Introduction

Respiratory failure necessitating mechanical ventilation and respiratory complications such as pneumonia are the main causes of morbidity and mortality following acute spinal cord injury (SCI) [1,2,3,4,5]. The prevention of pneumonia and reaching ventilator liberation early in the rehabilitation course not only increases life expectancy, but also improves functional outcomes, quality of life, and decreases health care costs [1, 6, 7]. Optimal ventilator settings have been shown to reduce pneumonia and facilitate early weaning in people with and without acute respiratory distress syndrome [8,9,10,11,12,13,14,15]. However, there is a lack of robust research with regards to optimal ventilator settings to prevent morbidity and mortality in people with acute SCI.

Current SCI clinical practice guidelines [16] recommend high tidal volume (HVt) of >15 ml/kg predicted body weight (PBW) ventilation which is based on a small retrospective study [17]. Further support for this practice has come from a recently published small randomized control trial (n = 33) in subacute SCI comparing Vts of 10 and 20 ml/kg PBW, which did not find a difference in rate of pneumonia and ventilator weaning days [18]. People with cervical level SCI experience weakness of inspiratory and abdominal muscles resulting in hypoventilation and inability to clear secretions. HVt is recommended to prevent atelectasis and improve lung capacity in preparation for ventilator weaning for people with acute SCI [16, 18]. In the USA, SCI model systems have adopted these recommendations based on biological plausibility without strong evidence to support benefits. In addition, there are no studies which investigated long-term outcomes associated with HVt ventilation in SCI.

In contrast to the SCI recommendations, the Acute Respiratory Distress Syndrome (ARDS) Network protocol recommends smaller tidal volume ≤ 6 ml/kg PBW and positive end expiratory pressure (PEEP) to prevent secondary respiratory complications and decrease mortality in people with ARDS, based on robust evidence from multicenter clinical trials [8, 9, 19]. Following the ARDS Network trial, there have been several randomized clinical trials (RCTs) conducted in other diagnoses in which Vt of >8–12 ml/kg PBW and <6–8 ml/kg PBW were compared [10,11,12,13,14,15]. Results from these trials revealed increased risk of acute lung injury (ALI) and pneumonia in HVt groups compared to low Vt groups [10,11,12,13,14,15]. The presumed mechanism for worsening morbidity with increased Vt ventilation has been called volutrauma—overstretched alveoli from HVt causing ventilator-associated lung injury and pneumonia. In animals, ventilation with the use of large Vt caused the disruption of pulmonary epithelium and endothelium, lung inflammation, atelectasis, hypoxemia, and the release of inflammatory mediators [20,21,22,23].

Because of the risk of injury supported by animal research and the studies in non-SCI human populations, as well as the lack of a high level of evidence to support the ventilator settings in the SCI population, it is prudent to revisit the recommendations that are being widely followed. To address this gap in evidence in SCI, we performed a retrospective cohort study to evaluate the safety and efficacy of high HVt used in patients with acute SCI in the rehabilitation setting. The primary hypothesis was that HVt ventilation would be associated with a higher incidence of pneumonia and poor respiratory outcomes.

Methods

A retrospective cohort study of patients with acute SCI who were consecutively admitted to an acute inpatient rehabilitation (AIR) facility from 2015 to 2019 on mechanical ventilation with a tracheostomy was conducted. We performed data collection to measure exposure and outcomes until final discharge from our AIR hospital. Data were extracted from the electronic medical records retrospectively. Patients were excluded if they only required nocturnal ventilation, duration of SCI > 1 year, if they had a diaphragmatic pacer, diagnosis of ARDS, age < 18 years, required long-term IV antibiotics, severe dysphagia due to cranial nerve injury, and emergent transfer within 72 h of admission.

Collected data included age, sex, weight, height, date of injury, date of admission to AIR, mechanism of injury, American Spinal Injury Association (ASIA) impairment scale (AIS), vital capacity at admission and maximum vital capacity achieved during AIR stay, ventilator settings including Vt, and discharge location. Patients were retrospectively divided into two groups based on the maximum Vt received, calculated as ml/kg of PBW. People who received <15 ml/kg PBW were included in moderate Vt (MVt) group and those who received >15 ml/kg PBW were included in HVt group. In general, Vt is increased gradually after admission until a target Vt is achieved which is determined by attending physician based on atelectasis on admission x-ray, age, weight, and patient tolerance. They generally remain on this maximal Vt for the duration of their mechanical ventilation days in AIR. In instances when the maximum Vt is not tolerated, the Vt is reduced. Hence, we defined the Vt as the maximum Vt tolerated, used for the longest duration for mechanical ventilation.

