Drug interactions in maternal intensive care: prevalence, risk factors, and potential risk medications

ABSTRACT Objective: To characterize severe potential drug interactions in maternal intensive care, and to determine their frequency, risk factors and potential risk medications. Methods: An observational and longitudinal study conducted between December 2014 and December 2015 in a maternal intensive care unit. Clinical data were collected and severe potential drug interactions were identified on pregnant inpatients. The drug interactions were classified by type, prevalence and exposure rate. A multivariate logistic regression model was used to identify the severe potential drug interactions and the related drugs (p<0.05). Results: A total of 95.1% of patients were exposed to, at least, one potential drug interaction; in that, 91.7% 33.9% were related to, respectively, moderate and severe potential drug interactions. The patients were exposed, on average, on 69.2% of days they were in the intensive care unit. The main drugs involved in more severe drug interactions were magnesium sulfate, metoclopramide, propranolol and diazepam. Conclusion: The severe potential drug interactions were observed in almost all patients of the study, and, approximately one third of those interactions were related to greater severity and resulted in exposure during long hospital stay. The higher number of prescribed drugs and its previous use of medications at home increase the occurrence of severe potential drug interactions.


❚ INTRODUCTION
Care of critically ill patients in intensive care units (ICU) entails extensive use of medications and consequent polypharmacy. Prescription of several drugs increase occurrence of adverse events, especially drug interactions, leading to greater costs and longer hospital stay. (1)(2)(3) A potential drug-drug interaction (PDI) is defined as the concurrent administration of two potentially interactive drugs, with a prevalence of 46% to 80% among intensive care patients. (1,(4)(5)(6) Because of clinical severity, critically ill patients have a lower capacity for drug elimination, increasing the risk of damage resulting from PDI. (3)(4)(5)7) The treatment of pregnant women at ICU is associated with emergency obstetric care. (8,9) In developed countries, it is estimated that 2% or less of pregnant women are admitted to the ICU, while in developing countries this rate is approximately 10%. (9) The main reasons for admission in maternal ICU are related to hemorrhage and hypertensive disorders of pregnancy. (8)(9)(10) Other less common reasons include heart disease, genitourinary infection, obstetric complications, such as ectopic pregnancy and miscarriages, non-genitourinary infection, sepsis, cerebrovascular disease, and pulmonary embolism. (9,10) The use of drugs in maternal ICU presents some peculiarities. Physiology changes during pregnancy, leading to modified actions of medications on the body. (11) Pregnant women present significant pharmacokinetic changes. The increase in adipose tissue increases the distribution volume and the half-life of lipophilic drugs. Other changes include increased activity of liver enzymes and glomerular filtration rate, with risk of fetal exposure to lipophilic drugs. (11)(12)(13) Most studies on PDI are limited to characterizing the frequency of medications and therapeutic classes involved, without highlighting which factors contribute to the occurrence of PDI and their possible consequences. However there is scarce literature on frequency of clinically significant PDI in pregnant women, in addition to characterization of these events in patients admitted to maternal ICU. To characterize severe potential drug-drug interactions  in maternal intensive care units and determine the  frequency, risks factors, and potential risk medications. ❚ METHODS An observational, descriptive, cross-sectional study conducted at the 6-bed maternal ICU of the Maternidade Escola Januário Cicco (MEJC), with an average of 53 admissions per month. All patients admitted to the ICU during the study period who stayed for more than 24 hours, and had more than two drugs prescribed were considered eligible. Patients previously included in the sample were excluded. A total of 348 patients were evaluated between December 2014 and December 2015.
Clinical data related to previous pregnancies, comorbidity score as per the Mortality Probability Model (MMP) upon admission, clinical diagnosis according to the International Classification of Diseases and Related Health Problems -Tenth Edition (ICD-10), and prescribed drugs were collected. All patients were daily monitored for occurrence of drug interactions, assessed as to the prescribed items and time of exposure to the PDI throughout the hospitalization period.
