Video Versus Face‐To‐Face Preoperative Anaesthetic Assessment: The VIDFACE Trial—A Statistical Analysis Plan

ABSTRACT

Background

The video versus face‐to‐face preoperative anaesthetic assessment (VIDFACE) trial is a randomised, multicentre, two‐arm, superiority, assessor‐ and data analyst blinded clinical trial. The routine use of preoperative assessments is accepted as the gold standard for elective surgery, reducing both morbidity and mortality. In recent years, video‐based assessments have been investigated with a focus on patient satisfaction and enhancing efficacy. The VIDFACE trial aims to compare video‐based preoperative anaesthetic assessment with face‐to‐face assessments, with a focus on patient safety and satisfaction.

Objective

To report the statistical analysis plan for the VIDFACE randomised clinical trial.

Methods/Design

We will enrol 2260 adult participants undergoing elective ear‐nose‐throat or oral and maxillofacial surgery. Participants will be randomised 1:1 stratified by ‘trial site’ and ‘known or suspected malignant/haematological condition’ to either intervention (video‐based preoperative anaesthetic assessment) or control (face‐to‐face preoperative anaesthetic assessment). Two blinded statisticians will analyse the data. Before unblinding, we will write two abstracts based on the initial blinded analysis, one assuming A is intervention and B is control and one with the opposite assumption. Primary outcome is a composite outcome including five predefined serious complications. We assume a 30% relative risk reduction of the primary composite outcome. An interim analysis will take place after inclusion and follow‐up on half of the intended sample size.

Conclusion

This statistical analysis plan of the VIDFACE trial will minimise analysis bias and add transparency to the statistics applied in the results of the trial.

Trial Registration: ClinicalTrials.gov identifier: NCT06765538

Abbreviations

ASA

American Society of Anaesthesiologists physical status classification

CI

confidence interval

CONSORT

consolidated standards of reporting trials

CTU

Copenhagen trial unit

EHR

electronic health record

ENT

ear–nose–throat

GA

general anaesthesia

IQR

interquartile range

POA

preoperative anaesthetic assessment

REDCap

research electronic data capture

VIDFACE

video versus face‐to‐face preoperative anaesthetic assessment

1. Background

Preoperative anaesthetic assessment (POA) is the gold standard for elective surgery and plays a key role in optimising risk factors and reducing morbidity and mortality while increasing patient satisfaction and the overall quality of anaesthesia during surgery [1, 2, 3]. Anaesthesiologists evaluate medical conditions, perioperative risks, and procedural readiness, order tests, manage comorbidities, and coordinate care. The routine use of POA reduces unnecessary testing [4, 5], postoperative complications [3, 6, 7], as well as surgery cancellations [8] and delays [7].

One of the key aspects of a comprehensive POA is an evaluation of the airway evaluations using anamnestic and clinical assessments (e.g., mouth opening, Mallampati class, neck mobility) to identify risks and choose appropriate equipment and monitoring strategies [9]. POAs end with informed consent for the anaesthetic plan, respecting patient autonomy.

Airway complications are particularly prevalent in ear–nose–throat (ENT) as well as oral and maxillofacial surgeries [10], where patients often present with complex airways and a high prevalence of comorbidities that demand specialised resources. These challenges contribute to higher rates of same‐day surgery cancellations and delays, which disrupt hospital workflows, reduce operating room efficiency, and negatively affect patient satisfaction and healthcare costs [3, 11].

The COVID‐19 pandemic dictated a shift in how POAs were conducted, with an emphasis on providing safe and effective services while adhering to public health restrictions that limited in‐person visits. The introduction of remote POA, such as video assessments, showed potential improvements in both efficiency and patient satisfaction [12, 13, 14]. However, the impact of video‐based POAs on patient safety remains an area that has yet to be thoroughly investigated.

This study aims to explore an approach to POA—specifically, video POA—and hypothesises that it may offer superior outcomes compared to traditional face‐to‐face POA, particularly in terms of serious complications.

Here, we describe the statistical analysis plan of the VIDFACE trial.

