In both phases, we did not collect data on patients who did not meet eligibility criteria. ICU indicates intensive care unit; SAQ, Safety Attitudes Questionnaire.
aWe used multiple imputation to analyze primary outcome data.
aAll effect estimates were adjusted for baseline values of outcome variables, except in-hospital mortality odds ratio, which was adjusted for baseline values of intensive care units’ (ICUs’) standardized mortality rate (calculated with Simplified Acute Physiology Score 3 [SAPS3]) and patients’ SAPS3.
bThe intracluster correlation coefficient calculated from a random-effects model for the primary outcome was 0.13, and the coefficient of variation K was 0.25.
RASS indicates Richmond Agitation-Sedation Scale.
aAll effect estimates were adjusted for baseline values of outcome variables.
Statistical analysis plan
eTable 1. Strategies for implementing the interventions of the study
eTable 2. Definitions of care processes and clinical outcomes
eTable 3. Characteristics of intensive care units
eTable 4. Outcomes reflecting clinical events, processes of care and culture in the baseline observational and randomized phase and between group differences in pre-post differences
eTable 5. Effect of the quality improvement intervention on in-hospital mortality in the prespecified subgroups
eFigure 1. Daily round checklist
eFigure 2. Daily goals and clinician prompting form
. Effect of a Quality Improvement Intervention With Daily Round Checklists, Goal Setting, and Clinician Prompting on Mortality of Critically Ill PatientsA Randomized Clinical Trial. JAMA. 2016;315(14):1480–1490. doi:10.1001/jama.2016.3463
The effectiveness of checklists, daily goal assessments, and clinician prompts as quality improvement interventions in intensive care units (ICUs) is uncertain.
To determine whether a multifaceted quality improvement intervention reduces the mortality of critically ill adults.
Design, Setting, and Participants
This study had 2 phases. Phase 1 was an observational study to assess baseline data on work climate, care processes, and clinical outcomes, conducted between August 2013 and March 2014 in 118 Brazilian ICUs. Phase 2 was a cluster randomized trial conducted between April and November 2014 with the same ICUs. The first 60 admissions of longer than 48 hours per ICU were enrolled in each phase.
Intensive care units were randomized to a quality improvement intervention, including a daily checklist and goal setting during multidisciplinary rounds with follow-up clinician prompting for 11 care processes, or to routine care.
Main Outcomes and Measures
In-hospital mortality truncated at 60 days (primary outcome) was analyzed using a random-effects logistic regression model, adjusted for patients’ severity and the ICU’s baseline standardized mortality ratio. Exploratory secondary outcomes included adherence to care processes, safety climate, and clinical events.
A total of 6877 patients (mean age, 59.7 years; 3218 [46.8%] women) were enrolled in the baseline (observational) phase and 6761 (mean age, 59.6 years; 3098 [45.8%] women) in the randomized phase, with 3327 patients enrolled in ICUs (n = 59) assigned to the intervention group and 3434 patients in ICUs (n = 59) assigned to routine care. There was no significant difference in in-hospital mortality between the intervention group and the usual care group, with 1096 deaths (32.9%) and 1196 deaths (34.8%), respectively (odds ratio, 1.02; 95% CI, 0.82-1.26; P = .88). Among 20 prespecified secondary outcomes not adjusted for multiple comparisons, 6 were significantly improved in the intervention group (use of low tidal volumes, avoidance of heavy sedation, use of central venous catheters, use of urinary catheters, perception of team work, and perception of patient safety climate), whereas there were no significant differences between the intervention group and the control group for 14 outcomes (ICU mortality, central line–associated bloodstream infection, ventilator-associated pneumonia, urinary tract infection, mean ventilator-free days, mean ICU length of stay, mean hospital length of stay, bed elevation to ≥30°, venous thromboembolism prophylaxis, diet administration, job satisfaction, stress reduction, perception of management, and perception of working conditions).
Conclusions and Relevance
Among critically ill patients treated in ICUs in Brazil, implementation of a multifaceted quality improvement intervention with daily checklists, goal setting, and clinician prompting did not reduce in-hospital mortality.
clinicaltrials.gov Identifier: NCT01785966
Quiz Ref IDChecklists have been proposed as tools to ensure that essential components of care are not omitted.1- 3 In intensive care units (ICUs), the use of checklists is associated with increased adherence to guidelines,4 reduced rates of central line–associated bloodstream infection,3,5 and earlier extubation.6 Using checklists combined with daily goals assessment and clinician prompting may improve communication, adherence to care processes, and clinical outcomes.7,8 However, the evidence supporting the use of checklists in critical care derives from before-after studies,3,4,8 or studies aimed at specific conditions,5 all conducted in high-income countries. Evidence from randomized trials is lacking,9 akin to that for quality improvement (QI) interventions in general.10
Low- or middle-income countries such as Brazil sustain 85% of the global burden of critical illness.11 In these settings, staff tend to operate under vertical, rather than horizontal, hierarchy, which contributes to a poor work climate and safety culture.12,13 Perhaps as a consequence, adherence with guidelines is lower, and severity-adjusted outcomes are worse than in high-income countries.14,15 Checklists, read out loud by team members, have successfully flattened hierarchy in a number of settings, improving work climate and process adherence.7,8 We therefore hypothesized that such an approach would improve work climate, care processes, and mortality in Brazilian ICUs. Thus, we conducted a cluster randomized trial to assess the effect of a multifaceted QI intervention that promoted a daily checklist, goal setting during multidisciplinary rounds, and clinician prompts on in-hospital mortality of critically ill patients.
The CHECKLIST-ICU study was conducted in 2 phases (Figure 1). Phase 1 was an observational study to assess baseline data on work climate, care processes, and clinical outcomes in Brazilian ICUs caring for adult patients. In phase 2, the ICUs were randomized to receive the multifaceted QI intervention or to routine care. The unit of concealed randomization was the ICU to minimize contamination, as we applied the intervention to the whole ICU multidisciplinary team. As part of the agreement to be included in the study, all ICUs randomized to control received the intervention after the study was completed.
