The establishment of Rapid Response Teams (RRTs) has proven to be critical for improving patient safety in hospital settings. These teams consist of trained professionals, including physicians, nurses, and respiratory therapists, who can quickly intervene during clinical deterioration, thereby potentially reducing severe adverse events.^(1-3)

Since their introduction in the 1990s, RRTs have been widely adopted in hospitals worldwide, with numerous studies reporting their positive impact across diverse hospital settings, including general, oncology, and specialty hospitals.^(1-3) These teams aim to identify and respond to early signs of clinical instability, thereby helping to prevent further deterioration and reduce mortality.

Rapid Response Team activation in general hospitals is often triggered by medical emergencies such as cardiopulmonary arrest, septic shock, and respiratory distress.⁽⁴⁾ In outpatient and specialized settings, these activations are less frequent but no less critical. The relevance of RRTs is particularly high in predominantly oncology-focused hospitals, where activation is frequently associated with adverse reactions to oncological treatments or procedural complications.⁽⁵⁾

Despite their established efficacy in inpatient settings, there is limited data on RRT activation in outpatient settings, particularly in oncology hospitals. Outpatient oncology care presents unique challenges owing to the complexity of available therapies, such as chemotherapy and immunotherapy, which are associated with a significant risk of acute adverse reactions. This knowledge gap underscores the importance of characterizing RRT activation to inform strategies that enhance patient safety.

This study aimed to describe Rapid Response Team activation in adult outpatients at an oncology hospital over one year. This included the analysis of activation patterns by location, timing, and underlying causes; assessment of documentation completeness by healthcare teams; and evaluation of clinical outcomes, such as in-hospital mortality or the need for intensive care unit transfers. By investigating these factors, this study sought to provide actionable insights to optimize Rapid Response Team protocols and resource allocation, with the ultimate goal of improving the quality of care in outpatient oncology settings.

This retrospective observational study was conducted at a community oncology hospital in São Paulo, Brazil, between January 1 and December 31, 2023. All RRT activations for adult patients treated in the ambulatory oncology care setting were included. The study was approved by the local Research Ethics Committee (Protocol 5804/2023), and the requirement for informed consent was waived owing to the retrospective nature of the study. The study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

The hospital provides highly complex cancer care and has dedicated outpatient clinics, imaging and laboratory units, infusion rooms, and rehabilitation facilities. It operates with a fully in-house medical team responsible for approximately 200 operational beds, 40 intensive care unit (ICU) beds, and outpatient services. The RRT includes hospitalist physicians trained in emergency response and is activated through an electronic system whenever acute clinical deterioration is suspected in a non-hospitalized patient. Activations, referred to as "code yellow" events, may be initiated by nurses or attending physicians according to predefined criteria or if there is a subjective concern about the patient's condition. In the outpatient chemotherapy sector, an oncologist is available during chemotherapy infusions and is capable of addressing the most frequent medical complications. Activation of the RRT in this sector was at the discretion of the oncologist if they deem that additional assistance is necessary.

Eligible activations for this study included those occurring in the outpatient, imaging, chemotherapy, rehabilitation, and laboratory collection areas. Activations involving admitted inpatients or those lacking essential information were excluded. Data were obtained from structured electronic RRT forms integrated into the hospital electronic medical record (Cerner®). Each activation record contained demographic data, activation site and time, clinical triggers defined by the nursing staff, time of physician arrival and care completion times, and interventions or subsequent transfers to ICUs. Events requiring the activation of additional hospital emergency codes (stroke, cardiac arrest, sepsis, or acute myocardial infarction) were also recorded. Supplementary Material 1 lists the criteria for RRT activation. The RRT consisted of a single physician (in all cases except for cardiac arrest) and one nurse and one respiratory therapist (in cases of cardiac arrest-"code blue"), along with a second physician.

The sample comprised the totality of RRT activations among non-admitted oncology outpatients during the study period (census sampling). No sample size calculation was performed. The evaluation criteria included form submission completeness by nursing staff (n=12 fields), form submission completeness by physicians (n=24 fields), days of the week and times of the day when activations occurred, and the duration of responses in minutes. We assessed ICU transfers and mortality during hospitalization. The outcomes were analyzed to assess the clinical impact of RRT activation in this patient population. The inclusion criteria were adult patients whose forms were available for review in the electronic medical records. The exclusion criteria were forms not filled electronically, pregnancy, and patients not receiving oncologic care.