2. Methods

This statistical analysis plan has been written according to the Guidelines for the Content of Statistical Analysis Plans in Clinical Trials [15]. It aims to give an overview and justification of the chosen variables and statistical methods used in the VIDFACE trial. All analyses will be conducted according to this plan. The VIDFACE trial has been approved by the Danish Research Ethics Committee (H‐24006615; 03 June 2024).

2.1. Trial Design

The VIDFACE trial is a randomised, parallel‐group, two‐arm, superiority, primary outcome assessor‐ and analyst‐blinded, clinical trial of video versus face‐to‐face POA. The trial population will be adults (18 years or older) who are scheduled for general anaesthesia in either ENT or oral and maxillofacial surgeries and need POA before surgery. Ineligible patients are those with under 24 h until surgery and those with severe mental challenges such as dementia. We will also exclude patients with hearing or sight impairment as well as patients who need an interpreter for POA. Patients will be eligible for enrolment if they meet the inclusion criteria and none of the exclusion criteria.

2.2. Randomisation and Blinding

All eligible participants will be randomised 1:1 according to a computer‐generated allocation sequence with permuted blocks of varying sizes from 4 to 8 participants per block. Randomisation was stratified by ‘trial site’ and ‘known or suspected malignant/haematological condition’. Patients with known or suspected malignant/haematological conditions have a higher incidence of difficult airway management cases and general anaesthesia complication rates [16]. These malignant or suspected conditions are both primary tumours in the maxillofacial area but also diagnostics of metastasising tumours/lymphomas. It is clearly stated in the electronic health record (EHR) if the reason for surgery is due to malignancy or suspected malignancy. The randomisation script was written by author O.R., with the allocation sequence list generated and uploaded to a secure web‐based Research Electronic Data Capture platform (REDCap), by a person (A.W.F) not affiliated with the study. For reproducibility purposes, the R‐script was saved.

Operating room surgical and anaesthetic personnel are blinded to the participant's randomisation allocation. Due to the trial setup, it is not possible to blind either the participants or the anaesthetist performing the POA. In the exported dataset, the randomisation groups will be renamed A and B by an unblinded person, otherwise not affiliated with the trial. All analyses and initial interpretations will be carried out using these scrambled allocations. The dataset with blinded allocation will be used for analysis by two independent trial statisticians from The Copenhagen trial unit (CTU) for blinded data analysis. Two independent reports will be compared for differences, and a final statistical report will be used to write two blinded abstracts formatted according to the initial target journal. Only after all authors agree upon both abstracts can unblinding occur.

2.3. Sample Size and Power Estimation

The sample size was determined to detect a clinically meaningful difference in the primary composite outcome (see Section 5.1) between the video group and face‐to‐face group on day 30 post‐surgery. We presume an incidence of 11.8% [17] of complications in the control group (face‐to‐face group). We chose a power of 80%, a two‐sided alpha of 5%, and a clinically meaningful relative risk reduction of 30%. We intend to be able to detect or reject a 30% risk reduction (corresponding to risk in the experimental group of 8.3%). Therefore, this trial requires a total of 2260 participants (i.e., 1130 in each group).

2.4. Statistical Interim Analysis and Stopping Guidance

When information on 30‐day follow‐up has been collected for 1130 participants (50% of the full sample size) an interim analysis will be performed. There will be an independent Data Safety Monitoring Committee (DSMC) arranging for an independent statistician to conduct a blinded interim analysis. If deemed necessary by the DSMC, they will be able to review unblinded data. The DSMC can initiate analysis at any time they request. In the DSMC analyses, the alpha will be adjusted based on Lan‐DeMets approximation of O'Brien–Fleming boundaries. The DSMC may terminate or pause the trial if a group difference for the primary composite outcome measure is found according to pre‐defined stopping rules:

• ○Stopping for efficacy. The video group shows significant positive results during an interim analysis.
• ○Stopping for harm. The video group shows evidence of harm during the interim analysis.

