Predictors of dysphagia screening and pneumonia among patients with intracerebral haemorrhage in China: a cross-sectional hospital-based retrospective study


Intracerebral haemorrhage (ICH) accounts for one-fifth of all stroke cases.1 Pneumonia is a complication presented by one-third of all patients with ICH2–4 and is associated with increased in-hospital mortality risk, prolonged hospital stay and higher healthcare costs.5–7 Prevention of pneumonia is fundamental to improve the prognosis of ICH patients.8 Identifying potentially modifiable factors associated with pneumonia could improve patient outcomes and reduce antibiotic use, which is a priority for the currently increasing antimicrobial resistance.9 The process of care, which should be in a concrete and organised manner and use multidisciplinary management, determines the frequency of several complications, including pneumonia in stroke patients.10 Therefore, intensification of preventive intervention in high-risk patients and improvement of the process of care can reduce pneumonia occurrence after ICH.11 Dysphagia is an important risk factor for pneumonia12 that affects 37%–78% of all patients with stroke.13 14 Dysphagia can cause significant complications, including dehydration, malnutrition and respiratory infections.15–18 Studies have shown that dysphagia assessment itself is an important rehabilitation measure.19 Dysphagia in patients with ICH remains a neglected research area, despite guidelines recommending dysphagia screening (DS) for all patients with acute stroke. Studies have shown that roughly one-quarter to one-third of all patients with ICH do not receive documented DS,20–22 with 43%–77% of patients with ICH failing the DS test.20 23 There is limited data regarding the number and type of patients with ICH who undergo DS, comparisons with DS among patients with ischaemic stroke and how failing a DS test affects the outcomes of patients with ICH.24–26

This study aimed to use data from the China Stroke Centre Alliance (CSCA) to examine factors influencing DS and pneumonia development, as well as the relationship between DS and pneumonia in patients with ICH.


Data sources and study sample

The CSCA is a national, hospital-based, multicentre, multifaceted intervention and continuous quality improvement initiative that is available to all Chinese secondary and tertiary grade hospitals.27 Details regarding the design and procedures of the CSCA programme have been previously described.27 28

Data were collected using a web-based patient data collection and management tool (Medicine Innovation Research Centre, Beijing, China), abstracted via chart review, coded, deidentified and transmitted securely to maintain patient confidentiality as per the national privacy standards. This patient data collection and management tool has two main functions. Its first function is to collect concurrent data. The data abstraction tool is characterised by predefined logic features, range checks and user alerts to identify a potentially invalid format or value entries and to optimise data quality at the time of entry. The second function is to analyse and provide data feedback. The China National Clinical Research Centre for Neurological Diseases (NCRCND) serves as the data analysis centre and has an agreement to analyse the aggregate deidentified data for care quality feedback and research purposes. All hospitals using the tool received an independent account and password to view the benchmark for adherence to evidence-based performance measures. Information on stroke care quality was also sent to hospital personnel via the WeChat App, a communication tool.27 Meanwhile, the CSCA primarily focuses on the construction of stroke centres at hospitals according to Chinese guidelines, organisational stakeholder and opinion leader meetings, collaborative workshops and webinars for hospital teams, hospital toolkits and hospital recognition. Data collection, decision support and hospital data feedback via multiple on-demand reports of performance on all key measures were conducted with the web-based patient data collection and management tool.27 We used the CSCA to obtain information regarding baseline patient characteristics, clinical variables, medical history, hospital-level characteristics and stroke severity. Stroke severity was assessed on admission using the Glasgow Coma Scale (GCS), which is a validated scale whose scores are positively correlated with stroke severity.

We included patients with ICH who were hospitalised between 1 August 2015 and 31 July 2019 based on the following criteria: (1) aged≥18 years, (2) a primary diagnosis of acute ICH confirmed by CT or MRI of the brain, (3) being within 7 days of symptom onset and (4) being admitted directly to the ward or through the emergency department. We excluded patients who died within 72 hours of admission. There were 1 006 798 patients with acute cerebrovascular events who were recruited into the registry; among them, 84 636 were diagnosed with ICH on admission and were eligible for this study. After excluding patients with missing DS data (n=170), without a history of pulmonary infection for 2 weeks prior to admission (n=450), without GCS scores (n=41 332) and with GCS scores≤8 (n=11 138), we included 31 546 participants (figure 1).

Figure 1
Figure 1

Flowchart of study participants. CSCA, China Stroke Centre Alliance; TIA, transient ischaemic attack; AIS,acute ischaemic stroke; SAH,subarachnoid hemorrhag.

Ascertainment of pneumonia and DS

According to the CSCA, pneumonia was defined as clinicians according to the Centres for Disease Control and Prevention (CDC) criteria, with support from clinical, biochemical, microbiological and radiological evidence, such as fever, cough, abnormal chest radiograph, gas exchange barrier, and isolation of a relevant pathogen, and identified on the clinical documentation as a discharge diagnosis.29 DS was defined as the assessment of a patient’s swallowing ability by a healthcare professional before oral intake of food, fluid or medications. The 30 mL water swallowing test was used to screen for dysphagia at all participating hospitals.30 Dysphagia was described as difficulty eating and drinking, and levels III–V on the 30 mL water swallowing test (when patients experienced choking or coughing or were unable to drink) indicated the presence of dysphagia.


