Method of constructing evaluation and grading system of medical-related infection in neonatal intensive care unit by Delphi method

Abstract

The application discloses a method for constructing a neonatal intensive care unit medical related infection evaluation and grading system by using the Delphi method, and belongs to the technical field of medical quality management. The method comprises the following steps: establishing a multidisciplinary research team; determining candidate indexes through systematic literature retrieval; carefully selecting and recruiting experts; designing an expert consultation questionnaire containing outcome indexes, process indexes, balance indexes and sentinel indexes; carrying out two rounds of Delphi expert consultation, and organizing online expert meeting to discuss uncertain indexes during the two rounds of Delphi expert consultation; adopting SPSS software to analyze data, and calculating the authority coefficient of experts, the group median and the divergence index; and finally determining the NICU-Shield evaluation system, which contains 11 core indexes (3 outcome indexes, 6 process indexes and 2 balance indexes) and 1 sentinel index. The application constructs a quality evaluation and grading system specially for neonatal intensive care unit medical related infections.

Description

Technical Field

[0001] This invention belongs to the field of medical quality management technology, specifically involving a method for constructing a medical-related infection evaluation and grading system for neonatal intensive care units using the Delphi method. Background Technology

[0002] Hospital-associated infections (HAIs) are a critical challenge commonly faced by neonatal intensive care units (NICUs), posing a serious threat, especially to high-risk newborns such as very low birth weight infants (VLBWIs). These infections not only significantly increase morbidity and mortality rates but also lead to prolonged hospital stays, higher healthcare costs, and a sustained burden on long-term prognosis and healthcare resources.

[0003] To reduce the incidence of hospital-acquired infections, global prevention and control strategies include implementing hand hygiene, establishing dedicated central venous catheter care teams, implementing bundled care protocols, and strengthening antibiotic management. However, due to inconsistent definitions of hospital-acquired infections and differences in core evaluation indicators across studies, infection rates reported in different regions vary significantly. This severely limits the comparability of data and hinders cross-regional exchange of prevention and control experiences and collaborative quality improvement.

[0004] Currently available hospital-acquired infection (HAI) assessment systems are mostly general-purpose tools applicable to various medical institutions or general intensive care units (ICUs), but specialized assessment tools specifically for HAI prevention and control in neonatal intensive care units (NICUs) are still lacking. Existing HAI assessment systems lack specificity for the unique needs of NICUs, such as failing to consider the special physiological characteristics of extremely low birth weight infants, NICU-specific infection types (such as late-onset sepsis), and neonatal-specific clinical procedures (such as umbilical vein catheterization and peripherally inserted central venous catheterization).

[0005] Therefore, establishing a clinically feasible and widely recognized NICU infection quality assessment system is key to achieving accurate monitoring and continuous quality improvement.

[0006] Similar studies have been conducted, such as CN202410871609.X, which discloses a method for constructing evaluation indicators for breastfeeding in extremely premature infants in the intensive care unit. The Delphi method was used to construct breastfeeding quality evaluation indicators. However, this patent is aimed at breastfeeding quality, not hospital-acquired infection quality evaluation, and does not involve the setting of sentinel indicators or the construction of a grading system.

[0007] For example, CN202310827496.9 discloses a method for constructing quality evaluation indicators for nutritional management of nasopharyngeal carcinoma patients during the peri-radiotherapy period. The evaluation indicators are constructed using a three-dimensional quality model of "structure-process-outcome" as the theoretical framework. However, this patent is aimed at nutritional management of nasopharyngeal carcinoma patients, and its application field and theoretical framework are different from those of this invention. Furthermore, it is not specifically designed for the special characteristics of hospital-acquired infections in NICUs.

[0008] In conclusion, there is an urgent need to develop a standardized quality assessment and grading system based on the Delphi method specifically for medical-related infections in neonatal intensive care units. Summary of the Invention

[0009] To address the lack of specialized quality assessment tools for healthcare-associated infections (HAIs) in existing technologies, this invention provides a method for constructing an evaluation and grading system for HAIs in neonatal intensive care units (NIHs) using the Delphi method. The HAI evaluation and grading system constructed in this invention establishes a consensus-based standardized HAI quality assessment framework for NHIs, aiming to improve data comparability among institutions and support continuous quality improvement.

[0010] The technical solution of this invention is as follows:

[0011] This invention presents a method for constructing a medical-related infection assessment and grading system for neonatal intensive care units (NICUs) using the Delphi method. Through the Delphi expert consultation approach, it constructs the "NICU-Shield System for Assessment and Grading of Hospital-Acquired Infections in Neonatal Intensive Care Units." This system focuses on NICU clinical practice, screening evaluation indicators with high correlation and feasibility to infection control quality, and standardizing the definition and measurement criteria of these indicators. This establishes a specialized and comparable benchmark for NICU infection control quality control. The system aims to improve the consistency and comparability of data between institutions, providing support for cross-sectional performance comparisons, longitudinal quality monitoring, and the promotion of high-level clinical research.

[0012] This method includes: establishing a multidisciplinary research team; identifying candidate indicators through systematic literature retrieval; carefully selecting and recruiting experts; designing an expert questionnaire including outcome indicators, process indicators, balance indicators, and sentinel indicators; conducting two rounds of Delphi expert consultations, during which online expert meetings were organized to discuss uncertain indicators; using SPSS software for data analysis to calculate expert authority coefficients, group medians, and dissent indices; and finally determining the NICU-Shield evaluation system, which includes 11 core indicators (3 outcome indicators, 6 process indicators, and 2 balance indicators) and 1 sentinel indicator. This invention constructs a quality assessment and grading system specifically for medical-related infections in neonatal intensive care units. It features a rigorous methodology, a comprehensive indicator system, high expert authority, and strong clinical operability, providing a standardized framework for improving data comparability among institutions and supporting continuous quality improvement.

[0013] The objective of this invention is achieved through the following technical solution:

[0014] The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method includes the following steps:

[0015] S1. Establish a multidisciplinary research team:

[0016] The research team consists of more than 10 members, including the director of the neonatal department, the neonatal nursing manager, the director of the hospital infection control department, the clinical microbiology expert, the second-line doctor of the neonatal intensive care unit, the part-time doctor of the neonatal intensive care unit infection control management, and the specialist nurse of the neonatal intensive care unit.

[0017] The research team members divided the work and were responsible for searching, reading and analyzing literature, developing expert inquiry questionnaires, selecting experts for inquiry and conducting expert inquiries, compiling and analyzing the feedback results of each round of expert inquiries, and determining whether indicators need to be deleted, modified or added.

[0018] S2. Determine candidate indicators:

[0019] Candidate indicators were determined through systematic literature retrieval, including databases such as PubMed, Web of Science, and Embase.

[0020] Search terms included “infant, newborn”, “healthcare-associated infection”, “Quality Improvement”, and “Quality Indicators, Health Care”.

[0021] The search was limited to original research articles published in English and conducted independently by two skilled researchers, with any disagreements resolved by a third researcher.