The ethics committees of all institutions approved the study. The protocol and statistical analysis plan were published previously (see trial protocol in Supplement 1 and statistical analysis plan in Supplement 2).16,17 To avoid selection bias, written consent was obtained at the cluster level from the director of each institution.18 The funders had no role in the analysis or publication decision. Nevertheless, the funders and ICU leaders requested an intervention period not longer than 6 months so that ICUs randomized to the control group would also receive the intervention in a timely fashion.
Adult ICUs were included from all Brazilian regions. Pediatric ICUs, cardiac ICUs, step-down units, ICUs whose leadership did not or would not implement multidisciplinary daily rounds, and ICUs that already used checklists during rounds were excluded. When 2 or more ICUs from a single institution participated in the study, they were considered a single unit, except for those with completely separate teams. Intensive care units that successfully collected data in the observational phase (ie, inclusion of ≥30 patients within 6 months and collection of the Safety Attitudes Questionnaire [SAQ] from >75% of their staff) were randomized.
In both phases, ICUs enrolled between 40 and 60 consecutive patients 18 years or older with an ICU stay longer than 48 hours. Patients in whom a high probability of early death was anticipated (defined as death occurring between the 48th and 72nd hours of the ICU stay), patients receiving exclusive palliative care, and those with a suspected or confirmed diagnosis of brain death were excluded.
The units of randomization were the ICUs. A statistician from Research Institute HCor generated the randomization list with random permuted blocks of 4 units using an appropriate statistical package. Stratification was performed according to the median in-hospital mortality determined in the observational phase. To ensure allocation concealment, the statistician who prepared the list received only the identification code of the unit and was not aware of the identity of the ICU. The allocation list was then sent to the research manager, who informed the ICUs about their randomization status in January 2014.
Caregivers were not blinded to group assignment. Research coordinators and the statistician who analyzed the data were also not blinded, because data regarding adherence to checklist and clinician prompting were collected exclusively for the intervention group ICUs. Ventilator-associated pneumonia and central line–associated bloodstream infection diagnoses were adjudicated by a blinded committee using standardized definitions.19,20
Quiz Ref ID
The intervention consisted of modifying daily multidisciplinary rounds to include the use of a checklist and discussion of goals of care, together with clinician prompting later in the day to ensure follow-through with checklist adherence and goals of care, for all patients during their entire ICU stay. The checklist (eFigure 1 in Supplement 3) was developed based on the clinical practice guideline development cycle.21 It targeted 11 care processes aimed at prevention of venous thromboembolism (VTE), ventilator-associated pneumonia (head-of-bed ≥30°), central line–associated bloodstream, and urinary tract infection (removal of unnecessary urinary or venous catheters); improvement in nutrition and analgesia; reduction of sedation; assessment of readiness for extubation; detection of severe sepsis and acute respiratory distress syndrome; optimization of antibiotics (indication to start, adjust, or stop); and reduction of tidal volume.
A strategy to leverage the potential of the checklist was used, not only to avoid errors of omission, but also to promote a flat hierarchy by empowering the entire team to participate in daily rounds. Thus, during weekday rounds, the checklist was read aloud by a nurse and answered by participants. Daily goals were registered on a standardized form (eFigure 2 in Supplement 3) and then read aloud to the team. Every afternoon, a nurse reviewed the daily goals and prompted the on-call physician when there were pending items. The goal was that the intervention be applied at least on weekdays (Monday through Friday).
The following actions were taken to ensure adherence to the QI intervention (eTable 1 in Supplement 3): an investigators meeting; a visit of 1 member of the steering committee to every ICU to train the multidisciplinary team and participate in the rounds; audit and feedback on adherence to care processes; a study website; and dissemination of videos with testimonials of opinion leaders that focused on the importance of the study, and successful QI experiences, with special emphasis on the principle that the whole team works better than isolated voices. In addition, electronic text messages were sent every 2 to 3 days to the multidisciplinary team at the sites regarding adherence to the intervention, and ICU medical and nursing directors were contacted if adherence was low. Intensive care units started implementing the study intervention soon after the investigators meeting (March 14, 2014) and no later than March 31, 2014. Control group ICUs maintained routine care and received no preintervention training.
The primary outcome was in-hospital mortality, truncated at 60 days. Mortality was chosen as the primary outcome for several reasons. First, it is the most important outcome for critically ill patients and their relatives. Second, mortality is high among patients admitted to ICUs in Brazil. Third, previous before-after studies have suggested decreased mortality with the use of a surgical safety checklist2 and clinician prompting in ICUs.8
Secondary exploratory outcomes were adherence to care processes, ICU safety climate, and clinical outcomes (eTable 2 in Supplement 3). Adherence to 7 care processes targeted by the checklist was measured: patients receiving enteral or parenteral feeding; head-of-bed elevated at 30° or more; patients receiving moderate to light sedation or alert and calm (Richmond Agitation-Sedation Scale score −3 to 0); mechanical ventilation tidal volume 8 mL/kg of predicted body weight or less; VTE prophylaxis; central venous catheter use; and indwelling urinary catheter use. Four items targeted by the checklist were not assessed due to feasibility constraints (timely screening of sepsis, adequacy of antibiotics, adequacy of analgesia, and daily spontaneous breathing trials). Six ICU safety climate domains (team work climate, safety climate, job satisfaction, stress recognition, perception of management, working conditions) were evaluated using the validated Brazilian-Portuguese version of the SAQ, administered anonymously to promote uninhibited answers.22,23 Furthermore, the following clinical secondary outcomes were assessed: ICU mortality, mechanical ventilation-free days within 28 days, central line–associated bloodstream infection rate, ventilator-associated pneumonia rate, urinary tract infection rate, length of ICU stay, and length of hospital stay.
Trained research coordinators not involved in the care of ICU patients collected data prospectively in a web-based data capture system in both phases. Quality control was guaranteed by automated data entry checks, weekly contact with investigators, and central statistical checks.
