The effect of the physiological reserve surrogate characteristics on non-airway extubation failure in patients with pneumonia with high Burns Wean Assessment Program scores

2.1 Study design

A retrospective observational cohort study was conducted between January 2019 and January 2020 at King Saud University affiliated teaching hospitals to identify any physiological or biochemical parameters as independent predictors of early extubation failure in patients with pneumonia. The study design was approved by the Institutional Review Board of the health science colleges at King Saud University (E-19-4221). Records were collected for statistical analysis through an electronic health records system. Samples size (n), considering population size (N) and a margin of error (e), was calculated with Solvin’s formula: n = N/(1 + Ne${}^2$).

2.2 Patient population and inclusion and exclusion criteria

We included all adult (aged $>$19 years) patients with pneumonia requiring IMV who scored $\ge$50 on the Burns Wean Assessment Program (BWAP) which is considered the cutoff limit of successful weaning [16]. BWAP, which was used and well validated in long-term mechanically ventilated patients that was defined as patients requiring IMV more than 3 days, was also used in this study for the patients with pneumonia who required intubation more than 72 hours before screening the patients to help using this validated score. BWAP was validated as good predictive tool if the score is more than 50. Therefore, we excluded all patients scoring $<$50 on the BWAP or those with acute events not related to the original pneumonia diagnosis or the complications associated with it that led to reintubation (e.g., Myocardial infarction (MI), new sepsis, hospital acquired pneumonia, ventilator associated pneumonia, upper gastrointestinal bleeding, and others). We also excluded those requiring non-invasive ventilation (NIV) after extubation more than 24 hours since it could potentially delay the need for intubation beyond the 72-hour period, those with unplanned self-extubation, those who underwent tracheostomy from the first intubation, or those who underwent palliative one-way extubation.

2.3 Non-airway extubation failure definition and criteria

Due to the variability of definition of the extubation failure in the literature and to minimize selection bias and to help enrollment of target patients centered around the physiologic reserve that is necessary to sustain the liberation from IMV for 72 hours and beyond, the data collectors adapted a stepwise approach to help identifying patients with non-airway extubation failure due the inadequate physiologic reserve resulting from the original pneumonia diagnosis [20]. First, the airway extubation failure due to post-extubation laryngeal edema, which is usually happened minutes to hours after the extubation trial, have been excluded. Second, the non-airway extubation failure due to an acute events post extubation not related to the initial pneumonia and the associated complications have been excluded on the chart review. Third, we include the distinct group of patients who had to be reintubated because of the general weakness which was assessed subjectively on the patients charts indicating frailty, which is reflective of inadequate physiologic reserve which contribute to the lack of sustainability from IMV. Successful extubation was defined as the removal of an endotracheal tube for more than 72 hours [8].

2.4 Extubation procedures

A spontaneous breathing trial was conducted in case the causes of mechanical ventilation were relieved or alleviated, PaO${}_2$/FiO${}_2$$>$ 150–200 at FiO${}_2$$\le$ 0.4–0.5 and PEEP $\le$ 5–8 cm H${}_2$O, stable hemodynamic, capability of spontaneous breathing patients were extubated in absence of (1) diaphoresis, anxiety, agitation, depressed mental status; (2) increase in accessory muscle activity; (3) PaO${}_2$$\le$ 50–60 mmHg at FiO${}_2$$\ge$ 0.5 or SaO${}_2$$<$ 90%; (4) PaCO${}_2$$>$ 50 mmHg or increase in PaCO${}_2$$>$ 8 mmHg; (5) pH $<$ 7.32 or decrease in pH of $\ge$0.07; (6) a respiratory rate $>$35 beats per minute or increased by $\ge$50%; (7) rapid shallow breathing index score $>$105 beats per minute; (8) heart rate $>$140 beats per minute or increased by $\ge$20%; (9) systolic blood pressure $<$90 mmHg or $>$180 mmHg or increased by $\ge$20%; and (10) cardiac arrhythmias [21, 22].

