Remote Monitoring in Automated Peritoneal Dialysis Improves Treatment Adherence and Reduces Hospitalization Burden: A Prospective Multicenter Cohort Study

Abstract

Introduction:

Remote patient monitoring (RPM) in peritoneal dialysis (PD) has the potential to improve adherence and clinical outcomes, yet comparative real-world data among PD modalities are limited. This study aimed to compare PD treatment adherence, hospitalization burden, fluid status, and health care costs among patients undergoing continuous ambulatory PD (CAPD), automated PD (APD), and remote monitoring-assisted APD (RM-APD).

Methods:

In this prospective multicenter cohort study, 73 patients maintained on a single PD modality for ≥1 year were enrolled (CAPD: n = 26; APD: n = 23; RM-APD: n = 24). Treatment adherence was defined as the proportion of prescribed PD treatments actually performed (sessions for APD/RM-APD; exchanges for CAPD), expressed on a per-treatment-day basis derived from each patient’s prescription and dialysate supply records. In RM-APD patients, supply-based estimates were cross-validated against Sharesource treatment records. Fluid status was evaluated using the overhydration-to-extracellular water ratio (OH/ECW%). Hospitalization days, emergency department visits, health care costs, and clinical parameters were assessed over 1 year.

Results:

The RM-APD group showed the highest treatment adherence (median 83.4%, interquartile range 81.0–88.2), compared with 81.0% in CAPD and 78.4% in APD (Kruskal–Wallis p = 0.010). Adjusted analysis confirmed superior adherence in RM-APD compared with APD (p = 0.013) and CAPD (p = 0.041). In a subset of RM-APD patients, the supply-based estimate correlated well with Sharesource-recorded treatment delivery. RM-APD patients had significantly fewer hospitalization days (adjusted incidence rate ratio 0.34; 95% confidence interval 0.23–0.49; p < 0.001) and lower inpatient costs than APD (p = 0.037). OH/ECW% reduction was greater in RM-APD than CAPD (p = 0.042), while nutritional markers, dialysis adequacy (Kt/V), and ejection fraction were comparable across groups.

Conclusions:

Integration of remote monitoring into APD was associated with improved adherence, reduced hospitalization burden, and better volume control compared with conventional PD modalities. These findings provide real-world evidence that RPM-enabled home dialysis can enhance both clinical outcomes and health care efficiency, supporting broader implementation of telehealth-based care models for patients with end-stage kidney disease.

Introduction

Peritoneal dialysis (PD) is a home-based renal replacement therapy that offers patients greater independence, better quality of life, and preservation of residual renal function compared with in-center hemodialysis. However, successful PD depends heavily on the patient’s ability to adhere to prescribed treatment regimens in an unsupervised setting. Patients are expected to actively manage fluid removal, blood pressure, and treatment parameters at home, and timely communication with health care providers is essential when complications arise. Delays in detection or reporting can preclude early intervention and adversely affect outcomes.

A major challenge with home-based PD is the lack of frequent, direct monitoring by health care teams, which may lead to suboptimal treatment delivery. Without regular feedback and support, patients may deviate from their prescribed regimens, either unintentionally or owing to difficulties in self-management.^1,2 Poor adherence has been associated with increased peritonitis incidence, reduced dialysis adequacy, increased hospitalization duration, higher rates of transition to hemodialysis, and increased mortality rates.^1,3,4 These risks underscore the importance of timely intervention and sustained engagement between patients and clinical teams.

Recent advances in digital health have enabled the integration of remote patient monitoring (RPM) platforms with automated peritoneal dialysis (APD) systems. These technologies allow real-time transmission of treatment data—including exchange volumes, ultrafiltration, dwell times, and alarm events—between patients and dialysis centers, enabling health care providers to track adherence, detect deviations from prescriptions, and intervene promptly. RPM systems in PD have been shown to improve patient safety, reduce unplanned health care use, and enhance overall treatment efficiency.^5–7 RPM has been successfully applied to the management of other chronic conditions such as heart failure, ⁸ diabetes, ⁹ and hypertension, ¹⁰ where continuous monitoring and timely treatment adjustments are required. However, prospective real-world evidence comparing remote monitoring-assisted APD (RM-APD) with conventional modalities such as continuous ambulatory peritoneal dialysis (CAPD) and nonmonitored APD remains scarce.

