Healthcare professionals’ knowledge, attitudes, and practices towards predictive diagnosis of early neurological deterioration

Abstract

This study aimed to investigate healthcare professionals’ knowledge, attitudes, and practices (KAP) towards predictive diagnosis of early neurological deterioration. We conducted a cross-sectional study at hospitals in Jiangsu region from December 1, 2023, to April 15, 2024. Demographic information and KAP scores were evaluated using a self-designed KAP questionnaire. A total of 299 valid responses were included in the final analysis. The median scores for knowledge, attitude, and practice were 44.00 [37.00, 47.00], 18.00 [17.00, 19.00], and 35.00 [31.00, 37.00], respectively. A total of 40.2% rarely used clinical scales, whereas 28.5% rarely relied on imaging for assessing early neurological deterioration risk. Multivariate logistic regression analysis revealed independent associations between poor knowledge and lack of professional title, other positions, and working in the ICU. Additionally, holding a college or Bachelor’s degree was independently linked to a negative attitude. Higher knowledge scores and holding doctor positions were independently associated with practice. Healthcare professionals generally possess adequate knowledge and demonstrate positive attitudes and proactive practices towards the predictive diagnosis of early neurological deterioration. Efforts to improve stroke predictive diagnosis should focus on targeted educational programs that address specific gaps in knowledge and attitudes, especially for professionals in critical care settings and those without advanced titles.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-09505-x.

Introduction

As the population ages, the incidence of stroke in China is on the rise, increasingly becoming a leading cause of disability and death. The China Stroke Report of 2020 indicated a stroke prevalence of 1114.8 per 100,000, with an annual incidence rate of 246.8 per 100,000 and a mortality rate of 149.49 per 100,000¹. This is coupled with an annual increase in stroke prevalence of 8.7%². Consequently, China now holds the global distinction of having the highest lifetime risk of stroke and associated disease burden³. Early neurological deterioration (END), characterized by an increasing severity or spread of neurological deficits after hospital admission, underscores the urgency for timely and accurate diagnosis to mitigate treatment delays and improve outcomes^4,5. END occurs in approximately 20–30% of ischemic stroke patients and is associated with significantly poorer outcomes, including higher mortality rates and more severe long-term disability^6,7. A multicenter study found that patients with END had longer hospital stays (median 12 vs. 8 days), higher in-hospital mortality (19.6% vs. 7.9%), and poorer functional outcomes at discharge compared to those with non-END⁸. Furthermore, the risk of progression is highest within the first 72 h after stroke onset, highlighting the critical time window for intervention and emphasizing the need for accurate predictive tools in the acute phase⁹.

Ongoing research into stroke prediction models has been notably enhanced by the integration of artificial intelligence (AI), which leverages machine learning (ML) techniques to refine and improve the precision of prognostic tools in this critical area of healthcare. These ML methodologies are employed for diagnosing stroke¹⁰, predicting the onset of stroke symptoms^11,12, assessing stroke severity^12,13, and forecasting outcomes^14,15. The integration of voluminous clinical and radiological data has necessitated that clinicians consider this information comprehensively in patient management. In the time-sensitive context of acute stroke, artificial intelligence (AI) provides clinicians with tools for rapid assessment and synthesis of available information, facilitating precise predictions from complex and noisy data sets¹⁶. As a result, clinicians are compelled to continually adapt, acquiring new competencies such as clinical assessments or image interpretation, staying abreast of the literature, and integrating these advances into daily practice¹⁷.

The Knowledge-Attitude-Practice (KAP) model is foundational in understanding and influencing health behaviors¹⁸. It utilizes the KAP questionnaire to thoroughly evaluate the knowledge, attitudes, and practices within the healthcare sector, facilitating the assessment of content demand and acceptance levels¹⁹. Rooted in the premise that enhanced knowledge positively influences attitudes, which in turn drive individual practices, this model is crucial for advancing health literacy²⁰. Moreover, it plays a significant role in modifying the practice patterns of physicians²¹. This study targets healthcare professionals in critical departments such as neurology, rehabilitation, emergency, neurosurgery, and ICU, owing to their direct roles in the diagnosis and management of END. The importance of understanding their knowledge, attitudes, and practices towards the predictive diagnosis of END cannot be overstated, as it is vital for identifying existing gaps and augmenting early intervention strategies.