2.5. Timing of Outcome Assessments and Final Analysis

Total duration of involvement for each participant includes the assessment and an online participant survey. The participant's electronic health record will be reviewed in relation to the POA, serious complications and a 30‐day follow‐up after surgery.

The registered baseline variables are presented in Table 1. The primary outcome and key secondary outcomes are presented in Tables 2 and 3. Primary analysis can begin after a 30‐day follow‐up on all included participants.

TABLE 1.

Baseline demographic and clinical characteristics.

Variable	Definition
Age	Calculated from birth at inclusion (year)
Sex	Genotypic (male/female)
Glasgow coma scale score a	3–15
Height a	Measured in centimetres
Weight a	Measured in kilogram
Active smoker a	More than zero cigarettes per day
Comorbidities a	Any of the following stated in the EHR
Neurological disease a	Epilepsy, psychological, stroke and other
Respiratory disease a	Chronic obstructive pulmonary disease, asthma, sleep apnoea and other
Cardiovascular disease a	Hypertension, angina, atrial fibrillation, heart failure, coronary artery disease and other
Gastrointestinal disease a	Reflux, ileus, liver‐ or kidney disease and other
Endocrine disease a	Diabetes, hypo‐ or hyperthyroidism and other
Musculoskeletal system diseases a	Rheumatoid arthritis, neck pain and other
Previous problems with anaesthesia	PONV, allergic reactions, cardiovascular complications and other
ASA physical status classification a	Number 1–5
Mallampati a	Number 1–4
Mouth opening a	< 2.5, < 4 or > 4 cm
Neck mobility a	Full or limited
Thyromental distance a	> 6.5 or < 6.5 cm
Ability to protrude jaw a	Full or limited

Abbreviation: ASA, American Society of Anaesthesiologists.

^a Data extracted from the anaesthetic journal from the POA.

TABLE 2.

Primary composite outcome.

Variable	Definition
30‐day post‐surgery mortality	Yes/No
30‐day post‐surgery all‐cause admission to intensive care	Yes/No
30‐day post‐surgery bleeding requiring transfusion	Clearly stated in the participant's medical record by a physician or assigned as a diagnosis
30‐day post‐surgery infection requiring antibiotics	Blood products entered in the EHR
30‐day post‐surgery hospital readmission	Counting day of surgery as index day

TABLE 3.

Secondary outcomes.

Variable	Definition
Hospital‐free days within 30 days post‐surgery	Days admitted to hospital, counting day of surgery as index day
The proportion of participants with one or more airway‐related serious complications	Peri‐intubation hypoxia with an oxygen saturation of SpO2 < 93% anytime from preoxygenation until 1 min after intubation Dental trauma Lip or tongue trauma (bruising, lacerations, bleeding, swelling etc.) Upper airway tract trauma (bleeding, laceration) Vocal cord injury (hoarseness, difficulty speaking, vocal cord paralysis) Oesophageal injury (perforation, laceration, bleeding)
Time necessary to spend from home to hospital	Minutes
Time spent on assessment	Minutes
Participant preference	Video or face‐to‐face POA
Participant satisfaction with information in the POA	1 = very dissatisfied, 2 = dissatisfied, 3 = neutral, 4 = satisfied 5 = very satisfied

3. Statistical Principals

All analyses will be conducted according to the intention‐to‐treat (ITT) principle, i.e., all randomised participants will be included. A per‐protocol sensitivity analysis will be performed if the number of participants not receiving the allocated POA exceeds 5% in either group.

We will assess if the thresholds for statistical significance and clinical significance are crossed (Bayes factor calculations will be reported in Supporting Information) [18]. Assessment of clinical significance will be based on the anticipated effects used in the sample size/power estimations [18]. Our primary conclusion will be based on the primary outcome, so all tests of statistical significance (including subgroup analyses) will be analysed using a two‐sided alpha of 0.05 and presented with 95% confidence intervals (CI) [18].

