Potential medication incompatibilities in pediatric oncology prescriptions

Introduction: Pediatric oncology patients have a limited number of venous access routes and need a large number of drugs during hospitalization. This study evaluates potential medication incompatibilities (MI) in pediatric oncology prescriptions and identifies possible factors associated with the risk of their occurrence. Methods: This cross-sectional study evaluated prescriptions from a tertiary universitary hospital from December 2014 to December 2015. The association between variables and the risk of potential incompatibilities between drugs was determined by Student’s t-test and Pearson’s chi-square, considering p < 0.05 significant. The odds ratio was calculated considering a 95% confidence interval for each drug. Results: 385 prescriptions were evaluated. The mean age of 124 patients was


INTRODUCTION
Cancer in children and adolescents from 0 to 19 years old represents a group of diseases with particular characteristics that are related to histopathological profiles and clinical manifestations.It is considered a rare disease and represents 1 to 4% of malignant tumors, with short periods of latency and more aggressive manifestations.However, it has a better response to treatments and, generally, has a good prognosis 1 .
Cancer treatment in children and adolescents includes different modalities, from systemic to diagnostic, according to histological type, local extension, and region.This disease can be treated with surgery, radiotherapy, and/or systemic treatment, such as chemotherapy (monotherapy or drug association), according to national and international guidelines 2 .Antineoplastic drugs are used in almost all treatments, but the undesirable effects of chemotherapy include a potential acute and late toxicity, as well as the possibility of medication interactions and drug incompatibilities.In these situations, dose reduction or even medication discontinuation should be considered, in order to limit the antineoplastic effect that impairs patient survival 2 .
Medication incompatibilities (MI) are physical or chemical reactions that occur between two or more drugs in vitro (before intravenous administration) when associated in the same syringe, vial, or bag 3,4 .Physical reactions are commonly visible, with precipitate formation, color change, or gas production.However, the identification of chemical reactions requires analytical techniques that show significant loss of active components during the drug mixing process 5 .
The consequences of MI can result in a reduction of the efficacy and safety of drug therapy, drug inactivation, the formation of a new innocuous or toxic active component, and in organoleptic changes in color, consistency, turbidity, as well as precipitation and crystal formation 4 .There are reports of fatal pulmonary embolism related to drug incompatibility and mechanical failure of venous catheters [5][6][7] .
MI should be even more considered in the pediatric population, since this group has a limited number of venous access routes, such as the totally implantable catheter (Port-a-Cath), which has only one route for drug administration.Moreover, this population needs a relatively large number of drugs, in addition to water supply via the catheter [8][9][10] .
Clinical pharmacists working on oncology wards play an important role on a multiprofessional team, with their knowledge, experience, and abilities concerning cancer patient care 11 .Among the activities that can be developed by clinical pharmacists in oncology with the focus on patient safety, we highlight the evaluation of drug incompatibilities, aiming at the rational use of drugs 12 .
Based on this context, this study aims to evaluate the potential medication incompatibilities within a pediatric oncology unit and identify possible risk factors for their occurrence.

Type of study
Cross-sectional study at the Pediatric Oncology Unit of a tertiary university hospital in Southern Brazil.

Population and sample
In order to calculate the sample size, a 50% prevalence (considering the worst-case scenario due to the lack of data about the prevalence of MI on medical prescriptions for pediatric oncology patients), 5% precision, and 95% confidence level were used.Retrospectively, 385 medical prescriptions were selected from patients aged 0 to 17 years and 11 months old admitted to the Pediatric Oncology Unit from December 2014 to December 2015.

Inclusion and exclusion criteria
The prescription chosen for analysis was related to the third day of hospitalization of patients with at least two intravenous drugs prescribed.It is important to consider that in the first and second day of hospitalization, there may be no intravenous drugs prescribed.The drug metamizole was not analyzed, since its use is not authorized in some countries, thus, it did not appear in the database used in this study to evaluate MI.

Data collection
Medical prescriptions were accessed via the hospital's electronic medical records system.A data collection form was used for the following information: patient's age and sex; reason for admission; type of venous access; number of venous access routes; oncological diagnosis described on the medical records system; number of intravenous drugs prescribed; number of intravenous drugs prescribed as "fixed" and "symptomatic"; total number of drug crossings per prescription; number of compatibilities, variable incompatibilities, untested incompatibilities, physical incompatibilities, and chemical incompatibilities; and the existence of total parenteral nutrition (TPN).MI were evaluated using the Drugdex database (Thomson Micromedex), accessed by the Periódicos Capes website 13 .Notably, it was not possible to observe if there was incompatibility; we can only evaluate the potentiality of its occurrence during prescription analysis.