2.5 Data collection

We collected baseline characteristics including age, sex, body mass index (BMI), history of diabetes mellitus, hypertension, history of chronic lung disease, and history of cardiac disease. Acute factors included admission APACHE II score, type of pneumonia, CURB-65, type of respiratory failure preceding the intubation, concomitant viral illness, and use of steroids. We also included biochemical laboratory measurements at admission including serum sodium (Na${}^{+}$), chloride (Cl${}^{-}$), bicarbonate (HCO${}_3$${}^{-}$) albumin, BUN, phosphorus (PO${}_3$${}^{-}$), and corrected calcium (Ca${}^{++}$). We also included biochemical lab abnormalities upon admission including hypoalbuminemia, hypophosphatemia, and corrected hypocalcemia.

2.6 Statistical analyses

Means (M) and standard deviations (SDs) were used to describe the continuous measured variables and the frequency and percentages for categorical variables. The Kolmogorov-Smirnov test of normality and histograms were used to assess continuous measured variables, and Levene’s test was used to assess the homogeneity of variance statistical assumption for continuous variables. Independent samples t-tests were conducted to assess the significant mean differences between continuous variables across the levels of binary dichotomous variables, and the Mann-Whitney U non-parametric test was used to assess the mean differences of length of stay and mechanical ventilation durations across the levels of reintubation. Chi-squared tests were used to assess the correlations between categorically measured variables. Multivariate binary logistic regression analyses were conducted to explore the association between patients’ sociodemographic and clinical characteristics at admission as well as their past medical history against their odds of failing the extubation process in the ICU. The dependent variable of interest was extubation failure, and the association between these risk factors and reintubation characteristics was expressed with odds ratios (ORs) with 95% confidence intervals. SPSS version 20 (SPSS Inc., Chicago, IL, USA) was utilized for data analyses, and EXCEL spreadsheets were employed to create figures and depictions. Significance was set at p $<$ 0.05.

3.1 Baseline characteristics

Table 1 displays patients’ sociodemographic and medical characteristics. Most of the final cohorts were males (60.7%), aged between 44 and 64 years old (55.2%). Patients with history of DM, HTN, chronic lung disease, and cardiac disease were 59.1%, 47.3%, 21.7%, and 25.6%, respectively. Patients’ BMI was also computed and classified, and obese patients were 38.3% of the enrolled patients.

3.2 Acute clinical factors

The mean APACHE II score on admission was 19.28 $\pm$ 8.35, and in-hospital mortality was 16.9% (Table 2). The results also indicated that 14.1% of the patients had presented with aspiration pneumonia, while 55% presented with community-acquired pneumonia and 31% had hospital-acquired pneumonia. The mean CURB-65 was 4.1 out of 5 (SD = 0.67). More than half (55.6%) of the admitted patients had a CURB65 of 4. Further, 60.7% of the patients were intubated owing to hypoxemic respiratory failure, while 39.3% had developed hypercapnic respiratory failure. Moreover, 31.3% had escalating vasopressor requirements beyond the 24-hour period. However, considering all patients scored $\ge$50 points on the BWAP tool, which indicated a high likelihood of successful weaning from the mechanical ventilator and extubation, the analysis revealed that 37.7% of the patients had failed extubation within 72 hours and the median (IQR) mechanical ventilation duration was 7 (5–12) days. Nonetheless, the median (IQR) length of ICU and hospital stay were 12 (6–23), and 26 (17–37) days, repectively. In addition, 16.9% of the sample had received steroid therapy upon admission.