It is generally acknowledged that regression analyses ought to be adjusted for the stratification variables used in the randomisation [19, 20]. The VIDFACE trial uses two stratification variables in the randomisation; ‘trial site’ and ‘known or suspected malignant/haematological condition’ to balance prognostic baseline characteristics across randomisation groups. We will also assess whether there are significant interactions between the VIDFACE trial and the stratification variables.

3.1. Analysis of Dichotomous Data

Dichotomous data will be presented as proportions of participants in each group, relative risks, and corresponding 95% CIs. Dichotomous outcomes will be analysed using generalised mixed effects logistic regression models with a log link—with ‘known or suspected malignant/haematological condition’ as fixed effect and ‘trial site’ as random effects. The primary outcome is a composite outcome and will secondly be analysed using win ratio [21].

3.2. Analysis of Continuous Data

Continuous outcomes will be presented as means and standard deviations for each group, along with 95% CIs for means of the groups and mean differences between the groups. For distributions that are not normally distributed, the median and interquartile range (IQR) will be reported as measures of central tendency and variability, respectively. Continuous outcomes will be analysed using generalised mixed effects linear regression models—with ‘known or suspected malignant/haematological condition’ as fixed effect ‘trial site’ as random effects.

3.3. Analysis of Count Data

Count data will be presented as medians, Hodges Lehman estimator for median differences, and interquartile ranges or means, mean differences, and 95% CIs depending on the observed distribution. Count data will be analysed by the van Elteren test with ‘trial site’ as stratification variable.

3.4. Assessments of Underlying Statistical Assumptions

We will systematically evaluate the underlying statistical assumptions for all statistical analyses [22, 23]. For both primary and secondary regression analyses, we will test for significant interactions between each covariate and the intervention variable. In examining interactions, we will include all possible first‐order interactions between included covariates and the video‐intervention variable. For each combination, we will assess its significance and effect size. We will consider an interaction to be statistically significant only if it meets Bonferroni adjusted thresholds (p < 0.025, calculated as 0.05 divided by the number of potential interactions). For example, the interaction between the treatment variable and ‘trial site’ or between the treatment variable and ‘known or suspected malignant/haematological condition’. We will evaluate if it shows a clinically meaningful effect. If an interaction is found to be significant, we will present separate analyses for each interaction as well as an overall analysis that includes the interaction term in the model [22, 23].

3.5. Assessments of Underlying Statistical Assumptions for Dichotomous Outcomes

We will assess if the deviance divided by degrees of freedom is significantly larger than 1 to evaluate for relevant overdispersion. Overdispersion is the presence of greater variability (statistical dispersion) in a dataset than expected based on a given statistical model. In this case, a maximum likelihood estimate of the dispersion parameter will be used. We cannot exclude the risk that some trial sites might have problems with recruitment. However, to avoid analytical problems such as zero events or an overrepresentation of events in the malignant group, we have planned large trial sites that operate on both malignant and benign conditions.

3.6. Assessments of Underlying Statistical Assumptions for Linear Regression

We will visually inspect quantile‐quantile plots of the residuals to assess if the residuals are normally distributed and use residuals plotted against covariates and fitted values to assess for homogeneity of variances. If plots show deviations from the model assumptions, we will consider transforming the outcome, e.g., using log transformation, square root, and/or robust standard errors.

3.7. Adherence and Protocol Deviations

Protocol adherence will be described in a table reporting the number of nonadherent participants in accordance with the Principles of Good Clinical Practice. Protocol deviations are defined as incorrect inclusion, wrong type of POA, and failure of informed consent. Wrong type of POA can happen if technical difficulties occur in the video group. Then the POA will be conducted either by telephone or face‐to‐face. Conversely, if participants fail to meet for face‐to‐face POA, they will receive video POA. In some cases, if participants fail to attend either video or physical POA, the EHR will be evaluated and no POA will be performed.

4. Trial Population

4.1. Screening, Eligibility, and Recruitment

Screening, eligibility, and recruitment will be conducted as stated in the trial protocol. The trial population will be screened based on bookings for a POA from the ENT, oral, and maxillofacial departments. In other words, patients will be screened daily when scheduled for surgery and subsequently booked for a POA. A template for a CONSORT [15] flow diagram is presented in Figure 1.