[0022] Experts extract specific indicators from the literature and list candidate indicators; then the research team reviews the list and proposes new indicators to add, reorganize or delete duplicate indicators or indicators deemed irrelevant; finally, the findings from all sources are summarized into a final list of indicators.

[0023] S3. Selection of Experts by Consultation:

[0024] The experts were carefully selected and recruited to participate in the Delphi study. The inclusion criteria for the experts are as follows:

[0025] S3-1. Possess a bachelor's degree or above;

[0026] S3-2. Possess an intermediate or higher professional title;

[0027] S3-3, More than 10 years of experience in neonatal intensive care, hospital infection control, or clinical microbiology;

[0028] S3-4, demonstrating high enthusiasm, interest, and willingness to participate in research;

[0029] According to the RAND / UCLA Appropriateness Method (RAM), the recommended size of the expert panel is 15 to 25 people.

[0030] S4. Questionnaire Design:

[0031] The questionnaire consists of four parts:

[0032] S4-1. Purpose and background of the study;

[0033] S4-2. Definition and types of candidate indicators, including outcome indicators, process indicators, balance indicators and sentinel indicators;

[0034] S4-3. Basic information about experts, including work experience, educational background, professional qualifications, identification criteria, and familiarity with the indicators;

[0035] S4-4, Indicator Evaluation Section: A nine-point Likert scale is used to assess the relevance and feasibility of each indicator.

[0036] S5, Delphi Inquiry Round 1:

[0037] Experts were asked to assess the relevance and feasibility of all candidate indicators through an online survey;

[0038] The experts were given specific instructions to use a nine-point Likert scale to evaluate each indicator, ranging from 1 (completely irrelevant / infeasible) to 9 (highly relevant / feasible).

[0039] Following the RAM approach, experts need to utilize available scientific evidence and best clinical judgment when assessing relevance and feasibility;

[0040] Each candidate metric is evaluated based on two questions:

[0041] (1) Whether the indicators are relevant, i.e. whether the quality of medical-related infections in the neonatal intensive care unit was assessed;

[0042] (2) Whether the indicator is feasible, that is, whether the required data can be collected, measured and reported in clinical practice, and whether the cost is controllable;

[0043] Experts assess the basis for their judgments and their familiarity with the indicators based on pre-determined scoring criteria.

[0044] An additional expert opinion section has been added to the consultation form, allowing experts to add or modify different indicators; experts are encouraged to provide additional indicator suggestions for each indicator.

[0045] S6. Online Expert Meetings:

[0046] After analyzing the results of the first round of Delphi inquiries, the research team discussed the relevance and feasibility of the indicators based on supporting literature and the experience of team members; and organized an online thematic discussion meeting with experts.

[0047] The indicators accepted in the first round, as well as those lacking consistency and with uncertain scores, were all included in the online meeting discussion. The meeting focused on the applicable scenarios of the indicators, the indicators with uncertain scores, and the indicators that needed to be modified, supplemented, or designated as sentinel monitoring indicators based on the expert feedback collected in the first round.

[0048] The aim of the group discussion was to clarify differences in professional understanding, identify challenges in clinical practice, and provide direction for improving indicators and promoting more accurate scoring in the second round.

[0049] The final scores were still collected independently and anonymously through questionnaires; throughout the discussion, all personal and institutional information of the experts was concealed, and participants were identified only through randomly assigned anonymous codes.

[0050] S7, Delphi Inquiry Round Two:

[0051] Following the online expert meeting, the research team developed a second round of questionnaires.

[0052] The questionnaire summarized and reported the statistical results of the first round of surveys, including each expert's initial score, the median score of the expert group, and the Disagreement Index (DI), so that experts could make a more in-depth evaluation based on their knowledge.

[0053] After the second round of questionnaires were collected, the team organized and analyzed the data, and sent the preliminary analysis results to all participating experts for member checking to ensure the accuracy and reliability of the research conclusions.

[0054] S8. Data Analysis:

[0055] Data analysis was performed using SPSS statistical software.

[0056] Categorical variables are summarized using counts and percentages;

[0057] Questionnaire response rate is used to measure the enthusiasm of experts;

[0058] The expert judgment coefficient Ca is calculated based on four criteria: theoretical analysis, practical experience, reference to domestic and international data, and intuitive feeling.

[0059] The expert familiarity coefficient Cs was assessed using a five-point Likert scale ranging from 1 (very unfamiliar) to 5 (very familiar);

[0060] The level of the expert authority coefficient Cr is represented by the arithmetic mean of the expert judgment coefficient Ca and the familiarity coefficient Cs;

[0061] Ca, Cs, and Cr were determined through expert self-assessment; when the expert authority coefficient Cr is equal to or greater than 0.70, the consultation rate is reliable, and the higher the Cr value, the higher the authority.

[0062] The relevance and feasibility of each candidate indicator are scored on a nine-point Likert scale.

[0063] For each indicator, a group median score was calculated to determine the degree of relevance and feasibility, and a divergence index (DI) was also calculated to determine the level of consistency.

[0064] According to RAM, DI is the ratio between percentile range IPR and adjusted symmetry IPR IPRAS; DI < 1 indicates consistency, and the closer the score is to zero, the stronger the consistency.

[0065] The indicators are divided into the following three categories:

[0066] (1) Exclusion: The median score of groups 1-3 and DI<1 indicate that the indicator is irrelevant / infeasible;

[0067] (2) Uncertainty: For indicators lacking consistency, DI≥1 and / or the group median score is 4-6 and DI<1, it indicates that the relevance / feasibility of the indicator is uncertain;

[0068] (3) Acceptance: A group median of 7-9 and a consistency DI < 1 indicate that the indicator is relevant / feasible;

[0069] The NICU-Shield evaluation system was finally determined, which includes 11 core indicators and 1 sentinel indicator.

[0070] In this invention:

[0071] Furthermore, the research team members mentioned in step S1 are preferably 12 members, including 3 directors of the neonatal department, 2 nursing managers of the neonatal department, 2 directors of the hospital infection control department, 2 clinical microbiology experts, 1 second-line doctor of the neonatal intensive care unit, 1 part-time doctor of hospital infection control in the neonatal intensive care unit, and 1 specialist nurse of the neonatal intensive care unit.

[0072] Furthermore, the literature retrieval described in step S2 covers the period from database creation to October 2025.

[0073] Furthermore, the candidate indicators mentioned in step S2 include 7 outcome indicators, 10 process indicators, and 4 balance indicators, for a total of 21 indicators.

[0074] Furthermore, in step S3, taking into account the possible decline in participation rates during multiple rounds of surveys, a total of 21 experts were invited.

[0075] Furthermore, in the online expert meeting described in step S6, whether a certain indicator can be identified as a sentinel indicator requires a consensus reached by more than two-thirds of the experts as the final determination basis.

[0076] Furthermore, the final NICU-Shield evaluation system determined in step S8 includes:

[0077] The three outcome measures were: incidence of hospital-acquired infections per 1,000 hospital stays, incidence of central venous catheter-related bloodstream infections per 1,000 catheter days, and hospital acquisition rate of certain multidrug-resistant organisms.

