Original Research•

Factors associated with cost of adverse events in the intensive care unit: an economic evaluation performed in Bogotá, Colombia

Abstract

Background The costs of adverse health events represent a significant burden of disease. Management of these adverse events also entails high costs to healthcare systems, reaching up to US$110 516.25 in some countries. The intensive care unit is a context where they occur most frequently, with some studies reporting rates as high as 50% of the population. It is of utmost importance to examine which aspects of the care process related to patients or the events themselves have the most significant impact on costs. The objective of this economic evaluation is to establish the clinical and healthcare factors that determine the cost of adverse events in the intensive care units of two university hospitals in Bogotá, Colombia.

Methods We carried out an economic analysis study evaluating the clinical factors of patient care and institutions that have the most impact on the costs of adverse events. The population consisted of patients treated in the intensive care units of two university hospitals in Bogotá during 2019 and 2022. The outcome variable was the costs of adverse events, measured as direct costs through the perspective of the institutions providing health services. To estimate the direction and degree of association between clinical variables, adverse events and care with the costs of adverse events, a generalised linear regression gamma link function log was performed.

Results A total of 369 patients were included, 232 men (62.9%) and 137 women (37.1%), with a median age of 63 years. The median APACHE II classification system score was 13 points, and the median Charlson comorbidity index was 5. Of the adverse events, 252 (68.29%) were passively reported, 223 (60.43%) were preventable and 267 events (72.36%) were non-serious. The events had a median cost of US$54.32 per patient, and the variables related to higher cost were overall hospital stay, stay until the event, Charlson comorbidity index, reporting system, severity, overall stay until the event, musculoskeletal major disease categories, and events related to safe care, pharmacovigilance and technovigilance.

Conclusion This study was able to establish that overall hospital stay, Charlson index, reporting system and severity are positively related to the costs of adverse events. The factors that were found to be associated with lower costs were length of stay until the event, admission diagnosis related to the musculoskeletal system, and presenting an event related to safe care, pharmacovigilance and technovigilance.

What is already known on this topic

  • Adverse events in intensive care units (ICUs) are frequent and significantly increase healthcare costs.

  • Previous studies have shown that the incidence of adverse events can be as high as 50% in some settings and that these events often lead to substantial economic burden to healthcare systems, with costs reaching up to US$110 516.25 in some countries.

What this study adds

  • This study identifies specific clinical and healthcare factors that significantly influence the costs of adverse events in ICUs in Bogotá, Colombia.

  • The study establishes that overall hospital stay, Charlson comorbidity index, reporting system and severity of events are positively related to increased costs, while certain factors such as length of stay until the event and specific types of events are associated with lower costs.

How this study might affect research, practice or policy

  • Understanding the cost determinants of adverse events can inform targeted prevention and mitigation strategies, optimising resource allocation in healthcare settings.

  • The findings advocate for the integration of cost considerations into event analysis protocols, potentially leading to more cost-effective and safer patient care practices.

Introduction

Adverse events, also known as reportable events with undesired effects (REUNE),1 are defined by the WHO as the result of actions in healthcare that cause physical or psychological injuries to patients, occurring within an unintentional context, thus generating an increase in the costs of care and a decrease in the quality standards related to patient safety.2

A study carried out between 2001 and 2004 in 54 academic and community hospitals worldwide engaged in the Institute for Healthcare Improvement (IHI) critical care collaboratives estimated that the incidence of adverse events in intensive care units (ICUs) was 11.3%.3 In Colombia, a study conducted in a high-complexity hospital found that the incidence of adverse events in the ICU was 52.1%, where 38.4% of the events produced temporary damage that required intervention and 10.8% produced damage that required life-saving intervention.4

The consequences of REUNE range from minor injuries to death, the latter being one of the most feared outcomes.5–7 The Harvard Medical Practice Study showed the incidence of mortality associated with events to be 13.6% (95% CI 11.6 to 15.7).8 In the National Study on Adverse Effects Linked to Hospitalisation (ENEAS) carried out in Spain, an incidence of 4.4% (95% CI 2.8 to 6.5) was found.9

As for the economic impact, it is evident that costs vary from US$324 million per year in the USA10 to a total incremental cost of €88.26 million in Spain, which corresponds to an additional 6.7% of the total healthcare expenditure for 2010.5 In Colombia, there are reports of an increase in the cost of care due to adverse events, ranging from US$16 687 to US$110 516.25 in 2010.6 7

By knowing the impact of adverse events, the importance of implementing damage prevention and mitigation actions is recognised. It is essential to create a safety culture where scientifically supported recommendations are applied. Some of the most frequent criteria for prioritisation of events to be analysed include the severity of the event or the frequency of presentation11; however, it should be recognised that health resources are finite and should be used responsibly. Therefore, it is justified to recognise other criteria that facilitate prioritisation of events to be analysed. Providing essential cost information to inform intervention and policy development, as well as evaluation, is a research priority. Considering these, we proposed to establish the patient, adverse event and care factors that determine the cost of adverse events in ICUs.

Methods

We developed an observational analytic study of a retrospective cohort that used explanatory modelling of the costs of adverse events to estimate their association with patient characteristics, adverse event characteristics and adverse event care.

