Süleyman Gökhan Kara1,2, Başak Bayram3, Şebnem Şakar Halaç4, Osman Sönmez5, Neşe Çolak6

1Department of Emergency Medicine, Ministry of Health Eskişehir City Hospital, Eskişehir, Türkiye
2Department of Stem Cell, Institute of Health Sciences, Eskişehir Osmangazi University, Eskişehir, Türkiye
3Chief Physician, İzmir Metropolitan Municipality Eşrefpaşa Hospital, İzmir, Türkiye
4Department of Emergency Medicine, Ministry of Health Fethiye State Hospital, Muğla, Türkiye
5Department of Emergency Medicine, Ministry of Health Tire State Hospital, İzmir, Türkiye
6Department of Emergency Medicine, İzmir Tepecik Training and Research Hospital, İzmir, Türkiye

Keywords: Drowning, emergency medicine, Glasgow Coma Scale, neurological outcomes, resuscitation

Abstract

OBJECTIVES: This study sought to identify risk factors linked to mortality, intensive care unit admission, and poor neurological outcomes among drowning victims and to find markers for safe discharge from the emergency department (ED).

METHODS: This retrospective cross sectional study evaluated all drowning victims presenting to both adult and pediatric EDs at a single center over an 11 year period. Variables such as arrival time at ED, age, type of water, comorbid diseases, vital signs, treatments given, and prehospital interventions were assessed.

RESULTS: The study found that early basic life support (BLS) by bystanders significantly improves survival and neurological outcomes. Respiratory rate, oxygen saturation, and Glasgow Coma Scale (GCS) were identified as independent risk factors for poor clinical outcomes. While the Szpilman clinical score is useful, it alone is not sufficient for predicting poor clinical outcomes.

CONCLUSIONS: For optimal management of drowning victims, immediate BLS is crucial. In the ED, respiratory rate, oxygen saturation, and GCS should be closely monitored. Drowning victims with a GCS of 15, normal respiratory rate, normal oxygen saturation, and Szpilman score below 3 can be safely discharged from the ED.

Introduction

Drowning is defined as the process of experiencing respiratory impairment from submersion/immersion in liquid.[1] The victim may survive or die after this process, but regardless of the outcome, has been involved in a drowning incident.[2] In studies to date, prognostic markers in drowning victims have been evaluated in prehospital and intensive care unit (ICU) settings. In addition, survival has been used as a predictor of prognosis in drowning; however, poor neurological outcomes in surviving patients have not been included in these evaluations. Because of these reasons, there are no defined discharge criteria for drowning patients in the emergency department (ED).

Although not suitable for the standard approach in the ED, there are two commonly used scoring methods for drowning. The Orlowski score, which is one of these scores, is used to determine the prognostic criteria of pediatric cases with out of hospital cardiac arrest (OHCA) due to drowning. The Orlowski score predicts mortality by incorporating several factors, including age, submersion time, no resuscitation period after the rescue, patient in a coma upon presentation to the ED, and arterial blood pH.[3] However, solely knowing the mortality rate without knowing neurological outcomes is insufficient to predict the prognosis and does not provide a significant benefit to patient management. Furthermore, this scoring system has limitations in its applicability to all age groups since the original Orlowski study only included pediatric age groups.[3] The Szpilman clinical score, the other scoring, predicts the mortality rate of drowning cases with the patient’s respiratory status and prehospital lung examination.[4] Unlike the Orlowski score, the Szpilman clinical score was created by incorporating data from drowning victims of all age groups. While the Szpilman score is a helpful tool in predicting mortality, it is limited in its scope due to its failure to provide information regarding neurological outcomes.

Prehospital findings, duration of submersion, and postrescue interventions can be used to predict mortality and thus determine the appropriate center for patient transfer. However, while these scores are effective in prehospital management, it is unclear how to decide on the discharge of patients after the presentation to the ED. Physical examination, laboratory testing, imaging studies, and interventions applied in the ED can be decisive in predicting the prognosis of the patient admitted to the ED.

