Introduction

Mechanical ventilation (MV) is a lifesaving treatment modality considered a cornerstone to support critically ill patients and is used most commonly in Intensive Care Units (ICU) [1,2,3,4,5]. Globally, the number of patients requiring mechanical ventilation is increasing, specifically patients with Acute Respiratory Failure (ARF). It is provided for over 20 million patients worldwide [6], and among ICU-admitted patients, 35–50% require MV during their admission [7]. Millions of patients are discharged from hospitals after surviving critical illnesses each year. However, patients with ARF had a low survival rate, with a death rate of 67.2%. As a result, these patients are often supported with MV to improve gas exchange, reduce the work of breathing, prevent complications, and improve their outcome [8]. On the contrary, due to complications resulting from mechanical ventilation, up to 40% of patients died in the hospital because of its harmful effect on the patients’ lungs [9].

Despite the advancement in technology to support medical practices in the ICU, the death rate among mechanically ventilated patients remains high [2]. This mortality might be due to factors existed prior to the initiation of MV or developed during the course of treatment, as well as its complications [10]. Moreover, survivors of mechanically ventilated critically ill patients had short-term survival and a poor quality of life [11]. In addition to the high mortality rate, mechanically ventilated patients had faced various health-related problems [12, 13], such as emotional signs when they are stressed, mental health problems [14], physiological disturbances, and physical changes [15]. Besides, these critically ill patients on MV had increased healthcare expenditures since these patients consume extensive medical resources [16, 17], such as medical equipment, supplies, and personnel costs required to provide care [18].

In Ethiopia, many studies have been conducted to determine the mortality outcome of patients admitted to the ICU. However, studies on the survival of patients on mechanical ventilation admitted to the ICU were scarce. Furthermore, even available studies are not specific and cross-sectional in nature, which couldn’t consider the time variable and other clinically significant variables. In this regard, evidence about the time to death and its predictors helps to improve patient outcomes, assist clinicians in designing evidence-based strategies, decision-making process, and better resource allocation, and is used as an input for future studies. Therefore, this study aimed to determine the time to death and its predictors among mechanically ventilated patients admitted to the ICU.

Methods and materials

Study area

The study was conducted in West Amhara comprehensive specialized hospitals. There are five comprehensive specialized hospitals found in West Amhara. Among these, Felege-Hiwot, Tibebe Ghion, and Debre Markos comprehensive specialized hospitals were included in the study. Felege-Hiwot and Tibebe Gion comprehensive specialized hospitals are found in Bahir Dar, which is located 565 km northwest of Addis Ababa, the capital city of Ethiopia, whereas Debre Markos comprehensive specialized hospital is found in Debre Markos town, 299 km away from Addis Ababa. These hospitals provide inpatient and outpatient healthcare services for more than 15 million people. The ICUs are providing similar level of care equipped with mechanical ventilators, noninvasive hemodynamic monitoring devices, portable ultrasounds, Electrocardiograms (ECG), defibrillators, and infusion pumps. The total number of ICU beds in Felege Hiwot, Tibebe Ghion, and Debre Markos is 12, 9, and 4, respectively.

Study design and period

The institutional-based retrospective follow-up study design was conducted from January 1, 2020, to December 31, 2022, and the actual data collection period was December 1–31, 2022.

Population

All adult patients who were admitted to the ICU in West Amhara comprehensive specialized hospitals and received MV were the source population. All adult patients on MV who were admitted to the ICU in selected comprehensive specialized hospitals during the study period were the study population. Records of patients on MV during the study period and whose age ≥ 18 years were included. However, records of patients on MV without the outcome variable were excluded from the study.

Sample size determination

A survival sample size calculation power approach using Stata 14 software with cox proportional assumptions was used.

$$N = \frac{E}{{P(E)}}\,where\,E = \frac{{{{\left( {\frac{{z\alpha }}{2} + Z\beta } \right)}^2}}}{{\rho (1 - \rho )Ln{{(HR)}^2}}}$$

Where: N = total number of sample size, E = total number of events required, p(E) = probability of event,ρ=proportion of subjects under exposure variable,HR=hazard ratio,\(\frac{{z\alpha }}{2}\) =critical value at 95% confidence level, which 1.96 and \(Z\beta\)= 0.842, standard deviation and correlation of covariate 0.5 and 0, respectively, and 0.1 for probability of withdrawal.

