Direct healthcare costs associated with reported patient safety incidents in two tertiary hospitals in Colombia

Abstract

Background

Patient safety incidents are a leading cause of morbidity and mortality, and impose a substantial financial burden on health systems. This study aimed to determine the direct healthcare costs associated with reported patient safety incidents in two tertiary hospitals in Colombia between 2020 and 2022.

Methods

A retrospective cohort study was conducted using hospital discharge records classified by diagnosis-related groups and linked to patient safety incidents reports. The outcome was the direct cost per hospital episode, adjusted for inflation and expressed in United States dollars as of October 2022. Descriptive, bivariate, and multivariate analyses were performed using a log-linear model with gamma distribution to estimate adjusted cost ratios.

Results

Among 68.521 hospital episodes, 9.002 (13.1%) included at least one patient safety incident report. Episodes with reported patient safety incidents accounted for 31.2% of total costs and had a median cost 2.3 times higher than episodes without reports ($2.799 vs. $1.194; p < 0.001). Most reported incidents (90%) caused negligible or mild harm but accounted for 86.7% of total patient safety–related costs. After adjustment, hospital episodes with patient safety incidents remained associated with higher costs (CR: 1.15; 95% CI: 1.12 to 1.18; p < 0.001). Longer hospital lenght of stay, admission to intensive care units, and invasive mechanical ventilation were the main cost drivers.

Conclusions

Patient safety incidents significantly increase hospitalization costs, even when harm is minimal. The findings highlight the cumulative financial impact of mild safety events and reinforce the need for systematic reporting and prevention strategies. Investments in patient safety represent not only an ethical imperative but also a potential cost-effective approach to improving health system sustainability.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-026-14212-y.

Introduction

Patient safety incidents are defined as events or circumstances that happen during the care process and could have resulted in, or did result in, unnecessary harm to a patient [1]. According to the World Health Organization (WHO), patient safety incidents are classified into three categories: near misses, in which the incident does not reach the patient; no-harm incidents, in which the incident reaches the patient without causing discernible harm; and harmful incidents, which result in patient harm [1]. Harmful incidents include adverse events, defined as patient safety incidents that lead to preventable harm [1]. These incidents may be associated with medication errors, healthcare-associated infections, venous thromboembolisms, falls, pressure ulcers, and other failures or omissions in care processes [2].

Patient safety incidents represent a major global health challenge across healthcare systems with different levels of resources. In high-income countries, including many members of the Organisation for Economic Co-operation and Development (OECD), approximately one in ten hospitalized patients experiences harm while receiving hospital care [3]. In hospitals in low- and middle-income countries, an estimated 134 million adverse events occur annually, resulting in approximately 2.6 million deaths [2]. Harm can be caused by a range of patient safety incidents, of which more than 40% are potentially preventable [4].

Hospital care is an essential component of healthcare systems, and its primary goal is to provide effective and safe medical care to patients. However, despite efforts to improve patient safety, patient safety incidents related to medical care continue to occur in hospital settings, resulting in negative consequences for both patients and healthcare institutions [2]. From an ethical perspective, the prevention of patient harm is grounded in the fundamental principles of healthcare ethics, including non-maleficence and beneficence, which obligate healthcare professionals, institutions, and health systems to minimize avoidable risks and protect patient safety [5]. Harm to patients burdens individuals, their families and the community. Therefore, optimizing safety is a fundamental responsibility of healthcare professionals, institutions, and the healthcare system.

Economically sound arguments can also be made in favor of safety, as harm to individuals burdens healthcare system resources and society in general [3, 6]. Beyond their clinical consequences, patient safety incidents impose a substantial economic burden on healthcare systems. Previous studies have shown that adverse events are associated with significant incremental costs at the episode level, typically ranging from approximately USD 2.000 to over USD 30.000 per event, depending on the type and severity of the incident [7, 8]. Higher costs have been reported for surgical complications, healthcare-associated infections, and hospital-acquired bloodstream infections [9]. Patient harm directly manifests in the need for additional treatment, diagnostic tests, hospital readmissions, prolonged hospitalization, and other uses of healthcare resources. In a broader approach, the economic effects of harm include prolonged morbidity and reduced productivity over the patients’ and their caregivers’ lifetimes. Additional system-level consequences include the implementation of risk management strategies, litigation, and claims costs, and the erosion of plublic trust in healthcare system and institutions, potentially affecting organizacional reputation and long-term sustainability [10–14].

