Inter- Ó 2024 The Authors. Published by Elsevier Inc. on behalf of the National
national), Kyoto, Japan. Kidney Foundation, Inc. This is an open access article under the CC BY-NC-
††
Department of Metabolism, Endocrinology, and Molecular Medicine, Osaka ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Metropolitan University Graduate School of Medicine, Osaka, Japan. 1051-2276
‡‡
Kidney Center, Ohno Memorial Hospital, Osaka, Japan. https://doi.org/10.1053/j.jrn.2024.07.017
Support: This manuscript was directly supported by the Kidney Foundation,
Japan. Global support for the ongoing DOPPS Program is provided without re-
striction on publication by a variety of funders.

Journal of Renal Nutrition, Vol 35, No 1 (January), 2025: pp 187-195 187

188 MORI ET AL

Introduction DOPPS (J-DOPPS). For the present study, the eligibility

criteria were maintenance hemodialysis for $90 days, age
H OSPITALIZATION IS MUCH more frequent in
patients undergoing dialysis than in the general pop-
ulation.1-3 The leading causes of hospitalization in these
.18 years, and available data on NRI-JH.13

patients are cardiovascular disease (CVD) and infection.4,5 Exposure and Outcomes
A recent report showed that hospitalization rates in patients The exposure of interest was the NRI-JH at baseline in
undergoing dialysis have increased over the last 8 years.6 the main and subanalyses. The NRI-JH at baseline and at
Risk of rehospitalization with subsequent mortality is also the closest time before hospitalization, evaluated within
very high in this population.3,4,7-9 In addition to serious 4 months before hospitalization, was also used for explor-
outcomes, hospitalization places a huge burden on health atory analyses. The NRI-JH was calculated as previously
expenditures.10 Therefore, prevention of hospitalization reported (Appendix 1).
in patients undergoing dialysis is critical to both improving The primary outcome was CVD-related or infection-
poor outcomes and reducing medical costs. To date, how- related hospitalization, including death. The secondary
ever, no implementable countermeasures in these patients outcome was all-cause mortality. In exploratory analyses,
have yet been taken. the outcome was all-cause mortality after CVD-related or
Nutritional disorders in patients undergoing dialysis are infection-related hospitalization. The definitions of
characterized by wasting (loss of muscle and fat mass), CVD-related and infection-related hospitalization are
which is known as protein energy wasting (PEW).11 Since shown in Appendix 2.
PEW is related to adverse outcomes,12 nutritional evalua-
tion of this condition and appropriate measures are critical. Statistical Analysis
However, a direct adaptation of PEW criteria for Asian pa- According to the original scoring system, patients were
tients, including Japanese, is problematic considering the categorized into low-risk, medium-risk, and high-risk
differences in ethnicity, lifestyle, and physique. In response, groups. Baseline characteristics are shown as means (stan-
the Nutritional Risk Index for Japanese Hemodialysis dard deviation) for continuous variables with a normal dis-
(NRI-JH) was developed based on 1-year mortality using tribution or medians (interquartile range [IQR]) for those
nationwide registry data.13 NRI-JH, which is calculated with skewed data. Categorical variables are presented as
based on body mass index (BMI), serum levels of albumin, proportions.
total cholesterol, and creatinine, is an objective screening First, to examine the relationship between NRI-JH risk
tool that is both reproducible and easy to measure. A recent groups at baseline and the main outcomes (CVD-related
study showed that higher NRI-JH, which indicates higher and infection-related hospitalization), Cox proportional
nutritional risk, was related to higher long-term mortality hazards models were employed to estimate unadjusted
in 3046 patients undergoing hemodialysis.14 Another study and multivariable adjusted hazard ratios (HRs) with 95%
reported that the high-risk group under NRI-JH was asso- confidence intervals (CIs). In the multivariable analyses,
ciated with poor survival in 133 hemodialysis patients dur- the models were adjusted with either of two sets of possible
ing long-term care.15 confounders as below: a) minimally adjusted models
To date, however, the association of NRI-JH with hos- included age, gender, dialysis vintage, presence of diabetes,
pitalization and subsequent outcomes has not been exam- and history of CVD; and b) fully adjusted models included
ined. In this study, we focused on the association of presence of hypertension, serum calcium, serum phos-
NRI-JH with CVD-related and infection-related hospital- phate, serum parathyroid hormone (categorized), serum
ization as well as all-cause mortality. Furthermore, we C-reactive protein (categorized), hemoglobin, use of
explored whether the baseline or latest NRI-JH, as evalu- vitamin D, use of calcimimetics, use of erythropoiesis stim-
ated within 4 months before hospitalization, was associated ulating agents, and use of iron preparations, in addition to
with death after hospitalization. the five factors used in the minimally adjusted models.
We also employed multivariable Fine-Gray model, with
death as the competing event, to examine the association
Materials and Methods of NRI-JH with CVD-related or infection-related hospi-
Study Design and Population talization for sensitivity analyses.
The dialysis outcomes and practice pattern study Second, Cox models were employed in a similar way to
(DOPPS) is an international prospective cohort study de- examine the relationship between NRI-JH at baseline and
signed to examine the relationship between practice pat- secondary outcome (all-cause mortality).
terns and patients outcomes with hemodialysis from a Third, exploratory analyses were performed to examine
representative sample of dialysis facilities in 21 participating the relationship between the baseline or the latest NRI-JH
countries. Detailed information regarding the protocol of and death after CVD-related or infection-related hospital-
DOPPS has been reported elsewhere.16,17 In the present ization. Cox models were employed with adjustment for a
study, participation was limited to patients in the Japanese minimal set of confounders because of the limited numbers
 NRI-JH AND INFECTION-RELATED HOSPITALIZATION 189

