Stroke-associated pneumonia: pathophysiology, diagnosis and management strategies

SAP arises from a complex interplay between neurological impairment and post-stroke immune dysregulation. Strokes involving the cerebral cortex, brainstem, or cerebellum frequently disrupt swallowing function (dysphagia) by damaging neural pathways responsible for mastication and bolus transit. Lesions in the precentral gyrus can cause contralateral facial and oropharyngeal motor deficits, while brainstem strokes may delay the swallowing reflex and diminish oral-pharyngeal sensation. These dysfunctions significantly increase the risk of aspiration, allowing food, fluids, or pathogens to enter the lower respiratory tract and lead to pneumonia (6).

Acute stroke also triggers systemic immunosuppression, primarily through stroke-induced immunodepression syndrome (SIDS). In the early phase, immune cells release cytokines in response to tissue injury, activating afferent vagal pathways that signal central immune-regulatory centres such as the hypothalamus. This initiates anti-inflammatory signalling via the efferent vagus nerve to suppress further cytokine production. However, prolonged stimulation leads to immune exhaustion and weakened host defences. SIDS is driven by the coordinated activation of three key systems: the sympathetic nervous system (SNS), the parasympathetic nervous system (PNS) and the hypothalamic-pituitary-adrenal (HPA) axis (7, 8).

SNS activation causes sustained catecholamine release (epinephrine, norepinephrine, and dopamine), which suppresses lymphocyte counts and disrupts early lymphocyte activation. This catecholamine-mediated immunosuppression plays a critical role in the diminished antibacterial immune response observed following stroke (8). Concurrently, the PNS releases acetylcholine, which binds to α7 nicotinic acetylcholine receptors (α7nAChR) on macrophages, inhibiting the release of pro-inflammatory cytokines and weakening pulmonary immune defences, thereby increasing susceptibility to respiratory infections (8, 9). Meanwhile, the HPA axis releases glucocorticoids in response to stress and inflammation. Excess glucocorticoid production can cause lymphocytopenia and disrupt immune balance, further compromising host immunity (8). Collectively, these neurologic and immunologic alterations undermine host immunity and heighten the risk of pneumonia, forming a central mechanism in the pathogenesis of SAP.

Several additional neurological and medical factors contribute to the risk of SAP. Among these, stroke type is a critical determinant, as patients with haemorrhagic stroke face a higher risk of SAP compared to those with ischemic stroke (6). An altered level of consciousness, especially a Glasgow Coma Scale (GCS) score below 8 on admission, is strongly associated with increased SAP incidence (6). Stroke severity, as indicated by a National Institutes of Health Stroke Scale (NIHSS) score above 15, is also a significant predictor (6, 10). Dysphagia, a common and serious post-stroke complication, markedly elevates SAP risk due to impaired swallowing and the likelihood of silent aspiration. This results from disrupted neural control of the swallowing network, leading to dysfunction in the oropharyngeal, laryngeal and oesophageal muscles. Pharyngeal phase deficits, in particular, can hinder bolus transit and cause secretion pooling near the airway, substantially increasing the risk of aspiration and respiratory infections such as pneumonia (11–13).

From a medical perspective, several demographic and clinical factors have been identified as significant contributors to SAP. Advanced age, particularly in patients over 75 years, and male gender are both recognised as non-modifiable risk factors (10, 11, 14). Ageing increases susceptibility to swallowing dysfunction due to physiological decline, such as reduced tongue strength, impaired pharyngeal constriction and inadequate pharyngeal shortening. These changes compromise swallowing efficiency, heightening the risk of aspiration and subsequent pneumonia (13). Cardiovascular and metabolic conditions such as atrial fibrillation, fasting hyperglycaemia and severe hypertension are consistently linked to increased SAP risk (11, 15). Additional comorbidities, including congestive heart failure (CHF), chronic pulmonary disease, renal failure, anaemia, and diabetes mellitus, also increase the risk of vulnerability. CHF and other disorders affecting pulmonary circulation may predispose patients to infection due to pulmonary oedema and compromised respiratory mechanisms (6, 14, 15). Lower serum iron levels have also been linked to higher SAP risk, with studies suggesting an L-shaped relationship. Although the mechanisms remain unclear, impaired iron homoeostasis potentially driven by altered macrophage function and microbial iron uptake may contribute to this association (16). Furthermore, the use of mechanical ventilation and nasogastric tubes significantly increases the risk of SAP (6, 14, 17). Mechanical ventilation impairs mucociliary clearance, facilitating bacterial colonisation of the lower respiratory tract (17). Nasogastric tube insertion increases the risk of pneumonia in stroke patients by promoting gastroesophageal reflux and impairing the natural protective mechanisms of the oropharynx. This can lead to microaspiration of infected gastric contents and facilitate colonisation by pathogenic bacteria, particularly in the absence of normal chewing and swallowing, which otherwise help reduce oral bacterial load (17). These risk factors can be categorised as modifiable or non-modifiable, as outlined in Table 1.

