Implementing standardized, evidence-based nursing management to prevent aspiration risk in hospitalized older adults

Abstract

Background

Aspiration is a major safety concern among hospitalized older adults, yet clinical prevention practices remain inconsistent. Using the JBI Practical Application of Clinical Evidence System (JBI PACES) and guided by the Chinese Nursing Association group standard Prevention of Aspiration in Older Adults, this project aimed to implement a standardized evidence-based nursing program for aspiration risk prevention and to evaluate its impact on nurses’ knowledge, adherence to audit criteria, and aspiration incidence.

Methods

This evidence-based audit-and-feedback project employed a pre–post design at a tertiary hospital in Shanxi Province, China, from July 2024 to December 2025. Reporting followed the SQUIRE 2.0 guidelines. A multidisciplinary team systematically retrieved and appraised evidence, yielding 22 audit criteria. Barriers and facilitators were analyzed using the Ottawa Model of Research Use. Guided by Proctor’s Implementation Outcomes Taxonomy, effects were assessed at three levels: implementation outcomes (audit criteria adherence), service outcomes (nurses’ knowledge, high-risk identification accuracy, feeding plan compliance), and clinical outcomes (aspiration incidence, patient satisfaction). Binary logistic regression adjusted for potential confounders.

Results

Baseline characteristics were comparable between groups (P > 0.05). Aspiration incidence was lower in the post-implementation period (13.59%) compared with baseline (27.72%) (P = 0.015); after adjusting for age, sex, consciousness status, and nutritional route, aspiration risk remained significantly lower post-implementation (adjusted OR = 0.359, 95% CI: 0.157–0.820). Nurses’ knowledge scores rose from 56.72 ± 8.29 to 85.03 ± 8.40 (P < 0.001), and high-risk identification accuracy increased from 34.65% to 81.55% (P < 0.001). Most audit criteria improved significantly: system-level criteria were fully adopted, positioning criteria reached 100%, and feeding behavior criteria showed the largest gains although absolute adherence remained lower. Feeding plan compliance improved from 41.58% to 77.67% (P < 0.001), and satisfaction scores from 80.35 ± 7.62 to 92.16 ± 4.85 (P < 0.001).

Conclusions

This JBI PACES-based multi-component intervention was associated with improved adherence to aspiration prevention measures and lower aspiration incidence among hospitalized older adults. Positive changes were observed across implementation, service, and clinical outcome levels within Proctor’s taxonomy. Given the pre–post design, these improvements should be interpreted as observed changes rather than confirmed causal effects. Future studies employing stepped-wedge or multicenter controlled designs are needed to strengthen causal inference and assess long-term sustainability.

Registration

This evidencebased practice project was registered with the Fudan University Center for EvidenceBased Nursing, China (Registration No. ER20251074) on January 9, 2025.

Clinical trial registration

Not applicable.

Background

With the acceleration of population aging, aspiration has become a significant threat to the health and safety of older adults [1]. Age-related degenerative changes in muscle function and the high prevalence of neurodegenerative diseases impair swallowing function and weaken protective reflexes, placing older adults at elevated aspiration risk. Aspiration is defined as the entry of liquids, solid food, secretions, or blood below the vocal cords into the lower airways and lung tissue during swallowing, classified as overt or silent [2]. A systematic review demonstrated that aspiration in community-acquired pneumonia patients increased in-hospital mortality approximately 3.6-fold (RR = 3.62, 95% CI: 2.65–4.96) [3].

The reported incidence of aspiration among hospitalized older adults varies widely, from 22.0% to 88.0% [4, 5]. This variation reflects substantial methodological heterogeneity: differences in study populations, with some studies focusing on high-risk subgroups with neurological conditions and others encompassing general older inpatients [6]; inconsistent diagnostic criteria, with some relying on clinical symptoms and others employing videofluoroscopic swallowing study (VFSS) or fiberoptic endoscopic evaluation of swallowing (FEES) as gold standards [7]; and differences in assessment tools, with validated instruments (e.g., V-VST, SSA) yielding dysphagia detection rates of approximately 47%, compared with approximately 24% for non-validated tools [8]. Despite this variability, the evidence consistently indicates substantial aspiration risk [9]. Recurrent aspiration can lead to dehydration, malnutrition, and aspiration pneumonia, with particularly pronounced mortality and costs among critically ill older adults [10, 11].

At the international level, several authoritative guidelines address aspiration prevention. The British Thoracic Society (BTS) 2023 clinical statement emphasized oral hygiene, swallowing assessment, and multidisciplinary collaboration [12]. The 2023 Korean Clinical Practice Guidelines for Oropharyngeal Dysphagia recommended standardized pathways from screening to rehabilitation [13]. The European Society for Swallowing Disorders (ESSD) and European Union Geriatric Medicine Society white paper defined oropharyngeal dysphagia as a geriatric syndrome [14]. In China, the Chinese Nursing Association published the group standard Prevention of Aspiration in Older Adults (hereafter “the Standard”) in 2023 [15], whose core recommendations are highly consistent with these international guidelines.

However, significant gaps persist between evidence and clinical practice. A survey of 965 Chinese nurses found that only 36.52% of stroke patients received guideline-recommended dysphagia screening [16], reflecting not merely a knowledge deficit but a combination of inadequate training in standardized assessment tools, low confidence in clinical decision-making for atypical presentations such as silent aspiration, and cognitive reliance on experiential judgment rather than structured protocols. At the organizational level, the fragmented nature of evidence sources—scattered across guidelines, consensus statements, and evidence summaries from different clinical specialties—creates a translation burden that individual practitioners cannot reasonably overcome without institutional support. Nursing staff awareness of aspiration risk factors and ability to identify silent aspiration remain insufficient. A mapping review noted that interdisciplinary collaboration between nurses and speech-language therapists is limited [17], partly because aspiration prevention has historically been treated as an ancillary nursing task rather than a shared clinical responsibility requiring structured interprofessional workflows. At the system level, standardized processes for individualized feeding plans, positioning management, and food texture modification are lacking, and the absence of integrated clinical pathways means that even when individual nurses possess the requisite knowledge, the practice environment does not reliably support its application. These multi-level barriers—cognitive, organizational, and systemic—collectively sustain a substantial evidence-practice gap that cannot be addressed by knowledge dissemination alone.

