Rapid Response Team activation after major hip surgery: patient characteristics and outcomes

Background: Rapid response teams (RRTs) are a critical care resource that review deteriorating patients within the hospital. We aimed to describe demographic, preoperative, surgical, anesthetic, and postoperative characteristics of patients who required RRT activation after major hip surgery. We also sought to assess whether these characteristics where associated with mortality during the index hospital admission. Methods: We reviewed an RRT database of adult patients undergoing orthopedic surgery at a university teaching hospital. We then retrospectively reviewed the medical records to extract a priori–dened patient, preoperative, surgical, anesthetic, and postoperative data of major hip surgery admissions between September 2014 and December 2017. Patients who survived the index hospital stay were compared to those who died. Results: Overall, 187 patients had postoperative RRT activations. Median (interquartile range) age was 84.0 (78-90) years; 125 (67%) were female, and most patients had at least one signicant comorbidity, median Charlson Comorbidity Index (CCI) 5.0 (4.0-7.0). The majority of patients were frail (68%), American Society of Anesthesiologists physical status Class 3 or greater (91%), and underwent nonelective surgery (88%). Median (interquartile range) time from surgery to RRT activation was 29.4 (11.3–75.0) hours, and 25 (13%) patients had unplanned admissions to intensive care or high dependency units. Compared to patients who survived RRT activation, those who died displayed higher mean CCI (6.5 [1.8] vs. 5.5 [2.1], p = 0.02), were more frail (80.1% vs. 56.5%, OR = 3.2, 95% CI: 1.2,8.1; p = 0.03), and received less intraoperative opioids (intravenous morphine equi-analgesia: median = 5.8 (0.1–8.20 vs. 11.7 (3.7–19.0) mg, p = 0.03). They were also more likely to have received an urgent medical review prior to RRT activation (62% vs. 40%, OR = 2.4, 95% CI: 1.1, 5.6); p = 0.05. Conclusions: Death after RRT activation occurred in 1 in 7 patients undergoing major hip surgery. Common patient characteristics included advanced age (> 82 years), frailty, high CCI, and emergency surgery. Further studies investigating perioperative surveillance teams in the identication of the high-risk patients before surgery and deteriorating patients after major hip surgery are warranted.


Background
A rapid response team (RRT) is an interdisciplinary team of critical care health professionals who manage deteriorating patients within the hospital (1). RRTs, also referred to as "medical emergency teams" or "emergency response teams," are commonplace in many modern hospitals. The composition of an RRT can include critical care physicians, anesthetists, critical care nurses, and respiratory therapists (2). Major surgery poses a signi cant physiological challenge to patients, which in turn can predispose them to an increased risk of postoperative deterioration (3,4). Past case-control and cross-sectional studies have implicated an array of preoperative and anesthetic factors that may be associated with postoperative patient deterioration and the need for RRT activation (5-7). However, these studies provide insu cient information on patient characteristics, and preoperative and postoperative anesthesia related variables, including perioperative hemodynamic data, use of uid, vasoactive drugs, and opioid medications. Furthermore, no studies to date have speci cally explored the perioperative factors and patient characteristics of major hip surgery in relation to RRT activation. Therefore, we sought to describe the perioperative course of patients who underwent major hip surgery and required a postoperative RRT review. We describe patient characteristics and the detailed preoperative, surgical, anesthetic, and postoperative factors of patients who required RRT activation after major hip surgery in an Australian university hospital. Speci cally, we evaluated the effects of uids, vasoactive medications, and opioids on the development of RRT activation and assessed whether these perioperative characteristics affected mortality during the index hospital admission. In addition, we assessed the incidence and severity of perioperative hypotension and whether this was associated with in-hospital mortality. This analysis may facilitate the identi cation of patients at risk of postoperative deterioration, guide intraoperative patient management, and allow for a focused allocation of critical care and hospital resources, all of which may provide opportunities for proactive prevention strategies.