In this multicenter prospective cohort study, we compared PD treatment adherence, hospitalization burden, fluid status, and health care costs among patients receiving CAPD, APD, and RM-APD. We hypothesized that the addition of remote monitoring to APD would translate into measurable improvements in adherence, clinical outcomes, and health care resource utilization in real-world practice.

Methods

Study design and setting

This was a multicenter prospective observational cohort study conducted in three tertiary hospitals in South Korea. All participating centers have well-established PD programs and prior experience with RPM technology. This study was approved by the Integrated Institutional Review Board of Korea University Medical Center (IRB no. 2020GR0553), which oversees the three participating hospitals. Written informed consent was obtained from all participants prior to enrollment. The study was conducted in accordance with the Declaration of Helsinki, and reporting follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines for observational studies.

Adult patients with end-stage kidney disease (ESKD) who initiated or continued PD (CAPD, APD, or RM-APD) were enrolled between February 2021 and August 2022. Patients were included if they had been on a stable PD regimen for at least 1 month and had no plans for imminent modality change. Patients were excluded if they had a life expectancy of <6 months, had an active malignancy, or were unable to complete follow-up due to transfer or other logistical reasons. The follow-up duration for each patient was 1 year. Patients who underwent kidney transplantation, transitioned to hemodialysis, or changed their PD regimen (e.g., from APD to RM-APD or from CAPD to APD) during the study period were excluded from the final analysis.

Remote monitoring system

RM-APD patients used cycler devices equipped with a cloud-connected RPM platform (Sharesource™, Baxter Healthcare). The system automatically transmits each treatment session’s data—including drained and infused volumes, ultrafiltration, dwell time, dextrose concentration, and alarm events—to a secure web portal accessible to the dialysis care team. Clinical staff at each center reviewed transmitted data on a regular schedule and contacted patients by telephone when prescription deviations, missed sessions, or technical alarms were identified. Patients in the CAPD and APD groups received standard care, with outpatient visits at monthly to bimonthly intervals and no remote data transmission.

Data collection

Baseline demographic and 1-year follow-up clinical data were collected, including age, sex, cause of ESKD, comorbidities (diabetes mellitus and hypertension), blood pressure, laboratory values (serum albumin and hemoglobin), nutritional status assessed by the Patient-Generated Subjective Global Assessment (PG-SGA, Korean version), and dialysis adequacy (Kt/V).

Treatment adherence was defined as the proportion of prescribed PD treatments actually performed over the follow-up period. The relationship between dialysate bags and delivered treatments differs across modalities. In CAPD, one bag corresponds to one exchange. In APD and RM-APD, a single automated session draws on a variable number of bags (typically two to four), plus a separate icodextrin bag for the daytime dwell. For this reason, adherence was computed on a per-treatment-day basis rather than by bag count. For each patient, prescribed and undelivered (unused) dialysate supplies were converted into treatment-days (exchanges for CAPD; sessions for APD and RM-APD) using the individual prescription, and adherence (%) was calculated as [(prescribed treatment-days − undelivered treatment-days)/prescribed treatment-days] × 100. In a subset of RM-APD patients with available day-level Share source records, this supply-based estimate was cross-validated against the actually delivered treatment record. This record was defined as the number of treatment-days with a recorded session divided by the number of prescribed treatment-days over the same follow-up window. Agreement was assessed by Spearman correlation (Supplementary Fig. S1). Sharesource records were not available for CAPD or APD patients.

Volume status was assessed by bioimpedance body composition monitoring (BCM) at baseline and at 1 year, with the overhydration-to-extracellular water ratio (OH/ECW%) calculated as fluid overload divided by extracellular water as previously described.^11,12 Echocardiography was performed at the same time points to measure left ventricular ejection fraction.

Hospitalization days, emergency department (ED) visits, and health care costs (total, outpatient, and inpatient) were recorded over 1 year.

Outcomes

The primary outcome was PD treatment adherence (per-treatment-day) over the 1-year follow-up period. The secondary outcomes were hospitalization duration; ED visits; health care costs (total, outpatient, and inpatient); 1-year change in fluid status (OH/ECW% by BCM); and 1-year change in clinical parameters over 1 year, including dialysis adequacy (Kt/V), serum albumin level, PG-SGA components, and echocardiographic ejection fraction. Kt/V was reported to confirm comparable dialysis adequacy across groups rather than as an efficacy outcome; because PD prescriptions are titrated to maintain a target Kt/V, similar values across groups are expected by design.