The persistence of this multi-level gap suggests that effective aspiration risk prevention requires not isolated knowledge transfer but a theory-informed implementation strategy that simultaneously addresses clinician competency, organizational infrastructure, and care delivery processes. The Ottawa Model of Research Use provides a conceptual framework for this approach: by systematically diagnosing barriers and facilitators across the innovation, adopter, and practice environment dimensions, implementation strategies can be tailored to the specific obstacles identified at each level [18]. We hypothesized that a bundled intervention—combining KAS-model knowledge training to address cognitive barriers, standardized protocol development to resolve organizational fragmentation, and a precision risk-stratification model (“GRADE”) to embed evidence into routine clinical workflows—would operate through three interconnected mechanisms: (1) restructuring clinician knowledge from experiential to evidence-based frameworks, thereby improving risk identification accuracy; (2) reducing the translation burden through pre-synthesized, actionable audit criteria aligned with the national Standard; and (3) creating system-level reinforcement through audit-and-feedback loops and digital early warning, which sustain behavioral change beyond initial training effects. Guided by Proctor et al.’s Implementation Outcomes Taxonomy, we evaluated whether these mechanisms produced observable changes across implementation outcomes (audit criteria adherence), service outcomes (knowledge, identification accuracy, feeding plan compliance), and clinical outcomes (aspiration incidence, patient satisfaction).

Based on the Standard and related evidence, integrating international guideline recommendations, we used the JBI Practical Application of Clinical Evidence System (JBI PACES) to develop and implement a standardized aspiration risk prevention program for hospitalized older adults. The objectives were to: (1) improve nurses’ aspiration-related knowledge and risk identification skills; (2) enhance adherence to prevention measures; (3) reduce aspiration incidence; and (4) improve care quality and patient safety. This project was approved by the Fudan University Center for Evidence-Based Nursing (registration number: ER20251074).

Methods

Study design and reporting standards

This best practice implementation project used a pre–post design guided by the JBI PACES audit-and-feedback framework, encompassing baseline audit, EBP program implementation, and follow-up audit. The pre–post design is the standard paradigm for JBI evidence implementation projects [18], whose core objective is systematic evidence translation rather than testing single-intervention causal effects. We acknowledge that this design does not permit causal inference equivalent to a randomized controlled trial; however, it is the methodologically appropriate framework for evaluating the real-world implementation of complex, multi-component evidence-based programs where randomization would compromise the ecological validity and system-level integration that the intervention requires. To partially mitigate the absence of a concurrent control group, we employed binary logistic regression adjusting for key patient-level confounders, compared baseline characteristics between groups, and maintained identical measurement procedures across both audit cycles. The project was conducted from July 2024 to December 2025. Reporting followed the SQUIRE 2.0 guidelines [19], which are specifically designed for quality improvement and implementation studies using before-after designs.

EBP team establishment

A multidisciplinary EBP team was established. Two nursing department directors served as project leads, responsible for overall coordination and nursing standard development. One head nurse oversaw process management, implementation supervision, and quality control. Seven multidisciplinary healthcare professionals participated as team members: one nutritionist, one rehabilitation physician, two geriatricians, and three senior nurses, who were responsible for topic-specific training, barrier analysis, evidence application, and data collection. In addition, two master’s-level nursing graduate students with formal JBI evidence-based nursing training—one of whom held certification from the Fudan University Center for Evidence-Based Nursing—conducted evidence retrieval, quality appraisal, data analysis, and report preparation.

Evidence retrieval

Clinical questions were formulated using the PIPOST framework: Population—patients aged ≥ 60 years; Intervention—aspiration risk prevention and management; Professionals—nurses and physicians; Outcomes—knowledge, audit criteria adherence, aspiration incidence; Setting—geriatrics and related departments; Type of evidence—guidelines, expert consensus, systematic reviews, and evidence summaries.

Databases searched included BestPractice, UpToDate, JBI EBP Database, National Guideline Clearinghouse, PubMed, the Cochrane Library, Embase, and Chinese databases (CNKI, Wanfang, Chinese Medical Current Contents). The search period was January 2004 to July 2024.

Evidence appraisal and audit criteria development

Guidelines were appraised using AGREE II [20]; expert consensus using JBI critical appraisal criteria (2016) [21]; clinical decision support documents were traced to original sources. Two trained researchers independently conducted appraisals, with a third resolving disagreements.

An evidence appraisal panel (four nurse managers, three senior clinical nurses) graded evidence using JBI Levels of Evidence (2014) [22]. Each statement was evaluated using FAME criteria (Feasibility, Appropriateness, Meaningfulness, Effectiveness), and overlapping evidence was synthesized into audit criteria.

EBP program development

The intervention comprised a multi-component bundled implementation strategy with three core modules: KAS-model knowledge training, standardized management system development, and the “GRADE” precision intervention model. This design is consistent with the definition of complex interventions in implementation science [23], aiming to evaluate the overall program rather than disentangle individual component contributions.

Study setting and participants

This single-center study was conducted in the geriatrics department and selected older-adult units at a tertiary hospital in Shanxi Province, China. A single-center design ensured implementation consistency and process quality.

Healthcare professionals with valid practice licenses were eligible; project team members and those absent during implementation were excluded. Patients aged ≥ 60 years who provided informed consent were included. Evaluation samples differed by level: practitioner level—all on-duty staff (436 at baseline, 412 at follow-up); process level—85 nurses per audit cycle; patient level—hospitalized older adults during each audit period (101 pre-implementation, 103 post-implementation), constituting independent samples from different time periods. Independent sampling was chosen for three reasons: (1) the average hospital stay (12.4 ± 5.8 days) was far shorter than the implementation period, precluding longitudinal tracking of the same patients across both phases; (2) this approach avoided repeated-measurement effects and selective attrition bias that would arise from a cohort design; and (3) consecutive enrollment of all eligible patients during each audit window minimized selection bias by ensuring that sampling reflected the natural patient flow rather than investigator judgment. We recognize that independent sampling introduces the possibility that unmeasured differences between the two patient groups could confound the comparison. To address this, baseline demographic and clinical characteristics were compared statistically (Table 1), confirming no significant differences in age, sex, disease type, consciousness status, nutritional route, or comorbidities (all P > 0.05). Additionally, binary logistic regression was employed to adjust for these variables in the primary outcome analysis.

Table 1.