In many high-income countries, particularly those whithin the OECD, approximately 15% of total hospital activity and expenditure is attributable to adverse events, underscoring their economic relevance at the system level [3]. It is important to distinguish between the occurrence of patient safety incidents and their detection or reporting, as not all incidents are detected and not all detected incidents are reported through institutional safety reporting systems. Consequently, analyses based on reported patient safety incidents reflect a subset of all incidents that occur during hospital care, rather than the total underlying burden of harm. Nevertheless, assessing the costs associated with hospital episodes with reported patient safety incidents might help to understand their financial impact, allocating resources more effectively, and developing quality improvement strategies focused on higher-risk areas. This study aimed to determine the direct healthcare costs associated with reported patient safety incidents in two tertiary hospitals in Colombia between 2020 and 2022.

Materials and methods

Study design

A retrospective cohort study of hospital episodes treated in two tertiary hospitals in Colombia between January 1, 2020, and October 31, 2022 was conducted, with follow-up from hospital admission to discharge or death. A hospital episode was defined as a single inpatient stay, including admissions through the emergency department with a length of stay longer than 24 h or direct inpatient admissions. The Diagnosis-Related Group (DRG) hospital discharges database was used, including their corresponding demographic, clinical, and cost variables for each hospital episode.

Hospital episodes were excluded if patient death ocurred within 24 h of a reported safety incident, if the episode requird referral or tranfer to another healthcare institutions, or if records had missing or insufficient patient identification or incomplete cost data.

Setting

The study was conducted using data from two tertiary-level hospitals providing high-complexity inpatient care. Both institutions have established patient safety programs with voluntary incident reporting systems aligned with the WHO International Classification for Patient Safety framework. The participating hospitals have fostered a positive reporting environment, in line with WHO recommendations [1]. Reported incidents are routinely reviewed by institutional patient safety committees responsible for classification, monitoring, and feedback as part of ongoing quality improvement activities.

Study variables

The outcome variable in the study was the direct cost of safety incidents related to healthcare reported during every hospital episode. Safety incident reports were taken from the patient safety program records and linked with the DRG database.

The primary independent variable analyzed in the study was the presence or absence of healthcare-related safety incidents, evaluated according to their severity level and incident type [15, 16], as classified by the patient safety program in the participating clinics and validated by the service or area assigned to develop the improvement plan. A safety incident was defined as an event that caused or could have caused unnecessary harm associated with the provided healthcare [1]. Additionally, the following independent variables were included: age, sex, DRG and major diagnostic category, Charlson comorbidity index, requirement for invasive mechanical ventilation, admission setting (emergency, scheduled, referral, or obstetric), need for intensive care unit (ICU) care, hospital, and healthcare insurance.

Sample size and sample selection

Based on literature reports, assuming the average approximate cost of care for patients without reported unsafe hospitalization-related actions is USD 6.332 with a standard deviation of USD 8.180 [6]. Similarly, for patients with reported unsafe hospitalization-related actions, the average incremental cost generated by adverse events per DRG is USD 6.943, with an approximate standard deviation of USD 10.400 [10]. With a sample size ratio (Group 2 with report / Group 1 without report) of 0.25 and 80% power at an IC 95%, the minimum sample size in each group to detect a difference in costs would be 2.626 patients with reported safety incident and 10.502 patients without reported safety incident during hospitalization. The entire study population was consecutively and sequentially selected from the defined cohort for the study, meeting predefined eligibility criteria. While a sample size estimation was conducted, it was considered the minimum number required to demonstrate differences between groups but our cohort was not limited to this estimated size.

Analysis

All analyses were conducted at the hospital episode level. A univariate and bivariate descriptive analysis was first performed to characterize the population. Continuous variables were summarized using medians and interquartile ranges (IQRs) when distributions were skewed, and means were reported for variables such as hospital length of stay to facilitate interpretation, accompanied by the 25th, 50th, and 75th percentiles. Categorical variables were expressed as absolute and relative frequencies. Comparisons between groups were made using the Wilcoxon rank-sum (Mann–Whitney U) test for continuous variables and the chi-square test for categorical variables. Statistical significance was defined as p < 0.05.

For the cost analysis, the starting point was the value in Colombian pesos (COP) reported by each institution for every patient hospital episode. Costs were adjusted to October 2022 values following the methodology recommended by the Central Bank of Colombia. Specifically, the October 2022 Consumer Price Index (CPI) was divided by the CPI corresponding to the month in which the costs were incurred [17]. This procedure allowed the value of one peso from a prior period to be expressed in current terms, thereby eliminating the effects of inflation and ensuring comparability across time. Each hospital cost was thus adjusted and expressed in constant Colombian pesos (October 2022 values).

Subsequently, costs were converted to U.S. dollars using the Representative Market Rate (TRM), which reflects the weighted average exchange rate in the Colombian foreign exchange market [18]. The mean TRM for the study period was used to obtain consistent and comparable values in constant U.S. dollars. This ensured that all costs were expressed in real terms and a standard currency for analysis.

Once standardized costs were obtained, non-parametric comparisons of median hospital costs were performed to assess differences according to the presence or absence of patient safety incident reports. Given the skewed distribution of the cost data, the median was selected as the measure of central tendency to best represent the typical cost per episode. Relative differences in costs between groups were estimated through median ratios.