Statistical analysis

Continuous variables are presented as mean ± standard deviation or median with interquartile range (IQR); they were compared using one-way analysis of variance or the Kruskal–Wallis test, depending on distribution. Categorical variables were compared using chi-square or Fisher’s exact test. Treatment adherence was compared across groups using the Kruskal–Wallis test with Dunn’s post hoc test (Bonferroni correction) as the primary analysis; rank-based analysis of covariance (ANCOVA) adjusted for the prespecified covariates was performed as a supportive sensitivity analysis. Adjusted comparisons for costs were made using rank-based ANCOVA. Agreement between the supply-based adherence estimate and the Sharesource-recorded treatment delivery in RM-APD patients was assessed using Spearman correlation. Count-based outcomes (ED visits and hospitalization days) were analyzed using Poisson regression. All models were adjusted for potential confounders, including age, sex, ESKD cause, comorbidities, baseline Kt/V, serum albumin level, and nutritional status. Results with two-tailed p value <0.05 were considered statistically significant. Statistical analyses were performed using R software (version 4.4.3).

Results

Baseline characteristics

Seventy-three patients were included in the final analysis: 26 in the CAPD group, 23 in the APD group, and 24 in the RM-APD group. Baseline demographics such as age, sex, cause of ESKD, and presence of comorbidities such as diabetes and hypertension were comparable across groups (Table 1). The baseline PG-SGA score was significantly lower (indicating better nutritional status) in the RM-APD group than in the other two groups (p = 0.034); other baseline parameters did not differ significantly. This baseline imbalance was addressed by inclusion of PG-SGA in all adjusted models.

Table 1.

Baseline Characteristics of the Study Population

Group	CAPD (N = 26)	APD (N = 23)	RM-APD (N = 24)	p
Age (years)	52.6 ± 13.5	54.5 ± 13.5	52.6 ± 9.5	0.833
Male sex, n (%)	12 (46.2%)	12 (52.2%)	14 (58.3%)	0.69
Cause of ESKD, n (%)				0.91
DM	7 (26.9%)	7 (30.4%)	6 (25.0%)
HTN	8 (30.8%)	7 (30.4%)	9 (37.5%)
CGN	9 (34.6%)	6 (26.1%)	6 (25.0%)
PKD	0 (0.0%)	1 (4.3%)	0 (0.0%)
Others	2 (7.7%)	2 (8.7%)	3 (12.5%)
DM, n (%)	12 (46.2%)	10 (43.5%)	11 (45.8%)	0.98
HTN, n (%)	20 (76.9%)	20 (87.0%)	23 (95.8%)	0.151
IHD, n (%)	1 (3.8%)	3 (13.0%)	1 (4.2%)	0.364
CHF, n (%)	2 (7.7%)	5 (21.7%)	4 (16.7%)	0.377
Liver disease, n (%)	4 (15.4%)	2 (8.7%)	0 (0.0%)	0.14
Body weight (kg)	63.1 [56.0; 72.0]	65.0 [60.9; 75.2]	66.0 [62.1; 78.9]	0.32
Body mass index (kg/m2)	24.6 ± 3.5	25.1 ± 3.6	25.6 ± 4.1	0.634
Baseline Kt/V	2.1 [1.8; 2.4]	1.9 [1.7; 2.3]	2.0 [1.7; 2.4]	0.493
Baseline SBP (mmHg)	126.3 ± 14.4	131.0 ± 18.0	132.3 ± 19.4	0.434
Baseline DBP (mmHg)	75.9 ± 11.7	79.5 ± 11.2	80.1 ± 10.2	0.354
Baseline Hb (g/dL)	10.4 ± 1.5	10.1 ± 1.3	10.5 ± 1.2	0.661
Baseline serum albumin (g/dL)	3.8 ± 0.5	3.8 ± 0.3	3.9 ± 0.3	0.809
Baseline urine amount (mL/day)	300 [0; 800]	50 [0; 1000]	750 [225; 1000]	0.088
Baseline PG-SGA score	3.0 [2.0; 5.0]	4.0 [3.0; 7.0]	2.0 [1.5; 3.5]	0.034
Baseline PG-SGA score (patient part only)	1.0 [0.0; 2.0]	2.0 [0.5; 4.5]	1.0 [0.0; 2.5]	0.345
Baseline OH in BCM	1.6 [0.5; 2.9]	0.9 [0.4; 3.5]	1.8 [1.0; 4.0]	0.276
Baseline OH/ECW % in BCM	10.6 [3.7; 15.0]	6.2 [2.1;16.7]	11.9 [6.4; 20.2]	0.231
Baseline ejection fraction (%) in echocardiography	62.5 [57.5; 62.5]	62.5 [62.5; 62.5]	61.2 [60.0; 62.5]	0.365

