In the last decades, the survival of extremely low birth weight (ELBW) neonates (BW ≤ 1000 g) has improved dramatically due to advances in perinatal and neonatal care and better understanding of their physiopathology . These neonates, characterized by a great immaturity, are more exposed to risks to develop different morbidities and in particular are vulnerable to bacterial and fungal infections (up to one third develop hospital acquired infections) due to the immaturity of the immune system and to predisposing factors such as maternal chorioamnionitis, ventilator care, catheterization and total parenteral alimentation .
Suspected infections are frequent in preterm newborns and the incidence of sepsis is inversely associated with BW : this leads to a common use of antiinfectives in ELBW neonates in the first days of life often on a prophylactic basis . If treatment is delayed, neonatal sepsis can be rapidly fatal, making optimal use of antibiotic essential. On the other hand, a prolonged antibiotic empirical therapy is associated to adverse effects (alteration of gut microflora, fungal colonization and subsequent invasive infection, necrotizing enterocolitis, late-onset sepsis and death) and could lead to unnecessary exposure causing a selective pressure for antibiotic resistance, especially in the case of late-onset infections . Therefore, empirical therapy should be applied only when necessary at the best possible option .
At this moment, a large variability in the use of antibiotics for the treatment of suspected/confirmed neonatal sepsis persists between different European Neonatal Intensive Care Units (NICUs) and most agents are still used in an off-label (OL) manner . In fact, with some exceptions antibiotics are licensed for use in the neonate, but are frequently administered with different modalities particularly as regards dosage and frequency .
This paper is focused on antibiotic prescriptions in a cohort of ELBW neonates included in a multicentre study involving a representative sample of Italian NICUs , with the purpose to describe practices concerning prevention and treatment of bacterial infections, in particular the extent and nature of OL antibiotic use in this setting. Moreover, the prescription behaviour was compared with indications contained in the British National Formulary for Children (BNFC) 2016-2017 and in a practical guide to the use of drugs in newborns prepared by the Neonatal Pharmacotherapy Study Group (NPSG) of the Italian Society of Neonatology (ISN).
2. Patients and Methods
2.1. Study Design
A sub-analysis of a multicentre one-day study involving all 107 level III Italian NICUs was performed .
2.2. Data Collection and Analysis
For current analysis, data from each ELBW infant present in NICUs and receiving at least an antibiotic treatment in the day chosen within each ward between May and July 2014 were retrieved from data collected and recorded in an online questionnaire (Google form), after sought and information to local ethics committees.
Anonymised demographic data included date of birth, sex, gestational age (GA) and birth weight (BW), post-natal age (PNA), Apgar score, diagnosis. Moreover, all information about each antibiotic administered during the day chosen was retrieved: formulation, route of administration, individual dosing regimen (unit dose and dosing interval), length of therapy, indication for use, tolerability. As personal identifying data of the infants could neither directly or indirectly be attributed to a specific individual and the study design did not affect the healthcare of the included patients, a formal written consent for participation in this study was not obtained.
For each antibiotic, the licensed or OL use was determined according to the Italian Drug Compendium 2013. This classification was based on information derived from product data sheets (package insert, Summary of Product Characteristics).
Antibiotic prescriptions were classified into three groups: 1) antibiotics following the marketing authorization (on-label prescriptions); 2) antibiotics with no information for use in neonatal population (off-label for age); 3) antibiotics licensed for use in neonates, but off-label for dose, frequency, route of administration, length of therapy and clinical indication.
In addition, every prescription was compared with the BNFC 2016-2017 , commonly accepted as one of the few dosing references, and with a practical guide proposed by the NPSG, containing information about all medicines commonly used in NICU and available both as book and online to all Italian neonatologists. Dosing recommendations for ELBW infants were based not only on BW, but also on GA and/or PNA.
Data were collected in a database and summarized using standard descriptive methods. Categorical variables related to prescription behaviour were compared by χ2 analysis: statistical significance was defined as p ≤ 0.05.
