Abstract
Introduction
Unplanned dialysis initiation (UDI) is associated with poor outcomes and high medical costs. Although aging is a prominent risk factor for UDI, the roles of age-related factors such as cognitive impairment and physical dysfunction remain underexplored. This study aimed to clarify the associations of cognitive impairment and physical dysfunction with UDI and additional medical costs.
Methods
This study used a Japanese administrative claims database to analyze 79,850 patients aged ≥ 65 years (median age: 76 ys; 31.6% females) who began receiving dialysis. UDI was defined as starting dialysis with a temporary catheter. Physical function and cognitive impairment were classified based on mobility and daily living abilities. We assessed the association using logistic regression. Additional medical costs were estimated via generalized linear regression.
Results
UDI occurred in 16,176 patients (20%). Compared with the normal group, the odds ratios (ORs) for UDI were 1.58 (95% confidence interval [CI]: 1.49–1.67) for low physical function, 1.70 (95% CI: 1.58–1.82) for very low, and 2.22 (95% CI: 2.09–2.35) for extremely low physical function. For cognitive impairment, the ORs were 1.02 (95% CI: 0.96–1.08) for mild impairment and 1.26 (95% CI: 1.14–1.39) for severe impairment relative to normal. The average marginal cost of UDI was $7178 [95% CI: $7019–$7338] per admission. A combination of physical dysfunction and cognitive impairment further increased UDI risk and inpatient care costs.
Conclusion
Older adults with cognitive impairment and physical dysfunction face a higher risk of UDI. Early intervention for these patients may reduce UDI and its associated costs.
Keywords: aging, cognitive impairment, dialysis, end-stage kidney disease, physical dysfunction
Graphical abstract
The total number of patients with chronic kidney disease (CKD) has increased to 843.6 million, including 3.9 million cases of end-stage kidney disease (ESKD).1,2 Because CKD and ESKD are more common in older adults, dialysis, as the predominant kidney replacement therapy for the elderly, now has a substantial impact on geriatric health care.1,3 It is important to study how age-related conditions affect transitions to kidney replacement therapy in older adults because there is a significant link between aging and CKD. Notably, CKD accelerates biological aging, increasing risks of age-related issues, such as physical dysfunction (e.g., frailty and sarcopenia) and cognitive impairment.3,4 Sarcopenia, that is, age-related loss of muscle mass and physical function, is closely associated with CKD progression, with severe sarcopenia being prominent in patients with ESKD.5 CKD is also a known risk factor for cognitive impairment, which is frequently observed in patients transitioning to kidney replacement therapy.4,6
Dialysis preparation is essential for managing ESKD effectively and planning dialysis initiation.4,7,8 A smooth transition is crucial because UDI is associated with higher mortality, lower quality of life, and higher medical costs than planned initiation.4,8, 9, 10 Older age is a major factor for suboptimal preparation in patients with ESKD, with studies showing that older patients face a higher risk of UDI.11, 12, 13 These findings highlight the need to identify UDI risks in elderly individuals to clarify the underlying mechanisms. In 2015, over half of US patients starting dialysis were aged > 65 years; the mean age of initiation was 70.88 years in Japan (2020) and 64.9 years in Europe (2019).14, 15, 16 As societies age, it is increasingly important to clarify how age-related issues contribute to UDI.
Physical dysfunction and cognitive impairment are key aging-related issues strongly associated with poor health outcomes and higher medical costs, not only in the general population but also in patients with CKD.3, 4, 5, 6,9,17, 18, 19, 20, 21, 22 Physical dysfunction is one of the most critical health problems in older adults, because it correlates with mortality, CKD progression, and reduced quality of life.17,20 Cognitive impairment is another major aging-related condition, closely linked to higher mortality.3,4,18 Besides older age, previous studies have identified several major risk factors for UDI, including the cause of kidney disease, body mass index (BMI), cardiovascular disease, cancer, diabetes, low serum albumin levels, estimated glomerular filtration rate decline, and fewer nephrology visits before initiating dialysis.23 However, the association of physical dysfunction and cognitive impairment with UDI has not been fully investigated. In particular, despite the close relationship between physical dysfunction and cognitive impairment,18 a comprehensive understanding of their combined impact on UDI is lacking.
To address this knowledge gap, this study explored the association between these conditions and UDI using a nationwide, real-world database. We hypothesized that the combination of physical dysfunction and cognitive impairment constitutes unrecognized risk factors for UDI in the elderly. We aimed to identify high-risk elderly populations by comprehensively analyzing physical dysfunction and cognitive impairment. This study classified physical dysfunction and cognitive impairment based on the intensity of care required and whether they were independent. Cognitive impairment was evaluated on a spectrum, ranging from mild, where individuals experience some memory or thinking problems without requiring nursing care, to severe, where individuals need nursing care because of significant memory or thinking difficulties. Furthermore, given that UDI, physical dysfunction, and cognitive impairment markedly impact medical costs, we assessed the additional costs associated with UDI at different levels of physical dysfunction and cognitive impairment. Overall, this study offers insights to improve health care strategies for elderly patients with ESKD.
Methods
Data Sources
This observational study used the Diagnosis Procedure Combination database, an administrative claims database covering over half of all admissions in Japan, including those from > 1000 hospitals, encompassing all 82 university hospitals.24 The Diagnosis Procedure Combination database contains information on admission reasons, comorbidities, and complications during admission based on the International Classification of Disease and Related Health Problems, 10th Revision.25 It also provides patient data, such as age, sex (defined as male or female based on biological characteristics), BMI, daily life activities (including physical function at admission and discharge), cognitive status before admission, procedure types, admission status (e.g., emergency admission or ambulance transfer), readmissions (defined as admission within 4 weeks of a prior admission), discharge status, and in-hospital mortality.24,25
Study Population
We analyzed 85,973 admissions from 2014 to 2021 in the Diagnosis Procedure Combination database. The inclusion criteria were patients aged ≥ 65 years who were primarily hospitalized for dialysis initiation because of ESKD. Dialysis initiation was defined as occurring within 14 days of admission and identified using the International Classification of Disease and Related Health Problems, 10th Revision codes N18.0 or N18.5, along with specific dialysis initiation procedure codes. After excluding patients who died during admission, those who withdrew from dialysis at discharge or transfer, those who began dialysis with continuous renal replacement therapy, and second or later admissions (Supplementary Figure S1), 79,850 eligible patients were included in the analysis.