BCM, body composition monitoring; BMI, body mass index; CAPD, continuous ambulatory peritoneal dialysis; CHF, congestive heart failure; CGN, chronic glomerulonephritis; DBP, diastolic blood pressure; DM, diabetes mellitus; ECW, extracellular water; ESKD, end-stage kidney disease; HTN, hypertension; IHD, ischemic heart disease; OH, overhydration; PG-SGA, Patient-Generated Subjective Global Assessment; PKD, polycystic kidney disease; RM-APD, remote monitoring automated peritoneal dialysis; SBP, systolic blood pressure.

Comparison of treatment adherence among the CAPD, APD, and RM-APD groups

The median follow-up duration differed slightly across groups (CAPD, 357.5 [343.0–370.0] days; APD, 331.0 [318.0–352.0]; RM-APD, 335.0 [319.0–352.0]; p = 0.013). The number of unused dialysate bags during follow-up was lowest in the RM-APD group (123.0 [86.5–148.5]) compared with CAPD (175.5 [143.0–284.0]) and APD (216.0 [101.0–402.0]; p = 0.001), indicating greater treatment delivery as prescribed. Because the number of bags issued does not map one-to-one onto delivered treatments—extra bags issued for priming, setup, alarms, or replacement of defective bags inflate apparent consumption—adherence was quantified on a per-treatment-day basis rather than by raw bag count (see Methods).

Per-treatment-day adherence differed significantly among groups (Fig. 1). The RM-APD group achieved the highest median adherence (83.4%, IQR: 81.0–88.2), compared with 81.0% (74.5–85.5) in the CAPD group and 78.4% (74.2–80.4) in the APD group (Kruskal–Wallis p = 0.010). In Dunn post hoc testing, adherence was significantly higher in RM-APD than in APD (p = 0.008), with no significant difference between CAPD and APD (p = 0.44). In the supportive rank-ANCOVA adjusted for age, sex, cause of ESKD, comorbidities (diabetes, hypertension), body mass index, baseline Kt/V, baseline serum albumin, and baseline PG-SGA score, RM-APD remained superior to both APD (p = 0.013) and CAPD (p = 0.041). In a subset of RM-APD patients with available day-level Sharesource records, the supply-based adherence estimate showed good correlation with the actually delivered treatment record (Supplementary Fig. S1), supporting the validity of the supply-based metric.

FIG. 1.

Comparison of per-treatment-day adherence among the CAPD, APD, and RM-APD groups. The boxplot displays the median, interquartile range (IQR), and individual data points. Kruskal–Wallis p = 0.010. Pairwise p values from Dunn’s test with Bonferroni correction are shown above the boxes. APD, automated peritoneal dialysis; CAPD, continuous ambulatory peritoneal dialysis; PD, peritoneal dialysis; RM-APD, remote monitoring assisted automated peritoneal dialysis.

Comparison of hospitalization burden and emergency department visits among the CAPD, APD, and RM-APD groups

Hospitalization burden differed notably among the three groups (Table 2). Total hospitalization days over 1 year were 137 in the APD group, 79 in the CAPD group, and 49 in the RM-APD group. The proportion of patients with no hospitalization was lowest in the APD group (56.5%), followed by RM-APD (66.7%) and CAPD (76.9%). In the adjusted Poisson regression, the APD group had significantly more hospitalization days than the RM-APD group (adjusted incidence rate ratio [IRR]: 2.92; 95% CI, 2.03–4.26; p < 0.001). The CAPD group also showed a nonsignificant trend toward more hospitalization days versus RM-APD (adjusted IRR: 1.37; 95% CI, 0.91–2.05; p = 0.128). ED visits were rare overall (RM-APD, n = 1; APD, n = 2; CAPD, n = 11) and did not differ significantly after adjustment.

