Factors Influencing Unplanned Discontinuation of Continuous Renal Replacement Therapy: A Descriptive Time‐Series Analysis

ABSTRACT

Background

Continuous renal replacement therapy (CRRT) is the preferred renal replacement therapy for critically ill patients with acute kidney injury; however, unplanned treatment discontinuation remains a major challenge that compromises treatment efficacy and patient outcomes.

Aims

To investigate the risk factors and temporal trends associated with unplanned CRRT discontinuation using descriptive time‐series analysis and to identify evidence‐based nursing interventions.

Study Design

This retrospective cohort study was conducted at a tertiary hospital in China between January and December 2025. Univariate analysis and binary logistic regression were used to identify factors independently associated with unplanned discontinuation. Descriptive time‐series analysis with 7‐day moving average smoothing was applied to evaluate temporal patterns throughout the study period.

Results

Unplanned discontinuation occurred in 77 of 3128 CRRT sessions (2.46%). After adjustment, anticoagulation regimen, treatment duration (p < 0.001) and net ultrafiltration rate (p = 0.032) were identified as independent predictors of unplanned discontinuation. Compared with no anticoagulation, nafamostat mesylate was associated with lower odds of unplanned CRRT discontinuation (p = 0.013), whereas sodium citrate was associated with higher odds (p = 0.004), likely reflecting centre‐specific implementation factors related to low citrate utilization. Descriptive time‐series analysis showed temporal fluctuations in unplanned discontinuation rates, with an apparent peak in November. Elevated circuit pressure was the leading cause of unplanned discontinuation.

Conclusions

Unplanned CRRT discontinuation was associated with anticoagulation regimen, treatment duration and ultrafiltration rate. Standardized anticoagulation protocols, staff training, safe ultrafiltration practices and enhanced circuit pressure surveillance may reduce unplanned discontinuation and improve patient outcomes.

Relevance to Clinical Practice

These findings provide nurses with evidence‐based strategies, including locally informed monitoring protocols, standardized anticoagulation checklists and circuit pressure early warning indicators, to reduce unplanned CRRT discontinuation.

Impact Statements

• What is known about the topic

  • ○
    Unplanned Continuous renal replacement therapy (CRRT) discontinuation due to circuit clotting occurs in 17%–74% of sessions and compromises treatment efficacy.
  • ○
    Regional citrate anticoagulation is recommended as the first‐line treatment according to the KDIGO guidelines based on randomized controlled trial (RCT) evidence of prolonged filter lifespans.
  • ○
    Nafamostat mesylate has shown efficacy similar to that of heparin, with a lower risk of bleeding in high‐risk patients.
• What this paper adds

  • ○
    Descriptive time‐series analysis revealed temporal fluctuations in unplanned CRRT discontinuation rates, with an apparent peak in November; however, the clinical significance of this pattern remains uncertain and requires further prospective validation.
  • ○
    At this centre, where citrate utilization was low, citrate anticoagulation was associated with higher odds of unplanned CRRT discontinuation than no anticoagulation (OR = 3.456, 95% CI: 1.466–7.949, p = 0.004). This unexpected association may reflect limited staff experience and centre‐specific challenges in implementing the citrate protocol, rather than an inherent lack of efficacy of citrate anticoagulation. These findings highlight the importance of standardized protocols, comprehensive staff training and regular competency assessment.
  • ○
    A three‐tiered nursing intervention bundle (organizational, educational and individualized) was used to reduce unplanned discontinuation rates.

1. Introduction

Continuous renal replacement therapy (CRRT) is the standard renal replacement therapy for critically ill patients with acute kidney injury in intensive care units worldwide. Unlike intermittent haemodialysis, CRRT provides continuous solute clearance and fluid removal, improving haemodynamic stability and volume control in patients with circulatory instability. The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines recommend CRRT for haemodynamically unstable patients [1].

