Patient Characterization Panel to Improve Satisfaction With the Intensive Care Unit Stay (CONNECT)

Study design:

Study CONNECT is a prospective, open-label, multicenter clinical trial with cluster-level randomization, using a stepped-wedge design. The randomization unit is the Intensive Care Unit (cluster). All clusters begin in the control phase and cross over to the intervention in a staggered, randomly determined sequence, one cluster every two months, with no transition period between phases. The study runs for 14 months, comprising seven two-month periods (including a baseline period in which all Intensive Care Units (ICUs) remain in the control condition). Each cluster's allocation sequence is disclosed to researchers and staff one week before its respective phase transition. Blinding of patients, family members, and care staff is not feasible given the nature of the intervention; however, personnel responsible for administering outcome questionnaires are blinded to the period the patient was admitted to the ICU (i.e., control vs. intervention period) and to the details of each patient's ICU stay, to minimize information bias.

Settings:

The trial is conducted in six ICUs across two hospitals in Brazil: four ICUs at Clinical Hospital of the University of São Paulo Medical School (HC-FMUSP), São Paulo, a public, academic, quaternary-care hospital; and two ICUs at Monte Klinikum Hospital, Fortaleza, a private, tertiary, non-academic hospital. This combination was selected to ensure diversity in hospital type, patient and family sociodemographic profiles, clinical case mix, and care team organization. All participating ICUs use closed (single-patient) rooms.

Intervention:

The intervention is the "Patient as a Person" Characterization Panel (PCPP) named "Come to Know me", a structured bedside instrument inspired by the "Get to Know Me" poster developed at Massachusetts General Hospital. The PCPP comprises 19 fields completed collaboratively by the multidisciplinary care team (medicine, nursing, and physical therapy) together with the patient and family members, and may include photographs. The panel documents the patient's identity, personal preferences, biographical details, and what matters most to them outside the hospital, making this information visible and accessible to the entire care team. The panel is displayed near the patient's bedside in a location that preserves data privacy. Implementation is preceded by structured staff training delivered prior to each ICU's phase transition, supplemented by monthly retraining sessions and printed and online instructional materials throughout the intervention period.

Control arm:

In the control phase, participating ICUs maintain standard care practices. No version of the PCPP or related training is introduced.

Eligibility:

Patients are eligible if they are 18 years of age or older and have been admitted to a participating ICU for more than 48 hours. Their family members or legally designated representatives, also 18 years of age or older, are eligible as co-participants. Exclusion criteria are pregnancy, incarceration, and inability to communicate in Portuguese. Written informed consent is obtained from family members after 48 hours of ICU admission, and from patients once they are alert and oriented after 48 hours of ICU admission. In case patients are transferred between study ICUs or readmitted, they are assigned to the arm/period of their index ICU admission (intention-to-treat).

Outcomes and data collection:

The primary outcome is family satisfaction with the ICU stay, assessed using the Family Satisfaction with Care in the Intensive Care Unit Questionnaire (FS-ICU 24R questionnaire), a validated instrument adapted for use in Brazil. Secondary outcomes include: patient-perceived quality of communication (Quality of Communication Questionnaire, QOC); ICU staff perceptions of the PCPP, assessed via a quantitative-qualitative questionnaire developed by the study team; and symptoms of anxiety and depression in patients (where feasible) and family members, assessed using the Hospital Anxiety and Depression Scale (HADS) at three months after ICU discharge. All questionnaires are administered in Brazilian Portuguese by a blinded assessor. Sociodemographic and clinical data, including age, sex, marital status, educational level, self-reported race/color, religion, admission diagnosis, ICU length of stay, ICU and hospital outcome, decisions regarding withholding or withdrawing life-sustaining treatment, and Simplified Acute Physiology Score 3 (SAPS 3) severity score, are collected via REDCap (version 14.5.21) for sample characterization.

Sample size:

The sample size was calculated using the steppedwedge command in Stata version 13.1, assuming a two-sided alpha of 5%, 80% power, a mean of 23 eligible patients per two-month period per ICU, a baseline mean FS-ICU 24R score of 75 points (SD 30), an expected absolute increase of 7.5 points (corresponding to a 10% relative increase, from 75 to 82.5 points), and a coefficient of variation (k) of 0.25. This yields an estimated total of 966 patients (6 ICUs × 7 periods × 23 patients per period). As is standard in stepped-wedge designs, study completion is determined by the elapsed time across the seven predefined periods, not by reaching a specific patient count.

Statistical analysis:

The primary analysis follows the intention-to-treat principle, with two-sided tests and a 5% significance level. The primary outcome (FS-ICU 24R questionnaire), analyzed at the individual patient level, will be estimated using a generalized linear mixed model (GLMM), with intervention status, calendar time period, and pre-specified adjustment covariates (SAPS 3 score, clinical versus surgical profile, ICU region, and number of ICU beds) as fixed effects. The error distribution will be selected based on best fit to the data (Poisson, negative binomial, or alternative distributions). Random effects include a per-cluster intercept, capturing between-ICU variability, and, where warranted, an ICU-by-period interaction term, capturing intra-cluster temporal correlation. If a substantial proportion of primary outcome data are missing, multiple imputation by chained equations (MICE) will be applied under the missing-at-random (MAR) assumption. Pre-specified subgroup analyses will assess potential effect modification by patient severity (SAPS 3), ICU outcome (discharge versus death), and presence or absence of a terminal or advanced chronic illness, using the same GLMM with an intervention-by-subgroup interaction term. A statistical analysis plan (SAP) will be finalized and submitted for publication before database lock.