Despite its critical importance, there is a notable lack of KAP studies focused on this area. Consequently, this study aims to investigate the KAP of healthcare professionals regarding the predictive diagnosis of END, addressing an essential gap in current research.

Methods

Study design and participants

This cross-sectional study was conducted at secondary and above hospitals in Jiangsu region from December 1, 2023, to April 15, 2024. Participants included healthcare professionals from the neurology, rehabilitation, emergency neurosurgery, and intensive care departments. Ethical approval for the study was granted by the Ethics Committee of RUGAO BOAI Hospital (2024-K007-01), and informed consent was obtained from all participants.

Sampling and recruitment

Participants included doctors, nurses, and rehabilitation specialists from the hospital’s neurology, rehabilitation, emergency neurosurgery, and intensive care units; interns and those about to leave their positions were excluded. Some participants with resident trainee without professional title were standardized training residents who hold physician licenses and actively participate in clinical care under supervision. Although they do not yet have formal job titles, they are considered part of the clinical workforce and were included in this study accordingly. A non-probabilistic convenience sampling method was employed, primarily through an online questionnaire promoted via WeChat using the snowball method, with supplementary direct recruitment within departments.

Sample size calculation

A frequently recommended guideline suggests a minimum ratio of 10 participants per observed variable (item) to ensure stable parameter estimates and adequate statistical power^22,23. As the questionnaire comprises 30 items (17 Knowledge, 5 Attitude, and 8 Practice items), this suggests a minimum sample size of approximately 30 × 10 = 300. Although the final number of valid responses was 299, which is one participant short of the target, this minor shortfall is unlikely to meaningfully affect statistical power or the generalizability of the findings.

Data collection

An online questionnaire was created using a WeChat mini-program, which also generated a QR code for access. Participants accessed the survey by scanning this QR code. To maintain the quality and completeness of the data, submission was restricted to one per IP address, and all questionnaire items were required to be completed before submission. The research team conducted thorough checks on the completeness, internal consistency, and reasonableness of all completed questionnaires, ensuring that all responses met the mandatory requirements for each question.

Questionnaire introduction

After creating the initial draft, the questionnaire was developed based on input from experts in the field of neurology. In designing items related to imaging predictors—particularly the use of diffusion-weighted imaging (DWI) in identifying END—we referred to published literature on the risk factors and neuroimaging features of END^8,24,25. And the instrument was subjected to two small-scale pilot tests with 30 participants each to refine its content based on the feedback received. The feedback from the second pilot test resulted in a Cronbach’s alpha of 0.8064, indicating good internal consistency based on the pilot sample.

The finalized questionnaire comprised four sections (Appendix): demographic information (including variables such as education level, gender, job type, institutional nature, professional title, and department), and dimensions assessing knowledge, attitude, and practice. The knowledge dimension included 17 questions addressing fundamental concepts of END and the predictive indicators of the condition. Questions K1-K4 were scored as follows: 2 points for correct answers, and 1 point for incorrect or unclear responses. Questions K5-K17 utilized a scoring system where responses of “very familiar” earned 3 points, “heard of” 2 points, and “never heard of” 1 point, with total scores ranging from 17 to 47 points. The attitude dimension consisted of 5 questions focused on healthcare professionals’ perceptions toward the current status of predictive diagnosis of END, future development directions, the clinical value of early prediction, and challenges in patient management. These questions used a five-point Likert scale from “very positive” (5 points) to “very negative” (1 point), allowing for a total score between 5 and 25 points. The practice dimension contained 8 questions that examined how healthcare professionals actually monitored for END, conducted risk assessments (using clinical, imaging, and biochemical data), communicated with patients, and participated in relevant research in their daily work. These questions employed a five-point Likert scale from “always” (5 points) to “never” (1 point), resulting in a score range from 8 to 40 points. Scores exceeding 70% of the maximum possible in each section were considered indicative of adequate knowledge, a positive attitude, and proactive practice^26,27.