Exploratory analyses examined the association between length of hospital stay and reported patient safety incidents, including the timing of incidents occurrence during hospitalization.

It is important to note that patient safety incidents in this study were identified through voluntary institutional reporting systems. Reported patient safety incidents therefore represent detected and documented events rather than the total underlying occurrence of unsafe care. Reporting is influenced by factors such as patient clinical complexity, documentation intensity, and organizational safety culture.

Because more complex, resource-intensive hospital episodes (for example, those involving ICU admission, invasive mechanical ventilation, or high-complexity DRGs) may both incur higher costs and have a greater likelihood of generating safety reports, the analysis was designed to estimate associations between reported patient safety incidents and hospital costs rather than to attribute costs causally to the incidents themselves.

Finally, a multivariate log-linear regression model (generalized linear model with gamma distribution and log link) was fitted to estimate adjusted cost ratios (CRs) and their 95% confidence intervals. The model included the key demographic, clinical, and institutional covariates described above to adjust for potential confounding. This multivariate analysis was exploratory in nature and aimed to identify adjusted associations between reported patient safety incidents and patient hospital episode costs; it did not seek to establish causal relationships. All statistical analyses were performed using Stata (StataCorp LLC, College Station, TX, USA).

Ethical approval

This study was reviewed and approved by the Research Ethics Committee of the Sanitas University Foundation (Fundación Universitaria Sanitas), under approval protocol number CEIFUS2729-23. The study was conducted in accordance with the Declaration of Helsinki and the International Council for Harmonization and Good Clinical Practice.

Results

Initially, 69.817 patient hospital episodes lasting more than 24 h were analyzed. Then, 1.264 records were excluded due to insufficient data, reducing the sample to 68.553 episodes. Subsequently, 32 records that did not belong to the included hospitals were removed, resulting in a final database of 68.521 episodes. Of these, 9.002 (13.14%) had patient safety incident reports and 59.519 (86.86%) did not (Fig. 1).

Fig. 1.

Selection of patient hospital episodes with lenght of stays > 24 h included in the study

Table 1 (Characteristics of patient hospital episodes with and without safety reports), compares the frequency of episodes of hospitalization with and without safety reports. It was observed that patients in episodes with safety reports tend to be older (median age of 63 years compared to 55 years), have a higher frequency of comorbidities (72.86% compared to 58.43%), and a higher proportion of emergency admissions (65.41% compared to 52.36%). Additionally, patients in hospital episodes with safety reports had a higher rate of infectious diseases (15.15% compared to 10.84%) and higher use of invasive mechanical ventilation (17.87% compared to 4.63%). On the other hand, patients in hospital episodes without safety reports had a higher percentage of obstetric diagnoses (13.85% compared to 5.61%).

Table 1.

Characteristics of hospital episodes with and without safety reports

Characteristics	Hospital episodes with safety reports (n = 9.002)	Hospital episodes without safety report (n = 59.519)
Age, median years [IQR]	63 [45,76]	55 [35,70]
Sex, no. (%)
Male	4.185 (46.49)	24.736(41.56)
Female	4.817 (53.51)	34.783(58.44)
Hospital, no. (%)
Hospital A	4.766 (52.94)	43.312(72.77)
Hospital B	4.236 (47.06)	16.207(27.23)
Diagnosis related group (DRG), no (%)
Clinical DRG - Major diagnostic category	4.549 (50.53)	27.609 (46.39)
Infectious Diseases	1.363(15.15)	6.446 (10.84)
Clinical and Surgical Diseases of the Cardiovascular System	562 (6.25)	5.154 (8.67)
Clinical and Surgical Diseases of the Digestive System	543 (6.03)	3.615 (6.08)
Other clinical DRGs	2.081(23.12)	12.394(20.82)
Procedural DRG - Major diagnostic category	3.798 (42.19)	21.740 (36.53)
Clinical and Surgical Diseases of the Digestive System	1.226 (13.62)	6.613 (11.11)
Clinical and Surgical Diseases of the Urinary and Genital Tract	351 (3.90)	4.549 (7.64)
Clinical and Surgical Musculoskeletal and Connective Tissue Diseases	656 (7.29)	3.672 (6.17)
Other procedural DRGs	1.565 (17.39)	6.906 (11.60)
Pregnancy, Childbirth, and Puerperium	505 (5.61)	8.236 (13.85)
Birth and Neonatal Period	146 (1.62)	1.873 (3.15)
Charlson Comorbidity Index, no. (%)
No comorbidity	2.443 (27.14)	24.741 (41.57)
One or two comorbidities	3.157 (35.07)	20.300 (34.11)
Three or four comorbidities	1.768 (19.64)	7.749 (13.02)
Five or more comorbidities	1.634 (18.15)	6.729 (11.31)
Invasive mechanical ventilation (IMV), no. (%)
Yes	1.609 (17.87)	2.758(4.63)
No	7.393 (82.13)	56.761 (95.37)
Admission setting, no. (%)
Emergency	5.887 (65.41)	31.159 (52.36)
Scheduled	1.545 (17.17)	15.142 (25.44)
Referral	1.195 (13.28)	7.127 (11.98)
Obstetrics	373 (4.14)	6.081 (10.22)
Intensive Care Unit (ICU) admission, no. (%)
No	5.409 (66.83)	40.263 (83.76)
Yes	2.915 (36.01)	9.465 (19.69)
Health Insurance*, no (%)
Payer 1	3.600 (39.99)	33.666 (56.56)
Payer 2	3.915 (43.49)	17.263 (29)
Payer 3	42 (0,47)	170 (0.29)
Others payers	1.445 (16.05)	8.420 (14.15)

