Time to death and its predictors among adults on mechanical ventilation at Hiwot Fana comprehensive specialized hospital,eastern Ethiopia: A retrospective cohort study

Abstract

Introduction

Mechanical ventilation is an essential lifesaving intervention for patients with severe illness. Mortality among patients receiving mechanical ventilation is high in low-resource settings, including Ethiopia. Evidence on predictors of time to death in such settings is limited. Therefore, this study aimed to assess time to death and its predictors among adult patients receiving mechanical ventilation in intensive care unit at Hiwot Fana comprehensive specialized Hospital, Eastern Ethiopia.

Method

An institution-based retrospective cohort study was conducted among 586 mechanically ventilated adult patients. Data were collected using structured checklist. Time to event analysis was performed using cox proportional hazards model to identify predictors of time to death. The proportional hazard assumption was assessed using schoenfeld residuals. Adjusted hazard ratios with 95% confidence interval were used to quantify associations.

Results

The incidence density of mortality was 6.74 per 100 person-days observations (95% CI: 5.92-7.68). Factors associated with a higher hazards of death included platelet count less than 150,000*10⁶/L (Adjusted Hazard Ratio (AHR)=1.74, 95% CI: 1.26–2.40), cardiovascular failure during mechanical ventilation (AHR=1.62, 95% CI: 1.08–2.43), cardiopulmonary resuscitation (AHR=2.36, 95% CI: 1.71–3.31), vasopressor use (AHR=2.94, 95% CI: 2.02–4.27), and higher fraction of inspired oxygen at initiation of mechanical ventilation (AHR=1.01, 95% CI: 1.007-1.019) had higher risk of death. Gastrostomy was associated wit lower hazard of death (AHR=0.35, 95% CI: 0.18-0.68).

Conclusion

Several clinical and treatment related factors were associated with time to death among mechanically ventilated Intensive Care Unit (ICU) patients. Close monitoring and appropriate management of high risk patients are essential needed.

Keywords

time to death mechanical ventilation mortality intensive care unit Ethiopia

Introduction

Mechanical ventilation (MV) is critical lifesaving intervention widely used in Intensive Care Units (ICU) to support patients with severe and life threatening conditions.¹ Despite substantial advances in critical care medicine over recent decades, ICU mortality remains a major global public health concern.² A considerable proportion of critically ill patients depend on MV for survival, with studies indicating that up to 70% of ICU admissions require ventilatory support at some point during their stay.^3–5 However, the use of MV is also associated with increased risk of complications and mortality, making outcomes among these patients a key indicator of ICU performance.⁶

Globally, mortality among mechanically ventilated patients in ICUs varies widely, ranging from 20.3% to 77.2% depending on patient characteristics, healthcare settings, and resource availability.^7,8 Although some improvements have been observed over time for instance, a decline in ICU mortality from 31% in 1998 to 28% in 2010 these gains are not uniformly distributed.⁷ In low and middle-income countries (LMICs), including those in sub-Saharan Africa, mortality rates remain disproportionately high.⁹ This disparity is often attributed to delayed healthcare-seeking behavior, limited access to advanced care, inadequate ICU infrastructure, and shortages of trained personnel and essential medical supplies.^1,10

Evidence from sub-Saharan Africa shows substantial variability in ICU mortality among mechanically ventilated patients, ranging from 24.4% in South Africa to over 60% in countries such as Nigeria, Kenya, and Rwanda, with Ethiopia reporting rates between 29.6% and 62.2%.^9,11–13 Such variation highlights the influence of contextual factors, including differences in ICU capacity, patient case-mix, and clinical management practices.^3,14

Beyond overall mortality, understanding when deaths occur during ICU stay is equally important. Previous studies indicate that mechanically ventilated patients often experience short survival durations, with median time to death ranging from 13 to 16 days in Ethiopian settings.^13,15 Moreover, patients on MV have been shown to have significantly higher hazards of death compared to non-ventilated patients.^16,17 Several predictors of mortality have been identified, including advanced age, low body mass index, severity of illness (e.g., APACHE II score), presence of comorbidities, sepsis, low Glasgow Coma Scale (GCS), need for vasopressors, and ICU-related complications.^18–20

However, most existing studies particularly in Ethiopia primarily report proportions of mortality without adequately accounting for the timing of death and censoring of observations. This limits the ability to fully understand patient survival patterns and may lead to incomplete or biased conclusions. Time-to-event (survival) analysis provides a more robust approach by incorporating both the occurrence and timing of death, allowing for better estimation of survival probabilities and identification of predictors influencing the hazard of death over time.

In Ethiopia, available studies on mechanically ventilated ICU patients are limited in scope, often restricted to specific populations, short follow-up periods, or single outcome measures such as mortality proportion.^15,21 Furthermore, there is limited evidence focusing specifically on predictors of time to death among mechanically ventilated patients, particularly in eastern Ethiopia. This represents an important gap in the literature, as understanding survival dynamics is essential for improving clinical decision-making, resource allocation, and patient management in resource-limited settings.