Methods
The study was conducted at Austin Health, a tertiary teaching hospital a liated with the University of Melbourne in Melbourne, Victoria, Australia. Austin Health performs approximately 38,000 surgical procedures annually, including complex cardiothoracic surgery, hepatobiliary-pancreatic surgery, liver transplantation, and major spinal and orthopedic surgery. The orthopedic surgical unit provides services to over 15,000 outpatients annually and performs over 2,500 operations per year, of which approximately 650 are major hip operations.
Following approval from the Austin Health Human Research Ethics Committee (LNR/17/Austin/616), we performed a retrospective cohort study of patients who required RRT activation following major hip surgery between September 2014 and November 2017. The need for informed written consent from participants was waived due to the observational and retrospective nature of the study.
The RRT at our institution is an intensive care-led service introduced in 2000. The RRT is governed by the Department of Intensive Care Medicine, and the RRT team comprises an intensive care registrar and critical care nurse. The RRT is also accompanied by the patient's admitting unit at every activation. Escalation of medical resources to assist with RRT activation are immediately available if required (e.g., anesthesia support for airway management). At our institution, the RRT is activated whenever a patient meets predetermined criteria, which include acute changes in any of the following: obstructed airway, noisy breathing or stridor, problems with a tracheostomy tube, any di culty in breathing, respiratory rate < 8 or > 25 breaths/min, oxygen saturation < 90% despite oxygen administration, heart rate (< 40 or > 120 beats/min), systolic blood pressure < 90 mmHg, urine output < 50 mL over 4 hours, sudden change in conscious state, patient cannot be roused, or if any member of staff is worried about imminent deterioration of the patient. Additionally, an RRT is activated for any "code blue." A "code blue" is activated whenever a patient suffers a cardiac or respiratory arrest. Our institution's RRT reviews approximately 3,000 patients annually, of which the majority are post-surgery.
For inclusion in our study, patients had to be adults (age > 18 years) undergoing major hip surgery who had an RRT or "code blue" activation post-surgery and were within the index hospital admission. In the case of multiple RRT activations, we only analyzed the rst event. We used the following procedures as listed in the International Statistical Classi cation of Diseases (10th revision) to select patients: total hip arthroplasty, partial hip replacement/hemiarthroplasty (unipolar or bipolar femoral head), revision of hip replacement not otherwise speci ed, arthrotomy for removal of prosthesis, revision of hip replacement (both acetabular and femoral components), revision of hip replacements (acetabular liner), resurfacing hip (total acetabulum and femoral head), resurfacing hip (partial femoral head or acetabulum), and insertion or removal of any internal xation device. We excluded super cial procedures of the hip joint including joint arthrocentesis and wound debridement.
As part of routine perioperative care for major hip surgery, patients were assessed by a multidisciplinary team consisting of a surgeon, anesthetist, perioperative physician, and ortho-geriatrician (if over 70 years of age). Routine preoperative investigations included biochemical, hematological, and coagulation tests, and where necessary, all patients were optimized from a cardiorespiratory perspective prior to surgery. All patients underwent preoperative hemoglobin optimization, based on the National Blood Authority of Australia's patient blood management initiative (8). When appropriate, standard perioperative care included strict transfusion practice in accordance with these guidelines. Further, as part of the Diabetes Discovery Initiative, all patients with a HbA1c of 8.3% (67 mmol/mol) and above were seen by the endocrinology unit, which generated a personalized plan for glycemic control according to our institution's guidelines. Patients with a HbA1c between 7.5% (58 mmol/mol) and 8.2% (66 mmol/mol), and those with newly diagnosed diabetes, were seen by a general physician. All patients were managed according to the hospital's perioperative guidelines for patients with diabetes, with an inpatient blood glucose target of 5-10 mmol/L based on the Australian Diabetes Society guidelines (9). In addition, for patients with decision-making capacity, an advance care plan was undertaken, which allowed patients to communicate their future preferences relating to medical treatment to their families, friends, and health professionals. In accordance with existing legislation, a legally de ned "person responsible" was appointed to make medical decisions on behalf of a patient who lacks the capacity to give their own consent to treatment.
Data were extracted from the patient's electronic medical records and from Austin Hospital's computerized laboratory results by two independent study investigators. Austin Health uses Cerner electronic medical records, which allows comprehensive electronic data capture and access to patient health information from the perioperative setting. We collected a priori-de ned data on patient characteristics, comorbidities, and preoperative management. All other comorbidities were extracted from patient medical records. Patient comorbidity was further de ned using the Charlson Comorbidity Index (CCI), a validated metric that predicts 1-year patient mortality (10). For the calculation of the CCI, moderate/severe chronic kidney disease was de ned as an estimated glomerular ltration rate of less than 60 mL/min (Stage 3 or worse), and chronic liver disease was de ned based on the Child-Pugh classi cation (11). Congestive cardiac failure was de ned as "heart failure with preserved ejection fraction" (i.e. diagnosed by combination of clinical acumen combined with either echocardiographic features of diastolic dysfunction, or elevated plasma B-type natriuretic peptide concentrations) and "heart failure with reduced ejection fraction" (i.e. left ventricle ejection fraction <40%), regardless of etiology. We used a modi ed Canadian Study of Health and Aging Clinical Frailty Scale to determine frailty (12).
Intraoperatively, we recorded the type of procedure, anesthesia (regional and/or general), as well as the use of uids, and vasoactive and opioid medications. Furthermore, the number of epochs of intraoperative hypotension, and the magnitude of each hypotensive event, were recorded. Similar data were collected from the post-anesthesia care unit (PACU). A hypotensive event was de ned as any reduction in systolic, diastolic, or mean arterial pressure by 30% or more as compared to preoperative values; severe hypotension was de ned as a reduction in any of the above-mentioned blood pressures by 50% or more. The duration of hypotensive episodes was not assessed, and each hypotensive measure was counted as a discrete epoch. Postoperatively, we collected postsurgical discharge destination, indication for RRT activation, time to RRT activation from surgery, as well as length of stay and inhospital mortality.
Due to the exploratory and observational design of this study, our primary objectives were to describe the demographic and perioperative pro le of patients who required RRT activation after major hip surgery. We also compared the perioperative characteristics of patients who survived the index hospital admission to those who did not. Speci cally, we further explored differences between these two groups with respect to the following a priori variables: i) preoperative comorbidities (including frailty), ii) type of anesthesia (general and/or regional), iii) surgical presentation (selective or emergency), iv) perioperative hypotension, and v) use of opioids and vasoactive drugs.