Despite its clinical benefits, CRRT is frequently complicated by unplanned discontinuation, defined as premature therapy termination due to filter clotting, circuit pressure elevation, mechanical failure or haemodynamic instability, preventing achievement of prescribed treatment goals [2, 3]. A recent systematic review and meta‐analysis reported rates ranging from 17% to 74% [4]. This variability highlights the multifactorial nature of unplanned discontinuation and the need for context‐specific risk assessment. Consequences include blood loss, increased treatment costs, interrupted solute clearance with potential underdosing and greater nursing workload. Repeated circuit clotting may also require blood transfusion and has been associated with increased mortality [4].

Nurses play a central role in CRRT circuit management. Their responsibilities include operating CRRT devices, monitoring circuit pressures, assessing clinical parameters such as haemoglobin, lactate and Sequential Organ Failure Assessment scores, and applying predictive models to guide individualized treatment [5].

1.1. Background/Justification for the Study

Among the modifiable factors affecting circuit longevity, anticoagulation strategy has received the greatest attention. International guidelines and meta‐analyses of randomized trials recommend regional citrate anticoagulation (RCA) for most patients, whereas nafamostat mesylate is a safe alternative in selected populations [6, 7, 8, 9, 10, 11, 12]. However, anticoagulation effectiveness depends not only on pharmacological properties but also on protocol adherence, staff training and local experience, which vary considerably across centres.

Previous studies primarily used cross‐sectional or short‐term cohort designs with limited follow‐up. Consequently, temporal patterns of unplanned discontinuation remain unclear, including monthly or seasonal variation and associations with patient or environmental factors. Identifying high‐risk periods could guide targeted prevention and optimize nursing resource allocation. In addition, filtration fraction—the ratio of ultrafiltration rate to plasma water flow rate, typically maintained below 20%–25% to reduce haemoconcentration and circuit clotting—has rarely been examined in temporal studies of CRRT circuit longevity.

Time‐series analysis provides an effective approach for evaluating temporal trends, identifying systematic patterns and visualizing outcome fluctuations over time [13]. In this study, descriptive time‐series analysis assessed temporal variation in unplanned discontinuation rates over 12 months without assuming a predefined intervention point.

1.2. Aims

This study aimed to characterize temporal trends in unplanned CRRT discontinuation over 12 months using descriptive time‐series analysis, identify independent patient‐ and treatment‐related factors, particularly anticoagulation regimens, examine the distribution of documented causes and develop evidence‐based nursing recommendations for CRRT circuit management.

2. Design and Methods

A retrospective cohort study design was conducted.

2.1. Setting and Sample

All patients who underwent continuous renal replacement therapy at a tertiary hospital in Beijing, China, between 15 January 2025 and 15 December 2025 were screened. Inclusion criteria were: (1) receipt of CRRT at this institution; (2) a hospital stay of at least 3 days; and (3) an age of 18 years or older. Exclusion criteria were: (1) severely incomplete clinical data preventing outcome assessment; (2) concurrent extracorporeal membrane oxygenation; and (3) baseline severe coagulation abnormalities or disseminated intravascular coagulation.

2.2. Data Collection Tools and Methods

Patient data were extracted from electronic medical records by two trained research nurses using a standardized data abstraction form. Discrepancies were resolved by consensus with a third investigator. Collected variables included demographic characteristics (age, sex), vascular access type (native arteriovenous fistula, arteriovenous graft or central venous catheter), treatment date (to determine monthly and seasonal patterns), anticoagulation regimen (none, low‐molecular‐weight heparin, argatroban, sodium citrate, nafamostat mesylate or a combined regimen of 2–3 h heparin‐free plus nafamostat mesylate), single‐treatment duration (hours) and net ultrafiltration volume (L/session). The primary outcome was unplanned discontinuation, with the documented primary reason recorded for each event. Data were entered into a secure database with range and logic checks to minimize entry errors. Missing or ambiguous data were verified against the original medical records by a second investigator.