Therefore, this study aims to assess the time to death and its predictors among adult ICU patients receiving mechanical ventilation at Hiwot Fana Comprehensive Specialized Hospital (HFCSH) in Harar, eastern Ethiopia. By applying survival analysis techniques, this study seeks to provide more comprehensive evidence that can inform targeted interventions, optimize patient care, and ultimately improve ICU outcomes in low-resource settings.

Methods and materials

Study setting and period

The study was carried out in HFCSH, located in Harar town, Eastern Ethiopia approxiamately 526 Km from Addis Ababa. Harari Regional State has an estimated population of 283,000 with an approximately equal male to female ratio. There are two public hospitals, two private hospitals, and one federal police hospitals in the region. The study was carried out from November 1 to November 30, 2024. The follow-up period for the cohort extended from January 1, 2019, to December 31, 2023.

Study design

A hospital-based retrospective cohort study design was employed to assess time to death and its predictors among mechanically ventilated adult ICU patients.

Source population

All adult patients admitted to the ICU and managed with mechanical ventilation at HFCSH between January 1, 2019, and December 31, 2023 were source popuation.

Study population

Selected adult ICU patients on mechanical ventilation whose medical records met inclusion criteria forms study population.

Inclusion criteria

Adult patients aged 18 years and above, patients who received mechanical ventilation in ICU were included.

Exclusion criteria

Charts with missing or unclear key dates (admission, ventilation initiation, or outcome) and incomplete records lacking critical variables were excluded from the study.

Sample size determination

The sample size was determined using Schoenfeld’s formula for survival analysis under the Cox proportional hazards model. A 95% confidence level (Zα/2 = 1.96) and 80% power (Zβ = 0.84) were assumed. The probability of event (death) was taken as 62.2% from a previous Ethiopian study conducted among mechanically ventilated patients²²: Since previously reported hazard ratios in similar studies vary and are often large, a more conservative effect size (HR = 1.36) was assumed to ensure adequate power to detect modest associations. Equal allocation between exposure groups (ρ = 0.5) was assumed.

Using the Schoenfeld formula for Cox proportional hazards models:

E = \frac{{(z \frac{α}{2} + Z β)}^{2}}{{P (1 - P) \ln (H R)}^{2}} = 331, N = \frac{E}{P (E)} = \frac{331}{0.622} = 533 . After adding 10 % for incomplete records, the final sample size was 586 .

Sampling technique

A sampling frame was constructed from ICU admission records of patients who received mechanical ventilation during the study period. Since the number of eligible patients exceeded logistical feasibility for full review, simple random sampling was applied using computer-generated random numbers to ensure representativeness and minimize selection bias.

Data collection methods

Structured and pretested data extraction format (that was developed from review of literatures^18,22–26 was used for data collection (Appendix A). In this study, two BSc Holder data collectors and one Master degree holder supervisor was participated. One day training was given about the tool. Then, the data collectors were deployed to the data collection after we agreed on the tool. The primary investigator of the study and the supervisor was following the data collection process critically to minimize data errors.

Medical record numbers was collected from admission-discharge registration books of HFCSH ICU Wards and based on that those selected sampling unit, individual patient folders data was collected as per checklists.

Study variables

The Dependent variable for this study was time to death among mechanically ventilated patients. Independent variables included sociodemographic characteristics (age, body mass index), clinical-related predictors (cause of mechanical ventilation initiation such as ARDS, COPD, septic shock, diabetes mellitus (DM), pneumonia, sepsis, neurological failure, hepatic failure, renal failure, hypotension, and cardiovascular conditions; and diagnoses developed during mechanical ventilation such as ventilator-associated pneumonia (VAP), ARDS, COPD, septic shock, diabetes mellitus, sepsis, and cardiovascular failure), and management-related predictors (tracheostomy, vasopressor use, neuromuscular blocking agents, haemodialysis, cardiopulmonary resuscitation, and systemic corticosteroid use). To account for baseline illness severity, additional clinical indicators such as Glasgow Coma Scale²⁷ and vital sign were included as proxy measures of patient severity.

Operational definition

Event: mechanically ventilated adult patients who died in the hospital during the follow-up period.²²

Censored: patients who did not develop the outcome of interest (death) such as Left Against Medical Advice,²⁸ discharged, transferred, or survived until the end of the follow-up period were considered censored.²² Censoring was assumed to be non-informative, meaning that the probability of being censored was independent of the risk of death, conditional on the observed covariates. This assumption is reasonable in this setting, as censoring mainly occurred due to discharge or referral; however, it cannot be fully verified and is acknowledged as a limitation.