[0078] The process indicators include six items: hand hygiene compliance, duration of invasive ventilation, duration of central venous catheter placement, bed occupancy rate, number of days of intravenous infusion, and management of intravenous antibiotics.

[0079] Two balance indicators: length of hospital stay and all-cause mortality rate;

[0080] One sentinel indicator: early warning of hospital-acquired infection outbreaks.

[0081] Furthermore, the evaluation population of the evaluation system is limited to extremely low birth weight infants, with a birth weight of <1500g.

[0082] Compared with the prior art, the present invention has the following beneficial effects:

[0083] The method described in this invention for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method:

[0084] 1. Highly specialized: This invention is the first to construct a quality evaluation and grading system specifically for medical-related infections in the neonatal intensive care unit (NICU), which fully considers the special physiological characteristics of extremely low birth weight infants and the unique infection types in the NICU, filling a gap in this field.

[0085] 2. Rigorous methodology: This invention strictly follows the RAND / UCLA Appropriateness Methodology (RAM) and the Delphi Research Implementation and Reporting Guidelines (CREDES), and employs two rounds of Delphi expert consultation to ensure the scientific validity and reliability of the indicator system.

[0086] 3. Improved indicator system: A four-dimensional indicator system has been constructed, which includes outcome indicators, process indicators, balance indicators and sentinel indicators. It not only focuses on the final result of infection control, but also emphasizes the monitoring of process quality. At the same time, the balance indicators ensure patient safety, and the sentinel indicators enable early warning of major risks.

[0087] 4. High Expert Authority: All participating experts come from tertiary-level Class A hospitals and have extensive clinical and quality management experience in NICUs. The authority coefficient of both rounds of expert consultation is >0.8, ensuring the reliability of the research results.

[0088] 5. High clinical operability: All indicators can be collected in routine clinical work, the data acquisition methods are clear, the calculation methods are standardized, and it is convenient to conduct horizontal comparisons and longitudinal monitoring between different medical institutions.

[0089] 6. Continuous Improvement Orientation: This system is not only used for quality evaluation, but more importantly, it supports continuous quality improvement. Through regular data feedback and benchmarking, it promotes the continuous improvement of infection control in NICU hospitals. Attached Figure Description

[0090] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0091] Figure 1 This is a flowchart of the method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method, as described in this invention.

[0092] Figure 2This is a framework diagram of the NICU-Shield evaluation index system in the method for constructing a medical-related infection evaluation and grading system for neonatal intensive care units using the Delphi method, as described in this invention. Detailed Implementation

[0093] The embodiments and examples of the present invention will be described in detail below with reference to the implementation methods and examples. However, those skilled in the art will understand that the following implementation methods and examples are only for illustrating the present invention and should not be regarded as limiting the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Unless otherwise specified, conventional conditions or conditions recommended by the manufacturer shall apply.

[0094] This invention employed the RAND / UCLA Appropriateness Methodology (RAM) and conducted two rounds of Delphi expert consultations. To enhance methodological rigor, the study strictly adhered to the requirements of the Delphi Research Implementation and Reporting Guidelines (CREDES). All research procedures complied with the ethical principles outlined in the Declaration of Helsinki, and data processing strictly complied with the General Data Protection Regulation (GDPR) to protect the privacy and information security of participants. The Ethics Review Committee of Bao'an District People's Hospital, Shenzhen, has exempted this invention from ethical approval requirements.

[0095] The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method includes the following steps:

[0096] sample:

[0097] To ensure the comparability of the NICU-shield assessment system in cross-institutional comparisons and time-series analyses, efforts are made to balance the tension between accuracy, fairness, and generalizability; the assessment system includes extremely low birth weight infants (weight <1500g) and minimizes the number of excluded cases as much as possible (see Supplementary Material 1 for specific exclusion criteria and their basis); all these criteria have been reviewed and confirmed by the expert panel described below;

[0098] Participants:

[0099] The selection and formation of the expert group for this invention follows the principles of professionalism, representativeness, practicality and voluntary participation. Experts are recruited nationwide based on pre-established inclusion criteria. The specific selection criteria are as follows: (1) Possess a bachelor's degree or above and have an intermediate or higher professional technical title; (2) Have been engaged in neonatal medical care for 10 years or more; (3) Have rich practical experience in the prevention and control of neonatal hospital-acquired infections and be able to provide professional and valuable opinions and suggestions; (4) Have a high level of enthusiasm for participating in research and are willing to cooperate with this invention.

[0100] The size of the expert panel for this invention was determined based on the RAND / UCLA suitability method (RAM); considering the possibility of expert attrition during multiple rounds of Delphi consultations, the research team initially invited 21 experts to participate; all invited experts confirmed their participation and completed the entire research process, ultimately forming a valid expert sample of 21 people;

[0101] Questionnaire preparation:

[0102] This invention employed a systematic literature search in PubMed, Web of Science, and Embase databases (keywords included "infant, newborn," "healthcare-associated infection," etc., with a search period from database inception to October 28, 2025, limited to original English studies). Two senior neonatologists independently screened the literature, with disagreements arbitrated by a third party. Based on the relevance standards for improving neonatal hospital-acquired infection quality, specific personnel extracted and defined indicators, prioritizing definitions from authoritative literature, and integrating inconsistencies or mismatches to form composite operational definitions. Subsequently, a review panel composed of three NICU directors with quality management experience and two senior physicians reviewed the indicators based on clinical evidence or literature support, proposing additions, deletions, and modifications. This resulted in an initial list of 21 indicators, including 7 outcome indicators, 10 process indicators, and 4 balance indicators, laying the foundation for subsequent Delphi consultations. The initial questionnaire draft was finalized after internal pre-testing to optimize its wording and logic, and included four parts: research description, expert demographic information, indicator definitions and evaluation scales, and expert self-assessment.

[0103] Delphi process:

[0104] The Delphi expert consultation of this invention was conducted in two rounds. The first round was conducted from December 1 to December 14, 2025, and the second round was conducted from January 26 to February 9, 2026. Before the research was conducted, electronic informed consent forms were obtained from all participants. Questionnaires were distributed using a combination of online surveys and offline paper questionnaires. All collected data were managed by designated personnel through a designated research computer and stored in an encrypted and secure environment.