Several risk factors have been independently associated with an increased risk of multidrug-resistant bacterial infections in patients with SAP. According to a multivariate analysis by Xu et al. (18), late-onset SAP, intensive care unit (ICU) admission, presence of an indwelling gastric tube, prolonged hospitalisation, prophylactic antibiotic use, and impaired consciousness significantly elevate the likelihood of Multidrug-Resistant (MDR) pathogen involvement.

The Pneumonia in Stroke Consensus (PISCES) Group recommends using modified Centres for Disease Control and Prevention (CDC) criteria to classify as either ‘probable’ or ‘definite’. Probable SAP is diagnosed when all CDC criteria are met, but initial chest X-ray (CXR) and serial/repeat CXR are non-confirmatory, and no alternative diagnosis or explanation is available. Definite SAP requires the presence of typical CXR changes (on at least one) alongside fulfilment of all CDC criteria. This differentiation addresses the challenge that early stage pneumonia in stroke patients may not exhibit typical CXR findings (1). The CDC criteria for the diagnosis of pneumonia are based on a combination of clinical, radiographic and laboratory findings. A diagnosis typically requires the presence of a new or progressive and persistent infiltrate, consolidation, or cavitation on chest radiography, accompanied by at least one of the following: fever (>38°C); leukocytosis or leukopenia; or altered mental status in adults aged ≥70 years without another recognised cause. Additionally, at least two of the following signs or symptoms should be present: new onset of purulent sputum or a change in the character of sputum; new or worsening cough, or dyspnoea, or tachyponea; rales, crackles, or bronchial breath sounds; or worsening gas exchange (1).

Early identification of SAP remains a significant clinical challenge due to the non-specific nature of its early symptoms and their overlap with other post-stroke complications. These overlapping presentations, combined with the limitations of conventional diagnostic tools, such as imaging and laboratory tests, often delay accurate diagnosis. Nevertheless, early and precise detection is crucial for initiating timely antibiotic therapy, which can significantly reduce morbidity, mortality and hospital length of stay (19).

Several clinical scoring systems have been developed to aid in risk stratification as outlined in Table 2. These tools aim to predict SAP risk based on clinical and demographic parameters and are primarily intended to assist clinicians in identifying high-risk patients for closer monitoring or prophylactic treatment (19). However, the clinical utility of these scores remains uncertain. Most have not been evaluated for their impact on clinical decision making, such as their ease of use, the time required for completion, or their influence on clinician behaviour. Moreover, none have been robustly validated in terms of improving patient outcomes, such as reducing SAP incidence or mortality. Sensitivity and specificity at commonly used cut-offs vary widely and may not meet the thresholds required for routine clinical implementation (19).

A study by Tu et al. (21) demonstrated that all evaluated scoring algorithms, specifically Kwon (cutoff = 3), Chumbler (cutoff = 2), acute ischaemic stroke-associated pneumonia score (AIS-APS) (cutoff = 11), age, atrial fibrillation, dysphagia, sex, stroke severity (A2DS2) (cutoff = 7), and prestroke independence, sex, age, National Institutes of Health Stroke Scale (ISAN) (cutoff = 11) exhibited low positive predictive values (PPV) for SAP at their respective optimal thresholds. This highlights their limited utility in reliably identifying high-risk patients. In contrast, all scores consistently achieved very high negative predictive values (NPV), ranging from 98% to 100%, indicating strong performance in excluding individuals unlikely to develop SAP. These findings suggest that such scoring systems may be more useful for excluding low-risk patients than for definitively identifying high-risk individuals.