FIGURE 1.

CONSORT flow diagram of screening, randomisation, and follow‐up.

4.2. Baseline Participant Characteristics

Baseline variables for all analysed participants are presented in Table 1. Data will be summarised with numbers and percentages for categorical variables and medians with IQRs for continuous variables.

5. Analysis

5.1. Outcome Definitions

Outcomes are defined as primary and key secondary outcomes. Sample size was based on the primary outcome, and our primary conclusions will be based on the results of the primary outcome. Definitions of the outcomes can be seen in Tables 2 and 3.

5.1.1. Primary Composite Outcome

1. 30‐day post‐surgery mortality.

2. 30‐day post‐surgery all‐cause admission to intensive care.

3. 30‐day post‐surgery bleeding requiring transfusion.

4. 30‐day post‐surgery infection requiring antibiotics.

5. 30‐day post‐surgery hospital readmission.

5.1.2. Key Secondary Outcomes

• Hospital‐free days within 30 days post‐surgery.

• Proportion of participants with one or more airway‐related serious complications defined as:

  • ○
    Peri‐intubation hypoxemia.
  • ○
    Dental trauma.
  • ○
    Lip or tongue trauma.
  • ○
    Upper airway tract trauma.
  • ○
    Vocal cord injury
  • ○
    Oesophageal injury
• Participant satisfaction survey.

  • ○
    Time necessary to spend from home to hospital.
  • ○
    Time spent on assessment.
  • ○
    Participant preference (video vs. face‐to‐face assessment).
  • ○
    Satisfaction with information given by the anaesthesiologist (Likert scale) [24].

5.1.3. Exploratory Outcomes

• Incidence of uncomplicated intubations defined as an intubation difficulty scale score ≤ 5 (easy to intubate).

• Change in anaesthetic plan on the day of surgery.

  1. Change in airway management to awake procedure.
  2. Change from non‐invasive to invasive blood pressure monitoring.
  3. Change to continuous vasopressor infusion at anaesthetic induction.

• Number of same‐day surgery cancellations.

• Time delay for surgical start (minutes). ¹

• Did not attend rate for video versus face‐to‐face POA.

5.2. Missing Data

All randomised participants will be included in the primary analysis of all outcomes. Missing data will be handled according to the recommendations by Jakobsen et al. [25] In short, we will consider using multiple imputation based on a flow diagram [25], and present best–worst and worst–best case scenarios if it is not valid to ignore missing data [25]. Best–worst and worst–best case scenarios assess the potential range of impact by the missing data on the trial results [25]. In the ‘best–worst’ case scenario, it is assumed that all participants lost to follow‐up in the video group have had a beneficial outcome (no serious complications, no higher number of hospital days, and so forth), and all those with missing outcomes in the control group have had a harmful outcome (have had serious complications, have had a higher number of hospital days). Conversely, in the worst–best scenario, it is assumed that all participants who were lost to follow‐up in the video group have had a harmful outcome and that all those lost to follow‐up in the control group have had beneficial outcomes [25]. When continuous outcomes are used, a beneficial outcome will be defined as the group mean minus two standard deviations of the group mean (fixed imputation) [25].

5.3. Subgroup Analysis

Six subgroups will be analysed. All subgroups will be analysed using ITT.

• Age(< 45, ≥ 45; hypothesising that there will be more serious complications in the older participant group) [26].

• Sex (male/female; hypothesising that there will be no difference between the participant group) [27].

• ASA classification (I–II, > II; hypothesising that ASA class > II will have more serious complications) [28].

• Cardiovascular (yes/no; hypothesising that participants with cardiovascular diseases will have more serious complications than the people without cardiovascular diseases) [29].

• Sleep apnoea (yes/no; hypothesising that participants with sleep apnoea will have more serious complications than the people without sleep apnoea) [30].

• Longer operation (≤ 1, > 1 h; hypothesising that participants with planned operations > 1 h will have more serious complications than participants with operations shorter than 1 h) [31].