[0105] Delphi Round 1: This invention explicitly requires 21 experts to focus on the overall relevance and feasibility of the indicators, rather than evaluating their applicability to individual patients; at the same time, the research team clarified that the constructed NICU-Shield could be used for internal quality improvement or comparative measurement of external quality; the experts used a 9-point Likert scale (1 = completely irrelevant / infeasible, 9 = highly relevant / feasible) to score each candidate indicator, and made judgments mainly based on two core questions: (1) Is the indicator effective in assessing the quality of infection prevention and control in NICU hospitals? (2) Is the indicator collectable, measurable, reportable, and cost-controllable? To assess the objectivity of the evaluation, experts evaluated their judgment criteria and familiarity with the indicators based on pre-determined scoring standards. The consultation questionnaire included a section for expert opinions, encouraging experts to suggest new indicators or modifications to existing ones. All suggestions that received approval from more than one-third of the participating experts were included in subsequent discussions. Furthermore, the study specifically solicited expert opinions on whether indicators outside the comprehensive indicator system could serve as sentinel indicators, clarifying that sentinel indicators are used to identify individual adverse events that reveal significant systemic deficiencies or risks in healthcare quality. These indicators themselves are not routinely scored or graded; once triggered, they require in-depth investigations such as root cause analysis to trace systemic problems and drive improvement. Ultimately, the designation of an indicator as a sentinel indicator is based on consensus reached by more than two-thirds of the experts.

[0106] Online Meeting: During the implementation of the Delphi method in this invention, an online thematic discussion was organized between two rounds of expert consultation. This meeting focused on the following core issues: applicable scenarios for indicators, indicators with scoring uncertainties, and indicators that need to be modified, supplemented, or designated as sentinel monitoring indicators based on expert feedback collected in the first round. Through collective discussion, the aim was to clarify differences in professional understanding, identify challenges in clinical practice, and provide direction for improving indicators and promoting more accurate scoring in the second round, thereby reducing repetitive disputes and improving the efficiency of reaching consensus. The purpose of this discussion was to exchange viewpoints. The final scores were still collected independently and anonymously through questionnaires. Throughout the discussion, all personal and institutional information of the experts was concealed, and participants were identified only through randomly assigned anonymous codes to ensure objectivity.

[0107] Delphi Round Two: Following the online expert meeting, the research team developed a second round of questionnaires. These questionnaires summarized and reflected the statistical results of the first round, including each expert's initial score, the median score of the expert group, and the Disagreement Index (DI), enabling experts to conduct more in-depth evaluations based on informed consent. After the second round of questionnaires was collected, the team organized and analyzed the data, and then submitted the preliminary analysis results to all participating experts for member checking to ensure the accuracy and reliability of the invention's conclusions.

[0108] Statistical analysis:

[0109] This invention uses SPSS 20.0 software for data analysis; the expert judgment coefficient (Ca) is determined based on experts' self-assessment results across four dimensions: theoretical analysis, work experience, literature reference, and intuitive judgment. Each dimension is categorized into high, medium, and low levels according to its degree of influence; the expert familiarity coefficient (Cs) is determined by experts using Likert. The self-assessment was conducted using a 5-point scale (1 point = extremely unfamiliar, 5 points = very familiar); the expert authority coefficient (Cr) was the arithmetic mean of Ca and Cs, used to comprehensively reflect the overall authority of the experts; Cr ≥ 0.70 indicated that the expert authority reached an acceptable level, and the higher the value, the stronger the credibility of the experts; this invention uses the median score of the expert group and the Disagreement Index (DI) to evaluate the relevance and feasibility of each candidate indicator; the Disagreement Index was calculated as the ratio of the percentile interval (IPR) to the skewness-corrected percentile interval (IPRAS), and when DI < 1, it was considered that the experts had reached a consensus (the lower the value, the higher the degree of consensus); the candidate indicators were divided into three categories accordingly: included category, median score 7-9 and DI < 1; pending category, median score 4-6 and / or DI ≥ 1; excluded category, median score 1-3 and DI < 1; for detailed evaluation criteria of Cr and calculation method of DI, please refer to Supplementary Material 2.

[0110] result:

[0111] Literature search:

[0112] The initial search yielded 2,327 relevant articles; after removing 120 duplicate records, 2,122 articles that did not meet the inclusion criteria were excluded by reading the titles and abstracts; the remaining 85 articles were evaluated in full, and 60 of them were excluded; finally, a total of 25 articles were included for indicator extraction.

[0113] The expert panel's demographic and authority

[0114] Twenty-one experts from 21 tertiary-level hospitals in 16 cities across China participated in this study. The majority of participants (n=12, 57.1%) were between 50 and 59 years old, and 61.9% (n=13) were female. Among these experts, 66.7% (n=14) were directors of neonatology departments, 42.9% (n=9) held doctoral degrees, and 42.9% (n=9) had 10-20 years of professional experience (Table 1). In this invention, 21 questionnaires were distributed in each of the two rounds of expert consultations. All questionnaires were collected and confirmed to be valid, with a 100% effective response rate in both rounds. Furthermore, the expert authority coefficient (Cr) for each round of consultations exceeded 0.8, indicating that the participating experts possessed high professional authority in their field and ensuring the reliability of the results of this invention (Table 2).

[0115] Delphi Round 1:

[0116] In the first round, the research team systematically organized the ratings of all experts on the 21 candidate indicators and their suggestions for modifying or adding indicators; the median score and the DI index were used to assess the relevance and feasibility of these indicators (Table 3); 11 indicators were considered relevant and feasible (52.4%), and 10 indicators were uncertain in terms of relevance or feasibility (47.6%); a total of 19 opinions were collected in the first round of Delphi method expert consultation; based on the preset criteria, 6 valid opinions were finally summarized (3 added indicators, 2 modified indicators, and 1 added sentinel indicator).

[0117] Online meeting:

[0118] Twenty-one experts participated in the online meeting, focusing on exchanging opinions on the rigor of indicator definitions and the feasibility of data collection. Based on the first round of feedback, the research team and experts optimized and adjusted one original indicator according to the pre-set revision specifications and added one sentinel monitoring indicator. All indicator definitions were reviewed and verified item by item to ensure that the expressions were accurate, clear, and reflected the consensus of the experts. Finally, a quality evaluation system containing 20 original indicators, 1 revised indicator, and 1 newly added sentinel indicator was formed, and the system entered the second round of Delphi expert consultation.

[0119] Delphi Round 2:

[0120] In the second round, experts evaluated 21 indicators (Table 3); 11 indicators (52.4%) were selected as quality indicators for nosocomial infections in neonatal intensive care units, with a median of ≥7 and DI <1, indicating their relevance and feasibility; the remaining 10 indicators (47.6%) were discarded; these indicators measured the outcome indicators (n=3), process indicators (n=6), and balance indicators (n=2) of nosocomial infections; 95.2% (n=20) of the experts agreed to set "nosocomial infection outbreak early warning" as a sentinel indicator.

[0121] discuss:

[0122] This invention constructs a hospital-acquired infection specialty quality assessment system (NICU-Shield) applicable to neonatal intensive care units. This system includes 11 core indicators and 1 sentinel indicator. This invention limits the evaluation population to very low birth weight infants (birth weight <1500g) to reduce potential bias caused by differences in neonatal baseline risk, thereby improving the accuracy and fairness of the evaluation system. The indicator system takes into account relevance and feasibility, ensuring operability and sustainability in a real NICU environment.

