To guide the evidence summarization, specific questions were established based on the PIPOST model (17).

P (Population): Hospitalized adult cancer patients (≥18 years).

I (Intervention): Administration of parenteral nutrition support.

P (Professionals applying evidence): Clinicians, nutrition support dietitians, pharmacists, nurses, and other members of the multidisciplinary team.

O (Outcomes): Incidence of complications, laboratory test indicators, nutritional status, and quality-management indicators.

S (Setting): Oncology specialty hospitals, oncology wards of general hospitals, and parenteral nutrition compounding centers.

T (Type of evidence): Thematic evidence summaries (including clinical decision-making, practice recommendations, and evidence summaries), clinical practice guidelines, systematic reviews, expert consensus statements, and expert opinions.

Definition note: Nutrition support dietitians are defined according to ASPEN as professionals who perform individualized nutrition assessment, develop and implement nutrition care plans, monitor patient responses to nutrition therapy, and coordinate transitional or discontinuation plans for nutrition support (18, 19).

Inclusion criteria:

Exclusion criteria:

Using the “5S” evidence-pyramid model (20), the terms “cancer,” “nutrition,” and related keywords were used to search relevant Chinese- and English-language guideline websites, professional society websites, and the UpToDate in a top-down manner. Systems-level resources searched included UpToDate and BMJ Best Practice. Synthesized-summary and guideline sources searched included the National Guideline Clearinghouse (NGC), the Scottish Intercollegiate Guidelines Network (SIGN), the National Institute for Health and Care Excellence (NICE), the Guidelines International Network (GIN), the Registered Nurses’ Association of Ontario (RNAO), Medlive, the European Society for Clinical Nutrition and Metabolism (ESPEN), the American Society for Parenteral and Enteral Nutrition (ASPEN), the Chinese Society for Parenteral and Enteral Nutrition (CSPEN), and the Chinese Certified Dietitian. Synthesis databases searched included the Cochrane Library and the Joanna Briggs Institute (JBI) Library. Primary study databases searched included PubMed, Embase, Web of Science, and CINAHL. Chinese databases searched included China National Knowledge Infrastructure (CNKI), Wanfang Data, VIP, and the China Biology Medicine Database (CBM). The search period extended from each database’s inception to 13 August 2025. The PubMed search strategy is detailed in the Supplementary material S1.

To ensure the reliability and interpretability of the evidence synthesis, quality-appraisal tools appropriate to each type of included literature were applied. The included evidence types comprised clinical decision resources, clinical practice guidelines, expert consensus statements, and systematic reviews. The appraisal methods were as follows: (1) Clinical decision resources: Because no universally accepted instrument exists for appraising system-level clinical decision–support content, items obtained from authoritative clinical decision platforms or system databases were provisionally classified as high-quality clinical decision evidence, and source provenance and last-update date were recorded (21). (2) Clinical practice guidelines: The Appraisal of Guidelines for Research and Evaluation (AGREE II) (Table 1) (22) instrument was used. AGREE II comprises six domains and 23 items; each item is scored from 1 to 7 (1 = strongly disagree; 7 = strongly agree). Domain scores were calculated and used to generate an overall assessment. Guidelines were classified as grade A (all six domain scores ≥60%), grade B (≥3 domains scoring 30–60%), or grade C (>3 domains scoring <30%). (3) Expert consensus (Supplementary Table S2) and systematic reviews (Supplementary Table S3) were evaluated using the criteria of the Authentic Assessment Tool (2016) (23) from the Australian JBI Center for Evidence-Based Health Care.

Two reviewers trained in evidence-based methods independently and blind to each other’s assessments appraised each included document, and item-level scores and principal methodological limitations were recorded. Disagreements were resolved through discussion; if consensus was not reached, a third reviewer with methodological expertise adjudicated. All final quality assessments and the rationale for judgments were documented.