Statistical analysis

The Kaiser-Meyer-Olkin (KMO) measure was calculated to evaluate sampling adequacy and the structural validity of the questionnaire, ensuring that the dataset was suitable for factor analysis. Data analysis was performed using R 4.3.2. Continuous data were assessed for normality and presented as means and standard deviations (SD) for normally distributed data, while non-normally distributed data were represented using the median, 25th percentile, and 75th percentile. Categorical data were expressed as n (%). For two-group comparisons of continuous variables, the t-test was utilized for normally distributed data and the Wilcoxon Mann-Whitney test for non-normally distributed data. In cases involving three or more groups with normally distributed continuous variables and homogeneous variances, ANOVA was employed, whereas the Kruskal-Wallis test was applied for non-normally distributed data. Univariate and multivariate regression analyses were conducted to examine the relationships between the scores from each dimension (knowledge, attitude, practice) and demographic characteristics. Variables that reached a significance level of P < 0.1 in the univariate analysis were included in the multivariate regression. Results were reported with P-values to three decimal places, considering P < 0.05 as statistically significant.

Results

Basic characteristics of the study participants

A total of 301 questionnaires were collected, from which two were excluded due to duplicate IP addresses, leaving 299 valid responses for analysis. The overall internal consistency of the scale and its subscales was high based on the final analysis sample, with Cronbach’s alpha coefficients of 0.9681 for the total scale, 0.9541 for the knowledge subscale, 0.8344 for the attitude subscale, and 0.8944 for the practice subscale. The Kaiser-Meyer-Olkin (KMO) measure for the total scale was 0.9439. Among the participants, 191 (63.9%) were doctors, 88 (29.4%) were aged between 28 and 34 years, 172 (57.5%) were male, 161 (53.8%) held a college or bachelor’s degree, and 122 (40.8%) did not have a professional title. Additionally, 132 (44.1%) reported daily encounters with stroke patients in their departments. Median [25th percentile, 75th percentile] scores for knowledge, attitude, and practice were 44.00 [37.00, 47.00], 18.00 [17.00, 19.00], and 35.00 [31.00, 37.00], respectively. By analyzing demographic characteristics, participants’ knowledge, attitude, and practice scores varied across age (P < 0.001, P < 0.001, P < 0.001), education (P < 0.001, P = 0.047, P < 0.001), professional title (P < 0.001, P < 0.001, P < 0.001), years of work experience (P < 0.001, P < 0.001, P < 0.001), and probability of stroke patients treated in their department developing into END (P < 0.001, P = 0.005, P < 0.001). Meanwhile, their knowledge and practice scores varied across nature of hospital (P < 0.001 and P < 0.001), position (P < 0.001 and P < 0.001), department (P < 0.001 and P < 0.001), frequency of receiving stroke patients (P < 0.001 and P < 0.001), and probability of stroke patients they treated developing into END (P < 0.001 and P < 0.001), and participation in academic conferences or training (P < 0.001 and P < 0.001). Moreover, participants with different marital status were more likely to have different knowledge and attitude scores (P < 0.001 and P < 0.001) (Table 1).

Table 1.

Basic information of participants and KAP score.