Statistical analysis
Continuous variables were tested for normality and normally distributed data were expressed as means and standard deviations (SD) and compared using a Student's t test; non-normally distributed data were expressed as medians and interquartile ranges (IQRs) and compared using the Mann-Whitney U test. Categorical variables were described as proportions and compared using the chi-square test or the Fisher's exact test. All p values of less than 0.05 were treated as indicative of statistical signi cance, and no correction for multiplicity of testing was undertaken due to the exploratory nature of the study. We reported this study using the STROBE guidelines for reporting observational studies (13). Analyses were performed using GraphPad Prism (version 7.00 for Mac, GraphPad Software, La Jolla, California, United States).

Results
Over the 3-year period, a total of 8,094 patients underwent elective and nonelective orthopedic surgery at our institution. A total of 1,825 patients underwent surgery on the hip. We excluded 39 patients who underwent minor/super cial hip procedures. Consequently, a total of 1,786 patients underwent major hip surgery, of which 187 (9%) had a postoperative RRT activation. Of these patients, seven (0.4%) ful lled criteria for a "code blue" activation. Of the seven "code blue" activations, six patients required cardiopulmonary resuscitation, which was unsuccessful in three patients (i.e., death). Three patients had return of spontaneous circulation but remained in the ward and were palliated. The one patient who did not receive CPR was transferred to a critical care setting for further management. The patient was discharged to a residential home on postoperative Day 11. A consort ow diagram is presented in Figure  1.
The operative course of patients who underwent an RRT activation is summarized in Table 2. Most patients underwent hip arthroplasty, with 40% of these being total hip arthroplasty. One quarter had surgery out of hours (between 18h00 and 08h00, or over a weekend). Almost 1 in 3 patients had surgery performed under regional anesthesia, one third under general anesthesia, and the rest had combined general and regional anesthesia. The median duration of surgery was 128 (99-163) minutes. Median intraoperative intravenous morphine equi-analgesia dose was 10.0 (2-16.7) mg. A detailed breakdown of opioid use is presented in Table 2. The mean lowest recorded temperature in theater was 36.2 (0.4) ºC.
During anesthesia, two thirds of patients had received an arterial line as part of intraoperative advanced hemodynamic monitoring. Almost all patients had received intraoperative vasopressor support, and 58% had a documented intraoperative hypotensive event. Of those patients who were hypotensive, the median number of hypotensive episodes was 3.5 (1-9). A detailed overview is presented in Table 3.
Postoperatively, in the PACU, 10% of patients had received vasopressor support, and 45% of patients had a documented hypotensive episode; the median number of hypotensive epochs was 2 (1-4). An overview of hypotension and vasopressor use in PACU is presented in Table 3. Almost all patients were transferred directly to a standard surgical ward; seven (4%) patients, directly to critical care services (intensive care unit [ICU] or high dependency unit [HDU]). The median time to RRT activation from discharge from theater was 29.4 (11.3-75) hours. The majority of RRT activations occurred within the rst 48 hours of surgery. The most common reason for RRT activation was hypotension (35%), followed by tachycardia (25%), and high respiratory rate (11%). Most of the RRT activations occurred after hours, and 25 (13%) patients had unplanned admissions to critical care services (ICU or HDU) after RRT activation. Although patients who received regional anesthesia has a lower median pain score (11-point Numerical Rating Scale [NRS]) within the rst 24 postoperative hours compared to patients receiving general anesthesia only (NRS score 2 [1][2][3] vs 4 [3][4][5], p=0.0001), severe pain was not a cause RRT activation in any patient. Of those who survived, the median length of hospital stay was 9 (6-14) days. Overall, 26 (14%) patients did not survive their acute hospital admission and died a median of 2.9 (0.2-8.9) days after RRT review. , compared to those who survived. There were no signi cant differences observed between those who survived and those who died in regard to the type of surgery or anesthesia (regional vs. general), number of perioperative hypotension episodes, or use of vasoactive medications, inotropes, or uid therapy. Similarly, there were no observed statistical differences in the time from surgery to RRT activation or in unplanned admissions to critical care services (ICU or HDU).