The study was conducted in the ICUs of two university hospitals in Bogotá. Data from patients hospitalised in the ICUs during 2019–2022 were included. Patients aged 18 years or older with adverse events occurring in the ICUs of two high-complexity university hospitals were included. Patients with adverse events occurring outside the study institution or outside the ICU were excluded.

We retrieved information on the following clinical and sociodemographic variables: sex, age, insurer (contributory regimen, out-of-pocket payment, prepaid plans, subsidised regimen), major disease categories (nervous system; eye; respiratory system; circulatory system; digestive system; musculoskeletal system; skin, subcutaneous tissue and breast; endocrine, nutrition and metabolism; male reproductive system; female reproductive system; pregnancy, childbirth and puerperium; blood and immune system; infectious and parasitic diseases; mental disorders; and urological disorders), COVID-19 diagnosis, Glasgow Coma Scale, Acute Physiology and Chronic Health Evaluation II (APACHE II), Charlson comorbidity index, overall hospital stay (length of stay in days), length of stay until the event (in days), readmission (any readmission occurring within 15 days of hospital discharge), device exposure, high-risk medications and surgical treatment.

The variables related to adverse events were the reporting personnel (doctors, researchers, nursing, auditing, laboratory), seriousness (any REUNE that results in death or endangers the patient’s life, or requires hospitalisation of the patient or prolongs an existing hospitalisation, or results in persistent or significant disability or incapacity, or results in a congenital anomaly or birth defect), preventability, reporting system (passive or active) and event classification (safe care, safe surgery, pharmacovigilance, infections, technology surveillance, reactovigilance, haemovigilance and biovigilance; see online supplemental appendix 1). Infections were defined as events that occur in a patient during the provision of healthcare services that were not present or were in the incubation period at the time of the patient’s admission to the healthcare facility. The outcome variable was the costs of adverse events, measured as direct costs through the perspective of the institutions providing health services.

To determine the cost of adverse events, a micro-counting technique was used by the authors of the study. This included the direct costs (medications, hotels, consultations, surgeries, procedures, among others) that were generated after the occurrence of the event. The perspective considered was that of the health service provider, for which the invoices paid directly by the institutions for the services were taken. In case of doubt as to whether a service was due to the occurrence of an event, this was consulted with clinical and billing experts from each of the corresponding institutions.

Having as a reference for the analysis a generalised linear regression gamma link function log and an estimated 35 independent variables, and in order not to incur p>N and according to the recommendations that allow the reproducibility of the model, it was decided to include 10 observations per variable in the study, which would result in a sample size of 350 patients; to avoid bias in the loss to follow-up, a final sample of 385 patients was adjusted by 10%.

To facilitate meaningful comparisons across the years 2019, 2020 and 2021, a cost adjustment was applied to account for the impact of inflation. The consumer price index (CPI) was used as the inflation metric. The adjustment involved calculating a factor by dividing the CPI of each specific year by the CPI of the base year, 2022. Subsequently, this factor was applied to each cost from the respective years, resulting in adjusted costs that reflect the purchasing power of the base year. This methodology ensures a more accurate assessment of cost trends by normalising for inflationary effects over the study period.

There was no loss to follow-up or missing data in the variables. A univariate analysis was performed according to the nature of the variable: in the case of categorical variables, absolute and relative frequencies were used, and for continuous variables measures of central tendency and dispersion, depending on the pattern of normality, which was established by means of the Shapiro-Wilk statistical test. Second, a bivariate analysis was performed by sex, severity of the event, type of event, diagnostic group and avoidability; Student’s t-test, Wilcoxon rank test and Kruskal-Wallis test (analysis of variance of ranks) were used. To estimate the direction and degree of association between clinical variables, adverse events and care with the costs of adverse events, a generalised linear regression gamma link function log was performed. To characterise sources of uncertainty in our analysis, a critical review of the model was conducted and experts in the field were consulted to validate the model assumptions, methods and results, helping to identify and address potential sources of uncertainty.

Patient and public involvement

No patient and public involvement.

Results

The population of this study consists of 369 patients who presented adverse events during the period from 2019 to 2022. The patients had a median age of 63 years, ranging from 18 to 95 years. The population consisted of 232 men (62.9%). The sociodemographic characteristics of the population are summarised in table 1. The median Charlson comorbidity index was 5, ranging from 0 to 15 points. Seventy patients (18.97%) had a diagnosis of SARS-CoV-2 infection. Hospital number 1 included 208 (56.37%) patients. The clinical characteristics of the population are summarised in table 1.

Table 1
|
Sociodemographic and clinical characteristics of the population

There was a wide range of events, including falls, pressure ulcers, delirium, hypotension, respiratory depression due to medication use, phlebitis and healthcare-associated infections (ventilation-associated pneumonia, bacteraemia, central venous catheter-associated bacteraemia, urinary tract infection associated with bladder catheter), among others, which were classified as safe care with 263 events (71.27%), infections with 36 events (9.76%), pharmacovigilance with 24 events (6.5%), technovigilance with 24 (6.5%), safe surgery with 19 events (5.15%) and haemovigilance with 3 events (0.81%). There were 223 preventable events (60.43%) and 267 non-serious events (72.36%). Of the 102 (27.64%) serious events, 71 (69.91%) were preventable. 252 events (68.29%) were passively reported. The characteristics of the REUNE are summarised in table 2.