For some variables, the number of hospital episodes does not add up to the total of the hospital episodes included due to missing information

*Payer categories were anonymized to preserve institutional confidentiality and represent different types of health insurance coverage within the Colombian health system

In the period under analysis, patient hospital episodes with safety reports accounted for 31.20% of the total cost ($58.22 millions of $186.60 millions, Table 2), although they only represented 13.14% of the episodes. The median cost for the group of patient hospital episodes with safety reports was 2.34 times that of episodes without reports (median cost $2.799 compared to $1.194, p < 0.001). Higher median cost in episodes with safety reports were consistently observed across categories of covariates (sex, hospital, DRG, Charlson Comorbidity Index, IMV, admission setting, ICU admission, and health insurance) and across different harm severity levels (negligible to severe harm; Tables 2 and 3). Among patient hospital episodes with safety reports, 90.11% (8.112 out of 9.002 episodes) were associated with negligible or mild harm. The median episode cost was higher even for episodes with negligible harm when compared to those without safety reports ($ 2.329 vs. $ 1.194, p < 0.001). The highest median cost was that of episodes with moderate harm ($ 6.937 per episode). The median cost was not higher for hospital episodes with severe or catastrophic harm ($ 5.181 per episode) as compared to those with mild harm ($ 5.473 per episode). Focusing on total costs, 86.74% of the cost of episodes with a safety report ($50.50 millions of $ 58.22 millions) corresponded to reports where the harm was classified as negligible or mild, and only 10.87% of the cost ($ 6.33 millions of $ 58.22 millions) corresponded to reports with moderate, severe or catastrophic harm (Table 2).

Table 2.

Comparison of hospital episode costs by presence and severity of patient safety incident reports

Type of event	n (% of total)	Total Cost (USD) (%)	Median Episode Cost (USD)	Comparison with “No Report”	p-value	Adjusted p-value
Without patient safety incident report	59.519 (86.86%)	$128.377.372 (68.80%)	$1.194	-	-	-
With patient safety incident report / harm severity level:	9.002 (13.14%)	$58.223.881 (31.20%)	$2.799	$1.605	< 0.001	-
- Negligible	5.814 (8.48%)	$25.833.546 (13.84%)	$2.329	$1.135	< 0.001	< 0.001
- Mild	2.298 (3.35%)	$24.667.050 (13.22%)	$5.473	$4.279	< 0.001	< 0.001
- Moderate	287 (0.42%)	$4.010.006 (2.15%)	$6.937	$5.743	< 0.001	< 0.001
- Severe / Catastrophic	26 (0.04%)	$2.321.100 (1.24%)	$5.181	$3.987	< 0.001	< 0.001
- Complications (Severity Unknown)	577 (0.84%)	$3.398.580 (1.82%)	$2.579	$1.385	< 0.001	< 0.001
Total	68.521 (100%)	$186.601.254 (100%)	$1.320	-	-	-

Comparison with “No Report” refers to the absolute difference in median episode cost versus the group without patient safety incident report

p-values calculated using Mann–Whitney U test; Bonferroni adjustment applied to comparisons within severity subcategories

Table 3.

Comparison of median hospital episode costs between cases with and without patient safety incident reports, by subgroup