3.1. General Data
Among the 36 NICUs participating to the multicentre study (34% of all Italian level III NICUs, comprising hospital and academic wards) , 30 wards admitted ELBW neonates: each NICU participating in this study recorded a median number of six charts (range 2 - 22) and should be considered representative of the regional distribution and of the number of beds/ward (in every case > 4, with a maximum of 36 beds in some cases).
3.2. Patient Data
A total of 93 ELBW neonates were treated with at least one drug in the day chosen. As specified in Table 1, 52 were male, 50 with a G.A. between 26 and 28 weeks and 39 with a BW between 601 and 800 g: 21/93 (22.6%) were small for gestational age (SGA). On the day chosen for data collection, the median PNA
Table 1. Baseline characteristics of ELBW infants.
was 2 weeks (range: 0 - 28 days).
3.3. Prescription Data
Among a total of 367 drug prescriptions (54 different medicines), 92 regarded antibiotics given to 56/93 (60%) neonates: in 89/92 cases the antibiotic was administered intravenously (mostly by infusion). Ampicillin, gentamicin, amikacin and vancomycin were the antibiotics most commonly used for the prevention/treatment of bacterial infections (56/92, 61% of all antibiotic prescriptions). In 36% of cases (33/92 prescriptions) a coadministration of ampicillin and gentamicin was applied. In other cases, neonates were treated with vancomycin (14/92 prescriptions, 15% of cases) or with a broad-spectrum antibiotic such as ceftazidime, cefotaxime, ceftriaxone, imipenem, meropenem, clarithromycin (15/92 prescriptions, 16% of cases). The mean duration of treatments was 5 - 7 days in case of suspected infections, >10 days (mean 12 - 14 days) when the infection was proven or a diagnosis of sepsis was made.
3.4. Off-Label Prescriptions
50/56 ELBW neonates (89.3%) received at least one OL antibiotic prescription.
Prescriptions related to 20 different antibiotics, analysed according to their license status, are reported in Table 2. Some differences have been observed taking into account the birth weight of neonates, with a higher but not significant percentage of on-label prescriptions in newborns with a BW < 800 g (p = 0.583, NS).
The only antibiotics used according indications were ampicillin/sulbactam, cefotaxime and ceftriaxone, piperacillin and teicoplanin (23/92 prescriptions, 25%). 13 prescriptions (14%) regarded five antibiotics (meropenem, imipenem, piperacillin/tazobactam, clindamycin, clarithromycin) used in absence of a specific indication for use in the neonatal population. 56/92 antibiotic prescriptions (61%) resulted off-label, mainly as regards dosing frequency, while in few other cases prescriptions deviated for dose or formulation.
As specified in Table 3, in the current study off-label prescriptions resulted >60%, relatively low as regards vancomycin (35.7%) and ceftazidime (33.3%), >90%
Table 2. Antibiotic prescriptions in ELBW infants analysed according to their license status (χ2 test, NS = not significant).
Table 3. Analysis of off-label (OL) antibiotic prescriptions of the current study in comparison with ISN practical guide.
for the other antibiotics (in particular ampicillin, aminoglycosides and metronidazole). Compared to the ISN practical guide, only about 20% of prescriptions deviated and differences mainly regarded frequency of administration: the adherence was >80% for ampicillin and vancomycin.
As regards the four antibiotics most commonly prescribed in Italian NICUs to ELBW infants (ampicillin, gentamicin, amikacin, vancomycin), differences have been observed in dosing regimens between current study and recommendations contained in BNFC, particularly as regards ampicillin and gentamicin (Table 4).
Bacterial infections account for a major part of neonatal morbidity and mortality worldwide and antibiotics are extensively prescribed in neonates admitted to NICUs, particularly in ELBW infants .
The current analysis of data, regarding a subpopulation of ELBW infants enrolled in a nation-wide study comprising about one-third of all Italian level III NICUs , confirms the existence of different approaches between Italian NICUs towards the prevention and management of infections in ELBW neonates, resulting in a large variability in antibiotic use and in a high number of off-label prescriptions.
In detail, from our data different prescription behaviour emerge by the comparison with indications contained in BNFC and ISN practical guide, with a major adherence to local guidelines.
