Drug therapy and adverse drug reactions to terbutaline in obstetric patients: a prospective cohort study in hospitalized women

Background Adverse drug reactions (ADR's) could be expected more frequently in pregnant women. This study was performed in order to identify ADR's to tocolytic drugs in hospitalised pregnant women. Methods A prospective cohort study was performed in two General Hospitals of the Instituto Mexicano del Seguro Social (IMSS) in Mexico City. Two hundred and seven women undergoing labor, premature labor, threatened abortion or suffering any obstetric related disease were included. Drug prescription and signs and symptoms of any potential ADR were registered daily during the hospital stay. Any potential ADR to tocolytic drugs was evaluated and classified by three of the authors using the Kramer's algorithm. Results Of the 207 patients, an ADR was positively classified in 25 cases (12.1%, CI95% 8.1 to 17.5%). All ADR's were classified as minor reactions. Grouping patients with diagnosis of threatened abortion, premature labor or under labor (n= 114), 24 ADR's were related to terbutaline, accounting for a rate of 21.1 ADR's per 100 obstetric patients. Obstetric patients suffering an ADR were older than obstetric patients without any ADR. However, the former received less drugs/day × patient-1 and had a shorter hospital stay (p < 0.05) whereas the dose of terbutaline was similar between the two groups. Terbutaline inhibited uterine motility in women with and without any ADR at a similar rate, 70 and 76% respectively (x2 = 0.07; p = 0.8). Conclusion Terbutaline, used as a tocolytic drug, was related to a high frequency of minor ADRs and to a high rate of effcicacy.


Background
The ADR's are considered as a major public health problem [1]. Being the fourth cause of general deaths [2], the ADR's are costly and represent a significant rate of hospital admissions [3,4]. Furthermore, the number of drugs available for prescription in the clinical setting is increasing every day. It is therefore important to acquire, interpret and report all ADR's identified with any drug [2]. Furthermore, pharmacovigilance in special populations, e.g. pregnancy women, could be useful to identify unexpected responses probably expressing unusual pharmacokinetic profiles as a result of their particular physiological state [5,6].
Multiple methods have been used for pharmacovigilance, and despite spontaneous reporting is the simplest one, it has been of low efficacy in Mexican medical practice [7]. Furthermore, in Latin-America only a limited number of drug utilization studies done in order to report any ADR during clinical practice are available. In Mexico, the population is growing at a yearly rate of 1.85, representing approximately more than 2 millions of newborns every year [8]. The IMSS, one of the two social health systems available in Mexico, is responsible to attend approximately 60% of the Mexican citizens and also a considerable rate of deliveries [9], resulting into a major source of information on drugs utilization. The present study was performed in order to identify any ADR to tocolytic drugs in a prospective cohort of hospitalised pregnant women requiring medical attention at the IMSS.

Materials and Methods
After approval by the National Research Committee of the IMSS, the study was performed at the Gynecology and Obstetric units of two secondary-care general hospitals. Selection of these hospitals was based on their similar characteristics of medical care while they had a different geographic distribution, one at the north and other at the south of Mexico City. A minimal sample size of 148 patients was estimated using previous studies on ADR's with an expected frequency of ≥ 2.5% and a significant level of 5% [10,11]. During the study period, 207 women in labor, premature labor, threatened abortion or suffering any obstetrical related disease referred to any of the two hospitals over a 4-month period were included into the cohort, and data were obtained during their hospital stay.

Data collection
General information including age (yr.), level of education (according to the basic Mexican scholar system, patients were grouped into ≤ 6 and >6 years of scholar level), diagnosis (cesarean section, labor, premature labor, threatened abortion, or post-cesarean complications), and hospital stay (days) were obtained. Drugs and dose administrated were obtained from the medical and nurs-ery records. Patients were questioned daily, in relation to the presence or not, of symptoms relating to drug administration. Nurses trained for purpose of the study collected data and an Obstetrician confirmed clinical information.

