Abstract
BACKGROUND: Liberal oxygenation during mechanical ventilation is harmful in critically ill patients and in certain subsets of patients, including those with stroke, acute myocardial infarction, and cardiac arrest. Surveillance through electronic medical records improves safety of mechanical ventilation in the ICU. To date, this practice has not been used for oxygen titration () in adults. We hypothesize that a surveillance system based on the electronic medical record to alert respiratory therapists to titrate is feasible, safe, and efficacious.
METHODS: In this pilot study, mechanically ventilated subjects were randomized to respiratory therapist-driven titration after an electronic alert versus standard of care (ie, titration based on physician order). An automated surveillance system utilizing a hyperoxemia-detection algorithm generated an electronic alert to a respiratory therapist’s pager. Hyperoxemia was defined as > 0.5 and > 95% for > 30 min. No other aspects of treatment were changed. We assessed feasibility, safety, and preliminary efficacy. Primary outcome was duration of hyperoxemia during mechanical ventilation. An unsafe outcome was identified as hypoxemia ( < 88%) within 1 h after titration per alert. Feasibility was assessed by a survey of respiratory therapists.
RESULTS: Of 226 randomized subjects, 31 were excluded (eg, programming errors of the electronic alerts, no consent, physician discretion). We included 195 subjects, of whom 86 were in the intervention arm. Alert accuracy was 78%, and respiratory therapists responded to 64% of the alerts. During mechanical ventilation, exposure to hyperoxemia significantly decreased in the intervention group (median 13.5 h [interquartile range 6.2–29.4] vs 18.8 h [interquartile range 9.6–37.4]). No episodes of significant hypoxemia were registered. Most respiratory therapists agreed that the alert was helpful in reducing excessive oxygen exposure.
CONCLUSIONS: Use of an electronic surveillance system to titrate was safe and feasible and showed preliminary efficacy in reducing hyperoxemia. Our study serves to justify larger randomized controlled trials for titration.
Introduction
Liberal oxygenation persists in ICUs, emergency departments, and in the pre-hospital environment.1-4 Hyperoxic injury from excessive oxygen exposure is well known; however, recent evidence shows clear association of even moderate hyperoxemia to higher mortality during mechanical ventilation.5,6 Given the growing body of data, clinical practice guidelines for oxygen management in acutely ill patients strongly recommend against supplemental oxygen () for peripheral oxygen saturation () over 96% in acutely ill patients.7 Most health care staff agree that oxygen titration is overlooked and needs further education and research.8
There have been attempts to address oxygen titration. A goal-directed oxygen strategy with daily physician orders within predefined goals led to target only 60% of the time.9 Strategies like using a silicone band on the patient’s arm stating oxygenation goals resulted in moderate adherence and variable oxygenation.10 Both of these studies indicated that one-time reminders are not effective and that a specific and directed strategy is needed.
Electronic assistance in the ICU has widely helped improve mechanical ventilation-related practices as well as adherence to guidelines (eg, low tidal volumes, detection and management of sepsis).11 The Cochrane Effective Practice and Organization of Care (EPOC) Review Group lists “reminders” as a specific professional intervention for implementing a change in practice or in provider behavior.12 Thus, automated electronic-alerts (e-alerts) based on electronic medical records that ascertain real-time oxygenation based on can serve as continuous reminders to address the need for a specific and directed strategy.
Our aim was to assess the feasibility, safety, and preliminary efficacy of utilizing e-alerts to reduce hyperoxemia and to enable greater adherence to the oxygenation protocol during invasive mechanical ventilation.
QUICK LOOK
Current knowledge
Hyperoxemia is prevalent in patients with mechanical ventilation and is associated with worse outcomes. Prioritization of oxygen titration is necessary. Current strategies fall short of translating physician orders to the bedside, and a more effective strategy is needed.
What this paper contributes to our knowledge
A real-time algorithm using electronic medical records was used to generate electronic alerts to respiratory therapists upon detecting hyperoxemia. We found that this approach was feasible, safe, and helpful in reducing excessive oxygen supplementation. This strategy improved adherence to the existing oxygenation protocol. Respiratory therapists supported the strategy and did not perceive it as increased work load.