Comparison of baseline demographic and clinical characteristics between the two groups

Variable	Pre-implementation (n = 101)	Post-implementation (n = 103)	t/χ²/P
Age (years, x̄ ± s)	74.36 ± 7.82	73.91 ± 8.15	t = 0.396, P = 0.693
Sex (male/female, n)	58/43	55/48	χ² = 0.427, P = 0.514
Disease type
Cerebrovascular disease	32 (31.68%)	29 (28.16%)
Respiratory disease	28 (27.72%)	31 (30.10%)	χ² = 1.052, P = 0.789
Digestive disease	22 (21.78%)	25 (24.27%)
Other	19 (18.81%)	18 (17.48%)
Consciousness (alert/impaired, n)	82/19	86/17	χ² = 0.218, P = 0.641
Nutritional route
Oral intake	63 (62.38%)	67 (65.05%)	χ² = 0.396, P = 0.820
Nasogastric/nasointestinal tube	32 (31.68%)	30 (29.13%)
Other	6 (5.94%)	6 (5.83%)

Note: No statistically significant differences were found between groups for any baseline characteristic (P > 0.05)

Baseline audit

A baseline audit (July–August 2024) assessed three levels: (1) Practitioner level: knowledge via a self-developed questionnaire; (2) Process level: adherence via a standardized checklist; (3) Patient level: aspiration events.

EBP program implementation

Barrier and facilitator analysis

Using the Ottawa Model of Research Use, barriers and facilitators to evidence implementation were analyzed across three dimensions: the evidence-based innovation, potential adopters, and the practice environment. Expert panel meetings and semi-structured interviews informed this analysis. Detailed results are presented in Table 2.

Table 2.

Evidence content, audit indicators, and audit methods for standardized prevention and management of aspiration risk in hospitalized older adults

Category	Evidence content	Audit indicators	Audit methods
Management system	1) Establish institutional policies for aspiration prevention in older adults, including the formation of a multidisciplinary aspiration management team (e.g., nurses, dietitians, rehabilitation therapists, physicians) to support clinical decision-making	1) A multidisciplinary aspiration prevention and management system is established in the department	Document review
	2) Provide education and training for healthcare staff involved in aspiration risk management, covering both theoretical knowledge and practical skills	2) Regular training and competency assessments related to aspiration prevention are conducted	Document review
Aspiration risk assessment	3) Conduct early assessment of aspiration risk in hospitalized older adults and implement preventive measures to reduce aspiration and aspiration pneumonia	3) Nurses perform timely aspiration risk assessments and accurately complete assessment records	Record review; on-site observation
	4) Identify high-risk factors for aspiration, including advanced age (> 70 years), mechanical ventilation, dysphagia, impaired consciousness, supine position, head-of-bed elevation < 30°, persistent hiccups, nausea/vomiting, impaired sphincter function, neurological or psychiatric disorders, sedative or muscle relaxant use, and patient transport	4) Nurses accurately assess aspiration risk based on risk factors, clinical manifestations, and criteria, and document findings	Record review; on-site observation
	5) Use the aspiration risk checklist recommended in the group standard to identify existing risks through interviews, observation, and testing		Record review; on-site observation
Dysphagia management	6) Use the Drooling Severity and Frequency Scale and implement graded interventions based on assessment results	5) Nurses apply the drooling assessment scale and implement corresponding interventions	Record review; on-site observation
	7) Apply the Modified Water Swallow Test to assess aspiration risk in older adults	6) Nurses conduct and document the Modified Water Swallow Test and related interventions	Record review; on-site observation
	8) Provide routine swallowing rehabilitation training for older adults	7) Nurses collaborate with rehabilitation physicians to deliver swallowing training and evaluate effectiveness	Interview; on-site observation
Cough effectiveness management	9) Assess cough effectiveness using a semi-quantitative cough strength scale	8) Nurses assess cough strength and implement interventions accordingly	Record review; on-site observation
	10) Guide older adults in respiratory muscle training, such as diaphragmatic and pursed-lip breathing	9) Nurses provide respiratory training guidance and document progress	Interview; on-site observation
	11) For mechanically ventilated patients, maintain cuff pressure at 25–30 cmH₂O using automated or manual monitoring	10) Cuff pressure is monitored every 6 h and maintained within the recommended range	Record review; on-site observation
	12) Adjust cuff pressure during suctioning, low airway pressure, weak spontaneous breathing, and after position changes	11) Cuff pressure is reassessed after suctioning, repositioning, or transport	Record review; on-site observation
	13) Use subglottic suctioning to prevent aspiration in mechanically ventilated patients	12) Subglottic suction endotracheal tubes are used as indicated	Record review; on-site observation
Reflux and positioning management	14) Maintain a semi-recumbent position (30°–45°) for ICU patients receiving mechanical ventilation or enteral nutrition when feasible	13) Head-of-bed elevation is maintained at 30°–45°	Record review; on-site observation
	15) Use post-pyloric feeding for patients at high risk of aspiration	14) Nasointestinal tubes are used for post-pyloric feeding in high-risk patients	Record review; on-site observation
	16) Maintain a semi-recumbent position during tube feeding and for 30 min afterward; avoid turning or suctioning during this period	15) Positioning requirements during and after tube feeding are followed	Record review; on-site observation
	17) Suspend enteral feeding before repositioning, percussion, or suctioning	16) Enteral feeding is paused prior to these procedures	Record review; on-site observation
Oral care management	18) Use standardized oral assessment tools (e.g., BOAS, MCM) every 12 h and document results	17) Oral cleanliness is assessed every 12 h using standardized tools	Record review; on-site observation
	19) Perform oral care with chlorhexidine twice daily for tube-fed patients	18) Chlorhexidine oral care is provided twice daily	Record review; on-site observation
	20) Use disposable suction toothbrushes or soft pediatric toothbrushes for oral care	19) Suction toothbrushes are used for patients with cognitive or consciousness impairment	Record review; on-site observation
Feeding behavior management	21) Maintain upright or semi-recumbent positioning (30°–60°) with neck flexion during oral intake and for 30 min afterward	20) Patients are guided to maintain safe positioning during and after meals	Interview; on-site observation
	22) Recommend soft foods and avoid dry, hard, or sticky foods; use spoons with 5–10 ml capacity	21) Appropriate food texture and feeding tools are used	Interview; on-site observation
	23) Feed hemiplegic patients from the unaffected side; alternate liquids and solids; stop feeding immediately if choking occurs	22) Caregivers are correctly instructed in safe feeding techniques	Interview; on-site observation

Practice change strategies

Based on identified barriers, two graduate researchers developed targeted action strategies informed by the literature and clinical context. These were reviewed and refined by the program development team before implementation.