According to the sample size calculation, with 102 ICUs and an average of 50 patients per unit, the study had 90% power with a type I error rate of 5% to detect an absolute reduction in in-hospital mortality of 6% (from 30% in the control group to 24% in the intervention group), considering a coefficient of variation K of 0.25.24 Baseline in-hospital mortality was estimated at 30% considering data of patients with an ICU stay longer than 48 hours from a large administrative database of Brazilian ICUs (Epimed Monitor System, Epimed Solutions). A 20% relative (6% absolute) risk reduction in mortality was considered to be a clinically relevant, biologically plausible, moderate effect size, and one typically observed in health care interventions.25
All analyses were prespecified and followed the intention-to-treat principle.17 Because ICUs were randomized rather than patients and we measured outcomes at the patient level, the analyses were adjusted for clustering of the data. The primary outcome was analyzed using a prespecified random-effects logistic regression model, with adjustment for patients’ Simplified Acute Physiology Score 3 (SAPS3) and for each ICU’s standardized mortality ratio (calculated as the observed hospital mortality divided by that predicted from the SAPS3 observed in the observational phase) to account for potential baseline imbalances in patients’ severity and ICUs’ performance. Multiple imputation with chained equations was used for missing in-hospital vital status, assuming data were missing at random (mice package in R, R Foundation for Statistical Computing).26 The imputation was conducted with 5 replicates using a logistic regression model with baseline covariates for sex and in-hospital mortality at the observational phase.
For all other analyses, the effect estimate was generated from a generalized linear mixed model that included a term for the observed outcomes within each ICU during the observational phase. Additional analyses included prespecified subgroups (2 strata of baseline in-hospital mortality [by median] observed for ICUs in the observational phase, public or private hospital, medical or surgical patient, 2 strata of the Sepsis-Related Organ Failure Assessment [SOFA] score, presence of sepsis at admission, and need for mechanical ventilation at admission) and subgroups defined a posteriori (academic or nonacademic hospitals, time of patients’ enrollment at or before 60 days or after 60 days) using formal interaction tests.
All tests were 2-sided, and a P value lower than .05 denoted statistical significance. Prespecified secondary outcome analyses and subgroup analyses were not adjusted for multiple comparisons; thus, their results should be interpreted as exploratory. In post hoc sensitivity analyses, the significance level for secondary outcomes was adjusted using a Sidak correction, a method to control overall type I error when there are several independent hypothesis tests.27 Analyses were performed with R version 2.10.1.
The observational phase ran from September 2013 through March 2014, and the randomization phase ran from April 2014 through November 2014. A total of 335 ICUs accepted to participate; however, 149 were ineligible (120 were already applying checklists during daily rounds, 12 did not conduct and would not start multidisciplinary daily rounds, 6 were coronary care unit or step-down units, and 11 were pediatric units), and 55 eventually declined to participate. Thus, 131 ICUs were enrolled in the observational phase, of which 118 collected the required data and were randomized (59 in the intervention and 59 in the control group) (Figure 1). There were 11 ICUs located in the same hospitals, which were considered different units in the study. Only 1 hospital had ICUs allocated to different groups (2 ICUs assigned to intervention and 4 to control). Characteristics of the ICUs were similar between groups, although control ICUs had more beds and were more often located in academic hospitals (Table 1). Characteristics of ICUs not included in the randomized phase were similar to randomized ICUs (eTable 3 in Supplement 3).
A total of 13 638 patients in 118 ICUs were enrolled, including 6877 patients (mean age, 59.7 years; 3218 [46.8%] women) in the observational phase and 6761 in the randomized phase (mean age, 59.6 years; 3098 [45.8%] women) (Figure 1). The mean duration each ICU was enrolled in the study was 4.4 months (median, 4.3; interquartile range [IQR], 3.7-5.0) and 4.5 months (median, 4.5; IQR, 3.5-5.4) in the control and intervention groups, respectively. The primary outcome was available for 6877 patients (100%) in the observational phase and 6758 patients (99.9%) in the randomized phase, although all patients were included in the primary outcome analyses. Responses to the SAQ were obtained from 6656 (85.3%) staff members in the observational phase and 6375 (78.8%) in the randomized phase, with response rates 75% or greater for 99.2% and 92.4% of ICUs, respectively, in both phases. Patient characteristics were similar between groups in both phases (Table 2). A total of 3435 (49.9%) and 3236 (47.9%) patients received mechanical ventilation in the observational and randomized phases, respectively.
Multidisciplinary rounds occurred on 55.3 days per 100 patient-weekdays in the 118 ICUs in the observational phase. During the randomized phase, multidisciplinary rounds increased to 92.8 days per 100 patient-weekdays in intervention group ICUs, compared with 61.5 days per 100 patient-weekdays in control ICUs (P < .001). In the intervention group, the checklist was applied on 90.6 days per 100 patient-weekdays and clinician promptings on 89.1 days per 100 patient-weekdays.
In the observational phase, in-hospital mortality was 32.5%, whereas the SAPS3-predicted risk was 27.1%. Intensive care unit mortality was 25.0%. Rates of central line–associated bloodstream infection, ventilator-associated pneumonia, and urinary tract infection were 7.4, 4.3, and 10.6 cases per 1000 patient-days, respectively. The mean (SD) lengths of ICU stay and hospital stay were 10.6 (11.2) days and 19.8 (16.6) days, respectively. Among patients receiving mechanical ventilation, the mean (SD) number of days receiving ventilator was 8.1 (9.3), and they had a mean (SD) 13.2 (7.7) ventilator-free days in 28 days.
In the randomized phase, there were 1096 in-hospital deaths among 3327 patients (32.9%) in the intervention group and 1196 deaths among 3434 patients (34.8%) in the control group (adjusted odds ratio [OR], 1.02; 95% CI, 0.82 to 1.26; P = .88) (Figure 2 and eTable 4 in Supplement 3). The between-group adjusted difference for in-hospital mortality was 0.39% (95% CI, −3.70 to 4.47; P = .84).
The QI intervention had no effect on secondary exploratory clinical outcomes: ICU mortality, central line–associated bloodstream infection, ventilator-associated pneumonia, urinary tract infection, mean ventilator-free days, mean days receiving mechanical ventilation, and mean ICU or hospital length of stay (Figure 2 and eTable 4 in Supplement 3).