Key ndings
We performed a retrospective observational study describing the perioperative characteristics of patients who required RRT activation after major hip surgery. We found that in-hospital mortality after RRT activation occurred in 1 in 7 patients. Moreover, we found that common patient characteristics associated with such activation included advanced age (> 82 years), frailty, high CCI score, and emergency surgery presentation. Finally, we found that overall mortality was close to 1 in 7 patients.

Relationship to other studies
To date, there have been no studies investigating RRT activation after major hip surgery. There is also limited research focusing on the perioperative determinants of postoperative RRT activation after major surgery. Three studies have identi ed perioperative characteristics affecting patient deterioration in the postoperative setting (5-7). Lee et al. conducted a retrospective case-control study investigating early postoperative emergencies requiring an intensive care team intervention (7), with 34 RRT activations identi ed for 32 patients. In the study, RRT participants were matched with a nested cohort of 126 controls. Similar to our ndings, there were signi cant preoperative associations with early RRT activation, such as high ASA status. The authors did not report on frailty or detailed patient comorbidity. Likewise, the associations with perioperative hypotension, and detailed anesthesia and surgical variables, were not assessed. More recently, in a tertiary children's hospital, Barry et al. performed a retrospective review of 100 RRT calls occurring within 24 hours of receiving anesthesia or procedural sedation (5). These patients' medical records were reviewed to obtain patient characteristics, etiology of the RRT call, and outcomes. Only nine patients (9%) had undergone orthopedic surgery, the type of which was not speci ed. The authors reported that high ASA status, general anesthesia administration, and the presence of acute or chronic conditions prior to anesthetic administration predisposed a patient to perioperative complications resulting in the need for an RRT review. Generalization to patients undergoing major hip surgery is limited by the lack of detailed anesthesia and surgical variables reported. Further, the pediatric context and low prevalence of orthopedic patients in their study limits its generalization to our adult population.

Study implications
Our ndings show that patients who required RRT activation after major hip surgery had an in-hospital mortality of 14%. These patients were likely to be older, frail, have multiple comorbidities, and undergoing nonelective surgery. Intraoperative hypotension and the use of vasoactive medications was ubiquitous, and surgery was frequently performed after hours. Speci cally, mortality after RRT activation following major hip surgery in our institution occurred in a signi cantly high-risk patient cohort, with a patient pro le of advanced age (> 82 years), frailty, and high CCI being pervasive. Further, despite these important risks, such patients had no structured critical care support. The identi cation of such patients may allow for effective preoperative risk strati cation, optimization of medical comorbidity, proactive planning regarding advanced care directives, and increased postoperative monitoring. Interestingly, we found that surgical factors (including type or duration of surgery) and anesthesia factors (including type of anesthesia, intraoperative hemodynamics, opioid use, and vasopressor use) did not differentiate patients who survived or died after RRT activation. These ndings may be of particular interest to perioperative clinicians and health organizations, as they highlight-in patients for whom escalation of care is appropriate-opportunities for patient risk strati cation and appropriate allocation of critical care resources (HDU or ICU) in the postoperative setting.
There has been a strong association reported between postoperative RRT activation and after-hours surgery (7); however, in our study, the majority of surgeries (77%) were undertaken during normal working hours. Further, almost all patients who required RRT activation were discharged postoperatively to a general surgical ward, and over one third of our patient cohort required an urgent medical review prior to RRT activation. These ndings may be of particular interest to perioperative clinicians and health organizations, as they highlight-in patients for whom escalation of care is appropriate-opportunities for patient risk strati cation and appropriate allocation of critical care resources (HDU or ICU) in the postoperative setting. Our ndings further highlight the need for enhanced postoperative ward surveillance and more effective early warning systems detecting postoperative patient deterioration.
We reported on detailed perioperative hemodynamic variables: in particular, the use of invasive hemodynamic monitoring, rate of perioperative hypotension, and the use of uid and vasoactive medications. Hypotension has been reported to be the most common indication for RRT activation in adult postsurgical patients (6, 7). There is mounting interest in intraoperative hypotension and its strong association with postoperative morbidity and mortality (14, 15). Over half of our patients experienced an episode of intraoperative hypotension, with 9% of patients having a severe hypotensive event. The almost ubiquitous use of invasive blood pressure monitoring possibly allowed for timely identi cation and treatment of perioperative hypotension, re ected in the frequent use of vasopressor medication, which was administered in nearly all patients (88%). The use of a regional technique in 69% of patients may have also affected the frequency of vasoactive medication use. Postoperatively, in the PACU, almost half of the patients had a documented hypotensive event, and only 10% of patients required vasopressor support. Intraoperative vasopressor use has been associated with postoperative RRT requirement (6). We also identi ed that, compared to those who survived, patients who died received a lesser intravenous morphine equi-analgesia dose. This likely re ects the more advanced age and signi cantly higher comorbidity pro le of these patients. However, analgesia use outside of the operating theater and PACU was not collected.