3.3 Biochemical factors

The mean serum Na${}^{+}$ among patients upon arrival was 138.92 $\pm$ 6.2 mmol/L, and the mean serum Cl${}^{-}$ was 102.97 $\pm$ 7.3 mmol/L. The mean serum HCO${}_3$${}^{-}$ was 25.67 $\pm$ 4.82 mmol/L, and 59.7% of the patients presented with a BUN that was higher than 19 mg/dL. Moreover, the mean serum albumin was 27.37 $\pm$ 7.1 g/L, and 79.6% had hypoalbuminemia (albumin level $<$25 g/L) upon arrival at the ICU. The mean serum PO${}_4$${}^{-}$ was 1.05 $\pm$ 0.43 mmol/L, and 35% of the patients presented with hypophosphatemia (PO${}_4$${}^{-}$$<$ 0.8 mmol/L) at admission. The mean corrected Ca${}^{++}$ was 2.11 $\pm$ 0.2 mmol/L, and 18.2% were hypercalcemic (Ca${}^{++}$$<$ 2 mmol/L). The hematological lab revealed that platelet count in this cohort was 274 $\pm$ 140.04 10${}^3$/$\mu$L and persistent hyperlactatemia ($>$2 mmol/L) beyond 24 hours in 8.62%.

3.4 Bivariate and multivariate regression analysis

To clarify why the patients with pneumonia failed the extubation process, we explored patients’ extubation success/failure and any significant associations with their measured sociodemographic or medical characteristics (Table 3). There was no significant difference in the mean APACHE II score (SD) between who failed extubation compared to those who had successful extubation [20.3 (8.45) vs. 18.6 (7.27); p = 0.064]. There was also no significant difference in in-hospital mortality between both groups [14.4% vs. 18.5% (p = 0.353)]. However, there was a significant association between patients’ sex and reintubation, as men were significantly more likely to have extubation failure than women (p = 0.001). Furthermore, the patients who failed the extubation were significantly older with the mean age of 65.9 $\pm$ 8.85 years vs. 62.26 $\pm$ 11.1 years for those who had a successful extubation; p = 0.003. However, patients’ age group was not significantly correlated with their reintubation outcome (p = 0.403). BMI and patients’ medical history of DM, and HTN were not correlated with extubation failure risk.

Patients with a history of chronic lung disease or cardiac disease were significantly more likely to have failed the extubation process compared to their counterparts (p = 0.037 and p $<$ 0.001, respectively). Unsurprisingly, patients who underwent reintubation had significantly longer median (IQR) duration on mechanical ventilation before the attempt of extubation compared to those who did not have extubation failure [12 (6–13) days vs. 6 (4–9) days; p $<$ 0.001)]. Moreover, patients who failed to be extubated stayed significantly longer median duration (IQR) in the ICU than those whose endotracheal tube was successfully removed [19 (11–37) days vs. 11 (5–16) days; p $<$ 0.001] and longer median hospital stay [50 (28–122) days vs.18 (16–26) days; p $<$ 0.001] (Table 3).

Patients who failed the extubation process had significantly higher mean serum Na${}^{+}$ (140.73 $\pm$ 4.80) and Cl${}^{-}$ (104.21 $\pm$ 6.73) at admission than those who were successfully extubated (137.83 $\pm$ 6.7, p $<$ 0.001; 102.22 $\pm$ 7.54, p = 0.019, respectively). Patients who presented with a high BUN ($>$19 g/dL) at admission were also significantly more likely to have extubation failure compared to those with a normal BUN (p $<$ 0.001). No differences were found concerning serum albumin, hypoalbuminemia, persistent hyperlactatemia, platelet count, and serum PO${}_4$${}^{-}$ (p = 0.731, p = 0.971, p = 0.449, p = 0.546, and p = 0.709, respectively); however, patients who presented with hypophosphatemia were significantly more likely to have failed the extubation process than those patients whose serum PO${}_4$${}^{-}$ was $>$0.8 mmol/L (p $<$ 0.001). Patients failed the extubation had also significantly lower mean serum Ca${}^{++}$ levels (2.08 $\pm$ 0.18) than patients who had successful extubation (2.13 $\pm$ 0.21, p = 0.028); however, hypocalcemic patients at admission (with serum Ca${}^{++}$$<$ 2 mmol/L) were significantly more likely to have failed the extubation process compared to those with a serum Ca${}^{++}$$>$ 2 mmol/L at admission (p = 0.028; Table 3).