Therefore, the sample size was calculated for the four predictor variables, i.e., use of vasopressors, neuromuscular blockers, hemodialysis, and body mass index ≤ 21 kg/m2 based on the study done in Korea [11] (Table 1).

Table 1 Sample size calculation to assess time to death and its predictors among mechanically ventilated patients in West Amhara comprehensive specialized hospitals, Ethiopia, 2022
Full size table

Accordingly, the largest sample size obtained was 355. Thus, by adding 10% for possible incomplete charts and lost medical records, the total sample size was 391.

Sampling techniques and procedures

A simple random sampling was employed to select three hospitals from a total of five comprehensive specialized hospitals. Moreover, after the sample size was proportionally allocated for each hospital, a simple random sampling technique using the computer generation method was used to select the patients’ charts. The list of patients who were admitted to the ICU and received MV was used as a sampling frame. The total number of patients who received MV and were admitted to the ICU in two years was 688 (Fig. 1).

Fig. 1
figure 1

Schematic presentation of sampling procedure to select mechanically ventilated patients admitted to the ICU in comprehensive specialized hospitals in West Amhara, 2022. NB: CSH (Comprehensive Specialized Hospital)

Full size image

Operational definitions

Prolonged use of mechanical ventilation

when patients experience difficulty of weaning from MV after 21 days (the use of mechanical ventilation for ≥ 21 days) [19].

Incomplete card

patient chart in which the outcome of the patient and the duration of mechanical ventilation were not registered.

Comorbidity

presence of disease conditions according to the Carlson comorbidity index. A “yes” to these disease conditions indicates comorbidity, and “no” answer was taken as the absence of comorbidity [20].

Event

mechanically ventilated adult patients who died in the hospital during the follow-up period.

Censored

patients who did not develop the outcome of interest (death) such as Left Against Medical Advice (LAMA), transferred out, and recovered during the follow-up period.

Time to death

The number of days starting from the initiation of the mechanical ventilator to the occurrence of an event (death) within the follow-up period.

Data collection tool and procedures

Data was collected using a data extraction tool adapted from different literature [2, 4, 11, 21, 22]. The questionnaire has five sections: factors related to socio-demographic characteristics, factors present at the initiation of mechanical ventilation, factors related to the patient’s clinical condition, factors related to the management of patients, and factors related to vital signs at admission (Supplementary material 1). The data were obtained from the ICU registration book and patients’ charts. Data collection was done by four BSc nurses, and one MSc nurse supervised the collection period process.

Data quality control

A one-day training was provided to data collectors and supervisor on the data collection process. Preliminary chart review was conducted in Debre Markos comprehensive specialized hospital to ensure the availability of variables on the patient’s chart. Moreover, the data were cross-checked for consistency, completeness, clarity, and accuracy by the supervisor and the principal investigator before analysis.

Data processing and analysis

Data were entered into Epi Data Version 4.6 and exported into Stata Version 14 statistical software for analysis. Descriptive statistics of numeric variables are presented using the median with Inter Quartile Range (IQR) and categorical variables expressed in frequency with percentages. The incomplete data were managed with the assumption of multiple imputations, after which it was ascertained that the missing data was completely at random and less than 15% of the records. The outcome of each participant was dichotomized into censored or event. The Kaplan-Meier failure curve was used to estimate failure time. The Kaplan-Meier failure curve and the log-rank test were fitted to test the presence of differences between different categories of explanatory variables. The proportional hazard assumption was checked both graphically and statistically by using the log (-log) plot and Schoenfeld residual test, respectively, and it was satisfied. Multicollinearity for independent variables was checked by using Variance Inflation Factor (VIF) and the mean VIF was 1.37. The Cox proportional hazard regression was used to explore the association between each independent variable and the outcome variable. The model’s fitness was tested using the Cox–Snell residual test. Variables having a p-value < 0.25 in the bivariable analysis were candidates for the multivariable analysis. An Adjusted Hazard Ratio (AHR) with 95% Confidence Intervals (CI) was computed to explore the strength of the association, and variables with a p-value less than 0.05 were considered statistically significant with time to death.

