Early Acute‐Phase Energy Intake Is Independently Associated With Swallowing Decline in Older Patients With Aspiration Pneumonia: A Retrospective Cohort Study

ABSTRACT

Aim

Although swallowing decline is common after aspiration pneumonia treatment, the impact of early acute‐phase energy intake on swallowing outcomes remains poorly understood. We aimed to clarify the association between acute‐phase energy intake (Days 3–7) and swallowing decline at discharge in older patients with aspiration pneumonia.

Methods

This single‐center retrospective cohort study included patients aged ≥ 65 years with aspiration pneumonia who maintained oral intake before admission (Functional Oral Intake Scale [FOIS] Score 4–7). The exposure was mean energy intake during Days 3–7 (kcal/kg/day). Swallowing decline was defined as ≥ 1‐point FOIS reduction at discharge. We compared the incidence of swallowing decline across energy‐intake quartiles and fitted a multivariable logistic regression model, with covariates selected using a directed acyclic graph. ROC and restricted cubic spline analyses were performed.

Results

Of 200 patients, 109 (54.5%) experienced swallowing decline at discharge. The incidence decreased across energy‐intake quartiles, from 68.0% to 36.0% (p for trend < 0.001). In multivariable analysis, lower energy intake was independently associated with swallowing decline (adjusted odds ratio 0.927, 95% confidence interval 0.885–0.971, p = 0.001). ROC analysis identified an optimal cutoff of 15.3 kcal/kg/day (area under the curve 0.648; sensitivity 69.6%; specificity 57.4%).

Conclusions

Insufficient early acute‐phase energy intake was independently associated with swallowing decline in older patients with aspiration pneumonia. Because energy and protein intake were strongly correlated, energy intake below 15.3 kcal/kg/day—possibly reflecting overall macronutrient insufficiency—may serve as a practical reference point for identifying at‐risk patients, although prospective validation is needed.

Early acute‐phase energy intake (Days 3–7) is independently associated with swallowing decline in older patients with aspiration pneumonia. Energy intake below 15.3 kcal/kg/day during this period may serve as a practical reference point for identifying patients at risk, highlighting nutritional management as a potentially modifiable factor for preserving swallowing function at discharge.

1. Introduction

Aspiration pneumonia is one of the leading causes of hospitalization and death among older adults in super‐aged societies [1, 2]. Although advances in acute care have improved survival outcomes, swallowing decline after treatment is common, and difficulty maintaining oral intake can have long‐term consequences for activities of daily living (ADL) and quality of life [3, 4]. Factors associated with swallowing outcomes after aspiration pneumonia include age, preadmission ADL, cognitive function, and comorbidities [3]. Dysphagia itself is also known to increase the risk of pneumonia recurrence and mortality [3]. Most of these factors are difficult to modify and offer limited opportunity for intervention.

Nutritional management is one of the few clinically modifiable factors in patients with aspiration pneumonia; however, the impact of early acute‐phase energy intake (Days 3–7) on swallowing outcomes remains insufficiently understood. During the acute catabolic phase, increased catabolism driven by cytokine release predisposes patients to negative energy balance [5]. Inadequate energy and protein intake contributes to the development of hospital‐associated sarcopenia [6, 7, 8], and it has also been implicated as a factor in acute sarcopenia [9]. The swallowing‐related muscles are skeletal muscles, and they are therefore susceptible to hypercatabolism and disuse, which reduce muscle mass and strength [10]. Conversely, adequate energy intake during this period may help preserve swallowing‐related muscle mass and function. The hypothesis that acute‐phase nutritional insufficiency adversely affects swallowing outcomes is clinically important, yet it remains insufficiently investigated.

Even among older patients who had maintained oral intake before admission, reduced swallowing reserve due to aging and underlying disease is common, making them vulnerable to nutritional insufficiency during acute illness. However, it remains unclear whether early acute‐phase energy intake (Days 3–7) is independently associated with subsequent swallowing decline in this population. Clarifying this relationship could identify a practical and modifiable target for preserving swallowing function during hospitalization. We therefore examined the association between acute‐phase energy intake (Days 3–7) and swallowing decline at discharge in older patients with aspiration pneumonia who had maintained oral intake before admission (FOIS 4–7) [11], and sought a clinically relevant cutoff value. We hypothesized that lower early energy intake would be associated with a greater risk of swallowing decline.