2.3. Assessment of Unplanned CRRT Discontinuation

Unplanned CRRT discontinuation was defined as premature termination of a CRRT session before achieving the prescribed treatment goal or planned duration [2, 3]. Bedside nurses documented each event using a standardized assessment protocol, and the research team verified it based on: (1) transmembrane pressure (TMP) > 300 mmHg; (2) visible clot formation obstructing blood flow; (3) blood pump malfunction or pressure alarm requiring circuit termination; or (4) Grade II or higher filter clotting according to the Blood Purification Standard Operating Procedure [14]. The grading system comprised Grade 0, no visible clotting or minimal fibrin deposition at the filter ends; Grade I, < 10% of hollow fibres clotted; Grade II, 10%–50% clotted; and Grade III, > 50% clotted. Grade II or higher clotting was considered unplanned discontinuation because circuit function was clinically compromised. Other documented reasons for premature termination were also recorded. Each event was further classified by primary cause based on nursing documentation and machine alarm records.

2.4. Data Analysis

Data were analysed using SPSS 29.0 and R 4.4.1. The normality of continuous variables was assessed using the Shapiro–Wilk test. Normally distributed variables are presented as the mean and standard deviation and were compared using the independent‐samples t‐test. Non‐normally distributed variables are presented as median (P25, P75) and compared using the Mann–Whitney U test. Categorical variables are presented as n (%) and compared using the chi‐square (χ²) or Fisher's exact test. Variables with p < 0.05 in univariate analysis were entered into a binary logistic regression model. Because temporal variation was descriptive and non‐linear, treatment month was excluded from the model and evaluated only by descriptive time‐series analysis. The final model included three parameters (anticoagulation regimen, treatment duration and net ultrafiltration rate), yielding an event‐per‐variable ratio of approximately 12.8, which was acceptable.

For time‐series visualization, daily unplanned CRRT discontinuation rates were calculated as the number of unplanned CRRT discontinuations divided by the total daily treatment sessions × 100%. A descriptive time‐series plot was generated to illustrate fluctuations over 12 months. A 7‐day moving average smoothed short‐term variability and highlighted sustained trends. No intervention point was defined, and the analysis was not intended to test causal hypotheses. Therefore, the results are presented solely as descriptive observations. All tests were two‐tailed, and statistical significance was set at p < 0.05.

2.5. Ethical and Institutional Approvals

This study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the participating tertiary hospital in Beijing, China (approval number: KY2023‐289‐01; approval date: 25 October 2023). Informed consent was waived because of the retrospective study design, and all patient data were anonymized before analysis.

3. Results

3.1. Patient Characteristics and Univariate Analysis

A total of 3128 CRRT sessions from 741 patients were included, of which 77 (2.46%) ended in unplanned CRRT discontinuation. Univariate analysis showed significant differences between groups in treatment month, anticoagulation regimen, treatment duration and net ultrafiltration volume (p < 0.05; Table 1).

TABLE 1.

Univariate analysis of factors associated with unplanned CRRT discontinuation.

