In this single-center retrospective cohort study, we included patients consecutively admitted to the Neurorehabilitation Unit of the ‘Istituti Clinici Scientifici Maugeri’ of Pavia, Italy, dedicated to the management and care of patients in the post-acute phase after sABI between December 1, 2019 and December 31, 2022. The inclusion criteria of the study were: (1) age ≥18 years; (2) admission to our rehabilitation unit within 3 months from sABI to continue clinical care and rehabilitation programs started at the Intensive Care Unit. We excluded patients with positive respiratory viral panel results in the absence of positive bacterial cultures or with mechanical ventilation. During the hospital stay, all patients underwent intensive multidisciplinary rehabilitation for up to 6 months consisting of an individual 90-min daily treatment program, 6 days a week.

The study design complied with the ethical guidelines of the Declaration of Helsinki, and was approved by the local Ethics Committee of Istituti Clinici Scientifici Maugeri (ICS Maugeri, ref. 2214 CE, June 19, 2018). All participants, or their authorized representatives, signed a written informed consent.

We collected from electronic clinical records the following data: age at injury, sex, LOS expressed in days of hospitalization, etiology of sABI, comorbidities (e.g., hypertension, diabetes mellitus, chronic renal failure, ischemic cardiopathy, hypothyroidism), neurological and functional assessments, presence of invasive devices (e.g., tracheostomy tube, nasogastric tube, percutaneous endoscopic gastrostomy), occurrence of HAP, death during hospitalization.

We defined HAP as any acute lower respiratory tract infection acquired after at least 48 h of admission to hospital (12), confirmed with positive bacterial microbiological culture and through chest radiography or computed tomography.

We used the Glasgow Coma Scale (GCS) to assess brain injury severity on admission. The GCS is a standardized system for evaluating the degree of neurological impairment and level of consciousness in all types of brain injury (16). The total score reflect the severity of brain damage, with scores of 3–8 indicating a severe injury, 9–12 a moderate injury, and 13–15 a mild injury (16).

Functional assessment was evaluated at admission and discharge with the Functional Independence Measure (FIM), the Levels of Cognitive Functioning (LCF), and the Glasgow Outcome Scale-extended (GOS-E). FIM is a standardized instrument designed to assess and monitor progress in functional status through 13 motor (FIM-m) and 5 cognitive items (FIM-c) in terms of independence (17). Each of the 18 items is graded on a scale of 1–7 based on the level of independence. The final total score ranges from 18 to 126, where 18 represents complete dependence/total assistance, and 126 represents complete independence (17). The LCF is a descriptive ordinal scale developed to monitor and classify deficit of responsiveness and degree of cognitive impairment in response to any stimulus (1, no response; 2, generalized; 3, localized; 4, confused–agitated; 5, confused-inappropriate; 6, confused–appropriate; 7, automatic–appropriate; 8 purposeful–appropriate) (18). Finally, GOS-E is an eight-point ordinal scale that broadly reflects the patient’s functional outcome with references to their dependence on others and community reintegration (1, death; 2, persistent vegetative state; 3, lower severe disability; 4, upper severe disability; 5, lower moderate disability; 6, upper moderate disability; 7, lower good recovery; 8, upper good recovery) (19).

Data cleaning was performed before the analysis process, which did not reveal any missing or incorrect data. Categorical variables were expressed as absolute numbers and percentages, and differences were compared with the chi-square test, whilst non-normally distributed data were presented as the median and interquartile range (IQR), and differences were compared with the Mann–Whitney U test. Data were tested for normality of distribution via the Shapiro–Wilk test and log-transformed when needed, to correct for skewness. Multivariable logistic regression analysis was used to identify the potential risk factors of HAP and mortality during hospitalization. The Odds ratio (OR), 95% confidence interval (95% CI), and related significant values obtained from the regression analysis were reported.

