Temporal trends and current practice patterns for intraoperative ventilation at U.S. academic medical centers: a retrospective study

This study received an exemption from institutional review board (IRB) approval by the Vanderbilt University Human Research Protection Program-Institutional Review Board, an exemption from IRB approval from the Colorado Multiple Institutional Review Board, IRB approval from Mayo Clinic’s Office for Human Research Protection-Institutional Review Board, IRB approval from Partners Human Research Committee, and a determination of non-human subject research from the Division of Human Subjects Protection at Thomas Jefferson University. At each institution, the requirement for written, informed consent from patients was waived. Patients aged 18 and older who received general anesthesia with tracheal intubation between 2005 and 2013 were screened. Each institution defined its own start date based on the availability of electronically captured ventilator data. Patients who underwent procedures with potential confounding ventilation considerations were not included, specifically those undergoing cardiac, thoracic, neurosurgical, laparoscopic and one lung ventilation procedures, as well as organ harvest. Cases without adequate data for analysis were excluded as described below.

Data were retrieved electronically from each institution’s anesthesia information management system (AIMS). At each site, AIMS data were extracted and de-identified. This was achieved by creating a de-identified case number, listing patients 90 years or older with an age of 90, removing date information with the exception of case year and calculating all physiologic and ventilator data timestamps relative to entry into the OR. Files were encrypted and transmitted using a secure website to Vanderbilt University, the coordinating institution, and were then loaded into a SQL Server database (Microsoft, Redmond, WA). The relevant data elements transmitted for each case were institution number, de-identified case number, year, American Society of Anesthesiologists (ASA) Physical Status, emergency vs. elective case status as defined by the presence of an ASA E flag, height in centimeters, weight in kilograms, age, gender, primary surgical service, surgical Current Procedural Terminology (CPT) code, anesthesia CPT code, total case duration (i.e., entry to exit from room), times of surgical incision and end of surgery, use of lung isolation, use of a laparoscopic approach, and whether the case was a cardiac, thoracic or neurosurgical case. For each case, the following ventilation data were retrieved: peak inspiratory pressure, PEEP, exhaled tidal volume, mean airway pressure, respiratory rate (RR), ventilation mode, fraction of inspired oxygen (FiO₂), oxygen saturation via pulse oximetry (SpO₂) and end-tidal carbon dioxide (EtCO₂). Not all data elements used to define the surgical procedure were available at all institutions; thus multiple data elements were included to define case exclusions.

The minimum, maximum, average and median of each data element at each institution between incision and end of surgery were examined to rule out systematic problems with the data extraction process. The exclusion criteria were sequentially applied to the screened cases (Figure 1). Patients missing heights, or with a height less than 152 cm or greater than 203 cm were excluded to be consistent with prior work [13], as were those with unknown or missing genders, weight below 40 kg or above 300 kg, oxygen saturation (SpO2) above 100% or with median values below 50%, and median tidal volumes below 3 mL per kg of PBW. PBW was determined using gender and height per ARDSnet definition [16]. Specifically, for males, PBW was calculated as [(height (cm) -154) × 0.9] + 50. For females, PBW was calculated as [(height (cm) -154) × 0.9] + 45.5.

We excluded cases where there may have been overriding patient or procedural factors that may have altered routine ventilation strategies, such as one lung ventilation, cardiopulmonary bypass, elevated intracranial pressure, and use of laparoscopy. Specifically, we excluded thoracic, cardiac, and neurosurgical cases, and cases with use of laparoscopy. Thoracic cases were defined based on the primary surgical service field, a positive value for the one lung ventilation flag, or an anesthesia CPT code indicating one lung ventilation. Cardiac cases were defined based on the primary surgical service field. Neurosurgical cases were defined based on the primary surgical service field or an anesthesia CPT code indicating an intracranial procedure. Non-intracranial procedures performed by neurosurgical services were excluded. Laparoscopic cases were defined as those with a positive value for the laparoscopic flag or the presence of ‘laparoscopic’ at any point in the text description of the surgical procedure.

Demographic variables were summarized with means and standard deviations for continuous variables and with percentages for categorical variables. Ventilator parameters were defined within each case at the 25^th, 50^th and 75^th percentiles, restricting the analysis to the interval between incision and end of surgery. All tidal volumes refer to exhaled tidal volumes. Cases were divided into a three category ordered outcome with median tidal volumes > 10 mL per kg of PBW (traditional), 8–10 mL per kg of PBW (intermediate) and < 8 mL per kg of PBW (physiologic tidal volume during quiet breathing at rest), breakpoints selected to differentiate ventilation consistent with ARDSnet recommendations of 6–8 mL per kg of PBW [16] from traditional strategies of > 10 mL per kg used to prevent atelectasis [15], while also including an intermediate category. The primary analysis was to determine if most anesthetics performed in 2013 used a median tidal volume of < 8 mL per kg of PBW and if most of those anesthetics used PEEP. As a secondary analysis, a proportional odds model was used to examine factors associated with a higher tidal volume category (traditional vs. intermediate, intermediate vs. physiologic) using predictors previously identified [17] and adding institution and year. The pre-specified covariates were year, institution, gender, height and actual body weight. Body weight was modeled rather than body mass index, as height was already included in the model. As the number of included cases was large, height and body weight were modeled using restricted cubic splines with 4 knots (placed at quartiles 0.05, 0.35, 0.65, 0.95) in order to identify non-linear effects. The remaining parameters were modeled as categorical variables. The odds of receiving a lower category of tidal volume (traditional vs. intermediate, intermediate vs. physiologic) were computed for height and weight as interquartile odds ratios, which represent comparisons between the 25^th and 75^th percentile values for those parameters. Similarly, odds ratios were computed for categorical variables, using female gender, the year 2005 and institution A as references. We determined the mean values of all ventilation parameters on a per-year, per-institution basis, in addition to a per-year basis for all institutions for tidal volume and PEEP. Tidal volume per mL of actual body weight and per mL of PBW were compared. Two-tailed, independent t-tests were used for comparison of continuous variables. Data are presented with means and standard deviations, except where otherwise noted. The significance of differences between proportions were compared using z-ratios with two-tailed, independent binomial proportions tests. Statistical inference was performed using a significance level of 0.05. All analyses were performed using R (R Foundation for Statistical Computing, Vienna, Austria).