Covariate and Outcome Definitions
Age, sex, and BMI were recorded upon hospital admission. Comorbidities were identified using the International Classification of Disease and Related Health Problems, 10th Revision coding algorithms and included congestive heart failure, hypertension, myocardial infarction, peripheral arterial disease, diabetes mellitus, cerebrovascular disease, chronic pulmonary disease, and any malignancy (Supplementary Table S1).26 Smoking status was categorized as never smoked or current/past smoker. Hospitals were divided into 4 categories based on the annual number of hospitalized individuals requiring dialysis: very high volume (> 584 individuals/yr), high volume (400–583 individuals/yr), low volume (229–399 individuals/yr), and very low volume (< 229 individuals/yr). Admission status included emergency admissions, ambulance transfers, year of admission, and readmission (previous admission within 4 weeks). Emergency admission was defined as hospitalization on a nonscheduled day.
Physical function was evaluated by clinicians based on the patient's ability on the day of admission and categorized into the following 4 groups based on mobility: normal (independently walking), low (requiring walking assistance), very low (wheelchair-bound), and extremely low (bedridden). The primary doctor responsible for each patient during hospitalization evaluates such impairments by gathering information about the patient’s preadmission status from the patient, their family, and any involved social workers, using the nationwide administrative classification of cognitive function for the elderly aged ≥65 years, known as “the degree of independence in daily living for the demented elderly.”27 This system, which is widely used in studies of the elderly, classifies patients based on communication capacity, symptoms, and behaviors (Additional File 1, Supplementary Table S2).28, 29, 30, 31, 32, 33 We divided patients based on cognitive impairment as follows: none, mild (some daily living difficulties but mostly independent; levels 1–2), and severe (not independent and requiring care; levels 3–M). A validation study demonstrated a strong correlation with the Mini-Mental State Examination.34 Another validation study comparing the tool with a neuropsychiatrist’s diagnosis found that it effectively identifies cognitive impairments.35 Our chosen cut-off (level 1) for cognitive impairments follows the recommendation from this validation study.
Dialysis during hospitalization was identified based on procedure codes for chronic maintenance hemodialysis (< 4 h per session, ≥ 4 h and < 5 h per session, ≥ 5 h per session), chronic maintenance hemodiafiltration, continuous peritoneal dialysis (PD), or continuous renal replacement therapy.24 All patients were followed-up until discharge or transfer. Withdrawal from dialysis was defined as the absence of a dialysis code during the last 5 days of follow-up.
The primary outcome was UDI during hospitalization. In Japan, permanent access is typically prepared before planned dialysis initiation. The Dialysis Outcomes and Practice Patterns Study III (i.e., DOPPS III), an international study on dialysis practices, suggested that Japan is the only country that meets the goal of having < 10% of prevalent patients with a catheter.36 Furthermore, hospitalization for initiating dialysis is common practice in Japan.37 Therefore, UDI was defined as starting hemodialysis with a nontunneled temporary catheter during hospitalization.13 The secondary outcome was the additional medical costs associated with UDI, physical dysfunction, and cognitive impairment. Costs were calculated using the currency exchange rate of ¥109.01 for $1, based on the 2019 annual average.
Data Analyses
Baseline characteristics were presented as numbers (percentages) or medians with an interquartile range. Values were missing for BMI (n = 1619), smoking history (n = 8116), cognitive impairments (n = 4), physical function (n = 1601), emergency admissions (n = 12), and ambulance transfers (n =12). Thus, to handle missing data, multiple imputations were performed via the chained equation.38 In total, 20 imputed complete datasets were generated using the covariates, age, sex, BMI, comorbidities, physical function, cognitive impairment, emergency admission, ambulance transfer, readmission, hospital volume, admission year, and outcomes.39 Rubin’s rule was applied to calculate the final estimates for each model.38
Logistic regression was performed to assess the association between the outcome and declines in physical function or cognitive impairment. We presented adjusted models accounting for age, sex, BMI, comorbidities, physical function, cognitive impairment, emergency admission, ambulance transfer, readmission, hospital volume, and admission year. Predicted probabilities were calculated using marginal standardization based on the fully adjusted model.40 The average marginal effect (AME) of medical costs between planned and UDI initiation was estimated using a generalized linear regression model, incorporating interaction terms between UDI and physical and cognitive status.41 AME is defined as the mean of the partial derivatives computed for each individual in the sample. It estimates the average difference in the outcome between UDI and planned dialysis. Sensitivity analyses were conducted on the following subsets: (i) “complete case samples,” which included only cases without missing values; (ii) “restricted sample #1,” which excluded emergency admissions; and (iii) “restricted sample #2,” which excluded patients who had a fistula placed or planned but did not mature at admission. The latter group was defined as patients who did not require any vascular access or peritoneal catheter procedures during hospitalization. Statistical analyses were performed using Stata software version 18.0 MP (Stata Corp., College Station, TX), with visualizations created with Stata and Python version 3.12.6 (Python Software Foundation, DE). P-values < 0.05 were considered statistically significant.
Ethics Approval and Consent to Participate
The study was conducted in accordance with the ethical principles of the Declaration of Helsinki. This study was approved by the ethics committee of the Tokyo Medical and Dental University (approval number M2000-788). Informed consent was not required because of the anonymity of the data.
Results
Patient Characteristics
In total, 79,850 participants were included in the analysis. In Table 1, we summarize patient characteristics overall and stratified by physical function and cognitive impairment. Among all participants, 70.2%, 11.3%, 6.2%, and 10.2% had normal, low, very low, and extremely low physical function, respectively. For cognitive impairment, 85.1% had none, 12.5% had mild impairment, and 3.0% had severe impairment. The median age at admission was 76 years (interquartile range: 70–81 yrs). Patients with normal physical function and no cognitive impairment had the youngest median age (75 yrs, interquartile range: 70–80 yrs).
Table 1.