and events for this analysis. In the exploratory analyses, the medium-risk and high-risk groups showed a significant as-
time of hospitalization was set as day 0. sociation with infection-related hospitalization (fully
Subgroup analyses for each age category (,65 years, adjusted HR 1.46, 95% CI: 1.09 to 1.97, P 5 .012; fully
$65 years) and for the presence or absence of comorbid adjusted HR 2.46, 95% CI 1.81 to 3.35, P ,.001, respec-
diabetes were conducted to examine the association of tively) (Table 2). The results by the Fine-Gray model were
NRI-JH with the primary and secondary outcomes. Sensi- closely similar to those by the Cox model (Table S2).
tivity analyses were also performed to ascertain the robust-
Association of NRI-JH with All-Cause
ness of the results. For that purpose, NRI-JH was
Mortality
recategorized in accordance with tertiles of NRI-JH score.
To confirm the association of NRI-JH with mortality as
In this study, we defined censoring as any patient who
previously reported,13-15 we conducted additional analyses
was lost to follow-up due to transfer to another dialysis fa-
using all-cause mortality as the outcome. During the study
cility or who was still alive at the end of the study period.
period, 453 (11.3%) patients died. Figure 1C shows the
Except for exploratory analyses, the first time of evaluation
Kaplan–Meier curves for all-cause mortality, as categorized
of NRI-JH was set as day 0. All missing covariates of the
by NRI-JH risk groups. The NRI-JH showed a significant
fully adjusted models were imputed based on the multiple
association with all-cause mortality in both the crude and
imputation method using chained equations. These esti-
adjusted Cox models (Table 2).
mates from 20 imputation datasets were then combined us-
ing Rubin’s rules.18 Further, we checked the assumption of Association of NRI-JH with Death after CVD-
the proportional hazards, using log cumulative hazard plots. Related or Infection-Related Hospitalization
All analyses were performed using Stata MP version 17.0 We explored that whether the baseline or most recent
software (Stata Corp., College Station, TX, USA). P NRI-JH before hospitalization was associated with death af-
values , .05 were considered statistically significant. ter hospitalization. Of 4021 patients, 566 and 375 patients
were hospitalized due to CVD and infection, respectively.
Results Of these, 447 and 289 patients had available NRI-JH scores
Baseline Characteristics at baseline, and similarly, 429 and 278 patients had available
Of total 7033 patients in J-DOPPS 4-6, 4021 patients the most recent NRI-JH obtained within 4 months before
met the inclusion criteria (Figure S1). The mean age was hospitalization, respectively (Figure S1). Figure 2 shows the
65.4 years, 65.6% were male, and the median vintage of Kaplan–Meier curves for all-cause mortality after CVD-
dialysis was 5.0 years. Table 1 shows baseline characteristics related or infection-related hospitalization, according to
of the patients categorized by NRI-JH risk group. baseline or latest NRI-JH risk group. Baseline NRI-JH
Hemodialysis-related conditions and treatments are also showed a significant association with death after both
shown in Table S1. CVD-related and infection-related hospitalizations in the
adjusted models (Table 3). Further, latest NRI-JH also
Association of NRI-JH with CVD-Related
showed a substantial association with death after both
Hospitalization
CVD-related and infection-related hospitalizations (Table 4).
During a median follow-up of 2.6 years, 566 (14.1%) pa-
tients were hospitalized due to CVD (CVD-related hospi- Subgroup Analysis
talization). Out of 566 patients with CVD-related When patients were stratified by age (,65 years, and
hospitalizations, 109 died (Figure S1). Figure 1A shows $65 years), results of multivariable Cox regression analyses
the Kaplan–Meier curves for the incidence of CVD- for the associations of NRI-JH with primary and secondary
related hospitalization, categorized by NRI-JH risk groups. outcomes were mostly consistent with the main results
No substantial relationships were observed between NRI- (Table S3). When patients were stratified by the presence
JH and CVD-related hospitalization (Table 2). The results or absence of diabetes, results were closely similar to the
of the Fine-Gray model, with death as the competing event main results, except with regard to the significant associa-
of CVD-related hospitalization, showed a similar trend to tion between the medium-risk NRI-JH group and
those of the Cox model (Table S2). CVD-related hospitalization (Table S3).
Association of NRI-JH with Infection-Related Sensitivity Analysis
Hospitalization When the original NRI-JH risk groups were replaced in
During the follow-up period, 375 (9.3%) patients were accordance with tertiles of NRI-JH score, results of multi-
hospitalized due to infection (infection-related hospitaliza- variable Cox regression analyses for the associations of
tion). Of 375 patients with infection-related NRI-JH, with infection-related hospitalization or all-
hospitalizations, 86 died (Figure S1). Figure 1B shows the cause mortality, were consistent with the main results
Kaplan–Meier curves for the incidence of infection- (Table S4). In contrast, a substantial association was
related hospitalization, as categorized by NRI-JH risk observed between NRI-JH tertiles and CVD-related hos-
groups. Compared to the low-risk group as referent, the pitalization in the sensitivity analysis. Compared to the first
190 MORI ET AL

