Neonatal adverse drug reactions: an analysis of reports to the French pharmacovigilance database

Abstract

Aim

Term and preterm neonates are at high risk for serious adverse drug reactions (ADRs).

Methods

A descriptive study of reports registered in the French pharmacovigilance database from 1986 to 2012 were obtained. All reports concerning neonates (≤1 month of life) with direct drug exposure were retrieved. Characteristics of the reports, including reported ADR(s), drug(s) and the causality assessment using the French causality assessment method, were described.

Results

A total of 1688 reports were analyzed and more than half of them were classified as serious (n = 995). Median age at ADR occurrence was 9 days. Overall, 3127 ADRs were described in these reports in relation to 2238 suspect/interacting drugs. The most commonly reported system organ classes (SOCs) were injury, poisoning and procedural complications (16%), general disorders and administration site conditions (12.5%) and blood and lymphatic system disorders (12%). In the majority of ADRs reported (73%), infants fully recovered and less than 4% of neonates deceased as a consequence of the reported ADR. One out of five ADRs was associated with drug administration errors. Therapeutic classes commonly incriminated were anti‐infectives, nervous system and alimentary tract drugs. Substances most frequently related to serious ADRs were zidovudine, ibuprofen and nevirapine. Among the 10 most frequently encountered drug−ADR pairs, two substances were mainly implicated, zidovudine in haematological adverse reactions and phytomenadione in maladministrations.

Conclusions

Anti‐infective drugs, mainly antiretroviral therapy, account for the majority of ADRs reported in neonates. The specific issue of drug maladministration and medication errors remains to be addressed in neonates.

What is Already Known about this Subject

• Previous pharmacovigilance studies reported that in children approximately one ADR notification out of four concerns infants under the age of 1 year.

• No study has focused on describing ADRs following direct drug exposure in neonates more in‐depth.

What this Study Adds

• The majority of neonatal ADR reports were classified as serious and 3.7% of neonates deceased as a consequence of an ADR.

• Anti‐infective medicines, mainly antiretroviral therapy, account for the large majority of reported ADRs and ADR fatalities.

• Vitamins such as phytomenadione also account for a large number of ADR reports mainly associated witho drug administration errors without clinical consequences.

Introduction

There are many reasons why neonates may face a higher risk for adverse drug reactions (ADRs) than other age groups. Their organ immaturity and the rapid developmental changes that occur after birth may impact on both drug pharmacokinetics and pharmacodynamics 1. Sick neonates often require treatment with a substantial number of drugs because of the complexity and seriousness of their conditions and most prescribed drugs have never been specifically evaluated in this neonatal population. Indeed, the percentage of patients receiving at least one unlicensed or ‘off‐label’ drug ranges from 80% to 100% in neonatology units 2, 3, 4.

Spontaneous reporting systems are an important source of information on drug safety and detection of safety signals. This is essential for products that have not undergone a marketing authorization process, the most commonly encountered situation in paediatric practice. Previous pharmacovigilance (PV) studies reported that in children approximately one ADR notification out of four concerns infants under the age of 1 year 5, 6, 7 but few studies specifically described reports in neonates 8, 9, 10. In these studies assessing ADRs after in utero and direct exposure, congenital disorders and reactions related to pregnancy were predominant. Currently, no study has focused on describing ADRs following direct drug exposure in neonates more in‐depth.

Consequently, the objective of this study was to describe the nature of all ADRs and drug/ADR pairs reported in neonates following direct drug exposure and included in the French pharmacovigilance database.

Methods

The French pharmacovigilance database

This is a descriptive study of reports registered in the French pharmacovigilance database (FPVDB). This national database was set up in 1985–1986 to allow online input of spontaneous ADR reports from a network of 31 regional PV centres. Reporting of serious and/or unexpected ADRs has been compulsory for physicians, dentists and midwifes since 1984 and for pharmacists since 1995. Also, nurses (since 1995) and patients, families and associations (since 2011) can contribute on a voluntary basis. PV reports include ADRs following drug use in usual conditions but also medication errors and overdoses.

Spontaneous ADR reports are sent to the regional PV centres mainly by means of a standardized form. Each report is validated and a drug causality assessment is performed by qualified personnel before anonymous storage in the national database.

Database search

For the present study, reports contained in the FPVDB were extracted from January 1 1986 to November 30 2012. To focus on the neonatal population, the database was searched for all notifications with a reported age at the occurrence of the ADR of equal to or less than 30 days, 1 month or 4 weeks. To avoid neonatal ADRs secondary to maternal drug intake, reports classified as related to ‘pregnancy’ or ‘breastfeeding’ were discarded.

