The study was approved by the Nursing and Behavioral Medicine Research Institutional Review Board, Xiangya Nursing School, Central South University (CSU) (IRB CSU: E202151). Written consent from the participants were obtained and confidentiality of their privacy was assured. The rights of participants to withdraw at any time were guaranteed.

This was a cross-sectional study carried out from September to November 2021 at Xiangya Hospital, CSU, Changsha, China. The sample size was determined according to two calculation methods. First, sample size = variables × (5–10) × (1 + 20%) [17]. There were 11 items in the questionnaire, and 17 items in the general data questionnaire of neurosurgical hospitalized patients, with a calculated sample size of 168–336. Second: sample size was calculated according to the statistical formula: n = \(_+_\right)S} \right]}^\) [18], Where ‘n’ is the sample size, ‘S’ is the standard deviation and ‘δ’ is the allowable error. The standard deviation (S) of VTE prevention knowledge score of neurosurgery inpatients obtained through pilot study validation was 19.281, The allowable error (δ) is set at 0.18S, α = 0.05, β = 0.10, t0.05/2 = 1.960, tβ = t0.10 = 1.282. According to the formula, n = 324, and on this basis, 15% was added to prevent sample loss. The sample size was calculated to be 373. Combining the two calculation methods, the maximum 373 were selected as the sample size to be investigated. The study sample consisted randomly selected patients from five neurosurgery units (numbers of patients selected ranged from 57 to 99) for a total of 386 hospitalized patients who met the inclusion and exclusion criteria. There was no significant difference in VTE knowledge score among the five neurosurgical units (P = 0.375).

Surveys of self-reported experiences and retrospective medical chart reviews were conducted with the sample of patients. Participants of the study were hospitalized adults aged 19 years or older in the department of neurosurgery. The routine VTE prevention measures in the hospital included chemical prophylaxis in the form of low molecular weight heparin (LMWH), and warfarin and mechanical prophylaxis as recommended by the 2020 American Society of Hematology (ASH) VTE prevention clinical guidelines [2]. Patients on the 48-beds neurosurgical units were screened for inclusion criteria: 1) hospitalization in the department of neurosurgery, 2) age > 18 years, 3) written informed consent, voluntary participation in this investigation, and 4) no delirium or dementia as per medical chart. Patients were excluded from the study if they were 1) experiencing a lack of language proficiency, 2) with coma and disturbance of consciousness, and 3) receiving palliative care or being a terminal patient.

First, the corresponding author contacted the head nurse of the five neurosurgical units of Xiangya Hospital, CSU. Permission was also received from Xiangya Hospital, CSU to apply the data collection tools, an interview was chosen as a method to complete the questionnaires.

Before the investigation: All investigators were convened by the researcher under the supervision of the supervisor of the research group to conduct unified training. The training contents and methods included VTE prevention knowledge, the contents and methods of this study, the hospital medical record information system, precautions for questionnaire collection and nurse-patient communication skills, Possible errors and problem strain methods and so on. After the training they were assigned to different neurosurgical wards to perform the study.

Patients were screened for inclusion criteria and exclusion for current cognitive impairment based on data from the medical charts, nurses on the unit and assessment by the interviewers. Neurosurgical hospitalized patients were approached by the investigator and after declaring their interest to participate in the study, they received detailed information about the aim of the study and were asked to provide informed consent.

In the investigation process: all the investigators followed the principle of objective facts to conduct the investigation, and checked the integrity and authenticity of the questionnaires collected on site. Patients could ask the investigators if they did not understand anything in the filling process, and the investigators explained the problems without any induction or inspiration, then the patients completed the work by themselves. A small number of subjects may had poor eyesight, low educational level or difficulty in filling out their own information, the investigator would read the questionnaire items one by one and explained the questions, the patient then explained the specific choice of each item.