Each activation was categorized according to the period of the day: morning (07:00-12:59), afternoon (13:00-18:59), or night (19:00-06:59). When more than one activation trigger was recorded, only the first listed trigger was used for classification. Time to physician arrival was calculated as the interval between activation and physician arrival, whereas the total duration of care corresponded to the period between arrival and the end of medical attendance. For patients who died during hospitalization, the time from activation to death was expressed in days.

Continuous variables were expressed as medians and interquartile ranges (IQR), and categorical variables were expressed as absolute and relative frequencies. The Shapiro-Wilk test was used to assess normality.

To explore the factors associated with the composite outcome of ICU admission or in-hospital death, we first performed univariate analyses comparing the demographic, clinical, temporal, and operational characteristics of patients who experienced the composite endpoint and those who did not. Categorical variables were compared using Fisher's exact test or the chi-squared test. Continuous variables were analyzed using the Wilcoxon rank-sum test, as they did not meet the normality assumptions on visual inspection.

Based on clinical plausibility and univariate results, three variables were selected a priori for multivariable modeling: sex, age, and activation trigger category. These variables were entered into a multivariate logistic regression model to estimate the adjusted odds ratios (ORs) and corresponding 95% confidence intervals (95%CIs) for composite outcomes. All analyses were performed using R version 4.5.1, and statistical significance was defined as a two-tailed p<0.05.

In 2023, there were 55,133 outpatient consultations, 23,993 imaging examinations, and 19,147 intravenous chemotherapy sessions. A total of 441 RRT activations met the eligibility criteria and were included in the analysis. Regarding the completeness of the electronic forms, the nursing staff completed an average of 9.9 (out of 12) fields, and the medical staff completed 19.8 (out of 24) fields physicians. Table 1 shows the baseline characteristics and clinical triggers for RRT activation during the study period. Most events occurred in female patients (53.3%), with a median age of 64 years (interquartile range [IQR] 53-72). Activation was predominantly triggered in the outpatient clinic (55.1%) or the imaging sector (37.4%). Regarding temporal patterns, events occurred most frequently during the daytime (92.4%) and weekdays (93.6%). Throughout the year, the period between May and August had the highest number of monthly activations (Figure 1). Table 2 presents the activation triggers, response characteristics, and short-term outcomes. The most common activation triggers were elevated nursing staff concern (19.7%), symptomatic hypertension with systolic pressure >180mmHg (17.7%), acute drop in O₂ saturation to <90% (8.2%), acute changes in heart rate to >130 or <40bpm (13.2%), and acute drop in systolic blood pressure to <90mmHg (9.1%) (Figure 2).

The median time from activation to physician arrival was 3 minutes (IQR= 2-4). In 406 (92.1%) activations, the time to physician arrival was less than the hospital goal of 5 minutes. The median duration of care was 48 min (IQR=25-93). Intervention by the anesthesia team was required in 0.7% (n=3) of patients, while the stroke code (a specific algorithm for suspected stroke) was activated in 1.13% (n=5) of the cases. In 11.5% (n=51) of the cases, the sepsis bundle was collected. Ten patients (2.3%) died during the study period, with a median interval of 17 days (IQR= 6-32.5) between activation and death. Finally, transfer to the ICU occurred in 39 activations (8.8%).

Table 3 shows the univariate analyses for demographic, temporal, and clinical characteristics (including sex, age, hospital sector, period of day, weekday versus weekend, activation trigger category, and time to physician arrival) associated with the composite outcomes of death or ICU admission. No characteristics showed statistically significant associations with the composite outcomes (all p>0.05).

Table 4 shows the adjusted associations between patient characteristics and composite outcomes. In the multivariate logistic regression model adjusted for age, sex, and activation trigger, no variable demonstrated an independent association with the composite outcome.