Variable	Planned CRRT Discontinuation Group (n = 3051)	Unplanned CRRT Discontinuation Group (n = 77)	Statistic	p
Sex (n %)			χ2 = 0.904	0.346
Male	1867 (61.19)	43 (55.84)
Female	1184 (38.81)	34 (44.16)
Age (years)	64.09 ± 15.16	65.48 ± 15.92	t = −0.794	0.427
Treatment Month (n %)			Fisher	< 0.001
January	366 (12.00)	0 (0)
February	273 (8.95)	2 (2.60)
March	279 (9.14)	13 (16.88)
April	310 (10.16)	13 (16.88)
May	257 (8.42)	7 (9.10)
June	241 (7.90)	5 (6.49)
July	295 (9.67)	3 (3.90)
August	195 (6.39)	6 (7.79)
September	260 (8.52)	7 (9.10)
October	215 (7.05)	2 (2.60)
November	312 (10.23)	18 (23.38)
December	48 (1.57)	1 (1.30)
Vascular Access (n %)			Fisher	0.425
Native Arteriovenous Fistula	953 (31.24)	19 (24.68)
Arteriovenous Graft	10 (0.33)	0 (0)
Central Venous Catheter	2088 (68.44)	58 (75.32)
Anticoagulation (n %)			Fisher	< 0.001
No Anticoagulation	324 (10.62)	20 (25.97)
Low‐molecular‐weight heparin	232 (7.60)	4 (5.19)
Argatroban	82 (2.69)	2 (2.60)
Sodium citrate	183 (6.00)	12 (15.58)
Nafamostat mesylate	2216 (72.63)	39 (50.65)
2‐3 h Heparin‐free + Nafamostat	14 (0.46)	0 (0)
Single‐treatment duration (h)	7.80 ± 2.08	6.22 ± 2.74	t = 6.506	< 0.001
Net ultrafiltration volume (L/session)	1.67 ± 0.82	1.23 ± 0.84	t = 4.676	< 0.001

Abbreviations: χ², chi‐square statistic; CRRT, continuous renal replacement therapy; n, number; SD, standard deviation.

3.2. Logistic Regression Analysis of Factors Influencing Unplanned CRRT Discontinuation

Multivariable logistic regression identified anticoagulation regimen, single‐treatment duration (OR = 0.638, 95% CI: 0.563–0.720, p < 0.001) and net ultrafiltration rate (OR = 0.145, 95% CI: 0.023–0.789, p = 0.032) as independent factors associated with unplanned CRRT discontinuation (Table 2). Compared with no anticoagulation, sodium citrate was associated with higher odds of unplanned CRRT discontinuation (OR = 3.456, 95% CI: 1.466–7.949, p = 0.004), whereas nafamostat mesylate was associated with lower odds of unplanned CRRT discontinuation (OR = 0.481, 95% CI: 0.274–0.870, p = 0.013).

TABLE 2.

Logistic regression analysis of factors associated with unplanned CRRT discontinuation.

Variable	B	Std. error	Wald χ2	p	OR	95% CI
Intercept	0.309	0.451	0.47	0.493	1.362	0.553–3.256
Anticoagulation (ref = No anticoagulation)
Low‐molecular‐weight heparin	−0.904	0.562	2.59	0.107	0.405	0.115–1.106
Argatroban	−0.079	0.769	0.01	0.918	0.924	0.143–3.400
Sodium citrate	1.240	0.428	8.38	0.004	3.456	1.466–7.949
Nafamostat mesylate	−0.731	0.293	6.23	0.013	0.481	0.274–0.870
Single‐treatment duration (h)	−0.450	0.063	51.5	< 0.001	0.638	0.563–0.720
Net ultrafiltration rate (L/h)	−1.930	0.900	4.60	0.032	0.145	0.023–0.789
Intercept	0.309	0.451	0.47	0.493	1.362	0.553–3.256

Note: The combined heparin‐free plus nafamostat mesylate group (n = 14, no events) was merged with the nafamostat mesylate group.

Abbreviations: χ², chi‐square statistic; B, regression coefficient; CI, confidence interval; CRRT, continuous renal replacement therapy; OR, odds ratio; SE, standard error.

3.3. Time‐Series Analysis of Unplanned CRRT Discontinuation Rate

Descriptive time‐series analysis showed fluctuations in the daily unplanned CRRT discontinuation rate over 12 months. After 7‐day moving average smoothing, a distinct peak occurred in November (Figure 1). Monthly incidence rates are shown in Figure 2, with the highest rate in November (5.45%).

FIGURE 1.

Time series of unplanned CRRT discontinuation rate. The blue line represents the raw daily incidence (%), and the red line the 7‐day moving average. The x‐axis shows dates from January to December 2025. The moving average peaks around November.

FIGURE 2.