N = 299	N (%)	Knowledge	P	Attitude	P	Practice	P
		Median [25%,75%] or mean (± SD)		Median [25%,75%] or mean (± SD)		Median [25%,75%] or mean (± SD)
Total Score	299 (100.0)	44.00 [37.00, 47.00]		18.00 [17.00, 19.00]		35.00 [31.00, 37.00]
1. Your age:			< 0.001		< 0.001		< 0.001
45 and above	73 (24.4)	47.00 [47.00, 47.00]		17.00 [16.00, 17.00]		35.00 [34.00, 37.00]
Less than 28	67 (22.4)	34.00 [27.00, 38.00]		18.00 [17.00, 19.00]		30.00 [27.00, 32.00]
28–34	88 (29.4)	42.50 [37.00, 45.00]		18.00 [17.00, 19.00]		36.00 [30.00, 38.00]
35–44	71 (23.7)	46.00 [44.00, 47.00]		18.00 [17.00, 19.00]		37.00 [35.00, 38.00]
2. Your gender:			0.098		0.097		0.134
Male	172 (57.5)	44.00 [36.00, 47.00]		18.00 [17.00, 19.00]		35.00 [30.00, 37.00]
Female	127 (42.5)	45.00 [39.00, 47.00]		17.00 [17.00, 18.50]		35.00 [32.00, 37.50]
3. Your marital status:			< 0.001		< 0.001		0.583
Unmarried	80 (26.8)	40.50 [34.00, 44.00]		18.00 [18.00, 19.00]		35.00 [30.00, 38.00]
Married	187 (62.5)	45.00 [38.00, 47.00]		18.00 [17.00, 18.00]		35.00 [31.00, 37.00]
Divorced/Widowed	32 (10.7)	46.50 [42.50, 47.00]		17.00 [16.00, 18.00]		35.50 [32.00, 38.00]
4. Your education level:			< 0.001		0.047		< 0.001
Master’s and above	138 (46.2)	46.00 [43.00, 47.00]		18.00 [17.00, 19.00]		37.00 [35.00, 38.00]
College/Undergraduate	161 (53.8)	39.00 [33.00, 46.00]		18.00 [17.00, 18.00]		32.00 [29.00, 36.00]
5. Your professional title:			< 0.001		< 0.001		< 0.001
Intermediate and above	83 (27.8)	47.00 [47.00, 47.00]		17.00 [16.00, 17.00]		35.00 [34.00, 37.00]
Resident Trainee without professional title	122 (40.8)	36.00 [28.00, 42.75]		18.00 [17.00, 19.00]		31.00 [28.00, 36.75]
Junior	94 (31.4)	45.00 [43.00, 46.00]		18.00 [17.00, 19.00]		37.00 [35.00, 38.00]
6. The nature of your hospital:			< 0.001		0.897		< 0.001
Public tertiary hospital	154 (51.5)	45.00 [40.00, 47.00]		18.00 [17.00, 19.00]		36.00 [33.00, 37.75]
Public primary/secondary hospital	115 (38.5)	44.00 [37.00, 47.00]		18.00 [17.00, 18.00]		35.00 [30.00, 37.00]
Private hospital	30 (10.0)	36.00 [27.50, 44.00]		18.00 [17.00, 19.00]		30.00 [28.00, 36.00]
7. Your years of work experience:			< 0.001		< 0.001		< 0.001
More than 20 years	67 (22.4)	47.00 [47.00, 47.00]		17.00 [16.00, 17.00]		35.00 [34.00, 37.00]
5 years or less	113 (37.8)	37.00 [29.00, 43.00]		18.00 [17.00, 19.00]		31.00 [28.00, 37.00]
5–10 years	48 (16.1)	43.00 [37.00, 45.00]		18.00 [17.00, 19.00]		35.50 [30.75, 37.25]
10–20 years	71 (23.7)	46.00 [45.00, 47.00]		18.00 [17.00, 18.00]		37.00 [35.00, 38.00]
8. Your position:			< 0.001		0.397		< 0.001
Doctor	191 (63.9)	46.00 [44.50, 47.00]		18.00 [17.00, 19.00]		37.00 [35.00, 38.00]
Nurse	53 (17.7)	37.00 [34.00, 40.00]		18.00 [17.00, 18.00]		30.00 [29.00, 32.00]
Rehabilitation therapist	55 (18.4)	27.00 [25.00, 36.00]		18.00 [17.00, 19.00]		28.00 [26.50, 30.00]
9. Your department:			< 0.001		0.944		< 0.001
Emergency department neurosurgery	130 (43.5)	46.00 [43.25, 47.00]		18.00 [17.00, 19.00]		36.00 [35.00, 38.00]
Neurology department	64 (21.4)	45.00 [39.75, 47.00]		18.00 [17.00, 18.25]		36.00 [31.00, 37.00]
Rehabilitation department	64 (21.4)	29.50 [25.00, 37.00]		18.00 [17.00, 19.00]		29.00 [27.00, 31.00]
ICU	41 (13.7)	46.00 [41.00, 47.00]		18.00 [17.00, 19.00]		36.00 [33.00, 37.00]
10. Frequency of receiving stroke patients in your department:			< 0.001		0.753		< 0.001
Daily	132 (44.1)	46.00 [42.75, 47.00]		18.00 [17.00, 19.00]		36.00 [34.00, 38.00]
Quarterly	30 (10.0)	40.50 [31.75, 47.00]		18.00 [17.00, 19.00]		33.00 [29.00, 37.00]
Monthly	63 (21.1)	36.00 [27.00, 42.00]		18.00 [17.00, 18.50]		30.00 [28.00, 35.00]
Weekly	74 (24.7)	45.00 [41.25, 47.00]		18.00 [17.00, 19.00]		36.00 [33.00, 37.75]
11. Probability of stroke patients treated in your department developing into progressive stroke patients:			< 0.001		0.005		< 0.001
Greater than 40%	60 (20.1)	45.50 [41.75, 47.00]		18.00 [17.00, 19.00]		36.00 [34.00, 37.25]
20-40%	125 (41.8)	45.00 [42.00, 47.00]		17.00 [17.00, 18.00]		36.00 [34.00, 38.00]
0–20%	114 (38.1)	38.00 [29.00, 45.00]		18.00 [17.00, 19.00]		31.00 [28.00, 36.00]
12. Probability of the stroke patients you treat developing into progressive stroke patients:			< 0.001		0.841		< 0.001
Greater than 40%	73 (24.4)	46.00 [43.00, 47.00]		18.00 [17.00, 19.00]		36.00 [34.00, 38.00]
20-40%	112 (37.5)	45.00 [42.75, 47.00]		18.00 [17.00, 19.00]		36.00 [34.00, 38.00]
0–20%	114 (38.1)	37.00 [29.00, 45.00]		18.00 [17.00, 18.00]		31.00 [28.00, 36.00]
13. Have you participated in academic conferences or training related to progressive stroke in the past six months?			< 0.001		0.848		< 0.001
Yes	278 (93.0)	45.00 [38.00, 47.00]		18.00 [17.00, 19.00]		35.00 [32.00, 37.00]
No	21 (7.0)	28.00 [25.00, 37.00]		17.00 [17.00, 19.00]		28.00 [26.00, 30.00]