Four hundred and seventy-one neurosurgical hospitalized patients were contacted, but 85 (18.04%) refused to participate in the study. Most of the reasons for rejection were that the hospitalized patients could not see clearly or were busy going through discharge relevant procedures. The investigators administered the questionnaires in the privacy of the patients’ rooms. Survey administration was completed in approximately 10–20 min. The medical records of participants were reviewed by the investigators to identify patients' disease diagnosis and hospital stay. Diagnosis was based on discharge diagnosis, which was divided into 8 categories: craniocerebral injury, intracranial tumor, cerebrovascular disease, scalp and skull disease, intracranial infection, spinal cord disease, functional diseases, and others, respectively. The length of hospitalization was from most recent hospital admission date to the survey date. If the patient had been admitted to the neurosurgery department multiple times, the length of previous hospitalization was added. Participants each received a toothpaste compensation for their time and efforts to complete the questionnaire.

Initial draft of questionnaire: We did not identify any VTE preventive knowledge instruments specifically designed for neurosurgical hospitalized patients. The research group members drafted the original version questionnaire after referring to the questionnaires utilized in similar contexts such as application in orthopedic patients [18]. From June to August 2021, the Delphi method [19] was used to carry out expert consultation through correspondence, a panel of 11 Clinical nurses (come from neurosurgery, neurosurgical intensive care unit, vascular surgery, cardiac surgery), 4 Clinical doctors (including Neurosurgery and Hematology), 4 nursing administrators and 1 schoolteacher were invited to ask any question, make comments to any part of the questionnaire, after 2 rounds of correspondence, the first draft of the questionnaire contains 4 dimensions and 11 questions were formed. Ten hospitalized patients of neurosurgery were invited to conduct a small sample test with a pre-test questionnaire, and the content of the pre-test questionnaire was adjusted according to the feedback of patients, to ensure the intelligibility of the questionnaire.

Pilot study validation: Then the second draft of the questionnaire was pilot-tested with a sample of 210 neurosurgical hospitalized patients who met the cross-sectional inclusion and exclusion criteria. The time required to complete the questionnaires was approximately 5–10 min for each individual. The discriminant degree analysis method, correlation coefficient method and exploratory factor analysis (EFA) were used to conduct item reduction. All items’ critical ratio > 3.000 (P < 0.001), and the correlation coefficient between each item and the total score was ranged from 0.454–0.748 (P < 0.001). The Kaiser-Mayer-Olkin (KMO) value was 0.905 and Bartlett’s test was 1271.228, degrees of freedom was 55 (P < 0.001), which was adequate for EFA. Two factors were extracted by using rotated factor analysis. The two factors jointly accounted for 62.234% variance observed. The loading value of all items were > 0.400 (Table 1). Factor 1 consists of three items, which are mainly about the recognition of VTE, DVT and PE. They are named as basic knowledge of VTE. Factor 2 includes 8 items, including the causes, risk factors, clinical manifestations, and prevention knowledge of VTE, which is named as VTE professional knowledge. The questionnaire finally formed consists of two dimensions and 11 items.

The survey consisted of two primary sections. The first section asked 17 questions concerning patient socio-demographic characteristics such as age, gender, area of residence, educational level, occupation, marital status, family monthly income (US dollars), payment manner of the medical expenses, the reason for admission, surgery, length of hospitalization, caregivers, sources of information related to VTE, whether there is a doctor/nurse in the family, personal history of VTE, family history of VTE, and whether the doctors/nurses have taught the patients about VTE knowledge. The second section consists of 11 questions measuring the knowledge of VTE (refer to the list in Table 2), including two sections: 1) The VTE basic knowledge, including three questions: ‘Have you heard of VTE?’, ‘Have you heard of DVT?’, and ‘Have you heard of PE?’ 2) The VTE professional knowledge, including nine questions: causes of VTE, the dangers of having VTE, symptoms and signs of DVT and PE, risk factors of VTE including ‘immobility or bed rest for more than three days, trauma, surgery, obesity, use dehydrating drugs, deep vein catheterization, have high blood pressure or diabetes and so on, infection, cancer/chemotherapy/radiotherapy, personal history of VTE’; preventive knowledge includes basic preventive knowledge, mechanical preventive knowledge, and drug preventive knowledge. For single choice, one point is given for "yes" and zero points is given for "no" or "uncertain". For sets of questions which could be marked, each choice except “not sure” is counted as one point (for example, “Which of the following are signs/symptoms of VTE: swelling, calf pain, redness and warmth of legs, lower extremity varicose veins, not sure). Finally, all the scores are added up to obtain the total knowledge score.