In this study, data from drowning patients admitted to a single center adult and pediatric ED over an 11 year period were analyzed. Our objective was to identify the risk factors associated with mortality, ICU admission, and poor neurological outcome, as well as to identify markers for safe discharge from the ED. The results of this analysis can improve the potential to enhance the management of drowning victims in the ED, thereby resulting in improved patient outcomes.

Material and Methods

This retrospective cross sectional study examined all drowning victims reported to the Adult and Pediatric EDs at our hospital over an 11 year span. Cases were discovered via hospital databases utilizing relevant International Classification of Diseases, Tenth Revision codes. Cases without sufficient documentation or not pertaining to drowning were excluded from the study.

This retrospective investigation examined the medical records of drowning victims. The collected data encompassed: ED arrival time, characteristics of the drowning site, comorbidities, initial vital signs including GCS score, ED interventions and oxygenation methods, diagnostic evaluations (laboratory and imaging studies), therapeutic interventions, prehospital management (intubation/cardiopulmonary resuscitation [CPR]), time to mortality, clinical outcomes, and discharge modified Rankin Scale (mRS) scores. The mRS was used to score the range from asymptomatic (0) to death (6).[5] In this study, the mRS was used to assess neurological outcomes. Good neurological outcomes (mRS 0–1) indicated no disability, moderate neurological outcomes (mRS 2–3) represented mild disability without assistance, and poor neurological outcomes (mRS 4–5) signified moderate to severe disability requiring assistance. mRS 6 (death) was excluded from the poor neurological outcome group. As mRS is not routinely documented in ED records, it was retrospectively derived from physician notes, nursing assessments, and discharge summaries. For discharged patients, functional status was inferred from documented neurological examinations, GCS scores, and mobility assessments. Cases with insufficient data were excluded to ensure accuracy.

The Szpilman clinical score was used to classify drowning severity.[4] Although the Szpilman clinical score was not explicitly documented in the medical records, it was retrospectively determined based on available clinical data. The score was derived from physician notes, prehospital and ED assessments, documented respiratory status, and imaging findings. Cases with insufficient data to accurately assign a Szpilman score were excluded to maintain the reliability of the analysis.

SPSS V22.0 (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY, USA: IBM Corp.) was used for data analysis. To enhance comparability across the results, missing values of covariates were imputed using the Expectation– Maximization Algorithm. The maximum percentage of missing values was 10%. Missing variables were imputed based on the entire study population (n = 292).

Descriptive statistics

For categorical variables, numbers and percentages are given. Numerical variables were given as median, minimum, and maximum values for age, and interquartile range for other parameters. A one sample binomial test was used for homogeneity of categorical data, and Pearson Chi square Fisher’s exact test was used for comparison. After evaluating the variability of numerical variables with the Kolmogorov–Smirnov test, the Mann– Whitney U test was used for comparisons because the data did not conform to a normal distribution.

Factors affecting the patients’ outcomes (for death and survival) were evaluated by univariate regression analysis, and the results were given with an odds ratio, 95% confidence interval (CI), and P values. After evaluating demographic data, laboratory, physical examination, and chest radiography findings in terms of poor clinical outcome (death, discharge as mRS ≥4, or need for hospitalization from ED) with univariate regression analysis in adult patients, those with P > 0.20 were evaluated in multivariate regression analysis, and independent risk factors were identified. Sensitivity, specificity, negative and positive predictive values for poor clinical outcomes were determined by examining the relationship between Szpilman scores and poor clinical outcomes in these patients. The data were analyzed at a 95% confidence level, and if the P < 0.05, it was considered significant.

Ethical Approval: The study commenced subsequent to obtaining approval from the Non-Interventional Ethics Committee of Dokuz Eylül University Faculty of Medicine (Decision No: 2019/11-12, Date: April 24, 2019).

This manuscript was prepared in compliance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for observational studies.

The primary objective of this study was to identify prognostic factors associated with mortality, ICU admission, and poor neurological outcomes in drowning victims. In addition, the study aimed to determine clinical parameters that could guide safe discharge decisions in the ED. To achieve these objectives, the primary endpoints included mortality, both in hospital and at 1 year follow up, ICU admission rates, and neurological outcomes assessed using the mRS. Secondary endpoints focused on identifying independent risk factors for poor clinical outcomes, such as vital signs, GCS, and Szpilman clinical score, as well as establishing criteria for the safe discharge of drowning victims from the ED based on these parameters.