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 quality control

To control data quality the data retrieval form was pre-tested on 5 % of total sample in Jugal Hospital. This was conducted in another hospital to evaluate the clarity, completeness, and consistency of the data extraction tool, as well as the availability of required variables in routine medical records, while avoiding contamination of the actual study sample. Based on the findings of pre-test some modification was done for the tools. The review was conducted by trained data collectors. Two days intensive training was given to data collectors and supervisors about the purpose, study tool and an overall data collection procedure. The supervisor had checked the completeness and consistency of the filled data retrieval immediately before submission. Manual checks were done for completeness during data collection. Range and consistency was also been checked during entry of data. In addition, double data entry was done by two data clerk and checked for validation by the principal investigator of the research following data entry.

Data analysis

Data were collected using Kobo Toolbox, checked for completeness, and processed using Microsoft Excel and Stata version 17. Descriptive statistics, including means (±SD), medians, Inter Quartile Range (IQR), and proportions, were used to summarize continuous and categorical variables as appropriate. The primary outcome was time to death among mechanically ventilated patients, measured in days from initiation of mechanical ventilation to the occurrence of death or censoring (e.g., recovery, transfer, or end of follow-up). Survival experience was described using Kaplan–Meier survival curves, and differences between groups were compared using the log-rank test. The incidence density rate was calculated per 100 person-days of observation.

Both bivariable and multivariable Cox proportional hazards regression models were fitted to identify predictors of time to death. Variables were selected for multivariable analysis based on a combination of statistical criteria (p-value <0.25 in bivariable analysis) and clinical relevance informed by prior literature and subject-matter knowledge. A purposeful selection approach was employed to ensure inclusion of potential confounders regardless of their bivariable significance.

The proportional hazards assumption was assessed using Schoenfeld residuals and the global test (χ² = 11.38, p = 0.9358), indicating no violation of the assumption. Additionally, graphical methods (e.g., log-minus-log survival plots) were used to support this assessment. Multicollinearity among independent variables was evaluated using the variance inflation factor (VIF), with a mean VIF <10 indicating no significant multicollinearity.

Model adequacy was assessed using Cox–Snell residuals, and the model was found to fit the data well. Adjusted hazard ratios with 95% confidence intervals were reported to quantify the strength of associations, and statistical significance was declared at a p-value <0.05. The dataset was examined for missing values using frequency distributions and summary statistics. No missing values were identified for the key variables included in the final analysis.Therefore, no imputation or other missing data handling techniques were required.

Results

Socio-demographic characteristics

A total of 586 patient charts were reviewed and included in the analysis. The median age of participants was 30 years (IQR: 25–45), with more than half (314, 53.6%) aged 25–45 years. Males accounted for 334 (57.0%) of the study population. The majority of participants (368, 62.8%) were from rural areas (Table 1).

Table 1.

Socio-demographic characteristics of patients on mechanical ventilation admitted to ICU in Hiwot Fana comprehensive specialized hospital, Ethiopia, 2024.

Variable	Category	Frequency	Percentage
Age	18-24	131	22.35
	25-45	314	53.58
	46-71	141	24.06
Sex	Female	252	43.0
Sex	Male	334	57.0
Residence	Rural	368	62.8
Residence	Urban	218	37.2

Clinical characteristics

Nearly half of the patients (286, 48.8%) were admitted from the emergency department. Most patients were medical cases (334, 57.0%), while 303 (51.7%) underwent surgical interventions.

At admission 465 (79.4%) had normal systolic blood pressure (90–140 mmHg), 395 (67.4%) had normal diastolic blood pressure (60–90 mmHg), 237 (40.4%) had hypothermia (<36.5°C), 373 (63.7%) had tachycardia (>100 beats/minute), 525 (89.6%) had normal respiratory rate (12–20 breaths/minute), 324 (55.3%) had oxygen saturation <90%, 307 (52.4%) had normal Glasgow Coma Scale^12–14 and comorbidity was present in 208 (35.5%) of patients (Table 2).

Table 2.

Clinical related characteristics of patients on mechanical ventilation admitted to ICU in Hiwot Fana comprehensive specialized hospital, Ethiopia, 2024.

Variable	Category	Frequency	Percentage
Source of admission	Emergency	286	48.81
	Operation room	227	38.74
	Ward	73	12.46
Patient’s case	Medical case	334	57
	Surgical case	303	51.71
	Both Medical and surgical cases	80	13.42
	Obstetric and gynecological case	43	7.33
Systolic blood pressure (mmHg)	<90	57	9.73
	90-140	465	79.35
	>140	64	10.92
Diastolic Blood pressure (mmHg)	<60	137	23.38
	60-90	395	67.41
	>90	54	9.22
Temperature (°C)	<36.5	237	40.44
	36.5-37.5	208	35.49
	>37.5	141	24.06
Heart Rate (beat/minute)	<60	9	1.54
	60-100	204	34.81
	>100	373	63.65
Respiratory Rate (beat/minute)	12-20	525	89.59
Respiratory Rate (beat/minute)	>20	61	10.41
Partial Oxygen Saturation	<90	324	55.29
Partial Oxygen Saturation	90-100	262	44.71
Glasgow Coma Scale	3-8	143	24.4
	9-12	136	23.21
	13-15	307	52.39
Comorbidity	No	378	64.51
Comorbidity	Yes	208	35.49

mmHg: millimeters of mercury; °C: degrees Celsius.

