Study of incidental errors that occur in the intensive care unit of Ghazi Al-Hariri Teaching Hospital

Abstract

This study aims to estimate the rate of incidental errors (InE) in the intensive care unit (ICU) and assess the impact of these errors on patient mortality. A prospective cohort study was conducted in the ICU of a tertiary hospital. The study included InE that occurred during patients’ admission to the ICU. A total of 1026 cases were admitted to the ICU; InE happened in 142 cases (13.84%). The total number of deaths was 306 cases (29.8% of total admissions), with 40 cases (28.2%) dying in the InE group. In the InE group, there is no difference in age, sex, and cause of admission regarding mortality. The reason for admission of those cases with adverse events was variable; the most common cause of admission was medical conditions requiring ICU, followed by traumatic brain injury and postoperative observation. Regarding the intervention required, the use of a bronchoscope and paracentesis was significantly associated with increased mortality, while receiving plasma pheresis was significantly associated with lower mortality. The mean APACHE II score was higher in cases that resulted in death (40.80 ± 4.72 vs 31.67 ± 8.97), with a range of 15% to 50%. Regarding incidental adverse events, bedsores were the most common adverse event in cases that resulted in death, followed by endotracheal tube blockage, tracheostomy blockage, and pneumothorax. At the same time, only blocked tubes and bed sores were significantly higher in the cases that died. The reported InE in this study were not associated with increased mortality. Mortality was more frequent in cases of blocked endotracheal tubes and cases that developed bed sores.

1. Introduction

Intensive care units (ICUs) represent a triumph of medicine, enabling the support and replacement of many bodily functions. And yet, these actions are not cost-free; nearly every intervention has side effects and risks, which must be mitigated. While rapid decision-making is often possible and desired in managing emergency center or ward patients, this approach can be prone to errors in the ICU environment and must be used in the appropriate clinical situations.^[1] Intensive care encompasses a spectrum of treatment from patient admission sources, necessitating regular vital sign assessments, invasive hemodynamic monitoring, intravenous pharmacotherapy, fluid management, and ventilatory and nutritional support to ensure safe and successful outcomes.^[2] Access to critical care is an essential element of healthcare systems; therefore, critically ill patients are admitted to ICUs to mitigate morbidity and death.^[3]

Human error is unavoidable in intricate systems, and the ICU is one such area characterized by its complexity. Over 1,300,000 individuals admitted to ICUs in United States hospitals have sustained injuries from unintended accidents, resulting in numerous complications for patients and their families. The primary issue with the errors is the substantial financial burden imposed on the healthcare system, patients, and their families.^[4] The underlying cause of admission primarily influences the epidemiology of ICU admissions. Mortality in the ICU constitutes a worldwide burden. The variation occurs globally based on ICU infrastructure, staff availability, training, admission patterns, and underlying causes of ICU hospitalization.^[5] In industrialized regions such as North America, Oceania, Asia, and Europe, ICU death rates are comparatively low, recorded at 9.3%, 10.3%, 13.7%, and 18.7%, respectively. Conversely, in other parts of the world, including South America and the Middle East, mortality rates are significantly higher at 21.7% and 26.2%.^[6] Adverse incidents and complications commonly arise in the ICU. The occurrence may reach 2 errors per patient per day, with as many as 18% of ICU patients experiencing significant adverse effects.^[7]

Patient safety, a critical element of hospital performance, is receiving heightened scrutiny across all levels of the healthcare system, particularly in the formulation of healthcare policy and hospital quality assurance initiatives.^[8] Iatrogenic adverse events significantly contribute to death, morbidity, extended hospital stays, and healthcare expenditures. The intricate nature of care and the severity of conditions in ICUs lead to a significant risk of iatrogenic incidents.^[9] Enhanced error reporting and disclosure are the key detection strategies recommended in Australia,^[10] the US by The Joint Commission,^[9] the United Kingdom,^[11] and France.^[9]

Due to underlying comorbidities, organ failure, medical equipment, and polypharmacy, critically ill patients face a heightened risk of adverse events, particularly adverse drug events.^[12–14] Moreover, the life-sustaining interventions and intricate routine care employed in ICUs present several opportunities for medical errors.^[15] Initiatives have been undertaken to establish uniform definitions of medical errors. For medical errors to serve as effective indicators, they must be prevalent, avoidable, reproducible, readily diagnosable and collectible, correlated with significant morbidity and mortality, and reportable without fear of retribution. A universally recognized list of medical errors displaying these features in critical care does not exist.^[16,17]

The current study aims to estimate the rate of incidental errors (InE) in the ICU, assessing the impact of these errors on patients’ mortality.

