INTRODUCTION
Adverse drug reactions (ADRs) are responses to a medicinal product that are noxious and unintended, and that occur at doses normally used in humans or arise from the use of a medicinal product within or outside the terms of the marketing authorization, including medication errors, misuse, abuse, and off-label use (1,2). ADRs are a serious public health issue due to their potential to greatly increase morbidity, death, and financial burden (3). Up to 0.1–1% of patients using systemic drugs experience ADRs, which are a global health concern. These are expensive for the healthcare systems and have an effect on patient care and health-related quality of life (4). Research indicates that hospital admissions were caused by 5–10% of ADRs. Hospital admissions for ADRs were 5.5% in underdeveloped countries and 6.3% in wealthy countries. According to an Indian study, the prevalence of ADRs was 3.7%, and 0.7% of hospitalizations were caused by ADRs (5).
Based on patient characteristics (gender, age, weight, creatinine clearance, and number of comorbidities) and drug administration (dosage, administration method, number of concurrent drugs), high-risk agents should be properly monitored. Extremes in age, gender, the use of many medications, a medical condition, a history of allergies or ADRs, genetics, high dosages, and numerous other factors can all raise the risk of ADRs. In some groups, particularly the elderly, stopping medication or altering dosage may have a significant role in the development of ADRs (6,7,8,9). Effective healthcare management and patient safety depend on an understanding of ADRs (10). Hospitalized patients are particularly vulnerable to ADRs due to the severity of their underlying diseases, polypharmacy, and the frequent use of high-risk medications. Antibiotics are among the most commonly prescribed drugs in hospital settings, and their use is often associated with a considerable number of adverse reactions (11). In addition to allergic reactions, antibiotics may cause gastrointestinal disturbances, hepatotoxicity, nephrotoxicity, hematological disorders, and neurological complications (12,13).
Reserve antibiotics, including carbapenems, glycopeptides, polymyxins, and oxazolidinones, are primarily used for the treatment of severe infections caused by multidrug-resistant microorganisms. Although these drugs are highly effective, they are also associated with a higher risk of ADRs, particularly in critically ill and hospitalized patients. Their toxicity profile and the complex clinical conditions of treated patients make pharmacovigilance and careful monitoring essential (14,15). In tertiary healthcare institutions, the use of reserve antibiotics has increased significantly due to the rising prevalence of antimicrobial resistance (16). Despite their clinical importance, data regarding the frequency and risk factors for ADRs with these drugs remain limited, particularly in hospitalized populations. Therefore, the aim of this study was to determine the frequency of ADRs during treatment with reserve antibiotics in hospitalized patients at the University Clinical Center Kragujevac and to identify factors associated with their occurrence.
MATERIALS AND METHODS
Study design, population, sampling procedure
This study had an observational design with a mixed prospective–retrospective approach and included spontaneous reporting of ADRs to reserve antibiotics by clinical pharmacologists during their routine consultations. The study was conducted among patients who were receiving or had received reserve antibiotics.
The study population consisted of patients hospitalized at the University Clinical Center Kragujevac between January 2022 and December 2023. Cases were defined as hospitalized patients who developed ADRs associated with reserve antibiotics. The control group consisted of patients hospitalized during the same period who were also treated with reserve antibiotics but did not develop ADRs. Controls were matched to cases in a 1:2 ratio.
The inclusion criteria were: hospitalization at the University Clinical Center Kragujevac, treatment with reserve antibiotics, and complete medical documentation. Exclusion criteria included incomplete medical records, confirmed drug overdose, and age under 18 years.
A consecutive convenience sampling method was used for the selection of cases, meaning that all patients with reported ADRs during the study period were included. For each case, potential controls were selected among patients of the same sex and similar age (± 5 years).
The dependent variable was the occurrence of ADRs, determined based on the consultation report of a specialist physician recorded in the patient's medical documentation. ADRs were defined as any harmful and unintended response to a drug that occurs at doses normally used in humans for the purposes of treatment, prevention, diagnosis, or modification of physiological functions, or occurring at any dose during a clinical trial.