The A2DS2 scoring system demonstrates a practical advantage in predicting SAP, with most studies consistently applying a cutoff score of ≥5, yielding moderate sensitivity and specificity (22). In a study by Zhang et al. (23), patients with high A2DS2 scores (5–10) had nearly ninefold increased odds of developing SAP compared to those with lower scores (0–4). Additionally, the risk of in-hospital death was more than seven times higher in the high-score group. These findings underscore a strong association between elevated A2DS2 scores, increased susceptibility to SAP and poorer clinical outcomes.

The AIS-APS has been consistently recognised as one of the most reliable tools for predicting SAP. A systematic review by Ni et al. (22) reported that AIS-APS demonstrated a pooled sensitivity of 79% and specificity of 74%, outperforming other commonly used models such as A2DS2 and ISAN. Furthermore, Zhang et al. (20) found that AIS-APS achieved an area under the curve (AUC) of 0.79, reflecting strong diagnostic accuracy, in comparison to lower AUCs for A2DS2 (0.74) and ISAN (0.76). Despite its strong performance, a standardised cutoff for AIS-APS has not been established (22).

A systematic review by Ni et al. (22) reported that the ISAN score demonstrated a pooled sensitivity of 76% and specificity of 74% in predicting SAP. The diagnostic odds ratio (DOR) was 9, indicating that patients classified as high risk were nine times more likely to develop SAP compared to those at low risk. However, the ISAN scoring system lacks a standardised cutoff value, with different studies reporting a wide range of optimal thresholds. This variability may limit its consistency and applicability in clinical practice.

A study by Wang et al. (24) compared six internationally recognised scoring systems for predicting SAP in patients with AIS. Among these, the Preventive Antibacterial Therapy in Acute Ischaemic Stroke (PANTHERIS) score demonstrated the weakest discriminatory ability, with the lowest Area Under the Receiver Operating Characteristic curve (AUROC) of 0.660. Despite a high sensitivity of 86.0% at a cutoff value of 1, suggesting it effectively identifies most SAP cases, it had a low specificity of 36.6% leading to a high rate of false positives. Its PPV and NPV were 0.067 and 0.020, respectively, indicating limited precision. This score may be useful in clinical contexts where minimising missed SAP cases is prioritised over reducing false alarms. In contrast, the Kwon Pneumonia score showed a more balanced performance, with an AUROC of 0.703, sensitivity of 63.4% and specificity of 68.9% at a cutoff of 2. It achieved a PPV of 0.097 and an NPV of 0.027, supporting its role as a moderately effective tool for both identifying and excluding SAP cases.

A study by Gong et al. (25) identified an optimal cutoff value of 1 for the Chumbler score in SAP. At this threshold, the score demonstrated a sensitivity of 71.49% and a specificity of 52.93%, with a PPV of 21.66% and an NPV of 91.07%. These findings suggest that the Chumbler score is relatively effective in excluding SAP in low-risk patients, as indicated by its high NPV. However, its ability to accurately identify high-risk individuals is limited due to a low PPV. Among the models evaluated, the Chumbler score exhibited the lowest overall discriminatory performance.

In light of the challenges associated with diagnosing SAP, recent research has increasingly focused on the exploration of biological markers as promising tools for its early detection. Several biomarkers including international normalised ratio (INR) (26), procalcitonin (PCT) (27), C-reactive protein (CRP) (28), neutrophil-to-lymphocyte ratio (NLR) (29), lactic dehydrogenase to albumin ratio (LAR) (30) have been found to be significantly elevated in patients who develop SAP. These markers reflect systemic inflammation, infection and tissue injury, suggesting a potential role in the early identification and risk prediction of SAP. While promising, these findings require further validation in large, prospective studies to determine their predictive accuracy and practical applicability in clinical settings.

Elevated INR at admission in patients with AIS exemplifies the link between coagulation abnormalities and inflammatory processes underlying SAP. Higher INR reflects enhanced coagulation activation and consumption of clotting factors triggered by brain injury and immune dysregulation after stroke. This heightened inflammatory and pro-coagulant state not only increases susceptibility to infections such as SAP but is also commonly observed in patients with more severe strokes, who face greater risks of mortality and poor recovery due to complications like impaired consciousness and respiratory compromise. Thus, INR serves as an important early marker connecting stroke severity with SAP development and prognosis (26).