Monthly unplanned CRRT discontinuation rate: Bars represent the monthly percentage of unplanned CRRT discontinuation. The highest rate occurred in November (5.45%). Months without bars (e.g., January) indicate no events.

3.4. Distribution of Unplanned CRRT Discontinuation Causes

Among 77 unplanned CRRT discontinuation events, elevated circuit pressure was the leading cause (44.17%). In these cases, pre‐filter pressure elevation often preceded TMP elevation, suggesting that pre‐filter pressure may be an early indicator of circuit coagulation risk (Figure 3).

FIGURE 3.

Causes of unplanned CRRT discontinuation.

4. Discussion

We analysed 3128 CRRT sessions. Descriptive time‐series analysis examined temporal fluctuations in unplanned CRRT discontinuation rates, and multivariable regression evaluated the independent associations of anticoagulation regimen, treatment duration and net ultrafiltration volume. The main findings are discussed below.

4.1. Temporal Variation in Unplanned CRRT Discontinuation Rates

Descriptive time‐series analysis showed temporal fluctuations in unplanned CRRT discontinuation rates, with a peak in November. Because no predefined intervention point was included, interrupted time‐series analysis was not performed. These monthly variations should be interpreted cautiously because few unplanned CRRT discontinuation events occurred and environmental or biological factors underlying seasonal variation were not measured. Prospective studies incorporating temperature, infection burden, coagulation markers and staffing factors are needed to determine whether these patterns are reproducible and clinically meaningful.

Although the November peak requires prospective validation, nursing administrators may consider preventive measures during locally identified high‐risk periods. These include more frequent circuit pressure monitoring, assessment of anticoagulation adequacy and staff education. Such measures are consistent with best‐practice recommendations for CRRT circuit management [3].

4.2. Anticoagulation Regimen and Unplanned CRRT Discontinuation: Interpreting the Citrate Finding

In the final multivariable model, sodium citrate was associated with a higher odds of unplanned CRRT discontinuation than no anticoagulation. This finding should be interpreted cautiously and does not indicate that citrate anticoagulation is inherently inferior or harmful. The KDIGO guidelines recommend RCA as the first‐line strategy for CRRT based on robust randomized controlled trial (RCT) evidence. A network meta‐analysis by Zhou et al. [6] (37 RCTs; 2648 patients) found that RCA prolonged filter lifespan by 12.0 h and reduced bleeding risk compared with unfractionated heparin. Likewise, Li et al. [7] reported citrate was superior to heparin, and Bhalla et al. [9] found longer mean filter lifespan with citrate than heparin (45.11 vs. 26.11 h). Our findings do not contradict this evidence but likely reflect centre‐specific implementation factors.

RCA is a complex protocol requiring precise citrate dosing based on pre‐filter ionized calcium, appropriate calcium supplementation to maintain systemic ionized calcium, regular monitoring of post‐filter ionized calcium (target, 0.25–0.35 mmol/L), acid–base management because citrate metabolism generates bicarbonate and adjustments for patient factors such as liver function [15, 16]. In a study conducted in the United Kingdom, Kirwan et al. [16] reported that simplified citrate protocols prolonged filter lifespan and reduced protocol deviations, highlighting the importance of standardized implementation. At this centre, citrate was used in only 6.0% of sessions versus 72.6% for nafamostat mesylate. This limited use may have reduced staff familiarity, increasing the likelihood of suboptimal citrate titration, delayed calcium supplementation or unrecognized post‐filter ionized calcium fluctuations, which could increase coagulation risk despite citrate efficacy. A recent cross‐sectional study found that inadequate nursing knowledge of CRRT interruption management may compromise treatment efficacy and safety, supporting targeted training programmes [17]. Furthermore, the retrospective design prevented assessment of liver function; unrecognized citrate accumulation, metabolic alkalosis and hypocalcaemia in patients with hepatic impairment can all impair circuit function [15].