Independent variables included length of hospitalization, number of consulted physicians, total number of consultations, and whether the patient had been treated in an intensive care unit. All variables were obtained from patients’ medical records.
Potential confounding variables included patient education level, marital status, place of residence (urban or rural), occupation, lifestyle habits (smoking, alcohol consumption, coffee intake), Charlson Comorbidity Index, and specific comorbidities such as chronic renal insufficiency, heart failure, type 2 diabetes mellitus, hyperlipidemia, malignancy, liver disease, and atrial fibrillation. Comorbidities were assessed using the Charlson Comorbidity Index, a validated method for classifying prognostic comorbidity in longitudinal studies (17). These variables were derived from patients’ medical records. The primary reason for hospitalization and the medications administered during hospitalization were also recorded.
Data were independently extracted by two researchers from computerized patient medical records stored in the hospital information system “ZIS” (Comtrade, Belgrade). Any discrepancies between the extracted datasets were resolved by consensus in order to ensure accuracy and consistency.
Statistical analysis
The minimum required sample size for the study was calculated using the G*Power software (version 3.1.9.7) (18) based on results from a previously published study (19). The calculation was performed using the following parameters: probability of type I error (α) of 0.05, minimum statistical power of 0.80, odds ratio of 2.6, outcome frequency of 0.26, coefficient of determination (R2) of 0.40, and normal distribution of variables. Based on these parameters, the calculated minimum sample size was 33 patients in the case group, while the control group was planned to include twice as many participants.
The collected data were numerically coded, tabulated, and subsequently checked for possible errors by the two investigators. Descriptive statistics were used to summarize the data. Continuous variables were presented as mean values with standard deviation when normally distributed, while median and interquartile range were used when data were not normally distributed. Categorical variables were expressed as frequencies, percentages, and relative proportions.
The influence of potential predictors and confounding variables on the occurrence of ADRs was assessed using multivariate binary logistic regression analysis. Prior to applying the regression model, the assumptions for multivariate analysis were verified, including binary outcome variable, independence of observations, absence of multicollinearity among predictors, lack of extreme outliers, and adequate sample size for logistic regression analysis.
The quality and goodness-of-fit of the regression model were evaluated using Cox & Snell R2, Nagelkerke R2, and the Hosmer–Lemeshow goodness-of-fit test.
A p-value ≤ 0.05 was considered statistically significant. All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS), version 18.0.
Ethical approval: The study was performed in line with all the relevant national regulations, institutional policies, and in accordance with the tenets of the Helsinki Declaration and has been reviewed and approved by the institutional review board, the Ethics Committee of University Clinical Center Kragujevac [approval number: 01/24-131].
Informed consent statement: Written informed consent was obtained from all individual participants included in the study.
RESULTS
Out of a total of 9,621 clinical pharmacologist consultations during the two-year study period, 156 patients were included in the analysis. Among them, 52 patients experienced an ADR, while 104 patients formed the control group. Overall, 47 patients died, whereas 11 required prolonged hospitalization. Data on the characteristics of the study sample are presented in Tables 1–4, which include demographic and clinical parameters.
Table 1.