Table 1. Baseline Characteristics of Patients by NRI-JH Risk Group

NRI-JH
All Subjects Low-Risk Group Medium-Risk Group High-Risk Group
Items (N 5 4,021) (N 5 3,163) (N 5 537) (N 5 321)

Age (years) 65.4 (11.8) 64.6 (11.8) 66.6 (11.4) 70.8 (11.8)
Male sex 2,638 (65.6%) 2,077 (65.7%) 373 (69.5%) 188 (58.6%)
BMI (kg/m2) 21.5 (3.5) 21.8 (3.4) 21.8 (3.6) 17.7 (1.6)
Current or former smoker 1,387 (47.4%) 1,097 (47.1%) 204 (54.0%) 86 (38.7%)
Hemodialysis vintage 5.0 [1.6-11.1] 5.2 [1.9-11.3] 3.6 [0.8-9.3] 4.1 [1.1-10.8]
(years)
Comorbidities
Diabetes mellitus 1,641 (40.8%) 1,205 (38.1%) 310 (57.7%) 126 (39.3%)
Hypertension 3,330 (82.8%) 2,631 (83.2%) 445 (82.9%) 254 (79.1%)
CVD 2,428 (60.4%) 1,837 (58.1%) 372 (69.3%) 219 (68.2%)
Laboratory
Hemoglobin (g/dL) 10.7 (1.2) 10.8 (1.2) 10.4 (1.4) 10.4 (1.4)
Albumin (g/dL) 3.7 (0.4) 3.8 (0.3) 3.3 (0.4) 3.1 (0.3)
Creatinine (mg/dL) 10.4 (2.9) 11.0 (2.7) 8.3 (2.3) 7.3 (1.9)
Total cholesterol (mg/dL) 154.2 (34.8) 156.3 (32.5) 144.5 (42.2) 150.3 (38.9)
HDL cholesterol (mg/dL) 48 (16) 48 (15) 46 (17) 49 (16)
Non-HDL cholesterol 106 (33) 107 (32) 100 (39) 102 (36)
(mg/dL)
Calcium (mg/dL) 8.8 (0.8) 8.9 (0.7) 8.6 (0.7) 8.6 (0.8)
Phosphate (mg/dL) 5.3 (1.4) 5.5 (1.3) 4.9 (1.3) 4.7 (1.4)
C-reactive protein
#0.1 mg/dL 1,346 (46.1%) 1,118 (50.1%) 145 (34.5%) 83 (31.2%)
0.1., #1.0 mg/dL 1,209 (41.4%) 910 (40.8%) 189 (45.0%) 110 (41.4%)
1.0, mg/dL 364 (12.5%) 205 (9.2%) 86 (20.5%) 73 (27.4%)
Intact parathyroid
hormone
#60 pg/mL 765 (22.4%) 587 (21.8%) 105 (23.0%) 73 (27.1%)
60., #240 pg/mL 2,020 (59.1%) 1,605 (59.6%) 252 (55.1%) 163 (60.6%)
240, pg/mL 634 (18.5%) 501 (18.6%) 100 (21.9%) 33 (12.3%)
Medication
Erythropoietin 3,455 (85.9%) 2,688 (85.0%) 469 (87.3%) 298 (92.8%)
stimulating agents
Iron 1,594 (39.6%) 1,252 (39.6%) 217 (40.4%) 125 (38.9%)
Vitamin D analogs 2,761 (68.7%) 2,222 (70.2%) 346 (64.4%) 193 (60.1%)
Calcimimetics 732 (18.2%) 624 (19.7%) 65 (12.1%) 43 (13.4%)
BMI, body mass index; CVD, cardiovascular disease; HDL, high-density-lipoprotein; NRI-JH, nutritional risk index for Japanese hemodialysis.
Values are mean (standard deviation), median [interquartile range], or number (proportion).