Ventilator weaning

Ventilator weaning is managed by board-certified SCI medicine rehabilitation physicians in our institution. Internal Medicine physicians are consulted on all of our mechanically ventilated patients. Pulmonologists are available for consultation on as needed basis. On admission, ventilators are placed on assist control mode. The tracheostomy cuff is deflated and PEEP is weaned within 1–2 days after admission to AIR. Tidal volumes are increased at 100–200 ml increments with close monitoring of peak pressures every 1–2 days. The ventilator weaning protocol begins once the patient is determined to be medically stable by the attending physician and satisfies the following institutional criteria: lung auscultation with clear or relatively clear breath sounds, vital capacity > 12 ml/kg PBW measured by Wright’s spirometer via mouth or tracheostomy, chest x-ray clear or stable, arterial blood gas analysis on room air and without PEEP is within normal limits, no active infection, hemoglobin > 10 gm/deciliter and should be tolerating tracheostomy cuff deflation. The institutional weaning protocol consists of 20 days progressive spontaneous breathing which is used as a guide. Weaning schedule is adjusted sometimes due to respiratory complications, anxiety, and inability to tolerate protocol.

The primary outcome measure was the incidence of pneumonia (count data) occurring at least 48 h after admission to AIR or if diagnosed within 48 h of transfer from AIR to the acute care hospital due to respiratory complications [24]. Pneumonia was defined as new or progressive with persistent infiltrate on chest x-ray plus 2 of the following: abnormal white blood cell count (<4000 or >12,000), presence of fever or hypothermia (<36 °C or >38 °C), purulent sputum, or deterioration of gas exchange [24]. We also created a composite binary variable of pulmonary adverse events consisting of at least one of the following: pneumonia, transferred to acute care due to respiratory complications (other than pulmonary embolism), or failed to wean off ventilator for any duration in spite of vital capacity of >12 ml/kg PBW. Secondary outcomes included preweaning days, weaning days, and total ventilator days as defined below:

  1. (1)

    AIR preweaning days: Number of days from AIR admission to start of ventilator weaning. This applies to only people who were successfully weaned. Days on ventilator prior to admission to AIR was not included in this data.

  2. (2)

    Ventilator weaning days: Number of days from initiation of weaning to end of weaning (24 h off ventilator). This applies to only people who were successfully weaned.

  3. (3)

    AIR Ventilator days: Ventilator days from time of AIR admission to AIR discharge. All patients were included in this analysis.

For the purpose of calculating AIR preweaning days, weaning days, AIR ventilator days, and AIR admission to discharge days, we included the days that lapsed due to an acute care transfer for any acute emergencies during their stay in AIR facility. In addition, we also collected data on improvement in vital capacity, peak pressures, discharge location and AIR admission to discharge days.

Statistical analysis

Descriptive analyses were performed to compare baseline demographics. Median values with interquartile range (IQR) or means with standard deviation (SD), when normally distributed, were used to describe continuous data. Categorical and ordinal data were reported as totals and frequencies. Baseline variables and variables of interest were compared using t test if data were normally distributed otherwise using Wilcoxon rank-sum test for continuous variables and chi square for categorical and ordinal variables. Shapiro–Wilk test was performed to test for normality for each continuous variable. Multivariable frequentist regressions were utilized to assess the relationship between outcomes and exposure (Vt) while adjusting for potential confounders.

We calculated risk ratios (RR) and 95% confidence interval (CI) for pneumonia and odds ratio (OR) for the binary composite outcome for HVt group compared to MVt group using Poisson regression models [25, 26] and logistic regression models, respectively.

Negative binomial regression models [25, 26] were fit separately for secondary outcomes of preweaning days, ventilator weaning days, and total ventilator days. We adjusted for age, sex and vital capacity at admission as potential confounders for analysis of following dependent variables: pneumonia, composite outcomes, preweaning days, ventilator weaning days, and total ventilator days. We used Akaike Information Criterion to compare models. All associations were reported as RR or OR with 95% CI. All data analyses were completed using StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP. Sample size calculations were not performed due to lack of information on effect size.