2. Methods

2.1. Study Design and Participants

This study was a single‐center retrospective cohort study conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [12]. Patients discharged between April 1, 2022 and March 31, 2026 with an admission diagnosis of aspiration pneumonia (including suspected cases) were consecutively enrolled. Eligible patients were identified by the medical records department, which extracted cases based on the ICD‐10 code assigned to the admission diagnosis. The study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of our hospital (approval number: 2025‐007). Because the study was retrospective, clinical research information was published on the hospital website, and patients were given the opportunity to opt out.

2.2. Eligibility Criteria

Among patients diagnosed with pneumonia based on imaging and clinical symptoms [13], those with features indicating aspiration risk (dysphagia, neurological disease, or impaired consciousness) and clinically diagnosed with aspiration pneumonia by the attending physician per Japanese Respiratory Society guidelines [14] were eligible. Eligibility was restricted to patients who had maintained oral intake before admission (FOIS 4–7); those requiring alternative nutrition (FOIS 1–3) were excluded to avoid a floor effect that could limit the detection of further swallowing decline. Patients were excluded if they met any of the following criteria: (1) in‐hospital death, because swallowing at discharge could not be assessed; (2) hospitalization of 7 days or fewer, because energy intake over Days 3–7 could not be fully measured; (3) age younger than 65 years; (4) a preadmission FOIS Score of 3 or lower; or (5) missing body weight data, because energy intake per body weight could not be calculated.

2.3. Variable Definitions

Exposure: The exposure was mean energy intake during the acute phase (Days 3–7; kcal/kg/day), calculated as the sum of all nutritional inputs, including oral intake, enteral nutrition, and parenteral nutrition. During Days 1–2, hemodynamic instability and intake fluctuations due to diagnostic procedures and fasting orders may make nutritional data unrepresentative of actual nutritional status. The ESPEN guidelines also recommend gradually increasing nutritional delivery after the initial acute catabolic phase [5]. Days 3–7 were therefore chosen, a period during which treatment protocols have stabilized and systematic nutritional management can begin. To minimize day‐to‐day variability from intravenous fluids and oral intake, the mean over Days 3–7 (kcal/kg/day) was used as the exposure variable.

Outcome: The outcome was swallowing decline, defined as a reduction of one or more points in FOIS score at discharge compared with the preadmission score [3, 11]. The outcome (FOIS at discharge) and the baseline (preadmission FOIS) were assessed retrospectively and independently by a registered dietitian and a speech‐language‐hearing therapist, based on dietary and food‐texture records in the charts. Discrepancies were resolved by consensus. Inter‐rater reliability was excellent (Cohen's weighted κ, 0.992). The Day 3 FOIS was assessed independently by a single registered dietitian, based on meal‐order records and nursing notes. These records objectively reflected the prescribed diet and actual intake and were considered less susceptible to inter‐rater variability than assessments requiring clinical judgment.

Covariates. The following covariates were collected: sex, age, body mass index (BMI), protein intake (mean g/kg/day over Days 3–7), C‐reactive protein (CRP), pneumonia severity (A‐DROP score) [15], comorbidity burden (Charlson Comorbidity Index [CCI]) [16], nutritional status (malnutrition assessed by the GLIM criteria) [17], preadmission FOIS, Day 3 FOIS, Day 3 ADL (Barthel Index [BI]) [18], dysphagia rehabilitation (performed or not during Days 3–7), and percentage weight loss. Percentage weight loss was calculated as (admission − discharge weight)/admission weight, reflecting in‐hospital weight change. Weight loss for the GLIM criteria was assessed separately, based on preadmission weight records documented in the medical chart.

2.4. Statistical Analysis

Patients were grouped into quartiles of mean Days 3–7 energy intake, and baseline characteristics were compared across quartiles. Continuous and ordinal variables were compared using the Kruskal–Wallis test, and categorical variables using the chi‐squared test. Because missing data accounted for 1% or less across all variables, a complete‐case analysis was performed.