Unsurprisingly, patients who failed the extubation presented with higher mean CURB65 scores (4.36 $\pm$ 0.61) compared to their counterparts (3.88 $\pm$ 0.63; p $<$ 0.001). Further, patients who scored 5 on the CURB65 were significantly more likely to have failed the extubation process compared to those who scored 3 or 4 (p $<$ 0.001). Furthermore, those who failed the extubation tend to have escalating vasopressor requirements beyond 24 hours (44.1% vs. 23.6%; p = 0.0002). Patients who presented with hospital-acquired pneumonia were also significantly more likely to have failed extubation compared to those who presented with community- or aspiration-acquired pneumonia (p $<$ 0.001; Table 3). Furthermore, the type of respiratory failure and the concomitant viral illness as well as the use of steroids did not correlate significantly with the risk of extubation failure.

The multivariate logistic binary regression analysis was significant overall, denoting that one or more of the tested predictors had a significant association with the extubation failure (Table 4). Patients’ sex correlated significantly with their odds of failing the extubation process (p = 0.033); although, men were significantly less likely to be re-intubated than women and patients’ age did not converge significantly with their odds of failing the extubation. Further, the analysis model indicated that the interaction between patients age and sex suggested that older men compared to younger women were significantly more likely to fail extubation (OR = 1.21; 95% CI, 1.03–1.22; p = 0.006). Fig. 2 reveals that after the age of 44 years, men were significantly more likely than women to have required reintubation; conversely, younger women aged $<$44 years were significantly more likely to require an endotracheal reintubation than young men aged $<$44 years. The analysis model also indicated that patients with DM were significantly more likely to have extubation failure (OR = 3.05; 95% CI, 1.22–7.62; p = 0.017).

Fig. 2.The association between age-sex interaction and the probability of extubation failure among patients with pneumonia.

In addition, patients with a history of cardiac disease were significantly more likely to have extubation failure (OR = 2.64; 95% CI, 1.11–6.3; p = 0.028). The duration of mechanical ventilation also converged significantly on their odds of requiring reintubation. For each additional day on mechanical ventilation, the odds of patients failing the extubation process increased by 1.35 times (i.e., was 35% times higher; p $<$ 0.001). Moreover, Na${}^{+}$ level at admission converged significantly and positively with the odds of extubation failure (p $<$ 0.001). Nonetheless, chloride level at admission did not correlate significantly with patients’ odds of failing the extubation process (p = 0.13).

Patients with high BUN scores ($>$19 mg/dL) were significantly more likely to fail the extubation (OR = 2.62; 95% CI, 1.08–6.33; p = 0.033), and hypophosphatemia at admission significantly predicted higher odds of failing extubation (OR = 2.39; 95% CI, 1.12–5.08; p = 0.024). Nonetheless, patients’ CURB-65 score correlated significantly and positively with their odds of failing the extubation process: for every increase on the CURB-65, the odds of failing the extubation trial increased by 3.1 times (OR = 3.05; 95% CI, 1.43–6.52; p = 0.0040) (Fig. 3). Patients with hypercapnic respiratory failure were more likely (OR = 1.99) to be re-intubated in the ICU compared to those who had hypoxemic respiratory failure; however, the difference was not significant (p = 0.090). Lastly, patients who had hospital-acquired pneumonia were significantly more likely to have extubation failure (OR = 5.90; 95% CI, 2.49–13.98; Fig. 4) compared to the patients who presented with community- and aspiration-acquired pneumonia (p $<$ 0.001).

Fig. 3.The association between the severity of pneumonia using CURB-65 and the probability of extubation failure. CURB-65, confusion, urea $>$7 mmol⋅L-1, respiratory rate $\ge$30⋅min-1, low blood pressure, and age $\ge$65 years.

Fig. 4.The association between the type of pneumonia and the model predicted probability of extubation failure. *Hospital-acquired pneumonia carries a higher risk of non-airway extubation failure higher than community-acquired pneumonia and aspiration pneumonia.