Introduction

It is estimated that critically ill newborns in neonatal intensive care units (NICUs) receive up to15 to 20 medicines routed intravenously [10, 15] and may be exposed to over 20 different excipients per day, depending on the number and dosage form of the drugs that they receive during hospitalisation [34]. This merits great concern, given the limited knowledge available regarding the full impact of developmental immaturity on the safety of drugs and their excipients in newborns, especially in those that are more immature and/or affected by diseases and specific conditions [1, 13, 14].

Serious and even life-threatening adverse events have been associated with exposure to the excipients present in drugs, when administered in higher doses or to vulnerable population groups such as neonates, particularly, those with low birth weight. These reactions may occur because of the immaturity of the organs responsible for biotransformation and elimination, which results in accumulation of these substances and their ensuing the toxicity, as has been observed in relation to propylene glycol (PG) [2, 13, 14, 26, 34].

Currently, certain excipients are recognised as toxic to newborns if they are given at a high dose. These include sodium benzoate, PG, parabens (methyl and propyl parahydroxybenzoate), sodium saccharin, benzyl alcohol (BA), benzalkonium chloride, polysorbate 80 and ethanol, which are present in medicines used commonly in this population worldwide [17, 21, 26, 28, 32–34]. Lass et al. [17] identified exposure of hospitalised neonates in Estonia to these and other excipients and made an attempt to categorise these substances in terms of safety, according to published information.

We are not aware of any data relating to excipient exposure among neonates in countries that are not included in the region covered by the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH), such as Brazil. This paper describes the exposure to excipients among neonates admitted to the NICU of a public care service in Brasilia, Brazil, applying the categories of Lass et al. [17] for risk analysis.

Methods

This was a retrospective study based on data from prescribed formulations, which was conducted in the NICU of a mother and child hospital in Brasilia, Federal District, Brazil, between January and March 2012. This is a tertiary-level institution specialising in caring for pregnant women and children, with 345 inpatient beds, of which, 46 beds are in the NICU (26 active currently) and 14 are in the paediatric surgery sector. The study subjects were neonates (postnatal age ≤28 days) who were admitted for more than 24 h during the study period. Data regarding intravenous hydration, vaccines, blood and parenteral nutrition were not recorded. Patients with incomplete clinical data, incomplete prescription or prescription of intravenous hydration alone were excluded. The research was approved by the Research Ethics Committees of the Health Department of the Federal District (Brazil), under Protocol No. 021/2012.

Data gathering

The prescription data and other information were gathered from the patients’ electronic medical records. The following data were recorded on specific forms: gestational age, birth weight, gender, date of birth, diagnoses, prescribed formulations data (name of the drug, laboratory, dosage form, administration beginning and end of treatment) and outcome within the observation period (discharge, death, transfer or remained hospitalised).

Neonates were classified according to the gestational age (GA) as preterm (<37 weeks of GA) and full-term (≥37 weeks); and according to gestational weight as very low weight (<1,500 g), low birth weight (1,500 to 2,500 g) and above (>2,500 g) [10].

Drugs were grouped into therapeutic classes in accordance with the Anatomical Therapeutic Chemical classification [35].

The composition of prescription formulations were determined from the drug package leaflets. If this information was not found, the electronic package leaflet database of the National Agency for Sanitary Surveillance (ANVISA) [3] and the website http://www.bulasmed.com.br [6] were consulted.

The excipients identified were categorised as proposed by Lass et al. [17], into four categories: (a) Potentially harmful and known to be harmful—adverse reactions reported; (b) Potentially safe—no adverse reactions reported; (c) No safety data found—no data found in the literature on human exposure and toxicity; (d) Manufacturer’s description of the excipients does not allow a specific literature search. The excipients for which no classification was included in the paper by Lass et al. [17] were categorised based on the information present in the Handbook of Pharmaceutical Excipients (6th edition) [24] and on the websites http://hazmap.nlm.nih.gov [31] and http://toxnet.nlm.nih.gov [32] (Fig. 1). The excipient classification was certified independently by three evaluators (MSGM, LAAB and ASSJ) and cases of discordance were resolved through discussion among them to reach a consensus.

