An observational, prospective, multicenter and national study conducted in 8 ICU from eight academic hospitals from Mexico including consecutive adult patients admitted to the ICU due to acute hypoxemic respiratory failure secondary to SARS-CoV-2 pneumonia between March 8, 2020–February 28, 2021. Patients with a acquired neuromuscular disease were excluded. The diagnosis was confirmed by positive result of real time reverse transcriptase-polymerase chain reaction (RT-PCR) testing of a nasopharyngeal swab.

The presence of frailty was specifically addressed. Frailty was defined according to the FRAIL scale (FS) definition and was obtained at admission to the ICU and 6-month after discharge from the ICU, and included five-item12: fatigue, resistance, ambulation, illnesses, and weight loss adapted and validated for Spanish language and Mexican population13 (eFig. 1 in Supplementary Appendix). The information was obtained from standardized interviews with patients or surrogate respondents in cases of inability to answer due to the need for invasive mechanical ventilation or sedation.

The FRAIL scale is a subjective judgment-based screening tool for frailty that has been proved to be valid, reliable, simple to perform, validated and successfully adapted to Mexican Spanish. For the purpose of the analysis, the categorization of this scale was as follows: health status zero point was considered as robust (non-frail), with to two points were considered pre-frail, and when three or more points were obtained, the participant was cataloged as frail. The follow-up to monitor frailty was performed at 6 months after discharge hospital through a telephone interview. Participants with missing data for one or more frailty criteria were excluded.

The primary endpoint was the frailty status assessed at 6-months after hospital discharge in ICU survivors. The secondary outcomes were the development of complications and organ failure during the ICU stay, ICU mortality, and 6-month mortality rates.

Study personnel at each site collected data by manual review of medical records and used a standardized case report form to enter data into a secure online database.

The clinical variables recorded were: age, sex, height, weight, severity at admission estimated by Simplified Acute Physiology Score (SAPS2) which ranges from 0 [lower severity] to 150 [higher severity], comorbidities (asthma, COPD, other chronic pulmonary disease, chronic renal disease, diabetes, hypertension, dyslipidemia, obesity, chronic cardiac failure, ischemic cardiomyopathy, permanent atrial fibrillation, cirrhosis, ischemic stroke, immunosupression, hematological neoplasia, autoimmune disease, HIV infection), chronic medication (angiotensin converting enzyme inhibitors, angiotensin receptor blockers, statins, antiplatelet therapy, oral anticoagulants, steroids), mode of respiratory support (high flow nasal cannula therapy and/or noninvasive positive pressure ventilation [NPPV]), and compassionate medication received [antivirals (lopinavir-ritonavir, hydroxychloroquine, remdesivir) and immunomodulatory agents (interleukin-6 receptor antagonists, Janus kinase inhibitor, and corticosteroids)], length of hospital stay prior to ICU admission, date of intubation, ventilatory settings within first week of mechanical ventilation, daily arterial blood gases, concentrations of plasma-based and serum-based biomarkers drawn within 7 days of ICU admission, including high-sensitivity C-reactive protein, d-dimer, ferritin, high-sensitivity troponin, procalcitonin, and IL-6, use of adjuvant therapies for acute respiratory failure (neuromuscular blocking agents, inhaled pulmonary vasodilators, prone-positioning ventilation, and extracorporeal membrane oxygenation), antiviral agents, and immunomodulatory agents, complications and organ dysfunction during the ICU stay.

The subjects were followed up to 6 months to assess for frailty and status outcomes.

The study was approved by the Institutional Ethical Board of each participating site, and waived the requirement for informed consent from individual patients considering the study design as minimal-risk research using data collected for routine clinical practice and ongoing public health emergency. All data except dates were deidentified.

The study was registered in ClinicalTrials.gov Identifier: NCT04379258. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement guidelines for observational cohort studies.14

The main outcome was the changes in frailty score 6 months after hospital discharge and to estimate the variables associated to increase of the FS.

Descriptive statistics were expressed by quantitative data as mean with standard deviation (SD) or median (IQR), and qualitative data as absolute and relative frequencies (proportions, %). To test for statistical association between baseline FS states (ordinal variable) and participant characteristics, the ANOVA or Willcoxon test were used for continuous and ordinal variables as required, and the Chi-squared test for trend test for dichotomous variables.