Continuous variables are presented as means with SD or medians with IQRs, and categorical variables are displayed as counts with percentages. For medical history, patients with undocumented or unclear medical history were merged and classified together with patients without a medical history. Given the large sample size, some significant differences may not be clinically meaningful. Therefore, we employed absolute standardised differences (ASDs) to determine between-group differences in baseline characteristics. Unlike the t-test, χ2 test and other statistical tests, the ASD is independent of the sample size. Significant between-group differences were indicated by an ASD of >10%.31 To identify predictors for pneumonia, predictors that may be related to the outcome were selected through literature review and clinical experience; these predictors were then included in multifactor analysis. Multivariable logistic regression models were used to identify patient characteristics that were independently associated with DS and pneumonia. All tests were two tailed, and p<0.05 was considered statistically significant.


DS predictors

Table 1 presents the comparisons between patients with and without DS. There were 25 749 (81.6%) patients who had undergone DS. Patients who underwent DS had a higher rate of GCS scores≥14 compared with those who did not undergo DS (69.2% vs 58%).

Table 1

Baseline characteristics and patients with intracerebral haemorrhage who did or did not have documented dysphagia screening

Multivariable analysis revealed that a higher GCS Score (OR, 1.12 per point increase in GCS; 95% CI, 1.11 to 1.14); older age (OR, 1.03 per 10-year increase; 95% CI, 1.01 to 1.05), admission to a stroke unit (OR, 1.62; 95% CI, 1.49 to 1.76), hypertension (OR, 1.14; 95% CI, 1.07 to 1.22) and a history of coronary heart disease (CHD) or myocardial infarction (MI) (OR, 1.23; 95% CI, 1.07 to 1.42) were associated with undergoing DS. Moreover, history of stroke/transient ischaemic attack (OR, 0.87; 95% CI, 0.81 to 0.92), arrival by emergency medical services (EMS) (OR, 0.81; 95% CI, 0.76 to 0.86), admission to a tertiary-grade hospital (OR, 0.79; 95% CI, 0.74 to 0.83) and admission to a hospital in the Eastern region of China (OR, 0.67; 95% CI, 0.62 to 0.73) or Central region of China (OR, 0.59; 95% CI, 0.55 to 0.64) were associated with lower odds of having DS (table 2).

Table 2

Multivariable analysis of factors associated with documented dysphagia screening in patients with cerebral haemorrhage

Pneumonia predictors

Since August 2015, 31 546 patients with ICH were admitted to 1476 hospitals. The average age of patients eligible for DS with complete data regarding pneumonia status was 62.5±12.9 years; moreover, 63.6% and 67.2% of the patients were male and had a GCS Score≥14 points, respectively. A total of 7257 (23.0%) patients developed pneumonia. Table 3 presents the patient demographics, hospital characteristics and clinical outcomes by pneumonia status. Compared with patients without pneumonia, those with pneumonia were significantly older (mean±SD: 65.9±13.0 years vs 61.5±12.7 years), more likely to arrive at the hospital by EMS (47.7% vs 32.2%), had a higher dysphagia incidence (32.8% vs 8.0%) and less likely to have a GCS Score≥14 (47.3% vs 73.1%).

Table 3

Univariate analysis of patient characteristics associated with pneumonia

Multivariable analyses revealed older age (OR, 1.30 per 10-year increase; 95% CI, 1.27 to 1.33); being male (OR, 1.35; 95% CI, 1.26 to 1.45); arrival in the hospital by EMS (OR, 1.56; 95% CI, 1.47 to 1.66); having dysphagia (OR, 4.34; 95% CI, 4.02 to 4.68); having a history of CHD/previous MI (OR, 1.18; 95% CI, 1.04 to 1.33), heart failure (OR, 1.85; 95% CI, 1.24 to 2.77), smoking (OR, 1.12; 95% CI, 1.05 to 1.20) were associated with an increased risk of pneumonia (table 4). Further, there was an independent association of documented DS (OR, 0.65; 95% CI, 0.44 to 0.95), GCS Score (OR, 0.83 per point increase in GCS; 95% CI, 0.81 to 0.84); admission to a stroke unit (OR, 0.87; 95% CI, 0.81 to 0.94), and being admitted to a hospital in the Central or Eastern regions of China (compared with the Western region) with a lower risk of pneumonia (table 4).

Table 4

Multivariable analysis of patient characteristics associated with hospital-acquired pneumonia among patients eligible for dysphagia screening


Although several national and international guidelines recommend early swallowing function screening in patients with stroke before orally taking food or tablets,32–34 we found that one-fifth of eligible patients with ICH in the CSCA did not undergo DS.