Reasons for mechanical ventilation

The most common indication for mechanical ventilation was acute respiratory distress syndrome (ARDS) (167, 28.5%), followed by delayed awakening (122, 20.8%) and neurological failure (121, 20.7%) (Table 3).

Table 3.

Reason of patients on mechanical ventilation admitted to ICU in Hiwot Fana comprehensive specialized hospital, Ethiopia, 2024.

Condition	Yes n(%)
ARDS	167(28.5)
COPD	10 (1.71)
Septic shock	58 (9.9)
Sepsis	52 (8.9)
Neurological failure	34 (5.8)
Hepatic failure	10 (1.71)
Renal failure	25 (4.27)
Cardiovascular failure	34 (5.8)
Delay awakening	122 (20.82)

ARDS: Acute Respiratory Distress Syndrome; COPD: Chronic Obstructive Pulmonary Disease

Time to death among mechanically ventilated patients

A total of 586 mechanically ventilated patients were included in the analysis. The overall mortality rate was 38.6%. The median time to death was 9.81 days (95% CI: 8.91–12.86).

Patients were followed for a minimum of 0.27 days and a maximum of 72.98 days, contributing a total of 3,350.67 person-days of observation.

The overall incidence density of mortality was 6.74 per 100 person-days (95% CI: 5.92–7.68).

Kaplan–Meier analysis showed that cumulative mortality increased over time, with probabilities of 12.2% at day 1, 24.7% at day 3, 40.2% at day 6, and 50.6% at day 10 (Figure 1).

Figure 1.

Overall Kaplan-Meir survival of patients on mechanical ventilation admitted to ICU in Hiwot Fana comprehensive specialized hospital, Ethiopia, 2024.

Patients receiving vasopressors and those who did not undergo gastrostomy had lower survival probabilities compared to their counterparts (Figures 2 and 3).

Figure 2.

Kaplan-Meir survival curve compares survival time of patients who received gastrostomy procedure during mechanical ventilation admitted to ICU in Hiwot Fana comprehensive specialized hospital, Ethiopia, 2024.

Figure 3.

Kaplan-Meir survival curve compares survival time of patients who received vasopressor and on mechanical ventilation admitted to ICU in Hiwot Fana comprehensive specialized hospital, Ethiopia, 2024.

Predictor of time to death among mechanically ventilated patients

In bivariable Cox regression, variables with p < 0.25 and clinically relevant covariates were included in the multivariable model. After adjustment, platelet count <150,000/mm³, cardiovascular failure during mechanical ventilation, cardiopulmonary resuscitation (CPR), vasopressor use, gastrostomy procedure, and fraction of inspired oxygen (FiO₂) were independent predictors of time to death.

Patients with platelet count <150,000/mm³ had a 74% higher hazard of death compared to those within the normal range (AHR = 1.74; 95% CI: 1.26–2.40).

Cardiovascular failure during mechanical ventilation increased hazard of death by 62% (AHR = 1.62; 95% CI: 1.08–2.43).

Patients who underwent Cardio Pulmonary Resuscitation (CPR) had more than twice the hazard of death (AHR = 2.36; 95% CI: 1.71–3.31), while vasopressor use was associated with nearly threefold increased hazard (AHR = 2.94; 95% CI: 2.02–4.27).

Each 1% increase in FiO₂ at initiation of mechanical ventilation was associated with a 1% increase in hazard of death (AHR = 1.01; 95% CI: 1.007–1.019).

Patients who received gastrostomy had a 65% lower hazard of death compared to those who did not (AHR = 0.35; 95% CI: 0.18–0.68) (Table 4).

Table 4.

Bivariable and multivariable analysis using cox regression model for predictors of time to death of patients on mechanical ventilation admitted to ICU in Hiwot Fana comprehensive specialized hospital, Ethiopia, 2024.