2. Methods

2.1. Study design and settings

A prospective cohort study was conducted at the ICU of Ghazi Al-Hariri Teaching Hospital, part of the Medical City Teaching Complex. The study commenced on January 5, 2023, and was completed on January 5, 2025.

The total number of cases admitted to the ICU during the eligible period of study was 1026 cases; in this research, the data has been collected from all cases who experienced an InE during their period of admission in the ICU, as when informed by the ICU staff (senior, intensivist, resident doctor, and ICU nurse) developed InE (that reported by nurses, residents, and senior intensivists) according to their registry and their daily monitoring work up.

This study investigated the data by filling a predesigned forma that included patient sex, age, date of admission, cause of admission, the use of interventional procedures (continuous renal replacement therapy [CRRT], plasmapheresis, bronchoscopy, and paracentesis), the incidental and medical errors that specifically investigated were airway problems (blockage of tube, O₂ failure, and pneumothorax), accidental removal of appliances (endotracheal, tracheostomy tubes, central venous line, feeding tubes, and Foley catheter), errors regarding drug administration, and bed sores. APACHE II score for all cases after measurement of temperature, mean arterial pressure, heart rate, respiratory rate, oxygenation, arterial pH, HCO₃, serum electrolytes (sodium, potassium), serum creatinine, white blood cell count, hematocrit percent, and Glasgow coma scale. Cases were followed until discharge or death, and the outcome was included in the form.

2.2. Eligibility criteria

The study included adults (aged ≥ 18 years) with critically ill cases who experienced an incidental iatrogenic error. The following cases were excluded from the study: those hospitalized for <24 hours, patients’ refusal to participate, and incomplete data.

2.3. Ethical approval

The study was approved by the Research Ethics Committee of the Department of Medicine, College of Medicine, University of Baghdad (Approval Number: 12, Date: December 5, 2023). Written informed consent was obtained from all participants, and the study was done following the Declaration of Helsinki.

2.4. Statistical analysis

The data were introduced into Microsoft Excel 2019 software (Microsoft Corporation, Redmond), and statistical analysis was performed using GraphPad Prism 10.1 software (GraphPad Software, LLC, Boston). The data are presented as mean ± standard deviation or number (percentage). The significance level (P-value < .05) was presented, and the significance level was calculated using Chi-square, Fisher exact, and independent t-test as appropriate.

3. Results

During the study period, 1026 cases were admitted to the ICU; InE happened in 142 cases (the incidence of InE was 13.84%). There was no significant difference between cases with or without InE in their age (45.77 ± 19.35 vs 48.93 ± 17.61 years) and their sex (female to male ratio 0.8 and 1.07). The total number of deaths was 306 cases (29.8% of total admissions); cases with an InE that died were 40 cases, representing 28.2% of the InE (13.1% of total deaths and 3.9% of total ICU admissions). The InE did not associate with an increased rate of death in comparison to those without InE (P-value = .642), as seen in Table 1.

Table 1.

Demographic variables.

Variables	No errors	Incidental errors	P-value
Number	884	142	–
Age (yr)*	48.93 ± 17.61	45.77 ± 19.35	.051§
18–29 yrs†	154 (17.4%)	40 (28.2%)
30–39 yrs†	156 (17.6%)	22 (15.5%)
40–49 yrs†	136 (15.4%)	22 (15.5%)
50–59 yrs†	134 (15.2%)	18 (12.7%)
60–69 yrs†	156 (17.6%)	16 (11.3%)
70–79 yrs†	148 (16.7%)	24 (16.9%)
Sex†			.080‡
Female	456 (51.6%)	62 (43.7%)
Male	428 (48.4%)	80 (56.3%)
Outcome†			.642‡
Survived	618 (69.9%)	102 (71.8%)
Died	266 (30.1%)	40 (28.2%)

^*Mean ± SD.