Fig. 1
figure 1

Categorisation procedure for excipients found in prescription drugs in the NICU

Full size image

Data processing and statistical analysis

The exposure was estimated using the number of medications containing an excipient divided by the total number of prescriptions and using the incidence rate (IR) for each excipient. In IR, the numerator was the number of neonates exposed to each excipient once or more times and the denominator was the total number of neonate days at risk of using an excipient. The IR made it possible to evaluate the exposure of newborn infants to each excipient. For analysis, the IR was subdivided according to the GA. The data were stored in Excel for Windows version 7 and analysed in the Statistical Package for the Social Sciences version 18, using simple descriptive analysis.

Results

Characteristics of the neonates

During the study period, 89 patients were admitted to the NICU, among whom 10 (11.2 %) were excluded (4 had postnatal ages of more than 28 days of postnatal age, 4 presented incomplete data, and 2 only received intravenous hydration or parenteral nutrition). Of the 79 neonates included in the study, 64 (81.0 %) were preterm and 15 (19.0 %) were full-term; the median GA was 33.4 weeks (GA range 22–40 weeks); 40 (50.6 %) were male; the mean length was 42.1 cm (standard deviation (SD) ± 5.3 cm); the mean birth weight was 1,793.0 g (SD ± 855.7 g); and 63 (80.7 %) had low birth weight (<2,500 g), including 35 of very low birth weight (<1,500 g). The median length of hospital stay was 18 days (range: 1–75 days; mean 22.6 days; SD ± 17.7 days), thus totalling to 1,791 neonate days, and 24 deaths occurred during the period. The most frequent diagnoses that led to admission to the NICU were: respiratory distress (30.1 %), jaundice (11.6 %), sepsis (8.4 %), bacterial/fungal infection (6.8 %) and hyaline membrane disease (5.3 %).

Prescribed formulations

The neonates in the study received 1,303 prescriptions (median: 10.0 prescription/patient; interquartile range (IQR) 8.5; range 1–70) for a total of 733 prescribed formulations (median nine formulations/patient; interquartile range 4.5; range 1–26) with 77 different formulations and 70 different active pharmaceutical ingredients (API). The preterm neonates had a higher number of prescriptions (median 11; IQR 22; range 1–70 prescription/neonate); for the full-term neonates, the median number was 9 (IQR 7; range 1–19 prescription/neonate). The therapeutic classes that were most prescribed were drugs for the alimentary tract and metabolism (n = 21; 29.2 %), anti-infective for systemic use (n = 16; 22.2 %) and drugs for the cardiovascular system (n = 11; 13.9 %). The five formulations that were most prescribed were aminophylline as an injection solution (n = 566; 9.8 %), multivitamin without minerals as an injectable solution (n = 409; 7.0 %), multivitamin without minerals as an oral solution (n = 404; 6.9 %), domperidone oral solution (n = 324; 5.6 %) and ranitidine as an injectable solution (n = 294; 5.0 %).

Regarding the dosage form of the 77 formulations, 51.9 % (40) were injectable, 22.0 % oral solutions, 9.0 % tablets, 7.8 % oral suspensions, 2.6 % oral emulsions, 2.6 % ophthalmic solutions, 1.3 % pulmonary inhalation solution, 1.3 % sterile suspension for intratracheal administration and 1.3 % ointments.

Classification of the excipients present in prescribed formulations

Of the 77 formulations investigated, it was only possible to obtain the drug package leaflet of 58 and only 53 of these cited the excipients. The mean number of excipients per formulation was 5.3 (range 1–14) with a total of 86 excipients, of which, 21 were mentioned in the paper by Lass et al. [17]. The excipients were categorised as follows: (a) potentially harmful and known to be harmful (n = 57; 66.2 %); (b) potentially safe (n = 8; 9.3 %); (c) no safety data found (n = 1; 1.2 %) and (d) description of the excipients in the leaflet is unspecific (n = 20; 23.3 %). Excipients in category “a” were subdivided into harmful excipients (HE, n = 9) and potentially harmful excipients (PHE, n = 48) which were found in 48 formulations, among which 4 formulations only presented HE, 24 only presented PHE and 20 had both types of excipients (Tables 1 and 2). The excipients citric acid, hydrochloric acid, stearic acid, water, starch, microcrystalline cellulose, sodium chloride and simethicone were included in category “b” and were present in 43 formulations, used as diluents, suspending agents, solvents and antioxidants. Category “c” included carmellose sodium present in two formulations and functioned as suspending agent. Category “d” included flavouring agents, scents, and colourants, such as mint essence, artificial orange flavour and iron oxide yellow colourant.