Variable	Category	Death (%)	Censored (%)	CHR (95% CI)	AHR (95% CI)	P-value
Age	18-24	38 (29.01)	93 (70.99)	1	1
	25-45	113 (35.99)	201 (64.01)	1.07 (0.74 – 1.56)	0.76 (0.52 – 1.16)	0.163
	46-71	75 (53.19)	66 (46.81)	1.65 (1.11 – 2.44)	1.40 (0.93 – 2.11)	0.103
Diastolic Blood pressure (mmHg)	<60	75 (54.74)	62 (45.26)	1.86 (1.4 – 2.47)	1.24 (0.89 – 1.72)	0.19
	60-90	133 (33.67)	262 (66.33)	1	1
	>90	18 (33.33)	36 (66.67)	1.07 (0.65 – 1.75)	1.14 (0.68 – 1.92)	0.602
Temperature	<36.49	107 (45.15)	130 (54.85)	1.65 (1.21 – 2.25)	1.06 (0.75 – 1.49)	0.738
	36.5-37.2	65 (31.25)	143 (68.75)	1	1
	>37.2	54 (38.30)	87 (61.70)	1.37 (0.95 – 1.97)	1.07 (0.73 – 1.56)	0.722
Platelet Count(*10⁶/L)	<150000	65 (47.10)	73 (52.90)	1.47 (1.09 – 1.98)	1.74 (1.26 – 2.40)	0.001**
	150000-450000	142 (36.32)	249 (63.68)	1	1
	>450000	19 (33.33)	38 (66.67)	0.92 (0.56 – 1.49)	0.97 (0.58 – 1.61)	0.911
Cause of MV initiation septic shock	Yes	34 (58.62)	24 (41.38)	1.9 (1.31 – 2.74)	0.83 (0.54 – 1.28)	0.403
Cause of MV initiation septic shock	No	192 (36.36)	336 (63.64)	1	1
Cause of MV initiation Neurological failure	Yes	45 (37.19)	76 (62.81)	0.69 (0.49 – 0.96)	0.93 (0.63 – 1.37)	0.726
Cause of MV initiation Neurological failure	No	181 (38.92)	284 (61.08)	1	1
Cause of MV initiation hepatic failure	Yes	19 (76.00)	6 (24.00)	2.81 (1.24 – 6.36)	1.18 (0.49 – 2.83)	0.71
Cause of MV initiation hepatic failure	No	207 (36.90)	354 (63.10)	1	1
Cause of MV initiation renal failure	Yes	6 (60.00)	4 (40.00)	2.6 (1.62 – 4.16)	1.62 (0.97 – 2.71)	0.063
Cause of MV initiation renal failure	No	220 (38.19)	356 (61.81)	1	1
Cause of MV initiation cardiovascular failure	Yes	17 (50.00)	17 (50.00)	1.86 (1.13 – 3.06)	0.59 (0.32 – 1.09)	0.094
Cause of MV initiation cardiovascular failure	No	209 (37.86)	343 (62.14)	1	1
Diagnosis during MV cardiovascular failure	Yes	53 (67.09)	26 (32.91)	3.02 (2.2 – 4.13)	1.62 (1.08 – 2.43)	0.019*
Diagnosis during MV cardiovascular failure	no	173 (34.12)	334 (65.88)	1	1
Diagnosis during MV neurological failure	Yes	33 (32.67)	68 (67.33)	0.60 (0.41 – 0.87)	1.16 (0.76 – 1.79)	0.479
Diagnosis during MV neurological failure	No	193 (39.79)	292 (60.21)	1	1
Diagnosis during MV septic shock	Yes	103 (60.95)	66 (39.05)	2.66 (2.04 – 3.47)	1.25 (0.88 – 1.77)	0.210
Diagnosis during MV septic shock	No	123 (29.50)	294 (70.50)	1	1
Gastrostomy done	Yes	11 (50.00)	11 (50.00)	0.51 (0.26 – 0.98)	0.35 (0.18 – 0.68)	0.002**
Gastrostomy done	No	215 (38.12)	349 (61.88)	1	1
Cardiopulmonary resuscitation done	Yes	68 (82.93)	14 (17.07)	3.93 (2.95 – 5.23)	2.36 (1.71 – 3.31)	<0.001**
Cardiopulmonary resuscitation done	No	158 (31.35)	346 (68.65)	1	1
Vasopressor administer	Yes	154 (66.38)	78 (33.62)	4.43 (3.35 – 5.87)	2.94 (2.02 – 4.27)	<0.001**
Vasopressor administer	No	72 (20.34)	282 (79.66)	1	1
Fraction of inspired Oxygen				1.02 (1.01 – 1.02)	1.01 (1.007 – 1.019)	<0.001**

AHR: Adjusted Hazard Ratio; CI: Confidence Interval;; CHR: Crude Hazard Ratio; mmHg: millimeters of mercury; MV: Mechanical Ventilation.*: P-value < 0.05 ; ** : P-value < 0.001

Discussion

The present study assessed time to death and its predictors among adult patients receiving mechanical ventilation in the ICU of Hiwot Fana Comprehensive Specialized Hospital. The median time to death was 9.81 days (95 CI: 8.91-12.86). This finding is comparable with studies conducted in Addis Ababa, Ethiopia (13 days:95% CI: 8.0, 18.0)²⁶ and Korea (9 days: IQR 7 – 10 days).¹⁸ The relatively short survival time observed in this study suggests that a substantial proportion of mortality occurs early during ICU admission among mechanically ventilated patients. However, this should be interpreted cautiously, as survival time is influenced by case mix, admission severity, and ICU resource availability rather than reflecting a uniform disease progression pattern.

