Metamizole Use in Children: Analysis of Drug Utilisation and Adverse Drug Reactions at a German University Hospital between 2015 and 2020

Metamizole utilisation was evaluated for the general wards (including one ward for infectious diseases) of a department of paediatrics and adolescent medicine of a German tertiary teaching hospital. We systematically reviewed the drug therapy of paediatric inpatients < 18 years of age. For the study period 01 June 2015 to 31 May 2020, we retrospectively analysed prescription data using electronic medical records (prescription software: VMobil [31], cf. https://​www.​advanova.​de).

The electronic data stored in the electronic medical record VMobil were extracted, processed and compiled into a standardised Microsoft^® Excel data file to analyse medication data systematically. Vitamins and food supplements were categorised with only one code; therefore, several different vitamins or food supplements in one patient cannot be distinguished in the exposure evaluation. Fluid and electrolyte infusions or parenteral nutrition were not included in this assessment. We introduced five age categories, which were adopted from the International Conference on Harmonisation (ICH) classification [32]: 0–27 days (newborns), 28 days to 23 months (infants and toddlers), 2–5 years (pre-school children), 6–11 years (school children) and 12 to < 18 years (adolescents).

In the scope of this analysis, a case is defined as an inpatient stay of a patient. If one patient was admitted to the hospital several times within the study period, all inpatient stays (cases) were included in our analysis. We retrieved data to quantify and characterise metamizole use per case (i.e. information on date of metamizole administration, route of administration, and the number of days with metamizole therapy). Descriptive patient data, age, body weight, gender, length of hospital stay and the number of different drugs per case were also collected.

In our evaluation, all cases and the subgroup of cases with any drug therapy as well as the subgroup of cases with metamizole therapy were characterised. To quantify metamizole exposure, we calculated the percentage of cases that used metamizole at least once among all included cases. Additionally, we determined the proportion of metamizole exposed cases in relation to the subgroup of cases with drug therapy.

We assessed the percentages of metamizole exposure overall, stratified by age and by study year. Descriptive results are presented with numbers, percentages or medians with interquartile range (IQR). For descriptive analysis, Microsoft^® Excel and IBM^® SPSS 24 were used.

In 2013, an in-house quality standard for reporting ADRs and medication errors (ME) was implemented at the department of paediatrics and adolescent medicine. According to this standard, every suspected ADR and ME observed at the paediatric department is documented systematically and, irrespective of whether it led to inpatient admission or occurred during the hospital stay, reported to the Drug Commission of the German Medical Profession (Arzneimittelkommission der deutschen Ärzteschaft, AkdÄ) as appropriate. As the quality standard for spontaneous reporting of ADRs is subject to regular training of hospital staff as part of the hospital’s quality management, the method used in the second part of the analysis is called a “stimulated spontaneous reporting system”.

All case reports of patients < 18 years of age recorded between 01 June 2015 and 31 May 2020 were included for our analysis. Data of the inpatient stay, including the length of hospital stay and patient demographics, such as date of birth, weight, laboratory parameters and anamnesis, were extracted from the hospital´s clinical information system. Medication data, including administered drug substance, dose and route of administration, were obtained from the medical records. Age and total length of metamizole intake were calculated for each case report from the extracted variables. Continuous variables were summarised by their mean and standard deviation. The World Health Organisation–Uppsala Monitoring Centre (WHO-UMC) system was used for standardised causality assessment of suspected ADRs [33]. Causality was assessed via a consensus-based approach by a team consisting of a clinical paediatric pharmacologist and experienced pharmacists. The case reports in which causality between the ADR and metamizole was assessed as ‘possible’ or ‘probable’ are presented in detail. Data were analysed pseudonymously.

The local ethics committee has approved the study (Ethics Committee of the Friedrich-Alexander-Universität Erlangen-Nürnberg, Germany, code 561_20 Bc). Data security and privacy were ensured, and patients’ identity was protected. Therefore, in accordance with the legal requirements, assent or informed consent of patients or their legal representatives was waived [34].

