Article Text
Abstract
Objectives We aimed to describe administration of eight potentially harmful excipients of interest (EOI)—parabens, polysorbate 80, propylene glycol, benzoates, saccharin sodium, sorbitol, ethanol and benzalkonium chloride—to hospitalised neonates in Europe and to identify risk factors for exposure.
Methods All medicines administered to neonates during 1 day with individual prescription and demographic data were registered in a web-based point prevalence study. Excipients were identified from the Summaries of Product Characteristics. Determinants of EOI administration (geographical region, gestational age (GA), active pharmaceutical ingredient, unit level and hospital teaching status) were identified using multivariable logistical regression analysis.
Results Overall 89 neonatal units from 21 countries participated. Altogether 2095 prescriptions for 530 products administered to 726 neonates were recorded. EOI were found in 638 (31%) prescriptions and were administered to 456 (63%) neonates through a relatively small number of products (n=142; 27%). Parabens, found in 71 (13%) products administered to 313 (43%) neonates, were used most frequently. EOI administration varied by geographical region, GA and route of administration. Geographical region remained a significant determinant of the use of parabens, polysorbate 80, propylene glycol and saccharin sodium after adjustment for the potential covariates including anatomical therapeutic chemical class of the active ingredient.
Conclusions European neonates receive a number of potentially harmful pharmaceutical excipients. Regional differences in EOI administration suggest that EOI-free products are available and provide the potential for substitution to avoid side effects of some excipients.
- Neonatology
- Pharmacology
Statistics from Altmetric.com
What is already known on this topic?
Some excipients have been recognised as causing side effects in neonates.
The few recent single centre or single country studies suggest that neonatal medicines contain significant amounts of potentially harmful excipients.
What this study adds?
Administration of eight potentially harmful excipients to European neonates is described.
While only 27% of products contained potentially harmful excipients, two-thirds of neonates received at least one.
Significant regional differences in excipient exposure indicate substitution possibilities.
Introduction
Pharmaceutical excipients are essential components of medicines necessary to maintain quality and patient acceptability.1 It is desirable for excipients to have limited pharmacological activity. However, some (in this study called excipients of interest (EOI)) have been associated with toxicity in neonates (table 1).1–8 Specifically, pharmacokinetics of excipients in neonates differ from that of adults/older children and can be additionally affected by underlying medical conditions.1
The extent of excipient use in neonatal medicines is largely unknown. A few single centre and/or single country studies suggest that neonates receive a significant amount of excipients, including those potentially harmful.22–25
We hypothesised that administration of EOI to neonates is not always justified. Solvents and solubilising agents, for example, would be expected to be consistently required for a specific active pharmaceutical ingredient and so should not show major variations in the extent of use in similar products. Antimicrobials, in contrast, may not be required for intravenous preparations (at least for single use formulations). Sweeteners are needed only for enteral formulations. Their use can be expected to vary between gestational age (GA) categories as preterm neonates are less likely to receive enteral medicines than term infants. Excipient administration should not vary between geographical regions after accounting for differences in choice of active ingredient and route of administration.
We aimed first, to describe the extent of the administration of EOI in European Neonatal Intensive Care Units (NICU); second, to explore covariates associated with the use of these excipients and third, to relate their administration to rational principles of excipient use.
Methods
A multicentre web-based point prevalence study was performed as a part of the European study of neonatal exposure to excipients.26 ,27
A network of national contact persons was built to invite as many sites as possible. NICUs with >50% of admissions consisting of neonates from all European Union countries plus Iceland, Norway, Switzerland and Serbia were invited to record prescriptions for neonates aged up to 28 days, in the unit at 08:00 of the study day.
Data were collected during 1 day chosen by the unit between 1 January and 30 June 2012. Demographic data and all prescriptions, except blood products, glucose and electrolyte solutions, vaccines and nursery care topical agents, were registered. Each medicine was classified according to trade name (product), manufacturer, pharmaceutical dosage form, strength and route of administration. Each unit provided also data on hospital teaching status (yes/no) and level of care (first, second and third28 ,29).