Characteristics	Median costs of hospital episodes with safety reports	Median costs of hospital episodes without safety reports	Ratio (With/Without)
Sex
Female	$ 2.426	$ 999	2.4
Male	$ 3.513	$ 1.491	2.4
Health service provider company Hospital
Provider A	$ 3.321	$ 1.187	2.8
Provider B	$ 2.456	$ 1.210	2.0
Diagnosis related group (DRG)
Clinical DRG - Major diagnostic category	$ 2.256	$ 1.146	2.0
Infectious Diseases	$ 3.391	$ 1.443	2.4
Clinical and Surgical Diseases of the Cardiovascular System	$ 1.917	$ 1.131	1.7
Clinical and Surgical Diseases of the Digestive System	$1.693	$ 894	2.0
Other clinical DRGs	$2.375	$1.181	2.0
Procedural DRG - Major diagnostic category	$ 4.701	$ 1.949	2.4
Clinical and Surgical Diseases of the Digestive System	$4.143	$2.106	2.0
Clinical and Surgical Diseases of the Urinary and Genital Tract	$1.664	$1.124	1.5
Clinical and Surgical Musculoskeletal and Connective Tissue Diseases	$3.285	$1.947	1.7
Other procedural DRGs	$3.369	$1.626	2.1
Pregnancy, Childbirth, and Puerperium	$ 640	$ 446	1.4
Birth and Neonatal Period	$ 3.184	$ 1.052	3.0
Charlson Comorbidity Index
None comorbidity	$ 1.616	$ 823	2.0
One or two comorbidities	$ 3.018	$ 1.536	2.0
Three or four comorbidities	$ 3.739	$ 1.644	2.3
Five or more comorbidities	$ 3.743	$ 1.532	2.4
Invasive mechanical ventilation (IMV)
No IMV	$ 2.234	$ 1.139	2.0
IMV < 96 h without tracheostomy	$ 6.296	$ 3.205	2.0
IMV < 96 h with tracheostomy	$ 26.260	$ 9.440	2.8
IMV > 96 h without tracheostomy	$ 12.502	$ 7.499	1.7
IMV > 96 h with tracheostomy	$ 29.025	$ 15.762	1.8
Admission setting
Emergency	$ 2.672	$ 1.188	2.3
Referral	$ 4.956	$ 1.458	3.4
Scheduled	$ 3.129	$ 1.747	1.8
Obstetrics	$ 666	$ 451	1.5
Intensive Care Unit (ICU) admission
Yes	$ 8.384	$ 2.850	2.9
No	$ 1.864	$ 1.019	1.8
Health Insurance*
Payer 1	$ 3.285	$ 1.210	2.7
Payer 2	$ 2.373	$ 1.093	2.2
Payer 3	$ 8.218	$ 3.006	2.7
Others payers	$ 3.348	$ 1.492	2.2

All comparisons were statistically significant (p < 0.001), calculated using the Wilcoxon rank-sum (Mann–Whitney U) test

*Payer categories were anonymized to preserve institutional confidentiality and represent different types of health insurance coverage within the Colombian health system

Patients with safety reports had a longer hospital stay than patients without safety reports (median 8 days, interquartile range 2 to 21 days, vs. median 3 days, interquartile range 2 to 6 days). Difference in the timing between admission, date of ocurrence of de incident, and discharge are described in Supplementary Table S1; these data should be interpreted cautiously, as they are derived from an administrative safety reporting system.

Episodes associated with four patient safety incident types accounted for 69.52% ($ 40.48 millions of of $ 58.22 millions) of the cost of episodes with safety reports: patient care issues (including falls, pressure ulcers, line and tube issues, and venous thromboembolism prevention), medication related problems, healthcare-associated infections, and diagnosis and report issues (Supplementary Table S2). Hospital episodes related to clinical and procedural DRGs accounted for 97.25% ($ 56.63 millions of $ 58.22 millions) of the total cost of episodes with safety reports. For clinical DRGs, the most important was “Infectious Diseases”, representing 15.15% of the episodes with a median and total cost of $ 3.391 and $ 9.26 millions, respectively. For procedural DRGs, “Diseases of the Digestive System” accounted for 13.62% of the episodes with a median and total cost of $ 4.143 and $ 10.15 millions, respectively.

After adjustment for all covariates, patient hospital episodes with a safety report were associated with 15.5% higher than those without a report (cost ratio [CR] = 1.2; CI 95%: 1.1 to 1.2; p < 0.001;). Each additional day of hospital stay was associated with a 6.7% increase in costs (CR = 1.1; CI 95%: 1.1 to 1.1; p < 0.001). Male sex was associated with a 3.1% increase in costs compared with females. A Charlson comorbidity index ≥ 1 was linked to higher costs. Diagnosis of COVID 19 was associated with a 2.7% increase in costs (CR = 1.0; CI 95%: 1.0 to 1.0; p = 0.007). Compared with obstetric admissions, scheduled admissions showed a 25.9% increase in costs, while emergency and referral admissions were associated with lower costs. Belonging to a DRG procedure category (CR = 2.4; CI 95%: 2.4 to 2.5; p < 0.001), and having ICU admission (CR = 1.9; CI 95%: 1.9 to 2.0; p < 0.001) were strongly associated with higher costs. Use of invasive mechanical ventilation for > 96 h without tracheostomy was associated with the largest cost increase (CR = 1.7; CI 95%: 1.6 to 1.8; p < 0.001). For interpretability, in Supplementary Table S3 we report exponentiated coefficients (cost ratios); the underlying log-linear coefficients (β), standard errors, and CI 95% are provided in Supplementary Table S4. As a robustness check, a fixed-effects model (Supplementary Table S5) yielded directionally consistent results, including a similar association for safety reports (β = 0.1375; CR = 1.1, CI 95%: 1.1 to 1.2; p < 0.001), while some covariates (e.g., sex, COVID-19, payer categories) were not statistically significant under within-unit estimation.