Multivariable logistic regression was performed with swallowing decline as the dependent variable to evaluate the independent association of energy intake. No a priori sample‐size calculation was performed; all eligible consecutive patients were enrolled. Variance inflation factors (VIF) were calculated to assess multicollinearity, and at least 10 events per variable were maintained to avoid model overfitting. Covariates were selected using a directed acyclic graph (DAG) to distinguish confounders from variables on or downstream of the causal pathway [19]. The adjustment set comprised age, A‐DROP score, CRP, nutritional status (malnutrition assessed by the GLIM criteria), preadmission FOIS, and Day 3 BI. Dysphagia rehabilitation was not included, because it is likely a consequence of early swallowing status and adjusting for it would risk overadjustment. Sex, CCI, BMI, and percentage weight loss were excluded because of events‐per‐variable constraints and, for BMI and weight loss, multicollinearity. Protein intake, strongly correlated with energy intake (Pearson's r = 0.866, p < 0.001), was excluded from the model and evaluated in a separate sensitivity analysis, as was Day 3 FOIS. Time zero was defined as the end of Day 7; because all included patients were alive and hospitalized until this point, energy intake was always ascertained before outcome assessment.

For the secondary analysis, a decline of two or more FOIS points at discharge was used as the dependent variable. A receiver operating characteristic (ROC) analysis was performed, and the optimal cutoff was determined using the Youden Index. To assess the shape of the dose–response relationship, a restricted cubic spline (RCS) with four knots at the 5th, 35th, 65th, and 95th percentiles (6.0, 13.6, 19.8, and 30.2 kcal/kg/day) was fitted; this analysis was exploratory and was not a test of a prespecified nonlinear hypothesis. For reference, energy intake quartiles were compared (chi‐squared test), and the linear trend was assessed (Cochran–Armitage trend test).

Effect modification by baseline nutritional status was assessed primarily by adding an interaction term (energy intake × malnutrition) to the model. Stratified analyses for malnourished patients and those without malnutrition, using the same covariates as the model, were also performed; these were not prespecified and are reported as exploratory and hypothesis‐generating. To address the exclusion of in‐hospital deaths, a composite outcome of swallowing decline or in‐hospital death occurring after Day 7 was analyzed as a sensitivity analysis, using the same adjustment set as the primary model. Of the 80 in‐hospital deaths, 51 occurred after Day 7 and met the other eligibility criteria; these were added to the primary cohort for this analysis. All analyses used a two‐sided 5% significance level and were performed using EZR [20].

3. Results

Of 324 patients admitted with aspiration pneumonia, we excluded 124 (80 in‐hospital deaths, 18 hospitalized for 7 days or fewer, 14 with a preadmission FOIS of 3 or lower, 9 with missing body weight data, and 3 aged under 65 years), and 200 patients remained for analysis (Figure 1). Swallowing decline at discharge occurred in 109 patients (54.5%). Baseline characteristics across intake quartiles are presented in Table 1. The proportion of patients with swallowing decline decreased stepwise across increasing quartiles, from 68.0% in Q1 and 66.0% in Q2 to 46.0% in Q3 and 36.0% in Q4 (Cochran–Armitage test for trend, p < 0.001). Day 3 FOIS, Day 3 BI, and protein intake were higher in the upper quartiles, whereas preadmission FOIS, age, sex, BMI, percentage weight loss, CRP, A‐DROP score, and CCI showed no consistent gradient. Malnutrition by the GLIM criteria, which could be assessed in 198 patients, was common in every quartile (at least 86%). Dysphagia rehabilitation was performed in 49 patients (24.5%) and was more frequent in the lower‐intake quartiles (Q1: 36.0%, Q2: 30.0%, Q3: 18.0%, Q4: 14.0%; p = 0.036).

FIGURE 1.

Flowchart of patient selection. FOIS, functional oral intake scale.

TABLE 1.

Baseline characteristics by quartile of acute‐phase (Days 3–7) energy intake.