1. Knowledge training: Using the KAS model, batch training over one month covered aspiration risk factors, identification, prevention, and emergency management.

2. System and protocol development: Institutional policies were established, including aspiration risk and prevention strategy checklists aligned with the Standard.

3. “GRADE” precision intervention model: Graded risk evaluation: Two-version risk assessment tools (alert vs. altered consciousness) stratified patients into low (10–12), moderate (13–18), and high risk (19–23) using an 8-dimension scale. Dynamic assessments occurred within 24 h, upon deterioration, and weekly.

   • Risk-intervention precisely matched: Low-risk patients received routine monitoring and education. Moderate/high-risk patients received bedside alerts, MDT consultation, and targeted interventions across five domains: dysphagia management, cough strength management, gastroesophageal reflux management, oral care, and feeding behavior management.
   • Active quality control: A three-tier system—ward-level daily monitoring, head nurse-led focused audits, and department-level outcome monitoring.
   • Digital early warning: A closed-loop platform for real-time assessment, automated alerts, personalized strategy delivery, and tracking.
   • Extended management: Hospital-to-home continuity including outpatient clinics, telephone follow-up, home visits, and community outreach.

Follow-up audit

A follow-up audit (November–December 2025) used identical methods and criteria across all three levels.

Outcome evaluation framework

Guided by Proctor et al.’s Implementation Outcomes Taxonomy [24, 25], indicators were classified as: (1) implementation outcomes—audit criteria adherence (fidelity) and hospital-wide adoption (penetration); (2) service outcomes—knowledge scores, high-risk identification accuracy, feeding plan compliance; (3) clinical outcomes—aspiration incidence, patient satisfaction.

Outcome definitions and measurement

The main outcome indicators and their operational definitions were as follows:

• Nurses’ knowledge level: Assessed using a self-developed 25-item questionnaire based on the Standard, covering five dimensions: aspiration definition (5 items), risk factors (6 items), assessment methods (5 items), prevention strategies (5 items), and emergency management (4 items), using true/false format with a total score of 100. The questionnaire was developed from an initial pool of 30 items, refined through two rounds of Delphi consultation with five geriatric nursing experts (≥ 10 years of clinical or teaching experience), resulting in deletion of 5 low-scoring items. The scale-level content validity index (S-CVI) was 0.92, with all item-level CVIs ≥ 0.80. Pilot testing among 30 healthcare professionals yielded a Kuder–Richardson 20 (KR-20) coefficient of 0.81 and a two-week test–retest intraclass correlation coefficient (ICC) of 0.85. A knowledge pass rate was calculated based on achieving ≥ 60% of the total score.

• Aspiration risk assessment tool: The aspiration risk assessment tool used in this project was adapted from the risk checklist recommended in the Chinese Nursing Association group standard Prevention of Aspiration in Older Adults (T/CNAS 27–2023) [31], incorporating risk factors consistently identified in international guidelines [12–14, 28–30]. The tool comprises eight assessment dimensions and stratifies patients into low (10–12), moderate (13–18), and high risk (19–23) categories. Content validity was established through the expert panel review during audit criteria development (using FAME criteria, as described above), and the tool demonstrated face validity in pilot testing. However, we acknowledge that a formal psychometric evaluation—including criterion validity against instrumental assessment (VFSS or FEES), inter-rater reliability testing, and assessment of sensitivity and specificity for predicting aspiration events—had not been completed prior to implementation. This represents a methodological limitation: while the tool’s content is grounded in the best available evidence and national standards, its measurement properties in our specific population remain to be formally quantified. Full psychometric validation is planned for the next project cycle and will be reported separately.

• High-risk identification accuracy: The proportion of nurses who correctly identified patients at high aspiration risk. Nurses were presented with 10 standardized clinical scenarios (5 typical and 5 atypical high-risk situations) and asked to judge risk status; accuracy was calculated as number of correct judgments / total scenarios × 100%.

• Audit criteria adherence rate: The proportion of audit criteria for which nurses demonstrated compliance, assessed through a combination of direct observation and nursing record review by two trained researchers independently, with discrepancies resolved by consensus.

• Aspiration incidence: Number of patients experiencing aspiration / total number of hospitalized older adults during the audit period × 100%. Aspiration was identified based on: (a) clinical manifestations during oral or tube feeding, including coughing, respiratory distress, or oxygen saturation decline ≥ 3% from baseline; (b) physician-confirmed aspiration diagnosis; (c) silent aspiration identified through bedside swallowing assessment abnormalities or VFSS findings. All suspected aspiration events were reported on a standardized form within 24 h by trained nurses and independently verified by two research team members.

• Feeding plan compliance rate: The proportion of patients whose oral or enteral feeding plans met evidence-based requirements (including positioning, food consistency, and feeding rate).

• Patient satisfaction: Assessed using a hospital-developed patient satisfaction questionnaire.

Statistical analysis

Data were double-entered into EpiData and verified for accuracy before analysis. All statistical analyses were performed using SPSS 26.0. Categorical variables were expressed as frequencies and percentages and compared between groups using chi-square tests or Fisher’s exact test when expected cell frequencies were below 5. Continuous variables were expressed as mean ± standard deviation and compared using independent-sample t-tests. To strengthen the comparability of the pre- and post-implementation groups, binary logistic regression was performed for the primary outcome (aspiration incidence), with group assignment as the independent variable and age, sex, consciousness status, nutritional route, and number of comorbidities entered as covariates. Results were reported as adjusted odds ratios (OR) with 95% confidence intervals (CI). A two-sided P < 0.05 was considered statistically significant.

Results

Baseline characteristics

A total of 101 pre-implementation and 103 post-implementation patients were enrolled. No significant differences were found in age, sex, disease type, consciousness status, nutritional route, or comorbidities (P > 0.05) (Table 1).

Evidence synthesis and audit criteria

A total of 1,247 records were initially identified through systematic searching. After deduplication, title/abstract screening, and full-text review, 19 publications were included: two clinical decision support documents [26, 27], three guidelines [28–30], six expert consensus statements and standards [31–36], six evidence summaries [37–42], and two systematic reviews/meta-analyses [43, 44]. The literature selection process is shown in Fig. 1 (PRISMA flow diagram).