Result

Socio-demographic characteristics of the study participants

During the study period, 688 patients were treated with invasive mechanical ventilation. From these, 391 patients’ charts were reviewed (Supplementary material 2). About 159 (40.7%) of the participants were less than 45 years old, with a median age of 50 years (IQR: 18–90). Nearly two-thirds (63.4%) were males, and more than half (54.5%) were rural residents (Table 2).

Table 2 Socio-demographic characteristics of patients on MV admitted to ICU in West Amhara Comprehensive specialized hospitals, Ethiopia, 2022
Full size table

Vital signs of mechanically ventilated patients during admission

Nearly half (45.5%) and three-fourths (73.40%) of study participants were hypotensive and hypoxic during admission, respectively (Table 3).

Table 3 Vital signs during admission among mechanically ventilated patients admitted to ICU in West Amhara Comprehensive specialized hospitals, Ethiopia, 2022
Full size table

Clinical-related characteristics of the study participants

More than a quarter (28.6%) of patients were admitted to the ICU with the diagnosis of respiratory failure, followed by neurological disorder (16.4%). Nearly half (48.3) of the study participants had a Glasgow Coma Scale (GCS) score of 3–8 (Table 4).

Table 4 Clinical-related characteristics of patients on MV admitted to ICU in West Amhara Comprehensive specialized hospitals, Ethiopia, 2022
Full size table

Management-related characteristics of patients on MV

More than one-fourth (27.62%) of the study participants were received vasopressors during their ICU admission. Tracheostomy procedure is performed for only (8.4%) of the mechanically ventilated patients (Table 5).

Table 5 Management-related characteristics of patients on MV admitted to the ICU in West Amhara Comprehensive specialized hospitals, Ethiopia, 2022
Full size table

Ventilation-related characteristics of the study participants

The majority of patients (90.0%) were received Synchronized Intermittent Mandatory Ventilation (SIMV) mode of mechanical ventilation. From the total mechanical ventilated patients, (15.5%) were developed ventilated associated complications (Table 6).

Table 6 Ventilation-related characteristics of the study participants of patients on MV admitted to ICU in West Amhara Comprehensive specialized hospitals, Ethiopia, 2022
Full size table

Assessing the proportional hazard assumption

The Cox proportional hazard assumptions were checked statistically and graphically using the global test and log (-log) plot, respectively. All the covariates met the proportional hazard assumptions, and the p-value overall Schoenfeld global test was 0.0795 (Table 7).

Table 7 Proportional hazard assumption test by Schoenfeld residuals for each predictor variable
Full size table

Model goodness of fit test

Cox Snell residuals were used to check the goodness of fit test. The hazard function follows 45˚ closest to the baseline hazard, which showed that the model was well fitted (Fig. 2).

Fig. 2
figure 2

Nelson-Aalen cumulative hazard graph against Cox-Snell residual for patients on MV and admitted to the ICU in West Amhara comprehensive specialized hospitals, 2022

Full size image

Time to death among mechanically ventilated patients

This study found that the overall mortality of patients on mechanical ventilation admitted to the ICU was 62.2%, with a median time to death of 16 days (95% CI: 11, 22). Patients were followed for a minimum of 1 day and a maximum of 36 days, with a total person-time at risk of 4098 days. The overall incidence density rate of mortality among mechanically ventilated patients admitted to the ICU was 5.93 per 100 person-day observations (95% CI: 5.23, 6.72). According to the Kaplan-Meier failure curve, the cumulative probability of death at the end of 8, 16, 24, 32 and 36 days was 5.8%, 3.6%, 7.3%, 23%, and 25.0%, respectively (Fig. 3).