Results

This retrospective analysis included 292 drowning cases (median age 36.0 years, range: 9 months to 89 years), comprising 204 adult and 88 pediatric patients. Peak presentations to the ED occurred throughout the summer months (July 32.2%, August 29.8%, June 15.1%) between 12:00 and 18:00 on weekdays. The primary location for drowning incidents was maritime environments (86.6%).

In this study, 56.8% of drowning patients arrived at the ED without comorbidities. Hypertension was the most prevalent preexisting condition at 19.5%, followed by diabetes mellitus and coronary artery disease at 10.6% each, epilepsy at 9.6%, and other neurological disorders at 7.9%. However, no statistically significant association was found between comorbid conditions and mortality (P > 0.05), so these findings were not included in further analysis.

An examination of respiratory interventions indicated that 18.5% of drowning victims necessitated intubation (8.6% prehospital, 3.4% reintubation, and 9.9% during follow up). Among the surviving patients, respiratory support methods utilized during ED management comprised invasive mechanical ventilation (IMV) (16.4%), non IMV (22.3%), and high flow nasal oxygen (5.8%).

We investigated mortality and hospitalization outcomes related to drowning. The allocation of patient dispositions was as follows: 41.1% were admitted to the ICU, 14% were admitted to wards, and 40.8% were discharged from the ED. The 1 year mortality rate was 14.7%, comprising 4.1% mortality in the ED and 2.1% posthospitalization mortality directly linked to drowning. In 8.5% of cases, the causal relationship between drowning and mortality was indeterminate.

The neurological outcomes of drowning victims were evaluated using the mRS on discharge. Individuals having an mRS score of 6 (indicating death) were omitted from the poor neurological outcome cohort. The results indicated that 90.7% of patients attained good neurological outcomes, 1.4% moderate neurological outcomes, and 1.7% poor neurological outcomes, accompanied by a mortality rate of 6.2%. Among patients not necessitating basic life support (BLS), 96% experienced good neurological outcomes, whereas only 56.4% of those requiring BLS received good neurological outcomes [Figure 1].

Within the study cohort, 39 patients (13.3%) had unresponsiveness with absent or abnormal breathing. Bystander initiated BLS occurred in 21 cases (53.9%), whereas emergency medical services (EMSs) initiated BLS accounted for the remaining 46.1%. Bystander initiated BLS resulted in markedly better outcomes, achieving 90.5% good neurological results compared to 16.7% with EMS initiated BLS (P < 0.05). Furthermore, mortality rates were significantly reduced with bystander initiated BLS (9.5%) in contrast to EMS initiated BLS (66.7%) (P < 0.05) [Figure 2].

We examined the relationship between life support interventions and mortality rates in drowning incidents. Advanced cardiac life support (ACLS) was provided to 13 patients (4.5%) at the ED. Prehospital ACLS started by EMS in 11 cases and continued upon arrival at the ED. The mortality rate in this group was 90.9% (n = 10). Of the two patients undergoing in hospital ACLS for cardiac arrest in the ED, one survived, resulting in a 50% mortality rate. An analysis of BLS patients among all providers indicated a 35.9% mortality rate. While this finding suggests a potential benefit of early BLS compared to delayed ACLS, the retrospective nature of the study and the complexity of drowning related cardiac arrests make it difficult to draw definitive conclusions. However, bystander initiated BLS before EMS arrival was associated with improved survival rates and neurological outcomes, emphasizing the importance of early intervention in the OHCA survival chain.

In this study, poor clinical outcome was operationally defined as cardiac arrest, poor neurological outcome at discharge, or ICU admission. Among 204 adult drowning victims analyzed, 107 (52.5%) experienced poor outcomes. Poor clinical outcomes were significantly more prevalent in freshwater drowning incidents (P < 0.05), early morning ED presentations (P < 0.05), and cases with diffuse infiltration on chest radiography (P < 0.001).