^†Number (%).

^‡Chi-square test.

^§Independent t-test.

In the InE group, there is no difference in age, sex, and cause of admission regarding mortality. The reason for admission of those cases with adverse events was variable; the most common cause of admission was medical conditions requiring ICU, followed by traumatic brain injury and postoperative observation. Regarding the intervention required, using a bronchoscope and paracentesis was significantly associated with increased mortality, while receiving plasma pheresis was significantly associated with lower mortality, as seen in Table 2.

Table 2.

Assessment of demographics, cause of admission, and type of intervention in patients who did not survive.

Variables	Alive	Died	P-value
Number	102	40	–
Age (yrs)*	44.33 ± 19.54	49.45 ± 18.58	.157§
18–29 yrs†	30 (29.4%)	10 (25.0%)	.045‡
30–39 yrs†	20 (19.6%)	2 (5.0%)
40–49 yrs†	14 (13.7%)	8 (20.0%)
50–59 yrs†	10 (9.8%)	8 (20.0%)
60–69 yrs†	14 (13.7%)	2 (5.0%)
70–79 yrs†	14 (13.7%)	10 (25.0%)
Sex†			.192‡
Female	48 (47.1%)	14 (35.0%)
Male	54 (52.9%)	26 (65.0%)
Cause of admission†			.136‡
Postoperative observation	28 (27.5%)	6 (15.0%)
Traumatic brain injury	26 (25.5%)	8 (20.0%)
Medical conditions requiring ICU	48 (47.1%)	26 (65.0%)
Interventions required†
CRRT	38 (37.3%)	22 (55.0%)	.054‡
Plasmaphereses	24 (23.5%)	2 (5.0%)	.010‡
Bronchoscope	52 (51.0%)	34 (85.0%)	<.001‡
Paracenteses	0 (0%)	8 (20.0%)	<.001‡

CRRT = continuous renal replacement therapy, ICU = intensive care unit.

^*Mean ± SD.

^†Number (%).

^‡Chi-square test.

^§Independent t-test.

Regarding incidental adverse events, bedsores were the most common adverse event in cases that resulted in death; next in frequency were endotracheal tube (ETT) blockage, tracheostomy blockage, and pneumothorax. At the same time, only blocked tubes and bedsores were significantly more prevalent in cases that resulted in death, as shown in Table 3.

Table 3.

Incidental adverse events.

Variables*	Alive	Died	P-value
Number	102	40	–
Accidental disconnection	10 (9.8%)	2 (5.0%)	.355†
Accidental extubation	2 (2.0%)	2 (5.0%)	.315†
Blocked tubes	26 (25.5%)	24 (60.0%)	<.001†
Blocked tracheostomy	48 (47.1%)	18 (45.0%)	.825‡
Bedsore	30 (29.4%)	26 (65.0%)	<.001‡
Accidental removal of the CV line	2 (2.0%)	0 (0%)	.999†
Leaking jejunostomy	14 (13.7%)	10 (25.0%)	.107‡
Discharge from jejunostomy	16 (15.7%)	10 (25.0%)	.197‡
Blocked jejunostomy	10 (9.8%)	4 (10.0%)	.972†
Pneumothorax	26 (25.5%)	16 (40.0%)	.088‡
Drug look-alike	4 (3.9%)	2 (5.0%)	.774†
O2 failure	4 (3.9%)	2 (5.0%)	.774†
Tracheal stenosis	2 (2.0%)	0 (0%)	.999†

^*Number (%).

^†Fisher-exact test.

^‡Chi-square test.