The researchers invited 10 experts (2 clinical doctors, 3 nursing administrators, and 7 clinical nurses) to help to assess the content validity by means of a quantitative method in which the Content Validity Index (CVI) was used. Version 1 of the questionnaire’s item-level CVI is 0.40–1.00, scale-level CVI is 0.86. All comments and suggestions were considered to omit potential misunderstandings. After revision the version 2 of the questionnaire contains 11 questions, item-level CVI is 0.70–1.00, scale-level CVI is 0.95, demonstrating good content validity. Confirmatory factor analysis (CFA) was performed on 210 pilot sample cases, and fitting results showed that 2/df (2, goodness of fit test) was 3.318, IFI (Increasing Fitting Index) was 0.921, CFI (Comparative Fit Index) was 0.920, AGFI (Adjusted Goodness of Fit Index) was 0.821, all of them reached the ideal standard, indicating that the fitting degree and stability of the questionnaire fitting model are good (Fig. 1). The reliability of the final version questionnaire was evaluated using fractional reliability coefficients, Cronbach’s alpha coefficients, and retest reliability. A value of 0.70 or above of Cronbach’s alpha was considered evidence of internal consistency. Cronbach’s alpha coefficient of the questionnaire is 0.851, the fractional reliability coefficient is 0.885. Thirty patients were selected and measured again 7–10 days later. The results showed that the retest reliability is 0.791, indicating that the questionnaire has good stability.

VTE knowledge questionnaire 2 factor confirmatory factor analysis model

All data analysis was conducted using Statistical Package for the Social Sciences (SPSS) version 26.0 and Analysis of Moment Structure (AMOS) version 27.0. For general data, frequency and percentage were used for descriptive analysis, mean, standard deviation and scoring rate was used for descriptive analysis of VTE knowledge score of neurosurgical hospitalized patients. (scoring rate = average score / total score × 100%, scoring rate < 60%, 60%-79%, > 80% corresponding low level, medium level, high level [17], respectively). The differences of VTE knowledge scores among hospitalized patients of neurosurgery in different age and gender, education level, economic status, and whether there was a doctor/nurse in the family, etc. were analyzed and compared using a t-test (t’ test was used when variance was not uniform) or Analysis of Variance (ANOVA), Kruskal–Wallis H test was used when variance was not uniform.

The primary outcome variable for our research is a count of correct answers based on responses to a series of 11 questions assessing knowledge of VTE. Five of these items were counted as correct or not based on Yes or No responses to simple one-sentence questions. Six items consisted of a list of possible answers to general questions (e.g., Which of the following help prevent VTE?). A total of 41 questions were scored with 1 point for each correct answer giving a possible range of knowledge scores from 0 to 41. The Fig. 2 below showed the distribution of the counts of correct answers in our sample. The distribution type of the dependent variable was an example of an over-disbursed count distribution in which the variance (132.76) was substantially greater than the mean (13.22). Stepwise regression analysis and Negative Binomial Regression (NBR) were performed to analyze the factors influencing the knowledge level of VTE prevention in neurosurgical hospitalized patients. The process was divided into two steps: 1) In the first stage, all variables were taken as independent variables, and the natural logarithm of VTE prevention knowledge score was taken as dependent variable to conduct stepwise regression analysis; 2) At the second stage, the variables that were significant predictors in the first stage were the predictors in the NBR. Table 3 shows how the independent variables were coded. P values < 0.05 were considered as statistically significant in this study. Because we make multiple tests, the true significance level should be 0.05 divided by the maximum number of tests: 0.05/12 = 0.004.

The histogram for the total knowledge score