Patients diagnosed with cardiovascular failure during mechanical ventilation had higher hazard of death compared those without this condition. Similar findings have been reported from Brazil.²³This association is likely explained by underlying disease severity, including hemodynamic instability, impaired tissue perfusion, and progression to multiorgan dysfunction. However, cardiovascular failure in this context should be interpreted as a marker of critical illness severity rather than an independent causal determinant of mortality, and residual confounding by unmeasured severity indices (e.g., APACHE II scores) is likely.

Patients who underwent CPR had a higher hazard of death compared to those who did not, consistent with findings from Ethiopia.²² This association should not be interpreted as CPR increasing mortality risk. Rather, CPR is typically performed in patients who are already in extremis, indicating reverse causality and confounding by severity of illness. Therefore, CPR reflects a terminal clinical state rather than a modifiable predictor of outcome.

Vasopressor administration was associated with increased hazard of death, consistent with findings from Korea and Colombia.^23,25 This relationship likely reflects the presence of shock and severe hemodynamic instability. Therefore, vasopressor use should be considered a proxy indicator of disease severity rather than an independent causal factor of mortality. The observed association is likely confounded by baseline physiological instability and organ failure severity.

Increasing FiO₂ at initiation of mechanical ventilation was associated with a slightly increased hazard of death. Similar findings have been reported in multicenter studies.^29,30 This may indicate that patients requiring higher oxygen concentrations at initiation of ventilation are more likely to have severe hypoxemic respiratory failure.³¹ However, the magnitude of effect was small, and this association should be interpreted in a clinical context rather than as a strong independent predictor of mortality.

Thrombocytopenia was associated with increased hazard of death, consistent with studies from Turkey and Denmark.^32,33 This finding may reflect underlying coagulopathy, sepsis-related platelet consumption, or organ dysfunction. However, platelet count may also act as a surrogate marker of systemic illness severity rather than a direct causal factor of mortality.

Patients who received gastrostomy had a lower hazard of death compared to those who did not.

This finding should be interpreted cautiously due to potential survivor bias (immortal time bias), as patients must survive long enough to receive the procedure. Therefore, the observed protective association may reflect selection of less severely ill or more stable patients rather than a true causal protective effect. Additionally, unmeasured confounding related to nutritional status and overall prognosis may have influenced this association.³⁴

Despite providing important evidence on time to death and its predictors among mechanically ventilated adult patients, this study has several limitations that should be considered when interpreting the findings. First, the study employed a retrospective cohort design based on medical record review, which is inherently subject to limitations such as incomplete documentation, missing variables, and potential recording errors. Second, important clinical variables such as body mass index (BMI), APACHE II score and other standardized severity of illness indicators were not available in the medical records. The absence of these variables limits the ability to fully adjust for baseline disease severity and increases the likelihood of residual confounding.

Third, although multivariable Cox regression was applied, the study may still be affected by unmeasured confounding, particularly due to the lack of comprehensive physiological and laboratory severity indicators. Fourth, the findings are based on a single-center study, which may limit the generalizability of the results to other ICU settings with different patient profiles, staffing levels, and resource availability. Finally, selection bias cannot be excluded, as only patients with complete and retrievable medical records were included in the analysis.

Conclusions

In summary, the median time to death among mechanically ventilated patients in this study was comparable to findings from similar settings. Several factors, including low platelet count (<150,000/mm³), cardiovascular failure during mechanical ventilation, cardiopulmonary resuscitation, vasopressor use, and higher fraction of inspired oxygen at initiation, were associated with an increased hazard of death, while receipt of gastrostomy was associated with a lower hazard of death.

However, these associations should be interpreted cautiously, as many of these variables likely reflect underlying severity of illness rather than direct causal effects. In particular, interventions such as vasopressor use and cardiopulmonary resuscitation may represent markers of critical clinical status.

From a clinical perspective, the findings highlight the importance of close monitoring and timely management of critically ill patients, especially those presenting with indicators of severe illness. Nevertheless, causal inferences cannot be established due to the observational nature of the study.

Future research using prospective study designs and incorporating standardized severity scoring systems, such as APACHE II is recommended to better clarify these relationships.

Footnotes

We would like to express our sincere gratitude to Haramaya University, College of Health and Medical Sciences, School of Public Health, and the Postgraduate Program Directorate for providing the opportunity to conduct this research. We also extend our appreciation to the staff and administration of Hiwot Fana Comprehensive Specialized Hospital for their cooperation during data collection.

ORCID iDs

Behailu Hawulte Ayele

Adisu Birhanu Weldesenbet

Ethical considerations

Ethical clearance was obtained from the Institutional Health Research Ethics Review Committee (IHRERC) of Haramaya University, College of Health and Medical Sciences (Ref. No: IHRERC/305/2024). As the study used secondary/anonymized data and involved no direct contact with participants, the requirement for written informed consent was waived by the Ethics Review Committee. Permission to access patient records was obtained from the hospital administration. All data were collected using a structured data extraction tool, and no personal identifiers (such as names, medical record numbers, or contact information) were recorded. To ensure confidentiality, data were anonymized and stored securely with access limited to the research team. The study was conducted in accordance with relevant ethical guidelines and regulations for research involving human subjects.