Regarding the incidental disconnection, it happened in 12 patients; 4 of them had a traumatic brain injury, 4 of them had Guillain-Barré syndrome, and 3 cases had a viral infection; all of them disconnected from the ventilator for <2 minutes, and no deaths occurred during or after this period. The last case was a 74-year-old male with multiple comorbidities, admitted to the ICU postoperatively (perforated viscus with sepsis); the tube was disconnected at the transfer for <2 minutes. The patient died on the 2nd day. Four cases were accidentally extubated; all were postoperative patients, were extubated at the theater, and reintubated after 2 to 3 minutes. Two died after 1 day of ICU admission. Other adverse events happened with varying frequency and are explained in further detail in Regarding the incidental disconnection, it happened in 12 patients; 4 of them had a traumatic brain injury, 4 of them had Guillain-Barré syndrome, and 3 cases had a viral infection; all of them disconnected from the ventilator for <2 minutes, and no deaths occurred during or after this period. The last case was a 74-year-old male with multiple comorbidities, admitted to the ICU postoperatively (perforated viscus with sepsis); the tube was disconnected at the transfer for <2 minutes. The patient died on the 2nd day. Four cases were accidentally extubated; all were postoperative patients, were extubated at the theater, and reintubated after 2 to 3 minutes. Two died after 1 day of ICU admission. Other adverse events happened with varying frequency and are explained in further detail in Table 3.

In the InE group, the mean APACHE II score was higher in cases that resulted in death (40.80 ± 4.72 vs 31.67 ± 8.97), with a range of 15% to 50%, as shown in Figure 1.

Figure 1.

Assessment of APACHE II score in patients with incidental errors.

4. Discussion

Human errors in healthcare socio-technical systems can seriously threaten patient safety.^[18] Detecting InE is another problem, as reporting may be ignored or run unnoticed in the best situations. In a postmortem study, the median error rate detected by autopsy was 23.5% (range, 4.1%–49.8%) for major medical errors,^[19] meaning many medical errors passed unnoticed.

This study aimed to estimate the rate of InE, their types, and risk factors for these errors. The incidence of InE was 13.8% in the current study. Ravi et al found that the rate of preventable and non-preventable ICU errors is 13%.^[20] Alghamdi et al found the human error rate in the neonatal ICU to be 14.6%.^[21]

During this study, 29.82% of patients died in the ICU, and 13.1% experienced InE; however, the ICU mortality rate was not statistically different between those with InE and those without these errors. This result indicates that InE did not increase mortality (i.e., nonfatal errors). Similarly, Ahmed et al found in their meta-analysis that InE in the ICU were not associated with a significant increase in mortality but with an increased hospital stay.^[22] This result could be explained by the fact that patients in the ICU were under intensive monitoring (although this monitoring did not completely prevent errors), which allowed for early detection of such errors.

While the overall analysis suggests that InE do not significantly affect mortality, certain errors, such as ETT obstruction and bed sores, appear to correlate with worse outcomes. The following could explain this discrepancy. Severity and Compounding Effects: Some errors may not be directly fatal but contribute to complications that escalate a patient’s condition. For example, ETT obstruction can lead to hypoxia, ventilator-associated pneumonia (VAP), or prolonged mechanical ventilation, which increases the likelihood of a fatal outcome. Underlying patient condition: ICU patients often have multiple underlying conditions. A seemingly minor error can worsen an already precarious situation. Bed sores, for instance, might not directly cause death, but they can lead to systemic infections like sepsis, which dramatically increases mortality risk. Prolonged ICU stay and associated risks: errors such as improper airway management or inadequate pressure ulcer prevention may extend a patient’s ICU stay, exposing them to secondary infections or complications that eventually lead to mortality. Error detection and late intervention: while ICU monitoring mitigates error-related harm to some extent, certain errors may only be detected at a late stage when they have already contributed to deterioration in the patient’s condition.