Knowledge, attitude, and practice

The distribution of knowledge dimensions shown that the three questions with the highest number of participants choosing the “Unclear”, “Uncertain”, or “Never heard of it” option were “Moderate to severe strokes, particularly in patients with an NIHSS score > 14.” (K6) with 31.8%, “DWI showing an infarct area involving more than three layers.” (K17) with 31.1%, and “Worsening consciousness (indicated by a lower GCS) is an independent predictor of neurological deterioration.” (K5) with 29.1% (Table 2).

Table 2.

Knowledge dimension of the participants.

Knowledge	Yes (correct option)	Wrong	Unclear
1. Stroke encompasses both ischaemic stroke and haemorrhagic stroke, with ischaemic stroke being the more prevalent type. Early Neurological Deterioration (END) is a specific subtype of acute ischaemic stroke.	283(94.6%)	16(5.4%)	0 (0%)
2. Early Neurological Deterioration typically occurs within a short duration following an acute ischaemic stroke, leading to the worsening of neurological function.	Within hours or days	Within days or weeks	Uncertain
	266(89%)	25(8.4%)	8(2.7%)
3. Early Neurological Deterioration is associated with higher rates of disability and mortality compared to other types of stroke.	275(92%)	24(8%)	0 (0%)
4. Accurate prediction of Early Neurological Deterioration can help delay disease progression and improve the prognosis of this condition.	273(91.3%)	23(7.7%)	3(1%)
Please assess your understanding of the predictive indicators for Early Neurological Deterioration:	Very familiar	Heard of it	Never heard of it
5. Worsening consciousness (indicated by a lower GCS) is an independent predictor of neurological deterioration.	170(56.9%)	42(14%)	87(29.1%)
6. Moderate to severe strokes, particularly in patients with an NIHSS score > 14.	170(56.9%)	34(11.4%)	95(31.8%)
7. Diabetes, pneumonia, increased body temperature within the first 24 h, absence of antithrombotic medication before the stroke, and a history of transient ischaemic attacks (TIA).	255(85.3%)	6(2%)	38(12.7%)
8. A blood BUN/Cr ratio > 15 is an independent predictor of END.	188(62.9%)	29(9.7%)	82(27.4%)
9. Elevated inflammatory markers are associated with END, including hsCRP, NLP, IL-6, and TNF.	206(68.9%)	30(10%)	63(21.1%)
10. CT scans indicating haemorrhagic transformation or cerebral oedema.	214(71.6%)	16(5.4%)	69(23.1%)
11. Vascular imaging showing proximal large artery occlusion.	210(70.2%)	10(3.3%)	79(26.4%)
12. Imaging findings supporting ischaemic stroke related to perforating artery disease: mainly infarct locations in the basal ganglia region (striatum, internal capsule, posterior limb of the internal capsule, corona radiata), and the mid-to-lower segments of the pons.	215(71.9%)	15(5%)	69(23.1%)
13. The responsible vessels include the lenticulostriate arteries or the paramedian pontine arteries.	176(58.9%)	39(13%)	84(28.1%)
14. Imaging findings of watershed infarcts may suggest END (severe stenosis/occlusion of the internal carotid artery and middle cerebral artery).	210(70.2%)	22(7.4%)	67(22.4%)
15. Imaging indicating thrombus progression (e.g., susceptibility vessel sign on MRI T2* images).	217(72.6%)	11(3.7%)	71(23.7%)
16. An infarct diameter > 15 mm in the striatum and internal capsule is an independent predictor of END.	166(55.5%)	52(17.4%)	81(27.1%)
17. DWI showing an infarct area involving more than three layers.	167(55.9%)	39(13%)	93(31.1%)