Patient characteristics
| Characteristics | Overall |
Physical function |
Cognitive impairment |
|||||
|---|---|---|---|---|---|---|---|---|
| Normal |
Low |
Very low |
Extremely low |
None |
Mild |
Severe |
||
| N = 79,850 | n = 56,091 | n = 9026 | n = 4939 | n = 8193 | n = 67,493 | n = 9947 | n = 2406 | |
| Age, yr | 76 (70–81) | 75 (70–80) | 79 (74–84) | 78 (72–83) | 79 (73–84) | 75 (70–80) | 80 (75–85) | 82 (77–86) |
| Sex (women) | 25,198 (31.6%) | 15,786 (28.1%) | 3624 (40.2%) | 1935 (39.2%) | 3290 (40.2%) | 20,354 (30.2%) | 3736 (37.6%) | 1107 (46.0%) |
| BMI | 23 (21–25) | 23 (21–25) | 22 (20–25) | 23 (20–25) | 22 (20–25) | 23 (21–25) | 22 (20–25) | 22 (19–24) |
| Smoking | 28,790 (40.1%) | 21,380 (42.3%) | 2848 (35.0%) | 1616 (36.5%) | 2476 (34.0%) | 25,042 (41.2%) | 3160 (35.8%) | 586 (27.5%) |
| Congestive heart failure | 18,690 (23.4%) | 11,370 (20.3%) | 2543 (28.2%) | 1450 (29.4%) | 2809 (34.3%) | 15,335 (22.7%) | 2701 (27.2%) | 653 (27.1%) |
| Hypertension | 55,791 (69.9%) | 41,658 (74.3%) | 5803 (64.3%) | 3046 (61.7%) | 4353 (53.1%) | 48,136 (71.3%) | 6355 (63.9%) | 1297 (53.9%) |
| Myocardial Infarction | 2254 (2.8%) | 1504 (2.7%) | 277 (3.1%) | 168 (3.4%) | 248 (3.0%) | 1929 (2.9%) | 278 (2.8%) | 47 (2.0%) |
| Peripheral arterial disease | 3741 (4.7%) | 2538 (4.5%) | 439 (4.9%) | 267 (5.4%) | 420 (5.1%) | 3146 (4.7%) | 487 (4.9%) | 108 (4.5%) |
| Diabetes mellitus | 33,768 (42.3%) | 23,654 (42.2%) | 3832 (42.5%) | 2172 (44.0%) | 3433 (41.9%) | 28,836 (42.7%) | 4054 (40.8%) | 876 (36.4%) |
| Cerebrovascular disease | 7843 (9.8%) | 4573 (8.2%) | 1154 (12.8%) | 691 (14.0%) | 1232 (15.0%) | 6129 (9.1%) | 1315 (13.2%) | 399 (16.6%) |
| Chronic pulmonary disease | 2236 (2.8%) | 1534 (2.7%) | 250 (2.8%) | 159 (3.2%) | 243 (3.0%) | 1888 (2.8%) | 284 (2.9%) | 64 (2.7%) |
| Malignancy | 9778 (12.2%) | 6760 (12.1%) | 1163 (12.9%) | 622 (12.6%) | 1036 (12.6%) | 8338 (12.4%) | 1,195 (12.0%) | 243 (10.1%) |
| Emergency admission | 25,243 (31.6%) | 12,341 (22.0%) | 3913 (43.4%) | 2421 (49.0%) | 5579 (68.1%) | 20,107 (29.8%) | 3927 (39.5%) | 1208 (50.2%) |
| Ambulance transfer | 4786 (6.0%) | 1058 (1.9%) | 606 (6.7%) | 360 (7.3%) | 2405 (29.4%) | 3517 (5.2%) | 902 (9.1%) | 367 (15.3%) |
| Readmission | 15,902 (19.9%) | 11,119 (19.8%) | 1931 (21.4%) | 1091 (22.1%) | 1506 (18.4%) | 13,356 (19.8%) | 2055 (20.7%) | 490 (20.4%) |
| Hospital volume | ||||||||
| Very low | 19,534 (24.5%) | 13,373 (23.8%) | 2355 (26.1%) | 1192 (24.1%) | 2229 (27.2%) | 15,878 (23.5%) | 2933 (29.5%) | 719 (29.9%) |
| Low | 19,843 (24.9%) | 14,072 (25.1%) | 2183 (24.2%) | 1213 (24.6%) | 2041 (24.9%) | 16,669 (24.7%) | 2592 (26.1%) | 582 (24.2%) |
| High | 20,184 (25.3%) | 14,424 (25.7%) | 2081 (23.1%) | 1232 (24.9%) | 1993 (24.3%) | 17,696 (26.2%) | 1926 (19.4%) | 562 (23.4%) |
| Very high | 20,289 (25.4%) | 14,222 (25.4%) | 2407 (26.7%) | 1302 (26.4%) | 1930 (23.6%) | 17,250 (25.6%) | 2496 (25.1%) | 543 (22.6%) |
| Admission year | ||||||||
| 2014–2015 | 17,420 (21.8%) | 11,816 (21.1%) | 2026 (22.4%) | 1181 (23.9%) | 1968 (24.0%) | 14,933 (22.1%) | 2027 (20.4%) | 460 (19.1%) |
| 2016–2017 | 19,815 (24.8%) | 13,911 (24.8%) | 2135 (23.7%) | 1243 (25.2%) | 2113 (25.8%) | 16,988 (25.2%) | 2291 (23.0%) | 536 (22.3%) |
| 2018–2019 | 20,590 (25.8%) | 14,652 (26.1%) | 2273 (25.2%) | 1218 (24.7%) | 2040 (24.9%) | 17,357 (25.7%) | 2551 (25.6%) | 678 (28.2%) |
| 2020–2021 | 22,025 (27.6%) | 15,712 (28.0%) | 2592 (28.7%) | 1297 (26.3%) | 2072 (25.3%) | 18,215 (27.0%) | 3078 (30.9%) | 732 (30.4%) |
BMI, body mass index.
Data are presented as numbers (percentage) or medians (interquartile range).
In groups with lower physical function and more severe cognitive impairment, the proportion of women was higher than that in the normal groups. In the normal physical function group, 28.1% of patients were female, with the proportions increasing to 40.2%, 39.2%, and 40.2% in the low, very low, and extremely low groups, respectively. For cognitive impairment, 30.2% of those with no impairment were female, relative to 37.6% and 46.0% in the mild and severe impairment groups, respectively. Hypertension (69.9%) was the most common comorbidity, followed by diabetes mellitus (42.2%). In this study, we identified 5586 patients who initiated maintenance PD. We observed that the proportion of PD decreased with advancing physical dysfunction and cognitive impairment (Supplementary Table S3). Furthermore, this proportion was notably lower in the UDI group.