tertile of the NRI-JH as the referent, fully adjusted HRs of involved in hospitalization or subsequent death, regardless
the second and third tertiles were 1.31 (95% CI 1.06 to of the cause of CVD-related or infection-related-
1.61, P 5 .011 for second tertile vs first tertile) and 1.25 hospitalization. NRI-JH may be useful in the planning of
(95% CI 1.01 to 1.55, P 5.041 for third tertile vs first ter- individual care to improve outcomes.
tile), respectively (Table S4). The NRI-JH was developed based on 1-year mortality in
Japanese patients undergoing hemodialysis.13 The significant
Discussion association of NRI-JH with long-term mortality was
In this analysis of 4021 patients who were receiving he- confirmed in an independent cohort of 3046 patients on he-
modialysis, we found that a higher nutritional risk, as eval- modialysis.14 In that study, NRI-JH was associated with
uated by NRI-JH, was significantly associated with CVD-related and infection-related deaths as well as all-
infection-related hospitalization but not CVD-related hos- cause death during the 4-year and 10-year follow-up pe-
pitalization. The NRI-JH also showed a significant associ- riods. Another study examined the association of NRI-JH
ation with all-cause mortality. Moreover, baseline and latest with nutritional intake in 133 in patients undergoing hemo-
NRI-JH were associated with mortality after both CVD- dialysis during long-term care.15 The NRI-JH high-risk
related and infection-related hospitalizations. These find- group was associated with low energy and protein intake.
ings suggest that nutritional disorder may be separately In addition, the NRI-JH was associated with mortality.
 NRI-JH AND INFECTION-RELATED HOSPITALIZATION 191

Figure 1. Kaplan–Meier curves showing the associations of NRI-JH groups with CVD-related hospitalization (A), infection-related
hospitalization (B), and all-cause mortality (C). Abbreviations: NRI-JH, Nutritional Risk Index for Japanese Hemodialysis; CVD,
cardiovascular disease.

However, the significant association between NRI-JH and Although an association of nutritional disorders, such as
mortality disappeared following adjustment for protein PEW with increased susceptibility to infection is plausible,
intake and C-reactive protein, suggesting the profound the precise mechanisms are largely unknown. PEW is pro-
involvement of wasting and inflammation in prognosis. In foundly associated with inflammation and hypercatabo-
this study, we show for the first time that the NRI-JH is lism, regardless of nutritional intake, possibly leading to
significantly associated with infection-related hospitalization immune deficiency. Low-grade inflammation, which re-
as well as mortality. Further, we show that nutritional risk, as flects persistent immune cell activation, dampens the
evaluated by NRI-JH, is associated with death after hospital- normal immune response in kidney disease.19,20 Persistent
ization due to both infection and CVD. inflammation disturbs the maturation of myeloid cells and
CVD and infection are known to be major causes of hos- consequent induction of myeloid-derived suppressor cells
pitalization in dialysis patients.4,5 A recent nationwide study (MDSCs), resulting in systemic immune paresis.21
from Taiwan showed that the leading cause of hospitalization Cachexia, a very severe form of PEW, is common in
was infections followed by CVD.6 An increased proportion advanced cancer. The expansion of immunosuppressive
of infection-related hospitalizations, compared to all-cause MDSCs in tumor-bearing mice was associated with
and CVD-related hospitalizations, was found in a large increased oxygen consumption, decreased adipose tissue,
cohort of Canadian patients on dialysis.10 Infection-related and increased susceptibility to fatal infection.22 A recent
hospitalization but not CVD-related hospitalization was study showed that increase in MDSCs was significantly
significantly associated with post-discharge mortality in correlated with infectious events in patients with end-
J-DOPPS phases 3 and 4.3 These recent trends suggest the stage kidney disease.23 Further studies are needed to
relative importance of preventing infection-related hospital- confirm the direct association of nutritional disorders
ization as an intervention target in dialysis patients. with immunosuppression.
192 MORI ET AL

Table 2. Association of NRI-JH Risk Group With CVD-Related or Infection-Related Hospitalization and All-Cause Mortality
NRI-JH
Outcome Index Low-Risk Group Medium-Risk Group High-Risk Group