Adherence varied across the 7 exploratory outcomes reflecting care processes during the observational phase (eTable 4 in Supplement 3). For only 36.1% of the total patient-days, patients received a moderate level of sedation or were alert and calm, and on 54.1% of patient-days, patients received a tidal volume of 8 mL/kg of predicted body weight or less. Head-of-bed was elevated 30° or more on 90.3% and VTE prophylaxis was adequate on 70.1% of patient-days. Central venous lines and urinary catheters were used on 74.1% and 73.2% of patient-days, respectively.
The QI intervention improved adherence for the 4 care processes that had poor baseline adherence: increased use of low tidal volume (67.5% vs 58.9% of patient-days; adjusted rate ratio [RR], 1.14; 95% CI, 1.03-1.26; P = .01) and patient-days receiving light sedation or alert and calm among patients under mechanical ventilation (40.5% vs 35.0% of patient-days; adjusted RR, 1.19; 95% CI, 1.00-1.42; P = .05) and decreased use of central venous catheters (72.4% vs 72.9% of patient-days; adjusted RR, 0.90; 95% CI, 0.83-0.98; P = .02) and urinary catheters (62.8% vs 74.8% of patient-days; adjusted RR, 0.86; 95% CI, 0.80-0.93; P < .001) (Figure 3 and eTable 4 in Supplement 3). The QI intervention did not affect the 3 care processes with better baseline adherence (bed elevation to ≥30° [95.6% vs 89.7% of patient-days; adjusted RR, 1.05; 95% CI, 0.99-1.11; P = .14]; VTE prophylaxis [74.8% vs 75.0% of patient-days; adjusted RR, 1.05; 95% CI, 0.91-1.22; P = .50]; and diet administration [79.2% vs 76.4% of patient-days; adjusted RR, 1.03; 95% CI, 0.89-1.20; P = .65]) (Figure 3 and eTable 4 in Supplement 3).
At baseline, the percentages of positive answers for exploratory SAQ domains were 47.9% for teamwork climate, 32.5% for safety climate, 72.5% for job satisfaction, 54.7% for stress recognition, 20.7% for management perception, and 45.6% for work conditions. There were significant differences between groups in teamwork and safety climate by the end of the randomization phase. In the intervention group, 53.8% of staff answered that teamwork climate was positive, compared with 45.8% in the control group (adjusted OR, 1.30; 95% CI, 1.08-1.57; P = .01) (Table 3 and eTable 4 in Supplement 3). In the intervention group, 36.5% of staff answered that safety climate was positive, compared with 31.9% in the control group (adjusted OR, 1.27; 95% CI, 1.02-1.57; P = .03). There were no differences between groups in the other SAQ domains. The effects on all secondary outcomes were not significant after adjustment for multiple comparisons, except for a reduction in the use of urinary catheters (Sidak-corrected significance level, .002).
There was no significant heterogeneity in the intervention effect on in-hospital mortality in the predefined subgroups of patients with higher or lower baseline in-hospital mortality; public, private nonprofit, or private for-profit hospitals; need for mechanical ventilation at admission or not; medical or surgical admission; lower or higher SOFA score; or presence of sepsis at admission or according to SAPS3 decile (eTable 5 in Supplement 3). However, in exploratory analyses not adjusted for multiple comparisons, the effects of the QI intervention on in-hospital mortality were significantly different according to baseline levels of safety climate and perception of management (eTable 5 in Supplement 3).
Because there was imbalance in the number of academic hospitals between groups, a post hoc analysis was conducted to assess the effect of the intervention on outcomes in the subgroup of academic vs nonacademic ICUs. Effects on in-hospital mortality, care processes, and SAQ domains were similar (all P values for interaction >.05) between academic and nonacademic hospitals. Additional post hoc analysis was conducted to assess whether the effect on mortality might differ among patients included earlier (within 60 days from the inclusion of the first patients in a given ICU) vs later in the randomized phase (>60 days since the inclusion of the first patient in a given ICU). There was no interaction between treatment effect on mortality and time of enrollment (enrolled ≤60 days: OR, 1.05; 95% CI, 0.81-1.36; enrolled >60 days: OR, 0.94; 95% CI, 0.68-1.3; P value for interaction = .42).
Quiz Ref IDIn this study of Brazilian ICUs, the introduction of daily checklists, goal setting, and clinician prompting did not decrease in-hospital mortality or other clinical outcomes. The large numbers of clusters included and events observed allowed us to reliably exclude a moderate to large effect on mortality (≥6% absolute reduction). In addition, potential improvements were observed in 4 of 7 care processes and 2 safety climate domains, although except for 1 outcome, urinary catheter use, these findings were not significant after adjustment for multiple comparisons.
Quiz Ref IDSeveral potential explanations were considered for the lack of effect on mortality. First, although adherence to the QI intervention was adequate, the effect on care processes was modest and questionable due to the multiple unadjusted comparisons. Second, it is possible that the intervention needs time to work and our observation period was too short. Although there was a signal suggesting improvement in care processes and safety culture, a longer intervention period might have led to a more horizontal hierarchy, greater trust and teamwork, and ultimately demonstrable effects on patient outcomes.3,5 However, both the government funder and local clinical leaders stipulated the short duration so that control ICUs could receive the intervention quickly. Third, it is possible that the items on the checklist have very modest or negligible effects on mortality. Although all the care processes included in the checklist used in this study are recommended by guidelines, most have uncertain effects on mortality.28- 32 For example, care processes such as semirecumbent position and VTE prophylaxis may decrease adverse events but have no proven effect on mortality.33,34
Quiz Ref IDStudies assessing checklists or clinician prompting have found contradictory results. Some suggest marked improvements on adherence to care processes and clinical outcomes,2- 5,8 while others found limited35 or no effect2,36 on care processes. A systematic review of 9 before-after studies of safety checklists in acute care found inconsistent improvements in patient safety.9 These studies differ in several ways. Haynes et al2 showed reductions in complication rates and mortality with the implementation of a checklist during surgery, but the study involved a different population. In addition, this and other studies involving critically ill patients were simplistic before-after studies, subject to bias by secular trends.2- 4,8,9 Indeed, the replication of either the safe surgery checklist in Ontario37 or the checklist to prevent central line infections in England36 did not change complications, infections, or deaths. Other studies using a cluster randomized design similar to that of the current study assessed different outcomes and interventions.5,35 Marsteller et al5 evaluated a multifaceted intervention exclusively focused on preventing catheter-related bloodstream infections. Scales et al35 tested the effectiveness of a QI program to improve adherence to 6 care processes, although checklists were used only during central line insertion. Thus, the robust assessment of the effects of daily round checklists and clinician prompting in a randomized trial is a unique feature of the current study.