Risk of UDI With Physical Dysfunction and Cognitive Impairment
During a median follow-up of 18 days (interquartile range: 12–30 days), 16,176 cases (20%) required UDI. In Figure 1, we present the crude and adjusted risks of UDI based on physical function and cognitive impairments. For physical function, compared with the normal group, adjusted ORs for UDI were 1.58 (95% CI: 1.49–1.67; P-value < 0.001] in the low group, 1.70 (95% CI: 1.58–1.82; P-value < 0.001) in the very low group, and 2.22 (95% CI: 2.09–2.35; P-value < 0.001) in the extremely low group (Figure 1; Supplementary Table S4). For cognitive impairment, compared with the no cognitive impairment group, adjusted ORs for UDI were 1.01 (95% CI: 0.96–1.08; P-value < 0.001) in the mild group and 1.26 (95% CI: 1.14–1.39; P-value < 0.001) in the severe group (Figure 1; Supplementary Table S4). Notably, in the mild cognitive impairment group, the crude model showed a significant difference (OR: 1.40, 95% CI: 1.34–1.48; P-value < 0.001); however, this was not observed after adjustment (OR: 1.02 [95% CI: 0.96–1.08; P-value 0.528]). Sensitivity analyses conducted on the complete case sample, restricted sample #1, and restricted sample #2 consistently demonstrated a strong association of lower physical function and severe cognitive impairment with a risk of UDI even after adjustment (Supplementary Table S4).
Figure 1.
Risk of unplanned dialysis initiation in patients with physical dysfunction and cognitive impairment. The association of physical dysfunction and cognitive impairment with unplanned dialysis initiation was estimated using logistic regression. The model for physical dysfunction was adjusted for age, sex, BMI, comorbidities, cognitive impairment, emergency admission, ambulance transfer, readmission, hospital volume, and admission year. The model for cognitive impairment was adjusted for age, sex, BMI, comorbidities, physical function, emergency admission, ambulance transfer, readmission, hospital volume, and admission year. BMI, body mass index; CI, confidence interval.
Predicted Probability of UDI Across Physical Dysfunction and Cognitive Impairment Levels
In Figure 2, we illustrate the predicted probabilities of UDI based on physical function and cognitive impairment. Patients with extremely low physical function and severe cognitive impairment showed the highest probability, nearly double those with normal physical function and no cognitive impairment (33.9%, [95% CI: 32.0–35.8] vs. 17.0%, [95% CI: 6.6–17.3], respectively; Supplementary Table S5). Declining physical function was associated with a more pronounced increase in predicted probabilities compared with worsening cognitive impairment.
Figure 2.
Predicted probabilities of unplanned dialysis initiation across physical function and cognitive impairment levels. A 3-dimensional bar graph illustrates the predicted probabilities of unplanned dialysis initiation. Marginal predicted probabilities were calculated using a logistic regression model adjusted for age, sex, BMI, comorbidities, physical function, cognitive impairment, emergency admission, ambulance transfer, readmission, hospital volume, and admission year. BMI, body mass index.
Medical Costs per Admission Based on Dialysis Initiation Status Stratified by Physical Function and Cognitive Impairment
Mean medical cost per admission for the cohort was $10,695 (¥1,165,817; standard error [SE]: $34 [¥3685]). In Table 2, we show the mean costs per admission and the unadjusted difference for UDI versus planned dialysis initiation, with higher costs observed for UDI across all groups. The mean cost for UDI was $18,176 (¥1,981,399; SE: $95 [¥10,367]) compared with $8793 (¥958,559; SE: $30 [¥3330]) for planned dialysis, resulting in a difference of $9383 (¥1,022,841; 95% CI: $9232–$9534 (¥1,006,333–¥1,039,348)]. This cost disparity persisted across subgroups stratified by physical function and cognitive impairment. As physical function and cognitive impairment worsened, the mean medical costs increased. The highest mean costs were observed in patients with extremely low physical function and severe cognitive impairment: $24,549 (¥2,676,052; SE: $1249 [¥136,127]) for UDI and $16,684 (¥1,818,716; SE: $684 [¥74,517]) for planned dialysis.
Table 2.
Unadjusted mean medical costs per admission with or without unplanned dialysis initiation, stratified across physical dysfunction and cognitive impairment levels
| Overall | UDI | Mean (SE), $ | Difference (95%CI), $ | |
| Yes | 18,176 (95) | 9383 (9232–9534) | ||
| No | 8793 (30) | |||
| Physical function | Cognitive impairment | |||
| Normal | None | Yes | 15,849 (99) | 8180 (8025–8336) |
| No | 7668 (28) | |||
| Mild | Yes | 17,432 (314) | 8697 (8046–9348) | |
| No | 8735 (139) | |||
| Severe | Yes | 19,748 (1222) | 9516 (7382–11 650) | |
| No | 10,231 (470) | |||
| Low | None | Yes | 18,645 (319) | 8319 (7767–8870) |
| No | 10,327 (123) | |||
| Mild | Yes | 19,740 (578) | 8692 (7639–9745) | |
| No | 11,048 (242) | |||
| Severe | Yes | 19,752 (949) | 7088 (4970–9206) | |
| No | 12,664 (566) | |||
| Very low | None | Yes | 18,723 (329) | 7602 (6916–8289) |
| No | 11,120 (179) | |||
| Mild | Yes | 20,719 (847) | 8840 (7194–10 486) | |
| No | 11,879 (412) | |||
| Severe | Yes | 22,183 (2198) | 9994 (6574–13 413) | |
| No | 12,189 (624) | |||
| Extremely low | None | Yes | 21,171 (244) | 7057 (6442–7673) |
| No | 14,114 (202) | |||
| Mild | Yes | 22,559 (544) | 7794 (6351–9238) | |
| No | 14,764 (480) | |||
| Severe | Yes | 24,549 (1249) | 7865 (5196–10534) | |
| No | 16,684 (684) |
BMI, body mass index; CI, confidence interval; SE, standard error; UDI, unplanned dialysis initiation.
AME of UDI on Medical Costs per Admission
To quantify the additional medical costs associated with UDI, we calculated the AME using a generalized linear regression model. Across the cohort, the AME of UDI was $7178 (¥782,520; 95% CI: $7019–$7338 [¥765,191–¥788,838]; P-value < 0.001) per admission. We also assessed the impact of physical function and cognitive impairment on the AME (Figure 3). Although extremely low physical function was associated with a lower AME than other groups, no consistent trend emerged across physical function levels. In contrast, AMEs increased with worsening cognitive impairment. In addition, we analyzed the AME of UDI on the length of hospital stay, revealing a distribution pattern similar to that observed for medical costs (Supplementary Figure S2).
Figure 3.
The average marginal effect of unplanned dialysis initiation on medical costs per admission, stratified by physical function and cognitive impairment. AME was derived from a generalized linear regression model adjusted for age, sex, BMI, comorbidities, physical function, cognitive impairment, emergency admission, ambulance transfer, readmission, hospital volume, and admission year. AME, average marginal effect; BMI, body mass index.