Distribution of patient comorbidities (n = 156)
| Comorbidity | Frequency (percentage) |
|---|---|
| Type 2 diabetes mellitus (yes/no) | 50/106 (32.1%/67.9%) |
| Myocardial infarction (yes/no) | 10/146 (6.4%/93.6%) |
| Heart failure (yes/no) | 64/92 (41%/59%) |
| Peripheral vascular disease (yes/no) | 22/134 (14.1%/85.9%) |
| Cerebrovascular disease (yes/no) | 12/144 (7.7%/92.3%) |
| Dementia (yes/no) | 10/146 (6.4%/93.6%) |
| Connective tissue disease (yes/no) | 5/151 (3.2%/96.8%) |
| Chronic obstructive pulmonary disease (yes/no) | 25/131 (16%/84%) |
| Peptic ulcer disease (yes/no) | 29/127 (18.6%/81.4%) |
| Liver disease (yes/no) | 5/151 (3.2%/96.8%) |
| Hemiplegia (yes/no) | 14/142 (9%/91%) |
| Solid tumor (yes/no) | 18/138 (11.5%/88.5%) |
| Chronic renal insufficiency (yes/no) | 38/118 (24.4%/75.6%) |
| Leukemia (yes/no) | 6/150 (3.8%/96.2%) |
| Lymphoma (yes/no) | 2/154 (1.3%/98.7%) |
Distribution of patient comorbidities are presented in Table 1.
The most frequently prescribed antibiotics were vancomycin, followed by meropenem. Their distribution is shown in Table 2.
Table 2.
Administered antibiotics (n = 156)
| Antibiotic | Frequency (percentage) |
|---|---|
| Meropenem (aayes/no) | 71/85 (45.5%/54.5%) |
| Imipenem/cilastatin (yes/no) | 22/134 (14.1%/85.9%) |
| Piperacillin/tazobactam (yes/no) | 12/144 (7.7%/92.3%) |
| Vancomycin (yes/no) | 82/74 (52.6%/47.4%) |
| Teicoplanin (yes/no) | 17/139 (10.9%/89.1%) |
| Linezolid (yes/no) | 9/147 (5.8%/94.2%) |
| Colistimethate sodium (yes/no) | 26/130 (16.7%/83.3%) |
| Tigecycline (yes/no) | 20/136 (12.8%/87.2%) |
The most notable laboratory abnormalities were observed for leukocyte count, serum creatinine, and C-reactive protein (CRP). Mean values are presented in Table 3.
Table 3.
Mean laboratory values (n = 156)
| Laboratory parameter | Mean ± SD, median, IQR | Units |
|---|---|---|
| Platelets | 277.69 ±161, 263.5, 212 | ×109/L |
| Urea | 10.6 ± 10.8, 7.55, 7 | mmol/L |
| Serum creatinine | 145.71 ± 173.68, 79, 80 | μmol/L |
| Leukocytes | 13.78 ± 17.99, 11.2, 8.3 | ×109/L |
| Neutrophils | 9.09 ± 7.72, 7.94, 6.2 | ×109/L |
| ALT | 65.68 ± 143.5, 22, 29 | U/L |
| AST | 76.68 ± 207.4, 27, 30 | U/L |
| GGT | 95.24 ± 121.2, 47.5, 98 | U/L |
| Total bilirubin | 25.9 ± 51.9, 13.45, 10 | μmol/L |
| Direct bilirubin | 12.84 ± 35.2, 4.1, 4.8 | μmol/L |
| Amylase | 73.15 ± 84.95, 56, 64 | U/L |
| Lipase | 103.13 ± 290, 8, 42 | U/L |
| CRP | 133.65 ± 92.83, 125, 140 | mg/L |
A slightly higher proportion of patients in the total sample had a fatal outcome, while a smaller percentage required prolonged hospitalization. Other relevant characteristics of the study population are presented in Table 4.
Table 4.
Other relevant characteristics of the study population (n = 156)
| Parameter | Frequency (percentage) for categorical variables, Mean ± SD, median, IQR for continuous variables |
|---|---|
| Adverse drug reaction (yes/no) | 52/104 (33.3%/66.7%) |
| Recovery (yes/no) | 33/123 (21.2%/78.8%) |
| Death (yes/no) | 47/109 (30.1%/69.9%) |
| Prolonged hospitalization (yes/no) | 11/145 (7.1%/92.9%) |
| Allergy history (yes/no) | 21/135 (13.5%/86.5%) |
| ICU stay (yes/no) | 74/82 (47.4%/52.6%) |
| Age | 64.12 ± 13.75, 65.5, 17 |
| Duration of ADR | 2.81 ± 5.52, 0, 4 |
| Charlson comorbidity index | 4.5 ± 2.36, 4, 3 |
| Duration of antibiotic therapy | 9.68 ± 5.83, 9, 6 |
| Length of hospitalization | 36.58 ± 36.59, 29, 27 |
| Total consultations | 17.27 ± 17.24, 11, 15 |
| Pharmacologist consultations | 4.93 ± 3.99, 4, 5 |
Table 5 presents the characteristics of patients according to the study outcome (presence of ADR). No statistically significant difference in sex distribution was observed between the groups. A considerable proportion of patients had diabetes, chronic renal insufficiency, and heart failure.