Therefore, the association between sodium citrate and increased unplanned CRRT discontinuation observed in this study should be considered a quality improvement signal rather than evidence of comparative efficacy. These findings emphasize the importance of standardized protocols, mandatory staff training with regular competency assessment and audits of anticoagulation protocol adherence when implementing RCA.

In contrast, nafamostat mesylate was independently associated with lower odds of unplanned CRRT discontinuation (OR = 0.481, 95% CI: 0.274–0.870, p = 0.013), consistent with previous evidence. A systematic review and meta‐analysis by Wang et al. [11] found that nafamostat and heparin had comparable filter lifespan and bleeding risk, with nafamostat potentially reducing hospital stay. Likewise, an RCT by Choi et al. [12] in patients at high bleeding risk reported longer filter lifespan with nafamostat (31.7 ± 24.1 vs. 19.5 ± 14.9 h) and a filter failure hazard ratio of 0.578. Nafamostat's ultra‐short half‐life (approximately 8 min) allows rapid loss of anticoagulant activity after discontinuation, making it suitable for patients at high bleeding risk [10]. Its high use at this centre may reflect both its favourable safety profile and accumulated local experience.

4.3. Treatment Duration and Ultrafiltration Rate: Interpretation Within the Context of Treatment Adequacy

Longer single‐treatment duration was associated with lower odds of unplanned CRRT discontinuation (OR = 0.638, 95% CI: 0.563–0.720, p < 0.001). However, because treatment duration is measured at session completion, shorter durations often reflect early circuit failure rather than a baseline predictor. Thus, this association primarily indicates that uncomplicated sessions were completed successfully.

After adjustment for treatment duration and other covariates, hourly net ultrafiltration rate was associated with lower odds of unplanned CRRT discontinuation (OR = 0.145, p = 0.032). This finding is counterintuitive because consistently higher ultrafiltration rates would be expected to increase filtration fraction and clotting risk [18]. A more likely explanation is that patients receiving higher hourly ultrafiltration were haemodynamically stable and had well‐functioning vascular access, enabling successful treatment completion. Therefore, the association likely reflects patient selection rather than a direct effect of ultrafiltration rate. Accordingly, ultrafiltration rate should not be interpreted as a modifiable factor that directly reduces unplanned CRRT discontinuation. Clinically, filtration fraction should remain below 20%–25% to minimize haemoconcentration [19, 20]. In our cohort, total ultrafiltration volume was only weakly correlated with treatment duration, indicating that these variables were largely independent.

4.4. Elevated Circuit Pressure: The Most Direct Technical Cause and Monitoring Strategies

Among the 77 unplanned CRRT discontinuation events, elevated circuit pressure was the leading cause (44.17%), consistent with Mateos‐Dávila et al. [21], who reported that pre‐filter pressure is a more reliable early warning indicator than TMP because TMP often rises too late for safe blood return. Filter coagulation accounted for 25.97% of events, and blood return was usually unsuccessful, resulting in iatrogenic blood loss.

Based on these findings, we recommend the following circuit pressure monitoring protocol for nursing staff:

1. Routine assessment: Record pre‐filter pressure, TMP and venous pressure hourly in the nursing flow sheet.

2. Threshold establishment: Establish centre‐specific alarm thresholds using the reference graphs reported by Mateos‐Dávila et al. [21], which provided risk scales applicable to all CRRT machines.

3. Early warning detection: Monitor for progressive pressure elevation without an identifiable cause (e.g., patient repositioning or tubing kinking). Pre‐filter pressure often rises 30–60 min before TMP, providing an opportunity for early intervention.

4. Troubleshooting: For patients with central venous catheters and abnormal arterial or venous pressure alarms, first reduce the blood flow rate. If ineffective, disconnect the circuit, switch to self‐circulation mode and reposition the catheter [22, 23].

5. Documentation: Record all pressure alarms and nursing interventions in the CRRT nursing record for quality improvement.