Variable	Q1 (n = 50)	Q2 (n = 50)	Q3 (n = 50)	Q4 (n = 50)	p
Age (years), mean (SD)	91.6 (6.5)	88.4 (7.9)	87.5 (7.7)	89.4 (6.1)	0.028
Sex (male), n (%)	26 (52.0)	35 (70.0)	29 (58.0)	23 (46.0)	0.093
BMI (kg/m2)	19.7 [18.3, 22.2]	20.1 [17.6, 22.3]	20.7 [18.4, 23.8]	18.5 [16.8, 20.1]	0.003
A‐DROP score	3 [2, 3]	2 [2, 3]	2 [2, 3]	2 [1, 3]	0.407
CRP (mg/dL)	6.6 [1.7, 9.9]	6.1 [1.2, 9.8]	5.6 [2.1, 10.9]	4.4 [1.2, 8.3]	0.334
Charlson Comorbidity Index	1 [0, 1]	0 [0, 2]	0 [0, 1]	0 [0, 1]	0.540
GLIM malnutrition, n/assessed (%)	44/50 (88.0)	45/50 (90.0)	35/49 (71.4)	47/49 (95.9)	0.003
Preadmission FOIS	5 [4, 6]	6 [4.3, 6.8]	6 [5, 6]	6 [4, 6.8]	0.286
Day 3 FOIS	2.5 [1, 3]	3 [2, 3]	3 [3, 4.8]	4 [3, 5]	< 0.001
Day 3 Barthel Index	0 [0, 0]	0 [0, 5]	5 [0, 23.8]	5 [0, 33.8]	< 0.001
Energy intake (kcal/kg/day)	7.9 [6.1, 9.7]	14.2 [12.2, 15.3]	19.6 [17.8, 20.3]	27.0 [24.3, 29.8]	< 0.001
Protein intake (g/kg/day)	0.28 [0.17, 0.44]	0.54 [0.36, 0.74]	0.85 [0.53, 0.98]	1.24 [1.02, 1.41]	< 0.001
Dysphagia rehabilitation, n (%)	18 (36.0)	15 (30.0)	9 (18.0)	7 (14.0)	0.036
Percentage weight loss (%)	7.1 [1.5, 10.8]	5.9 [0.9, 10.7]	2.1 [0.0, 7.1]	0.0 [0.0, 3.1]	< 0.001
Length of stay (days)	48 [27.5, 76.0]	37 [15.5, 76.3]	19.5 [14.0, 47.8]	17 [12.3, 35.5]	< 0.001
Swallowing decline, n (%)	34 (68.0)	33 (66.0)	23 (46.0)	18 (36.0)	0.002

Note: Data are median [interquartile range] for continuous and ordinal variables, mean (SD) for age, or n (%) for categorical variables. Quartiles of mean Days 3–7 energy intake: Q1 ≤ 11.1, Q2 11.2–16.4, Q3 16.5–21.6, and Q4 ≥ 21.7 kcal/kg/day. Comparisons across quartiles used the Kruskal–Wallis test (continuous and ordinal variables), one‐way ANOVA (age), and the chi‐squared test (categorical variables). GLIM malnutrition was available for 198 patients (n = 49 in Q3 and Q4).

Abbreviations: BMI, body mass index; CRP, C‐reactive protein; FOIS, Functional Oral Intake Scale; GLIM, Global Leadership Initiative on Malnutrition; SD, standard deviation.

In the multivariable logistic regression model, lower energy intake was independently associated with swallowing decline (adjusted OR 0.927 per kcal/kg/day, 95% confidence interval [CI] 0.885–0.971, p = 0.001; Table 2). This model was constructed to estimate the effect of energy intake; the coefficients for the other covariates are therefore not interpreted as independent causal effects [21].

TABLE 2.

Multivariable logistic regression for swallowing decline.

Variable	Adjusted OR	95% CI	p
Energy intake (per 1 kcal/kg/day)	0.927	0.885–0.971	0.001
Age (years)	0.952	0.908–0.998	0.043
CRP (mg/dL)	1.030	0.974–1.080	0.325
A‐DROP score	0.872	0.650–1.170	0.360
GLIM malnutrition (vs. normal)	1.550	0.584–4.100	0.380
Preadmission FOIS	2.570	1.800–3.670	< 0.001
Day 3 Barthel Index	0.975	0.958–0.993	0.006

Note: The model was built to estimate the effect of energy intake; the coefficients for the remaining covariates are not interpreted as independent causal effects. Covariates were selected with a directed acyclic graph, and dysphagia rehabilitation was not entered, as adjusting for it would risk overadjustment.