[0123] In the screening process of outcome indicators, three core indicators were ultimately included: "incidence rate of hospital-acquired infections per 1,000 hospital stays," "incidence rate of central venous catheter-related bloodstream infections per 1,000 catheter days," and "hospital-acquired infection rate of specific multidrug-resistant organisms (HO-MDRDs)." These indicators together constitute the core dimension for evaluating the direct consequences of infection. "Incidence rate of hospital-acquired infections per 1,000 hospital stays" is a fundamental indicator for assessing the overall infection burden. After collective expert review, this invention focuses its monitoring scope on severe hospital-acquired infections, with the criteria being: detection of viral pathogens through molecular diagnostic techniques, or detection of bacterial / fungal pathogens through culture of blood, cerebrospinal fluid, or urine samples. "Incidence rate of central venous catheter-related bloodstream infections per 1,000 catheter days" is an internationally recognized and widely used core indicator used to assess the risk of central venous catheter-related infections and evaluate the effectiveness of related prevention and control measures. This indicator is determined based on the monitoring definition and does not require consideration of signs or symptoms of infection. This characteristic may lead to a potential overestimation of the true incidence rate of central venous catheter-related bloodstream infections in practical applications. Some studies use catheter-related bloodstream infection (CRBSI) as an indicator for improving the quality of hospital-acquired infections. However, in the neonatal population, the large-scale application of CRBSI in epidemiological studies faces many challenges due to the low specificity of catheter tip culture, the low rate of collecting multiple blood cultures, and the complexity of confirming the catheter as the source of bloodstream infection. In 2024, the U.S. Centers for Disease Control and Prevention (CDC) proposed a new surveillance indicator—hospital-acquired bacteremia (HOB)—to expand the surveillance scope of hospital-acquired bloodstream infections in hospitalized patients. Related studies have also confirmed that hospital-acquired bacteremia events in neonatal intensive care units are associated with increased infant mortality, but current strategies for prevention and risk reduction of this infection are still lacking. In contrast, the surveillance procedure for central venous catheter-related bloodstream infection is relatively simple and more suitable for medical institutions such as neonatal intensive care units that need to routinely monitor central venous catheter-related bloodstream infection, and can provide reliable data support for medical institutions to conduct nosocomial infection trend analysis.

[0124] Based on expert feedback, this invention has revised the relevant indicator of "Methicillin-resistant Staphylococcus aureus (MRSA) hospital acquisition rate," expanding it to "Hospital Acquisition Rate of Specific Multidrug-resistant Organisms" (HO-MDROs). This adjustment aims to ensure that this monitoring indicator can systematically reflect the key directions of current infection control efforts. For details on the specific types of specific multidrug-resistant organisms covered by this indicator, please refer to the indicator definition section of Supplementary Material 4. Existing research indicates a high carrier rate of antibiotic-resistant pathogens in premature infants. Based on this characteristic, monitoring the detection of these pathogens (rather than limiting it to confirmed infection cases) can provide a reference for the rational selection of antimicrobial drugs and the strict implementation of contact isolation measures in clinical practice. Currently, there is no unified definition for neonatal ventilator-associated pneumonia (VAP). Standardized diagnostic protocols are required. Although this indicator can be used for in-hospital quality improvement monitoring in medical institutions, its validity and reliability are insufficient to support external quality assessment and benchmarking. The National Health Safety Network (NHSN) no longer lists ventilator-associated pneumonia as a neonatal quality assessment indicator, and mainstream neonatal quality collaboration organizations such as the Vermont-Oxford Network have not recognized its application value. Similarly, catheter-related urinary tract infection (CAUTI) has not been included in the core indicator system, mainly because the rate of catheterization in newborns is much lower than in adults and older children, which limits the application value of this indicator in cross-institutional comparisons and longitudinal monitoring. For the above reasons, this invention does not include ventilator-associated pneumonia and catheter-related urinary tract infection as outcome indicators in the NICU-Shield evaluation system.

[0125] Process indicators are tools for measuring adherence to best clinical practices. They help identify deficiencies in quality improvement projects and clinical audits, and monitor their progress. The basic clinical processes measured by these indicators are the core drivers for improving outcome indicators. Hand hygiene is widely recognized as a core measure in hospital-acquired infection control systems, and hand hygiene adherence rate is a key indicator for quantitatively evaluating the effectiveness of this control measure. Extensive evidence confirms a significant negative correlation between improved hand hygiene adherence and a decrease in hospital-acquired infection rates; the magnitude of this increase directly reflects the quality of infection control implementation in medical institutions. In hospital-acquired infection control practices in neonatal intensive care units, optimization of several core clinical and operational indicators is also closely related to reduced infection rates, including shortening invasive ventilation time, reducing central venous catheter indwelling time, scientifically managing bed occupancy rates, and appropriately shortening the number of days of intravenous infusion. Simultaneously, the clinical application of intravenous antibiotics is also... It can indirectly reflect the trend of hospital-acquired infections; therefore, the standardized management of intravenous antibiotics is not only the core content of clinical antibiotic management, but also an important basis for evaluating the effectiveness of hospital-acquired infection control in neonatal intensive care units. Although the aseptic technique compliance rate of central venous catheter insertion and maintenance is a key process indicator for preventing catheter-related infections, it was ultimately not included in the core indicator system. This is mainly based on the following consensus: at the operational level, continuous and objective monitoring requires a high investment of manpower and time, and standardized collection across institutions is difficult; in terms of definition and evaluation, the lack of unified and operable implementation and judgment standards easily introduces subjective bias; in addition, during the system construction process, experts emphasized the universality and clinical scalability of the indicators, so the overall feasibility score of this indicator is lower than that of other core process indicators; similarly, the "environmental hygiene compliance rate" was also not included due to insufficient monitoring feasibility and limited promotion.

[0126] In the selection of balancing indicators, length of hospital stay and all-cause mortality were ultimately included, while hospitalization costs and the incidence of bronchopulmonary dysplasia (BPD) were excluded. Length of hospital stay was included because it directly reflects the efficiency of healthcare services and the utilization of healthcare resources. Existing research has shown a significant correlation between a decrease in infection rates and a reduction in hospitalization duration; while longer hospital stays may increase the risk of hospital-acquired infections, and improvements in the quality of infection control efforts can also help reduce neonatal mortality. Including all-cause mortality in the balancing indicators aims to alert and ensure that the implementation of infection control measures does not compromise patient safety. Hospitalization costs were excluded because they are significantly affected by non-medical factors such as regional policies, which not only limits the cross-institutional comparability of this indicator but may also introduce bias into the evaluation. The incidence of bronchopulmonary dysplasia (BPD) was excluded because its correlation with the balancing effect of infection control interventions is relatively weak.

[0127] Neonatal hospital-acquired infection outbreaks are a globally recognized serious healthcare safety issue. Although the incidence of such events is low, they can easily lead to clusters of cases and even rapid deterioration or death in newborns. To enhance the early identification and warning of such extreme risks, this invention adds "hospital-acquired infection outbreak warning" as a core sentinel indicator. This aims to compensate for the shortcomings of conventional quality indicators in warning of low-probability, high-risk events, providing crucial evidence for the early identification of systemic infection risks, root cause analysis, and emergency intervention.