Two reviewers trained in evidence-based methods independently extracted and summarized evidence from the included documents. The following rules guided extraction and synthesis: retain original wording for evidence items that are independently stated; split multi-element statements into discrete items; when multiple sources report essentially the same finding, adopt the clearest and most concise wording; merge complementary findings according to logical relations; for conflicting evidence, trace back to sources and prioritize the evidence with higher methodological quality and more recent publication date, documenting the rationale for selection. Original citation numbers are retained for traceability. Evidence was regraded using the Joanna Briggs Institute (JBI) Levels of Evidence. Discrepancies in extraction or grading were resolved by discussion between the two reviewers, with a third reviewer adjudicating unresolved disagreements.

A total of 2,248 records were identified through database and website searches. After deduplication with EndNote version 20.0, 2,005 unique records remained. Two reviewers (BL and NW) independently screened titles and abstracts, excluding 1,929 records judged irrelevant and leaving 76 articles for full-text assessment. Upon full-text review, 58 articles were excluded: 44 were not relevant to the topic, 9 were updated versions of existing guidelines or consensus statements, and 5 were guideline interpretations or commentaries. A total of 18 articles met the inclusion criteria and were retained for evidence extraction and synthesis. These comprised one systems-level clinical decision resource, four clinical practice guidelines, nine expert consensus statements, and four systematic reviews. The literature selection process is shown in the PRISMA flow diagram (Figure 1).

The main characteristics of the included studies are summarized in Table 2.

After extraction, comprehensive synthesis, and analysis of evidence regarding the rational use of parenteral nutrition in hospitalized oncology patients, five key domains were identified: individualized nutritional decision-making; PN prescription and review; PN preparation and compounding; PN infusion and monitoring; and safety assessment and quality management. In total, 46 evidence items were identified (Table 3).

In hospitalized oncology patients, nutritional screening followed by comprehensive assessment remains essential for determining the appropriate timing of PN (39). Validated tools such as NRS-2002 and PG-SGA should be applied at diagnosis and upon admission, with reassessment triggered by clinical instability. Abnormal screening results require multidisciplinary evaluation of dietary intake, body-composition changes, metabolic markers, and refeeding-syndrome risk, owing to the highly dynamic nature of cancer trajectories. Differences among current guidelines indicate variability in operational thresholds for PN initiation. For example, CSPEN recommends initiating supplemental PN in high-risk patients (e.g., NRS-2002 ≥5 or NUTRIC ≥6) when enteral nutrition fails to provide ≥60% of required energy and protein within 48–72 h (24). ASPEN similarly advises initiating PN within 3–5 days in nutritionally at-risk patients unlikely to achieve sufficient oral intake or EN, while recommending deferring PN in cases of severe metabolic instability until clinical stabilization is achieved (9).

Although a retrospective cohort study suggested that early PN initiation may benefit patients with moderate to severe malnutrition (40), decisions regarding PN in advanced cancer should also incorporate expected survival, functional status, and patient preferences, as emphasized by multiple expert consensus statements (9, 26, 27, 34, 35). These differences reflect variation in both the certainty of the supporting evidence and the clinical priorities across regions. Implementation challenges persist, including limited availability of Nutrition Support Teams (NST), inconsistent training in PG-SGA and GLIM, and inadequate resources to support serial reassessment. These constraints diminish the reliability and generalizability of guideline-based PN initiation. Establishing structured screening–assessment pathways, ensuring NST oversight, and standardizing reassessment intervals may improve the appropriateness and consistency of PN initiation.

Parenteral nutrition formulations for oncology patients must be individualized to meet metabolic demands while minimizing metabolic burden. Energy targets for adult patients generally range from 25 to 30 kcal/kg/day, and protein requirements typically fall between 1.0 and 1.5 g/kg/day, with higher amounts indicated in hypercatabolic states (27, 41). Appropriate macronutrient distribution and strict control of glucose infusion rates (42) are essential to prevent metabolic complications such as hyperglycemia, while the selection of lipid emulsions (43–45) should reflect metabolic and immunological status and be adjusted in cases of hypertriglyceridemia. Compared with fixed formula patterns, adopting a personalized formulation strategy is critical, incorporating tumor type, disease trajectory, treatment tolerance, organ function, and dynamic metabolic status. This individualized approach highlights the need for stronger empirical evidence, as many current formulation recommendations are grounded in expert consensus rather than high-quality comparative trials. More robust RCTs evaluating formulation components in specific cancer subtypes would substantially strengthen the evidence base.