Responses to the attitudinal dimension showed that 32.8% strongly agreed and 41.8% agreed that there is a lack of consensus guidelines regarding END, and the clinical definition, prediction, and diagnostic criteria for END are not well established (A1). Regarding managing the frustration and helplessness of patients with END to be a challenging task (A5), 30.8% were neutral (Table 3).

Table 3.

Attitude dimension of the participants.

Attitude	Strongly Agree	Agree	Neutral	Disagree
1. Currently, there is a lack of consensus guidelines regarding progressive stroke, and the clinical definition, prediction, and diagnostic criteria for progressive stroke are not well established.	98(32.8%)	125(41.8%)	76(25.4%)	0 (0%)
2. Translating indicators of progressive stroke prediction into clinical practice is of great significance for the early identification of such patients.	164(54.8%)	128(42.8%)	7(2.3%)	0 (0%)
3. I am willing to participate in early prediction studies on progressive stroke to discover suitable biomarkers for clinical application.	177(59.2%)	118(39.5%)	4(1.3%)	0 (0%)
4. Early prediction and diagnosis of progressive stroke can enable doctors to provide more support and education to patients and their families, increase patient and family cooperation, and ease doctor-patient relationships.	156(52.2%)	127(42.5%)	16(5.4%)	0 (0%)
5. I consider managing the frustration and helplessness of patients with progressive stroke to be a challenging task.	60(20.1%)	130(43.5%)	92(30.8%)	17(5.7%)

When it comes to related practices, 21.1% just sometimes and 19.1% occasionally used at least 1 scale to determine the presence of END in newly admitted stroke patients (P3). On the other hand, 17.1% just sometimes and 11.4% occasionally used imaging results to assess a patient’s risk of developing a END (P4) (Table 4).

Table 4.

Practice dimension of the participants.