The large variability in antibiotic use undoubtedly derives from the difficult standardization of treatments in this neonatal subpopulation characterized by immaturity and rapid physiologic changes in the first postnatal weeks, but also by the lack of a unique source for prescribing guidance. This heterogeneity was
Table 4. Dosing regimens of the four more frequently used antibiotics in ELBW infants admitted to Italian NICUs in comparison with BNFC and ISN guide (intravenous infusion).
*serious infections or sepsis.
previously underlined by some authors as regards dosage schemes of gentamicin, vancomycin and ciprofloxacin and by other authors who compared antibiotic prescriptions in a NICU with recommendations contained in three commonly used reference sources (Pediatric Dosage Handbook, Neonatal Drug Formulary and Neofax), detecting divergent information on dosage schemes in preterm newborns. More recently, dosage regimens of antibiotics were analysed in French NICUs and a considerable inter-centre variability was observed, with doses and/or dosage frequency varying significantly for 12 antibiotics (aminoglycosides, vancomycin, penicillin G, oxacillin and cloxacillin, ceftazidime, imipenem/cilastatin, clindamycin and metronidazole) . The same variability has been observed in 89 NICUs of 21 different European countries: from the analysis of 586 antibiotic prescriptions referred to 37 different systemic antibiotics, deviations from BNFC dosage recommendations have been observed between European countries, with the use of higher doses for antibiotics well tolerated and of lower doses when safety concerns were evident .
Off-label drug use is a common practice in neonatal care and in many situations this is the only therapeutic alternative due to the lack of availability of suitable licensed/labelled drugs.
As regards antibiotics, the most commonly prescribed medications in NICUs, the high prevalence of off-label prescriptions in newborns derives in some cases by a lack of registered clinical trials , leading to absence of indication for use in this paediatric subpopulation. Despite some encouraging initiatives (such as the introduction of the European Paediatric Regulation) have been taken with the aim to reduce the use of off-label/unlicensed drugs in the paediatric population , new antibiotics approved in the last years in the EU have been rarely studied in the newborn (only 6/31 pediatric clinical trials enrolled neonatal population) and very few labelling changes specific for the neonatal population have been introduced . More frequently, in presence of indication for use in neonates off-label prescriptions are due to deviations from information contained in data sheets that rarely reflects clinical practice.
Our data, regarding a cohort of ELBW neonates where antibiotics were given on the basis of a diagnosis of bacterial sepsis (12 neonates) or of a suspected/proven bacterial infection (44 newborns), are in line with previously published results on off-label antibiotic use : only about one-fourth of prescriptions followed the terms of the marketing authorization and more than 80% of newborns received at least one off-label antibiotic prescription, with a 75% prevalence for use of antibiotics off-label for dosage schemes or for age (no indication in preterm infants).
As regards dosage schemes, the choice of the interval in administering antibiotics have important efficacy and safety concerns in newborns: if the dosing interval is too long antibiotic concentrations may drop below the MIC, while if the dosing interval is too short the antibiotic may accumulate in the body causing toxicity. Extended dosing intervals rather than divided doses are more commonly used in preterm newborns, even if supported by limited data . This approach, based on the conviction of potentially less toxicity, results also by our data and the deviation from indications reported in data sheets translates in a better adherence to ISN recommendations, derived by clinical practice, that take into account both gestational age and postnatal age.
In our cohort of ELBW neonates, the combination ampicillin+gentamicin was the most commonly applied regimen (33/56) for an empirical therapy of early-onset infections, in some cases (15/56) substituted with a unique broad-spectrum antibiotic (ceftazidime or another cephalosporin, meropenem or imipenem, clarithromycin).
Ampicillin prescriptions resulted almost all off-label (94.4%) for frequency (12 - 24 h instead of 8 h), while deviated by the suggested ISN recommendations in only 11% of cases. The most common dosage scheme applied was 50 mg/kg every 12 h and this suggests a discrepancy between generic information contained in data sheets (100 mg/kg/daily divided in three doses) and how this antibiotic is effectively given to newborns taking into account the characteristics of the patient and the indication (prophylactic or therapeutic use and severity of the infection) as specified in ISN guide: 50 - 100 mg/kg every 8 - 12 h in preterm newborns depending both on gestational age and postnatal age.