CVD-related Incident proportion 439/3,163 (13.9%) 81/537 (15.1%) 46/321 (14.3%)

hospitalization Crude HR (95% CI) reference 1.28 (1.01 to 1.62), P 5 .04 1.25 (0.92 to 1.69), P 5 .16
or death (primary outcome) Minimally adjusted HR* reference 1.07 (0.84 to 1.36), P 5 .58 1.06 (0.78 to 1.43), P 5 .73
(95% CI)
Fully adjusted HR† (95% reference 1.17 (0.92 to 1.51), P 5 .21 1.18 (0.86 to 1.62), P 5 .30
CI)
Infection-related Incident proportion 235/3,163 (7.4%) 81/537 (11.7%) 59/321 (18.4%)
hospitalization Crude HR (95% CI) reference 1.92 (1.45 to 2.54), P , .001 3.26 (2.45 to 4.34), P , .001
or death (primary outcome) Minimally adjusted HR* reference 1.68 (1.27 to 2.23), P , .001 2.79 (2.09 to 3.73), P , .001
(95% CI)
Fully adjusted HR† (95% reference 1.46 (1.09 to 1.97), P 5 .01 2.46 (1.81 to 3.35), P , .001
CI)
All-cause mortality Incident proportion 264/3,163 (8.4%) 97/537 (18.1%) 92/321 (28.7%)
(secondary outcome) Crude HR (95% CI) reference 2.61 (2.07 to 3.30), P , .001 4.38 (3.46 to 5.56), P , .001
Minimally adjusted HR* reference 2.16 (1.71 to 2.74), P , .001 3.37 (2.64 to 4.29), P , .001
(95% CI)
Fully adjusted HR† (95% reference 2.02 (1.57 to 2.60), P , .001 3.14 (2.42 to 4.08), P , .001
CI)
CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; NRI-JH, nutritional risk index for Japanese hemodialysis.
All missing covariates of the fully adjusted models were imputed based on the multiple imputation method using chained equations.
*Minimally adjusted models were adjusted by age, gender, vintage, diabetes, and CVD.
†Fully adjusted models were adjusted by age, gender, vintage, diabetes, CVD, hypertension, serum calcium, serum phosphorus, serum para-
thyroid hormone (categorized), serum C-reactive protein(categorized), hemoglobin, vitamin D use, calcimimetics use, erythropoiesis stimulating
agents use, and iron use.

The association of NRI-JH with CVD-related hospi- related hospitalization. Why was NRI-JH associated with
talization was weaker than that with infection-related hos- death after hospitalization regardless of cause of hospitaliza-
pitalization, although NRI-JH classified by tertile was tion? One possible explanation is that, NRI-JH may reflect
significantly associated with CVD-related hospitalization a lack of resilience against unfavorable events, such as CVD
in this study. Nutritional disorders, including PEW, are a and infection. Indeed, high-risk NRI-JH groups show
nontraditional risk factor for CVD events.11,12 Such a rela- fragility or vulnerability to lethal outcomes when patients
tionship is known as malnutrition-inflammation athero- are exposed to stressful events. For example, insulin-like
sclerosis syndrome.24 Nevertheless, nutritional disorders growth factor 1 (IGF-1) is known as a representative
may be more closely involved in infectious events than anabolic hormone and has been associated with mortality26
CVD events. Indeed, NRI-JH in previous studies showed and muscle strength27 in hemodialysis patients. A lower
higher HRs for infection-related deaths than for CVD- IGF-1 was associated with death after CVD events but
related deaths.13,14 not new CVD events in hemodialysis patients,28 suggesting
While prognosis after infection-related hospitalization is that IGF-1 may be a biomarker of fragility or frailty. Nutri-
markedly poor,8,9 data concerning risk factors for mortality tional status and IGF-1 may share common pathways to
after infection-related hospitalization are limited. One fatal outcomes after the onset of stressful events.
report focused on the association of nutritional status The type of vascular access is known to be significantly
with infection-related hospitalization and subsequent death associated with mortality and hospitalization risk in patients
in 513 patients undergoing hemodialysis.25 Although the undergoing hemodialysis. A previous study reported that
geriatric nutritional risk index (GNRI), a simple nutri- patients using catheters had a 39% greater risk of death
tional screening tool, was not associated with infection- and 45% greater risk of hospitalization compared to those
related hospitalization, the GNRI showed a significant using an arteriovenous fistula (AVF).29 It has also been re-
association with death after infection-related hospitaliza- ported that vascular access types other than AVF were asso-
tion in that study.25 Our present study showed, firstly, ciated with a $20% increase in infection-related
that NRI-JH at baseline was associated with infection- hospitalization.9 In our present study, the percentages of pa-
related hospitalization based on data from a larger cohort. tients using AVF, arteriovenous graft, and catheter were
Moreover, the baseline and most recent NRI-JH were 91.7%, 7.2%, and 1.1%, respectively. These ratios are similar
significantly associated with death after infection-related to those reported in the annual dialysis report by the Japa-
hospitalization. Surprisingly, the baseline and most recent nese Society for Dialysis Therapy,30 and this result may
NRI-JH were also associated with death after CVD- therefore reflect typical dialysis conditions in Japan.
 NRI-JH AND INFECTION-RELATED HOSPITALIZATION 193