Data collection

Data were recorded in four relational tables: ‘report’, ‘drugs’, ‘ADRs’ and ‘medical history’ tables. Thus, a single report can be linked to one or more drugs, ADRs and previous medical history. The ‘report’ table contains information about 1) the notification: date of initial reporting, date of last follow‐up, regional PV centre of registration, type of the reporter, type of the reporter's practice, global seriousness of the ADR report and criteria for seriousness 11 and type of the ADR and 2) the patient: date of birth, reported age at ADR occurrence, gender, date and cause of death when applicable. Patient information was completed by data from the ‘medical history’ table.

For each reported drug, information from the ‘drugs’ table was retrieved: active substance, brand name, dates of beginning of treatment and drug withdrawal, drug indication, frequency and route of administration, role of the drug (suspect, concomitant, interacting) and causality assessment according to the French causality assessment method 12. This is an algorithmic method that combines semiological and chronological criteria to give an ‘intrinsic score’ (0: excluded, 1: doubtful, 2: possible, 3: probable, 4: certain) and then adds bibliographic data from standardized sources to give an ‘extrinsic score’ (0: never described before, 1: not described, 2: not well known, 3: well known). The ‘ADRs’ table includes information on each reported ADR: date of occurrence, date of resolution if applicable, type of the ADR and outcome of the patient.

All medical history, causes of death, drug indications and type of ADRs are coded according to the Medical Dictionary for Regulatory Activities (MedDRA), version 15.1 (http://www.meddra.org/). This dictionary is built as a five‐level scale comprising 26 system organ classes (SOC) divided into high level group terms (HLGT), high level terms (HLT), preferred terms (PT) and finally lowest level terms (LLT). The term often used to describe an ADR in spontaneous reporting systems is the PT because it corresponds to a single medical concept. Anatomical therapeutic chemical (ATC) codes for drugs were computed as the FPVDB does not include this information. Recoding to the ATC coding system was possible for 97.6%, not applicable for 0.5% and not possible for 2.1% of reported drugs. Throughout the paper, the word ‘drug’ will denote the ATC substance level or else as mentioned. Medical history of neonates was classified based on PTs to three major groups: prematurity, infectious disease and genetic and/or congenital abnormality. We excluded from analysis 26 reports (1.5%) where time between occurrence of the reported ADR and date of reporting to the FPVDB had a negative value or a value of more than 10 years. Time between the beginning of drug administration and occurrence of the reported ADR was assessed only for 767 drugs as from the remaining information on dates was missing.

Data analysis

A primary descriptive analysis was performed on the entire dataset between 1986 and 2012. However, during the calendar year 2001, an over‐reporting of ADRs in neonates was observed mainly due to the retrospective reporting of ADRs in relation to antiretroviral therapy. Therefore, a secondary descriptive analysis was performed excluding ADRs reports from that specific year. Results of the primary and secondary analysis were compared.

Descriptive data include absolute numbers (percentages) for categorical variables and medians [1^st quartile (Q1)3^rd quartile (Q3)] for continuous variables or else as mentioned. To apprehend the impact of impaired medical condition on adverse drug outcomes, differences in mortality between neonates with and those without medical history were tested using the chi‐square test. For the description of drug‐ADR pairs, we only retained suspect and interacting drugs. Analysis was performed with SAS version 9.2 software (SAS Inc, Cary, North Carolina, USA.).

Results

Characteristics of ADR reports

From 1986 to 2012, a total of 2019 notifications were retrieved from the FPVDB. After discarding cases of maternal exposure to drugs during pregnancy and duplicate reports, notifications for 1688 neonates were analyzed. More than half of these reports (995, 59%) were rated as serious. Trends over time in the number and characteristics of notified ADRs are reported in Figure 1 and Table 1, respectively. Taking into account the number of neonates in France (http://www.insee.fr), 778.468 and 792.996 in 1986 and 2011 respectively, the notification rate increased from 0.03 to 0.17 per 1000 neonates. Median age at occurrence of ADR was 9 days and a slight predominance of male neonates was noted (male : female ratio: 1.3). A medical history was specified for 240 (14%) neonates. Among those, 111 presented a preterm birth (46%), 26 an infection (11%) and 50 a genetic and/or congenital abnormality (21%). The median number of reports per regional PV centre was 36 (20.5–48.5). However, two centres accounted for 37% of all reports (320 and 297, respectively) assessed over the study period, whereas the total number of reports for the remaining centres ranged between 5 and 102.