Discussion
This study examined the association between UDI and both physical dysfunction and cognitive impairment in the elderly. Our results showed that both factors increased the risk of UDI, although physical dysfunction had a stronger association than cognitive impairment. Medical costs increased with worsening physical function and cognitive impairment, with UDI associated with markedly higher medical costs, and the AME of UDI per admission being $7178 (¥782,520; 95% CI: $7019–$7338 [¥765,191–¥788,838]; P-value <0.001). Cognitive impairment was linked to increasing AMEs of UDI, although the impact of physical dysfunction was not.
Our findings highlight the challenges faced by elderly patients suffering from physical dysfunction and/or cognitive impairment regarding planning for dialysis initiation. Our approach aimed to distinguish the specific impacts of physical and cognitive function on the risk of UDI. This distinction can be difficult to achieve using common frailty indicators, such as the Clinical Frailty Scale, Frailty Index, or the Fried Frailty Phenotype. Although these tools offer a convenient and comprehensive assessment that incorporates both cognitive and physical components, they do not readily separate the individual contributions of each. In contrast, our study highlights the risk of UDI in older adults from both physical and cognitive perspectives. Our study findings suggest that UDI may negatively affect the selection of PD, especially among patients with declining physical or cognitive function. Reducing UDI could therefore enhance patients’ ability to choose PD.
The strong association between physical dysfunction and UDI may be attributed to the link between loss of muscle strength and CKD progression, as previously reported.17 Muscle weakness, combined with multiple comorbidities, can mask the rapid progression of CKD. In addition, muscle loss can lead to an overestimation of kidney function. Given that creatinine originates from the muscles, older patients with muscle loss will show falsely high estimated glomerular filtration rate when assessed via serum creatinine levels.42 A previous study reported that estimated glomerular filtration rate based on creatinine has the lowest sensitivity (68.4%) in patients with low muscle mass.42 To address this, routine estimated glomerular filtration rate measurements using cystatin C, which is unaffected by muscle mass, is considered to be beneficial.43
Our study demonstrated the association between cognitive impairment and UDI. Generally, cognitive impairment can affect doctor-patient communication, treatment adherence, the likelihood of medical follow-up, medication selection, and potential side effects, thereby influencing overall health and increasing both caregiver burden and health care costs.44 In patients with CKD, cognitive impairment affecting orientation, attention, language, memory, and executive functions is common even from the early stages and can worsen as the disease progresses.45 As a result, the presence of cognitive impairment reduces a patient’s ability to make optimal health care decisions.45 Furthermore, previous studies suggest that cognitive impairments are linked to difficulties in achieving self-care among patients with advanced CKD.46 Loss of ability in self-care, communication, treatment adherence, and health care decision-making negatively influences the process of receiving optimal medical care. This subsequently increases the risk of UDI. For patients with suspected cognitive impairment, closer follow-up and early referral to a specialist are recommended.
Cognitive impairment may increase the risk of UDI through its negative impact on health literacy, which is strongly associated with poor health outcomes in patients with CKD and ESKD.45,47 Health literacy relies on higher-order cognitive functions, enabling individuals to interpret, assess, compare, and select the most advantageous health information. In patients with CKD, these higher-order cognitive abilities tend to decline early in the course of the disease.45 Consequently, patients with advanced CKD and cognitive impairment are likely to have compromised cognitive functions essential for health literacy. Investigating specific patterns of cognitive impairment and low health literacy contributing to UDI could provide valuable insights into the underlying mechanisms.
Consistent with previous studies, UDI, alongside physical dysfunction and cognitive impairment, was strongly associated with increased medical costs.10,19,21 We found that the financial burden of UDI becomes even greater in patients with severe cognitive impairment. The increasing medical costs may be partly because of a higher risk of complications when using temporary catheters, including infections, bleeding, or accidental catheter removal via self-manipulation.48 The extended hospital stay required for physical rehabilitation often overlaps with the time needed to create vascular access. As a result, the overall effect of UDI on total hospitalization days may be diminished, which could explain the observed reduction in AME. Further research is needed to explore the underlying mechanisms driving the increase in medical costs in patients with cognitive impairment undergoing UDI. Our findings suggest that optimizing care strategies for elderly patients, particularly addressing age-related physical and cognitive decline, is vital for preventing UDI and reducing health care costs.
To reduce UDI, fostering a collaborative health care framework is essential, because delayed referrals to nephrologists are known to increase the risk of UDI and mortality.49 Previous studies have shown that older patients are referred later, with nonnephrologists being less likely to refer them compared with nephrologists.49,50 Cognitive impairment also poses a potential risk, because it has been shown to lower the likelihood of timely referral by nonnephrologists.50 However, our study suggests that physical function plays a more prominent role than cognitive impairment in UDI among the elderly. Late referrals can delay the timely creation of vascular access before starting dialysis. The DOPPS study indicates that a longer wait between referral and access creation is associated with a higher risk of UDI.36 Because Japan is among the countries with the shortest wait times, the impact of physical and cognitive function on UDI may be even greater in countries with longer wait times.
Some older patients, especially those with physical dysfunction or cognitive impairment, may not consider dialysis as their preferred option, viewing it as an added burden. For such individuals, conservative kidney management (CKM) offers an alternative, nondialysis approach, focusing on symptom relief in ESKD.4,8 However, decision-making regarding CKM is often complicated by a lack of evidence, limited end-of-life care training among nephrologists, and widespread misconceptions regarding CKM and dialysis in both the medical community and the general public.51 Both patients and health care providers require ample time to engage in shared decision-making, necessitating a collaborative, discussion-based model involving patients, clinicians, and caregivers in the decision-making process.8 Our findings suggest that patients with declining physical function and cognitive impairments are at higher risk for unpleasant and unexpected dialysis initiation. Given the connection between cognitive impairment and the challenges it presents for health care decision-making and self-care, implementing CKM in these patients will be difficult and require further discussion. Developing a structured care pathway for older predialysis patients with physical dysfunction and/or cognitive impairment will require a multidisciplinary approach and evidence-based care guidelines. Educational programs for both nephrologists and nonnephrologists focused on CKM and dialysis options could facilitate early referrals and promote smoother shared decision-making processes.