Figure 2. Kaplan–Meier curves showing the associations of baseline NRI-JH groups with death after CVD-related hospitalization
(A) and infection-related hospitalization (B); and associations of the latest NRI-JH groups with death after CVD-related hospital-
ization (C) and infection-related hospitalization (D). Abbreviations: NRI-JH, Nutritional Risk Index for Japanese Hemodialysis;
CVD, cardiovascular disease.

When Cox models were adjusted for the type of vascular shown). Accordingly, the type of vascular access might
access as an addition to fully adjusted models, respective not have significantly contributed to infection- or CVD-
HRs were almost the same as those in Table 2 (data not related hospitalization in this study.

Table 3. Association of NRI-JH Risk Group (at Baseline) With All-Cause Mortality After CVD-Related or Infection-Related
Hospitalization
NRI-JH
Outcome Index Low-Risk Group Medium-Risk Group High-Risk Group

All-cause mortality after Incident proportion 56/369 (15.2%) 11/55 (20.0%) 7/23 (30.4%)
CVD-related Crude HR (95% CI) reference 1.52 (0.80 to 2.91), 3.18 (1.44 to 7.03),
hospitalization P 5 .20 P 5 .004
Minimally adjusted HR* reference 1.37 (0.71 to 2.64), 3.19 (1.40 to 7.26),
(95% CI) P 5 .35 P 5 .006
All-cause mortality after Incident proportion 35/202 (17.3%) 14/46 (30.4%) 15/41 (36.6%)
infection-related Crude HR (95% CI) reference 1.98 (1.07 to 3.69), 2.49 (1.36 to 4.56),
hospitalization P 5 .03 P 5 .003
Minimally adjusted HR* reference 1.82 (0.97 to 3.41), 2.31 (1.25 to 4.27),
(95% CI) P 5 .06 P 5 .007
CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; NRI-JH, nutritional risk index for Japanese hemodialysis.
*Minimally adjusted models were adjusted by age, gender, vintage, diabetes, and CVD.
194 MORI ET AL

Table 4. Association of NRI-JH Risk Group (Just Before Hospitalization) With All-Cause Mortality After CVD-Related or
Infection-Related Hospitalization
NRI-JH
Outcome Index Low-Risk Group Medium-Risk Group High-Risk Group

All-cause mortality after Incident proportion 51/333 (15.3%) 12/62 (19.4%) 8/34 (23.5%)
CVD-related Crude HR (95% CI) reference 1.59 (0.85 to 2.99), P 5 .15 2.50 (1.18 to 5.31), P 5 .02
hospitalization Minimally adjusted HR* reference 1.45 (0.77 to 2.74), P 5 .25 2.51 (1.14 to 5.52), P 5 .02
(95% CI)
All-cause mortality after Incident proportion 27/171 (15.8%) 13/52 (25.0%) 19/55 (34.5%)
infection-related Crude HR (95% CI) reference 1.78 (0.92 to 3.46), P 5 .09 2.71 (1.51 to 4.88), P 5 .001
hospitalization Minimally adjusted HR* reference 1.73 (0.88 to 3.38), P 5 .11 2.57 (1.41 to 4.70), P 5 .002
(95% CI)
CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; NRI-JH, nutritional risk index for Japanese hemodialysis.
*Minimally adjusted models were adjusted by age, gender, vintage, diabetes, and CVD.