Methods
Trial Design and Oversight
From January 2012 to March 2014, we conducted a randomized feasibility pilot trial at the medical ICU at the Mayo Clinic, in Rochester, Minnesota. In this trial we compared oxygen titration with the help of e-alert based on electronic medical records sent to a respiratory therapist’s pager versus titration per physician orders, which was the standard of care. This study was approved by the Mayo Clinic Institutional Review Board (Protocol 10–008711). Because the study intervention did not propose new oxygen targets and was directed to improve adherence to the current protocol, a waiver of consent was approved. An explanatory script was placed outside patient rooms, and a HIPAA authorization was obtained. Given the methodology, the study could not be blinded. The study was planned as a pilot to assess feasibility; therefore, at the time of initiation, this study was not subject to FDAAA 801 for an applicable clinical trial. Data were extracted from the electronic medical record and reviewed for accuracy by blinded study personnel.
All mechanically ventilated adult patients (≥ 18 y old) in the medical ICU were eligible for inclusion. We excluded patients who were pregnant and those who denied research authorization. Patients intubated solely for procedural purposes such as esophagogastroduodenoscopy, colonoscopy, bronchoscopy, and minor surgical interventions were also excluded. Patients were screened and randomized electronically within 1 h of endotracheal intubation. All eligible subjects were randomized via the electronic medical record to the intervention or control arm.
Interventions
In the intervention arm, e-alert were sent to respiratory therapists, and was titrated per a decision support tool to prevent an upper limit of > 95% (Table 1). In the control arm, was titrated via one-time physician orders, per the usual standard of care in the ICU. No alerts were issued for subjects in the control arm. The standard of care for oxygenation in the ICU was = 88–95% or = 55–80 mm Hg during mechanical ventilation per the ARDSNet clinical trials network.13 No other changes were made to the ventilator-management protocol or other aspects of patient care. Any subject in the intervention arm could be excluded per treating physician discretion. Subjects readmitted to the ICU within 28 d after transfer during the same hospitalization were re-included and analyzed in the previously assigned arm.
E-Alerts and Decision Support
E-alerts were created by the “sniffer,” a smart automation system developed in-house for Mayo Clinic ICUs in Minnesota, which continually executes hyperoxia algorithms via electronic medical records in real time. The sniffer has been validated and has high sensitivity and moderate positive predictive value.14 The e-alert system uses a near real-time data repository15 in which all necessary vital signs (including and ) are recorded at 15-min intervals. Excess oxygen exposure was defined as > 0.5, > 95% for 30 min consecutively after the initial hour of mechanical ventilation. When the criteria for excess oxygen exposure was met for 30 min, an e-alert was sent to the respiratory therapist’s pager, which indicated a room number and that oxygen titration was needed.
The directions for titration were given on the decision support tool (Table 1). This tool was developed in the form of a portable pocket chart, based on expert opinion and physiologically known oxygenation ranges. It indicated an excessive (hyperoxic) zone, a necessary zone, and a hypoxic zone based on and ranges. It carried instructions on the degree of adjustment based on the oxygenation zone.
Outcomes
Primary outcome was hyperoxemia, defined as durationof over the upper limit of the current protocol (> 95%) associated excessive exposure of (> 0.5) during mechanical ventilation.1 Adherence to the current oxygenation protocol was calculated as the time during mechanical ventilation that was maintained at 88–95%, irrespective of the used. Feasibility was assessed with respiratory therapy survey for workflow efficiency. Safety was assessed as the number of hypoxic episodes in first hour after titration, and this was assessed per alert. Secondary outcomes of ICU length of stay, duration of mechanical ventilation, and in-hospital mortality were also noted.
Statistical Analysis
The statistical analysis was performed as an intention-to-treat analysis. Continuous variables are presented as medians with interquartile ranges. Categorical variables are presented as count or percentages. Statistical comparisons between the intervention and control arms were performed using Wilcoxon test for continuous variables, whereas chi-square tests or Fisher exact tests were used for comparison of categorical variables. For comparison of serial and values on a subject between groups, a linear mixed model with a random effect for subject was used. A P value of < .05 was considered to be statistically significant. We used the JMP 10.0 statistical software package (SAS Institute, Cary, North Carolina) for all analyses.
Results
Subjects
A total of 226 mechanically ventilated patients from the medical ICU were screened for eligibility over 14 months; 23 subjects were excluded during the refinement of the screening algorithm (exclusion due to noninvasive ventilation, false positive alerts, hyperoxia range only during specific procedures including bronchoscopy, and esophagogastroduodenoscopy), and 8 subjects were excluded due to refusal or inability to obtain HIPAA authorization. After randomization, 86 subjects were included in the intervention group, and 109 subjects were included in the control group. One subject was excluded from the interventional arm after initiation of protocol due to physician’s discretion (Fig. 1).