Blood loss during filter replacement increases iatrogenic anaemia and transfusion requirements; therefore, early recognition of pressure abnormalities is essential for patient safety.

5. Study Limitations

This study had several limitations. First, its retrospective single‐centre design limits generalisability, and all associations should be interpreted as correlational rather than causal. This is particularly relevant for treatment duration and hourly net ultrafiltration rate, which may reflect successful therapy continuation (survivor bias) rather than baseline predictors. Likewise, the unexpected citrate findings likely reflect centre‐specific implementation rather than inherent drug effects. Second, important confounders, including disease severity, platelet count, fibrinogen, D‐dimer and liver function, were unavailable because of the retrospective design. Third, some anticoagulation regimens had small sample sizes, limiting statistical power. Fourth, nursing factors, including nurse‐to‐patient ratio, CRRT experience and shift timing, were unavailable and may have influenced circuit pressure detection. Fifth, the time‐series analysis did not measure mediating factors, such as temperature or infection rates. Sixth, the time‐series analysis was descriptive only; month was excluded from the multivariable model because no linear trend was assumed. Therefore, temporal or seasonal effects cannot be inferred and the November peak should be interpreted cautiously. Finally, each CRRT session was analysed as an independent observation, although 3128 sessions were derived from 741 patients, introducing potential within‐patient correlation. Because only 77 unplanned discontinuation events occurred, mixed‐effects logistic regression and generalized estimating equations were unstable, failing to converge or producing unreliable variance estimates. Consequently, standard errors may be underestimated and confidence intervals overly narrow. Future multicentre prospective studies with larger samples should validate these findings using clustering‐adjusted analyses and incorporate nursing workforce factors.

6. Recommendations for Practice and Further Research

Level 1: Organizational interventions: Healthcare institutions should consider implementing locally informed monitoring protocols, increasing CRRT nurse staffing during high‐risk periods, establishing a standardized anticoagulation protocol with regular audits to support adherence to regional citrate anticoagulation, and developing a CRRT circuit pressure early warning system with clear escalation pathways.

Level 2: Educational interventions: Intensive care unit nurses should receive mandatory CRRT training, with competency reassessment every 6 months. Clinicians should also consider RCA simulation training to address common protocol deviations, and circuit pressure monitoring should be incorporated into daily nursing handovers.

Level 3: Individualized patient interventions: Validated prediction models should be used to guide individualized treatment; net ultrafiltration volume per session should be tracked as a quality indicator; and single‐treatment duration should be maintained at ≥ 6 h whenever clinically feasible.

Future multicentre prospective studies should incorporate haemostatic markers, develop risk prediction models and evaluate machine learning‐based real‐time early warning systems using nursing data. They should also assess the impact of nursing education programmes on unplanned CRRT discontinuation and the cost‐effectiveness of the proposed nursing intervention bundle.

7. Conclusion

Unplanned CRRT discontinuation was associated with anticoagulation regimen, treatment duration and ultrafiltration‐related factors. Descriptive time‐series analysis suggested temporal variation, with a peak in November; however, its clinical significance and underlying mechanisms remain uncertain. Prospective studies are needed to validate these findings and identify modifiable patient‐, treatment‐ and nursing‐related factors.

Author Contributions

Wanning Jia: conceptualization, data curation, formal analysis, writing – original draft. Hongjiao Xu: data curation, investigation, methodology. Ruihan Peng: data curation, investigation, validation. Haifeng Wang: formal analysis, software, visualization. Haitao Lu: methodology, supervision, writing – review and editing. Yingyi Bi: conceptualization, project administration, supervision, writing – review and editing.

Funding

The authors have nothing to report.

Ethics Statement

This retrospective study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of China‐Japan Friendship Hospital (Approval No.: KY2023‐289‐01) (Date: 25 October 2023). The requirement for informed consent was waived due to the retrospective nature of the study, and all patient data were anonymized.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.