Strengths and Limitations
Although this study has several strengths, including the use of a large-scale database with a national representation of the Japanese population, which has been rigorously validated,25 it also has several limitations. First, the population was limited to a single race and ethnicity in Japan, which may reduce the generalizability of the findings to other populations. Second, we were unable to assess laboratory and preadmission data including actual kidney functions. This limitation made it challenging to strictly differentiate AKI cases from the database. To address this issue, we restricted the admission diagnosis to ESKD to exclude patients admitted for conditions that could lead to severe AKI (e.g., severe infections or cardiac surgery). In addition, we excluded individuals who required temporary dialysis or those who initiated continuous renal replacement therapy, to minimize the inclusion of cases of otherwise normal kidney function experiencing severe AKI or atypical critically ill patients starting dialysis. As a result, most patients were likely to have advanced CKD at the time of admission. This approach may have helped reduce the confounding effects of acute, critical illness on physical function measurements. Third, our database lacked nutritional status, exercise habits, and education, all of which could influence physical and cognitive function. Fourth, although the tool used to identify cognitive impairments is widely utilized and well-validated in Japan,28, 29, 30, 31, 32, 33, 34, 35 it is not commonly employed outside the country. Furthermore, to the best of our knowledge, this study is the first to apply this tool to patients with CKD and ESKD. Therefore, further validation studies are necessary to ensure its generalizability. Fifth, our definition of UDI has limitations in identifying patients with planned time-limited trials, goal-directed care or patient preferences, or a lack of vascular access options. However, these scenarios likely have minimal impact. In Japan, time-limited trials are uncommon because of ongoing legal and ethical debate.52 In addition, cases of limited vascular access are rare because tunneled catheters are typically used in such situations. In this study, we differentiated between tunneled catheters, which are used as permanent devices, and untunneled catheters, which are used as temporary devices. Distinguishing primary failure from failure to mature is also challenging. To address this, we conducted a sensitivity analysis excluding patients with fistula that failed to mature at admission (n = 1688). Because the results were consistent with our primary analysis, we concluded that failure to mature does not significantly impact our findings. Furthermore, planned fistula procedures with primary failure at admission are even less common than failures to mature, making it unlikely that they would alter our conclusions.
Conclusion
When examining both physical and cognitive functions, they were each associated with the risk of UDI in the elderly, with physical dysfunction being more prominently associated with this risk compared to cognitive impairment. UDI in the elderly was strongly associated with higher medical costs. To improve smooth transitions to kidney replacement therapy in these patients, we should establish a framework for integrating multidisciplinary care teams that include nephrologists, geriatric specialists, and social workers. It is also important to enhance physical and cognitive assessments, particularly in elderly patients with advanced CKD, as part of routine clinical practice. Overall, further research is required to refine ESKD treatment strategies in the elderly.
Disclosure
All the authors declared no competing interests.
Acknowledgments
We would like to thank all the study participants.
Funding
This study was partly supported by the Health and Labor Sciences Research Grant (Grant No. Seisaku-Sitei-24AA2006 to KF) of the Japan Ministry of Health, Labor and Welfare. For the remaining authors, none were declared.
Data Availability Statement
The datasets analyzed during the current study are not publicly available because a third party provided them and cannot be deposited in a public repository without permission from the data custodians. However, they are available from the corresponding author upon reasonable request.
Author Contributions
Conceptualization was by YN and SM. Study design was done by YN and SM. Formal analysis was done by YN and SM. Data acquisition was done by YN, SM, and KF. Data curation was done by YN, SM,YM, FA, KS, TM, SI, SN, ES, KF, and SU. Software use was done by YN. Visualization was done by YN. Writing of the original draft was done by YN. Writing-review and editing was done by SM, YM, FA, KS, TM, SI, SN, ES, KF, and SU. Supervision was by SM and SU. Funding acquisition was by KF. All authors read and approved the final manuscript.
Footnotes
Figure S1. Flowchart of the study.
Figure S2. The average marginal effect of unplanned dialysis initiation on length of hospital stay per admission, stratified by physical function and cognitive impairment.
Table S1. ICD-10 coding algorithms for comorbidities.
Table S2. Degree of independence in daily living for the demented elderly.
Table S3. Proportion of the initiation of maintenance peritoneal dialysis among patients with and without unplanned dialysis initiation.
Table S4. The association of unplanned dialysis initiation across levels of physical function and cognitive impairment in full samples, complete case samples, and restricted samples excluding emergent admission.
Table S5. The predicted probabilities of unplanned dialysis initiation across levels of physical function and cognitive impairment.
Supplementary Material
Figure S1. Flowchart of the study. Figure S2. The average marginal effect of unplanned dialysis initiation on length of hospital stay per admission, stratified by physical function and cognitive impairment. Table S1. ICD-10 coding algorithms for comorbidities. Table S2. Degree of independence in daily living for the demented elderly. Table S3. Proportion of the initiation of maintenance peritoneal dialysis among patients with and without unplanned dialysis initiation. Table S4. The association of unplanned dialysis initiation across levels of physical function and cognitive impairment in full samples, complete case samples, and restricted samples excluding emergent admission. Table S5. The predicted probabilities of unplanned dialysis initiation across levels of physical function and cognitive impairment.