The software Lexi-Interact was used to identify PDI in medical prescriptions. It consists of a database with restricted online access, with updated information on more than 8,000 unique drug terms. The database enables classifying the PDI as minor, moderate or severe, according to severity level. Severe PDI can be life-threatening and/or require intervention to reduce or prevent damage; moderate interactions imply exacerbation of the clinical condition, and possible changes in the prescription; and minor PDI imply a limited clinical effect with the possibility of worsening, but with no need for intervention. (14) Statistical analysis was carried out using Stata release 11 (Stata Corporation, College Station, TX, USA). In the descriptive analysis, the characteristics of the patients were presented as mean and standard deviation, or relative or absolute frequencies, when appropriate. The rate of exposure to drug interactions was calculated by the ratio between the number of days exposed to the event and the total length of hospital stay. For each variable studied, the number of drug interactions per day of hospitalization was obtained by multiplying the number of events by the exposure einstein (São Paulo). 2019;17(3):1-7 rate. The identification of risk factors for severe drug interactions was conducted by univariate and multivariate analysis using logistic regression. Only the variables presenting a p value <0.10 were included in the multivariate model (p<0.05). To determine the prescription drugs at the highest risk of causing severe drug interactions we used regression adjusted by the risk factors identified and those already reported as being associated with drug interactions (number of medications, length of stay and age).

❚ RESULTS
During the study, 348 patients were assessed ( Table 1). The mean age was 26.4 years, with a mean of two pregnancies; 26.2% reported at least one miscarriage. In the sample, there was a predominance of cesarean section (57.8%), but roughly 37% were still pregnant. Gestational hypertension was characterized as the most frequent diagnosis (65.9%), followed by eclampsia (7.5%), and urinary infection (2.6%). Patients had a mean length of ICU stay of 3 days (standard deviation of 2.1 days).
Approximately 8±3.5 medications were administered during hospitalization. About 95.1% of patients were exposed to at least one PDI; in that, 52.9% of minor, and 91.7% of moderate severity ( Table 2). About one-third of patients (33.9%) had a more severe PDI. Table 2 also emphasizes that during hospital stay patients were, on average, 69.2% of period exposed to at least one severe PDI.
In the univariate logistic regression analysis, the occurrence of PDI was significantly associated with number of drugs administered, use of medications at home before hospitalization, age and diagnosis of urinary infection. These variables were included in the multivariate model (Table 3), in which a significant association was found between the occurrence of severe PDI and the number of drugs administered (odds ratio -OR: 1.106; 95% confidence interval -95%CI: 1.073-1.139). There was also a significant association between the occurrence of severe PDI and the medications used at home (OR: 1.317; 95%CI: 1.022-1.696). The diagnosis of urinary tract infection was, however, less likely to be associated with occurrence of severe PDI (OR: 0.334; 95%CI: 0.136-0.854).
Among the drugs related to severe PDI, diazepam, magnesium sulfate, propranolol, and metoclopramide presented greater risk in pregnant women under     intensive care (Figure 1)

❚ DISCUSSION
The occurrence of PDI in maternal intensive care is observed in almost all patients, and the exposure time covers 69.2% of hospitalization period. Compared to other intensive care modalities, the prevalence of PDI detected in pregnant women was markedly higher. For adult and cardiac ICU, e.g., the prevalence of patients exposed to at least one PDI ranges from 50 to 80%, and severe PDI occur in percentages lower than those found in this study. (3)(4)(5) A systematic review included nine studies and 3,150 intensive care patients evaluated, found that about 63% of patients were exposed to at least one PDI. (15) It is important to emphasize that prescription of drugs with potential drug interactions does not necessarily imply harm to patients. Nonetheless, the risk is higher when PDI are classified as severe. (5) In this study, about one-third of pregnant women had at least one severe PDI and were exposed for more than two-thirds of the hospitalization period. Obstetric patients present physiological changes that alter the pharmacokinetics and pharmacodynamics of numerous drugs, thus increasing the risk of harm associated with use of drugs, especially when there is interaction between them, which may affect the dyad motherfetus. (11,12,16) Therefore, considering the prevalence and duration of exposure, severe PDI should be considered in the pharmacological approach of these patients.
Approximately 33.9% of patients presented severe PDI, and the prior use of medications before hospitalization and a high number of drugs prescribed during the hospitalization period were identified as risk factors for these events. Polypharmacy is a risk factor already described, and is related to the occurrence of drug interaction and other incidents associated with the use of medications. (17) The use of medications at home as a risk factor probably indicates greater clinical severity, leading to the use of multiple drugs during hospitalization. In contrast, urinary tract infection appeared as a reducing factor, which may be justified by the treatment approach applied in those cases in which a smaller number of drugs related to severe PDI are used.