According to age, there was no difference in mortality rates. However, when the age groups were divided, the study found that mortality increased with advancing age, which may be related to the increased comorbidities that accompany aging. Nielsson et al found that the elderly had increased mortality in the ICU even after the elimination of the effect of other factors (confounders such as cause of admission and medical illnesses).^[23]

The mortality in cases of InE was not different according to the cause of admission. Although the cause of admission to the ICU had a significant effect on survival, the InE itself did not alter the mortality rate. No study in the literature was found to investigate this relationship. The requirement of CRRT was not associated with increased mortality in cases of InE. Nash et al found that CRRT is not associated with increased mortality in the ICU in general.^[24] Hall et al found that using CRRT in fluid-overloaded ICU patients was associated with decreased mortality.^[25] The requirement for plasma phrases in patients with InE did not increase the mortality rate. Strużyna et al found similarly that plasmaphereses were not associated with increased mortality in ICU patients.^[26]

The study revealed that cases requiring bronchoscopy had higher mortality rates than those that did not. However, this mortality could be attributed to the bronchoscopy itself, as the majority of cases, especially those with VAP, require bronchoscopy. Thus, the cause of death may be related to the primary diagnosis rather than an InE or bronchoscopy per se. Bruyneel et al found that bronchoscopy had a positive impact on COVID-19 patients’ survival.^[27]

Accidental disconnection from the ventilator was 8.5% of InE (1.17% of total ICU admissions); this rate was lower than in previous studies. Alonso et al found the rate of accidental disconnection (from all cases of InE) to be 26.9%, and all cases were considered non-harmful events.^[28] Ravi et al found that no Accidental disconnection from the ventilator went unnoticed in their study. These findings highlight the possibility that many incidental disconnections could be passed unreported, as these nonfatal errors are easily managed incidents.^[20]

The rate of accidental extubation was 2.8% of cases, representing 0.39% of total ICU admissions. The rate of extubation was not different statistically regarding mortality. Lin et al found the unplanned extubation rate to be 3.5%, with no increase in the mortality rate.^[29] Li et al found in their meta-analysis that the unplanned extubation rate was 6.69%, and most extubations were self-extubations (84.2%).^[30]

The rate of blocked ETTs was 35.2% of InE (4.87% of total ICU admissions), and mortality was significantly higher in these cases. The causes of ETT obstruction in the current study were attributed to disease-related factors (cases of sepsis and VAP/hospital acquired pneumonia ), a large quantity of bronchial secretions, hot weather, which led to the inspissation of the secretions, or insufficient sedation that caused kinking of the tube due to movement, obstruction due to patient position, and inadequate care, including insufficient suctioning of secretions. Notably, ETT obstruction developed despite the combination of routine aspiration and humidification protocols aimed at maintaining airway patency and reducing complications.

Alonso et al found that the rate of ETT obstruction was 1.8% in cases of InE, which increased to 15.8% in cases of prone position.^[28] Wiles et al found that cases of COVID-19, adult respiratory distress syndrome, and prolonged intubation were more prone to ETT obstruction due to the thick consistency of bronchial secretions.^[31] Lellouche et al suggested that better suctioning on the tube, with proper humidification of inspired air, could dramatically decrease the rate of ETT obstruction.^[32] Although the tracheostomy tube’s blockage rate was higher than that of ETT obstruction, no difference in mortality was found.

The rate of bed sores in the current study was 39.5% of cases, representing 5.45% of total ICU admissions. The mortality was significantly higher in cases of bed sores. In our center, the standard of care regarding optimal patient positioning to reduce bed sores was applied, which focuses on frequent repositioning, skin assessment, and support surfaces to minimize prolonged pressure on vulnerable areas. Woźniak et al found the rate of bedsore in ICU patient to be 8.8%.^[33] Manzano et al found bedsores associated with an increased overall mortality of ICU-admitted patients.^[34] Khoshfetrat et al found that old age and prolonged ICU stay are associated with an increased rate of bedsores.^[35] Accidental removal of the central line happened in 1 case; tracheal stenosis happened in 1 case. Tracheal stenosis tends to occur after prolonged mechanical ventilation.^[36] Jejunostomy complications were not associated with an increased rate of mortality. Similarly, Paydar et al found that complications of jejunostomy in critically ill patients did not increase the mortality rate.^[37]

The rate of iatrogenic pneumothorax was 29.6% of InE (4.09% of total ICU admissions), and no increase in mortality was found. Pnömotoraks et al found the total rate of iatrogenic pneumothorax in ICU to be 2.35%.^[37] The rate of errors in medication (drugs look alike) was 4.2% of the InE (0.58% of total ICU admissions). Moudgil et al found that the rate of drug look-alike errors was 19% of all medication errors in the ICU. No significant increase in mortality was found in the current study.^[38] The rate of O₂ system failure was 4.2% of the InE (0.58% of total ICU admissions); it was transient and did not affect mortality.

