Prediction Model for Postoperative Headache After Transsphenoidal Pituitary Surgery

1. Quality Control and Assurance of the Study

To ensure the authenticity, integrity, and accuracy of the study data, as well as to ensure that the study process fully complies with the protocol and ethical standards, the following quality control and assurance measures have been established:

1.1 Development and Implementation of Standard Operating Procedures (SOPs)

1. . Study-specific SOP: A dedicated Clinical Research Data Collection Standard Operating Procedure will be developed for this study. It will clearly define the definitions, collection methods, time points, and recording standards for each observation indicator (e.g., VAS scoring instructions, nasal patency grading criteria).
2. . Data entry SOP: Norms for entering data into the electronic data capture system will be established to ensure that data is transferred from case report forms to the database in a timely, accurate, and consistent manner.

1.2 Investigator Training and Consistency Assessment

1. . Initiation training: Before the study begins, all researchers, nurses, and data managers involved in the study will undergo unified training to ensure they fully understand the study protocol, master the assessment methods for various scales, and are aware of the requirements for data entry.
2. . Consistency assessment: For subjective indicators such as pain assessment and nasal examination, internal consistency assessments will be conducted to ensure that different evaluators have consistent criteria for judgment and to reduce measurement bias.

1.3 Management of Participant Compliance

1. . Clear communication: During the informed consent process, participants will be thoroughly informed about the significance of the study, the procedures involved, and what their cooperation entails. Reasonable expectations will be set.
2. . Convenient follow-up: The HIT-6 scale assessments at 1 and 3 months postoperatively will be integrated with participants' routine outpatient follow-ups. Reminders and follow-ups will be conducted through convenient means such as phone calls and WeChat to reduce the dropout rate.
3. . Good relationship maintenance: The research team will maintain good communication with participants, promptly answering questions and enhancing their trust and enthusiasm for participating in the study.

1.4 Monitoring of the Study Process and Data Verification

1. . Internal monitoring: A study coordinator who is not directly involved in data entry will be appointed to conduct source data verification regularly (e.g., every two weeks). A random sample of 10-15% of case report forms will be checked against the source data in the hospital's electronic medical record system (progress notes, nursing records, laboratory reports) to ensure data accuracy.
2. . Data review: The data manager will regularly run logic checks to verify the completeness, rationality, and logical consistency of the data (e.g., discharge date should not be earlier than the surgery date, VAS scores should be within a reasonable range). Any discrepancies found will be promptly queried and corrected by the researchers.

1.5 Ethical Compliance Supervision This study will strictly adhere to the principles of the Declaration of Helsinki and will be subject to the full supervision of the Ethics Committee of Beijing Tiantan Hospital. Any modifications to the study protocol, as well as any serious adverse events that occur, will be reported to the ethics committee in accordance with regulations.