This study has several strengths. Over a relatively short time frame, a countrywide government-sponsored initiative with modest funding was able to successfully introduce and assess the effects of a multifaceted QI intervention. The trial was large, and the cluster randomization worked well, with a good balance of baseline characteristics between the 2 groups. The intervention was effectively deployed, and all relevant data were captured, including high response rates from staff, both in the observational and randomized phases. Bias was minimized through allocation concealment and through independent data collection by staff uninvolved with patient care. Furthermore, this study is one of the first of this scale to be conducted in an intensive care setting in a middle-income country, demonstrating important proof of feasibility.
This study also has several limitations. The duration of intervention was limited. In addition, restricting the sample to patients with more than 48 hours of ICU stay may have lead to a theoretical risk of imbalance between groups. Nevertheless, baseline characteristics were well balanced. Moreover, the care processes addressed in the checklist (eg, VTE prophylaxis or semirecumbent position) were unlikely to benefit those who were discharged early from the ICU or to prevent early deaths. Information on excluded patients or on discharge location for survivors was not collected, although hospital discharge to locations other than home is rare in Brazil. Four items targeted by the checklist were not assessed because of feasibility constraints. Qualitative assessments of the root causes of a suboptimal safety climate were not conducted. In addition, these results may not be applicable to settings with different baseline levels of safety climate.
Implementation of a multifaceted QI intervention with daily checklists, goal setting, and clinician prompting did not reduce in-hospital mortality compared with routine care among critically ill patients treated in ICUs in Brazil.
Corresponding Author: Alexandre B. Cavalcanti, MD, PhD, Research Institute, HCor–Hospital do Coração, Rua Abílio Soares 250, 12th Floor, 04005-000, São Paulo, SP, Brazil (email@example.com).
Authors/Writing Group for the CHECKLIST-ICU Investigators and the Brazilian Research in Intensive Care Network (BRICNet): Alexandre B. Cavalcanti, MD, PhD; Fernando Augusto Bozza, MD, PhD; Flavia R. Machado, MD, PhD; Jorge I. F. Salluh, MD, PhD; Valquiria Pelisser Campagnucci, MD, PhD; Patricia Vendramim, MSc; Helio Penna Guimaraes, MD, PhD; Karina Normilio-Silva, MSc; Lucas Petri Damiani, MSc; Edson Romano, MD, PhD; Fernanda Carrara, RN; Juliana Lubarino Diniz de Souza, RN; Aline Reis Silva, PharmD, MSc; Grazielle Viana Ramos, PharmD; Cassiano Teixeira, MD, PhD; Nilton Brandão da Silva, MD, PhD; Chung-Chou H. Chang, PhD; Derek C. Angus, MD, MPH; Otavio Berwanger, MD, PhD.
Affiliations of Authors/Writing Group for the CHECKLIST-ICU Investigators and the Brazilian Research in Intensive Care Network (BRICNet): Research Institute, HCor–Hospital do Coração, São Paulo, Brazil (Cavalcanti, Guimaraes, Normilio-Silva, Damiani, Romano, Berwanger); D’Or Institute for Research and Education, Rio de Janeiro, Brazil (Bozza, Salluh, Silva, Ramos); Oswaldo Cruz Foundation, Rio de Janeiro, Brazil (Bozza); Latin American Sepsis Institute, São Paulo, Brazil (Machado, Carrara, Lubarino Diniz de Souza); Hospital Samaritano, São Paulo, Brazil (Campagnucci, Vendramim); Hospital Moinhos de Vento, Porto Alegre, Brazil (Teixeira, Brandão da Silva); University of Pittsburgh, Pittsburgh, Pennsylvania (Chang, Angus); Associate Editor, JAMA (Angus).
Author Contributions: Dr Cavalcanti had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Cavalcanti, Bozza, Machado, Salluh, Campagnucci, Vendramim, Guimaraes, Normilio-Silva, Damiani, Romano, Carrara, Lubarino Diniz de Souza, Teixeira, Brandão da Silva, Angus, Berwanger.
Acquisition, analysis, or interpretation of data: Cavalcanti, Bozza, Machado, Salluh, Campagnucci, Normilio-Silva, Silva, Ramos, Teixeira, Chang, Berwanger.
Drafting of the manuscript: Cavalcanti, Bozza, Salluh, Campagnucci, Vendramim, Normilio-Silva, Romano, Carrara, Lubarino Diniz de Souza, Teixeira, Angus, Berwanger.
Critical revision of the manuscript for important intellectual content: Cavalcanti, Bozza, Machado, Salluh, Campagnucci, Guimaraes, Normilio-Silva, Damiani, Silva, Ramos, Teixeira, Brandão da Silva, Chang, Berwanger.
Statistical analysis: Damiani, Chang, Berwanger.
Obtained funding: Cavalcanti, Bozza, Salluh, Vendramim, Berwanger.
Administrative, technical, or material support: Cavalcanti, Salluh, Campagnucci, Vendramim, Guimaraes, Normilio-Silva, Carrara, Lubarino Diniz de Souza, Silva, Ramos, Teixeira.
Study supervision: Cavalcanti, Bozza, Machado, Campagnucci, Brandão da Silva, Berwanger.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This study was conducted as part of the Program to Support Institutional Development of Universal Health System (PROADI) from the Brazilian Ministry of Health. It was funded mainly by the Brazilian Health Surveillance Agency (ANVISA), PROADI, and Brazilian Development Bank (BNDES). D’Or Institute for Research and Education also contributed with additional funding.