Practice	Always	Often	Sometimes	Occasionally	Never
1. For newly admitted stroke patients, especially within 72 h, I would pay close attention and promptly observe their condition.	232(77.6%)	66(22.1%)	1(0.3%)	0 (0%)	0 (0%)
2. I would assess the risk of progressive stroke in patients based on their medical history, clinical results, and the latest research findings.	165(55.2%)	114(38.1%)	19(6.4%)	1(0.3%)	0 (0%)
3. For newly admitted stroke patients, I would use at least one scale, such as the NIHSS score, to determine if the patient has progressive stroke.	46(15.4%)	130(43.5%)	63(21.1%)	57(19.1%)	3(1%)
4. I would assess the risk of progressive stroke in patients based on imaging results, such as determining the location and size of the infarction.	90(30.1%)	118(39.5%)	51(17.1%)	34(11.4%)	6(2%)
5. I would assess the risk of progressive stroke in patients based on biochemical indicators such as blood sugar, blood pressure, and blood lipids.	121(40.5%)	143(47.8%)	26(8.7%)	8(2.7%)	1(0.3%)
6. I would refer to the latest research findings to comprehensively assess the risk of progressive stroke in newly admitted stroke patients.	183(61.2%)	61(20.4%)	55(18.4%)	0 (0%)	0 (0%)
7. I would patiently share information about the risk and adverse outcomes of progressive stroke with newly admitted stroke patients and their families to ensure that patients maintain reasonable expectations about the disease.	103(34.4%)	146(48.8%)	50(16.7%)	0 (0%)	0 (0%)
8. I actively participate in clinical research or trials to explore suitable biomarkers and diagnostic methods to assist in the early prediction and identification of progressive stroke patients.	200(66.9%)	71(23.7%)	27(9%)	1(0.3%)	0 (0%)

Univariate and multivariate analysis of knowledge, attitude, and practice

The median scores for the knowledge, attitude, and practice dimensions were used as the cut-off value for each dimension to divided the groups in order to ensure relatively balanced sample sizes for logistic regression modeling, and the number of participants above the cut-off value were 165 (55.18%), 171 (57.19%), and 169 (56.52%), respectively. Multivariate logistic regression showed that resident trainee without professional title (OR = 0.027, 95% CI: [0.002, 0.228], P < 0.001), other position (OR = 0.010, 95% CI: [0.002,0.041], P < 0.001), and worked in ICU (OR = 0.035, 95% CI: [0.001,0.947], P = 0.046) were independently associated with poor knowledge. Concurrently, with college/Bachelor’s degree (OR = 0.401, 95% CI: [0.196,0.795], P = 0.009) was independently associated with negative attitude. Furthermore, knowledge score (OR = 1.480, 95% CI: [1.123,2.045], P = 0.004) and other position (OR = 0.089, 95% CI: [0.011,0.535], P = 0.007) were independently associated with practice (Table S1).

Discussion

Healthcare professionals possess adequate knowledge, maintain positive attitudes, and engage in proactive practices regarding the predictive diagnosis of END. It is crucial to enhance targeted educational and training programs, especially for professionals without titles and those in specific roles such as working in the ICU, to uniformly elevate knowledge and optimize practices across all healthcare staff involved in stroke care.

In this study, significant differences in KAP toward predictive diagnosis of END were observed across various demographics and professional metrics, with multivariate logistic regression supporting these findings. Older healthcare professionals (45 and above) demonstrated higher knowledge levels, suggesting that experience might enhance understanding in this domain. This trend was consistent with the findings from logistic regression, which indicated that professionals without a title tended to have poorer knowledge, possibly due to less exposure or specialized training in END management^28,29.

In addressing the observed lower knowledge levels among healthcare professionals working in ICUs, it is plausible to consider the demanding nature of ICU environments. These professionals are often tasked with acute care priorities, which may limit their opportunities for ongoing education and focused training in more specialized areas such as END management. The hectic and unpredictable nature of ICU workloads can impede the ability to stay updated with specific advancements outside immediate critical care protocols, thus explaining the lower knowledge scores in this subgroup^28,30. This suggests that tailored educational interventions that accommodate the unique demands of ICU settings might be necessary to enhance knowledge in specialized areas like END diagnosis.

Interestingly, despite the variations in knowledge and practice, attitudes across different age groups remained consistently positive, indicating a universal recognition of the importance of stroke management. This suggests that while knowledge and practice can vary with professional development and environmental factors, attitudes towards stroke care may be uniformly positive across different age sectors of healthcare professionals. The regression analysis further highlighted that having a college degree was associated with a slightly more negative attitude, which could be attributed to a possible disconnect between theoretical education and practical expectations^31,32. Conversely, higher knowledge scores and diverse professional roles correlated with better practice outcomes, emphasizing the importance of comprehensive education and varied professional exposure in improving patient care practices.