Table 5.
Characteristics of study participants according to the outcome – occurrence of ADRs.
| Parameter | Cases (n=52) Frequency (%), median, [IQR] | Controls (n=104) Frequency (%), median, [IQR] | p-value |
|---|---|---|---|
| Sex (M/F) | 33/19 (63.5%/36.5%) | 65/39 (62.5%/37.5%) | 0.907 |
| Acute myocardial infarction | 4/48(7.7%/92.3%) | 6/98(5.8%/94.2%) | 0.644 |
| Heart failure | 15/37(28.8/71.2%) | 49/55(47.1%/52.9%) | 0.029 |
| Peripheral vascular disease | 8/44(15.4%/84.6%) | 14/90(13.5%/86.5%) | 0.745 |
| Cerebrovascular stroke | 2/50(3.8%/96.2%) | 10/94(9.6%/90.4%) | 0.202 |
| Dementia | 3/49(5.8%/94.2%) | 7/97(6.7%/93.3%) | 0.817 |
| Connective tissue disease | 1/51(1.9%/98.1%) | 4/100(3.8%/96.2%) | 0.520 |
| Peptic ulcer disease | 5/47(9.6%/90.4%) | 24/80(23.1%/76.9%) | 0.042 |
| Chronic obstructive pulmonary disease | 7/45(13.5%/86.5%) | 18/86(17.3%/82.7%) | 0.537 |
| Liver disease | 3/49(5.8%/94.2%) | 2/102(1.9%/98.1%) | 0.199 |
| Diabetes mellitus | 16/36(30.8%/69.2) | 34/70(32.7%/67.3%) | 0.808 |
| Hemiplegia | 5/47(9.6%/90.4%) | 9/95(8.7%/91.3%) | 0.843 |
| Chronic renal insufficiency | 18/34(34.6%/65.4) | 20/84(19.2%/80.8%) | 0.035 |
| Solid tumor | 5/47(9.6%/90.4%) | 13/91(12.5%/87.5%) | 0.595 |
| Leukemia | 0/52(0%/100%) | 6/98(5.8%/94.2%) | 0.077 |
| Lymphoma | 0/52(0%/100%) | 2/102(1.9%/98.1%) | 0.314 |
| Allergy | 4/48(7.7%/92.3%) | 17/87(16.3%/83.7%) | 0.135 |
| Meropenem | 23/29(44.2%/55.8) | 48/56(46.2%/53.8%) | 0.820 |
| Imipenem | 8/44(15.4%/84.6%) | 14/90(13.5%/86.5%) | 0.745 |
| Piperacillin/tazobactam | 5/47(9.6%/90.4%) | 7/97(6.7%/93.3%) | 0.524 |
| Vancomycin | 24/28(46.2%/53.8) | 58/46(55.8%/44.2%) | 0.257 |
| Teicoplanin | 3/49(5.8%/94.2%) | 14/90(13.5%/86.5%) | 0.146 |
| Linezolid | 2/50(3.8%/96.2%) | 7/97(6.7%/93.3%) | 0.466 |
| Colistin | 10/42(19.2%/80.8) | 16/88(15.4%/84.6%) | 0.543 |
| Tigecycline | 7/45(13.5%/86.5%) | 13/91(12.5%/87.5%) | 0.866 |
| ICU stay | 27/25(51.9%/48.1) | 47/57(45.2%/54.8%) | 0.427 |
| Recovery | 33/19(63.5%/36.5) | 0/104(0%/100%) | 0.000 |
| Death | 15/37(28.8%/71.2) | 32/72(30.8%/69.2%) | 0.805 |
| Prolonged hospitalization | 11/41(21.2%/78.8) | 0/104(0%/100%) | 0.000 |
| Age | 64.5[16] | 67.1[19] | 0.507 |
| Duration of ADR | 7[8] | 0[0] | 0.000 |
| Charlson comorbidity index | 4[3] | 4.5[3] | 0.436 |
| Duration of antibiotic therapy | 6.5[8] | 10[5] | 0.005 |
| Length of hospitalization | 30[26] | 28.5[26] | 0.801 |
| Total consultations | 12[14] | 11[16] | 0.509 |
| Pharmacologist consultations | 4[6] | 4[4] | 0.722 |
| AST | 30[29] | 26[31] | 0.719 |