Fig. 1.

PRISMA flowchart

Quality appraisal yielded the following results: all three guidelines scored > 60% across AGREE II domains (recommendation grade B); all six expert consensus documents received affirmative ratings on all JBI appraisal items, indicating high quality; both systematic reviews scored ≥ 9/11 on JBI quality appraisal, demonstrating good methodological quality. From these, 36 evidence statements were extracted, covering aspiration risk assessment and multidimensional management strategies. After panel discussion using the FAME criteria, 13 statements were excluded due to insufficient clinical feasibility or low evidence levels, and 23 best-evidence statements were selected and translated into 22 audit criteria (Table 2).

Baseline audit results

Baseline audit findings were as follows: (1) Practitioner level: among 436 healthcare professionals, the mean knowledge score was 56.72 ± 8.29, with a knowledge pass rate of 26.72%; (2) Process level: among 298 patients audited, overall adherence to prevention measures was 24.67%; (3) Patient level: the aspiration incidence was 27.72%.

Barriers and facilitators

Key barriers included: complexity of the evidence system and difficulty in workflow integration; insufficient nurse competency in assessment tools and specialized skills; and lack of standardized assessment and intervention protocols at the ward level. Key facilitators included: the program’s foundation in the Standard and high-quality research evidence; strong administrative leadership and commitment; and multidepartmental collaboration providing resource support (Table 3).

Table 3.

Analysis of barriers and facilitators to evidence-based practice for standardized prevention and management of aspiration risk in hospitalized older adults

Category	Evidence-based changes	Potential adopters	Practice setting
Barriers	• Complexity of the evidence base, involving multidimensional risk assessment and stratified interventions (e.g., swallowing, cough, and oral care), making workflow integration challenging • High operational requirements for some standardized assessment tools (e.g., Modified Water Swallow Test, semi-quantitative cough strength scale) • Certain implementation pathways require further localization, including multidisciplinary collaboration models and information-based closed-loop processes	• Insufficient knowledge and skills among nurses regarding assessment tools and aspiration prevention techniques (e.g., respiratory muscle training) • Limited acceptance and resistance to change, with some nurses perceiving increased documentation and assessment workload • Patient-related factors, such as cognitive impairment, hearing loss, and poor adherence among older adults, increasing the difficulty of education and intervention delivery	• Lack of unified departmental workflows for aspiration risk assessment and intervention • Insufficient or delayed availability of aspiration-prevention resources (e.g., suction oral care toothbrushes, foods with varying viscosities) • Inadequate training and supervision mechanisms, with limited routine feedback and quality monitoring
Facilitators	• Authoritative evidence sources, including national group standards and high-quality studies, aligned with existing basic nursing procedures (e.g., admission assessment, oral care) • Clear evidence translation, with evidence operationalized into structured management pathways and strategy checklists, facilitating clinical understanding and execution	• Strong leadership support from department directors and head nurses, providing organizational endorsement for practice change • Structured capacity-building through tiered training based on the Knowledge–Attitude–Skills (KAS) model, incorporating workshops and case discussions • Diversified patient education approaches (e.g., QR-code videos, visual education materials), improving patient engagement and adherence	• Organizational and resource support through collaboration among nursing administration, clinical departments, logistics, and information technology units • Established multidisciplinary collaboration with rehabilitation and nutrition teams, supporting coordinated care • Adequate infrastructure, including multimedia education equipment and ongoing development of smart wards, enabling integration of information-based and intelligent workflow support

Intervention implementation

Targeted interventions were implemented across five dimensions based on the mechanisms of aspiration: (1) Dysphagia management: introduction of a drooling severity and frequency scale, administration of the modified water swallow test, and collaborative swallowing rehabilitation with rehabilitation physicians; (2) Cough strength management: application of a modified semi-quantitative cough strength scoring scale, combined with artificial airway cuff management and respiratory muscle training; (3) Gastroesophageal reflux management: development of positioning protocols and enteral feeding route selection pathways; (4) Oral care management: implementation of the Beck Oral Assessment Scale and suction toothbrush for oral hygiene; (5) Feeding behavior management: development of an unsafe feeding behavior management protocol and patient/family education on appropriate feeding practices and food consistency.

Aspiration incidence

Following implementation, the aspiration incidence decreased from 27.72% (28/101) to 13.59% (14/103), with a statistically significant difference (χ² = 5.907, P = 0.015) (Table 4).

Table 4.

Comparison of aspiration incidence before and after implementation [n (%)]

Group	n	Aspiration [n (%)]	No aspiration [n (%)]
Pre-implementation	101	28 (27.72)	73 (72.28)
Post-implementation	103	14 (13.59)	89 (86.41)
χ²		5.907
P		0.015

The absolute risk reduction (ARR) was 14.13% points (95% CI: 2.91–25.35), corresponding to a number needed to treat (NNT) of approximately 7.1 (95% CI: 3.9–34.4), indicating that for every seven patients managed under the post-implementation protocol, one additional aspiration event was prevented. The relative risk reduction was 51.0%.

To control for potential confounders, binary logistic regression was performed with aspiration occurrence as the dependent variable, group (pre = 0, post = 1) as the independent variable, and age, sex, consciousness status, nutritional route, and number of comorbidities as covariates. After adjustment, the post-implementation group showed significantly lower aspiration incidence after covariate adjustment, indicating that the effect was independent of these clinical characteristics (Table 5).

Table 5.

Binary logistic regression analysis for aspiration incidence

Variable	B	SE	Wald χ²	P	Adjusted OR (95% CI)
Group (post vs. pre)	−1.023	0.421	5.902	0.015	0.359 (0.157–0.820)
Age	0.034	0.022	2.389	0.122	1.035 (0.991–1.080)
Sex (female vs. male)	−0.187	0.362	0.267	0.605	0.829 (0.408–1.686)
Consciousness (impaired vs. alert)	0.892	0.415	4.621	0.032	2.440 (1.082–5.502)
Nutritional route (tube vs. oral)	0.756	0.387	3.816	0.051	2.130 (0.997–4.550)
Constant	−3.247	1.685	3.714	0.054	—

Note: Dependent variable: aspiration occurrence (0 = no, 1 = yes). After adjusting for age, sex, consciousness status, and nutritional route, group remained a significant predictor (adjusted OR = 0.359, 95% CI: 0.157–0.820, P = 0.015), indicating an approximately 64.1% lower aspiration incidence in the post-implementation group after covariate adjustment. Impaired consciousness was an independent risk factor for aspiration (OR = 2.440, P = 0.032)

Nurses’ knowledge

After implementation, the mean knowledge score among 412 on-duty healthcare professionals improved from 56.72 ± 8.29 to 85.03 ± 8.40 (t = − 49.38, P < 0.001). The two survey cohorts did not differ significantly in professional title composition, departmental distribution, or years of work experience (P > 0.05), confirming sample comparability.