The results of this study show that RRTs play a key role in managing clinical deterioration among oncology outpatients. The high number of activations, resulting in 39 transfers to the ICU and 3 cases that progressed to cardiac arrest, highlights the importance of having a dedicated team capable of quickly intervening in these situations. Analyzing the reasons for nursing and physician activation provides valuable insights into the types of clinical deterioration that occur in this patient population. An important operational feature of the RRT system at our institution is that any healthcare professional, regardless of their role, can initiate activation in outpatient areas outside the chemotherapy suite. This open activation policy may contribute to the early identification of clinical deterioration, reinforce a culture of patient safety, and reduce delays in emergency responses.

The primary reasons for activations, including "serious concern for the patient's status" and "hypertension," reveal that many triggers are related to acute clinical issues rather than direct oncological complications. As these are outpatients, there are no early signs of clinical deterioration (such as the NEWS score used for hospitalized patients). In this context, it is relevant to compare the absence of independent predictors and observed outcomes in our cohort with those in previous inpatient and outpatient RRT studies. In contrast to several inpatient RRT studies in oncological and general hospital populations, we did not identify the independent predictors of ICU admission or in-hospital mortality among ambulatory patients. Notably, our adjusted model included only age, sex, and trigger category, without time-updated early warning scores or other granular severity variables. Prognostic discrimination in RRT cohorts improves when contemporaneous physiological severity markers, laboratory abnormalities, or escalation intensities are available, particularly in hospitalized oncology patients.^(1,6-7) The absence of such granular data in outpatient activation forms, combined with the relatively small number of adverse events, likely limited our ability to detect independent associations. Similar findings have been reported in other ambulatory oncology RRT series, where triggers are often nonspecific and outcomes are influenced by factors beyond the acute event itself.⁽⁵⁾ The distribution of activation triggers in our cohort also differed from that described in hospitalized oncology populations, where RRT activation is commonly driven by advanced physiological deterioration such as respiratory failure, sepsis, or shock.^(1,2,4) In the outpatient setting, nonspecific triggers such as nursing staff concerns and isolated hypertension were more prevalent. This pattern likely reflects a proactive activation culture rather than reduced clinical acuity, as previously suggested by studies demonstrating wide inter-institutional variability in the interpretation and prognostic meaning of "staff concern" triggers.

Concerns about patient status have been previously documented⁽⁸⁾ but primarily in inpatient cohorts. In outpatients, adverse reactions to cancer treatments are often cited as the predominant triggers, which were not observed in our cohort. A notable difference may be attributed to the presence of oncologists in chemotherapy suites to manage adverse events without requiring RRT activation, as they likely address many early stage complications and thus prevent patient deterioration. Other reasons for activation (syncope, falls, and contrast reactions) were frequent in other cohorts,^(1,5) but were not particularly frequent in our series. Furthermore, our study was conducted inside a hospital, which is different from what has been reported for ambulatory cancer patients⁽⁵⁾ in facilities outside of hospitals.

The rates of ICU transfer and in-hospital mortality observed in this study indicate that ambulatory oncology patients requiring RRT activation are clinically distinct from inpatient oncology cohorts, which typically exhibit higher baseline severity and worse short-term outcomes.^(1,2) Direct comparisons should be interpreted cautiously because of differences in the case mix and outcome definitions. Comparisons with published ambulatory oncology RRT experiences suggest broadly comparable downstream escalation and mortality rates,⁽⁵⁾ although longer-term outcomes in oncology are strongly influenced by disease stage, treatment intent, and goal-of-care decisions rather than by the acute trigger alone.

Temporal analysis showed peaks in activation during normal business hours and on Mondays, suggesting a need for optimized staffing during these periods. In addition, the low incidence of overnight activation reflects operational dynamics in outpatient settings.

To the best of our knowledge, the figures we report 44.08 activations per 10,000 outpatient visits, 68.77 activations per 10,000 imaging procedures, and 13.58 activations per 10,000 chemotherapy sessions – have not been previously reported.

The Rapid Response Teams plays a vital role in managing clinical deterioration among non-admitted oncology patients in an outpatient setting. The high number of activations, with more than 30 patients requiring transfer to the intensive care unit and a few progressing to cardiac arrest, emphasizes the importance of having a dedicated team capable of quickly intervening in these situations. Form completeness was an issue, which led us to reconsider the mandatory nature of each item. Further multicenter studies may help validate these findings in different settings.