This study has several limitations. First, the definitions In conclusion, we found that NRI-JH at baseline is asso-
of CVD-related and infection-related hospitalizations ciated with infection-related hospitalization but not CVD-
were not prespecified. Thus, cause-specific hospitalization related hospitalization in hemodialysis patients. Moreover,
was at the discretion of the individual physician and facil- the baseline and latest NRI-JH just before hospitalization
ity. Second, we used only the NRI-JH as a nutritional in- were significantly associated with death after both CVD-
dex in this study. The International Society of Renal related and infection-related hospitalizations. These
Nutrition and Metabolism has referred to malnutrition findings suggest the usefulness of NRI-JH as a nutritional
inflammation score (MIS) as a potential tool for assessment indicator for infectious events in hemodialysis patients. In
of PEW.11 Although MIS is a comprehensive scoring sys- addition, they may be useful of planning of individual
tem, it requires subjective assessment by a well-trained nutritional care to improve outcomes after hospitalization.
examiner, hampering its integration into routine clinical
practice. In contrast, the GNRI is an objective screening
tool that is easily calculated based on BMI and serum albu- Practical Application
min.25 The NRI-JH, calculated using BMI, serum albu- The NRI-JH, an objective screening tool for nutritional
min, serum total cholesterol, and serum creatinine, is disorders, was associated with infection-related hospitaliza-
also considered an objective tool. Nevertheless, further tion in patients undergoing hemodialysis. Moreover, a
studies are required to confirm whether an increased num- higher nutritional risk, as evaluated by NRI-JH, was associ-
ber of nutrition-related factors can predict various out- ated with mortality after both CVD-related and infection-
comes more accurately and to compare the NRI-JH related hospitalizations. Routine and repeated measurement
with MIS, a subjective assessment of nutritional status. of the NRI-JH may be helpful in the planning of individu-
Third, the causes, types, and severities of infection were alized nutritional care in improving outcomes.
not clearly identified. Fourth, the observational nature
of this study does not allow for any determination of
causality between nutritional condition and hospitaliza- CRediT authorship contribution statement
tion. Fifth, we lack information on muscle mass obtained Katsuhito Mori: Conceptualization, Methodology,
through dual energy X-ray absorptiometry or bioelectrical Writing – original draft. Yosuke Yamamoto: Formal
impedance analysis, even though serum creatinine levels analysis, Writing – original draft. Norio Hanafusa:
may reflect skeletal muscle mass to some extent in patients Writing – review & editing. Suguru Yamamoto:
undergoing hemodialysis.31 Finally, the present study Writing – review & editing. Shingo Fukuma: Writing
examined the association of nutritional index for Japanese – review & editing. Yoshihiro Onishi: Data curation,
hemodialysis patients with clinical outcomes using Visualization. Masanori Emoto: Writing – review & ed-
independent data from a Japanese hemodialysis cohort. iting. Masaaki Inaba: Supervision, Writing – review &
Therefore, worldwide studies are required to confirm editing.
the generalizability of the results. At the same time,
development of a common nutritional risk index for use
Supplementary Data
among patients with different ethnicity, lifestyle, and
physique is expected, for example, through the use of Supplementary data related to this article can be found at
modified cutoff values. https://doi.org/10.1053/j.jrn.2024.07.017.
 NRI-JH AND INFECTION-RELATED HOSPITALIZATION 195