The mean improvement of 28.31 points corresponded to a large effect size (Cohen’s d = 3.39, 95% CI: 3.09–3.69), far exceeding the conventional threshold for large effects (d = 0.80). While this magnitude is notable, it should be interpreted with the caveat that pre-implementation scores were low (mean 56.72), suggesting substantial room for improvement, and that post-implementation measurement may have been influenced by questionnaire familiarity among partially overlapping respondent cohorts.

High-risk identification accuracy

The accuracy of high-risk aspiration identification improved from 34.65% (35/101) to 81.55% (84/103) (χ² = 46.150, P < 0.001) (Table 6).

Table 6.

Comparison of nurses’ high-risk aspiration identification accuracy before and after implementation [n (%)]

Group	n	Correct [n (%)]	Incorrect [n (%)]
Pre-implementation	101	35 (34.65)	66 (65.35)
Post-implementation	103	84 (81.55)	19 (18.45)
χ²		46.150
P		< 0.001

Note: Identification accuracy refers to the proportion of nurses who correctly identified patients at high aspiration risk

The absolute improvement in identification accuracy was 46.90% points (from 34.65% to 81.55%), with an odds ratio of 8.37 (95% CI: 4.55–15.39), representing a large effect. This improvement suggests that the structured training using standardized scenarios, combined with the GRADE risk stratification model, facilitated a shift in identification practices from intuitive to criterion-based approaches.

Audit criteria adherence

Different types of audit criteria demonstrated varying patterns of improvement. System-level criteria (criteria 1–7, 9, 18) were fully adopted after implementation, indicating that institutional and procedural measures were readily embraced at the organizational level. Positioning and airway management criteria (criteria 10–17) showed significant improvement, with criteria 14 and 17 both reaching 100%. However, criterion 11 (cuff pressure monitoring) had a high baseline (88.24%) and improved to 94.12%, without reaching statistical significance (P = 0.176). Feeding behavior management criteria (criteria 20–23) demonstrated the greatest magnitude of improvement, yet absolute adherence remained lower than other categories (e.g., criterion 21 increased from 29.41% to 72.94%; criterion 23 from 24.71% to 74.12%), suggesting that interventions involving patient behavior change are more challenging to implement. Criteria 13 and 19 showed 100% adherence at both time points and were excluded from statistical comparison. With the exception of criterion 11 (P = 0.176), all other criteria showed statistically significant improvement (P < 0.05) (Table 7).

Table 7.

Healthcare providers’ adherence to audit criteria before and after implementation

Audit criterion	Pre (n = 85)	Post (n = 85)	χ²	P
8	25 (29.41%)	63 (74.12%)	34.02	< 0.001
10	33 (38.82%)	71 (83.53%)	35.76	< 0.001
11	75 (88.24%)	80 (94.12%)	1.83	0.176
12	52 (61.18%)	75 (88.24%)	16.47	< 0.001
14	71 (83.53%)	85 (100%)	15.26	< 0.001
15	33 (38.82%)	61 (71.76%)	18.66	< 0.001
16	51 (60.00%)	78 (91.76%)	23.43	< 0.001
17	68 (80.00%)	85 (100%)	18.89	< 0.001
20	0 (0%)	85 (100%)	170.00	< 0.001
21	25 (29.41%)	62 (72.94%)	32.33	< 0.001
22	0 (0%)	77 (90.59%)	140.75	< 0.001
23	21 (24.71%)	63 (74.12%)	41.51	< 0.001

Note: Specific audit criteria content is described in Table 2. Fisher’s exact test was used for criteria with zero-cell frequencies. Criterion 11 did not reach statistical significance (P = 0.176); all other criteria showed statistically significant differences (P < 0.05)

Audit criteria adherence was assessed through a combination of direct observation and nursing record review. To minimize measurement bias, assessments were performed independently by two trained researchers, with discrepancies resolved by consensus. Given the open nature of the implementation project, the Hawthorne effect cannot be entirely excluded, particularly for criteria reliant on direct observation, where nurses aware of being observed may have demonstrated higher adherence.

Feeding plan compliance

Feeding plan compliance improved from 41.58% (42/101) to 77.67% (80/103) (χ² = 27.624, P < 0.001) (Table 8).

Table 8.

Comparison of feeding plan compliance before and after implementation [n (%)]

Group	n	Compliant [n (%)]	Non-compliant [n (%)]
Pre-implementation	101	42 (41.58)	59 (58.42)
Post-implementation	103	80 (77.67)	23 (22.33)
χ²		27.624
P		< 0.001

The absolute improvement in feeding plan compliance was 36.09% points, with an odds ratio of 4.93 (95% CI: 2.72–8.92). The NNT for achieving one additional compliant feeding plan was 2.8, indicating high practical efficiency in this domain.

Patient satisfaction

Patient satisfaction scores improved from 80.35 ± 7.62 to 92.16 ± 4.85 (t = 13.177, P < 0.001) (Table 9).

Table 9.

Comparison of patient satisfaction scores before and after implementation (x̄ ± s)

Group	n	Satisfaction score	t / P
Pre-implementation	101	80.35 ± 7.62
Post-implementation	103	92.16 ± 4.85
t			13.177
P			< 0.001

The mean improvement of 11.81 points corresponded to a large effect size (Cohen’s d = 1.84, 95% CI: 1.52–2.16). While satisfaction scores are subject to social desirability bias and may have been influenced by patients’ awareness of the quality improvement project, the magnitude of improvement suggests a meaningful perceived difference in care quality.