During the study period, a total of 21,717 cases of 14,197 patients from 0 to < 18 years of age were electronically recorded. Among those, 11,857 cases of 7809 patients received at least one drug therapy. Metamizole was administered at least once in 17.3% (3759/21,717) of all cases. Related to cases with drug therapy, 31.7% (3759/11,857) of cases involved metamizole. We observed some variability in metamizole exposure across the different years: 34.3% metamizole exposure in the first, 31.1% in the second, 35.0% in the third, 31.6% in the fourth and 26.5% in the fifth study year. Characteristics of the study population overall and of those exposed to metamizole during the inpatient stay are presented in Table 1, with results also stratified by age category. Metamizole exposure was highest in adolescents and lowest in newborns (37.9% [1144/3019] of adolescent cases vs 9.9% [25/252] of newborn cases with any medication). Overall, 48.8% of the patients exposed to metamizole were male. The median length of stay of cases involving metamizole administration was longer (5 days; IQR 3–8) compared with cases overall (2 days; IQR 2–5) and cases with any medication (3 days; IQR 2–6). The difference in days of hospitalisation between patients with and without metamizole was statistically significant (p < 0.001, all cases; p < 0.001, cases with any medication; Mann–Whitney test). Overall, metamizole was administered parenterally in about 90% of the cases. The median number of days with metamizole therapy was 2 days (IQR 1–3). Metamizole cases received a median of 4 (IQR 2–6) different drugs.

Between June 2015 and May 2020, the stimulated spontaneous reporting system of the paediatric hospital resulted in seven registered case reports involving metamizole. Four of these were agranulocytosis, one allergic shock, one rash and one intoxication with suicidal character. According to the World Health Organisation (WHO) causality assessment, five cases (71.4%) were rated to have a possible or higher causality to metamizole. The remaining two cases were one of the agranulocytosis, as this was due to cyclic neutropenia, and the suicidal intoxication with metamizole. These two cases were therefore excluded from this analysis.

Patients’ baseline characteristics, information about metamizole intake and concomitant medication are displayed in Table 2. None of the reported cases ended fatally. All patients recovered without remaining harm and were discharged in good general health condition. No ME associated with metamizole was reported via the stimulated reporting system within the study period.

The analysis of the electronic medical records revealed that metamizole was administered in nearly one-third of the inpatient stays receiving medication (3759/11,857) within a 5-year period on paediatric general wards of a large German university hospital. This proves that metamizole is commonly used and an important substance in paediatric drug therapy, even though licensed indications are restricted in Germany and its use is still prohibited in some other European countries.

Oehme et al. [17] compared the differences in drug exposure and ADRs at one of the general paediatric medical wards included in the present analysis between 1999 and 2008. One interesting finding was the general increase in prescription of analgesics and anti-inflammatory drugs between 1999 and 2008, especially paracetamol, ibuprofen and metamizole. The increase in antipyretic and analgesic medication was associated with a drop in antibiotic prescriptions [17]. To our knowledge, there are no up-to-date data describing the patterns of metamizole prescriptions in a German paediatric hospital. Therefore, one of the main questions motivating this work was whether metamizole inpatient prescription patterns in children and adolescents have changed since 2008. Our study revealed slightly lower metamizole prescription rates compared with the 2008 data by Oehme et al. (39.2% [2008] vs 31.7% [2015–2020]). However, differences in the patient cohorts of the two studies need to be taken into account. The 2008 cohort only included data from one ward with the main focus on infectious disease, where more antipyretic medication is usually required. In the present study, various general paediatric wards were included besides the infectious disease ward, such as those with a focus on neuropaediatric diseases. The median age (5.0 years [2008] vs 4.9 years [2015–2020]), as well as the number of prescribed drugs per patient/case (3 drugs, each), was similar for both cohorts [17]. In our study, the use of metamizole fluctuated from year to year in the 5-year observation period but remained in the 26–35% range in patients receiving medication during their inpatient stay. In the last observation year (06/2019–05/2020), there was a decrease in metamizole exposure and the number of hospitalised patients. This might be explained by the fact that the first lockdown of the coronavirus pandemic was imposed in March 2020. It has already been shown that the lockdown has led to a decrease in paediatric hospitalisation, in part due to a decline in communicable infectious diseases or cancellation of planned procedures [35]. For a more detailed investigation of a possible trend, indications and the use of alternative therapy options would have to be considered. However, given these comparisons, it can be assumed that there has been no significant increase in metamizole exposure in paediatric inpatients since 2008, as has been reported for the adult population in Germany [8].