Discussion

Higher costs of patient hospital episodes associated with patient safety reports

The study was based on a large number of observations in a retrospective cohort of more than 68.000 patient hospital episodes in two tertiary-level institutions in Colombia, with and without reported patient safety incidents. In our study cohort, we observed safety reports in 13% of patient hospital episodes, while fewer than 2% were associated with moderate or greater harm. The frequency of voluntarily reported patient safety incidents in our study was higher than that reported in other studies relying on voluntary incident reporting sytstems, and closer to the frequencies observed in studies using systematic screening methods [6, 19, 20].

Patient hospital episodes with safety reports had a higher median cost than episodes without safety reports. We also observed distinct patient profiles among episodes with safety reports, characterized by older age, greater acuity of illness, and higher likelihood of complex health conditions. Comorbidities, planned and procedure-related admissions, and ICU admission were associated with higher costs. Each of these identified covariates could confound the relationship between the occurrence of safety reports and costs; however, after adjustment, patient hospital episodes with safety reports remained approximately 15% more costly than episodes without reports. This finding is consistent with studies from high-income countries, where care for patients experiencing adverse events has been estimated to account for 13%- 16% of total healthcare expenditures, even after adjustment for age and comorbidities [10, 21, 22]. In specific contexts, such as postoperative adverse events, costs have been reported to increase by up to 50% compared with uncomplicated cases [13]. Evidence from Latin America also indicates that patient safety incidents represent a substantial economic burden for health systems [23–26].

Relationship between safety reports, length of stay, and costs

In our study, patient hospital episodes with safety reports were associated with an average prolongation of hospital stay of eight days compared with episodes without safety reports. Differences in length of stay were observed mainly during the period following the reporting date recorded in the safety system, and cost increased progressively with each additional day of hospitalization. Similar findings have been reported in previous studies, which documented prolongations of 7 to 10 days associated with adverse events during hospitalization [22, 27–29].

However, although the temporal sequence based on reporting dates suggests that part of the prolonged stay followed the reported incident, reverse causation cannot be ruled out. Longer and more complex hospitalizations may increase clinical exposure, documentation intensity, and opportunities for incident detection, which in turn may raise the likelihood of reporting. Therefore, extended length of stay may act both as a consequence of, and a contributor to, the probability of reported incidents. Additionally, the recorded date of ocurrence may not always correspond perfectly to the exact moment when an incident occurred, as it is derived from administrative safety reporting system and may be influenced by local reporting practices and workflow.

Cost impact of incidents with minimal or negligible harm

Prior studies have reported heterogeneous finding regarding the relationship between the severity of patient safety incidents and their associated costs. In our study, the median cost of patient hospital episodes with safety reports was higher than that of episodes without reports even when harm was classified as negligible, defined as minimal injury requiring no or minimal intervention [15]. Moreover, more than 85% of total safety-related costs were attributable to incidents associated with negligible or mild harm, reflecting their high frequency, despite higher median costs for incidents causing moderate harm. Severe or catastrophic incidents accounted for a smaller proportion of total costs. This pattern contrasts with findings from ICUs studies [23], where severe incidents have been shown to generate the highest costs. While the literature often prioritizes severe incidents due to their clinical impact and regulatoty implications [30], other authors emphasize that incidents causing minimal o no harm provide critical insights into system vulnarabilities and latent failures [31]. Consistent with the evidence, our results confirm that the economic impact depends on both the cost per safety incident and the total volume of patient safety incidents [6, 27].

Types of patient safety incidents associated with the highest costs

While some studies have focused on specific types of patient safety incidents [20, 29], others have comprehensively addressed all reported incident types [6, 19], similar to our approach. In our study, four types of safety incidents, patient care-related issues, medication problems, healthcare-associated infections, and diagnostic issues, accounted for the majority of additional costs, a finding consistent with the frequency of incident types and the greater importance in costs of care issues and medications.

Our results aling with the those of previous studies showing that adverse events related to falls, vascular access and catheter complications, and medication errors tend to generate the greatest financial impact [6], while healthcare-associated infections, procedure-related events, and care management issues are reported as the most frequent occurrences [19]. These findings reinforce that the most common patient safety incidents not only entail substantial economic burden but also remain priority areas for patient safety interventions.