Identification of excipients
Excipient content for each medicine was identified from the summary of product characteristics (SmPC). If needed, searches from the homepages of manufacturers and credible public databases (eg, http://www.diagnosia.com) were performed.
EOI were categorised in a non-exhaustive way into three groups based on their main function (solvents and solubilising agents, antimicrobial preservatives and sweetening agents) as shown in table 1.
Statistical analysis
The analysis was performed with Stata Software (V.12.1). Descriptive statistics were used to describe general excipient administration. To estimate the influence of covariates on the administration of each EOI, first, the effect of geographical region (North, South, West and East Europe according to the United Nations Statistics department30), GA category (<28 weeks, called extremely preterm; 28 to <32 weeks, very preterm; 32 to <37 weeks, late preterm and >37 weeks, term neonates31), department level and hospital teaching status on the administration of each EOI was studied in univariate analysis.
Furthermore, a three-step multivariable logistical regression analysis was applied. All covariates with p<0.05 from the univariate analyses (geographical region, GA category, department level) were fitted into mixed effects models. Model 1 aimed to identify variables associated with EOI administration to individual neonates (outcome variable administration of a specified excipient to a neonate, yes/no). East region and extremely preterm neonates were used as the reference. Model 2 aimed to describe the associations between above-mentioned covariates and the presence of a particular excipient in each prescription. In addition to covariates explored in model 1, active ingredient was included to adjust for variations in EOI administration by active ingredient (outcome variable presence of a specified excipient in a prescription, yes/no). Active ingredients were classified according to the first level of Anatomical Therapeutic Chemical (ATC) classification; the third-level grouping was used for antibiotics. The ATC group with the largest number of prescriptions with respective excipient was used as reference. Finally, the route of administration was added into the model 2 to estimate the potential for product substitution. Topical prescriptions were excluded from the latter analysis due to small numbers of prescriptions.
The likelihood ratio statistic for testing the null hypothesis of zero between-group variance confirmed the need for the addition of department to the models as a random effect to adjust for between-department variance.
Results
Altogether, 21 countries (Austria, Belgium, Bulgaria, England, Estonia, France, Greece, Hungary, Ireland, Italy, Latvia, the Netherlands, Lithuania, Malta, Norway, Portugal, Romania, Serbia, Slovenia, Spain and Switzerland) with 89 NICUs participated. Out of 5 572 859 live births in invited countries in the year 2010, 349 465 (6%) were covered by participating units. Third-level NICUs (73%) predominated; 21% and 6% of units provided second-level and first-level care, respectively.
The extent and nature of EOI administration
In total, 726 neonates received 2199 prescriptions for 562 products containing 246 active ingredients. Demographic characteristics of patients and prescription data are shown in table 2.
The number of prescriptions per neonate was inversely related to GA (Spearman correlation=−0.3949; p<0.0005).
Information on excipients was available for 530 (94%) products with 2095 (95%) prescriptions. EOI were found in 638 (31%) prescriptions, 452 (22%) contained more than one EOI. Enteral and topical formulations contained EOI more frequently than parenteral (OR 6.3, 95% CI 4.1 to 9.7 and OR 10, 95% CI 4.4 to 22.9, respectively) with no difference between enteral and topical medicines (figure 1).
In total, 27% of products contained EOI, almost two-thirds (n=456) of the neonates who received medicines were exposed to at least one EOI. Parabens, used in 397 (19%) prescriptions administered to 313 (43%) neonates, were used most, and benzalkonium chloride least frequently (table 3).
In univariate analysis GA category, geographical region and department level were significant covariates of EOI administration (data not shown).