Results

There were a total of 140 patients with SCI who were consecutively admitted to the AIR facility on mechanical ventilation from 2015 to 2019. Fifty-six patients were excluded based on exclusion criteria, leaving 84 tracheostomized patients with acute SCI dependent on mechanical ventilation at the time of admission (Fig. 1).

Fig. 1: Flow chart with reasons for exclusion.
figure 1

SCI spinal cord injury; ARDS acute respiratory distress syndrome; CN cranial nerve; Vt tidal volume.

For comparison purpose, patients were retrospectively divided into two groups based on Maximum Vt received in AIR. There were 50 and 34 patients in the Moderate, and HVt groups, respectively. No one in this cohort received low Vt of <6–8 ml/kg PBW [8]. Patients in MVt group received Vt between 8.4 and 14.9 ml/kg PBW; and HVt group received 15–20 ml/kg PBW. The mean (SD) tidal volume for MVt group was 13 (1.3) ml/kg PBW, and for HVt was 17 (1.4) ml/kg PBW. The 95% CI for MVt group was 12.7–13.4 and for HVt group was 16.4–17.4 ml/kg PBW. Demographics and baseline characteristics are compared in Table 1. There were no significant demographic differences or SCI characteristics between the groups except for mean tidal volume (p < 0.00). However, patients were predominantly male. Median age was higher for HVt (43 years) compared to MVt group (33 years), though not statistically significant (p = 0.2). To account for these differences at baseline, we adjusted our analysis for age and gender. We also included vital capacity at admission to account for severity of respiratory impairment at admission to AIR.

Table 1 Demographic data.

Primary outcomes: HVt group had increased incidence (Table 2) and the risk of pneumonia (Table 3) compared to MVt group. The risk of pneumonia in HVt group compared to MVt group was 4.3 times higher (95% CI: 1.5–12). We also found that higher vital capacity at admission was associated with lower risk of pneumonia. An increment in vital capacity on admission by 1 ml/kg PBW was associated with 10% decreased risk of developing pneumonia (RR: 0.9, 95% CI: 0.83–0.98) (Table 3). Similarly, incidence (Table 2) and odds of the composite outcome of pulmonary adverse events were higher in HVt group compared to MVt group with OR of 5.4 (95% CI: 1.8–17) (Table 3). Regression analysis with tidal volume as continuous outcome revealed that for every 1 ml/kg PBW increment in Vt, risk of pneumonia increased by 28% (RR: 1.28, 95% CI: 1.1–1.6) and odds of the composite outcome of developing any pulmonary adverse events increased by 42% (OR: 1.4, 95% CI: 1.1–1.8) (Table 4).

Table 2 Results: unadjusted analysis.
Table 3 Results: primary and secondary outcomes regression analysis after adjusting for potential confounders.
Table 4 Primary outcomes: regression analysis after adjusting for confounders with tidal volume as continuous dependent variable.

Secondary outcomes: Overall, 87% of patients were successfully weaned off mechanical ventilator with no significant difference in rates of weaning between two groups (Table 2). Reason for failure to wean off ventilator was low vital capacity of <12 ml/kg PBW for all except two patients in HVt group who failed to wean in spite of vital capacity of 14 and 35 ml/kg PBW at the time of discharge from AIR. On review of documentation, anxiety off ventilator was interfering with weaning. Median days from time of SCI to time of admission to AIR facility was similar in both groups (Table 1).

There was no statistical difference in AIR preweaning days, ventilator weaning days, or AIR ventilator days between the two groups in either unadjusted or adjusted regression analysis (Tables 2 and 3). However, higher vital capacity at admission was associated with lower AIR preweaning and AIR ventilator days (Table 3; Secondary outcomes). There was no statistical difference in AIR admission to discharge median days between MVt group and HVt group (Table 2). In addition, neither ARDS nor barotrauma occurred in any patients in the two groups. Median peak pressures were <35 cm H2O for all patients with slightly lower peak pressures in MVt group compared to HVt group (median 19 vs. 21, p = 0.00) (Table 2). Though this difference is statistically significant, it is not clinically important given that values were below harmful range. In MVt group, 80% were discharged to home compared to 65% in HVt group (p = 0.04) (Table 2).

