Clinical outcomes of acute respiratory distress syndrome in a university hospital

Between 2010 and 2014, 266 of 4,090 (6.5%) patients admitted in our medical ICUs met the Berlin criteria for ARDS diagnosis. Approximately 11.7%, 44.4%, and 44.0% of the patients fulfilled mild, moderate, and severe ARDS criteria, respectively. The severity of ARDS correlated with APACHE II score (r² = 0.235, P < 0.001). Demographic data of the patients are summarized in Table 1.

The majority of the ARDS patients had pneumonia and were assisted with pressure-controlled ventilation with the mean peak inspiratory pressure (PIP) and PEEP on the first day of MV of 31 ± 6 and 10 ± 4 cm H₂O, respectively. The plateau pressure (Pplat) was recorded in only 4.1% of the patients. The mean TV on the first day was 8.56 ± 2.48 mL/kg ideal body weight (IBW), and the average TV during the first 3 days of MV was 8.81 ± 2.39 mL/kg IBW. Approximately 55.4% of the patients had TV on the first day of MV of >8 mL/kg IBW and 60.9% of the patients had average TV during the first 3 days of MV of >8 mL/kg IBW.

Neuromuscular blockades and sedative drugs were prescribed to 35.3% and 86.09% of the patients, respectively. Adjunctive interventions including prone position, recruitment maneuver, and rescued modes such as airway pressure release ventilation (APRV) and HFOV were applied to patients with moderate and severe ARDS (Table 1). Noticeably, recruitment maneuver was performed in a half of the patients, especially in the patients with moderate to severe ARDS. However, the incidence of pneumothorax was not different between the patients who received recruitment maneuver (79.7%) and who did not receive recruitment maneuver (82%).

Overall, the crude 7-day mortality was 31.1%, associated with age, ARDS severity, APACHE II score, PaO₂/FiO₂ ratio on day 1, mean PaO₂/FiO₂ ratio during the first 3 days, fluid balance on day 1 and cumulative fluid balance during the first 3 and 7 days, and average daily calories during the first week (Table 2). The crude 28-day mortality was 68.9%, associated with age, ARDS severity, APACHE II score, PaO₂/FiO₂ ratio on day 1, mean PaO₂/FiO₂ ratio during the first 3 days, TV/kg IBW on day 1, PIP on day 1, neuromuscular blockade use, fluid balance on day 1 and cumulative fluid balance during the first 3 and 7 days, and average daily calories during the first week (Table 2). After adjusting for confounding factors, the mean PaO₂/FiO₂ ratio during the first 3 days, cumulative fluid balance on day 3, and average daily calories during the first week were independent predictors for 7-day mortality, whereas APACHE II score, fluid balance on day 1, cumulative fluid balance during the first 7 days, and average daily calories during the first week were independent predictors for 28-day mortality (Table 3). Furthermore, the patients who received average daily calories during the first week of <60% of estimated daily energy requirement had a greater risk of death (OR = 3.28, 95% CI 1.79–6.01, P < 0.001). Moreover, interestingly, the risk of death within 28 days increased with increasing 7-day cumulative fluid balance as shown in Figure 1.

Figure 1

This study demonstrated that the 28-day mortality of the ARDS patients admitted in our medical ICUs was high, compared with the mortalities in other studies [15, 16, 17, 18, 19] (Table 4). We postulated that the higher mortality rate resulted from limited ICU beds, delayed ICU admission, delayed sedative and neuromuscular blockade administration, and shortage of medical staff. Our mortality rate tended to be increasing with the greater ARDS severity; however, it was not significant. It might be explained that there were multiple factors affecting the mortality rate such as ventilator-induced lung injury, ventilator-associated pneumonia, and positive cumulative fluid balance. Noticeably, 42.3% of our patients received TV greater than 8 mL/kg IBW on the first day, compared with 30% of the patients in the PRoVENT study [19]. Moreover, there were limited data of the Pplat in our study (4%), compared with 17.2% in ALIEN [18], 40% in LUNG SAFE [15], 41% in PRoVENT [19], and 90.7% in APRONET [20]. These findings might indicate that the physicians less concerned about ventilator-induced lung injury. Furthermore, the prevalence of prone position in our severe ARDS patients was only 4%, compared with 16% in LUNG SAFE [15] and 32.9% in APRONET [20]. Prone position was infrequently used in our study because of different study periods. The PROSEVA [6] reported the mortality benefit of prone position in 2013, but we retrospectively reviewed the data between 2010 and 2014. In addition, the prevalence of the use of neuromuscular blocking agent was 35.3% in our study, higher than LUNG SAFE (21.7%) [15] due to the higher proportion of severe ARDS patients in our study. Despite the greater use of neuromuscular blockade, the mortality rate was higher, which might be from inadequate doses of this medication and no ventilator asynchrony monitoring. Recruitment maneuver was performed to 48.12% of the patients, which was quite high, compared with 20.9% in the LUNG SAFE [15]. Despite the fact that the recruitment maneuver potentially increased the risk of barotrauma and death [21, 22], our study did not show the association between the procedure and the complications. Finally, similar to other studies, there were low prevalence of HFOV, APRV, and extracorporeal membrane oxygenation (ECMO) use, which less affected the ARDS outcomes [7, 8, 23, 24, 25, 26, 27, 28, 29].

According to the ventilator complications, barotrauma was not different between ARDS survivors and non-survivors. Nevertheless, ventilator-associated pneumonia was higher in the non-survivors group (18.4% vs 13.4%).

Besides, most of the patients had positive cumulative fluid balance, which contributed to the significantly greater mortality in our study (Figure 1). Compared with FACTT [10], our ARDS patients had cumulative fluid balance over 7 days close to the fluid balance of the patients in the liberal fluid management group (6,634.97 ± 6,718.06 mL in our study vs 6,992 ± 502 in liberal FACTT), who had worse clinical outcomes, including less ventilator-free days and deteriorated oxygenation index and lung injury score. Nevertheless, FACTT did not demonstrate the mortality risk from liberal fluid management. The different findings might be explained by the fact that there were limited data of hemodynamic status, which possibly confounded to the significantly higher mortality rate in our study.

Furthermore, our study showed that the patients receiving daily calories less than 60% of the estimated daily caloric need during the first week had higher mortality. This result was different from the findings in the EDEN study [30], which demonstrated that initial trophic enteral feeding for up to 6 days in patients with acute lung injury resulted in similar ventilator-free days, 60-day mortality, and infectious complications, compared with full enteral feeding. It might be explained that our patients with lower caloric intake possibly had more severe conditions such as hemodynamic instability or severe lung injury treated with prone position, which had an impact on mortality outcome.

Our study had limitations. Due to the retrospective design, the information was obtained from the recorded documents. Thus, there were some missing data such as height, body weight, Pplat, and driving pressure. In addition, there might be uncontrolled factors confounding to the mortality rate such as inadequate antibiotics and source control, inappropriate resuscitation of shock, or the different severity of comorbidities.

The 28-day mortality of ARDS was high. In addition, the average TV and fluid balance were greater than the protective limits. These findings indicated the potential improvement of ARDS outcomes in our hospital. The educational program focusing on ARDS management, follow-up, and feedback activities should be provided to our medical personnel.