Fig. 3
figure 3

Kaplan-Meier failure function for patients on MV admitted to ICU in West Amhara comprehensive specialized hospitals: Ethiopia, 2022

Full size image

Test for equality of failure function

A Kaplan-Meier failure curve together with the log-rank test was done to test the presence of a difference in the time to death among predictor variables. According to the current study, those patients without a tracheostomy procedure had a shorter time to death with a median survival time of 12 days (95% CI: 10, 18) as compared to their counterparts (23 days, 95% CI: 22, 25) (Fig. 4).

Fig. 4
figure 4

Kaplan-Meier failure curve for tracheostomy among adult patients on mechanical ventilation admitted to ICU comprehensive specialized hospitals in West Amhara, 2022

Full size image

In addition, those patients who received CPR had a shorter time to death with a median survival time of 5 days (95% CI: 3, 7) as compared to their counterparts (24 days, 95% CI: 23, 25) (Fig. 5).

Fig. 5
figure 5

Kaplan-Meier failure curve for CPR among patients on MV admitted to the ICU in comprehensive specialized hospitals in West Amhara, 2022

Full size image

Patients who had a respiratory rate less than 12 had a shorter time to death with a median survival time of 3 days (95% CI: 3, 8), as compared to those who had 12–20 breaths/minutes (23 days, 95% CI: 23, 24). Similarly, those patients with a respiratory rate greater than 20 breaths/ min had a shorter time to death with a median survival time of 12 days (95% CI: 9, 15) compared to those who had a respiratory rate of 12–20 breaths/min (Fig. 6).

Fig. 6
figure 6

Kaplan-Meier failure curve for respiratory rate among adult patients on mechanical ventilation admitted to ICU comprehensive specialized hospitals in West Amhara, 2022

Full size image

Moreover, those hypotensive patients at admission had a shorter time to death with a median survival time of 3 days (95% CI: 2, 6) as compared to normotensive patients (23 days, 95% CI: 15, 24). Similarly, patients having a higher systolic blood pressure (> 140mmHg) had also a shorter time to death with a median survival time of 12 days (95% CI: 10, 22) compared to normotensive patients (Fig. 7).

Fig. 7
figure 7

Kaplan-Meier failure curve for systolic blood pressure among adult patients on mechanical ventilation admitted to ICU comprehensive specialized hospitals in West Amhara

Full size image

Predictors of time to death among mechanically ventilated patients admitted to the ICU

Based on the multivariable Cox proportional hazard regression analysis, patients who underwent tracheostomy, CPR, SBP < 90 mmHg, and respiratory rate < 12 rate/min were significant predictors of time to death among mechanically ventilated patients admitted to the ICU.

Keeping other variables constant, underwent tracheostomy procedure among mechanically ventilated patients decreased the hazard of death by 60% as compared to their counterparts (AHR: 0.40, 95% CI: 0.20, 0.80). The hazard of death among those cardiopulmonary resuscitated patients was 8.8 times higher as compared to their counterparts (AHR: 8.78, 95% CI: 5.38, 14.35). Moreover, those mechanically ventilated patients who had hypotension at admission had 2.96 times a higher hazard of death as compared to those patients who were normotensive (AHR: 2.96, 95% CI: (1.11–7.87). Similarly, those patients who had a respiratory rate of < 12 breath/min had 2.7 times higher hazard of death as compared to those who had a respiratory rate of 12–20 breath/min (AHR: 2.74, 95% CI: (1.48–5.07)) (Table 8).

Table 8 Bivariable and multivariable Cox-proportional hazard regression analysis of predictors of time to death among mechanically ventilated patients admitted to the ICU in West Amhara, Comprehensive specialized hospitals, Ethiopia, 2022
Full size table

Discussion

This study aimed to determine the time to death and identify its predictors among patients on MV admitted to the ICU. According to the current study, the median time to death among patients on mechanical ventilator was 16 days. Those mechanically ventilated patients who underwent tracheostomy, received CPR, having a respiratory rate less than 12 breaths per minute, and were hypotensive were significant predictors for time to death.