[0128] Advantages and limitations

[0129] The composition of the expert panel significantly enhanced the clinical applicability of this indicator system. All participants in the Delphi method were frontline clinicians, including NICU directors with quality management experience and senior neonatologists with over 10 years of experience, covering areas such as neonatal care, infection control, and clinical management. These experts possessed a deep understanding of the infection control needs and operational challenges of NICUs, while also assessing the feasibility of implementation from a management perspective. They provided authoritative judgments on the relevance, definition, and measurement standards of the indicators, identified limiting factors in practical application (such as data collection and cross-institutional standardization), and proposed revisions aligned with workflows. This dual focus on clinical practice and quality management ensures that the system prioritizes key infection control processes while adapting to different healthcare environments, thereby maximizing its practicality and scalability. The NICU-Shield evaluation system constructed in this invention provides a quantitative reference standard for evaluating the quality of NICU infection control management. Neonatal intensive care unit (NICU) teams need systematic training to truly understand and apply various indicators to clinical practice. This will enable the systematic collection, analysis, and reporting of indicator-related data, avoiding inconsistencies in implementation standards that could affect the comparability and reliability of evaluation results. For medical institutions with small sample sizes of extremely low birth weight infants and where individual case variations may lead to unstable outcome indicator evaluation results, the focus of evaluation can shift from outcome indicators to the monitoring and optimization of process indicators. Furthermore, medical institutions must understand that the core purpose of this system is to promote continuous quality improvement in infection control, and to verify the effectiveness of intervention measures through regular data analysis and feedback, ensuring that evaluation results can be translated into actionable actions to improve patient safety.

[0130] Although the expert panel comprises members from multiple regions, all experts are from China. While this ensures the system's local applicability, it may limit its international universality. Future cross-cultural research could be conducted. Currently, the system is still a "theoretical framework" based on literature review and expert consensus. Its reliability, validity, and actual impact on clinical outcomes need to be validated through prospective application of NICU-Shield in multiple NICUs at different levels. As medical practice continues to evolve, the indicator system should be updated and revised regularly (e.g., every 3-5 years) to maintain its timeliness. Furthermore, the present invention did not directly incorporate the perspective of patients' families when constructing the indicator system, which may result in deficiencies in terms of humanistic care and family collaboration. For example, in family-participatory care and kangaroo care, significant differences exist in family participation rates and average daily participation time across institutions, and key behaviors such as family hand hygiene compliance have not been translated into assessable indicators, potentially affecting the comprehensiveness and clinical guidance value of the indicator system. Future research could use methods such as qualitative interviews or participatory design to systematically integrate the experiences and needs of patients and their families, incorporating them into the revision and improvement of indicators, thereby enhancing the system's humanistic content and practical effectiveness.

[0131] in conclusion:

[0132] NICU-Shield provides a consensus-based, easy-to-use framework for standardized hospital-acquired infection (HAI) quality monitoring in neonatal intensive care units (NICUs). Adopting this framework facilitates meaningful benchmarking and guides quality improvement efforts. Prospective validation studies are needed to assess its implementation fidelity and relevance to improved patient outcomes.

[0133] Example:

[0134] A method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method, such as... Figure 1 As shown, it includes the following steps:

[0135] 1. Establish a multidisciplinary research team:

[0136] The research team consists of 12 members, including 3 directors of neonatology, 2 neonatology nursing managers, 2 directors of hospital infection control, 2 clinical microbiology experts, 1 second-line physician in the neonatal intensive care unit, 1 part-time physician in infection control in the neonatal intensive care unit, and 1 specialist nurse in the neonatal intensive care unit.

[0137] The research team members divided the work and were responsible for searching, reading and analyzing literature, developing expert inquiry questionnaires, selecting experts for inquiry and conducting expert inquiries, compiling and analyzing the feedback results of each round of expert inquiries, and determining whether indicators need to be deleted, modified or added.

[0138] 2. Determine candidate indicators:

[0139] A systematic literature search was conducted in PubMed, Web of Science, and Embase databases, with keywords including "infant, newborn" and "healthcare-associated infection," and the search period was from the database inception to October 2025, limited to original English studies.

[0140] The literature was independently screened by two senior neonatologists, and any disagreements were arbitrated by a third party.

[0141] Based on the standards for improving the quality of neonatal nosocomial infections, a designated person extracted and defined indicators, prioritizing definitions from authoritative literature, and integrating them into a composite operational definition when there are inconsistencies or mismatches.

[0142] Subsequently, a review panel consisting of three NICU department directors with quality management experience and two senior physicians reviewed the indicators based on clinical evidence or literature support and made suggestions for additions, deletions and modifications.

[0143] The final list of 21 initial indicators includes 7 outcome indicators, 10 process indicators, and 4 balance indicators.

[0144] The initial draft of the questionnaire was finalized after internal pre-testing to optimize its wording and logic. It includes four parts: research description, expert demographic information, indicator definitions and evaluation scales, and expert self-evaluation.

[0145] 3. Selection of experts for consultation:

[0146] The selection and formation of the expert panel for this invention follows the principles of professionalism, representativeness, practicality, and voluntary participation, and experts are recruited nationwide based on pre-established inclusion criteria.

[0147] The specific selection criteria are as follows: (1) Possess a bachelor's degree or above and have an intermediate or higher professional technical title; (2) Have been engaged in neonatal medical and nursing work for 10 years or more; (3) Have rich practical experience in the field of neonatal hospital-acquired infection prevention and control and be able to provide professional and valuable opinions and suggestions; (4) Have a high level of enthusiasm for participating in the research and be willing to cooperate with the research.

[0148] Considering the potential loss of experts during the multiple rounds of Delphi consultations, the research team initially invited a total of 21 experts to participate.

[0149] 4. Questionnaire Design:

[0150] The questionnaire consists of four parts:

[0151] (1) Purpose and background of the study;

[0152] (2) Definition and types of candidate indicators, including outcome indicators, process indicators, balance indicators and sentinel indicators;

[0153] (3) Basic information of experts, including work experience, educational background, professional qualifications, identification criteria and familiarity with indicators;

[0154] (4) In the indicator evaluation section, a nine-point Likert scale is used to evaluate the relevance and feasibility of each indicator, ranging from 1 (completely irrelevant / infeasible) to 9 (highly relevant / feasible).

[0155] 5. Delphi Inquiry Round 1:

[0156] The first round will be held from December 1st to December 14th, 2025;

[0157] The 21 experts focused on the overall relevance and feasibility of the indicators, rather than evaluating their applicability to individual patients.

[0158] Experts used a 9-point Likert scale (1 = completely irrelevant / infeasible, 9 = highly relevant / feasible) to score each candidate indicator and made judgments based on two core questions: (1) Is the indicator effective in assessing the quality of infection prevention and control in NICU hospitals? (2) Is the indicator collectable, measurable, reportable, and cost-effective?

[0159] To assess the objectivity of the evaluation work, the experts evaluated their judgment criteria and familiarity with the indicators based on the pre-determined scoring standards.