Role of the Funders/Sponsors: The funding sources had no role in the conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Nevertheless, the funders requested an intervention period not longer than 6 months so that intensive care units randomized to the control group would also receive the intervention in a timely fashion. The teams of the Research Institute HCor, D’Or Institute for Research and Education, Hospital Samaritano of São Paulo, and the Latin American Sepsis Institute are the sponsors and coordinators of the study, in association with the Brazilian Research in Intensive Care Network. The Research Institute HCor managed the data and conducted all analyses.
CHECKLIST-ICU Site Investigators: Cardioeste: Pedro Aniceto Nunes Neto, Carlos Alberto Fernandes, Cristiano Ávila Luchesi; Casa de Saúde Santa Marta: Natalia Rodrigues de Sa, Roberta Antunes Faria Rodrigues, Larissa Stefany Radespiel; Casa de Saúde São Bento: Vinicius Leonardo dos Santos Iorio, Danielle Bhering, Viviane Ribeiro Linhares; Centro Hospitalar Unimed: Glauco Adrieno Westphal, Geonice Sperotto; Complexo Hospitalar São Mateus: Soraya Byana Rezende da Silva Rossi, Thaismari Escarmanhani Ferreira; Fundação Hospital de Clinicas Gaspar Vianna: Edgar de Brito Sobrinho, Helder Jose Lima Reis, Mauricio Soares Carneiro, Adriana de Oliveira Lameira Veríssimo; Fundação Hospital Getúlio Vargas: Juliana Fernandez Fernandes, Rodrigo Lopes Ferreira; Hospital Aristides Maltez: Sylvania Campos Pinho, Leonardo Dultra, Lise Oliveira Hizumi; Hospital Aviccena: Neiva Fernandes de Lima, Alexsandra Raimunda da Silva, Liliane dos Santos; Hospital Casa de Caridade de Carangola: Sidiner Mesquita Vaz, José Marcio Oliveira, Miria Bonjour Laviola; Hospital Casa de Portugal: Wania Vasconcelos de Freitas, Leonardo Passos; Hospital Copa D’Or: Ligia Sarmet Cunha Farah Rabello, Carollina Resende de Siqueira; Hospital da Sagrada Familia: Luiz Carlos de Oliveira Silva, Francisco Felix Barreto Junior, Eduarda Maria Alves Cruz; Hospital das Clinicas–FMUSP: Luiz Marcelo Sa Malbouisson, Fabiola Prior Caltabeloti, Estevão Bassi, Patrícia Regiane da Silva, Filipe Cadamuro, Renata Graciliano dos Santos Cagnon, Yeh-Li Ho, Lucas Chaves Neto, Bruno Azevedo Randi; Hospital das Clinicas da UFPE: Michele Maria Goncalves de Godoy, Pollyanna Dutra Sobral, Evônio de Barros Campelo Júnior; Hospital das Clínicas da Faculdade de Medicina de Botucatu/UNESP: Laercio Martins de Stefano, Ana Lucia Gut, Greicy Mara Mengue Feniman de Stefano; Hospital das Clinicas de Goiás: Denise Milioli Ferreira, Fernanda Alves Ferreira Gonçalves; Hospital das Clinicas Luzia de Pinho Melo–SPDM: Fernanda Rubia Negrao Alves, Wilson Nogueira Filho, Renata Ortiz Marchetti; Hospital de Base do Distrito Federal: Sheila Sa, Nadia Tomiko Anabuki; Hospital de Clinicas de Porto Alegre: Silvia Regina Rios Vieira, Edino Parolo, Karen Prado, Natalia Gomes Lisboa; Hospital de Messejana Dr Carlos Alberto Studart Gomes: Simone Castelo Branco Fortaleza, Maria Liduina Nantua Beserra Porfirio, Marcela Maria Sousa Colares; Hospital de Urgência e Emergência de Rio Branco: Rosicley Souza da Silva, Marcia Vasconcelos, Fabio de Souza; Hospital Distrital Dr Evandro Ayres de Moura: Lanese Medeiros de Figueiredo, Niedila Pinheiro Bastos Seabra, Paula Celia Pires de Oliveira; Hospital do Coração Balneário Camboriú: Marcio Andrade Martins, Ricardo Beduschi Muller, Manoela Cristina Recalcatti; Hospital e Pronto Socorro Dr Aristóteles Platao Bezerra de Araújo: Riani Helenditi Fernandes Camurça Martins, Leatrice Emilia Ferreira de França; Hospital Eduardo de Menezes: Frederico Bruzzi de Carvalho, Gustavo Cesar Augusto Moreira; Hospital Estadual Américo Brasiliense: Evelin Drociunas Pacheco Cechinatti, Aline Cristina Passos, Mariana da Costa Ferreira; Hospital Estadual Getúlio Vargas–Secretaria de Estado de Saúde: Antonio Carlos Babo Rodrigues, Vladimir dos Santos Begni; Hospital Estadual Monsenhor Walfredo Gurgel: Alfredo Maximo Grilo Jardim, Amanda Carvalho Maciel; Hospital Evangélico de Cachoeira de Itapemirim: Marlus Muri Thompson, Erica Palacio Berçacola Pinheiro, Claudio Henrique Pinto Gonçalves; Hospital Federal da Lagoa: Ricardo Schilling Rosenfeld, Valéria Abrahao Rosenfeld, Leticia Japiassú; Hospital Federal Servidores do Estado: Maria Inês Pinto de Oliveira Bissoli, Rosemary da Costa Tavares, Marta Rocha Gonçalves; Hospital Fernandes Tavora: Laercia Ferreira Martins, Maria Helena Oliveira, Adriana Kelly Almeida Ferreira; Hospital Geral de Camaçari: Emídio Jorge Santos Lima, Tárcio de Almeida Oliveira, Milena Teixeira Campos; Hospital Geral de Fortaleza: Nilce Almino de Freitas, Stephanie Wilkes da Silva, Vera