6. Discussion

The primary aim of this present publication is to minimise the risk of outcome reporting bias, p‐hacking, and erroneous data‐driven results. We, therefore, present a pre‐defined description of the statistical analysis plan for the VIDFACE trial.

6.1. Strengths

Our trial has several strengths. The methodology is described in detail and is predefined, and our chosen outcomes are patient centred. We will analyse data in accordance with the ITT principle and, if necessary, use multiple imputation and best–worst/worst–best case scenarios to assess the potential impact of missing data on the results. Furthermore, we plan to systematically assess whether underlying statistical assumptions are fulfilled for all statistical analyses. Our stratification for known or suspected malignant/haematological conditions will ensure that possible difficult airways are equally distributed into the two groups and thereby reduce the likelihood of confounding variables.

6.2. Limitations

A potential limitation of the VIDFACE trial is the potential biased impact of non‐blinded participants and investigators on the treatment of the participants. We will assess a composite primary outcome on which our primary conclusions will be based, and we will assess several secondary outcomes, exploratory outcomes, and subgroup analyses that increase the risks of type 1 errors. Our trial population is ENT and oral and maxillofacial patients. Therefore, it may not apply directly to other patient groups, such as abdominal patients, who may have a different set of health concerns (e.g., cardiovascular risks), which may limit the generalisability of the results.

7. Statistical Software

All trial data will be collected through EHR and entered into the REDCap system. Participant data will include baseline demographics, clinical measurements, airway evaluation, and serious complications. Data will be entered by authorised trial site personnel and monitored for completeness weekly. The data will be handled with the analysis population as intent‐to‐treat.

7.1. Changes to the Statistical Analysis Plan

Throughout the trial, the statistical analysis plan may be subject to revisions based on emerging data or regulatory requirements. The anticipated types of modifications that could occur are adjustments of statistical methods, changes in the handling of missing data, and revised statistical significance thresholds. The corresponding author of the final manuscript will clearly document and date all changes to the statistical analysis plan. Principal investigator K.B.B. will approve any modifications made after the trial has commenced. This section ensures that all modifications to the initial statistical analysis plan are transparently recorded, providing an accurate account of how the statistical methodology evolved during the trial period.

8. Dissemination

The data presented in this manuscript will constitute the basis for the primary publication of the study, which will be submitted to a peer‐reviewed medical journal. All results will be sought to be published, including both positive, negative, and inconclusive results.

9. Status

The VIDFACE trial began recruiting on 04 November 2024. As of 06 May 2025, 203 participants have been included. We expect the last participant to be included in early 2027.

Author Contributions

Conceptualisation: Sofie A. Bosholdt, Janus C. Jakobsen, Morten H. Møller, Rasmus T. Hesselfeldt and Katrine B. Buggeskov. Methodology: Sofie A. Bosholdt, Janus C. Jakobsen, Morten H. Møller and Katrine B. Buggeskov. Software: Sofie A. Bosholdt and Oscar Rosenkrantz. Validation: Janus C. Jakobsen, Markus Harboe Olsen and Katrine B. Buggeskov. Formal analysis: Janus C. Jakobsen, Markus Harboe Olsen. Investigation: Sofie A. Bosholdt and Moritz Kilian German Denneborg. Resources: Sofie A. Bosholdt, Moritz Kilian German Denneborg, Rasmus T. Hesselfeldt, Katrine B. Buggeskov. Data curation: Sofie A. Bosholdt and Moritz Kilian German Denneborg. Writing – original draft: Sofie A. Bosholdt, Oscar Rosenkrantz, Janus C. Jakobsen and Katrine B. Buggeskov. Writing – review and editing: Sofie A. Bosholdt, Moritz Kilian German Denneborg, Oscar Rosenkrantz, Janus C. Jakobsen, Morten H. Møller, Markus Harboe Olsen and Katrine B. Buggeskov Supervision: Katrine B. Buggeskov. Project administration: Katrine B. Buggeskov. Funding acquisition: Katrine B. Buggeskov.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.