[0160] The consultation questionnaire includes a special section for expert opinions, encouraging experts to propose new indicators or suggestions for modifying existing indicators; all suggestions that receive approval from more than one-third of the participating experts will be included in the subsequent discussion.

[0161] In addition, the study specifically consulted experts on whether other indicators outside the comprehensive indicator system could be used as sentinel indicators, and clearly stated that sentinel indicators are used to identify individual adverse events that can reveal significant systemic defects or risks in medical quality; these indicators themselves are not routinely scored or graded, and once triggered, in-depth investigations such as root cause analysis must be initiated.

[0162] In the first round, the research team systematically organized all the experts' ratings of the 21 candidate indicators and their suggestions on modifying or adding indicators; the median score and the DI index were used to assess the relevance and feasibility of these indicators; 11 indicators were considered relevant and feasible (52.4%), and the relevance or feasibility of 10 indicators was uncertain (47.6%).

[0163] The first round of expert consultation using the Delphi method yielded 19 comments; based on the pre-set criteria, these were ultimately summarized into 6 valid comments (3 added indicators, 2 modified indicators, and 1 added sentinel indicator).

[0164] 6. Online expert meetings:

[0165] Twenty-one experts participated in the online meeting, focusing on exchanging opinions on the rigor of indicator definitions and the feasibility of data collection.

[0166] Based on the initial feedback, the research team and experts optimized and adjusted one original indicator according to the pre-set revision specifications, and added one new sentinel monitoring indicator.

[0167] All indicator definitions have been reviewed and approved item by item to ensure accurate and clear expression and to reflect expert consensus;

[0168] The final quality evaluation system, comprising 20 original indicators, 1 revised indicator, and 1 newly added sentinel indicator, was formed and entered the second round of Delphi expert consultation.

[0169] Whether an indicator can be identified as a sentinel indicator requires a consensus reached by more than two-thirds of the experts as the final determination basis;

[0170] 7. Delphi Inquiry Round Two:

[0171] The second round will be implemented from January 26 to February 9, 2026;

[0172] In the second round, experts evaluated 21 indicators; 11 indicators (52.4%) were selected as quality indicators for nosocomial infections in neonatal intensive care units, with a median of ≥7 and a DI <1, indicating that they were relevant and feasible; the remaining 10 indicators (47.6%) were discarded.

[0173] 95.2% (n=20) of experts agreed to set "hospital infection outbreak early warning" as a sentinel indicator;

[0174] 8. Data Analysis:

[0175] SPSS 20.0 software was used for data analysis;

[0176] The expert judgment coefficient Ca is calculated based on four criteria: theoretical analysis, practical experience, reference to domestic and international data, and intuitive feeling.

[0177] The expert familiarity coefficient Cs was assessed using a five-point Likert scale ranging from 1 (very unfamiliar) to 5 (very familiar);

[0178] The level of the expert authority coefficient Cr is represented by the arithmetic mean of the expert judgment coefficient Ca and the familiarity coefficient Cs;

[0179] The relevance and feasibility of each candidate indicator were scored on a nine-point Likert scale; for each indicator, a group median score was calculated to determine the degree of relevance and feasibility, and the Disagreement Index (DI) was also calculated to determine the level of consistency.

[0180] According to RAM, DI is the ratio between percentile range IPR and adjusted symmetry IPR IPRAS; DI < 1 indicates consistency, and the closer the score is to zero, the stronger the consistency.

[0181] The indicators are divided into the following three categories: (1) Exclusion: The median score of 1-3 and DI<1 indicates that the indicator is irrelevant / infeasible; (2) Uncertainty: The lack of consistency of the indicator and DI≥1 and / or the median score of 4-6 and DI<1 indicates that the relevance / feasibility of the indicator is uncertain; (3) Acceptance: The median score of 7-9 and consistency DI<1 indicates that the indicator is relevant / feasible.

[0182] result:

[0183] Twenty-one experts from 21 tertiary-level hospitals in 16 cities across China participated in this study; the majority of participants (n=12, 57.1%) were between 50 and 59 years old, and 61.9% (n=13) were female; among these experts, 66.7% (n=14) were directors of neonatology departments, 42.9% (n=9) held doctoral degrees, and 42.9% (n=9) had 10-20 years of professional work experience.

[0184] Twenty-one questionnaires were distributed in each of the two rounds of expert consultations; all questionnaires were collected and confirmed to be valid, with a 100% effective response rate in both rounds; in addition, the expert authority coefficient (Cr) in each round of consultations exceeded 0.8, which indicates that the participating experts have high professional authority in the field and also ensures the reliability of the results of this invention.

[0185] The finalized NICU-Shield evaluation system includes 11 core indicators and 1 sentinel indicator, such as... Figure 2 As shown:

[0186] The three outcome measures were: incidence of hospital-acquired infections per 1,000 hospital stays, incidence of central venous catheter-related bloodstream infections per 1,000 catheter days, and hospital acquisition rate of certain multidrug-resistant organisms.

[0187] The process indicators include six items: hand hygiene compliance, duration of invasive ventilation, duration of central venous catheter placement, bed occupancy rate, number of days of intravenous infusion, and management of intravenous antibiotics.

[0188] Two balance indicators: length of hospital stay and all-cause mortality rate;

[0189] One sentinel indicator: early warning of hospital-acquired infection outbreaks.

[0190] Table 1: Demographic Information of Experts

[0191]

[0192] Table 2: Delphi Expert Authority Coefficient

[0193]

[0194] Note: Ca is the judgment coefficient; Cs is the familiarity coefficient; Cr is the expert authority coefficient.