The predictive value of the Szpilman clinical score was evaluated for drowning victims’ outcomes in the ED. A score ≥3 significantly correlated with poor clinical outcomes (P < 0.001) [Figure 3]. No mortality occurred in patients scoring <3; however, only 8.4% (n = 9) of patients with scoring ≥3 were discharged. A score of ≥3 exhibited a sensitivity of 66.4% (95% CI: 57.9– 74.2), specificity of 89.0% (95% CI: 83.0–93.5), positive predictive value of 84.3% (95% CI: 77.1–89.5), and negative predictive value of 75.0% (95% CI: 70.2–79.3) for poor clinical outcomes. In adult patients, the results were 67.3% (95% CI: 57.6–76.1), 92.8% (95% CI: 85.7–97.1), 91.1% (95% CI: 83.3–95.5), and 72.0% (95% CI: 66.1–77.1), respectively.

This study sought to discover prognostic factors associated with poor clinical outcomes in adult drowning victims, omitting patients who arrived at the ED in cardiac and respiratory arrest. Univariate analysis of vital signs, physical examination results, and laboratory parameters was succeeded by multivariate regression analysis. The latter found respiratory rate, oxygen saturation, GCS score, and chest radiograph local infiltrates as independent risk factors for poor clinical outcomes.

Discussion

The data of our study show that drowning victims are most frequently presented to our ED in the afternoons and evenings of July and August. This can be explained by factors such as the increase in population in the region due to our hospital being located in popular holiday destinations, schools being closed during the summer months, and the opening of the sea season. Similarly, drowning accidents are often observed in summer months and the frequency of incidents increases from midday to evening, as shown in literature in around the world. In a study conducted in Türkiye, the most common months for drowning incidents were found to be July, June, and August, respectively.[6] In a study conducted in South Africa, it was reported that drowning incidents occur frequently during summer months and official holidays.[7] In a study examining drowning related deaths in children, it was found that they occurred between May and August and most of them took place between 12:00 and 20:00 on weekends.[8] These results indicate that EDs should be more prepared for drowning accidents during the summer months.

Many studies have shown an increased risk of drowning in adults and children diagnosed with epilepsy.[9,10] The global incidence of epilepsy is between 0.4% and 1%,[11] and our study observed that epilepsy patients accounted for 9.6% of drowning victims, with no deaths among drowning cases diagnosed with epilepsy. While our findings suggest that epilepsy may be a contributing factor to drowning incidents, the absence of mortality in this subgroup could be due to sample size limitations or unmeasured confounders. Further studies with larger cohorts are needed to determine the impact of epilepsy on drowning related mortality.

El Sibai et al. found that drowning victims with chronic diseases had a higher likelihood of poor clinical outcomes.[12] In contrast, our study found that chronic diseases did not affect mortality. However, our study only evaluated drowning victims presented to our ED, and we do not have data on patients who did not respond to resuscitation before arriving at the hospital. Due to this limitation, we can only comment on ED outcomes regarding comorbid illnesses.

Our study has shown that 53.9% of BLS performed on drowning victims are started by bystanders. In studies conducted in different countries, the rates of BLS by bystanders in OHCA cases related to drowning incidents were reported as 42.5% in Japan,[13] 47.1% in the USA,[14] and 74% in Sweden.[15] While the results of our study indicate that the BLS rates by bystanders in Türkiye are better compared to some other countries, there are also studies showing that the BLS rates by bystanders in OHCA cases are low in Türkiye. In a study conducted in Türkiye, it was reported that BLS was performed by a bystander in only 1.7% of OHCA cases.[16] The higher rates of BLS in drowning victims may be due to the nature of drowning accidents, which may make bystanders more willing to intervene.

Since there is no correlation between the lung X ray findings and arterial blood gas parameters in the initial evaluation of drowning victims, the use of lung X rays for predicting prognosis is not recommended.[17,18] This investigation of chest radiograph findings in drowning victims (n = 278, excluding 14 cases without radiographs) revealed that 32.4% exhibited normal chest radiographs. Bilateral infiltrates were prominent among mortalities at 80%, in contrast to unilateral infiltrates at 20%. The total mortality rate was 5.6% in patients with bilateral infiltrates and 4.4% in those with unilateral infiltrates. Although radiographic abnormalities were associated with poor clinical outcomes and functioned as an independent risk factor in the whole cohort, they did not achieve statistical significance in forecasting outcomes for patients who did not necessitate CPR or intubation in the ED. These findings support prior research, demonstrating that chest radiographs alone are inadequate for prognostic assessment in drowning mortalities.