Statistical analysis
The data were entered in Microsoft Office Excel and analyzed using the SPSS 20.0 program.The equality of proportions of categorical variables was analyzed by Pearson's chi-square test (with continuity correction for 2×2 tables) and the comparisons of averages of continuous numerical variables were performed using Student's t test.The odds ratio was calculated with a 95% confidence interval for each drug with a statistically significant difference in the chi-square test for differences between proportions.The age-adjusted odds ratio was also calculated.Statistical tests were performed considering a significance level of α = 0.05.

Ethical aspects
The study was approved by the Research Ethics Committee of the Hospital de Clínicas de Porto Alegre (no.15-0583).
Regarding the type of venous access, the totally implantable catheter (Port-a-Cath) was used in 61.30% of cases, followed by the peripherally inserted central catheter (PICC) (16.10%), the peripheral venous catheter (15.32%), the double-lumen tube (4.93%), and the Hickmann line (2.33%).In 92.73% of cases, the venous access used had only one lumen (one access route), while 7.27% had two routes.Only 0.52% of cases had a TPN prescription.
Table 1 shows that age, diagnosis, the reason for hospital admission, and the type of venous access were risk factors for potential MI, since they presented a statistically significant difference in the chi-square test for the proportions and Student's t-test.
In 87.5% of the prescriptions evaluated, there was the possibility of occurrence of at least one MI.A total of 2108 drug incompatibilities were evaluated, of which 300 different incompatibilities were between two drugs.Table 2 shows the prevalence of the main incompatibilities, as well as the types of MI found.Considering the variables related to incompatibilities, the mean of the total number of drugs, of both "fixed" and "symptomatic" drugs, the total number of drug crossings, and total compatibilities were higher in prescriptions with MI.The difference between the means were statistically significant (Table 3).For drugs with a statistically significant difference in the chi-square test for proportions, the odds ratio was calculated in order to confirm the risk of incompatibilities of these drugs individually.