Complementing INR’s reflection of coagulation and inflammation, PCT offers a more direct measure of bacterial infection and systemic inflammatory burden. Produced by epithelial cells in response to infection, elevated PCT levels in AIS patients signify an active inflammatory response that exacerbates neuronal damage, endothelial dysfunction and blood-brain barrier breakdown. These pathophysiological changes contribute to worse clinical outcomes and higher mortality in SAP patients. Therefore, PCT not only predicts the risk and severity of SAP but also indicates a poor prognosis linked to amplified inflammation and coagulation abnormalities following stroke (27).

In a similar vein, CRP, an acute-phase protein synthesised in response to pro-inflammatory cytokines, provides valuable prognostic information regarding SAP. Elevated CRP levels on admission and during the early post-stroke phase correlate with greater diagnostic accuracy and increased severity of pneumonia. Moreover, CRP levels have been shown to independently predict worse functional outcomes and higher mortality rates. CRP levels exceeding 25 mg/L demonstrate high sensitivity, while levels above 65 mg/L show excellent specificity for early SAP diagnosis in severe stroke patients. Furthermore, elevated CRP independently predicts worse functional outcomes and higher mortality, with each 10 mg/L increase significantly raising the risk of developing SAP. These findings underscore CRP’s utility in early detection and as a prognostic marker of poor SAP outcomes (31).

NLR further expands our understanding of SAP by integrating cellular immune changes after stroke. Elevated NLR reflects the combined effects of post-stroke immunosuppression and systemic inflammation, characterised by neutrophil proliferation and lymphocyte apoptosis that predispose patients to infections like SAP. Additionally, higher NLR correlates with stroke severity and infarct size, both of which are established predictors of SAP and adverse clinical outcomes. Importantly, NLR may also reveal early, subclinical pneumonia not yet clinically apparent, highlighting its role in the timely identification of high-risk patients. Thus, NLR serves as a comprehensive marker linking immune dysregulation, stroke severity and SAP prognosis (29).

Building upon these inflammatory and immune biomarkers, LAR integrates systemic inflammation with nutritional status, two critical determinants of SAP risk and recovery potential. Elevated Lactate Dehydrogenase (LDH) levels reflect ongoing inflammation and immune suppression, while low albumin indicates poor nutritional reserves and weakened host defences. The combined LAR thus offers a superior predictive value for SAP development, severity and outcomes compared to either marker alone. Moreover, both LDH and albumin influence neuroinflammatory pathways and stroke severity, indirectly modulating SAP risk and patient prognosis. Therefore, LAR not only predicts SAP occurrence but also serves as a marker for poor clinical outcomes, including increased mortality and impaired functional recovery (30).

SAP is closely associated with poor clinical outcomes, including a significantly increased risk of mortality (2). Reported mortality among SAP patients reaches approximately 19% at 30 days and 44% at 6 months. A study by Tinker et al. (38) identified several independent predictors of both short-and intermediate-term mortality, including advanced age, haemorrhagic stroke subtype and pre-existing disability. Comorbid conditions such as dementia, lung cancer and a prior history of pneumonia were additionally associated with increased risk of 6-month mortality. Furthermore, elevated levels of inflammatory markers, particularly CRP, were found to correlate with greater functional impairment at discharge, highlighting the prognostic significance of systemic inflammation in SAP (38).

Admission of patients with SAP to the ICU is typically not based solely on the diagnosis of pneumonia itself, but rather on the presence of acute clinical decompensation requiring intensive support. Several precipitating factors beyond general predisposing risks, such as advanced age, dysphagia, or comorbidities, have been identified as significant triggers for ICU admission. These include acute respiratory failure necessitating mechanical ventilation, severe neurological impairment (e.g. NIHSS scores >15), reduced consciousness (GCS ≤8) and the development of sepsis or septic shock. Both higher NIHSS and lower GCS on admission were independent predictors of in-hospital mortality – underscoring their role in necessitating ICU-level care (39, 40).