The Szpilman clinical scoring system, established by Szpilman in 1997, predicts the mortality of drowning victims at the accident site by evaluating vital signs and physical examinations.[3] Mott and Latimer investigated the effectiveness of the Szpilman clinical scoring system and recommended that victims with grade 1 be left at the accident site, those with grade 2 be taken to the ED, and those with grade 3 or higher be admitted to the ICU.[19] Analysis indicated no mortality among patients with Szpilman scores below 3, but only 8.4% of patients with scores of 3 or higher were discharged from the ED. Poor clinical outcomes were noted in 25% of participants with scores <3 compared to 84.3% of those with scores ≥3 (P < 0.001). The Szpilman score exhibited a high specificity of 89.0% but a low sensitivity of 66.4% in forecasting poor clinical outcomes. A Szpilman score ≥3, however inadequate as the only discharge criterion, is a significant clinical indicator for emergency physicians.

Previous studies evaluating prognostic indicators in drowning victims have identified initial vital signs, GCS, pupillary response, and clinical scoring systems (APACHE 2) as valuable predictors.[17,20] Our analysis revealed 100% survival among patients presenting with normal respiratory rate, oxygen saturation, GCS of 15, and clear lung X ray. Following the exclusion of patients requiring initial CPR or intubation, we performed univariate regression analysis on demographic data, vital parameters, laboratory values, and physical examination findings [Table 1]. Multivariate regression analysis of statistically significant variables identified respiratory rate, oxygen saturation, and GCS as independent predictors of mortality [Table 2]. Our findings demonstrate that respiratory rate, oxygen saturation, and GCS serve as significant prognostic indicators for drowning victims in the ED. However, blood gas parameters and electrolyte levels were not found to be predictive of adverse outcomes in the adult population.


Limitations

This study has several limitations inherent to its retrospective nature and data collection methodology. The single center design and exclusive focus on ED presenting drowning cases potentially restrict result generalizability. In addition, the absence of prehospital mortality data and comprehensive neurological outcome assessments limits our understanding of drowning associated morbidity.

Conclusion

Our study demonstrates that bystander initiated BLS plays a crucial role in improving the survival and neurological outcomes of drowning victims. To enhance survival rates, public education on BLS is essential, particularly in regions with high drowning incidence. Based on our findings, respiratory rate, oxygen saturation, and GCS should be closely monitored in the ED, as they are independent predictors of poor clinical outcomes. While the Szpilman clinical score provides valuable prognostic insight, it is not sufficient as a standalone tool for predicting poor outcomes. Our results suggest that drowning victims presenting with a GCS of 15, a respiratory rate within the normal range, normal oxygen saturation, and a Szpilman score below 3 can be safely discharged from the ED.

How to cite this article: Kara SG, Bayram B, Halaç ŞŞ, Sönmez O, Çolak N. Predicting mortality and safe discharge in drowning victims: A comprehensive analysis of neurological and clinical outcomes in the emergency department. Turk J Emerg Med 2025;25:208-15.

Ethics Committee Approval

Ethical approval for this study was obtained from the Non‑Interventional Ethics Committee of Dokuz Eylül University Faculty of Medicine (Decision No: 2019/11‑12, Date: April 24, 2019).

Author Contributions

S.G.K. (Süleyman Gökhan Kara): Conceptualization (lead), Methodology (lead), Data Curation (lead), Writing – Original Draft (lead), Investigation (equal).
B.B. (Başak Bayram): Project Administration (lead), Supervision (lead), Validation (equal), Writing – Review and Editing (equal).
Ş.S.H. (Şebnem Şakar Halaç): Formal Analysis (equal), Data Collection (equal), Writing – Review and Editing (equal).
O.S. (Osman Sönmez): Resources (equal), Visualization (equal), Writing – Review and Editing (supporting).
N.Ç. (Neşe Çolak): Project Administration (equal), Supervision (equal), Funding Acquisition (equal).