Baseline characteristics including demographics, illness severity, and etiology for ICU admission were similar in both groups (Table 2). The most common reasons for ICU admission was sepsis and pneumonia. No statistically significant differences were found for variables related to ICU care, such as degree of neuromuscular blockade and Richmond Agitation Sedation Score, vasopressors use, minimum hemoglobin level, and tracheostomy in the ICU.
E-Alerts
During the study period, 323 e-alert were sent for 86 subjects in the intervention arm. Overall, alert accuracy (valid alerts) was 78% (251 of 323). Oxygen desaturation ( < 88%) within 1 h of titration was noted after response to 8 alerts (2%). Oxygenation improved after increasing . No adverse events were noted for any subjects. Respiratory therapists responded to 208 alerts (64% of alerts); they did not respond to 115 alerts, of which 60 were valid alerts. The reasons for not responding to alerts were e-alert during a procedure, most commonly bronchoscopy (n = 21); concern for hypoxia (n = 1), subject status changed to comfort care (n = 2), central line being placed (n = 2), and no documentation (n = 34). There were 72 inaccurate alerts, and the most common reasons for inaccurate alerts were representation of noninvasive as invasive mechanical ventilation in the electronic medical record (5), loss of connection between the electronic medical record and data mart (n = 1), control subject (n = 2), and changes in frequency of electronic data outflow resulting in recurrent alerts (n = 8) or alerting at the wrong criteria (n = 56) (Fig. 2).
Feasibility of E-Alerts
Respiratory therapists were surveyed to assess satisfaction and alert fatigue via a 10-question survey utilizing a 5-point Likert scale (see the supplementary materials at http://www.rcjournal.com). Nineteen of 29 therapists responded to the survey (66%): 84% (16 of 19) reported that they found the e-alerts helpful overall; 79% (15 of 19) agreed or strongly agreed that e-alert served as a reminder for oxygen titration, and 68% (13 of 19) did not think that e-alert interfered with their other duties.
Outcomes
Duration of hyperoxemia as specified in the study (ie, time with > 0.5 when > 95%) during mechanical ventilation was significantly reduced by a median 5.3 h in the intervention arm (median 13.5 h [interquartile range (IQR) 6.2–29.4] vs 18.8 h [IQR 9.6–37.4], P = .042) (Table 3). Exposure to excessively high oxygen (ie, > 0.7 when > 95%) was reduced by a median 4.9 h (4.4 h [IQR 0.4–11.0] vs 9.3 h [IQR 2.8–16.9], P = .009) in the intervention group (see the supplementary materials at http://www.rcjournal.com). Total time with > 95% with supplemental oxygen ( > 0.21) in the intervention arm was 64.7 h (IQR 32.2–132.7) versus 71.6 h [IQR 20.2–128.7] (P = .62) in the control arm (Table 3). Adherence to current oxygen protocol (ie, 88–95%, irrespective of ) during mechanical ventilation increased by 9% in the intervention arm (43.3% vs 34.5%, P = .030). The alerts were not intended to reduce to < 0.5. The median in both groups remained similar (intervention arm versus control arm: 0.5 (0.4–0.6) versus 0.5 (0.4–0.65). The secondary outcomes of days on mechanical ventilation and length of ICU and hospital stay were similar in both arms (Table 3).
Discussion
This feasibility trial utilizing e-alert for oxygen titration in critically ill, mechanically ventilated adult subjects led to a reduction of hyperoxemia and enabled greater adherence to the oxygenation protocol. The absence of significant hypoxemic events, good adherence, and respiratory therapist satisfaction indicate the safety and feasibility of this approach.
Utilization of an e-alert as a real-time reminder helped reduce exposure to excess oxygen in a timely manner. This intervention limits the time factor of liberal oxygen supplementation and emphasizes prioritization. Eighteen reviews looked at “reminders” alone as an intervention, from both paper and computer-based clinical decision support systems, and reported that they are 73–78% effective at creating an effective change in provider behavior.12 In our study, the duration of excess oxygen exposure during mechanical ventilation was reduced by about 5 h in the intervention arm, which may have an impact on reducing harm. Girardis et al6 reported that liberal oxygen resulted in higher mortality and worse organ failure in critically ill subjects. Similarly, hyperoxemia in critically ill subsets, such as patients with sepsis or traumatic brain injury or post-cardiac arrest patients, has also shown worse outcomes.16-21 A recent Cochrane meta-analysis suggested that higher may increase mortality.22 Although 2 recent multi center trials (ROX- ICU and LOCO2 trials), indicated no mortality benefit of conservative oxygenation, there was also no harm noted from this strategy.22-24 Additionally, subjects with hypoxic ischemic encephalopathy performed worse with hyperoxemia.23 Given that there is lack of equipoise and an existing knowledge gap, avoidance of hyperoxemia may be a reasonable approach to optimize patient outcomes, resource utilization, and oxygen consumption until further data are available.