Lucia Bento Ferreira; Hospital Geral de Nova Iguaçu: Alexander de Oliveira Sodré, Cid Leite Villela, Eduardo Duque Estrada Medeiros; Hospital Geral de Palmas: Nairo Jose de Souza Junior, Jhocrenilcy de Souza Maya Nunes, Rones de Souza Monteiro; Hospital Guilherme Álvaro: André Scazufka Ribeiro, Carlos Cesar Nogueira Giovanini, Elisete Tavares Carvalho; Hospital Instituto Dr Jose Frota: Lenise Castelo Branco Camurça Fernandes, Domitilha Maria Coelho Rocha, Cristiane Maria Gadelha de Freitas; Hospital Joari: Marcia Adélia de Magalhaes Menezes, Rosa Imaculada Stancato, Guilherme Brenande Alves Faria, Giovanna Asturi; Hospital Maternidade e Pronto Socorro Santa Lucia: Ricardo Reinaldo Bergo, Frederico Toledo Campo Dall’Orto, Gislayne Rogante Ribeiro; Hospital Maternidade Municipal Dr Odelmo Leão Carneiro: Cidamaiá Aparecida Arantes, Michelle Aparecida dos Santos Toneto; Hospital Moinhos de Vento: Cassiano Teixeira, Juçara Gasparetto Maccari; Hospital Municipal Dr Carmino Caricchio: Sergio Tadeu Górios, Julliana Pires de Morais, Daniela Marangoni Zambelli; Hospital Municipal de Paracatu: Roberta Machado de Souza, Eduardo Cenísio Teixeira de Paiva, Alessandra Gonçalves Ribeiro; Hospital Municipal Dr Mario Gatti: Marcus Vinicius Pereira, Leoni Nascimento, Rosangela da Silva; Hospital Municipal Padre Germano Lauck de Foz do Iguaçu: Roberto de Almeida, Karin Aline Zilli Couto, Izabella Moroni Toffolo; Hospital Municipal Pedro II: Jorge Eduardo da Rocha Paranhos, Antonio Ricardo Paixão Fraga; Hospital Municipal Souza Aguiar: Alberto Augusto de Oliveira Junior, Roberto Lannes, Andrea da Silva Gomes Ludovico; Hospital Naval Marcilio Dias: Luiz Fernando Costa Carvalho, Leticia de Araújo Campos, Patricia Soboslai; Hospital Nereu Ramos: Israel Silva Maia, Tatiana Rassele, Christiany Zanzi; Hospital Nossa Senhora Auxiliadora: Valeria Nunes Martins Michel, Sinésio Pontes Gonçalves; Hospital Nossa Senhora dos Prazeres: Ricardo Rath de Oliveira Gargioni, Rosangela Zen Duarte; Hospital 9 de Julho: Mariza D’Agostino Dias, Andrea Delfini Diziola, Daniela Veruska da Silva, Melissa Sayuri Hagihara; Hospital Primavera: André Luís Veiga de Oliveira, Diego Leonnardo Reis, Janaina Feijó; Hospital Procordis: Marco Antonio da Costa Oliveira, Luiza Veiga Coelho de Souza; Hospital Pronto Socorro 28 de Agosto: Liane de Oliveira Cavalcante, Jacilda Rodrigues, Moises Cruz de Pinho; Hospital Regional da Unimed Fortaleza: Marcos Antonio Gadelha Maia, Vladia Fabiola Jorge Lima, Emilianny Maria Nogueira, Antonielle Carneiro Gomes; Hospital Regional de Juazeiro: Katia Regina de Oliveira, Jose Antonio Bandeira, Carla Cordeiro Botelho Mesquita; Hospital Universitário Regional dos Campos Gerais: Guilherme Arcaro, Camila Wolff, Délcio Caran Bertucci Filho; Hospital Regional de Presidente Prudente: Gustavo Navarro Betonico, Rafaela Pereira Maroto, Leonardo Fantinato Menegon; Hospital Regional Deputado Manoel Gonçalves Abrantes: Patrício Junior Henrique da Silveira, Germana Estrela Gadelha de Queiroga Oliveira, Wallber Moreno da Silva Lima; Hospital Regional Público do Araguaia: Lilian Batista Nunes, Sotero Gonçalves Martins Neto, Liwcy Keller de Oliveira Lopes Lima; Hospital Regional João Penido: Maria Augusta de Mendonça Lima, Lidiane Miranda Milagres, Vivian Gribel D’Ávila; Hospital Santa Casa de Campo Mourão: Paulo Marcelo Schiavetto, Paulo Alves dos Santos, Francislaine de Matos Raimundo; Hospital Municipal Santa Isabel–Joao Pessoa: Carmen Leonilia Tavares de Melo, Aline Albuquerque de Carvalho, Cynthia Franca de Santana; Hospital Santa Lucia de Divinópolis: Adriana Lessa Ventura Fonseca, Martha Aparecida da Silva; Hospital Santa Rosa: Mara Lilian Soares Nasrala, Eloisa Kohl Pinheiro, Mara Regina Pereira Santos; Hospital Municipal São Francisco de Assis: Guilherme Abdalla da Silva, Rener Moreira, Marcia Loureiro Sebold; Hospital São Joao de Deus: Marcone Lisboa Simões da Rocha, Marco Antonio Ribeiro Leão, Jaqueline de Assis; Hospital São Jose–Criciúma: Felipe Dal Pizzol, Cristiane Damiani Tomasi; Hospital São Jose de Doenças Infecciosas: Jose Nivon da Silva, Luciana Vladia Carvalhedo Fragoso, Denise Araújo Silva Nepomuceno Barros; Hospital São Jose de Teresópolis: Mauricio Mattos Coutinho, Robson Sobreira Pereira, Veronica Oliveira dos Santos; Hospital São Jose do Avaí: Sergio Kiffer Macedo, Diego de Souza Bouzaga Furlani, Eduardo Silva Aglio Junior; Hospital São Paulo: Aécio Flavio Teixeira de Góis, Kathia Teixeira, Paula Zhao Xiao Ping; Hospital São Vicente de Paula–Cruz Alta: Paulo Ricardo Nazario Viecili, Simone Daniela Melo de Almeida; Hospital São Vicente de Paulo–Vitória da Conquista: Geovani Moreno Santos