These findings align with existing literature that underscores the importance of continuous professional development and targeted training to address gaps in healthcare provision^33,34. Given the lack of differences in attitude despite varying knowledge and practice levels, future interventions should focus more on enhancing knowledge and practical skills rather than altering attitudes, which are already positive.

In the knowledge dimension, healthcare professionals generally demonstrated a high level of understanding regarding the characteristics and implications of END. However, some critical gaps were noted, particularly in less common or complex predictors of END, such as the specific impact of biomarkers like hsCRP, NLP, IL-6, and TNF, and the interpretation of imaging findings related to specific artery diseases. These areas had a notably lower percentage of correct responses. To improve knowledge in these less understood areas, targeted educational programs should be developed. These programs could include detailed workshops focusing on the interpretation of advanced diagnostic imaging and the role of inflammation in stroke progression, potentially integrated into regular continuing medical education (CME) sessions^35,36.

The attitude dimension revealed a strong overall positive response towards the significance of early prediction in stroke management. However, there was a notable portion of respondents who felt neutral about the challenges associated with managing the frustration and helplessness of patients with END. This could be indicative of a lack of confidence or a perceived lack of resources to effectively manage patient emotions, which are critical to patient care. Literature supports that enhanced training in patient communication and psychological support can improve healthcare outcomes by fostering better patient-provider relationships^37,38. Practical steps could include the implementation of specific training modules focused on psychological aspects of patient care in stroke units and the incorporation of multidisciplinary teams including psychologists or counselors who specialize in chronic illness management³⁹.

The practice dimension indicated diligent observation and assessment practices among professionals, particularly within the initial 72 h of stroke admission. However, the use of scales such as the NIHSS to determine END risk was not consistently applied, with a significant number of respondents only sometimes or occasionally using these scales. Research highlights the importance of standardized assessment tools in the early detection and management of stroke^40,41. To enhance the use of these tools, hospitals could develop protocols that mandate the use of NIHSS scores for all stroke admissions and provide training sessions to ensure all healthcare providers are proficient in these assessments. Additionally, integrating these protocols into electronic health records with automated reminders could ensure compliance and improve the accuracy of early stroke assessment.

This study has several limitations that warrant consideration. First, the use of a self-designed questionnaire may introduce bias, as it has not been validated externally, which could affect the reliability and generalizability of the findings. Second, the study’s setting at a Single area limits its generalizability to other healthcare settings or regions, which may have different levels of resources and training. Lastly, the cross-sectional design of the study restricts the ability to infer causality between healthcare professionals’ characteristics and their KAP scores, making it difficult to determine the direction of the relationships observed. Additionally, although the study followed a recommended minimum sample-to-variable ratio for logistic regression, the wide confidence intervals observed in the multivariate analysis suggest possible limitations in statistical power. These results should therefore be interpreted with caution. Furthermore, while variables such as marital status were included based on univariate significance, they may not be strongly related to clinical decision-making ability and should be interpreted accordingly. Moreover, the study relied on self-reported data, which may introduce recall bias or social desirability bias, particularly in practice-related items such as self-assessed use of diagnostic tools or perceived proportions of patients developing END. This may affect the accuracy and validity of certain measures.

In conclusion, healthcare professionals generally demonstrated adequate knowledge, positive attitudes, and proactive practices towards the predictive diagnosis of END, although some groups, particularly those resident trainee without professional title or working in specific departments like the ICU, showed significantly poorer knowledge. To enhance the predictive diagnosis of END, it is crucial to implement targeted educational programs focusing on healthcare professionals with lower KAP scores, especially those resident trainee without professional title and those working in critical care settings, to ensure uniform expertise across all personnel.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

Xin Xia, Feng Li, Binbin Yuan, Zhengxie Dong and Yihuan Chen carried out the studies, participated in collecting data, and drafted the manuscript. Xin Xia, Feng Li, Binbin Yuan, Ya Zhang and Zixuan Wang performed the statistical analysis and participated in its design. Xinxin Huang and Qing Zhu participated in acquisition, analysis, or interpretation of data and draft the manuscript. All authors read and approved the final manuscript.

Funding

None.

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

This work has been carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association. Ethical approval for the study was granted by the Ethics Committee of RUGAO BOAI Hospital (2024-K007-01), and informed consent was obtained from all participants.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.