Regarding the APACHE II score, it was significantly higher in patients who died. Numerous studies have investigated the effect of this score, including the study conducted by Wong et al,^[39] Kumar et al,^[40] and Czajka et al.^[41] All the studies confirmed this correlation between increased mortality and APACHE II score.

This study highlights the prevalence and nature of InE in the ICU, underscoring the importance of implementing structured error-detection mechanisms and standardized safety protocols. While InE were not associated with increased mortality, they contributed to extended hospital stays and potential complications. Clinical practice can evolve based on these findings in several ways: enhanced error detection and reporting: the high percentage of errors going unnoticed underscores the necessity for improved surveillance systems. Integrating automated monitoring tools and fostering a culture of transparent reporting can help identify and address errors more effectively. Targeted preventive strategies: although the study does not specify direct intervention measures, its findings suggest that errors such as ETT obstruction and bedsores are correlated with increased mortality. Clinical teams can implement enhanced airway management, secretion clearance protocols, and stricter pressure ulcer prevention practices to mitigate their impact. Age-specific risk management: since mortality rates increase with age, ICU protocols can be refined to incorporate more individualized care strategies tailored to patients’ age and comorbidities. Process optimization for InE: although errors such as accidental ventilator disconnection and unplanned extubations did not increase mortality, minimizing their occurrence can reduce unnecessary interventions, resource utilization, and hospital burden. This study, therefore, advocates continuous improvements in ICU monitoring, error reporting, and preventive healthcare strategies.

Based on the current study funding, several recommendations can be drawn: better training of ICU staff to reduce preventable human errors, adaptation of more structured task reporting in ICU to minimize InE, offering better health education to the staff, and implementation of anonymous reporting system of InE, for better identification as it is the most important part to suggesting the solutions, and application of larger multicenter audit to estimate better the effect of these InE on patient survival.

4.1. Study limitations

The current study is a single-center study, which has several inherent limitations that affect its generalizability and applicability to broader populations. Additionally, the study included only 1 ethnic group (Iraqi patients), which also limits the generalizability of the study.

5. Conclusion

The reported InE in this study were not associated with increased mortality. Patients who required a bronchoscope may have had a more severe illness, which is thus associated with increased mortality rather than InE. Blocked ETTs and bedsores are both associated with increased mortality in patients with InE in the ICU.

Acknowledgments

We are grateful to the medical staff of the Critical Care unit, Ghazi Al-Hariri Teaching Hospital, Medical City Teaching Complex, and AlMustada University for their help completing this study.

Author contributions

Conceptualization: Waleed Ibraheem Ali, Layla Ali Hakeem, Zainab Taha Mohammed, Hayder Adnan Fawzi.

Data curation: Waleed Ibraheem Ali, Zainab Taha Mohammed, Hayder Adnan Fawzi.

Formal analysis: Hayder Adnan Fawzi.

Investigation: Waleed Ibraheem Ali, Layla Ali Hakeem, Zainab Taha Mohammed, Hayder Adnan Fawzi.

Methodology: Waleed Ibraheem Ali, Layla Ali Hakeem, Zainab Taha Mohammed, Hayder Adnan Fawzi.

Resources: Waleed Ibraheem Ali, Layla Ali Hakeem, Zainab Taha Mohammed, Hayder Adnan Fawzi.

Software: Waleed Ibraheem Ali, Layla Ali Hakeem, Zainab Taha Mohammed, Hayder Adnan Fawzi.

Supervision: Waleed Ibraheem Ali.

Validation: Waleed Ibraheem Ali.

Visualization: Waleed Ibraheem Ali.

Writing – original draft: Waleed Ibraheem Ali, Layla Ali Hakeem, Zainab Taha Mohammed, Hayder Adnan Fawzi.

Writing – review & editing: Waleed Ibraheem Ali, Layla Ali Hakeem, Zainab Taha Mohammed, Hayder Adnan Fawzi.