The same differences in dose and frequency of administration also regard gentamicin and other aminoglycosides (netilmicin and amikacin). Aminoglycosides have been licensed with the recommendation to divide the total daily dose into two administrations, while in clinical practice (ISN indications) it is preferred to adopt once-daily dosing or longer intervals in the first days of life (in particular for GA ≤ 29 weeks), as demonstrated in some clinical trials and also suggested by BNFC and by the National Institute of Health and Care Excellence . From our data, dosage schemes of these antibiotics resulted variable among NICUs (12 different combinations), with often the application of longer intervals of administration (36 - 48 h) and consequently lower total daily doses: this may reflect a “fear” of known adverse effects, such as nephrotoxicity, preferring to reduce the risks by limiting dosage as previously reported . This different prescription behaviour was previously underlined in a UK  and French survey and by other authors who compared antibiotic prescriptions related to 110 newborns admitted to some NICUs in the UK, Italy and Greece: the number of off-label prescriptions resulted significantly higher in Italy and Greece (92% compared to 63% in the UK), mostly regarding gentamicin and amikacin given at different total daily doses or frequency .
Some prescriptions (14/92) regarded vancomycin, a narrow-spectrum antibiotic usually administered for late-onset sepsis caused by Gram-positive bacteria in neonates presenting cardiorespiratory instability and in areas where methicillin-resistant Staph. aureus is prevalent . Although it has been used for >50 years, the dosing regimen remains a challenge in NICU and also for this antibiotic the use in clinical practice results variable . In our NICUs, 36% of prescriptions resulted off-label as regards frequency: every 18 - 24 h in preterm infants < 30 weeks (as partially suggested by ISN guide and BNFC), while data sheets generically reports every 8 - 12 h based only on postnatal age.
Finally, the use of antibiotics in absence of a specific indication in the neonatal population regarded five antibiotics (mainly meropenem, but also in specific situations imipenem, piperacillin/tazobactam, clindamycin and clarithromycin) and was cautious (in total 13 prescriptions, 14%) mainly referred to ELBW neonates with a higher BW: this kind of use was applied only in one case in presence of a BW < 600 g, where meropenem (20 mg/kg every 8 h) was given after diagnosis of late-onset sepsis. Although meropenem has been widely successfully used to treat severe infections such as sepsis and meningitis ≥ 3 months of life , until now it has not yet been registered for use, at least in Italy, in the neonatal population.
This survey, despite some limitations (in particular the self-report nature of the study and the number of ELBW neonates included) that do not allow a more accurate analysis of the data, confirms the high prevalence of off-label prescriptions and the variability in the use of antibiotics in ELBW neonates. The reasons for this high variability in dosage schemes of antibiotics in newborns are multiple. First, high quality data is lacking for many antibiotics, being current guidelines mainly based on expert opinion and small studies rather than on large clinical trials . Additional difficulties derive from changes in the pharmacokinetics of antibiotics in the first weeks of life . Other reasons regard problems in adoption and dissemination of evidence-based knowledge .
Given the paucity of data available on this neonatal sub-population, our recording and analysis of antibiotic prescriptions in a cohort of ELBW neonates could be a first step to introduce a data collection system useful to harmonize prescription behaviour.