Discussion

Cross-level outcome integration

Viewed through Proctor’s Implementation Outcomes Taxonomy, the results demonstrate a coherent pattern across the three outcome levels, though the observational design precludes definitive causal attribution. At the implementation level, audit criteria adherence improved substantially across most domains, with system-level and positioning criteria reaching near-complete adoption. This implementation fidelity appears to have created the necessary conditions for service-level changes: nurses who practiced under standardized protocols demonstrated markedly higher knowledge scores (Cohen’s d = 3.39) and identification accuracy (OR = 8.37), which in turn supported more appropriate clinical decision-making. At the service-to-clinical interface, the improvement in feeding plan compliance (OR = 4.93) represents a translation of enhanced knowledge and identification accuracy into patient-care behavior, and the concurrent reduction in aspiration incidence (adjusted OR = 0.359, ARR = 14.13%) is consistent with the hypothesis that improved adherence to evidence-based prevention measures reduces aspiration risk. The one notable exception—feeding behavior criteria, which showed the largest relative improvement yet the lowest absolute adherence—suggests that patient-level behavioral change remains the most resistant link in the implementation chain, and may partially explain why the aspiration incidence, while significantly reduced, was not eliminated. This cross-level pattern is consistent with implementation science theory [24, 25] but should be interpreted as suggestive of a plausible causal pathway rather than as confirmation of causality, given the pre–post design.

Improved knowledge and behavioral change

Nurses’ knowledge scores improved by 28.31 points, and high-risk identification accuracy nearly tripled, from 34.65% to 81.55%. Conventional didactic approaches typically yield retention rates of 60–70% that tend to erode over time [45]. The more durable gains observed in the present study may be attributable to the KAS model, which weaves together knowledge acquisition, attitudinal reinforcement, and hands-on skills practice rather than treating them in isolation [46]. The national Standard served as a common content backbone, reducing the kind of practice variation that arises when individual nurses rely on personal experience alone. Meanwhile, automated digital alerts for moderate- and high-risk patients helped bridge the gap that manual judgment inevitably leaves in busy clinical environments [47].

That said, these gains should not be taken at face value. Because the two survey cohorts partially overlapped, repeated exposure to the questionnaire and growing familiarity with its items may have contributed to the score increase. More importantly, without a concurrent unexposed control ward, we cannot rule out the influence of parallel activities—hospital-wide quality campaigns, national policy shifts in geriatric care, or updated dysphagia management guidance—that may have coincided with our implementation period. Even so, a 28-point jump far exceeds what conventional training studies have reported, and the steep rise in identification accuracy points to genuine improvement in clinical reasoning rather than simple recall.

Improved adherence to prevention measures

Audit criteria adherence rose substantially, consistent with a shift from experience-driven to protocol-guided aspiration management. Earlier work in this area concentrated on boosting awareness, yet sustained compliance has remained stubbornly difficult to achieve [48–50]. What distinguished the present project was the translation of scattered evidence into a coherent set of audit criteria and step-by-step clinical workflows that nurses could follow in real time [51]. Layering on regular audit-and-feedback cycles gave the quality control team a mechanism for catching and correcting deviations early [52], and the deliberate involvement of bedside nurses in protocol design helped turn externally imposed rules into something staff felt ownership over.

The pattern of improvement, however, was far from uniform. Institutional and procedural criteria were adopted across the board once management endorsed them—an unsurprising finding, since such measures hinge on organizational decisions rather than individual behavior. Positioning criteria likewise reached full compliance, probably because the actions involved are straightforward and easy to verify. Feeding behavior criteria, by contrast, showed the steepest climb in relative terms yet remained the weakest in absolute adherence. Patient-side factors—cognitive decline, limited cooperation, and insufficient family engagement—are the most plausible explanations [50, 53]. One might argue that the Hawthorne effect inflated adherence across the board; but if observation bias were operating uniformly, feeding behavior criteria should have benefited just as much as other categories. The fact that they lagged behind suggests that the differential pattern reflects real differences in how difficult each type of intervention is to sustain at the bedside.

Reduced aspiration incidence

Aspiration incidence fell by more than half, from 27.72% to 13.59%, a trajectory consistent with findings from related implementation studies [50]. After adjusting for age, sex, consciousness status, and nutritional route, the association held (adjusted OR = 0.359, 95% CI: 0.157–0.820). By measuring aspiration events directly rather than relying on the downstream endpoint of aspiration pneumonia [54], this study provides a more proximal indicator of aspiration risk [51].

A few methodological caveats are in order. Aspiration was identified on clinical grounds combined with physician confirmation, not through VFSS or FEES on every patient; silent aspiration events were almost certainly missed in both periods. Because the same criteria and reporting procedures applied before and after implementation, however, this underascertainment should not distort the between-group comparison. A subtler concern is detection bias: post-implementation nurses, now more alert to aspiration, may have flagged cases that would previously have gone unnoticed, artificially raising the post-implementation rate and thus making our effect estimate conservative rather than inflated. We also did not formally assess inter-rater reliability for aspiration identification; if different physicians judged events before and after the intervention, diagnostic thresholds may have shifted. Kappa statistics or similar measures should be incorporated in future work. Beyond patient-level confounders addressed through logistic regression, several system-level factors merit consideration. During the implementation period, no hospital-wide quality improvement campaigns specifically targeting aspiration prevention were conducted outside this project, and no changes in departmental staffing ratios, admission policies, or case-mix composition were documented. However, we cannot entirely exclude the influence of broader temporal trends—such as evolving national geriatric care policies, increased general awareness of patient safety following institutional accreditation activities, or seasonal variation in respiratory illness admissions—that might have independently influenced aspiration incidence. The concurrent introduction of the digital early warning platform, while integral to the intervention bundle, also constitutes a system-level change whose independent contribution cannot be isolated from the other intervention components. These system-level factors represent inherent limitations of the single-center pre–post design and underscore the need for multicenter studies with concurrent control sites to disentangle program effects from temporal and contextual trends. Finally, the pre–post design leaves open the possibility that temporal trends or concurrent policy changes played a role. These results are best read as observed improvements associated with the implementation, not as established causal effects.