[0195] It will be readily understood by those skilled in the art that the above description is merely for clarity and is not intended to limit the implementation. For those skilled in the art, other variations or modifications can be made based on the above description. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for constructing an evaluation and grading system for medical-related infections in the neonatal intensive care unit using the Delphi method, characterized by: Includes the following steps: S1. Establish a multidisciplinary research team: The research team consists of more than 10 members, including the director of the neonatal department, the neonatal nursing manager, the director of the hospital infection control department, the clinical microbiology expert, the second-line doctor of the neonatal intensive care unit, the part-time doctor of the neonatal intensive care unit infection control management, and the specialist nurse of the neonatal intensive care unit. The research team members divided the work and were responsible for searching, reading and analyzing literature, developing expert inquiry questionnaires, selecting experts for inquiry and conducting expert inquiries, compiling and analyzing the feedback results of each round of expert inquiries, and determining whether indicators need to be deleted, modified or added. S2. Determine candidate indicators: Candidate indicators were determined through systematic literature retrieval, including databases such as PubMed, Web of Science, and Embase. Search terms included "infant, newborn", "healthcare-associated infection", "QualityImprovement", and "Quality Indicators, Health Care". The search was limited to original research articles published in English and conducted independently by two skilled researchers, with any disagreements resolved by a third researcher. Experts extract specific indicators from the literature and list candidate indicators; then the research team reviews the list and proposes new indicators to add, reorganize or delete duplicate indicators or indicators deemed irrelevant; finally, the findings from all sources are summarized into a final list of indicators. S3. Selection of Experts by Consultation: The experts were carefully selected and recruited to participate in the Delphi study. The inclusion criteria for the experts are as follows: S3-1. Possess a bachelor's degree or above; S3-2. Possess an intermediate or higher professional title; S3-3, More than 10 years of experience in neonatal intensive care, hospital infection control, or clinical microbiology; S3-4, demonstrating high enthusiasm, interest, and willingness to participate in research; Based on the recommendations of the RAND Corporation / UCLA suitability approach, the recommended size of the expert panel is 15 to 25 people. S4. Questionnaire Design: The questionnaire consists of four parts: S4-1. Purpose and background of the study; S4-2. Definition and types of candidate indicators, including outcome indicators, process indicators, balance indicators and sentinel indicators; S4-3. Basic information about experts, including work experience, educational background, professional qualifications, identification criteria, and familiarity with the indicators; S4-4, Indicator Evaluation Section: A nine-point Likert scale is used to assess the relevance and feasibility of each indicator. S5, Delphi Inquiry Round 1: Experts were asked to assess the relevance and feasibility of all candidate indicators through an online survey; The experts were given specific instructions to use a nine-point Likert scale to evaluate each indicator, ranging from 1 (completely irrelevant / infeasible) to 9 (highly relevant / feasible); Following the RAM approach, experts need to utilize available scientific evidence and best clinical judgment when assessing relevance and feasibility; Each candidate metric is evaluated based on two questions: (1) Whether the indicators are relevant, i.e. whether the quality of medical-related infections in the neonatal intensive care unit was assessed; (2) Whether the indicator is feasible, that is, whether the required data can be collected, measured and reported in clinical practice, and whether the cost is controllable; Experts assess the criteria and familiarity with the indicators based on pre-determined scoring standards. An additional expert opinion section has been added to the consultation form, allowing experts to add or modify different indicators; experts are encouraged to provide additional indicator suggestions for each indicator. S6. Online Expert Meetings: After analyzing the results of the first round of Delphi inquiries, the research team discussed the relevance and feasibility of the indicators based on supporting literature and the experience of team members; and organized an online thematic discussion meeting with experts. The indicators accepted in the first round, as well as those lacking consistency and with uncertain scores, were all included in the online meeting discussion. The meeting focused on the applicable scenarios of the indicators, the indicators with uncertain scores, and the indicators that needed to be modified, supplemented, or designated as sentinel monitoring indicators based on the expert feedback collected in the first round. The aim of the group discussion was to clarify differences in professional understanding, identify challenges in clinical practice, and provide direction for improving indicators and promoting more accurate scoring in the second round. The final scores were still collected independently and anonymously through questionnaires; throughout the discussion, all personal and institutional information of the experts was concealed, and participants were identified only through randomly assigned anonymous codes. S7, Delphi Inquiry Round Two: Following the online expert meeting, the research team developed a second round of questionnaires. The questionnaire summarized and reported the statistical results of the first round of surveys, including each expert's initial score, the median score of the expert group, and the inconsistency index, so that experts could make a more in-depth evaluation based on informed consent. After the second round of questionnaires were collected, the team organized and analyzed the data, and sent the preliminary analysis results to all participating experts for member verification to ensure the accuracy and reliability of the research conclusions. S8. Data Analysis: Data analysis was performed using SPSS statistical software. Categorical variables are summarized using counts and percentages; Questionnaire response rate is used to measure the enthusiasm of experts; The expert judgment coefficient Ca is calculated based on four criteria: theoretical analysis, practical experience, reference to domestic and international data, and intuitive feeling. The expert familiarity coefficient Cs was assessed using a five-point Likert scale ranging from 1 (very unfamiliar) to 5 (very familiar). The level of the expert authority coefficient Cr is represented by the arithmetic mean of the expert judgment coefficient Ca and the familiarity coefficient Cs; Ca, Cs, and Cr were determined through expert self-assessment; when the expert authority coefficient Cr is equal to or greater than 0.70, the consultation rate is reliable, and the higher the Cr value, the higher the authority. The relevance and feasibility of each candidate indicator are scored on a nine-point Likert scale. For each indicator, a group median score was calculated to determine the degree of relevance and feasibility, and a divergence index (DI) was also calculated to determine the level of consistency. According to RAM, DI is the ratio between percentile range IPR and adjusted symmetry IPR IPRAS; DI < 1 indicates consistency, and the closer the score is to zero, the stronger the consistency. The indicators are divided into the following three categories: (1) Exclusion: The median score of groups 1-3 and DI<1 indicate that the indicator is irrelevant / infeasible; (2) Uncertainty: For indicators lacking consistency, DI≥1 and / or the group median score is 4-6 and DI<1, it indicates that the relevance / feasibility of the indicator is uncertain; (3) Acceptance: A group median of 7-9 and a consistency DI < 1 indicate that the indicator is relevant / feasible; The NICU-Shield evaluation system was finally determined, which includes 11 core indicators and 1 sentinel indicator.

2. The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method according to claim 1, characterized in that: The research team members mentioned in step S1 consist of 12 people, including 3 directors of the neonatal department, 2 nursing managers of the neonatal department, 2 directors of the hospital infection control department, 2 clinical microbiology experts, 1 second-line doctor of the neonatal intensive care unit, 1 part-time doctor of hospital infection control management in the neonatal intensive care unit, and 1 specialist nurse of the neonatal intensive care unit.

3. The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method according to claim 1, characterized in that: The literature retrieval described in step S2 covers the period from database creation to October 2025.

4. The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method according to claim 1, characterized in that: The candidate indicators mentioned in step S2 initially include 7 outcome indicators, 10 process indicators, and 4 balance indicators, for a total of 21 indicators.

5. The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method according to claim 1, characterized in that: In step S3, taking into account the possible decline in participation rates during multiple rounds of surveys, a total of 21 experts were invited.

6. The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method according to claim 1, characterized in that: In the online expert meeting described in step S6, whether a certain indicator can be identified as a sentinel indicator requires a consensus reached by more than two-thirds of the experts as the final determination basis.

7. The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method according to claim 1, characterized in that: The final NICU-Shield evaluation system determined in step S8 includes: The three outcome measures were: incidence of hospital-acquired infections per 1,000 hospital stays, incidence of central venous catheter-related bloodstream infections per 1,000 catheter days, and hospital acquisition rate of certain multidrug-resistant organisms. The process indicators include six items: hand hygiene compliance, duration of invasive ventilation, duration of central venous catheter placement, bed occupancy rate, number of days of intravenous infusion, and management of intravenous antibiotics. Two balance indicators: length of hospital stay and all-cause mortality rate; One sentinel indicator: early warning of hospital-acquired infection outbreaks.

8. The method for constructing a medical-related infection assessment and grading system for neonatal intensive care units using the Delphi method according to claim 1, characterized in that: The evaluation population of the evaluation system is limited to extremely low birth weight infants, with a birth weight of <1500g.

Images