| ALT | 22[30] | 21.5[30] | 0.328 |
| GGT | 50.5[116] | 40.5[97] | 0.594 |
| Total bilirubin | 12[9.7] | 13.9[11.1] | 0.293 |
| Direct bilirubin | 3.55[4.1] | 4.1[4.9] | 0.344 |
| Amylase | 61.5[47] | 49[62] | 0.250 |
| Urea | 7.75[7.7] | 7.4[7] | 0.870 |
| Creatinine | 86.5[135.8] | 77[65] | 0.307 |
| Leukocytes | 9.5[6.5] | 11.7[9.3] | 0.037 |
| Neutrophils | 6.8[4.8] | 8.52[6.64] | 0.068 |
| Platelets | 256[166] | 267[241] | 0.936 |
| CRP | 133.3[149.5] | 120.4[116.1] | 0.897 |
The relationship between independent variables, potential confounders, and the occurrence of ADRs was examined using multivariate binary logistic regression analysis. A forward conditional stepwise method was applied to construct the regression model including the following predictors: heart failure, peptic ulcer disease, chronic renal insufficiency, age. sex, Charlson comorbidity index, length of hospitalization, leukocyte count, CRP level.
All assumptions for logistic regression analysis were satisfied. There were no extreme outliers, the sample size was adequate, no multicollinearity was detected (variance inflation factor < 2 for all predictors), observations were independent, and residuals followed an approximately normal distribution. Six atypical observations were excluded from the analysis.
The Hosmer–Lemeshow test showed good model fit (χ2 = 1.916, df = 5, p = 0.861). The Cox & Snell R2 was 0.118 and the Nagelkerke R2 was 0.164.
Variables included in the final model are presented in Table 6. Heart failure, peptic ulcer disease, and chronic renal insufficiency were identified as factors associated with the occurrence of ADRs. Heart failure and peptic ulcer disease were identified as protective factors, whereas chronic renal insufficiency was identified as a risk factor for ADRs.
Table 6.
Statistically significant predictors of ADR occurrence identified by multivariate logistic regression
| Risk factor | Crude OR (95% CI) | Adjusted OR (95% CI) | P value |
|---|---|---|---|
| Heart failure | 0.455 (0.223–0.928) | 0.262 (0.115–0.599) | 0.002 |
| Peptic ulcer disease | 0.355 (0.127–0.992) | 0.268 (0.091–0.790) | 0.017 |
| Chronic renal insufficiency | 2.224 (1.049–4.713) | 3.356 (1.415–7.955) | 0.006 |
DISCUSSION
Our study evaluated factors associated with ADRs related to reserve antibiotics in hospitalized patients in a tertiary healthcare institution. Our findings indicate that chronic renal insufficiency represents a significant risk factor for the occurrence of ADRs, while heart failure and peptic ulcer disease were associated with a lower risk of these reactions.