References
- 1.Thurlow J.S., Joshi M., Yan G., et al. Global epidemiology of end-stage kidney disease and disparities in kidney replacement therapy. Am J Nephrol. 2021;52:98–107. doi: 10.1159/000514550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Jager K.J., Kovesdy C., Langham R., Rosenberg M., Jha V., Zoccali C. A single number for advocacy and communication-worldwide more than 850 million individuals have kidney diseases. Kidney Int. 2019;96:1048–1050. doi: 10.1016/j.kint.2019.07.012. [DOI] [PubMed] [Google Scholar]
- 3.Ortiz A., Mattace-Raso F., Soler M.J., Fouque D. Ageing meets kidney disease. Clin Kidney J. 2022;15:1793–1796. doi: 10.1093/ckj/sfac151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Schaeffner E. Smoothing transition to dialysis to improve early outcomes after dialysis initiation among old and frail adults-a narrative review. Nephrol Dial Transplant. 2022;37:2307–2313. doi: 10.1093/ndt/gfab342. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Duarte M.P., Almeida L.S., Neri S.G.R., et al. Prevalence of sarcopenia in patients with chronic kidney disease: a global systematic review and meta-analysis. J Cachexia Sarcopenia Muscle. 2024;15:501–512. doi: 10.1002/jcsm.13425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Singh-Manoux A., Oumarou-Ibrahim A., Machado-Fragua M.D., et al. Association between kidney function and incidence of dementia: 10-year follow-up of the Whitehall II cohort study. Age Ageing. 2022;51 doi: 10.1093/ageing/afab259. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kurella Tamura M., Li S., Chen S.C., et al. Educational programs improve the preparation for dialysis and survival of patients with chronic kidney disease. Kidney Int. 2014;85:686–692. doi: 10.1038/ki.2013.369. [DOI] [PubMed] [Google Scholar]
- 8.Kanbay M., Basile C., Battaglia Y., et al. Shared decision making in elderly patients with kidney failure. Nephrol Dial Transplant. 2024;39:742–751. doi: 10.1093/ndt/gfad211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Loos C., Briançon S., Frimat L., Hanesse B., Kessler M. Effect of end-stage renal disease on the quality of life of older patients. J Am Geriatr Soc. 2003;51:229–233. doi: 10.1046/j.1532-5415.2003.51062.x. [DOI] [PubMed] [Google Scholar]
- 10.Wong L.P., Anindita G., Li J., Rizk M.K., Hohman J.A. Dialysis costs for a health system participating in value-based care. Am J Manag Care. 2023;29(8):e235–e241. doi: 10.37765/ajmc.2023.89410. [DOI] [PubMed] [Google Scholar]
- 11.Arulkumaran N., Navaratnarajah A., Pillay C., et al. Causes and risk factors for acute dialysis initiation among patients with end-stage kidney disease-a large retrospective observational cohort study. Clin Kidney J. 2019;12:550–558. doi: 10.1093/ckj/sfy118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Mendelssohn D.C., Curtis B., Yeates K., et al. Suboptimal initiation of dialysis with and without early referral to a nephrologist. Nephrol Dial Transplant. 2011;26:2959–2965. doi: 10.1093/ndt/gfq843. [DOI] [PubMed] [Google Scholar]
- 13.Hassan R., Akbari A., Brown P.A., Hiremath S., Brimble K.S., Molnar A.O. Risk factors for unplanned dialysis initiation: a systematic review of the literature. Can J Kidney Health Dis. 2019;6 doi: 10.1177/2054358119831684. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wachterman M.W., O’Hare A.M., Rahman O.K., et al. One-year mortality after dialysis initiation among older adults. JAMA Intern Med. 2019;179:987–990. doi: 10.1001/jamainternmed.2019.0125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Boenink R., Astley M.E., Huijben J.A., et al. The ERA Registry Annual Report 2019: summary and age comparisons. Clin Kidney J. 2022;15:452–472. doi: 10.1093/ckj/sfab273. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hanafusa N., Abe M., Joki N., et al. Annual dialysis data report 2020, JSDT renal data registry. Ren Replace Ther. 2024;10:1–31. [Google Scholar]
- 17.Nakano Y., Mandai S., Naito S., et al. Effect of osteosarcopenia on longitudinal mortality risk and chronic kidney disease progression in older adults. Bone. 2024;179 doi: 10.1016/j.bone.2023.116975. [DOI] [PubMed] [Google Scholar]
- 18.Vahedi A., Eriksdotter M., Ihle-Hansen H., Wyller T.B., Øksengård A.R., Fure B. Cognitive impairment in people with physical frailty using the phenotype model: a systematic review and meta analysis. Int J Geriatr Psychiatry. 2022;37 doi: 10.1002/gps.5822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Leibson C.L., Long K.H., Ransom J.E., et al. Direct medical costs and source of cost differences across the spectrum of cognitive decline: a population-based study. Alzheimers Dement. 2015;11:917–932. doi: 10.1016/j.jalz.2015.01.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Chen L.K., Woo J., Assantachai P., et al. Asian Working Group for Sarcopenia. Consensus update on sarcopenia diagnosis and treatment. J Am Med Dir Assoc. 2019;21:300–307.e2. doi: 10.1016/j.jamda.2019.12.012. [DOI] [PubMed] [Google Scholar]
- 21.Han L., Clegg A., Doran T., Fraser L. The impact of frailty on healthcare resource use: a longitudinal analysis using the Clinical Practice Research Datalink in England. Age Ageing. 2019;48:665–671. doi: 10.1093/ageing/afz088. [DOI] [PubMed] [Google Scholar]
- 22.Ofori-Asenso R., Chin K.L., Mazidi M., et al. Global incidence of frailty and prefrailty among community-dwelling older adults: a systematic review and meta-analysis. JAMA Netw Open. 2019;2 doi: 10.1001/jamanetworkopen.2019.8398. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Magadi W., Birnie K., Santhakumaran S., Caskey F.J., Ben-Shlomo Y. An updated systematic review of the risk factors for unplanned dialysis initiation. Clin Kidney J. 2024;17 doi: 10.1093/ckj/sfae333. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Nakano Y., Mandai S., Genma T., et al. Nationwide mortality associated with perioperative acute dialysis requirement in major surgeries. Int J Surg. 2022;104 doi: 10.1016/j.ijsu.2022.106816. [DOI] [PubMed] [Google Scholar]
- 25.Yamana H., Moriwaki M., Horiguchi H., Kodan M., Fushimi K., Yasunaga H. Validity of diagnoses, procedures, and laboratory data in Japanese administrative data. J Epidemiol. 2017;27:476–482. doi: 10.1016/j.je.2016.09.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Quan H., Sundararajan V., Halfon P., et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43:1130–1139. doi: 10.1097/01.mlr.0000182534.19832.83. [DOI] [PubMed] [Google Scholar]
- 27.Ministry of Health Labour and Welfare Outline of survey of Institutions and Establishments for long term care 2010. Published 2010. https://www.mhlw.go.jp/english/database/db-hss/dl/siel-2010-01.pdf
- 28.Ihira H., Sawada N., Inoue M., et al. Association between physical activity and risk of disabling dementia in japan. JAMA Netw Open. 2022;5 doi: 10.1001/jamanetworkopen.2022.4590. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hikichi H., Aida J., Kondo K., et al. Increased risk of dementia in the aftermath of the 2011 Great East Japan Earthquake and Tsunami. Proc Natl Acad Sci U S A. 2016;113:E6911–E6918. doi: 10.1073/pnas.1607793113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Sakaniwa R., Shirai K., Cador D., et al. Socioeconomic status transition throughout life and risk of dementia. JAMA Netw Open. 2024;7 doi: 10.1001/jamanetworkopen.2024.12303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Tani Y., Fujiwara T., Kondo K. Association between adverse childhood experiences and dementia in older Japanese adults. JAMA Netw Open. 2020;3 doi: 10.1001/jamanetworkopen.2019.20740. [DOI] [PubMed] [Google Scholar]