Table 2.

Comparison of Hospitalization Days and Emergency Room Visits Among CAPD, APD, and RM-APD Groups

Comparison	Unadjusted		Adjusted a
	IRR (95% CI)	p	IRR (95% CI)	p
Hospitalization days
CAPD vs. RM-APD	1.49 (1.05–2.12)	0.029	1.37 (0.91–2.05)	0.128
APD vs. RM-APD	2.92 (2.03–4.26)	<0.001	2.92 (2.03–4.26)	<0.001
Emergency room visits
CAPD vs. RM-APD	10.15 (1.31–31.44)	0.026	1.55 (0.18–12.74)	0.737
APD vs. RM-APD	2.09 (0.02–5.00)	0.548	0.18 (0.01–2.17)	0.422

Data are presented as IRRs with 95% CIs and p values derived from Poisson regression models.

^a Adjusted analyses were controlled for age, sex, cause of end-stage kidney disease, diabetes mellitus, hypertension, body mass index, baseline Kt/V, baseline serum albumin, and baseline PG-SGA score.

APD, automated peritoneal dialysis; CAPD, continuous ambulatory peritoneal dialysis; CI, confidence interval; IRR, incidence rate ratio; RM-APD, remote monitoring automated peritoneal dialysis.

Health care costs among the CAPD, APD, and RM-APD groups

Total and outpatient health care costs did not significantly differ among the groups. In the adjusted rank ANCOVA, the difference in total cost between the RM-APD and APD groups was −169,050 Korean won (KRW) (95% CI, −3,740,418 to 3,402,318; p = 0.925), where the difference between the CAPD and APD groups was −2,383,909 KRW (95% CI, −5,993,814 to 1,225,995; p = 0.192). Outpatient cost showed a borderline but nonsignificant increase in the RM-APD group compared with that in the APD group (coefficient: 2,146,201 KRW; 95% CI, −81,078 to 4,373,479; p = 0.065), likely reflecting more frequent provider engagement enabled by the remote monitoring system.

Among patients who had to be hospitalized, inpatient costs showed a trend toward differences among the three groups (Kruskal−Wallis χ² = 5.12, p = 0.077). Post hoc pairwise comparisons using Dunn’s test with Bonferroni correction indicated that the RM-APD group had significantly lower inpatient costs than the APD group (adjusted p = 0.037), consistent with the reduced hospitalization burden observed in the RM-APD group.

Changes in clinical outcomes over 1 year

Changes in clinical parameters from baseline to 1 year are summarized in Table 3. Differences from baseline to 1 year in PG-SGA score, serum albumin level, and Kt/V were not significant among the groups. Median change in PG-SGA was −0.5 (APD), 1.0 (CAPD), and 0.5 (RM-APD) (p = 0.758); patient-reported PG-SGA component changes were similarly comparable (p = 0.849). Kt/V remained stable in all groups (p = 0.552). Changes in OH/ECW% over 1 year showed a borderline difference among groups (Kruskal–Wallis p = 0.06); in pairwise comparison, the RM-APD group had a significantly greater reduction in OH/ECW% than the CAPD group (adjusted p = 0.042; Fig. 2), suggesting better volume control with remote monitoring. Echocardiographic ejection fraction did not differ across groups (p = 0.775).

Table 3.

Changes in Clinical Parameters over 1 Year in the CAPD, APD, and RM-APD Groups

Group	CAPD (N = 26)	APD (N = 23)	RM-APD (N = 24)	p
Difference in SBP (mmHg)	8.0 ± 14.8	0.3 ± 24.4	−1.6 ± 24.8	0.276
Difference in DBP (mmHg)	5.6 ± 12.1	0.1 ± 16.0	−4.4 ± 14.4	0.061
Difference in serum albumin levels (g/dL)	0.1 [0.0; 0.2]	−0.1 [−0.2; 0.1]	0.0 [−0.1; 0.3]	0.167
Difference in PG-SGA score	1.0 [−1.0; 2.0]	−0.5 [−2.0; 2.5]	0.5 [−0.5; 2.5]	0.758
Difference in PG-SGA score (patient part only)	1.0 [−1.0; 2.0]	0.0 [−2.0; 2.0]	0.5 [−0.5; 2.0]	0.849
Difference in Kt/V	−0.2 [−0.3; 0.0]	−0.0 [−0.3; 0.3]	0.0 [−0.4; 0.4]	0.552
Difference in OH/ECW %	1.0 [−2.2; 5.8]	0.9 [−1.8; 4.4]	−1.7 [−6.6; 1.1]	0.06
Difference in ejection fraction %	0.0 [0.0; 3.5]	0.0 [0.0; 0.0]	0.0 [−0.3; 6.0]	0.775