Special thanks to all neonatologists who collaborated to the collection of the data (in alphabetical order): Antonietta Auriemma (AO Bolognini, Seriate, Bergamo), Adriano Azzali (Ospedale San Giovanni di Dio, Agrigento), Manuela Bedetta (Policlinico Casilino, Roma), Tatiana Boetti (Ospedale Sant’Anna, Torino), Lina Bollani (Policlinico San Matteo, Pavia), Raffaele Borrelli (Spedali Civili, Brescia), Angela Bossi (Ospedale Del Ponte, Varese), Roberto Bottino (Fondazione Poliambulanza, Brescia), Matteo Bruschettini (Istituto Gaslini, Genova), Elsa Buffone (AO San Camillo-Forlanini, Roma), Anna Casani (AO Rummo, Benevento), Giacomo Cavallaro (Fondazione IRCCS Ca’ Granda Ospedale Maggiore, Milano), Natalia Chukhlantseva (Ospedale Bambin Gesù, Roma), Elena Ciarmoli (Fondazione Monza e Brianza, Monza), Gloria Cristofori (Fondazione IRCCS Ca’ Granda Ospedale Maggiore, Milano), Angelica Dessì (AOU Cagliari), Andrea Dotta (Ospedale Bambin Gesù, Roma), Giancarlo Gargano (Arcispedale Santa Maria Nuova, Reggio Emilia), Paolo Ghirri (AOU Pisana, Pisa), Nicola Laforgia (AOU Policlinico, Bari), Stefania Liguori (Ospedale Maria Vittoria, Torino), Valeria Anna Manfredini (Ospedale di Rho, Milano), Luca Massenzi (Fatebenefratelli-Isola Tiberina, Roma), Anna Claudia Massolo (Ospedale Bambin Gesù, Roma), Federico Matina (Spedali Civili, Brescia), Fabio Natale (Azienda Policlinico Umberto I, Roma), Luisa Pieragostini (Ospedale San Filippo Neri, Roma), Matteo Rinaldi (Ospedali Riuniti di Foggia), Daniele Roncati (AOU Careggi, Firenze), Cristina Ruspaggiati (AOU Parma), Vincenzo Salvo (AOU Policlinico G. Martino, Messina), Elena Sorrentino (Fatebenefratelli-Ospedale San Pietro, Roma), Paolo Tagliabue (Ospedale San Gerardo, Monza).
Contributions of Author Statement
Rocco Agostino and Laura Cuzzolin defined the study and the online questionnaire, Laura Cuzzolin performed the analysis of the data and wrote the first draft of the manuscript, Rocco Agostino critically reviewed the manuscript.
ELBW―Extremely Low Birth Weight
NICU―Neonatal Intensive Care Unit
BNFC―British National Formulary for Children
NPSG―Neonatal Pharmacotherapy Study Group
ISN―Italian Society of Neonatology
 Marchant, E.A., Boyce, G.K., Sadarangani, M. and Lavoie, P.M. (2013) Neonatal Sepsis Due to Coagulase-Negative Staphylococci. Clinical and Developmental Immunology, 2013, Article ID: 586076.
 Tsai, M.H., Hsu, J.F., Chu, S.M., Lien, R., Huang, H.R., Chiang, M.C., et al. (2014) Incidence, Clinical Characteristics and Risk Factors for Adverse Outcome in Neonates with Late Onset Sepsis. The Pediatric Infectious Disease Journal, 33, e7-e13.
 Kuppala, V.S., Meinzen-Derr, J., Morrow, A.L. and Schibler, K.R. (2011) Prolonged Initial Empirical Antibiotic Treatment Is Associated with Adverse Outcomes in Premature Infants. Journal of Pediatrics, 159, 720-725.
 Russell, A.B., Sharland, M. and Heath, P.T. (2012) Improving Antibiotic Prescribing in Neonatal Units: Time to Act. ADC Fetal & Neonatal Edition, 97, F141-F146.
 Pandolfini, C., Kaguelidou, F., Sequi, M., Jacqz-Aigrain, E., Choonara, I., Turner, M.A., et al. (2013) Wide Intra- and Inter-Country Variability in Drug Use and Dosage in Very-Low-Birth-Weight Newborns with Severe Infections. European Journal of Clinical Pharmacology, 69, 1031-1036.
 Leroux, S., Zhao, W., Betremieux, P., Pladys, P., Saliba, E., Jacqz-Aigrain, E., The French Society of Neonatology (2015) Therapeutic Guidelines for Prescribing Antibiotics in Neonates Should Be Evidence-Based: A French National Survey. Archives of Disease in Childhood, 100, 394-398.