Causality assessment
For identifying any ADR, the Kramer's algorithm was used [12][13][14]. This system was previously translated into Spanish and successfully used in pediatric patients [15]. Briefly, the algorithm contains 56 questions grouped in six decision-making axes, and it evaluates previous experiences with any drug, potential etiologies of the ADR, a temporary relationship between drug administration and the presence of the ADR, the possibility of an overdoses, and any re-challenge with the suspected drug. Each axe is graded and a total score is obtained to classify the reaction as improbable (<0), possible (0,+1,+2,+3), probable (+4,+5) and definitive (+6,+7). For patients receiving more than one drug, each drug was evaluated by means of the algorithm. Any possibility of drugs interaction was evaluated by means of the Drug interaction program (The Medical Letter, Inc., New Rochelle, NY, USA). An ADR was positively qualified if two or all the evaluators qualified a suspected ADR as either probable or definitive. The file of each patient was reviewed independently by three of the authors trained to use the algorithm (AA Nava-Ocampo, JA Palma-Aguirre and H Sumano-López). Inter-observer agreement among the three evaluators in relation to the scores given to every potential ADR was computed by means of a Kappa analysis at a p < 0.05 level [16,17]. Finally, severity of each ADR was scored according to Capellá and Laporte into mortal, severe (any life-threatening reaction), moderate (any reaction requiring hospitalization or requiring urgent atention), or mild [18].

Statistical analysis
Data from all patients were summarized by using descriptive statistics. Patients were further goruped into patients suffering or not any ADR. Except for vitamins, all drugs received by patients were presented as only one active principle. We therefore counted all drugs daily received by each patient and obtained a mean value. Results were then summarized and a final mean value and SD of number of drugs/day × patient -1 was obtained. The unpaired Student t test was used to compare age and drugs/day × patient -1 between patients suffering or not an ADR. The dose of terbutaline was also compared between the two groups by means of the unpaired Student t test. The Fishers' exact chi-square test was used to evaluate differences in diagnosis and type of drug used for uterine activity inhibition (terbutaline, indometacine, or none) between groups. The significant level for all statistical analyses was fixed at a p < 0.05. When used, the parametric 95% confidence interval for the difference was computed.

Software
Data were collected in a predesigned Microsoft ® Excel 97 form. For statistical anaylsis, we used the Epi-lnfo ® 6 v. 6.04d (The Center for Disease Control and Prevention, Atlanta, Georgia, USA).

Results
Demographic data were summarized in Table 1. Patients were young people, most of them have the basic education level, and patients undergoing cesarean section was the the major clinical condition.
In relation to the ADR's, agreeement among the reviewers for classification of each ADR was satisfactory (Kappa > 0.92). Of the 207 patients included in the cohort, 28 presented any suspected ADR, being 25 positively classified (12.1%, 95%CI 8.1 to 17.5%). Grouping patients with diagnosis of threatened abortion, premature labor or under labor (n= 114), 24 ADR's were related to terbutaline, accounting for a rate of 21.1 ADR's per 100 obstetric patients.
Patients suffering an ADR were slightly but significantly older than those wihout any ADR ( Table 2). They also were attended mainly for premature labor, have a lower hospital stay, and they were mainly receiving terbutaline. However, the dose of terbutaline was similar between the two groups. The tocolytic therapy with terbutaline inhibited uterine motility in women with with and without any ADR in a similar rate, 70 and 76% (x 2 0.07; p = 0.8), respectively. Of the 24 patients suffering an ADR to terbutaline, tremor was present in all patients, dizziness in seven, confusion in six, depression in five, adynamia in three, astenia in other three and headache also in three patients, and loss of equilibrium in one, irritability in one, and palpitations and aggressiveness in another patient. The only patient suffering an ADR to indomethacin referred abdominal discomform. All ADR's were classified as minor reactions, and according with the obstetricians did not merit to prolong the hospital stay, any additional treatment or drug discontinuation, and no fetal complications was reported by the patient or at the maternal records during the hospital stay.  * 3.1 ± 0.7 * data expressed as mean ± sd