PN advancement should follow a gradual, risk-stratified escalation strategy, particularly for patients at high risk of refeeding syndrome (RFS), where conservative initiation and intensified metabolic monitoring are required (25–27). Studies have shown that the energy administered on the first day of PN is independently associated with the development of RFS (46, 47). Effective implementation of individualized PN requires coordinated multidisciplinary collaboration. Daily documentation, reassessment of prescribed goals versus actual infusion, and continuous metabolic evaluation are essential to guide timely adjustments. In practice, limited capacity for routine monitoring of triglycerides, electrolytes, and liver function restricts metabolism-guided titration, while insufficient NST staffing and fragmented workflow integration impede effective interdisciplinary collaboration. Establishing standardized procedures under NST supervision, together with risk-stratified advancement strategies, may enhance safety, reduce variability, and support the translation of evidence into clinical practice (32, 36, 37). Optimizing collaboration among multidisciplinary team members and implementing an NST-led prescription review system may further strengthen consistency and promote evidence-based PN formulation.

The preparation of parenteral nutrition is a high-risk compounding procedure that should be performed within centralized, standardized processes under pharmacist supervision to prevent microbial contamination and physicochemical incompatibilities; any breach of aseptic technique can result in contamination and increase the risk of serious infection (48). Whenever possible, manual compounding should be conducted in an institutional IV drug-admixture service (e.g., Pharmacy Intravenous Admixture Service (PIVAS)) within a Class B (ISO 5) laminar-flow hood or equivalent USP Chapter <797>−compliant sterile compounding environment (30, 49). The central principle is that safe PN preparation depends on robust environmental control, validated aseptic technique, and consistently applied compounding workflows (28). Pharmacist verification is essential, given the complexity of PN electrolyte–nutrient matrices and the high risk of calcium–phosphate precipitation (50–52). In clinical practice, PN admixtures may be used as a vehicle for drug infusion to reduce the patient’s fluid burden (14, 30, 53). However, the multi-component nature of PN makes compatibility unpredictable, and any medication addition must undergo formal compatibility and stability assessment with pharmacist approval (8, 29, 32).

Growing evidence (48, 54) shows that pharmacist-led services improve patient safety and reduce healthcare expenses, and ASPEN recommends pharmacist engagement in nutrition-support safety and quality-improvement programs (55). Pharmacists are positioned to identify root causes of compounding errors, develop targeted corrective strategies, and coordinate interdisciplinary execution. Given these responsibilities, optimizing PN compounding practices requires not only adherence to technical standards but also organizational systems that support pharmacist oversight and team-based coordination. To enhance safety and scalability, institutions should establish standardized compounding Standard Operating Procedures (SOPs), strengthen pharmacist-led prescription review, integrate intelligent verification technologies (e.g., barcode/RFID checks, automated compatibility databases), reinforce NST collaboration, and accelerate PIVAS capacity development (56). The transition toward smart compounding centers that integrate automation, robotics, and Internet-of-Things (IoT) technologies may further reduce error rates and improve operational efficiency, thereby enhancing the overall safety of PN preparation.

The correct choice of vascular access is a prerequisite for the safe infusion of PN. Although guidelines commonly recommend central venous access (CVAD) for prolonged or hyperosmolar formulations (8, 9, 25, 28, 57), device selection should be individualized by considering anticipated PN duration, lumen requirements, vascular status, and the patient’s oncologic treatment plan. Implantable ports may be advantageous for oncology patients who require both PN and chemotherapy, as they eliminate the need for additional catheter placement (8, 9, 24, 34, 57). When selecting catheters, the principle of using the fewest lumens necessary should be followed (9, 24, 57); however, further research is required to clarify and confirm whether reserving a dedicated PN lumen influences infection risk.