Abbreviations: CI, confidence interval; CRP, C‐reactive protein; FOIS, Functional Oral Intake Scale; GLIM, Global Leadership Initiative on Malnutrition; OR, odds ratio.

We then assessed the robustness of this finding through secondary and sensitivity analyses. Using a decline of two or more FOIS points as the outcome, energy intake showed a stronger association (adjusted OR 0.887, 95% CI 0.839–0.937, p < 0.001). ROC analysis identified an optimal cutoff of 15.3 kcal/kg/day (area under the curve [AUC] 0.648, 95% CI 0.571–0.724; sensitivity 69.6%; specificity 57.4%; Figure 2); intake below this threshold was associated with swallowing decline in an unadjusted analysis (OR 2.97, 95% CI 1.65–5.32, p < 0.001). The RCS showed no significant nonlinearity (p = 0.101; Figure 3), indicating a broadly linear negative association. To address possible reverse causation, we added Day 3 FOIS to the model; the association of energy intake was maintained (adjusted OR 0.934, 95% CI 0.891–0.979, p = 0.004). Substituting protein intake for energy intake with all other covariates unchanged, protein intake was also significantly associated with swallowing decline (adjusted OR 0.456, 95% CI 0.213–0.979, p = 0.044); because energy and protein intake were strongly correlated (r = 0.866), their individual contributions cannot be separated. To address the exclusion of in‐hospital deaths, we examined a composite outcome of swallowing decline or in‐hospital death occurring after Day 7 (157 events among 249 patients with complete data); lower energy intake remained associated with the outcome (adjusted OR 0.912, 95% CI 0.875–0.950, p < 0.001), consistent with the primary analysis.

FIGURE 2.

Receiver operating characteristic (ROC) curve for acute‐phase energy intake predicting swallowing decline. We determined the optimal cutoff value (15.3 kcal/kg/day; sensitivity 69.6%, specificity 57.4%) using the Youden Index. The area under the curve (AUC) was 0.648 (95% confidence interval, 0.571–0.724).

FIGURE 3.

Restricted cubic spline curve for the association between acute‐phase energy intake and the risk of swallowing decline. The solid blue line represents the estimated log odds, and the dashed red lines represent the 95% confidence intervals. The test for nonlinearity was not significant (p = 0.101). This analysis was exploratory.

In an exploratory analysis, the interaction between energy intake and malnutrition status was not significant (p for interaction = 0.668). Stratified estimates within nutritional‐status subgroups are provided as Supporting Information.

4. Discussion

Insufficient energy intake during the early acute phase (Days 3–7) was independently associated with swallowing decline at discharge, even after adjusting for pneumonia severity (A‐DROP), inflammatory markers (CRP), and early functional status (Day 3 BI). This suggests that the association reflects more than disease severity alone. When we additionally adjusted for Day 3 FOIS in the sensitivity analysis, the association remained significant, indicating that the finding is robust to potential reverse causation.

That preadmission FOIS was higher in the swallowing‐decline group may appear counterintuitive. However, even among older patients who had maintained oral intake alone before admission, latent reductions in swallowing reserve due to aging and underlying sarcopenia are common, and acute illness or nutritional deficits can readily lead to decompensation of swallowing function. Moreover, because of a ceiling effect in this ordinal scale, patients with higher baseline function have more room to decline and are thus more likely to show a detectable decline. Good preadmission swallowing function therefore does not mean that nutritional management can be given lower priority during the acute phase. In addition, conservative food‐texture modifications driven by clinical concerns about aspiration risk or reduced consciousness may have contributed to the observed FOIS changes independently of true functional decline.

These findings are consistent with prior literature. Nutritional management that incorporates rehabilitation contributes to swallowing recovery in older patients with sarcopenic dysphagia [22], and adequate energy intake has been associated with clinical trajectories and swallowing outcomes even in end‐stage aspiration pneumonia [23]. Although the study populations differ in disease stage, both suggest that nutritional intake may influence swallowing function [24]. The present study extends this evidence to the early acute phase (Days 3–7), immediately after the initial hypercatabolism, when intervention is clinically feasible.