Discussion
Kramer et al., created an algorithm to identify and qualify any ADR without any drug assay [12][13][14]. However, some problems have emerged with its use. It needs to be translated and adapted for using in different countries, as it happened to us. The algorithm is extensive, and therefore time-consuming. It requires multiple specific information that was collected due to the prospective nature of the present study. Some data, however, could not be available in the clinical file for a retrospective evaluation.
Despite the Council of the International Organization of Medical Sciences has provided definitions and basic requirements for the proper use of ADR terminology [19], a diagnosis is often difficult to establish due to the presence of clinical conditions or prescription of two or more drugs. Casuality assessment for ADR's is therefore often difficult in the clinical setting [20]. Comparisons among the reported ADR's frequencies are also problematic due to the differences observed among the studies, including the population (e.g. pediatric, adults or old patients), gender, set of the study (e.g. emergency rooms or hospitalized patients), and method of measurement (e.g. therapeutic drug monitoring, spontaneous reports) [4,10,11,[21][22][23][24][25][26][27][28] . The frequency of 12.1% of ADR's reported in the current study was lower than the 33.3% reported for all admissions at an Indian hospital [29], and than 28.2% reported at a Universitary hospital [11]. However, it resulted higher than 2.4 to 3.7% of ADR's observed in other studies for hospitalized patients [4,10]. Differences could be explained by the fact that obstetric patients do receive a lesser amount or less agressive theraphy than patients attended e.g. at internal medical wards.
We studied women patients requiring hospital attention for non-accidental causes, and despite a high incidence of ADR's was identified in patients receiving terbutaline for premature labor and threatened abortion, mortality was not present. Terbutaline is worldwide formerly approved for the treatment of asthma. As it was recently reviewed by Lam et al. [30], in the United States the off-label utility of terbutaline as a tocolytic agent has been known by clinicians for more than 20 years, estimating that at least 260,000 women are yearly receiving terbutaline during pregnancy, being the most popular prescribed β-mimetic for tocolysis in the USA. In Mexico, it is also extensively used in the obstetric wards provably favored by the fact that other therapuetic options as ritodrine did never arrive to our country. In fact, ritodrine was removed from the marked in the United States [30].
Terbutaline is clearly an effective inhibitor of uterine activity [31,32], and its ADR's are abated with discontinuing treatment only. In our study, no differences in dose of terbutaline were detected between patients suffering or not an ADR. Although efficacy was not our goal, the uterine inhibitor effects of terbutaline resulted in a high rate of patients. The study was performed during a period of 4 months and we did not observe any patient re-entering into the hospital because of the presence of a new episode of uterine activity. Whether the patients underwent another period of uterine activity and received medical care in another hospital or successfully completed the pregnancy period, cannot be clarified in our study. Also, the fetal and maternal long-term morbidity was unknown. Obstetrician service is the main request of medical care in Mexican hospitals, and births have been 31.2% of total hospital discharges, being 53.9% of total births registered in 1997 attended at the IMSS [9]. A careful selection of pregnant women in order to avoid a dangerous impact to both or either the mother and the fetus due to the production of palpitations counterbalancing the benefits between its use for managing a threatened abortion or a premature labor and costs of minor ADR's, must be mandatory. Furthermore, the small range of current options to be used as tocolytics should stimulate this area in order to identify drugs with lesser production of side effects. In fact, terbutaline could not only be undangerous for the fetus but to promote neonatal respiration and metabolic adaptation after elective cesarean section and to reduce the number of fetal heart abnormalities [33,34].
In relation to indomethacin, this drug has proved safety and efficacy to inhibit uterine contractions of premature labor [35,36]. Uterine contractility at term and preterm results from an activation of myometrium through several process varying from mechanical stimulation to a complex cascade of endocrine processes [37]. Prostaglandinds are important regulators of the labor process [38], and therefore its manipulation has resulted into a direct effect favoring or inhibiting uterine activity [39,40]. However, it is well known the adverse effects of all nonsteroidal antiinflamatory drugs administered at the third trimester of pregnancy [41][42][43], including constriction of the ductus arteriosus, persistent fetal circulation, impairment of renal function and bleeding. Furthermore, brain maldevelopment and neurobehaviour deviations have been experimentally demonstrated after neonatal exposure to indomethacin [44]. Therefore, despite our results seem likely to favour indomethacin administration because it was better tolerated than terbutaline for tocolysis, the serious adverse effects potentially produced in the fetus by indomethacin make this drug a greatly dangerous option for preterm labor management.
Additionally, incidence of preterm birth is greatly increased among the socially disadvantaged women, probably explained by two major factors [45]. First, the presence of chronic and acute social stressors which in turn are translated into organc responses. Second, the presence of a gene-environment interaction based on a highly prevalence mutation in the gene for methylentetrahydrofolate reductase. Even more, maternal education level could decrease infant mortality rate by preventing preterm births [46], without affecting fetal growth [47]. In the present study, most patients have the basic education level, and therefore if any effect was present this would be protector. Finally, there is a need for a simple, efficient and low-cost of ADR's reporting system covering a wide range of the population receiving any drug. The spontaneous reports probably satisfy these conditions and participation of nurses in the design of strategies of recognizing any ADR is undoubtedly necessary [48,49].
In conclusion, terbutaline was responsible of a high rate of mild ADR's in women receiving this drug as a tocolytic agent. However, the lack of well recognized options makes terbutaline the major tocolytic drug currently available.

Competing interests
None declared