Strengths and limitations

This study presents several important strengths. It was based on a large number of observations in a retrospective cohort of more than 68.000 patient hospital episodes in two tertiary-level institutions in Colombia, with and wihtout reported patient safety incidents. The inclusion of all categories of safety incidents provides a comprehensive and representative view of the hospital safety landscape, improving the generalizability of the findings. The linkage between the DRG database and institutional patient safety records enabled an integrated analysis of incidents and their associated costs, reducing measurement bias.

It is important to note that in our study, the detection of patient safety incidents was based on an incident reporting system. Therefore, our data reflect voluntary reports submitted by healthcare staff rather than a systematic detection of patient safety incidents. It has been observed that the frequency of patient safety incidents is lower for incident reporting systems as compared to systematic collection of patient safety prevalence information [32]. However, the observed pattern of voluntary reports in the two participating hospitals made it possible to identify a group of patients with distinct patterns in length of stay and hospital costs.

The multivariable analysis was exploratory in nature and aimed to identify associations between the occurrence of safety incidents and hospital costs while adjusting for relevant clinical and contextual factors. Although it does not allow for causal inference, this approach provides valuable insights into the main variables influencing cost differences between hospitalization episodes with and without safety reports. In addition, the economic methodology, using CPI adjustments and conversion to constant U.S. dollars, ensured temporal comparability and precision in cost estimation. Overall, this work provides relevant evidence in the Latin American context, offering a robust assessment of the financial impact of hospital safety incidents and supporting data-driven decision-making to improve quality and efficiency in healthcare delivery.

Despite its strengths, this study has several limitations. The retrospective design relies on administrative and institutional data, which may be subject to misclassification (the safety incident severity might have been underestimated) or underreporting (particularly for minor safety incidents). Because reporting systems are voluntary and differ across hospitals, variability in the frequency and severity of reported incidents is possible. Nevertheless, the relatively high frequency of voluntary reports in the participating hospitals provides valuable insights into real-world reporting practices.

Beyond these operational limitations, there are also conceptual considerations regarding what incident reporting systems capture. From a conceptual perspective, traditional incident reporting systems are largely aligned with a Safety-I approach, which focuses on identifying failures after they occur. Contemporary patient safety theory, however, emphasizes Safety-II and resilient healthcare frameworks, which understand healthcare as a complex sociotechnical system where variability, adaptation, and everyday adjustments are inherent to safe performance [33–35]. Within this broader view, voluntary incident reports, such as those analyzed in our study, partly reflect clinical environments with greater complexity, workload, and operational pressure, and do not constitute a direct or exhaustive measure of patient harm. Reported incidents may therefore be associated with underlying system demands and performance adjustments that also influence healthcare costs. In contrast to Safety-I, Safety-II is based on the principle that performance adjustments are ubiquitous and that performance not only always is variable but that it must be so. The variability should not be interpreted negatively, as in ‘performance deviations’, ‘violations’, and ‘noncompliance’. On the contrary, the ability to make performance adjustments is an essential human contribution to work, without which only the most trivial activity would be posible. This perspective supports cautious interpretation of the observed associations and reinforces that reported incidents represent only a partial window into overall system performance.

On the other hand, the study period coincided with the COVID-19 pandemic, which may have influenced case-mix, care processes, hospitalization patterns, and the behavior of voluntary incident reporting. Periods of high clinical workload and staff strain could have reduced reporting or altered reporting practices. Therefore, the observed frequency and characteristics of reported incidents may not fully represent usual reporting patterns outside pandemic conditions. Although our model adjusted for COVID-19 diagnosis, pandemic-related changes in care delivery and reporting behavior may not be fully captured by the available variables. Furthermore, although the data correspond to the period 2020 to 2022, the participating hospitals have continued operating with similar levels of clinical complexity and patient safety reporting structures since then. Still, changes in healthcare delivery, technology, or safety practices over time may limit the full temporal generalizability of the findings.

Another important limitation is the absence of detailed measures of baseline clinical severity at admission beyond the variables captured in DRGs and comorbidity indices. Patients with greater underlying severity are more likely to require complex care, generate higher costs, and receive closer monitoring, which may also increase the probability of incident detection and reporting. Residual confounding due to unmeasured severity may therefore partially explain the observed cost differences.

The study only considered costs incurred during inpatient stays. However, it is known that harm experienced during hospitalization can increase both in-hospital and post-discharge healthcare costs, especially in adverse events of moderate or greater severity, whose effects may persist for several months [20, 30]. Previous research has estimated that up to 20% of the costs attributable to adverse events arise in outpatient settings [13]. Because post-discharge costs were not included, the total economic burden associated with patient safety incidents is likely underestimated, particularly for events with longer-term consequences.

Finally, although the models were adjusted for multiple covariates, residual confounding cannot be ruled out, as prospective data on baseline illness severity or quality of care, were not available for more precise adjustment.