Patients who were prescribed magnesium sulfate, diazepam, propranolol and metoclopramide were at increased risk for severe PDI. Magnesium sulfate is not commonly used in other intensive care modalities and is indicated for hypertensive disorders of pregnancy. The other drugs are more frequently related to more severe PDI in intensive care in general. (6,18) einstein (São Paulo). 2019;17(3):1-7 Magnesium sulfate is the drug with the highest risk of causing severe PDI, especially when it is associated with nifedipine. The association can cause hypotension, hypocalcemia, neuromuscular blockade, myocardial depression, as well as pulmonary edema. (19,20) The mechanism of interaction involves the reduction of flow in the calcium channels in myocardial cells, in addition to blocking the adrenal release of norepinephrine, antagonizing the vasopressor effects. (21) The association of propranolol and methyldopa is not recommended, since it may cause hypertensive reaction, tachycardia or arrhythmia due to peripheral blockade of beta-adrenergic receptors, warranting the effect of vasoconstriction by alpha adrenergic receptors. (22) Propranolol should also be avoided with cabergoline, due to block of normal sympathetic vasodilatation stimulated by beta-2 receptors, resulting in excessive vasoconstriction, and may cause hypertension or even peripheral ischemia. (23,24) Severe PDI involving diazepam are associated with depression of neural activity and significant changes in drug metabolism. Diazepam and phenobarbital act to potentiate the gamma-aminobutyric acid (GABA) inhibitory effect, favoring the opening of the chlorine channels in the neural membrane. (25)(26)(27) Simultaneous use of these drugs may lead to potentiation of neuronal inhibition and, clinically, to central nervous system depression. (27) Benzodiazepine, when associated with antipsychotic drugs, like levomepromazine, causes increased blocking of dopaminergic receptors, which may lead to extrapyramidal adverse events or other effects associated with these receptors. (28) Drug interactions between diazepam and phenytoin occur at the pharmacokinetic level, with hydantoin being metabolized by hepatic isoenzymes CYP2C9 and CYP2C19. (29,30) Benzodiazepines are metabolized by a distinct isoform (CYP3A4), however, diazepam is specifically metabolized extensively by CYP2C19. (31) Diazepam may, therefore, compete with the metabolizing enzyme and significantly increase phenytoin concentrations, inducing toxicity. (29,30) Carbamazepine, may compromise the therapeutic action of diazepam through metabolic activity of the CYP3A4 isoenzymes. (31)(32)(33)(34) Metoclopramide also had a high frequency of more severe PDI, especially when used simultaneously with antipsychotics. The antiemetic action of metoclopramide is due to dopamine antagonism in D2 receptors. (35) The association with neuroleptics can cause intense dopaminergic blockade in the striatum and hypothalamus, favoring the occurrence of extrapyramidal effects and, in extreme cases, neuroleptic malignant syndrome. (36) It is important to emphasize that this study characterizes PDI in pregnant women in intensive care, and data on this topic is scarcely found in the literature. Likewise other intensive care modalities, drug interactions are a major concern in medical practice, since they may have detrimental consequences for health care. (4,6,7) The identification, description and monitoring of potential interactions are, therefore, important in order to minimize pharmacotherapy-related risks. Similar studies may contribute to establishing automated alert systems aimed to the multiprofessional team at ICU, enabling less exposure of patients to severe drug interactions. The findings also indicated that a clinical pharmacist should monitor the patients, adequately evaluating the occurrence of PDI in the prescription and making it possible to prevent damage. (1,37) This paper presents some limitations. Data was collected in one single center, which may compromise generalizing the findings. Despite the prospective collection, some information may have been missed due to failures in the process of medical records. It is important to emphasize that the interactions described are potential, that is, they were classified from the medical prescription, and did not necessarily imply the occurrence of negative clinical outcomes. The detection of PDI, even those classified as severe, should be contextualized with the patient's clinical presentation, and the risk-benefit of the treatment should always be considered. Another aspect that may decrease the capacity of generalizing the results involves possible disagreement between the databases and compendia regarding the class and severity of drug interactions. (38) However, the literature indicates that the software used in this study is solidly based on clinical evidence and is constantly updated. (38,39) New investigations should be conducted in order to evaluate clinical outcomes related to drug interactions, especially correlating them with clinical parameters, length of stay, and mortality.
❚ CONCLUSION Almost all pregnant women in maternal intensive care are exposed to at least one potential drug-drug interaction. More than one-third of potential drug-drug interactions are severe and occur during most of the hospitalization period. The factors associated with severe potential drug-drug interactions are prior and use of medication at home, as well as a greater number of drugs prescribed during hospitalization, whereas the diagnosis of urinary tract infection einstein (São Paulo). 2019;17(3):1-7 is inversely related to the occurrence of potential drug-drug interaction. The drugs most implicated in potentially severe drug-drug interaction were magnesium sulfate, metoclopramide, propranolol and diazepam. It is important to emphasize that the identification, description, and monitoring of potential interactions are important to ensure safety of pregnant woman under intensive care.