In aspiration pneumonia, inflammation‐driven hypercatabolism adversely affects the swallowing‐related muscles, and it has been implicated in the development of acute sarcopenia [7, 8, 10]. In this cohort, 171 of 200 patients (85.8%) met the GLIM criteria for malnutrition at baseline. Acute illness combined with insufficient energy intake may amplify catabolism in the swallowing‐related muscles, making functional recovery more difficult. In the exploratory stratified analyses, the association between energy intake and swallowing decline appeared stronger among malnourished patients than among those without malnutrition; however, the interaction term was not significant (p for interaction = 0.668). This difference may therefore be due to chance and should be confirmed in future studies, particularly given the small nonmalnourished subgroup (n = 29).

Dysphagia rehabilitation reached a minority of patients (24.5%) and was more frequent in the lower‐intake quartiles. We did not enter rehabilitation into the model, because our DAG indicated that it is a likely consequence of early swallowing status, and adjusting for it would risk overadjustment. Its imbalance across exposure groups is consistent with confounding by indication, whereby clinicians preferentially provided rehabilitation to patients who were already at higher risk of decline [25], This pattern can make any benefit of rehabilitation harder to detect [26]. We instead addressed early swallowing status, the likely common cause, by including Day 3 BI in the model and Day 3 FOIS in a sensitivity analysis; the association of energy intake with swallowing decline held in both. Energy intake therefore appears important for preserving swallowing function during the acute phase, regardless of whether patients received rehabilitation.

The cutoff that the ROC analysis identified (15.3 kcal/kg/day) is substantially below standard energy requirements for older adults (27–30 kcal/kg/day) [27] and rehabilitation‐nutrition recommendations (≥ 30 kcal/kg/day) [28]. The discriminative performance was moderate (AUC 0.648; sensitivity 69.6%; specificity 57.4%), so this value should not be applied as a definitive threshold for individual patients. It may instead serve as a practical reference point for minimum energy targets during the hemodynamically unstable acute phase, and intake below this level could act as a simple screening indicator of risk. A large‐scale study showing that adequate energy and amino acid delivery reduces in‐hospital mortality in aspiration pneumonia [29] supports the clinical relevance of these findings.

This study has several limitations. As a single‐center retrospective study, it cannot establish causality, and residual reverse causation cannot be excluded despite adjustment for Day 3 FOIS. FOIS reflects prescribed food texture rather than swallowing function assessed by videofluoroscopy or videoendoscopy, and conservative food‐texture downgrades made out of concern for aspiration, rather than for true functional decline, may have inflated the observed changes. Oral intake was recorded by visual estimation, a validated but imprecise method [30], We used body weight–based intake (kcal/kg/day) for its reproducibility; although we also calculated achievement relative to estimated requirements, neither measure fully captures each patient's relative degree of energy insufficiency. Because energy and protein intake were strongly correlated (r = 0.866), their individual contributions could not be separated.

The exclusion of in‐hospital deaths and short‐stay patients limits generalizability to more severely ill populations, although a composite outcome including post‐Day 7 death gave consistent results. Finally, as a single‐center study in Japan, our findings may not generalize to settings with different healthcare resources, rehabilitation practices, or cultural contexts.

5. Conclusion

In older patients with aspiration pneumonia, insufficient energy intake during the early acute phase (Days 3–7) was independently associated with swallowing decline at discharge. Because energy and protein intake were strongly correlated, energy intake below 15.3 kcal/kg/day—possibly reflecting overall macronutrient insufficiency—may serve as a practical reference point for identifying patients at risk, although this indicator has limited precision and prospective validation is needed.

Author Contributions

M.S. conceived and designed the study, acquired the data, performed the statistical analyses, and drafted the manuscript. Y.Y. assessed FOIS scores and contributed to data collection. J.K., Y.H., and S.T. assisted with data interpretation. All authors critically revised the manuscript for important intellectual content and approved the final version for submission.

Funding

The authors have nothing to report.

Ethics Statement

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board of Shobara Red Cross Hospital (approval number: 2025‐007). Given the retrospective nature of the study, the requirement for informed consent was waived. Patients were provided the opportunity to opt out of participation via information published on the hospital website.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Stratified multivariable analysis of energy intake and swallowing decline by nutritional status.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.