- 32.Watanabe T., Tamiya N. Utilization of Japanese long-term care-related data including Kaigo-DB: an analysis of current trends and future directions. Glob Health Med. 2024;6:63–69. doi: 10.35772/ghm.2023.01135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yoshikawa M., Goto E., Shin J.H., Imanaka Y. Regional disparities in Dementia-free Life Expectancy in Japan: an ecological study, using the Japanese long-term care insurance claims database. PLoS One. 2023;18 doi: 10.1371/journal.pone.0280299. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Hisano S. The relationship between revised Hasegawa dementia scale (HDS-R), mini-mental state examination (MMSE) and bed-fast scale, dementia scale. Jpn J Geriatr Psychiatry. 2009;20:883. https://cir.nii.ac.jp/crid/1370016861578773522 [Google Scholar]
- 35.Noda H., Yamagishi K., Ikeda A., Asada T., Iso H. Identification of dementia using standard clinical assessments by primary care physicians in Japan. Geriatr Gerontol Int. 2018;18:738–744. doi: 10.1111/ggi.13243. [DOI] [PubMed] [Google Scholar]
- 36.Ethier J., Mendelssohn D.C., Elder S.J., et al. Vascular access use and outcomes: an international perspective from the Dialysis Outcomes and Practice Patterns Study. Nephrol Dial Transplant. 2008;23:3219–3226. doi: 10.1093/ndt/gfn261. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Iimuro S., Kaneko T., Ohashi Y., et al. Analysis of 2897 hospitalization events for patients with chronic kidney disease: results from CKD-JAC study. Clin Exp Nephrol. 2019;23:956–968. doi: 10.1007/s10157-019-01730-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Royston P., White I. Multiple imputation by chained equations (MICE): implementation inStata. J Stat Softw. 2011;45:1–20. doi: 10.18637/jss.v045.i04. [DOI] [Google Scholar]
- 39.Austin P.C., White I.R., Lee D.S., van Buuren S. Missing data in clinical research: a tutorial on multiple imputation. Can J Cardiol. 2021;37:1322–1331. doi: 10.1016/j.cjca.2020.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Muller C.J., MacLehose R.F. Estimating predicted probabilities from logistic regression: different methods correspond to different target populations. Int J Epidemiol. 2014;43:962–970. doi: 10.1093/ije/dyu029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Isumi A., Fujiwara T., Kato H., et al. Assessment of additional medical costs among older adults in Japan with a history of childhood maltreatment. JAMA Netw Open. 2020;3 doi: 10.1001/jamanetworkopen.2019.18681. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Nankivell B.J., Nankivell L.F.J., Elder G.J., Gruenewald S.M. How unmeasured muscle mass affects estimated GFR and diagnostic inaccuracy. EClinicalmedicine. 2020;29-30 doi: 10.1016/j.eclinm.2020.100662. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Groothof D., Post A., Polinder-Bos H.A., et al. Muscle mass and estimates of renal function: a longitudinal cohort study. J Cachexia Sarcopenia Muscle. 2022;13:2031–2043. doi: 10.1002/jcsm.12969. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Perry W., Lacritz L., Roebuck-Spencer T., et al. Population Health Solutions for assessing cognitive impairment in geriatric patients. Clin Neuropsychol. 2018;32:1193–1225. doi: 10.1080/13854046.2018.1517503. [DOI] [PubMed] [Google Scholar]
- 45.Berger I., Wu S., Masson P., et al. Cognition in chronic kidney disease: a systematic review and meta-analysis. BMC Med. 2016;14:206. doi: 10.1186/s12916-016-0745-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Thancharoen O., Waleekhachonloet O., Limwattananon C., Anutrakulchai S. Cognitive impairment, quality of life and healthcare utilization in patients with chronic kidney disease stages 3 to 5. Nephrology (Carlton) 2020;25:625–633. doi: 10.1111/nep.13705. [DOI] [PubMed] [Google Scholar]
- 47.Toapanta N., Salas-Gama K., Pantoja P.E., Soler M.J. The role of low health literacy in shared treatment decision-making in patients with kidney failure. Clin Kidney J. 2023;16(suppl 1):i4–i11. doi: 10.1093/ckj/sfad061. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Stroz M.J., Nathoo B., Stroz P.M. Posteriorly tunneled dialysis catheters for permanent use in cognitively impaired patients undergoing hemodialysis. J Vasc Interv Radiol. 2014;25:1621–1625. doi: 10.1016/j.jvir.2014.06.017. [DOI] [PubMed] [Google Scholar]
- 49.Baer G., Lameire N., Van Biesen W. Late referral of patients with end-stage renal disease: an in-depth review and suggestions for further actions. NDT Plus. 2010;3:17–27. doi: 10.1093/ndtplus/sfp050. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Visser A., Dijkstra G.J., Huisman R.M., Gansevoort R.T., de Jong P.E., Reijneveld S.A. Differences between physicians in the likelihood of referral and acceptance of elderly patients for dialysis-influence of age and comorbidity. Nephrol Dial Transplant. 2007;22:3255–3261. doi: 10.1093/ndt/gfm382. [DOI] [PubMed] [Google Scholar]
- 51.Grubbs V., Tuot D.S., Powe N.R., O’Donoghue D., Chesla C.A. System-level barriers and facilitators for foregoing or withdrawing dialysis: A qualitative study of nephrologists in the United States and England. Am J Kidney Dis. 2017;70:602–610. doi: 10.1053/j.ajkd.2016.12.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Watanabe Y., Hirakata H., Okada K., et al. Proposal for the shared decision-making process regarding initiation and continuation of maintenance hemodialysis: JSDT proposal for withholding/withdrawal from hemodialysis. Ther Apher Dial. 2015;19(suppl 1):108–117. doi: 10.1111/1744-9987.12295. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Flowchart of the study. Figure S2. The average marginal effect of unplanned dialysis initiation on length of hospital stay per admission, stratified by physical function and cognitive impairment. Table S1. ICD-10 coding algorithms for comorbidities. Table S2. Degree of independence in daily living for the demented elderly. Table S3. Proportion of the initiation of maintenance peritoneal dialysis among patients with and without unplanned dialysis initiation. Table S4. The association of unplanned dialysis initiation across levels of physical function and cognitive impairment in full samples, complete case samples, and restricted samples excluding emergent admission. Table S5. The predicted probabilities of unplanned dialysis initiation across levels of physical function and cognitive impairment.
Data Availability Statement
The datasets analyzed during the current study are not publicly available because a third party provided them and cannot be deposited in a public repository without permission from the data custodians. However, they are available from the corresponding author upon reasonable request.