DISCUSSION
This study was developed to identify possible MI in pediatric oncology prescriptions and, thus, highlight issues related to safety in the administration of intravenous drugs.It is important to emphasize that there are few studies about MI, especially in pediatrics.To our knowledge, this is the first study to address MI in pediatric oncology.The identification of possible risk factors for the occurrence of MI may help clinical pharmacists in pediatric oncology.
The epidemiological data of the population involved in this study coincide with national statistics for this age group, with predominance of leukemia (27.30% of cases), followed by osteosarcoma (18.20%), and lymphomas (9.90%).The higher prevalence in male is also in line with Brazilian data 1 .
The reasons for hospital admission are similar to another study performed in a pediatric oncology unit, in which chemotherapy was the main cause of hospitalization (36.10% vs. 67.5%),followed by complications related to the treatment, such as fever and infections (15.32% vs. 24.8%) 14.The reasons for hospitalization were risk factors for the occurrence of MI, proving to be an important information for clinical pharmacists to analyze the MI, when prioritizing prescriptions for patients.Hospitalizations for chemotherapy and the management of complications require the use of intravenous drugs and, therefore, MI should be a concern in this group of patients.
The most frequent type of venous access in our study was the totally implantable catheter (Port-a-Cath), which was used in 61.32% of cases.This longterm central venous access is commonly used in oncology due to its comfort, as it avoids frequent venipunctures, especially in patients that need long-term treatments 15 .In 61.4% of prescriptions for patients with Port-a-Cath, the possibility of MI was identified.A study published in 2014 showed that MI represented the major problem related to intravenous drug administration in patients in a general hospital.The authors also observed that patients with central access had more incompatibilities than patients with peripheral access 16 .
TPN prescriptions occurred in only 0.52% of cases.The frequency of TPN was small due to its indication only for situations with impossibility of using oral or enteral routes in medium-and long-term periods.Moreover, the objective of using TPN is to offer favorable conditions for the therapeutic plan, maintaining the patient's vital functions 17 .Our study analyzed only the medical prescription from the third day of hospitalization, when the indication of TPN is uncommon.
We observed the possibility of the occurrence of MI in 87.5% of the prescriptions analyzed.As expected, the number of drugs in medical prescriptions with MI was significantly higher when compared with prescriptions without MI (a mean of 5.63 vs. 2.81 drugs per prescription), being a risk factor for the occurrence of MI.This data is similar to a study conducted in a Pediatric Intensive Care Unit, which showed that the number of drugs prescribed (>3 drugs) was a risk factor for MI 18 .
MI are classified into four categories.They are "untested" when there is no data in the literature on the combination of two certain drugs.In these situations, it is recommended, for safety reasons, for them to not be administered concomitantly.MI are "variable" when the compatibility between two drugs depends on the type of diluent used and/or the concentration of drugs.Finally, they are "physical" and "chemical" when the incompatibility leads to a physical or chemical reaction, respectively 12 .In this study, the number of "untested" incompatibilities was higher compared with other categories, with an average of 3.69 incompatibilities per prescription.This is similar to the study performed in 2016 by Leal et al., in which they also observed a higher prevalence of "untested" incompatibilities 18 .
The main possibility of incompatibility identified was between two antiemetics, dimenhydrinate and ondansetron, found in 65.19% of prescriptions.This high prevalence occurs because antiemetics are commonly prescribed for patients undergoing chemotherapy, as described in other studies 19,20 , since nausea and vomiting are frequent and debilitating adverse effects directly causing a meaningful impact in patients' quality of life 20 .Antiemetics are advisable from the first treatment cycle and are also used as prophylaxis for anticipatory vomiting in other stages of treatment 21 .Another frequent incompatibility occurred between dimenhydrinate and a solution of potassium chloride 10% + sodium chloride 20% + glucose (33.77% of medical prescriptions).This solution is often prescribed for pediatric patients, for water replacement and/or maintenance and is used to maintain patients' body homeostasis while fasting, offering the necessary amount of water and electrolytes to restore the losses due to physiological processes, such as diuresis, sweating, bowel movements, and respiration 22 .
The presence of leucovorin, sodium bicarbonate, cefepime, diphenhydramine, dimenhydrinate, hydrocortisone, and ondansetron in medical prescriptions allowed the occurrence of possible MI, characterizing these drugs with a higher risk for potential MI.
The presence of ondansetron, dimenhydrinate, and diphenhydramine can be explained by the fact that they were the most frequently prescribed drugs (83.89%, 74.30%, and 56.10%, respectively), despite being considered incompatible with each other.Similarly, the solution 10% potassium chloride + 20% sodium chloride + glucose was frequently prescribed (51.94% of prescriptions), despite being incompatible with dimenhydrinate, another frequently prescribed drug.
Leucovorin and sodium bicarbonate were drugs with higher risk for MI.These drugs are incompatible with each other and are part of the treatment for osteosarcoma 23 , the second most frequent disease in the population of this study (18.20% of cases).Leucovorin is used as an antidote to methotrexate, which blocks the conversion of folic acid to tetrahydrofolate by binding to the enzyme dihydrofolate reductase, and is indicated after 24 hours of the infusion of methotrexate.Sodium bicarbonate is used for urine alkalinization, facilitating the excretion of methotrexate 23 .
Cefepime was observed in 16.62% of medical prescriptions and presented a higher risk for MI.The use of cefepime in pediatric oncology is recommended in cases of febrile neutropenia, which is a frequent complication due to periods of spinal immunosuppression after chemotherapy 24,25 .
MI requires attention from all professionals involved in patient care.Clinical pharmacists can contribute to the prevention of MI through the use of some strategies 17 .Some tools that can be adopted are standard operating procedures with specific guidelines for medication administration, analysis of prescriptions informing potential MI via a drug compatibility charts, and nursing training in the preparation and administration of drugs 18,[26][27][28][29][30] .
It is important to reinforce the interaction of clinical pharmacists with the nursing team regarding IM since knowledge about compatibility varies depending on the professional and inpatient unit 31 .Medication errors related to medication administration are frequent 32 .Mendes et al. showed that, among medication errors, in cases of concomitant administration, only 17.86% were compatible 32 .Specific nursing education interventions should be planned, including the use of applications for smartphones 33 .
This study had some limiting factors, mainly due to its retrospective design.The main limitation was the fact that the analysis was restricted to drugs in medical prescriptions.Thus, it was not observed whether these drugs were actually administered concomitantly, causing MI.The cross-sectional design and the non-inclusion of metamizole in the analyses, due to its absence in the database were other limitations of the study.Further research is needed to better understand the problems related to MI, especially laboratory studies to evaluate MI that are currently considered "untested." This study showed that the possibility of the occurrence of MI is frequent in intravenous drug administration in pediatric oncology and should be the focus of attention of professionals in safety of drug use.

Table 2 :
Prevalence of the main MI between two intravenous drugs and types of incompatibilities (n = 385).
* Untested: the association was not tested in laboratory yet.MI: Medication incompatibilities.

Table 4 :
Drugs prescribed and the risk of potential MI (n = 385).
a p-value in the chi-square test; b p-value in the odds ratio calculated by the chi-square test; * significant odds ratio; ┼ Calculated and ageadjusted odds ratio.It was not possible to adjust for the other drugs due to the distribution of variables between groups.MI: Medication incompatibility; OR: Odds ratio; 95%CI: 95% confidence interval.