A multivariable linear regression analysis was carried out to identify the variables’ potential predictive role on functional outcomes in terms of FIM, LCF, and GOS-E at discharge from the Neurorehabilitation Unit. The models included combinations of independent variables encompassing sex (M = 0, F = 1), age (<65 = 0, ≥65 = 1), severity and etiology (non-traumatic = 0, traumatic = 1) of sABI, LOS, presence of comorbidities, and occurrence of HAP. Moreover, each outcome measure was adjusted for its baseline value. Statistical analyses were performed using IBM SPSS Statistics version 21 (Somers, NY, USA). The threshold for statistical significance was set at p < 0.05.

The cohort included 169 patients (Figure 1). Among these, 78 (46.2%) developed HAP during hospitalization.

Demographic and clinical characteristics of patients included are presented in Table 1.

The comparison of baseline characteristics shows no significant differences between the patients with and without HAP related to age, etiology (traumatic or non-traumatic), LOS, and presence of clinical comorbidities considered. On the contrary, a higher prevalence of HAP was observed in patients with tracheostomy tube (p = 0.0003) and percutaneous endoscopic gastrostomy (p = 0.0034) on admission.

Figure 2 shows that patients who developed HAP during hospitalization presented a worse assessment on admission in terms of GCS (p < 0.0001), FIM (p < 0.0001), LCF (p = 0.001), and GOS-E (p < 0.0001).

The potential risk factors identified for the development of HAP are reported in Table 2.

The multivariable logistic analysis showed that male sex (OR 0.313, p = 0.003), severity of brain injury on admission (OR 0.825, p = 0.011), and presence of nasogastric tube or percutaneous endoscopic gastrostomy (OR 14.336, p = 0.024) were significantly associated with the development of HAP during inpatient rehabilitation, regardless of age, injury etiology, length of stay, and comorbidities.

Table 3 shows the possible predictors of recovery and functional outcome at discharge identified by the multivariable linear regression analysis.

Occurrence of HAP was a main predictor of worse recovery at discharge in terms of FIM (p = 0.003) and LCF (p = 0.009). Other relevant predictors of poor functional outcome included older age, severity of neurological impairment on admission, non-traumatic etiology, chronic renal failure, and LOS.

Death during inpatient rehabilitation was documented in 36 (21.3%) patients after sABI, with a higher prevalence (p = 0.0006) in subjects that developed HAP (25 patients) compared to the other group (11 patients) (not shown). Table 4 reports the potential risk factors for mortality identified during hospitalization among the variables considered in our cohort.

Occurrence of HAP was independently associated with a threefold increase in mortality (OR 3.088; p = 0.020) during inpatient rehabilitation. Other risk factors of mortality were older age at diagnosis (OR 4.493; p = 0.001), non-traumatic etiology (OR 0.260; p = 0.034), and severity of brain injury (OR 0.834; p = 0.029) on admission.

This study has some limitations. First, the retrospective design implies a review of clinical files not originally aimed at collecting data for research, with a risk of selection and recall biases, and possible missing information. In particular, the severity of dysphagia is only reported through indirect indicators (presence of nasogastric tube or percutaneous endoscopic gastrostomy). However, the study sample size was very large, and the well-characterized cohort of patients was hospitalized in a tertiary referral center. Additionally, we included variables such as age, sex, comorbidities, and invasive devices, which may influence the risk of HAP and rehabilitation outcomes. However, we reduced this bias by weighing the multivariable analysis for the severity of the neurological damage. Finally, the retrospective nature of the study did not allow us to clarify the causal relationship between HAP and poor outcome after sABI. Further studies are needed to investigate the mechanisms underlying these associations.

In summary, our findings highlight HAP as a potential risk factor affecting recovery and mortality in patients with sABI. These results underline the importance of careful monitoring and preventive strategies to reduce the incidence of HAP in patients undergoing rehabilitation after sABI. Implementing a targeted rehabilitation strategy, in particular in patients with medical devices and dysphagia due to severe neurological status on admission, may not only improve functional outcomes but also reduce mortality.