Figure 1.

Trends over time in the number of ADR reports in neonates. number of reports with non‐serious ADRs, number of reports with serious ADRs

Table 1.

Characteristics of reports of adverse drug reactions (ADRs) in neonates

Characteristics of reports (n = 1688)	Median (Q1 ‐ Q3) or number (%)
	n	%
Gender
Male	922	55
Female	735	44
Note Reported	32	2
Age (days)	9 (2‐21)
Serious ADR
Yes	995	59
Death	88	5
Life‐threatening	166	10
Hospitalization (initial or prolonged)	632	37
Disability	10	1
Other clinically relevant conditions	99	6
No	693	41
Type of reported ADR
Adverse drug reaction	1331	79
Medical errors‐overdoses	347	20.4
Drug interaction	10	0.6
Type or reporter
General practitioner	84	5
Medical specialist	1302	77
Pharmacist	137	8
Nurse	29	1.7
Dentist	2	0.1
Other health professional	37	2.2
Non‐health professional	56	3.3
Unknown	41	2.4
Type of practice
University hospital	964	57
Other hospital	307	18
Private clinic	8	0.5
Private practice	286	17
Other	31	2
Unknown	92	5.5

Overall, 88 neonates deceased at a median age of 14 days 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27 of life, after median of 1 day (0–9 days) from first reported ADR occurrence and with a predominance of male neonates (ratio 1.5). Approximately 72% (n = 63) of these deaths were related to a reported ADR (not ADR related deaths = 17, missing information = 8). The majority of fatal ADRs were reported by medical specialists (n = 76, 92%). Fatal outcomes were more frequent in neonates with medical history (20/240, 8.3%) than in neonates without (68/1455, 4.7%, P value = 0.02).

Type of ADRs reported

A total of 3127 ADRs were reported (median of one ADR per report [1–3)]) which corresponded to 479 distinct PT codes. Median time between date of reporting and date of occurrence of the ADR was 21 days (6–103). The most commonly reported system organ class was injury, poisoning and procedural complications (n = 492, 16%), followed by general disorders and administration site conditions (n = 391, 12.5%), blood and lymphatic system disorders (n = 373, 12%), gastrointestinal disorders (n = 255, 8.2%), investigations (n = 251, 8%) and nervous system disorders (n = 245, 8%). The other system organ classes were involved less commonly (<8%) (Figure 2). Distribution of fatal and non‐fatal ADRs according to the different system organ classes is shown in Figure 2 and the 10 most commonly reported ADRs are given in Table 2. Regarding outcomes, neonates fully recovered without sequelae in 73%, recovered with sequelae in 6%, were recovering in 2%, deceased in 4% (ADR‐related 97; not‐ADR related 24; undetermined origin 12) and outcome was unknown/not yet established in 15% of ADRs.

Figure 2.

Types of reported ADRs in neonates according to MedDRA system organ class (SOC). no serious ADRs, serious, non fatal ADRs, serious, fatal ADRs

Table 2.

Ten most frequently reported ADRs according to the preferred terms (PTs) of MedDRA coding system

All ADRs reported†† (n = 3127)		ADRs with serious outcome (n = 1969)		ADRs with fatal outcome (n = 133)
	Number		Number		Number
Medication error	148	Neutropenia	120	Death†	11
Neutropenia	128	Anaemia	86	Renal failure acute	8
Anaemia	91	Blood lactic acid increased	66	Sepsis neonatal	6
No adverse event *	82	Bradycardia	46	Intestinal perforation	5
Blood lactic acid increased	66	Anaemia macrocytic	43	Bradycardia	4
Bradycardia	62	Malaise	39	Enterocolitis	4
Incorrect drug administration rate	56	Hypertriglyceridaemia	33	Blood creatinine increased	3
Accidental overdose	51	Medication error	32	Cardiac failure	3
Malaise	47	Accidental overdose	31	Hyperkalaemia	3
Wrong drug administered	47	Hypotonia	31	Hypotension	3
Incorrect dose administered	47

^* These ADRs correspond to errors of drug administration without clinical consequences

^† Reaction was coded as such with no further details

^†† In this column 11 ADRs are presented because the last threeADRs were reported in an equal number of reports.