Table 1 Harmful excipients identified in formulations prescribed to neonates (1,791 neonate days) in a NICU in Brasilia, Brazil, January–March 2012
Full size table
Table 2 Potentially harmful excipients identified in formulations prescribed to neonates (1,791 neonate days) in a NICU in Brasilia, Brazil, January–March 2012
Full size table

Neonates’ exposure to harmful and potentially harmful excipients

Sixty-nine neonates (87.3 %) were exposed to one or more excipients known to be harmful (HE), with median of three HE per neonate (IQR 8; range: 1–8) and a median of three formulations with HE per neonate (IQR 8; range: 1–9). Almost all the preterm neonates were exposed to HE (87.5 %) and 86.6 % of full-term neonates were exposed at least once. Among the formulations that were prescribed with HE, preterm neonates presented a median number of three formulations with HE (IQR 2; range: 1–9), and for the full-term neonates, the median was also three formulations with HE (IQR 2.5; range: 1–9).

Table 1 shows the IR of excipients per 1,791 neonate days during the study period. Polysorbate 80 had an IR of 3.23 per 100 neonate days, meaning that for every 100 neonates who remained hospitalised for 1 day, three were exposed to this excipient. The risk of exposure to other excipients was less than this, in the following decreasing order: PG, propylparaben, methylparaben, BA, saccharin, ethanol, sodium benzoate and benzalkonium chloride.

According to the IR and GA, the preterm neonates (1,502 neonate days) were at increased risk of exposure to polysorbate 80 (3.26/100 neonate days), followed by PG (3.19), propylparaben (3.06), methylparaben (2.39), ethanol (2.33), BA (2.06), sodium saccharin (1.86), sodium benzoate (1.19) and benzalkonium chloride (0.33). The full-term neonates (289 neonate days) presented greater risk in relation to ethanol (3.8), methylparaben (3.11), polysorbate 80 (3.11), propylparaben (2.76), BA (2.42), PG (2.42), sodium saccharin (1.73) and sodium benzoate (1.38). The full-term neonates were not exposed to benzalkonium chloride.

Nearly all the newborns (98.7 %) were exposed to one or more PHE. Among the formulations prescribed, the median number of PHE included in prescriptions for the neonates was 7 (IQR 6.2; range 1–19). Among these, the median for preterm neonates was 7 (IQR 6.0; range 1–19) and for full-term neonates were 2.5 (IQR 6.0; range 1–7). The risk of exposure was higher for sodium hydroxide, sodium citrate, sodium bisulphite, acetic acid and phenol (Table 2).

In relation to IR and PHE, the preterm neonates (1,502 neonate days) presented increased risk of exposure to sodium hydroxide (3.39/100 neonate days), ethylenediamine (2.99), sodium citrate (2.92), acetic acid (2.92) and sodium acetate (2.92). The full-term neonates (289 neonate days) presented increased risk of exposure to phenol (4.84/100 neonate day), sodium citrate (3.46), anhydrous sodium hydrogen phosphate (3.11), sodium hydroxide (2.42) and sodium bisulphite (2.42).

Discussion

To our knowledge, this is the first report on excipient exposure from a country that is not included in a region represented in the ICH and the first to express the rate of excipient exposure according to neonate days. The present study describes the exposure of an unselected group of neonates that were hospitalised in a NICU in Brasilia, the capital of Brazil, to a wide variety of HE or PHE that are present in commonly used medicines. This is a situation similar to that found in other countries [17, 21, 26, 33, 34]. The study population consisted primarily of preterm neonates, which increases the significance of these observations, given the greater vulnerability of this population to the toxicity of drugs and excipients, and the little that is known about this subject [1, 13].

Prescribing in public institutions in Brazil is standardised at each institution based on the list of essential medicines in the country, with variations in the profile with regard to local factors, product availability and prices [7]. The registration of new drugs is regulated by ANVISA, yet there are no specific rules for paediatric medicines or restrictions on the presence of excipients in them.