Authors’ contributions

Conception and design of the work, acquisition of data, analysis, and interpretation of data were done by Rahma Esmael. Data curation, drafting the article, revising it critically for intellectual content, validation, and final approval of the version to be published were done by Adisu Birhanu Weldesenbet, Behailu Hawulte Ayele and Rahma Esmael. All authors read and approved the final manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

All necessary information’s were included with the manuscript.*

Appendix

References

Debebe

Goffi

Haile

, et al. Predictors of ICU Mortality among Mechanically Ventilated Patients: An Inception Cohort Study from a Tertiary Care Center in Addis Ababa, Ethiopia. Crit Care Res Pract 2022; 2022: 7797328. https://doi.org/10.1155/2022/7797328

Vincent

Marshall

Namendys-Silva

, et al. Assessment of the worldwide burden of critical illness: the intensive care over nations (ICON) audit. Lancet Respir Med 2014; 2(5): 380–386. https://doi.org/10.1016/S2213-2600(14)70061-X

Kübler

Maciejewski

Adamik

, et al. Mechanical ventilation in ICUs in Poland: a multi-center point-prevalence study, Med Sci Monit 2013; 19: 424–429. https://doi.org/10.12659/MSM.883930

Tobi

Ekwere

Ochukpe

CJJACR

. Mechanical ventilation in the intensive care unit: a prospective study of indications and factors that affect outcome in a tertiary hospital in Nigeria. 2017; 8(718): 2.

Pisani

Algera

Neto

, et al. Epidemiological characteristics, ventilator management, and clinical outcome in patients receiving invasive ventilation in intensive care units from 10 Asian middle-income countries (PRoVENT-iMiC): an international, multicenter, prospective study. 2021; 104(3): 1022.

Liang

Dong

, et al. Prognostic factors associated with mortality in mechanically ventilated patients in the intensive care unit: A single-center, retrospective cohort study of 905 patients. Medicine (Baltimore) 2019; 98(42): e17592. https://doi.org/10.1097/MD.0000000000017592

Esteban

Frutos-Vivar

Muriel

, et al. Evolution of mortality over time in patients receiving mechanical ventilation. American journal of respiratory and critical care medicine 2013; 188(2): 220–230. https://doi.org/10.1164/rccm.201212-2169OC

Zhu

Jiang

, et al. A contemporary assessment of acute mechanical ventilation in Beijing: description, costs, and outcomes. Critical care medicine 2017; 45(7): 1160–1167. https://doi.org/10.1097/CCM.0000000000002360

Parker

Mwachiro

, et al. Mortality prediction in rural Kenya: a cohort study of mechanical ventilation in critically ill patients. Crit Care Explor 2019; 1(12): e0067. https://doi.org/10.1097/CCE.0000000000000067

10.

Chiwhane

Diwan

SJTEJCCM

. Characteristics, outcome of patients on invasive mechanical ventilation: A single center experience from central India. The Egyptian Journal of Critical Care Medicine 2016; 4(3): 113–118. https://doi.org/10.1016/j.ejccm.2016.10.003

11.

Riviello

Kiviri

Fowler

, et al. Predicting mortality in low-income country ICUs: the Rwanda Mortality Probability Model. R-MPM) 2016; 11(5): e0155858. https://doi.org/10.1371/journal.pone.0155858

12.

Abate

Assen

Yinges

, et al. Survival and predictors of mortality among patients admitted to the intensive care units in southern Ethiopia: A multi-center cohort study, Ann Med Surg 2021;65:102318. Surgery. https://doi.org/10.1016/j.amsu.2021.102318

13.

Demem

Tesfahun

Nigussie

, et al. Time to death and its predictors among adult patients on mechanical ventilation admitted to intensive care units in West Amhara comprehensive specialized hospitals, Ethiopia: a retrospective follow-up study. BMC Anesthesiol 2024; 24(1): 114. https://doi.org/10.1186/s12871-024-02495-9

14.

Sison

Sivakumar

Caufield-Noll

, et al. Mortality outcomes of patients on chronic mechanical ventilation in different care settings: A systematic review. Heliyon 2021; 7(2): e06230. doi:10.1016/j.heliyon.2021.e06230.

15.

Nega

Sibhat

Mekonnen

, et al. Survival status and predictors of mortality among mechanically ventilated COVID-19 patients in Addis Ababa COVID-19 Care Centers, Ethiopia: A survival analysis. F1000Res 2023; 11: 1329. https://doi.org/10.12688/f1000research.124586.1

16.

Colpan

Akinci

Erbay

, et al. Evaluation of risk factors for mortality in intensive care units: a prospective study from a referral hospital in Turkey. Am J Infect Control 2005; 33(1): 42–47. https://doi.org/10.1016/j.ajic.2004.09.005

17.