Previous studies have found that comorbidities represent an important factor influencing the development of ADRs (20,21). In the present study, chronic renal insufficiency was identified as a significant risk factor for the occurrence of ADRs. This finding is consistent with previous research showing that impaired renal function may increase the risk of drug accumulation, toxicity, and altered pharmacokinetics of many antimicrobial agents. Consequently, careful dose adjustment and monitoring of renal function are essential when prescribing reserve antibiotics to patients with renal impairment (22).
Reduced glomerular filtration, tubular secretion, and reabsorption in patients with chronic kidney disease lead to altered pharmacokinetics and increased systemic exposure to drugs that are primarily eliminated through the kidneys. Accumulation of antibiotics and their metabolites may occur, increasing the likelihood of toxic effects and adverse reactions (23). Furthermore, uremia may alter protein binding of drugs due to decreased serum albumin levels, resulting in a higher fraction of pharmacologically active free drug in circulation (24). These mechanisms may partly explain the increased incidence of ADRs observed in patients with chronic renal insufficiency in our study.
On the other hand, heart failure and peptic ulcer disease were identified as protective factors in the multivariate analysis. One possible explanation for this finding is that patients with these conditions are more frequently monitored and receive more cautious pharmacotherapy due to their already recognized clinical vulnerability. Increased clinical attention and closer therapeutic monitoring may therefore contribute to a lower observed incidence of ADRs in these patients.
In our study, vancomycin and meropenem were the most frequently prescribed antibiotics. This observation reflects current clinical practice in tertiary care hospitals, where these agents are commonly used in the treatment of severe infections caused by multidrug-resistant microorganisms. However, the use of these antibiotics is often associated with a higher risk of adverse reactions due to their pharmacological properties, potential nephrotoxicity, and the complexity of patients’ clinical conditions. A major concern is high dosage, since plasma trough levels of vancomycin show a linear association with the risk of acute kidney injury and is significantly correlated with nephrotoxicity, particularly at dosages of >20 mg/L or >4 g/day. Additionally, it has been discovered that a high dose of vancomycin (≥15 mg/L) is an independent risk factor for nephrotoxicity (25).
An important aspect of this study is the role of clinical pharmacologists in the detection, assessment, and prevention of ADRs. In our study, ADRs were identified through routine consultations and spontaneous reporting, highlighting the importance of integrating clinical pharmacology consultations into everyday hospital practice. Clinical pharmacologists contribute to early recognition of ADRs, optimization of pharmacotherapy, and individualized dose adjustment, particularly in high-risk patients receiving reserve antibiotics (26). The effectiveness of these interventions depends not only on the identification of drug-related problems but also on the successful implementation of recommendations in clinical practice. In line with previous studies, clinical pharmacologists and pharmacists play a crucial role in improving medication safety among hospitalized patients. Their recommendations are more likely to be accepted when they address clinically evident ADRs or involve high-risk drug combinations, particularly in the context of regularly prescribed therapies (27).
The findings of this study have important clinical implications. Identifying patients at increased risk for ADRs can help clinicians implement targeted monitoring strategies, adjust antibiotic dosing more appropriately, and potentially reduce the occurrence of adverse events. This is particularly relevant in tertiary healthcare settings where reserve antibiotics are frequently used in critically ill patients.
Limitations
Several limitations of this study should be acknowledged. First, the relatively small sample size may limit the generalizability of the findings. Second, the study was conducted in a single tertiary healthcare institution, which may reduce external validity. In addition, the retrospective component of the data collection may have introduced information bias due to incomplete or inconsistent medical records. Future studies with larger sample sizes and multicenter designs are needed to further clarify the risk factors associated with ADRs and to develop strategies for improving the safe use of reserve antibiotics in clinical practice.
CONCLUSION
The use of reserve antibiotics in hospitalized patients at tertiary healthcare institutions is associated with a measurable risk of ADRs, which is likely related to the severity of illness and complexity of these patients. Identification of risk factors may contribute to safer and more individualized antibiotic therapy in hospital settings.