In 2016, a survey by Witschi et al. [1] found that 68.6% of anaesthetists (n = 1467) use metamizole for perioperative, and 36.9% for postoperative pain treatment in children < 14 years of age. Intraoperatively, metamizole was used significantly more often compared with paracetamol and other NSAIDs [1]. These numbers once more prove the high relevance given to metamizole in paediatrics in Germany. Besides its excellent analgesic and antipyretic efficacy, other reasons for the high metamizole use may be its low organ toxicity, therapeutic index, few contraindications, low costs, and, especially for Germany, the possible reimbursement by health insurance companies [1, 8, 36, 37]. Another argument for metamizole use may be the reduction of opioid use after trauma surgery, which is, however, more important in the adult population [38].

Rashed et al. [39] showed that metamizole was the second most prescribed drug after ibuprofen in Germany. Reasons for that may be the lack of alternatives. The intravenous ibuprofen formulation was only recently licensed for patients aged from 6 years for symptomatic short-term treatment of acute moderate pain and fever when intravenous use is clinically justified and other routes of administration are not possible [40]. Previously, intravenous ibuprofen was only licensed for ductus arteriosus occlusion in premature infants [41], but was already used off-label as an intravenous analgesic and antipyretic drug in younger patients [42]. Although intravenous paracetamol formulations are available and licensed for short-term treatment of moderate pain and fever [43], there are concerns about its hepatotoxicity [44]. Metamizole is available as intravenous formulations and is licensed from the age of 3 months [45], which may be reasons for its popularity as a non-opioid analgesic in Germany.

Our data demonstrated that patients receiving metamizole had a statistically significant longer median length of stay than other patients. This could be partly explained by the severity of the illness of patients receiving metamizole therapy or the association with a prolonged length of stay due to a surgical procedure. However, our data also showed that the median number of days with metamizole therapy was only 2 days overall. We did not evaluate the indications in which metamizole was administered. Nevertheless, the low number of days with metamizole therapy and the high amount of parenteral use support our assumption that metamizole was only used short term and within the scope of its licence. In addition, a meta-analysis on metamizole-associated ADRs in adults concluded that short-term metamizole use in the hospital setting seems to be safe [9]. Except for the allergic shock, in our analysis all patients with a metamizole-associated ADR had started metamizole therapy at least 14 days prior to the event. This supports the hypothesis that long-term use should be considered critically. Further studies are needed evaluating the prolonged use of metamizole.

To our knowledge, this is the first study evaluating metamizole drug utilisation in combination with stimulated spontaneous reporting system data in paediatric inpatients. As drug utilisation from the same hospital was already published for 1999 and 2008, we could compare metamizole exposure between the studies directly. Drug utilisation data were automatically extracted from the electronic medical records, avoiding errors due to manual transcriptions. In addition, our methodology only included the doses that were actually administered. Especially in the case of on-demand medication such as analgesics, a considerable difference can occur between the doses prescribed and those actually administered.

Our data do not reflect the picture of all inpatients, as not all wards could be included in the analysis. This is because the electronic medical record from which the prescription data were extracted was only used on general paediatric wards. Therefore, oncology, surgical or intensive care patients were not included, and no conclusions can be drawn about the frequency of metamizole use on these wards. Moreover, our data only allowed the evaluation of metamizole exposure, but not a precise characterisation of clinical use by indication and dosage. However, from clinical experience, we know that metamizole is mainly used for short-term therapy following the hospital’s standard operating procedures (SOPs), such as for ‘postoperative pain’, ‘care after PEG placement’ and ‘first aid for scald and burn’. In these SOPs, the dosage information is as follows: single doses of 10–15 mg/kg, maximum 60 mg/kg per day [59]. Ziesenitz et al. showed that infants < 1 year of age need lower metamizole doses (5 mg/kg) to achieve equivalent adult exposure [60].

To date, there are only a few studies combining drug utilisation with stimulated reporting system data, as we did. By doing so, the individual drug risk can only be estimated on a population level but not on an individual level. Also, combining these sources for risk estimates, one has to be very cautious as the data come from different databases and may be biased. In the present study, it is hardly possible to define the population at risk, as we only analysed the inpatient data. As agranulocytosis can appear more than 28 days after metamizole intake [53], it could be the case that the patient has already been discharged, especially when taking into account the median length of hospital stay of 3 days (IQR 2–6) in our study. Moreover, three of our patients had taken metamizole in the ambulatory setting and are therefore not included in the utilisation data. For these reasons, calculating the reporting rate or the incidence is not possible from our point of view.