Variation of excipient administration by GA
Administration of EOI by GA category is shown in figure 2; the median number of studied excipients per neonate was 1 (IQR 0–2) for all GA categories. In model 1, term neonates were less likely to receive parabens (OR 0.5, 95% CI 0.3 to 0.9), ethanol (OR 0.3, 95% CI 0.1 to 0.7) and benzoates (OR 0.3, 95% CI 0.1 to 0.8), while late preterm were less likely to receive ethanol (OR 0.3, 95% CI 0.1 to 0.9) and very preterm more likely to receive saccharin sodium (OR 3.3, 95% CI 1.3 to 8.4), compared with extremely preterm neonates. In prescription-based ATC group-adjusted analysis (model 2) these associations were not significant except for saccharin sodium (OR 2.8, 95% CI 1.1 to 7.3). The latter association was lost after adjusting for route of administration. In contrast to model 1, in model 2, prescriptions administered to late preterms were more likely to contain polysorbate 80 (OR 4.5, 95% CI 1.6 to 12.2) and those administered to term neonates were to contain polysorbate 80 (OR 13.8, 95% CI 4.9 to 39.2), propylene glycol (OR 8.4, 95% CI 3.2 to 21.8), saccharin sodium (OR 3.2, 95% CI 1.1 to 9.3) and sorbitol (OR 8.9, 95% CI 2.4 to 33.3) compared with prescriptions administered to extremely premature neonates. After adjusting for route of administration, these associations remained significant for all the above-mentioned EOI except for saccharin sodium (data not shown).
Variation of excipient administration by geographical region
Neonates in North and in all other regions were more likely to receive parabens (OR 3.1, 95% CI 1.1 to 8.9) and saccharin sodium (data not shown), respectively, compared with those in East. The lower use of polysorbate 80 was reported in South (OR 0.1, 95% CI 0.1 to 0.5) and propylene glycol in North (OR 0.04, 95% CI 0.01 to 0.2) compared with those in East. These regional effects were not changed in model 2 adjusted for ATC group and route of administration (data not shown).
Regional differences in paraben administration were driven by choice of vitamin preparations (118/477 containing parabens), responsible for 118 of 397 paraben-containing prescriptions. The proportion of vitamin prescriptions containing parabens varied from 3% to 24% (three of 98 in East and 33 of 137 in West) to a third (56/164 in North and 26/78 in South).
Domperidone, prescribed in North and South, and carnitine, prescribed in West, were responsible for 30% of prescriptions containing saccharin sodium, but were not used in the East. No reasons could be identified for variations in polysorbate 80 and propylene glycol administration.
Variation of excipient administration by active ingredient
ATC group was a significant determinant of administration for all EOI except polysorbate 80. The highest proportion of prescriptions with parabens was found in group J01G (gentamicin; OR 26.0, 95% CI 14.3 to 47.4), benzoates in group J01F (lincomycin, clindamycin; OR 84.5, 95% CI 8.8 to 812.7), sorbitol in group B (phytomenadione, iron; OR 4.8, 95% CI 2.1 to 11.2) and ethanol in group G (dinoprostone; OR 69.4, 95% CI 1.5 to 3249.1). However, the largest number of prescriptions with the above-mentioned excipients was associated with ATC group A (mainly enteral vitamins). The highest number and proportion of prescriptions with propylene glycol, saccharin sodium and benzalkonium chloride was found in ATC group B, A and S (tobramycin, chloramphenicol, ciprofloxacin: all topical), respectively. Medicines from ATC groups J01C (ampicillin, benzylpenicillin), J01D (cefotaxime, ceftazidime, meropenem), J01E (trimethoprim), J01M (ciprofloxacin), J01X (metronidazole, teicoplanin, vancomycin), L (filgrastim) and M (pancuronium) did not contain EOI.
Variation of excipient administration by route of administration
The parenteral route was associated with a lower likelihood of administration for parabens (OR 0.2, 95% CI 0.1 to 0.3), polysorbate 80 (OR 0.4, 95% CI 0.2 to 0.95), benzoates (OR 0.3, 95% CI 0.2 to 0.7), propylene glycol (OR 0.4, 95% CI 0.2 to 0.7) and sorbitol (OR 0.003, 95% CI 0.00003 to 0.03). Saccharin sodium was found exclusively and ethanol in 79% (37/47) in enteral formulations and benzalkonium chloride in 85% (23/27) of prescriptions in topical formulations.