Strengths and limitations
There are several strengths and limitations of this study. This is a single-center study of patients undergoing major hip surgery, performed in a high-volume center for orthopedic surgery, which limits the external validity of our ndings to other institutions and to other types of surgery. Importantly, given the exploratory design of this study, we only collected data on patients who had an RRT activation after major hip surgery. We did not compare these patients to those who underwent similar surgeries but did not have an RRT activation, which limits the broad application of our ndings to all patients undergoing hip surgery. We did not collect data on comorbidity severity outside that of kidney disease and liver disease, which limits the detail of comorbidity description. Data relating to adverse post-operative complications such as pulmonary embolism or acute coronary syndromes were not collected as these were considered outside the scope of this report. The large number of distinct variables collected gives a detailed overview but severely limits the extrapolation of any causal relationships between perioperative course, adverse postoperative events, RRT activation, and post-RRT activation mortality.
However, our study also has several strengths. To date, this is the largest review of patients undergoing RRT activation after major hip surgery that speci cally combines detailed patient, surgical, and anesthesia factors describing the perioperative course of patients. The collection of data on detailed preoperative comorbid conditions, including frailty, provides a comprehensive evaluation of the baseline health characteristics of our patient population. The detailed overview and rates of hypotension, vasopressor use, and use of opioids provide an in-depth insight into this patient cohort and their perioperative journey. By reporting the rate of RRT activation and mortality in these patients, we have de ned both a need for enhanced postoperative ward surveillance and more effective early warning systems detecting postoperative patient deterioration. Our ndings are hypothesis-generating and may provide valuable data for power calculations in future studies on RRT activation in the major orthopedic setting. Given the exploratory nature of the study, we cannot establish a causal relationship between the perioperative variables we assessed and their impact on RRT activations or any other postoperative outcomes. However, our in-depth insights, comparing patients who survived hospital admission to those who died, may provide data for sample size calculations for future RRT prospective studies.

Conclusion
Death following RRT activation occurred in 1 in 7 patients who had undergone major hip surgery in a tertiary referral hospital. We have identi ed several important ndings relevant to RRT activation after major hip surgery. Most patients were older (> 82 years), frail, with a high CCI, and undergoing out-of-hours emergency surgery. Our ndings suggest that surgical (type or duration) and anesthesia (type, intraoperative hemodynamics, opioid use, vasopressor use) factors do not differentiate patients who died or survived following RRT activation. Given the high rate of RRT activations and high mortality rate after these activations, the ndings, speci c to major orthopedic surgery, provide good opportunities for the implementation of strategies aimed at improving postoperative outcomes in these at-risk patients.

Consent for publication
Not applicable.

Availability of data and material
The datasets generated and analysed during the study are not publicly available due to individual privacy concerns but are available from the corresponding author on reasonable request.   Note. Data are presented as number (proportion), mean (standard deviation), or median (interquartile range). a n = 127, 108, and 19 for total, survived, and did not survive, respectively.
b Missing values, n = 30. Note. Data are presented as number (proportion), mean (standard deviation), or median (interquartile range).  Study ow diagram