The low DS rate may be related to the allocation of medical resources, including medical and health service personnel, and inadequate training of those performing DS. Consistent with our findings, Joundi et al assessed 1091 patients with ICH in the Ontario Stroke Registry and reported a DS rate of 67.6% (vs 81.6% in our study).22 The slightly lower DS rate reported by Joundi et al could be attributed to the previous study limiting DS to within 72 hours of admission; contrastingly, we did not impose such a limit. Given that swallowing function requires dynamic management and patients may recover over the hospitalisation course, more patients may receive DS after the 72-hour period. Further, compared with patients with ischaemic stroke, patients with ICH were less likely to undergo DS.35 Compared with patients with ischaemic stroke, patients with ICH experience more serious and rapid changes in their condition, including the level of consciousness and swallowing function. Therefore, in patients with ICH, clinical medical staff are likely to overlook swallowing function in favour of stabilising the level of consciousness and recovering language function. We also found in patients who arrived by EMS had lower DS rates, possibly indicating that they had more severe strokes, so dysphagia was not a priority in their care. Contrastingly, patients with higher GCS scores had higher DS rates, since they had lower stroke severity, and the patients were conscious and could better cooperate with swallowing screening; thus, DS was given more treatment priority on hospital arrival. However, there were lower DS rates in hospitals in the Central and Eastern regions of China, as well as tertiary-grade hospitals, where medical resources are relatively abundant and medical service personnel were more likely to be specialised, which may be related to severe nerve injury, low level of consciousness, inability to evaluate swallowing and direct nasal feeding after admission. Further research is required to further elucidate the reasons underlying these findings.

Patients with ICH have a high dysphagia incidence (40%–70%).36–38 Martino et al reported that patients with swallowing or inhaling difficulties were more likely to develop pneumonia.16 Delays in screening and assessment were associated with pneumonia; further, pre–post comparison revealed that screening for swallowing disorders by nurses reduced the pneumonia rate and the length of hospital stay.35 39 Therefore, early screening and rehabilitation of dysphagia can improve cough reflex, as well as reduce the aspiration and pneumonia incidence.8 40 41 Our findings further confirm that dysphagia is an independent risk factor for pneumonia; further, the dysphagia rate in patients with pneumonia was about four times that of patients with normal swallowing function. Screening for swallowing function may reduce pneumonia incidence. This demonstrates the importance of DS in patients with ICH and supports the inclusion of DS as a performance indicator for improving stroke quality.11 34 Among them, DS and management is part of the process of quality of care. Nursing processes ensure the collection of complete and accurate information during evaluation, facilitate the development of personalised treatment plans and contribute to the continuous and coordinated long-term management, allowing for timely adjustments to meet the changing needs of patients. Therefore, screening all patients with stroke for swallowing disorders is critical for identifying patients at risk, mitigation strategies and means of active prevention of pneumonia and other complications. As demonstrated in our study, patients having varying results of swallow function give a more objective perspective of the relationship between pneumonia and dysphagia.

Our results are consistent with previous findings regarding risk factors for pneumonia (older age, dysphagia, male, smoking, history of CHD/MI and heart failure).21 42 Admission to a stroke unit reduced the odds of pneumonia, which could be attributed to the rapid development of stroke unit organisation and management, including diversified treatment with drug therapy, physical rehabilitation, language training, psychological rehabilitation and health education.43 44 Similarly, we observed that hospitals in the Central and Eastern regions of China have lower incidence rates of pneumonia compared with those in the Western region of China. This may be attributed to factors such as greater economic development, fewer restrictions on medical resources, more advanced diagnostic technologies, higher-level medical institutions, professional healthcare teams, strengthened disease monitoring and management, provision of professional nursing and the implementation of more comprehensive health management plans and prevention strategies.45 46

This study had several limitations. First, we collected data from many different hospitals in China, and there were individual differences in the results of swallowing evaluation by different doctors. Second, we lacked information regarding ICH bleeding volume, bleeding site and aetiology, which are factors that affect the DS rate and pneumonia. There is a need for further studies with detailed information regarding the ICH volume location. Although we defined pneumonia as the outcome variable, the specific time of pneumonia occurrence was not recorded. However, this limitation was mitigated by excluding patients with pneumonia at the time of admission. Additionally, there were no data on patients with intubation/mechanical ventilation and diet during hospitalisation, and these factors affect DS, although we excluded patients with a GCS≤8 on admission to reduce some confounders. Nonetheless, this is the study with the largest cohort of patients with ICH to assess factors related to DS, as well as the association between DS and pneumonia. Therefore, it provides important insight into the clinical characteristics and care requirements for patients with ICH.

In summary, we observed frequent pneumonia occurrence in patients hospitalised with ICH. The DS rate remained low; moreover, dysphagia is one of the strongest predictors of pneumonia. Preventing pneumonia development in patients with dysphagia during hospitalisation requires specialised and prompt monitoring, nursing and management.

This post was originally published on