The therapeutic classes that were found to be most prescribed in this study were drugs for the alimentary tract and metabolism, anti-infective drugs for systemic use and drugs for the cardiovascular system, which are commonly used in NICUs [5, 9, 22]. We identified a large number of excipients with different functions and safety profiles [12, 17, 23, 33, 34]. About two-thirds of the excipients identified were classified as HE and PHE, which were present in formulations that were used frequently and concurrently. This highlights the risk of addition, accumulation and enhancement of toxic effects, including interaction with the API present in the formulations [12, 19, 26–28].

Nearly all the neonates were exposed to HE and PHE that whose toxicity appears to be higher at lower gestational age [2, 17, 26, 34]. The number of formulations containing HE and PHE were similar to what was found by Lass et al. [17], differing only in the comparison between groups of GA in relation to HE. The neonates of the present study were at increased risk of exposure to polysorbate 80 (7 formulations/610 prescriptions), thus differing from the findings of Lass et al. [17] (4 formulations/70 prescriptions). This was despite the limited overlap in excipients between the two surveys (24 % of excipients seen in Brazil were reported in Estonia). The European Study of Neonatal Exposure to Excipients preliminary results of the point prevalence survey (ESNEE-PPS) reported that 42 % of 825 neonates were exposed to parabens [21]. These differences may have occurred because we have included all the formulations that were prescribed (e.g., phytomenadione injections). This product contains polysorbate, an excipient that is associated with E-Ferol syndrome which occurs in neonates that receive vitamin E preparations with large amounts of polysorbate 80 [24].

With regard to exposure to alcohols, the neonates in our study had a higher risk of being exposed to PG, BA and ethanol, differently from Lass et al. [17] who demonstrated lower exposure to BA, PG and ethanol. However, our findings were similar to the preliminary ESNEE-PPS results, if benzoic acid is considered to be a BA metabolite [21]. The acceptable daily intakes of PG and ethanol are not well known for this population, but their use is not recommended for children under 4 years of age [11]. The pharmacokinetics of PG have been described, but the “safe” circulating concentration remains uncertain [8]. Furthermore, recent data suggest that there is a lower limit of PG tolerance in neonates [16]; for example, no adverse reactions were associated with short-term exposure to a median PG dose of 34 mg/kg/day [2]. It has been acknowledged that large amounts of PG and its interaction with ethanol, as reported by the United States Food and Drug Administration in the case of lopinavir/ritonavir (Kaletra®) [30], possibly contributes towards the toxicity of PG. BA (which is contraindicated by the FDA and the European Medicines Agency in formulations for neonates) is present in heparin sodium injection solution, which is usually used to extend the access of an umbilical arterial catheter in neonates. This was also used in Estonia according to Lass et al. [17]. In a study conducted in the USA, Shebab et al.[26] observed that neonates received daily intake of BA and PG 21 to 180 that was times higher than the maximum recommended daily dose for adults (5 and 25 mg/kg/day, respectively). Also, Whittaker et al. [34] in England observed exposure to significant daily intake of ethanol among neonates who weighed between 1.0 and 3.5 kg (0.2 to 1.8 mL/week, not adjusted by weight).

The toxicity among neonates relating to the preservative sodium benzoate is assumed to derive from benzoic acid accumulation, which may cause metabolic acidosis and neurotoxicity [20]. We observed that sodium benzoate was present in four of the prescribed formulations, among which domperidone oral solution is frequently used in the treatment of gastrointestinal reflux, although its use is controversial [18]. Lass et al. [17] reported that benzoic acid was present in four formulations, among which simethicone oral suspension was the one most prescribed.