References 16. Young EW, Goodkin DA, Mapes DL, et al. The dialysis outcomes and
1. Plantinga LC, Jaar BG. Preventing repeat hospitalizations in dialysis pa- practice patterns study (DOPPS): an international hemodialysis study. Kidney
tients: a call for action. Kidney Int. 2009;76:249-251. Int. 2000;57:S74-S81.
2. Molnar AO, Moist L, Klarenbach S, et al. Hospitalizations in dialysis pa- 17. Pisoni RL, Gillespie BW, Dickinson DM, Chen K, Kutner MH,
tients in Canada: a national cohort study. Can J Kidney Health Dis. Wolfe RA. The dialysis outcomes and practice patterns study (DOPPS):
2018;5:2054358118780372. design, data elements, and methodology. Am J Kidney Dis. 2004;44(5 Suppl
3. Shimizu S, Fukuma S, Ikenoue T, Akizawa T, Fukuhara S. Increased 2):7-15.
mortality rate after hospitalization among chronic hemodialysis patients: a 18. Rubin DB. John Wiley & Sons; 2004.
prospective cohort study. Nephron. 2018;140:194-202. 19. Betjes MG. Immune cell dysfunction and inflammation in end-stage
4. Daratha KB, Short RA, Corbett CF, et al. Risks of subsequent hospital- renal disease. Nat Rev Nephrol. 2013;9:255-265.
ization and death in patients with kidney disease. Clin J Am Soc Nephrol. 20. Steiger S, Rossaint J, Zarbock A, Anders HJ. Secondary Immunodefi-
2012;7:409-416. ciency related to kidney disease (SIDKD)-Definition, Unmet need, and
5. Schoonover KL, Hickson LJ, Norby SM, et al. Risk factors for hospitaliza- mechanisms. J Am Soc Nephrol. 2022;33:259-278.
tion among older, incident haemodialysis patients. Nephrology. 2013;18:712-717. 21. Bronte V, Brandau S, Chen SH, et al. Recommendations for myeloid-
6. Lee CC, Hsu CC, Lin MH, Chen KH, Wu IW. Hospitalization in pa- derived suppressor cell nomenclature and characterization standards. Nat
tients with dialysis in Taiwan: a nationwide population-based observational Commun. 2016;7:12150.
study. J Formos Med Assoc. 2022;121(Suppl 1):S39-S46. 22. Cuenca AG, Cuenca AL, Winfield RD, et al. Novel role for tumor-
7. Plantinga LC, King L, Patzer RE, et al. Early hospital readmission induced expansion of myeloid-derived cells in cancer cachexia. J Immunol.
among hemodialysis patients in the United States is associated with subse- 2014;192:6111-6119.
quent mortality. Kidney Int. 2017;92:934-941. 23. Xing YF, Cai RM, Lin Q, et al. Expansion of polymorphonuclear
8. Allon M, Radeva M, Bailey J, et al. The spectrum of infection-related myeloid-derived suppressor cells in patients with end-stage renal disease
morbidity in hospitalized haemodialysis patients. Nephrol Dial Transplant. may lead to infectious complications. Kidney Int. 2017;91:1236-1242.
2005;20:1180-1186. 24. Stenvinkel P, Heimburger O, Paultre F, et al. Strong association be-
9. Dalrymple LS, Mu Y, Nguyen DV, et al. Risk factors for infection- tween malnutrition, inflammation, and atherosclerosis in chronic renal failure.
related hospitalization in in-Center hemodialysis. Clin J Am Soc Nephrol. Kidney Int. 1999;55:1899-1911.
2015;10:2170-2180. 25. Machiba Y, Mori K, Shoji T, et al. Nutritional disorder evaluated by
10. Lafrance JP, Rahme E, Iqbal S, Elftouh N, Laurin LP, Vallee M. Trends the geriatric nutritional risk index predicts death after hospitalization for
in infection-related hospital admissions and impact of length of time on dial- infection in patients undergoing maintenance hemodialysis. J Ren Nutr.
ysis among patients on long-term dialysis: a retrospective cohort study. CMAJ 2022;32:751-757.
Open. 2014;2:E109-E114. 26. Nilsson E, Carrero JJ, Heimburger O, Hellberg O, Lindholm B,
11. Fouque D, Kalantar-Zadeh K, Kopple J, et al. A proposed nomencla- Stenvinkel P. A cohort study of insulin-like growth factor 1 and mortality
ture and diagnostic criteria for protein-energy wasting in acute and chronic in haemodialysis patients. Clin Kidney J. 2016;9:148-152.
kidney disease. Kidney Int. 2008;73:391-398. 27. Delanaye P, Bataille S, Quinonez K, et al. Myostatin and insulin-like
12. Ikizler TA, Cano NJ, Franch H, et al. Prevention and treatment of pro- growth factor 1 are biomarkers of muscle strength, muscle mass, and mortality
tein energy wasting in chronic kidney disease patients: a consensus statement in patients on hemodialysis. J Ren Nutr. 2019;29:511-520.
by the International Society of Renal Nutrition and Metabolism. Kidney Int. 28. Nakaya R, Shoji T, Nagata Y, et al. Associations of serum insulin-like
2013;84:1096-1107. growth factor 1 with new cardiovascular events and subsequent death in he-
13. Kanda E, Kato A, Masakane I, Kanno Y. A new nutritional risk index modialysis patients: the DREAM cohort. J Atheroscler Thromb. 2022;29:1153-
for predicting mortality in hemodialysis patients: nationwide cohort study. 1165.
PLoS One. 2019;14:e0214524. 29. Lacson E Jr, Wang W, Hakim RM, Teng M, Lazarus JM. Associates of
14. Shimamoto S, Yamada S, Hiyamuta H, et al. Association of the nutri- mortality and hospitalization in hemodialysis: potentially actionable labora-
tional risk index for Japanese hemodialysis patients with long-term mortality: tory variables and vascular access. Am J Kidney Dis. 2009;53:79-90.
the Q-Cohort Study. Clin Exp Nephrol. 2022;26:59-67. 30. Nitta K, Masakane I, Hanafusa N, et al. Annual dialysis data report
15. Kitabayashi K, Yamamoto S, Narita I. Association of the nutritional 2017, JSDT renal data registry. Renal Replacement Therapy. 2019;5:1-44.
risk index for Japanese hemodialysis with mortality and dietary nutritional 31. Patel SS, Molnar MZ, Tayek JA, et al. Serum creatinine as a marker of
intake in patients undergoing hemodialysis during long-term hospitalization. muscle mass in chronic kidney disease: results of a cross-sectional study and
Clin Exp Nephrol. 2022;26:1200-1207. review of literature. J Cachexia Sarcopenia Muscle. 2013;4:19-29.