The standardized use of infusion devices forms another essential safeguard for PN safety. ASPEN now recommends a single 1.2-micron filter for total nutrient admixtures (TNA), dextrose–amino acid admixtures, and lipid emulsions (28, 30, 32, 58). Regular replacement of tubing and filters has been shown to reduce catheter-related bloodstream infection (13, 58), which highlights the importance of consistent device-management protocols. Nurses, as the primary providers of PN at the bedside, play a central role throughout the infusion process by maintaining aseptic technique, assessing catheter sites, detecting early signs of infectious or metabolic complications, verifying pump parameters, managing infusion equipment, educating patients, and communicating promptly with pharmacists, dietitians, and physicians (59). This continuous, coordinated oversight is critical for translating technical standards into safe clinical practice.

The mode and rate of PN infusion should be tailored to the patient’s tolerance (24, 25, 28, 32). Continuous infusion is recommended for critically ill or acutely unstable patients, with gradual titration to target rates over 24–48 h (25, 28, 57). Cyclic PN demonstrates biologically plausible hepatoprotective effects in animal studies (60–62); however, high-quality clinical evidence remains limited. Systematic monitoring of electrolytes, triglycerides, and liver function is essential for dynamic adjustment of PN composition and infusion rates (25, 28, 57). To promote consistent implementation, institutions should establish standardized infusion pathways, strengthen interdisciplinary coordination, and adopt smart-pump technologies to reduce manual variation. Structured training in aseptic technique, vascular-access management, complication recognition, and PN infusion protocols is essential for maintaining personnel competency and ensuring adherence to safety standards.

The 2024 International PN Safety and Quality Summit Consensus pointed out that PN quality management remains uneven across regulatory and operational levels, with substantial cross-country variation that directly affects patient safety and therapeutic consistency (63). The rational and safe implementation of parenteral nutrition requires integrating objective safety indicators, patient-centered outcomes, technological safeguards, and multidisciplinary governance into a structured continuous quality improvement (CQI) framework (63, 64). Core indicators—including catheter-related bloodstream infection (CRBSI), intestinal failure–associated liver disease (IFALD), refeeding syndrome (RFS), infusion reliability, and metabolic control—provide healthcare organizations with clear benchmarks for evaluating effectiveness and identifying areas requiring improvement (9, 28, 32, 33). In addition, a systematic review (38) reported that PN catheters may provoke anxiety and concerns about body image, underscoring the need to incorporate patient-experience and quality-of-life outcomes, particularly for individuals who rely on long-term PN.

At the operational level, standardized prescription-review tools, infusion checklists, smart pumps, barcode systems, and automated alerts contribute to more consistent and reliable PN delivery (28–30, 32). To strengthen safety oversight, institutions should implement transparent error-reporting systems capable of capturing both adverse events and near-misses, supported by regular NST-led reviews. Routine audits, PDSA cycles, and medication-use evaluations can facilitate root-cause analyses and guide iterative improvement (65–67). Modern digital tools—including automated compatibility checks, real-time dashboards, and electronic PN quality monitoring—may further reduce variability. By embedding PN care within a mature CQI framework that links safety metrics, patient experience, and system-level learning, healthcare organizations can establish a sustainable infrastructure that reduces preventable harm and supports better long-term outcomes.

To further illustrate how the five evidence-based domains operate within an end-to-end clinical workflow, we developed a visual schematic (Figure 2) that links each domain to the sequential clinical steps from nutritional screening to quality improvement. This diagram supports practical implementation by showing how decision-making, prescription, compounding, infusion management, and safety monitoring are integrated across the PN care pathway.

Despite providing consolidated evidence on PN use in hospitalized oncology patients, this study has certain limitations. First, restricting the literature search to English and Chinese publications may introduce language bias. Second, the applicability of the synthesized framework may vary across institutions due to differences in resources, staff competencies, PN infrastructure, and oncology care models, potentially contributing to applicability bias. In addition, evidence for specific tumor subgroups and complex clinical scenarios remains limited, which may affect the generalizability of some recommendations. Future updates should expand database coverage, include non-English sources, and incorporate real-world clinical data to enhance external validity.