Propyl and methylparaben were present in eight and nine formulations, respectively: injectable fentanyl solution (251 prescriptions) and multivitamins without minerals in oral solution (409 prescriptions) were the most frequently used of these formulations. These excipients are widely used as preservatives in the pharmaceutical and food industries, and are often encountered in medications used in NICUs, as suggested by some authors [4, 17, 21]. Lass et al. [17] also reported that there was a high frequency of prescribed formulations containing propyl and methylparaben, with gentamicin injection as the type most prescribed (200/1,971 prescriptions). Also, the preliminary ESNEE results [21] showed that there was high prevalence of exposure to parabens (42 % out of 825 neonates in 21 countries). In research on paediatric formulations marketed in the Netherlands (3,542 formulations), van Riet-Nales et al. [33] detected the presence of methyl and propylparaben in oral liquid formulations (77 and 45, respectively) and injectable formulations (9 and 1, respectively). Our results may differ from those of Lass et al. [17] and from the preliminary ESNEE results [21] for a number of reasons, including the way in which parabens were recorded. However, even considering the frequency of parabens as a group, the polysorbate exposure appears to be greater in Brazil than in Europe.

Sodium cyclamate, sorbitol and sodium saccharin were identified. The last of these was present in the most widely prescribed products (33 neonates/909 prescriptions), i.e., multivitamins without minerals in oral solution (40 neonates/404 prescriptions) and domperidone oral solution (20 neonates/324 prescriptions). It was also found by Lass et al. [17] in six formulations (173/1,971 prescriptions), among which simethicone oral suspension stood out (108 prescriptions). The use of sweetening agents in formulations for neonates is not recommended because of the lack of established safety data and the known risks of hypersensitivity (sodium saccharin) and laxative action (sorbitol) [12, 24]. However, it also needs to be taken into consideration that the production of excipient-free formulations may not be economically viable.

The clinical significance of these results is unclear, mainly because of the lack of quantitative information about excipients in medicines in our setting. This problem has been reported by other authors [2, 17, 25, 33, 34]. Manufacturers regard this information as proprietary and commercially sensitive. This makes it difficult to assess the extent of exposure to excipients, which was also reported by the ESNEE [25] and the PG experience [16]. The lack of quantitative information about excipients in medicines hinders research and prevents rational risk management by clinicians and regulators [29]. The present study was conducted over a short period and it should be borne in mind that exposure to excipients may continue during and after hospitalisation, as demonstrated by Whittaker et al. [34]. Moreover, it should be noted that the reports of excipient toxicity in neonates reflect unusually high exposures. Routine surveillance for the known adverse effects of excipients among neonates has not been reported often, although no problems were revealed as effects of PG [2]. Excipients are sometimes required to solubilise the medicine or to preserve the formulation, so as to extend the shelf-life of the medicines and improve the marketability of these medicines.

This paper provides the first detailed description of neonate exposure to toxic excipients in a NICU in Brazil and calls their risks into question. Our study applies the classification proposed by Lass et al. [17] to the reported excipients and complements that report with the inclusion of excipients not mentioned by these authors. There is clearly a significant difference between excipient exposure in different countries which needs to be accounted for in regulatory action and research. These differences may reflect differences in choice of active agents, differences in availability (due to market availability or price) or regulatory differences.

We recognise that this study has limitations that were inherent to its retrospective nature and to the fact that it was based on records from a single NICU. On the other hand, the electronic medical records with the analysis of over 1,303 prescriptions accumulated during 3 months, gives more reliability to the data. These results can be generalised to neonates in other public NICUs in Brazil, although there are regional and local differences in terms of the characteristics of these populations and health service practices.

It is important to remind prescribers about the need to take into account the excipients present in the medicines prescribed for neonates. Monitoring for adverse events potentially related to excipients should be part of the drug use routine in the NICU, as proposed by Giacoia et al. [13]. Neonates have not yet benefited appreciably from international efforts to promote the development of better medicines for children. Therapeutic arsenal for this population group remains very limited and the risk-benefit balance of excipients is not adequately addressed in many situations. However, promising studies are being conducted in Europe, like the European Study of Neonatal Excipient Exposure, which aim to assess and mitigate the risk posed by excipients in neonates [28]. Despite the risks, it is important to remember that excipients are necessary in medicines and it may not be economically feasible to develop medicines that do not contain excipients that are targeted exclusively to neonates. The risks can be minimised by actions such as excluding excipients associated with harm whenever possible, limiting the concentrations of excipients which are indispensable and informing clinicians about the quantitative composition of medicines. These actions should be regulatory requirements and good practice in the pharmaceutical industry in all countries, including Brazil.