Junior, Djalma Novaes Araújo Segundo, Marielle Xavier Santana; Hospital São Vicente de Paulo–Passo Fundo: Jose Oliveira Calvete, Luciana Renner; Hospital Unimed Bebedouro: Vinicius Vandré Trindade Francisco, Danytieli Silva de Carvalho, Adriana Neri da Silva Batista Campos; Hospital Unimed Costa do Sol: Sergio Leôncio Fernandes Curvelo Jr, Nayara Ribas de Oliveira, Suelem de Paula Freitas Deborssan; Hospital Unimed Maceió: Maria Raquel dos Anjos Silva Guimaraes, Luiz Claudio Gomes Bastos, Luciene Moraes Gomes; Hospital Unimed Natal: Erico de Lima Vale, Dionísia Arianne Vieira da Silva; Hospital Unimed Rio: Renato Vieira Gomes, Marco Antonio Mattos, Pedro Miguel Matos Nogueira, Viviane Cristina Caetano Nascimento; Hospital Universitário–Universidade Estadual de Londrina: Cintia Magalhaes Carvalho Grion, Alexsandro Oliveira Dias, Glaucia de Souza Omori Maier; Hospital Universitário Cassiano Antonio Moraes–Universidade Federal do Espirito Santo: Paula Frizera Vassallo, Maria Helena Buarque Souza de Lima, Wyllyam Loss dos Reis, Walace Lirio Loureiro, Andressa Tomazini Borghardt; Hospital Universitário Lauro Wanderley: Ciro Leite Mendes, Paulo Cesar Gottardo, Jose Melquiades Ramalho Neto; Hospital Universitário Onofre Lopes: Eliane Pereira da Silva, Maria Gorette Lourenço da Silva Aragao, Elisângela Maria de Lima; Hospital Universitário Polydoro Ernani de São Thiago–Universidade Federal de Santa Catarina: Rafael Lisboa de Souza, Ken Sekine Takashiba; Hospital Universitário Prof. Edgar Santos–Universidade Federal da Bahia: Dimitri Gusmão-Flores, Taciana Lago Araújo, Rosana Santos Mota; Hospital Universitário Regional de Maringá: Almir Germano, Flavia Antunes, Sandra Regina Bin Silva; Hospital Universitário São Francisco de Paula–Pelotas: Marcio Osório Guerreiro, Marina Peres Bainy, Patrícia de Azevedo Duarte Hardt; Hospital Vila da Serra–Instituto Materno Infantil de Minas Gerais: Hugo Correa de Andrade Urbano, Camila Amurim de Souza; Hospital Vivalle: Claudia Mangini, Fernando Jose da Silva Ramos, Luany Pereira de Araújo; Instituto Nacional de Infectologia Evandro Chagas/Fiocruz: André Miguel Japiassú, Denise Machado Medeiros, Michele Fernanda Borges da Silva; Irmandade da Santa Casa de Misericórdia de São Paulo: Fabiano Hirata, Roberto Marco, Elzo Peixoto; Irmandade do Hospital Nossa Senhora das Dores: Marcio Luiz Fortuna Esmeraldo, Leide Aparecida Damásio Pereira, Raquel Carvalho Leal; Irmandade Santa Casa de Misericórdia de Maringá: Paulo Roberto Aranha Torres, Maricy Morbin Torres; Santa Casa de Belo Horizonte: Mara Rubia de Moura, Claudio Dornas de Oliveira, Andressa Siuves Gonçalves Moreira, Brisa Emanuelle Silva Ferreira, Carolina Leticia dos Santos Cruz, Patrícia Moreira Soares, Paulo Cesar Correia, Lorena Lina Silva Almeida; Santa Casa de Caridade de Diamantina: Marcelo Ferreira Sousa, Andrey Antonio Santiago Vial, Marcia Maria Ferreira Souza; Santa Casa de Misericórdia de Anapolis: Diogo Quintana, Ana Cecilia Barbosa Guimaraes, Maria Dolores Menezes Diniz; Santa Casa de Misericórdia de Feira de Santana: Paulo Henrique Panelli Ferreira, Rosa Maria Rios Santana Cordeiro, Murilo Oliveira da Cunha Mendes; Santa Casa de Misericórdia de Guaxupé: Sergio Oliveira de Lima, Silvia Aparecida Bezerra, Aurélia Baquiao dos Reis da Silva; Santa Casa de Misericórdia de Pelotas: Cristiano Correa Batista, Thaís Neumann, Rafael Olivé Leite; Santa Casa de Misericórdia de Porto Alegre: Thiago Costa Lisboa, Martha Hädrich, Edison Moraes Rodrigues Filho; Santa Casa de Misericórdia de São Joao Del Rei: Jorge Luiz da Rocha Paranhos, Carlos Henrique Nascimento dos Santos, Hélia Cristina de Souza; Sociedade Beneficente Hospitalar Maravilha: Robson Viera de Souza, Jose Luís Toribio Cuadra, Jonas Spanholi; Unidade de Emergência, Hospital das Clinicas de Ribeirão Preto–FMRP-USP: Marcos de Carvalho Borges, Wilson Jose Lovato, Tania Mara Gomes, Luís Artur Mauro Witzel Machado; Hospital de Clinicas–Universidade Estadual de Campinas: Thiago Martins Santos, Marco Antonio de Carvalho Filho, Karina Aparecida Garcia Bernardes; Vila Velha Hospital: Jose Roberto Pereira Santos, Aline Esteves Mautoni Queiroga Liparizi, Patrícia Venturim Lana; Vitoria Apart Hospital: Claudio Piras, Luiz Virgílio Nespoli, Aparecida Silva Taliule.
Disclaimer: Dr Angus, JAMA Associate Editor, had no role in the evaluation of or decision to publish this article.
Additional Contributions: We thank the Brazilian Association of Intensive Care Medicine (AMIB) for facilitating the invitations to Brazilian intensivists. We thank all multidisciplinary teams in participant centers for their enthusiastic support and for following the study interventions.