2.Detailed description for variables

• age, years
• gender, 0=female 1=male
• Primary surgery/recurrent surgery, 0=NO 1=YES
• coffee/tea consumption, 0=NO 1=YES
• alcohol consumption, 0=NO 1=YES
• Smoking, 0=NO 1=YES
• diabetes, 0=NO 1=YES
• hypertension, 0=NO 1=YES
• heart disease, 0=NO 1=YES
• dyslipidemia, 0=NO 1=YES
• American_Society_of_Anesthesiologists_score_for_anesthesia, 0=I 1=II 2=III
• baseline_Beck_Depression_Inventory, 0-63
• baseline_Headache_Impact_Test, 36-78
• preoperative_headache_frequency; The frequency of monthly headache attacks
• preoperative_headache_location
• preoperative_headache_duration, minute
• baseline_VAS_score, 0-10
• preoperative_headache_associated_symptoms, 1=disgusting 2=vomiting 3=Photophobia and phonophobia
• preoperative_analgesics _usage, 0=NO 1=YES
• preoperative_analgesics_frequency
• preoperative_analgesics_effect, 1=Complete remission 2=Partial remission 3=no effect
• preoperative_nasal_passage_patency, 1=Fully patent, 2=mildly obstructed, 3=severely obstructed
• preoperative_nasal_congestion, 1=Bilateral patency 2=unilatera obstruction 3=bilateral obstruction
• preoperative_rhinorrhea, 0=NO 1=YES
• deviated_nasal_septum, 1=NO 2=right 3=left
• hypertrophy_of_middle/inferior_turbinate, 0=NO 1=YES
• sinusitis, 0=NO 1=YES
• nasal_polyps, 0=NO 1=YES
• concha_bullosa, 0=NO 1=YES
• nasal_passage_stenosis, 0=NO 1=YES
• preoperative_hormone_levels, 0=normal 1=Hypofunction 2=Hyperfunction
• tumor_size, cm3
• knosp_scale, 0-4 level
• Hardy-Wilson_extension, 0-4 level
• Hardy-Wilson_invasion, A-E level
• duration_of_surgery, hours
• blood_loss, ml
• Intraoperative_CSF_leakage, 0=NO 1=YES
• sellar_reconstruction, 0=NO 1=YES
• nasal_septal_mucosal_flap, 0=NO 1=YES
• fat_extraction, 0=NO 1=YES
• artificial_dura_mater, 0=NO 1=YES
• dura_suturing, 0=NO 1=YES
• resection_of_the_superior_turbinate, 0=NO 1=YES
• resection_of_the_middle_turbinate, 0=NO 1=YES
• nasal_packing, 0=NO 1=YES
• bilateral_nasal_packing, 0=NO 1=YES
• type_of_anesthetic_drugs
• dosage_of_anesthetic_drugs
• postoperative_intracranial_hemorrhage, 0=NO 1=YES
• postoperative_CSF_leakage, 0=NO 1=YES
• reoperation, 0=NO 1=YES
• extent_of_tumor_resection, 0=total resection 1=subtotal resection
• tumor_pathology
• postoperative_infection, 0=NO 1=YES
• daily_Visual_Analogue_Scale
• daily_headache_location
• daily_headache_type, 1=paroxystic 2=continuous 3=paroxystic \& continuous
• daily_headache_duration
• daily_headache_frequency
• daily_headache_associated_symptoms, 1=disgusting 2=vomiting 3=Photophobia and phonophobia
• daily_nasal_passage_patency, 1=Fully patent, 2=mildly obstructed, 3=severely obstructed
• daily_nasal_congestion, 1=Bilateral patency 2=unilatera obstruction 3=bilateral
• daily_analgesics _usage, 0=NO 1=YES
• daily_analgesics_frequency
• daily_analgesics_effect, 1=Complete remission 2=Partial remission 3=no effect
• daily_Na, mmol/L
• daily_K, mmol/L
• daily_urine_volume, ml
• postoperative_hormone_level, 0=normal 1=Hypofunction 2=Hyperfunction
• VAS_1_month, 0-10 point
• VAS_3_month, 0-10 point
• HIT_1_month, 36-78 point
• HIT_3_month, 36-78 point
• Postoperative_Olfaction_1_month, 0=NO 1=YES
• Postoperative_Olfaction_3_month, 0=NO 1=YES
• Postoperative_Massive_Epistaxis_1_month, 0=NO 1=YES
• Postoperative_Massive_Epistaxis_3_month, 0=NO 1=YES 3.Standard Operating Procedures for Clinical Research Data Collection 3.1. Introduction This standard operating procedure (SOP) is designed to regulate the collection and management processes of data in prospective observational clinical trials, ensuring the accuracy, integrity, and reliability of trial data. The collection of clinical trial data is a crucial step in evaluating outcomes and must adhere to SOPs to ensure the reliability and reproducibility of results.

3.2. Data Collection Process 3.2.1 Patient Recruitment and Enrollment Patients must undergo rigorous screening before enrollment according to the inclusion and exclusion criteria. Detailed records of patients' basic information and disease status should be maintained to ensure the completeness and accuracy of data.

3.2.2 Data Collection Tools The paper case report forms (CRFs) and electronic data capture systems (e.g., Excel) are used for data collection. CRFs should be complete, accurate, and easily understandable to ensure uniformity of data.

3.2.3 Data Collection Procedures

1. Before data entry, patient identity and study numbers should be verified to avoid confusion of information.
2. Data should be recorded and entered by qualified researchers to ensure accuracy and completeness.
3. Data recording should promptly and accurately reflect changes in patient status, especially outcome-related observation indicators.
4. Alteration or deletion of recorded data is prohibited. If corrections are necessary, they must be explained and confirmed with a signature.
5. If issues or missing data are identified during data entry, they should be promptly verified and supplemented with relevant personnel to ensure the integrity and reliability of the data.

3.2.4 Data Sources and Collection Points Data sources and collection points should be clearly defined, including medical records, laboratory tests, and patient self-reports.

Data sources: 1. Hospital electronic medical record system; 2. Patient self-reports; 3. CRF forms One researcher records data on the paper CRF, and another researcher enters the data into the database. The database includes standardized definitions and ranges for each variable.

3.2.5 Data Quality Management

1. During data collection, checks should be performed for logical consistency, reasonableness, and internal and external consistency to ensure data quality.
2. Outliers and missing data should be flagged and verified, with necessary supplements and corrections made.
3. Data validation should be conducted regularly to ensure accuracy and completeness.
4. After data collection is complete, data cleaning and review should be performed to correct any issues and ensure reliability.

3.3. Data Management and Storage 3.3.1 Data Management To ensure data security and confidentiality, a comprehensive data management system should be established, including data backup, access control, and secure data transfer.

3.3.2 Data Storage Data should be stored on solid-state drives for a period of three years. The security and reliability of data storage must be ensured to maintain data integrity and prevent loss.