Implications for policy, practice and research

This study provides robust evidence to support policymakers and healthcare leaders in recognizing the economic dimension of patient safety. Investing in safety initiatives is not only an ethical and regulatory imperative, but also a financially sound strategy, as safety incidents generate substantial additional hospitalization costs. We should view an investment in safety as an investment in productivity. However, our findings suggest that safety investments should not focus exclusively on rare catastrophic events. A large proportion of total costs was associated with high-frequency incidents classified as negligible or mild harm. Therefore, system-level strategies that reduce the volume of recurrent low-harm events, such as failures in basic care processes, medication management, and infection prevention, may yield meaningful cumulative economic benefits.

Efforts to reduce the financial burden of unsafe care should start with the reinforcement of safety reporting systems that can identify errors, near misses, or deviations before harm occurs. Strengthening such systems enables a more proactive approach of risk management and continuous improvement of patient safety activities.

For reporting systems to support both learning and economic impact, their design should encourage consistent reporting across all levels of harm severity and clinical areas, while minimizing blame and documentation burden. Standardization of incident classification, feedback loops to clinical teams, and integration with administrative and cost databases may enhance the ability to identify high-impact patterns of risk. However, hospitals should not use only incident reporting systems to estimate prevalence of harm to patients. Health systems should incorporate multiple patient safety data sources, including active surveillance, to prioritize interventions and promote a cycle of action and improvement based on data rather than merely just collecting and analysing information [32].

Further investigation is required to understand the true impact of safety events initially categorized as causing minimal or negligible harm. These incidents may contribute to patient distress, loss of trust in healthcare institutions, and hidden increases in system-wide costs.

Future research should also differentiate between cost avoidance and cost-effectiveness. While preventing safety incidents may reduce excess resource use (cost avoidance), safety interventions themselves require investment. Evaluations that compare the costs of specific safety programs, such as infection prevention bundles, medication reconciliation systems, or early warning and monitoring protocols, against the costs potentially avoided would provide more actionable evidence for decision-makers.

It is also essential to assess whether targeted investments in specific quality and safety domains can lead to measurable cost savings for healthcare systems, as suggested by previous research [36, 37].

In particular, interventions aimed at high-risk hospital contexts identified in this study, such as ICU care, complex procedural admissions, and prolonged hospitalizations, may offer greater economic returns when combined with strengthened surveillance and reporting practices.

The findings of this study could guide the prioritization of investments in the most impactful areas of patient safety, helping decision-makers and regulators allocate resources more efficiently by weighing both economic and human outcomes.

Conclusions

Hospital episodes involving patient safety incidents were substantially more expensive than those without, even when the associated harm was minor. Most of the additional costs originated from frequent patient safety incidents associated with negligible or minor harm, rather than rare severe or catastrophic events. These results demonstrate that unsafe actions (which are always preventable) impose a measurable economic burden on the healthcare system. Strengthening patient safety incident surveillance, reporting, and prevention initiatives could generate both clinical and financial benefits. Implementing effective safety strategies in hospitals is therefore essential for improving patient outcomes while promoting sustainability of healthcare.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

DRG

Diagnosis-Related Group

GRD

Diagnosis-Related Group (Spanish acronym used in institutional databases)

ICU

Intensive Care Unit

IMV

Invasive Mechanical Ventilation

USD

United States Dollar

COP

Colombian Peso

CPI

Consumer Price Index

TRM

Representative Market Rate

IQR

Interquartile Range

CR

Cost Ratio

CI

Confidence Interval

USA

United States of America

COVID-19

Coronavirus Disease 2019

Author contributions

J.R.A.P. and L.S.M.D. contributed to the conceptualization, methodology development, and formal analysis of the study, and both prepared the original draft of the manuscript. L.S.M.D. and N.Y.G. provided supervision and contributed to writing – review and editing. K.R.C.A. participated in the conceptualization and methodological design of the study and contributed to writing – review and editing. D.O.A.R. contributed to the methodological processes, carried out data curation and formal analysis, and participated in writing – review and editing. Y.M.T.C., S.M.R.R., A.M.G.B., A.Y.A.R., and E.B.G.L. were responsible for the investigation activities and contributed to writing – review and editing. All authors read and approved the final manuscript.

Funding

Not applicable.

Data availability

The datasets generated and analyzed during the current study were used to perform all statistical analyses reported in the manuscript and are included in the Supplementary Materials. Additional data are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was reviewed and approved by the Research Ethics Committee of the Sanitas University Foundation (Fundación Universitaria Sanitas; CEIFUS2729-23). The use of secondary, fully anonymized databases meant that no personal identifiers were available at any stage of the research process. Therefore, individual informed consent was not required, in accordance with national regulations and the policies of the ethics committee. All procedures complied with the Declaration of Helsinki and the International Council for Harmonisation Good Clinical Practice guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.