Discussion

This is the first study to investigate the association between tidal volumes on rates of pneumonia and other undesirable respiratory outcomes in an AIR facility as the primary outcome. We report findings contrary to the accepted guidelines for ventilator weaning in SCI. Our data show higher risk of pneumonia and negative respiratory outcomes with HVt. Our outcomes suggest that lower tidal volume ventilation may decrease incidence of pneumonia and adverse pulmonary events in the SCI patient population.

Many studies revealed that most alveolar cells are capable of producing inflammatory mediators when ventilated with a large tidal volume (Vt) in ex vivo and in vivo animal studies [20, 22, 23, 27, 28] and in vitro studies of human lung cells [29,30,31]. In vivo human studies suggest that the use of lower VTs and PEEP may limit pulmonary inflammation in mechanically ventilated patients without preexisting lung injury [32] and mechanical ventilation with high tidal volume induces inflammation in patients without lung disease [33].

Inflammatory mediators released during volutrauma may play a role in the development of various mechanical. ventilation related complications, such as sepsis and ventilator-associated pneumonia [34].

Pneumonia prevention is of primary importance for people with acute SCI to decrease morbidity and mortality. In a large prospective cohort study of the SCI Model System, pneumonia or postoperative infections suffered during the initial injury through acute rehabilitation was associated with reduced gain in ASIA motor scores recovery when compared to similarly injured people who did not develop pneumonia [6]. In another prospective cohort study of the SCI Model Systems (n = 1203) pneumonia and postoperative wound infection suffered during the initial injury through to acute rehabilitation was associated with reduced functional recovery and increased risk of death seen up to 10 years post-SCI when compared to similarly injured patients who did not develop pneumonia. It is theorized that pneumonia and other infections shortly after injury leads to worsening neurologic injury and functional recovery [6, 7]. It is imperative to develop evidence on optimal ventilator mode and settings to mitigate the incidence of pneumonia and reduce the time on the ventilator.

Similar to Fenton et al., we did not find a significant difference in ventilator weaning days between HVt and MVt groups. In their randomized control study, Fenton et al. compared tidal volumes of 10 ml/kg PBW and 20 ml/kg PBW in 33 participants with subacute SCI [18]. Importantly, they showed mean ventilator days prior to initiating weaning was longer in the high tidal volume group compared to low tidal volume group by nearly 10 days (37.8 days vs. 28.4 days, p = 0.197). We report a similar increase in mean days to initiate weaning in the high tidal volume group compared to the medium tidal volume group. Although not statistically significant, these are clinically meaningful differences. We recommend further exploring total days required from time of admission to AIR to end of weaning (off ventilator for 24 h) as an outcome in future randomized control trials.

The existing literature that guides SCI ventilator management has several limitations and requires further investigation. In the two studies that compared high and low Vt in SCI population, there were important limitations. In the study by Peterson et al. [17], they used current body weight to calculate tidal volume instead of the widely recommended PBW [35, 36]. This will likely cause patients to be misclassified into their respective groups and cause significant errors in analysis. The Fenton et al. study was limited by underpowered sample size. Therefore, results from this study should be interpreted with caution.

Likewise, this retrospective study has several limitations, primarily hampered by a small sample size and lack of generalizability due to the single institution setting, as well as the inherent limitations found in any retrospective study. Nonetheless, the results question both the safety and effectiveness of high tidal volume ventilation for patients with SCI, which concurs with much of the non-SCI literature. Another limitation was that the median age was older in the HVt group, although not found to be significant. Furthermore, the regression models adjusted for age to account for this difference.

In conclusion, our cohort study suggests that HVt  is associated with increased risk of pneumonia and higher odds of adverse pulmonary events in tracheostomized patients with SCI. We believe that further investigation of higher and lower Vt in a large, well-designed adequately powered randomized control trial with long-term follow up data on functional and survival outcomes is warranted to prevent morbidity. In addition, inflammatory biomarkers have the potential to identify patients at high risk for development of mechanical ventilation related complications. In the future clinical trials, measurement of biomarkers can guide identification of optimal ventilator settings for people with acute SCI.