3.4. Data Analysis and Reporting After the completion of data collection, data analysis and result reporting should be conducted. Data analysis should comply with statistical principles and relevant regulations to ensure the accuracy and interpretability of results.

3.5. Changes and Revisions This SOP may be amended or revised as needed, but such changes must be reviewed and approved by the responsible authorities and promptly communicated to trial personnel for SOP updates.

4.Sample size assessment 4.1 Calculation Basis: The primary objective of this study is to construct a predictive model for short-term postoperative headache. The sample size calculation is based on the commonly accepted standard for developing clinical prediction models - events per variable (EPV). Specifically, at least 10 positive outcome events are required for each predictor variable to be examined, to ensure the stability and reliability of the model.

4.2 Parameter Settings:

• Expected Positive Outcome Events: Based on preliminary observations and literature, the proportion of patients expected to experience moderate-to-severe headache (VAS ≥ 4) postoperatively is approximately 25%.
• Number of Predictor Variables: Initially, about 5-6 potential predictor variables (including general information, preoperative, intraoperative, and postoperative variables) will be included in the model.

4.3 Sample Size Estimation:

• Required Positive Events: 12 variables × 10 EPV = 120 events.
• Total Sample Size: Required positive events / Expected incidence rate = 120 / 0.25 = 480 cases.

4.4 Final Sample Size Determination: Considering the 1-year study period(December 2025 - December 2026) and the single-center nature of this study, based on our center's annual surgical volume (over 1,100 cases) and feasibility, while ensuring statistical power, the target sample size is set at 500 cases. With this sample size, even if the incidence rate of positive events is slightly lower than expected (e.g., 20%), it would still generate 100 positive events, meeting the basic requirement of EPV ≥ 10 and ensuring the accuracy of model construction.

5.Plan for missing data. To handle missing data and variable selection, the following methods will be adopted: 5.1 Multiple Imputation: Missing values for secondary variables will be imputed to fully utilize all available data information.

5.2 Exclusion Criteria Patients who meet any of the following criteria after enrollment will have their data excluded and not included in the final analysis

1. The postoperative pathology confirms that the tumor is not of pituitary origin.
2. The patient is transferred to the ICU for severe complications (such as intracranial hemorrhage or intracranial infection), resulting in an inability to conduct standardized daily pain assessments for more than 48 hours.
3. There is a severe loss of postoperative clinical data, with a missing rate of key variables (such as the primary outcome measure, VAS score) exceeding 20%.
4. The patient or their family voluntarily requests to withdraw from the study during the follow-up period.

6.Statistical analysis 6.1Statistical Software: All statistical analyses will be conducted using R (version 4.3.0 or higher) or IBM SPSS Statistics (version 26.0 or higher). A two-sided test will be applied, with P < 0.05 considered statistically significant.

6.2 Analysis Data Sets:

• Full Analysis Set: This set includes all participants who provided informed consent and completed at least one postoperative VAS assessment. It will be used primarily for descriptive statistics and analysis of missing data patterns.
• Complete Case Set: This dataset, used for predictive model construction and primary analyses, includes cases with no missing values for key variables. For secondary variables, missing values will be handled using multiple imputation.

6.3 Statistical Analysis Methods: 6.3.1 Descriptive Statistics:

• Continuous Data: Normally distributed data will be described using mean ± sd and analyzed with t-tests or analysis of variance (ANOVA). Non-normally distributed data will be described using median (interquartile range) and analyzed with Mann-Whitney U tests or Kruskal-Wallis H tests.
• Categorical Data: Data will be presented as counts (percentages) and analyzed using chi-square (χ²) tests or Fisher's exact tests.

6.3.2 Predictive Model Construction and Validation (for the primary objective):

• Variable Selection: Univariate analysis will be performed first, with variables having P < 0.1 included in multivariate analysis. To address overfitting and multicollinearity, LASSO (Least Absolute Shrinkage and Selection Operator) regression will be used for variable selection.
• Model Construction: Variables selected by LASSO will be included in a multivariate logistic regression model, with the occurrence of moderate-to-severe postoperative headache (VAS ≥ 4) as the dependent variable. Odds ratios and their 95% confidence intervals for each independent variable will be calculated.
• Model Performance Assessment:
• Discrimination: Receiver operating characteristic (ROC) curves will be plotted, and the area under the curve (AUC) will be calculated to assess discrimination.
• Calibration: Calibration curves will be drawn, and the Hosmer-Lemeshow goodness-of-fit test will be used to evaluate the consistency between predicted probabilities and actual observed probabilities.
• Internal Validation: Bootstrap resampling (1000 times) will be used for internal validation of the model, calculating the optimism-corrected AUC to assess the model's generalizability.

6.3.3 Correlation Analysis (for secondary objectives):

• Spearman rank correlation analysis will be used to examine the correlation between the highest postoperative VAS score and HIT-6 scores at 1 and 3 months postoperatively.