 Metsvaht, T., Nellis, G., Varendi, H., Nunn, A.J., Graham, S., Rieutord, A., et al. (2015) High Variability in the Dosing of Commonly Used Antibiotics Revealed by a Europe-Wide Point Prevalence Study: Implications for Research and Dissemination. BMC Pediatrics, 15, 41-51.
 Spyridis, N., Syridou, G., Goossens, H., Versporten, A., Kopsidas, J., Kourlaba, G., et al. (2016) ARPEC Project Group Members. Variation in Paediatric Hospital Antibiotic Guidelines in Europe. Archives of Disease in Childhood, 101, 72-76.
 de Jong, J., van den Berg, P.B., Visser, S.T., de Vries, T.W. and de Jong-van den Berg, L.T. (2009) Antibiotic Usage, Dosage and Course Length in Children between 0 and 4 Years. Acta Paediatrica, 98, 1142-1148.
 Porta, A., Esposito, S., Menson, E., Spyridis, N., Tsolia, M., Sharland, M., et al. (2010) Off-Label Antibiotic Use in Children in Three European Countries. European Journal of Clinical Pharmacology, 66, 919-927.
 Kadambari, S., Heath, P.T., Sharland, M., Lewis, S., Nichols, A. and Turner, M.A. (2011) Variation in Gentamicin and Vancomycin Dosage and Monitoring in UK Neonatal Units. Journal of Antimicrobial Chemotherapy, 66, 2647-2650.
 Cuzzolin, L. and Agostino, R. (2016) Off-Label and Unlicensed Drug Treatments in Neonatal Intensive Care Units: An Italian Multicentre Study. European Journal of Clinical Pharmacology, 72, 117-123.
 Cheng, C.L., Yang, Y.H., Lin, S.J., Lin, C.H. and Lin, Y.J. (2010) Compliance with Dosing Recommendations from Common References in Prescribing Antibiotics for Preterm Neonates. Pharmacoepidemiology and Drug Safety, 19, 51-58.
 Kaguelidou, F., Turner, M.A., Choonara, I., van den Anker, J., Manzoni, P., Alberti, C., et al. (2013) Randomized Controlled Trials of Antibiotics for Neonatal Infections: A Systematic Review. British Journal of Clinical Pharmacology, 76, 21-29.
 Olski, T.M., Lampus, S.F., Gherarducci, G. and Saint Raymond, A. (2011) Three Years of Paediatric Regulation in the European Union. European Journal of Clinical Pharmacology, 67, 245-252.
 Garazzino, S., Lutsar, I., Bertaina, C., Tovo, P.-A. and Sharland, M. (2013) New Antibiotics for Paediatric Use: A Review of a Decade of Regulatory Trials Submitted to the European Medicines Agency from 2000—Why Aren’t We Doing Better? International Journal of Antimicrobial Agents, 42, 99-118.
 Laughon, M.M., Avant, D., Tripathi, N., Hornik, C.P., Cohen-Wolkowiez, M., Clark, R.H., et al. (2014) Drug Labelling and Exposure in Neonates. JAMA Pediatrics, 168, 130-136.
 Knoderer, C.A., Everett, J.A. and Buss, W.F. (2003) Clinical Issues Surrounding Once-Daily Aminoglycoside Dosing in Children. Pharmacotherapy, 23, 44-56.
 Rao, S.C., Srinivasjois, R. and Moon, K. (2016) One Dose per Day Compared to Multiple Doses per Day of Gentamicin for Treatment of Suspected or Proven Sepsis in Neonates. Cochrane Database of Systematic Reviews, 12, CD005091.
 Jacqz-Aigrain, E., Leroux, S., Zhao, W., van den Anker, J.N. and Sharland, M. (2015) How to Use Vancomycin Optimally in Neonates: Remaining Questions. Expert Review of Clinical Pharmacology, 8, 635-648.
 de Hoog, M., Mouton, J.W. and van den Anker, J.N. (2005) New Dosing Strategies for Antibacterial Agents in the Neonate. Seminars in Fetal & Neonatal Medicine, 10, 185-194.