APD, automated peritoneal dialysis; CAPD, continuous ambulatory peritoneal dialysis; DBP, diastolic blood pressure; ECW, extracellular water; OH, overhydration; PG-SGA, Patient-Generated Subjective Global Assessment; RM-APD, remote monitoring automated peritoneal dialysis; SBP, systolic blood pressure.

FIG. 2.

Changes in OH/ECW% over 1 year in the CAPD, APD, and RM-APD groups. The boxplot displays the median, interquartile range (IQR), and individual data points. Kruskal−Wallis p = 0.06. Pairwise p values from Dunn’s test with Bonferroni correction are shown above the boxes. APD, automated peritoneal dialysis; CAPD, continuous ambulatory peritoneal dialysis; OH/ECW%, overhydration-to-extracellular water ratio; RM-APD, remote monitoring assisted automated peritoneal dialysis.

Discussion

In this prospective multicenter cohort study, integration of RPM into APD was associated with significantly higher treatment adherence, fewer hospitalization days, and better volume control compared with conventional CAPD and nonmonitored APD. Although the direction of these findings is consistent with prior international reports, to our knowledge, this is the first prospective real-world study to compare these three PD modalities concurrently with respect to adherence, fluid status, and actual health care costs. The findings position RM-APD as a clinically meaningful application of telehealth-enabled care in nephrology and provide direct evidence supporting the broader trend toward digitally augmented home therapies.

Poor adherence to PD regimens has long been recognized as a major clinical issue, with nonadherence rates ranging from 2.6% to 53% in a systematic review. ¹³ Nonadherent patients are known to have lower dialysis adequacy, increased hospitalization burden, higher rates of transition to hemodialysis, and higher mortality rates than adherent patients.^1,14 Therefore, identifying the underlying causes of nonadherence and addressing barriers to self-care are critical. RM-APD may play a pivotal role by facilitating more frequent and timely interactions between patients and health care providers, thereby supporting patients in maintaining consistent and effective PD treatment.

Previous studies have suggested that RM-APD can lead to improved clinical outcomes compared with APD.^14–19 Reduced emergency visits and hospitalization rates, ¹⁶ better blood pressure control, ¹⁷ and improved dialysis adequacy ¹⁷ following RM-APD have also been reported. In a recent multicenter prospective study, RM-APD was associated with superior survival, ¹⁸ and a randomized controlled trial demonstrated lower mortality and hospitalization rates in patients receiving RM-APD. ¹⁹ Our findings extend this growing body of evidence by showing that RM-APD was associated with a marked reduction in hospitalization days compared with both CAPD and APD. Furthermore, inpatient health care costs were significantly lower in the RM-APD group than in the APD group.

We also assessed 1-year changes in dialysis adequacy, fluid status, and cardiac function. While between-group differences were generally nonsignificant, the reduction in OH/ECW% over 1 year was significantly greater in the RM-APD group than in the CAPD group, suggesting a potential benefit of RM-APD in volume management. Given that volume overload is a well-established predictor of cardiovascular events and mortality in dialysis populations, even modest improvements in fluid control may carry long-term clinical significance. This finding warrants confirmation in larger cohorts with longer follow-up.

The strengths of this study include the use of real-world cost data and a strict modality-stable cohort. Previous cost-effectiveness analyses have been simulation-based, ²⁰ whereas we evaluated actual hospitalization and outpatient costs in routine clinical practice. Additionally, inclusion of only patients who remained on the same PD modality for 1 year strengthened the validity of between-group comparisons by minimizing modality-switch bias.