Discussion
This is the largest multi-country prospective study looking at the administration of pharmaceutical excipients to neonates. The diversity of the current pharmaceutical market was well reflected by the surprisingly broad range of medicines with highly variable excipient compositions. We demonstrate that while a quarter of used products contained at least one EOI, two-thirds of treated neonates were exposed to at least one of them. This suggests that a few commonly used medicines were responsible for a large part of the EOI ‘load’, meaning that substitution or reformulation of a relatively small number of products may spare a large number of neonates from unnecessary exposure.
Importantly, we observed regional variations in administration of four EOI (parabens, polysorbate 80, propylene glycol and saccharin sodium). Some of them may be driven by differences in active ingredients within the ATC group, whereas others more likely rise from hospital routines and policies. Region remained a significant independent determinant of EOI administration even after adjusting for covariates that might explain these differences. Some active ingredients are administered with EOI in some regions while EOI-free formulations of the same active ingredients are used in others. This indicates that EOI-free formulations are feasible and available on the European market and suggests the possibility for product substitution.
Other factors associated with the excipient use were route of administration and GA. Parenteral medicines, often produced as single dose vials where preservatives can be avoided, contained fewer EOI compared with topical and enteral formulations. In the latter additives are often needed to achieve acceptable palatability.32 Some of the variations between GA bands in EOI administration are explained by the use of different medicines in different GA categories, as was observed for parabens, benzoates and ethanol, which extremely preterm neonates were more likely to receive compared with term infants. Specifically, prescribing of enteral nystatin (contains parabens and ethanol) is more frequent in extremely preterm than in term neonates.
Our findings in multi-country settings were similar to previous studies looking at the administration of potentially harmful excipients to neonates in a single unit/country. Lass et al24 described the use of 123 excipients with 88% of neonates receiving at least one of the eight EOI in Estonian NICUs during 6 months. Similarly, in a Brazilian neonatal unit, 87% of neonates were exposed to at least one EOI.25
The pattern of covariates associated with the administration of EOI suggests rational use in many medicines, while a margin for improvement is suggested for others. First, we anticipated that solubilising agents or solvents are consistently required for a specific active ingredient. Our data suggest that only ethanol was used consistently. Both polysorbate 80 and propylene glycol use was associated with geographical region in that prescriptions in some regions were less likely to lead to exposure to these excipients, even after adjusting for ATC group. This suggests that formulations free of these excipients are available.
Second, antimicrobials were expected not to be required for parenteral formulations. The absence of parabens and benzoates in 85% of parenteral prescriptions suggests that administration of these excipients can largely be avoided. Regional differences in parabens administration suggest the availability of paraben-free alternatives on the European market. For example, the administration of parabens with some parenteral gentamicin formulations but not others suggests that administration of this EOI with gentamicin can be avoided.
Third, sweeteners were expected to vary only between routes of administration (not needed for parenteral medicines) and GA (enteral medicines less likely to be given to extremely preterm neonates). This was true for sorbitol, but regional differences were noted in the use of saccharin sodium, confirming our hypothesis that its use may not be essential. Specifically, saccharin sodium-free versions of enteral furosemide are available.
The absence of evidence of harm for many excipients is likely to indicate safety rather than being a cause for concern.27 It is accepted that excipients may only show side effects above a toxic levels.11 Unfortunately, these levels are often not known. The need for excipient toxicokinetic studies has been increasingly recognised.5 ,33 ,34 However, only 8% of SmPC, reviewed by us, contained quantitative data on excipients.
Some limitations need to be noted. First, of the 31 invited countries, about two-thirds participated in the study, omitting some large European countries (eg, Germany, Poland). Second, grouping of active ingredients to the first level of ATC classification was feasible due to the relatively small number of prescriptions for each separate ingredient. Therefore, some effects may be driven by variations within the ATC group. This may explain some findings at the GA level.