To reduce hyperoxemia during mechanical ventilation, a protocol combining both and is necessary. In our study, reduction in exposure to excessively high (ie, > 0.7) was also reduced significantly. Even short bursts of high oxygen exposure promotes lipid peroxidation and propagates oxidative stress, therefore such reduction in exposure is valuable.25 During the study period, there were no protocol violations directly attributed to PEEP management. The similarity in PEEP levels observed between both arms provides reassurance that reduction in is not the result of a major imbalance of a co-intervention such as adjusted airway pressure. Also, no additional requirement for sedation or neuromuscular blockade was noted in the intervention group. We have not collected red blood cell transfusion data; however, the lowest hemoglobin levels were similar in both arms.
No significant adverse effects were observed as a result of ventilator changes due to the e-alert. The e-alert was designed to improve adherence to the current protocol and to reduce exposure greater than the upper limit of the current protocol. There were no lower set targets. The 2 subjects in whom transient hypoxemia was noted had acute respiratory failure in the setting of pulmonary fibrosis and pulmonary hypertension.
Our results indicate that use of e-alert that ascertain real-time oxygenation based on is feasible and safe. Adherence to e-alert was about 64%. We designed our alerts in alignment with the following constructs of the normalization process theory used for implementation of electronic health systems and guidelines.26,27 First, for contextual integration, we used electronic medical records and pagers, which are ubiquitous and used by health care staff for patient care. Second, to match skill set workability, e-alert were directed to respiratory therapists who specifically focus on ventilator management. Next, for relational integration, respiratory therapy leadership was involved in determining the targets and decision support for titration. Lastly, for interactional workability, e-alert had high positive predictive value and specificity.14 Clear and concise information was sent with each alert, and simple, single-step directions were provided on the decision support tool. Additionally, to ensure feasibility and adherence, we included the following measures.28 To allow prioritization of other critical duties by therapists, we did not start the alerting process until 1 h after intubation; after 30 min of exposure to excess oxygen, there was a 15-min time lag to allow therapists to titrate oxygen before the alert was sent; also, there were no more than 3 alerts per patient in a 6-h period to prevent alert burden and fatigue. Given the above factors, e-alert were perceived well by respiratory therapists.
The need of specific informatics infrastructure for the e-alert puts some limits on the generalizability of our approach. The study could not be blinded. Some carry-over effect could be present because the respiratory therapists were involved in the care of subjects assigned to both arms. Conservative oxygen targets may increase ventilator-free days and other organ failure-free days. However, our study was not powered to extrapolate these findings. We did not perform a cost-effectiveness analysis of our intervention. Other limitations include the inability to stop alerts during procedures or when subjects were transitioned to comfort care status.
Randomized trial design with the availability of continuous and values every 15 min and automated time-sensitive electronic randomization are strengths of our study. Utilization of e-alert resulted in reduced exposure and improved adherence to the oxygenation protocol. Reduction in exposure may consequently lead to reduced oxygen toxicity. Limiting oxygenation to conservative levels is known to reduce mortality and organ failure. Our study shows a feasible and safe method to accomplish oxygen titration on the basis of behavioral modification and utilization of electronic medical records. Underscoring the effect of behavioral modification, prioritization of titration may encourage expedited weaning from mechanical ventilation. This has important implications for patient outcomes and resource utilization in the ICU. We recognize from this study that an effective method to achieve precision in oxygen titration is necessary, and future studies should be directed at achieving a tighter range in oxygenation to prevent both hypoxemia and hyperoxemia.
Conclusions
An automated electronic alert tool to assist the titration of oxygen in adult ICU patients during invasive mechanical ventilation was feasible, efficacious, and safe. There was a cumulative reduction in exposure to hyperoxemia in the intervention arm without any safety concerns. Improved adherence to an oxygenation protocol and prevention of hyperoxemia during mechanical ventilation can be achieved through the concept of reminders and prioritization of oxygen titration.
Footnotes
- Correspondence: Sonal R Pannu MD MSc, Davis Heart and Lung Institute, 473, west 12 th avenue, suite 201, Columbus, Ohio 43210. E-mail: sonal.pannu{at}osumc.edu
Supplementary material related to this paper is available at http://www.rcjournal.com.
The authors have disclosed no conflicts of interest.
- Copyright © 2021 by Daedalus Enterprises