This study found that the median time to death of mechanically ventilated ICU patients was 16 days (95% CI: 11–22). This finding is in line with the studies done in Saudi Arabia, Spain, and Ethiopia [5, 23, 24]. However, it is lower than the study conducted in Southern Brazil [2]. The reason for this discrepancy might be due to the difference in the quality of the health care delivery system [2, 25]. Additionally, it might also be due to the patients’ clinical status during admission to the ICU, which can increase the hazard of death [9, 25].

This study also showed that the overall incidence density rate of death among mechanically ventilated patients admitted to the ICU was 5.93 per 100-day observations (95% CI: 5.23–6.72), which is in line with the study conducted in Ethiopia [26]. The overall mortality of patients on mechanical ventilation was 62.2%, which is consistent with the previous study conducted in Ethiopia (60.7%) [27]. However, this finding is higher than the studies conducted in Ethiopia [24] Sub-Saharan Africa [28], and Brazil [2], which reported that the prevalence of mortality among mechanically ventilated patients admitted to the ICU was 33.78%, 49%, and 51%, respectively. The reason for this discrepancy might be due to the difference in the level of ICU organization and duration of follow-up.

Mechanically ventilated patients who underwent tracheostomy procedure had decreased the hazard of death by 60% as compared to their counterparts. This finding is supported by studies done in Nigeria and Taiwan [9, 29]. This might be due to the protective effects of tracheostomy procedure such as improved patient comfort, better oral hygiene, less dental damage and tracheal injury, easier and faster nursing care, and lower airway resistance, which may facilitate the weaning process and avoid ventilator-associated pneumonia [30].

This study also found that the hazard of death among those cardiopulmonary resuscitated patients was 8.8 times higher as compared to their counterparts. This finding is supported by the studies conducted in Ethiopia, Berlin, and France [24, 31, 32]. This might be due to the fact that those patients who were cardiopulmonary resuscitated had the problems of hemodynamic failure, lack of oxygenation [32], and a higher severity of illness [31], which facilitates patient mortality. The time of CPR initiation could be another possible reason since the chance of survival would be low if CPR couldn’t be initiated immediately after cardiac arrest due to severe and permanent brain damage [31, 33].

Mechanically ventilated patients who were hypotensive at admission had 2.96 times a higher hazard of death as compared to those who were normotensive. This is supported by studies done in France and China [32, 34]. This might be due to the scientific evidence that hypotension decreases tissue perfusion, which leads to multi-organ failure such as renal failure, cardiac failure, and brain death [35, 36]. Therefore, a healthcare provider should focus on maintaining the blood pressure of critically ill patients who have received mechanical ventilation.

Moreover, this study revealed that those patients who had a respiratory rate less than 12 breaths per minute had 2.7 times higher hazard of death as compared with those who had a respiratory rate of 12–20 breaths per minute. This is supported by the studies done in Denmark and Sweden [37, 38]. The possible reason for this might be attributed to the decreasing respiratory rate, which is an indicator of brain dysfunction from many causes, like structural or non-structural processes affecting the central nervous system, which leads to death [37, 38]. Therefore, measuring the respiratory rate is important for early identification of high-risk patients because the respiratory rate is a superior indicator to other physiologic parameters.

Strengths and limitations of the study

This is a multi-center follow-up study, which increases its generalizability to the target population. However, this study has some limitations. Since it was institutional-based and the data were obtained from patients’ medical records, mortality-predicting variables such as the Acute Physiology and Chronic Health Evaluation (APACHE) and Sequential Organ Failure Assessment (SOFA) scores couldn’t be addressed.

Conclusion

Patients treated with mechanical ventilation had a shorter time to death. Being cardiopulmonary resuscitated, hypotensive, and had lower respiratory rate were significant predictors of time to death, whereas patients who underwent tracheostomy was negatively associated with time to death. Therefore, we strongly recommended that the healthcare provider should perform a tracheostomy procedure early for critically ill patients who need prolonged ventilation, and a special attention should be given for patients who had lower systolic blood pressure and respiratory rate. Furthermore, a prospective follow up study is required by incorporating the Acute Physiology and Chronic Health Evaluation (APACHE) and Sequential Organ Failure Assessment (SOFA) scores to predict the mortality of mechanically ventilated patients.