Endeshaw

Fekede

Gesso

, et al. Survival status and predictors of mortality among patients admitted to surgical intensive care units of Addis Ababa governmental hospitals, Ethiopia: a multicenter retrospective cohort study. Front Med (Lausanne) 2023; 9: 1085932.

18.

Lee

Kim

Jeong

, et al. Outcomes and prognostic factors in patients with prolonged acute mechanical ventilation: a single-center study in Korea. J Crit Care 2015; 30(5): 1016–1020. https://doi.org/10.1016/j.jcrc.2015.05.020

19.

Urner

Jüni

Hansen

, et al. Time-varying intensity of mechanical ventilation and mortality in patients with acute respiratory failure: a registry-based, prospective cohort study. Lancet Respir Med 2020; 8(9): 905–913. https://doi.org/10.1016/S2213-2600(20)30325-8

20.

Machado-Alba

Usma-Valencia

Sánchez-Ramírez

, et al. Factors associated with survival in patients undergoing invasive mechanical ventilation in an intensive care unit in Colombia, 2017–2018: a retrospective cohort study. Drugs Real World Outcomes 2021; 8(3): 417–425. https://doi.org/10.1007/s40801-021-00241-y

21.

Demass

Guadie

Mengistu

, et al. The magnitude of mortality and its predictors among adult patients admitted to the Intensive care unit in Amhara Regional State. Northwest Ethiopia 2023; 13(1): 12010. https://doi.org/10.1038/s41598-023-39190-7

22.

Demem

Tesfahun

Nigussie

23.

Fialkow

Farenzena

Wawrzeniak

, et al. Mechanical ventilation in patients in the intensive care unit of a general university hospital in southern Brazil: an epidemiological study. Clinics 2016; 71: 144–151. https://doi.org/10.6061/clinics/2016(03)05

24.

Urner

Jüni

Hansen

25.

Machado-Alba

Usma-Valencia

Sánchez-Ramírez

, et al. Factors Associated with Survival in Patients Undergoing Invasive Mechanical Ventilation in an Intensive Care Unit in Colombia, 2017-2018: A Retrospective Cohort Study. Drugs Real World Outcomes 2021; 8(3): 417–425. https://doi.org/10.1007/s40801-021-00241-y

26.

Nega

Sibhat

Mekonnen

, et al. Survival status and predictors of mortality among mechanically ventilated COVID-19 patients in Addis Ababa COVID-19 Care Centers, Ethiopia: A survival analysis. F1000Res 2023.

27.

Teasdale

Jennett

. Assessment of coma and impaired consciousness: a practical scale. The lancet 1974; 304(7872): 81–84. https://doi.org/10.1016/s0140-6736(74)91639-0

28.

Karthikeyan

Kadhiravan

Deepanjali

, et al. Case-mix, care processes, and outcomes in medically-ill patients receiving mechanical ventilation in a low-resource setting from southern India: a prospective clinical case series. PLoS One 2015; 10(8): e0135336. https://doi.org/10.1371/journal.pone.0135336

29.

Adams

Rogers

Schuler

, et al. Association between peripheral blood oxygen saturation (SpO2)/fraction of inspired oxygen (FiO2) ratio time at risk and hospital mortality in mechanically ventilated patients. The Permanente Journal 2020; 24: 19.113. https://doi.org/10.7812/TPP/19.113

30.

Hueda-Zavaleta

Copaja-Corzo

Miranda-Chávez

, et al. Determination of PaO2/FiO2 after 24 h of invasive mechanical ventilation and ΔPaO2/FiO2 at 24 h as predictors of survival in patients diagnosed with ARDS due to COVID-19. PeerJ 2022; 10: e14290. https://doi.org/10.7717/peerj.14290

31.

de Jonge

Peelen

Keijzers

, et al. Association between administered oxygen, arterial partial oxygen pressure and mortality in mechanically ventilated intensive care unit patients. Critical care 2008; 12(6): R156. https://doi.org/10.1186/cc7150

32.

Burunsuzoğlu

Saltürk

Karakurt

, et al. Thrombocytopenia: A Risk Factor of Mortality for Patients with Sepsis in the Intensive Care Unit. Turk Thorac J 2016; 17(1): 7–14. https://doi.org/10.5578/ttj.17.1.002

33.

Russell

Holst

Kjeldsen

, et al. Risks of bleeding and thrombosis in intensive care unit patients with haematological malignancies. Ann Intensive Care 2017; 7(1): 119. https://doi.org/10.1186/s13613-017-0341-y

34.

Kostadima

Kaditis

Alexopoulos

, et al. Early gastrostomy reduces the rate of ventilator-associated pneumonia in stroke or head injury patients. Eur Respir J 2005; 26(1): 106–111. https://doi.org/10.1183/09031936.05.00096104