Table 2
|
Characteristics of REUNE

The total cost generated by the events in the 369 patients was US$104 035.51 for the healthcare provider institutions, with a median of US$54.32, ranging from US$0 to US$10 600.54 (figure 1). In women, the median cost was US$58.87, ranging from US$0 to US$10 610.85. It was found that both medians correspond to the same population, that is, there is no statistically significant difference between the two groups. When age was divided into quartiles, the median cost in the first group (18–47 years) was found to be US$54.32. For the second quartile (48–63 years), the median was US$66.95, for the third (63–74 years) US$60.20 and for the last group (75–95 years) US$45.66. Likewise, it was found that there was no statistically significant difference between the medians of the groups. A summary of costs by group is shown in table 3.

Figure 1
Request permissionrequest permission icon
Figure 1

Cost generated by the reportable events with undesired effects (REUNE).

Table 3
|
Summary of costs by group

A generalised linear gamma model with log link that allows extending the general linear model is seen in table 4. It shows that overall hospital stays, Charlson index, reporting system and severity are positively related to costs. The factors that were found to be associated with lower costs were length of stay until the event, admission diagnosis related to the musculoskeletal system, or presenting an event related to safe care, pharmacovigilance and technovigilance.

Table 4
|
Generalised linear regression generalised gamma link function with a logarithmic function

Discussion

Statement of principal findings

This study found that the included population exhibits characteristics consistent with existing literature.4 12 13 In this study, 60.43% of the events were avoidable, aligning with the 60% found in the Iberoamerican Study of Adverse Events (IBEAS) and the studies by the Health Technology and Policy Evaluation Group.1 12 13 Regarding severity, 27.64% of the events were serious, a decrease compared with the 49.2% reported in an ICU in Medellín.4 Despite the lower prevalence in these institutions, severity was significantly related to the cost of the event.

Interpretation within the context of the wider literature

The study found that adverse events increased healthcare costs by an average of US$104 035.51, ranging from US$0 to US$10 660 per case. This is similar to the findings in Bogotá by Dr Pinzón and colleagues6 and Dr Ovalle.7 Notably, these previous studies were conducted in less complex hospitals and in earlier years, suggesting improvements in event management and reduced care costs. Furthermore, the median cost per patient in this study is lower than in other countries, such as in the USA and Denmark.14 15 However, it is important to avoid narrative fallacies; with an avoidability rate of 60%, clinics still incur significant costs on preventable outcomes.

The median cost of serious events was higher than non-serious events, consistent with previous studies.16 Interestingly, no significant differences were found between the median costs of preventable and non-preventable events. Literature indicates that the cost difference between these groups can be as high as US$5984 in the USA.17

The generalised linear gamma model with a log link identified overall hospital stay, Charlson index, reporting system and severity as factors determining higher costs of adverse events. These covariates could be used in a selection matrix to identify events for analysis by protocols like London or Failure Mode and Effects Analysis (FMEA). The general recommendation is to implement prevention and mitigation strategies. Factors negatively associated with higher costs included length of stay until the event, admission diagnosis related to the musculoskeletal system, and events related to safe care, pharmacovigilance or technovigilance.

Strengths and limitations

A limitation of this study is that it only considered direct costs, excluding costs associated with years of life lost per patient and intangible costs. This limitation challenges the modelling of costs, as it would be expected that no patient would have a quantified cost of 0.

Implications for policy, practice and research

Optimising health resources is imperative in the current healthcare system. Strategies are needed to enable clinics to achieve positive balances for investment in innovation and research and to implement improvements for preventing or mitigating adverse events. Currently, event studies prioritise severity and frequency; this study introduces a third tool—costs—which could greatly benefit these institutions.

One successful example is the adoption of the Lean healthcare principles. Lean methodologies focus on eliminating waste and improving processes. For instance, the Virginia Mason Medical Center implemented Lean strategies to redesign workflows, resulting in significant cost savings and improved patient outcomes.18

Understanding the costs associated with adverse events can directly inform quality improvement initiatives. By identifying high-cost events, healthcare providers can prioritise interventions that not only reduce expenses but also enhance patient outcomes. For example, allocating resources towards comprehensive training programmes for staff or investing in advanced monitoring technologies can prevent high-cost adverse events and improve overall care quality. This dual focus on cost and quality ensures that financial resources are used effectively to promote better patient care and safety.

Conclusions

These results provide insights into the factors influencing healthcare costs and adverse events in our setting. While specific contexts may vary across hospitals and health systems globally, the principles of cost management and quality improvement strategies discussed can be adapted and applied with consideration to local healthcare contexts. Future studies in diverse settings would further enhance the understanding of these dynamics and their applicability worldwide.

Overview

Article metrics
Altmetric data not available for this article.
Dimensionsopen-url