Multi-component intervention evaluation

Because the intervention was delivered as a bundled package, the present design cannot tease apart which components drove which outcomes. Craig et al. [23] made this point within the MRC complex intervention framework: the components of such programs are interdependent, and pulling them apart may destroy the very synergy that makes the whole thing work. Still, the differential adherence patterns offer some indirect clues. The rapid, complete uptake of system-level criteria suggests that organizational endorsement was a necessary foundation. The high compliance with positioning measures reflects the power of clearly standardized, easily monitored procedures. And the persistent shortfall in feeding behavior adherence tells us that policy change and knowledge training, on their own, are not enough to shift patient-level behaviors. Answering the question of how much each piece contributed with any precision would require factorial or stepped-wedge designs—an avenue worth pursuing.

Proctor framework evaluation

The parallel rise in feeding plan compliance and patient satisfaction signals that this care pathway improved not only safety but also the quality of the care experience. Viewed through Proctor’s Implementation Outcomes Taxonomy [24, 25], the project yielded gains at all three tiers, with effect sizes that merit clinical attention: implementation outcomes showed near-universal adoption of system-level criteria; service outcomes demonstrated large effects in knowledge (Cohen’s d = 3.39) and identification accuracy (OR = 8.37); and clinical outcomes showed a meaningful reduction in aspiration incidence (ARR = 14.13%, NNT ≈ 7). The logical coherence across these levels—from implementation fidelity through service improvement to clinical benefit—strengthens the plausibility of a programmatic effect, even though the pre–post design does not permit definitive causal conclusions. The weakest link in this chain, feeding behavior adherence, identifies a specific target for the next implementation cycle: interventions that more effectively engage patients and family caregivers in behavioral change. What remains unaddressed is sustainability. The follow-up audit was conducted shortly after the program ended, and whether these gains will hold once the research team steps back is an open question. Implementation science consistently warns that without ongoing organizational reinforcement, hard-won adherence can erode. A third audit at six to twelve months would help clarify this.

This study was reported in line with the SQUIRE 2.0 guidelines endorsed by the EQUATOR Network, to support transparency and completeness of the methodological account.

Limitations

This study has several limitations that should be acknowledged.

First, the pre–post design without a concurrent control group limits causal inference. Although this is the standard methodological paradigm for JBI PACES projects and binary logistic regression was used to adjust for key patient-level confounders, system-level factors such as temporal trends and concurrent policy changes cannot be fully excluded.

Second, this was a single-center study conducted in a well-resourced tertiary teaching hospital with established multidisciplinary collaboration infrastructure, dedicated quality control personnel, and access to digital information systems. These conditions may not be representative of primary care, community hospital, or resource-limited settings where staffing ratios are lower, multidisciplinary support is less available, and digital infrastructure is absent. The transferability of our findings is therefore uncertain for settings that differ substantially in organizational capacity. To facilitate external validity assessment, we have described the implementation context in detail, including the staffing structure, resource requirements, and institutional prerequisites for each intervention component, following the recommendations of the SQUIRE 2.0 reporting guidelines. Multicenter replication across diverse healthcare settings—including secondary hospitals and community care facilities—is needed to determine which program components are setting-dependent and which are generalizable.

Third, some outcomes—particularly audit criteria adherence—relied on direct observation and nursing record review, which may have introduced Hawthorne effects. Additionally, aspiration was identified through clinical assessment rather than instrumental gold standards (VFSS or FEES) for every patient, potentially missing silent aspiration events. However, because identical diagnostic criteria and reporting procedures were applied in both audit cycles, this systematic underascertainment does not compromise the validity of the between-group comparison.

Fourth, the multi-component bundled strategy does not permit disaggregation of individual component effects. Determining the independent contribution of each intervention module would require more complex study designs, such as factorial or stepped-wedge trials.

Fifth, the aspiration risk assessment tool, although adapted from the risk checklist in the Chinese Nursing Association group standard, had not undergone a formal psychometric evaluation in our specific population prior to implementation; full validation of this tool — including content validity, inter-rater reliability, and predictive validity against VFSS findings — is planned for the next cycle of this project.

Conclusions

Based on the JBI PACES framework and Ottawa Model of Research Use, and reported per SQUIRE 2.0 guidelines, this project implemented evidence-based aspiration risk prevention for hospitalized older adults. The program was associated with improved adherence and lower aspiration incidence, with positive changes across all three Proctor taxonomy levels. Given the pre–post design limitations, these improvements should be interpreted as observed changes associated with the implementation program rather than confirmed causal effects. The coherent pattern of improvement across Proctor’s three outcome levels is consistent with programmatic impact but does not constitute causal proof. Future research should employ more rigorous designs to strengthen causal inference—specifically, stepped-wedge cluster randomized trials that allow sequential rollout across multiple sites while maintaining concurrent controls, or multicenter controlled before-after studies with matched comparison wards. Such designs would also permit formal mediation analysis to test the hypothesized causal pathway from implementation fidelity through service improvement to clinical outcomes. Additionally, formal psychometric validation of the aspiration risk assessment tool and longer-term follow-up (6–12 months post-implementation) are needed to assess both measurement quality and sustainability of the observed improvements.

Abbreviations

EBP

Evidence-based practice

JBI PACES

JBI Practical Application of Clinical Evidence System

KAS

Knowledge–Attitude–Skills

FAME

Feasibility, Appropriateness, Meaningfulness, Effectiveness

MDT

Multidisciplinary team

H2H

Hospital-to-home

VFSS

Videofluoroscopic swallowing study

FEES

Fiberoptic endoscopic evaluation of swallowing

Author contributions

Dan Liu, Binrong Zhang, and Jiajia Xu jointly conceived and designed the study. Dan Liu, Sisi Yuan, Jianghong Zhang, Fenfen Li, and Xiumin Ye were responsible for data collection and the implementation of the evidence-based practice. Dan Liu and Binrong Zhang conducted the literature review and data analysis. Dan Liu drafted the initial manuscript. Jiajia Xu and Xiuhong Zhang critically reviewed and revised the manuscript for important intellectual content. All authors discussed the findings and contributed to the final manuscript.

Funding

This work was supported by the Nursing Research Fund of Shanxi Bethune Hospital (Grant No. 2024YH23).

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to ethical restrictions and the need to protect participant privacy but are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Shanxi Bethune Hospital (Approval No. YXLL-2025-189). The study was conducted in accordance with the Declaration of Helsinki. Informed consent to participate was obtained from all participants in the study, and this is clearly stated in the manuscript.

Consent for publication

Not applicable. No identifying images or personal or clinical details of participants are included in this manuscript.

Competing interests

The authors declare no competing interests.