For the visualization of the longitudinal trajectories of the three frailty states components, alluvial charts were created using the R alluvial package. In each alluvial plot, the height of the stacked bars at each wave (which represent whether participants’ status for the given frailty state or component was yes, no, missing or died) is proportional to the number of participants identified as belonging to this state at each wave. The thickness of the streams connecting the stacked bars between waves are proportional to the number of participants who have the state identified by both ends of the stream.

To evaluate the predictive variables during ICU admission associated with frailty (estimated from robust category to frail or prefrail categories, and from prefrail category to frail), a multivariate analysis was performed, using logistic regression models and taking as dependent variables the development of frailty at 6 months after discharge. In turn, the independent variables were defined as those variables that were identified in the univariate analysis as being associated to the outcomes of interest (frailty) with a value P < .10, as well as those variables of clinical interest. Since the number of variables to be entered in the multivariate model were limited by the prevalence of the outcome, we were only able to include one to two variables. For this reason, fitting of the model (maximum model) was limited to: sex, immunesupression, chronic renal disease, cardiopathy, SAPS2 score, basal assessment of frailty, duration (in days) of hospital stay prior to ICU admission, need of invasive ventilatory support, use of neuromuscular blockers, and prone positioning.

Odds ratios (ORs) and 95% confidence intervals (CIs) for the estimated covariate effects of age, sex, SAPS2, tracheotomy and duration of ventilatory support were obtained. ORs were considered significant when their CIs did not include 1.

Statistical significance was considered at P < .05. Except for the alluvial graph, analyses were performed using Stata version 17.0 (StataCorp LLC).

In the period of study 524 patients were enrolled. From those, 194 patients (37%) who died during their hospital stay were excluded from the study. From the cohort of 330 survivors, 196 patients had an evaluation for frailty at admission in the ICU, and finally included in this analysis (Fig. 1). Baseline characteristics are showed in Table 1. Of these, 139 (71%) were male, 17 (9%) were older than 75 years, 65% received invasive mechanical ventilation and the median duration of ventilatory support was 8 days (IQR, 6–12 days).

Figure 1.

Flow chart of patients.

(0.1MB).

The complications, during ICU stay and outcomes are shown in Table 2. The characteristics of patients that were not included in the analysis are shown in eTable 1 and eTable 2 in Supplementary Appendix.

At ICU admission the baseline frailty was as follows: 181 patients (92%; 95% confidence interval [CI] 88%–96%) were classified as non-frail, 14 patients (7%; 95% CI 4%–12%) as pre-frail and 1 patient (0.5%; 95% CI 0.01%–2%) as frail. The prevalence of frailty status for each age category increased with age (robust was defined in 30% of older than 60 years old, prefrail affected 35% of patients older than 60 years, and frail in 100% of patients older than 60 years) (eTable 3, in Supplementary Appendix).

The alluvial plots are shown in Fig. 2. The cumulative proportion of frailty increased over time. At 6-months after discharge from the ICU, 164 patients were evaluated for frailty: 40 patients (24%) were classified as robust, 67 patients (41%) as pre-frail and 57 patients (35%) as frail (Fig. 2). In total, 114 patients (69%) worsened their condition after ICU admission.

Figure 2.

Alluvial chart of the longitudinal transitions of FS states in survivors within 6 months after ICU discharge. FS: frailty scale.

(0.23MB).

Table 3 shows the univariate analysis of the comparison between patients who showed deterioration in frailty score at six months (n = 114) versus those who did not have deterioration after ICU treatment (n = 50). Significant differences were found in the severity scale at ICU admission (SAPS2), in the treatment used (propofol infusion, continuous analgesics infusion and neuromuscular blockers use), and in the CPR levels at ICU admission.

The adjusted analysis showed that baseline frailty was not associated with ICU mortality (OR 0.43; 95% CI 0.11–1.59, P = .207).

After adjustment of basal covariates and those related with ICU treatment, the need of invasive mechanical ventilation was independently associated with frailty at 6 months of follow-up (Odds Ratio [OR] 3.70, 95% confidence interval 1.40–9.81, P = .008) (Fig. 3).

Figure 3.

Forest plot of odds ratios for frailty at day 6 months after multivariate logistic regression in survivor patients with COVID-19 after ICU treatment (n = 196). SAPS2, Simplified Acute Physiologic Score; CI: confidence interval.

(0.16MB).

The sensitivity analysis that included the baseline frailty assessment did not show differences in the predictive variables associated with the deterioration of frailty at 6 month after hospital discharge, and was associated as well independently with the impairment of frailty (OR 0.03; 95% CI 0.01−0.18, P < .01) (eFig. 2 in Supplementary Appendix).