Conflict of Interest

None Declared.

Financial Disclosure

None.

References

  1. van Beeck EF, Branche CM, Szpilman D, Modell JH, Bierens JJ. A new definition of drowning: Towards documentation and prevention of a global public health problem. Bull World Health Organ 2005;83:853 6.
  2. Orlowski JP. Prognostic factors in pediatric cases of drowning and near drowning. JACEP 1979;8:176 9.
  3. Szpilman D. Near drowning and drowning classification: A proposal to stratify mortality based on the analysis of 1,831 cases. Chest 1997;112:660 5.
  4. Papa L, Hoelle R, Idris A. Systematic review of definitions for drowning incidents. Resuscitation 2005;65:255 64.
  5. Orlowski JP, Szpilman D. Drowning. Rescue, resuscitation, and reanimation. Pediatr Clin North Am 2001;48:627-46.
  6. Çoban H. Drowning incidents in 2017. In: Drowning and Rip Currents Workshop. İstanbul, Turkey: Yıldız Technical University; 2018.p. 24–7.
  7. Saunders CJ, Adriaanse R, Simons A, van Niekerk A. Fatal drowning in the Western Cape, South Africa: A 7-year retrospective, epidemiological study. Inj Prev 2019;25:529-34.
  8. Brenner RA. Prevention of drowning in infants, children, and adolescents. Pediatrics 2003;112:440-5.
  9. Bell GS, Gaitatzis A, Bell CL, Johnson AL, Sander JW. Drowning in people with epilepsy: How great is the risk? Neurology 2008;71:578-82.
  10. Brehaut JC, Miller A, Raina P, McGrail KM. Childhood behavior disorders and injuries among children and youth: A populationbased study. Pediatrics 2003;111:262-9.
  11. World Health Organization. Epilepsy. Available from: https:// www.who.int/news-room/fact-sheets/detail/epilepsy. [Last accessed on 2023 Jun 03].
  12. El Sibai R, Bachir R, El Sayed M. Submersion injuries in the United States: Patients characteristics and predictors of mortality and morbidity. Injury 2018;49:543-8.
  13. Fukuda T, Ohashi-Fukuda N, Hayashida K, Kukita I. Association of bystander cardiopulmonary resuscitation and neurological outcome after out-of-hospital cardiac arrest due to drowning in Japan, 2013-2016. Resuscitation 2019;141:111-20.
  14. Tobin JM, Ramos WD, Pu Y, Wernicki PG, Quan L, Rossano JW. Bystander CPR is associated with improved neurologically favourable survival in cardiac arrest following drowning. Resuscitation 2017;115:39-43.
  15. Claesson A, Lindqvist J, Herlitz J. Cardiac arrest due to drowning – Changes over time and factors of importance for survival. Resuscitation 2014;85:644-8.
  16. Erdur B, Ergin A, Turkcuer I, Ergin N, Parlak I, Serinken M, et al. Evaluation of the outcome of out-of-hospital cardiac arrest resuscitation efforts in Denizli, Turkey. J Emerg Med 2008;35:321-7.
  17. Causey AL, Tilelli JA, Swanson ME. Predicting discharge in uncomplicated near-drowning. Am J Emerg Med 2000;18:9-11.
  18. Schmidt AC, Sempsrott JR, Hawkins SC, Arastu AS, Cushing TA, Auerbach PS. Wilderness Medical Society practice guidelines for the prevention and treatment of drowning. Wilderness Environ Med 2016;27:236-51.
  19. Mott TF, Latimer KM. Prevention and treatment of drowning. Am Fam Physician 2016;93:576-82.
  20. Ballesteros MA, Gutiérrez -Cuadra M , M u ñ o z P , Miñambres E. Prognostic factors and outcome after drowning in an adult population. Acta Anaesthesiol Scand 2009;53:935-40.
Acknowledgments

We extend our gratitude to the teams at Dokuz Eylül University Hospital Adult and Pediatric Emergency Departments for their unwavering support throughout this study. Special thanks to our colleagues for their invaluable insights and critiques, which greatly enriched our research.