Several limitations should be acknowledged. First, the sample size was relatively small, particularly for cost subgroup analyses, limiting statistical power for some comparisons. Second, as an observational study, residual confounding may persist despite multivariable adjustment, notably given the baseline imbalance in PG-SGA score across groups. Third, the study was conducted within a single national health care system, and the cost findings may not generalize to settings with different reimbursement structures or remote monitoring infrastructure. Fourth, the technology assessed was a single commercial RPM platform; outcomes with other systems may vary depending on data granularity, alert configuration, and clinical workflow integration. Finally, although the patient-reported component of the PG-SGA was assessed for nutritional self-evaluation, we did not measure telehealth-specific patient-reported outcomes such as treatment satisfaction, health-related quality of life, or perceived usability of the remote monitoring system, which are increasingly recognized as important endpoints in telehealth research and warrant evaluation in future work. In addition, modality assignment was not randomized; CAPD versus APD and the addition of remote monitoring were determined by shared clinical decision-making, which may have introduced selection or indication bias that adjustment cannot fully remove. Blood pressure and volume status were captured only at baseline and 1 year; because these parameters are regulated dynamically day to day, two-point measurements cannot capture the continuous monitoring-and-adjustment process that is the presumed mechanism of benefit of remote monitoring, and the corresponding findings (including the OH/ECW% comparison) should be regarded as hypothesis-generating. We also did not record operational process measures of the remote-monitoring workflow. These include the frequency and duration of data review, number of patient contacts, medication and target-weight adjustments, and time to correction of blood-pressure deviations, as well as the associated staff or operational costs. Our cost analysis therefore reflects patient-side health care costs only and does not capture the provider-side resource investment required to deliver remote monitoring. Finally, Sharesource treatment records were available only for RM-APD patients, so the cross-validation of the supply-based adherence metric could not be extended to the CAPD and APD groups, which have no equivalent electronic record of delivered treatments.

In summary, this study provides prospective real-world evidence that integration of RPM into APD improves treatment adherence, reduces hospitalization burden, and enhances volume control compared with conventional PD modalities. As home-based therapies and telehealth continue to expand, RM-APD exemplifies how digital infrastructure can be leveraged to deliver continuous, data-informed care for complex chronic conditions. Larger long-term studies and pragmatic trials are needed to confirm these benefits, evaluate cost-effectiveness across diverse health care systems, and inform the design of next-generation telehealth platforms for ESKD.

Authors’ Contributions

J.E.K.: Writing—original draft. G.-J.K.: Conceptualization and methodology. M.G.K. and J.J.C.: Data curation. E.J.C. and Y.J.K.: Supervision and writing—review and editing. All authors reviewed and approved the final article.

Ethical Considerations

This study was approved by the Integrated Institutional Review Board of Korea University Medical Center (IRB no. 2020GR0553), which oversees all three participating hospitals. Written informed consent was obtained from all participants prior to enrollment. The study was conducted in accordance with the Declaration of Helsinki.

Data Availability Statement

The data that support the findings of this study are not publicly available owing to privacy and ethical restrictions. Data are available from the corresponding author upon reasonable request and with permission from the Korea University Medical Center Institutional Review Board.

Abbreviations Used

ANCOVA

Analysis of covariance

ANOVA

Analysis of variance

APD

Automated peritoneal dialysis

BCM

Body composition monitoring

BMI

Body mass index

CAPD

Continuous ambulatory peritoneal dialysis

CGN

Chronic glomerulonephritis

CHF

Congestive heart failure

CI

Confidence interval

DBP

Diastolic blood pressure

DM

Diabetes mellitus

ECW

Extracellular water

ED

Emergency department

ESKD

End-stage kidney disease

HTN

Hypertension

IHD

Ischemic heart disease

IQR

Interquartile range

IRB

Institutional review board

IRR

Incidence rate ratio

KRW

Korean won

Kt/V

Dialysis adequacy index

OH

Overhydration

OH/ECW%

Overhydration to extracellular water ratio

PD

Peritoneal dialysis

PG-SGA

Patient-Generated Subjective Global Assessment

PKD

Polycystic kidney disease

RM-APD

Remote monitoring–assisted automated peritoneal dialysis

RPM

Remote patient monitoring

SBP

Systolic blood pressure

STROBE

Strengthening the Reporting of Observational Studies in Epidemiology

Cite this article as: Kim JE, Kim MG, Cha JJ, Cho EJ, Kwon YJ, Ko G-J (2026) Remote monitoring in automated peritoneal dialysis improves treatment adherence and reduces hospitalization burden: A prospective multicenter cohort study, Telemedicine Reports, 2026, 7, 26924366261483461, DOI: 10.1177/26924366261483461.