Conclusion
Neonates in European NICUs receive a number of excipients considered to be harmful. Comprehensive benefit/risk assessments for the use of excipients cannot be implemented until quantitative information is made available to professionals. Regional variations in neonatal administration of some potentially harmful excipients suggest a possibility to reduce exposure to parabens, polysorbate 80, propylene glycol and saccharin sodium through product substitution. However, a joint effort of all stakeholders will be required. Current administration rates and known developmental risks in neonates warrant further toxicokinetic studies as well as exploration of precise substitution possibilities.
Acknowledgments
We thank all national contact persons who provided lists of neonatal units and helped run the study in their own country: Bernhard Resch (Austria), Pieter De Cock (Belgium), Nelly Jekova (Bulgaria), Elisabeth Iyore (Denmark), Pascal Vaconsin (France), Kosmas Sarafidis (Greece), Aranka Vegso (Hungary), Noreen O'Callaghan (Ireland), Rocco Agostino (Italy), Daiga Kviluna (Latvia), Rasa Tameliene (Lithuania), Rene F. Kornelisse (Netherlands), Dag Bratlid (Norway), Almerinda Pereira (Portugal), Maria Livia Ognean (Romania), Milica Bajcetic (Serbia), Darja Paro (Slovenia), Elizabeth Valls (Spain), Per Nydert (Sweden), Hans Ulrich Bucher (Switzerland) and Maria Cordina (Malta). We also thank local pharmacists for providing data on excipient content: Caroline Fonzo-Christe (Switzerland), Domenico Tarantino (Italy), Velina Grigorova (Bulgaria), Milica Bajcetic (Serbia), Elizabeth Valss (Spain), Claudine Milstein (France), Jennifer Duncan (England), Sabina Zalar (Slovenia) and Per Gustaf Hartvig Honoré (Denmark).
References
Footnotes
Collaborators The following are members of ESNEE (European Study of Neonatal Exposure to Excipients): MAT (project leader, UK), Susan Graham (UK), Jennifer Duncan (UK), AJN (UK), Utpal Shah (UK), Hussain Mulla (UK), Hitesh Pandya (UK), James McElnay (UK), Jeff Millership (UK), Shirish Yakkundi (UK), Andre Rieutord (France), Thomas Storme (France), Pascal Vaconsin (France), IL (Estonia), TM (Estonia), HV (Estonia), GN (Estonia). Susan Graham; Jennifer Duncan;Utpal Shah; Hussain Mulla; Hitesh Pandya; James McElnay; Jeff Millership; Shirish Yakkundi; Andre Rieutord; Thomas Storme; Pascal Vaconsin; Bernhard Resch; Pieter De Cock; Nelly Jekova; Elisabeth Iyore; Kosmas Sarafidis; Aranka Vegso; Noreen O'Callaghan; Rocco Agostino; Daiga Kviluna; Rasa Tameliene; Rene F. Kornelisse; Dag Bratlid; Almerinda Pereira; Maria Livia Ognean; Milica Bajcetic; Darja Paro; Elizabeth Valls; Per Nydert; Hans Ulrich Bucher; Maria Cordina
Contributors Study design: TM, IL, GN, HV and MAT. Conduct of study: GN, TM and HV. Identification of excipients: GN, JL and IM. Analysing data: KT, GN, TM, IL and AJN. Writing the manuscript: GN, IL, TM and MAT. All members of ESNEE designed the study, monitored data collection and revised the draft paper.
Funding ESNEE is funded through ERA-NET PRIOMEDCHILD by the following national agencies: MRC from the UK, Grant G1100158, Estonian Science Foundation from Estonia, Estonian Target Financing SF0180004s12, project SARMB11016E, Agence Nationale de la Recherche from France.
Competing interests JL, AJN and IL are members of the Paediatric Committee (PDCO) of the European Medicines Agency; MAT is a Chair of the European Network for Paediatric Research at the European Medicines Agency (EnprEMA).
Ethics approval Ethics Committee approval was obtained in compliance with national guidelines. All data were anonymised before leaving the study sites.
Provenance and peer review Not commissioned; externally peer reviewed.