Of note, 14% (437/3127) of the reported ADRs were coded as maladministrations (n = 189, incorrect drug administration rate, wrong drug administered, incorrect drug dosage form administered, etc.) or medication errors (n = 182; drug dispensing errors, drug prescribing errors, etc.) or overdoses (n = 66). Thirty percent of these ADRs (n = 135/437) were serious and they resulted in three neonatal deaths. Also, 3% (82/3127) of the reported ADRs were coded as ‘no adverse event’ or ‘adverse effect absent’ and referred to errors of drug administration without clinical consequences.

Medicinal drugs reported

A total of 2797 drugs were reported (median of one drug per report 1, 2) which corresponded to 446 distinct ATC drug codes (5^th level codes, n = 417; 3th or 4^th level codes, n = 29; not applicable/missing, n = 3). Median time between the beginning of drug administration and the occurrence of the reported ADR was 7 days (1–21 days). A total of 2238 (80%) drugs was classified as suspect or interacting drugs and 559 (20%) as concomitant. Reported drugs by anatomical main group and role (suspect/interacting or concomitant) in the occurrence of adverse reactions are given in Figure 3. There was no particular trend over time with respect to the main therapeutic classes or the specific drugs involved (data not shown) except for anti‐infectives, mainly antiretroviral molecules, that accounted for the majority of drugs reported in the FPVDB in 2001 (58% of all ATC codes reported that year).

Figure 3.

Reported drugs by anatomical main group and role (suspect/interacting or concomitant) in the occurrence of adverse reactions (A) alimentary tract and metabolism, (B) blood and blood forming organs, (C) cardiovascular system, (D) dermatologicals, (G) genito‐urinary system and sex hormones, (H) systemic hormonal preparations, excluding sex hormones and insulins, (J) anti‐infectives for systemic use, (L)antineoplastic and immunomodulating agents, (M) musculo‐skeletal system, (N) nervous system, (P) antiparasitic products, insecticides and repellents, (R) respiratory system, (S) sensory organs, (V) various, (U) unknown (ATC code could not be attributed because the commercial name or active substance of the drug were missing) and (NA) not applicable (no ATC codes attributed to these substances). suspect/interacting drugs, concomittant drugs

For 90% of the suspect (n = 2215) and interacting (n = 23) drugs the route of administration was reported: 7% topical, 43% enteral, 39% parenteral and <1% other routes. The most commonly reported drugs (>1% of suspect or interacting drugs) are given in Table 3.

Table 3.

Most frequently reported suspect or interacting drugs (n = 2238)

Drug	Number	% of suspect or interacting drugs (n = 2238)	ATC code
Zidovudine	217	9.7	J05AF01
Phytomenadione	106	4.7	B02BA01
Ibuprofen	73	3.2	C01EB16
Chlorhexidine	41	1.8	D08AC02
Vancomycine	38	1.7	J01XA01
Caffeine	36	1.6	N06 BC01
Ergocalciferol	35	1.6	A11CC01
Cefotaxime	35	1.6	J01DD01
Nevirapine	33	1.5	J05AG01
Combinations of vitamins	31	1.4	A11JA
Lamivudine	31	1.4	J05AF05
BCG vaccine	31	1.4	L03AX03
Amikacin	29	1.3	J01GB06
Amoxicillin	28	1.2	J01CA04
Stavudine	27	1.2	J05AF04
Paracetamol	27	1.2	N02BE01
Midazolam	27	1.2	N05CD08
Combinations of solutions for parenteral nutrition	25	1.1	B05BA10
Morphine	25	1.1	N02AA01
Fentanyl	23	1.0	N01AH01

All other drugs accounted for less than 1% of the reported suspect or interacting drugs.

ATC recoding not possible: n = 51

Use of a total of 1423 suspect/interacting drugs were mentioned in reports classified as serious (n = 996) and the most frequently reported drugs were zidovudine (n = 195, 14%), ibuprofen (n = 65, 5%) and nevirapine (n = 33, 2%). Use of a total of 166 suspect/interacting drugs were mentioned in reports with a fatal outcome (n = 88). Again, three drugs were mainly reported: ibuprofen (n = 19, 11%), zidovudine (n = 7, 4%) and electrolyte solutions (n = 5, 3%).

Analysis of drug–ADR pairs

Overall, 4106 drug–ADR pairs were reported in neonates from 1986 to 2012. The 10 most frequent drug‐ADR pairs are given in Table 4. Two drugs are implicated in eight out of these 10 pairs, zidovudine and phytomenadione (vitamin K1).

Table 4.

Most frequently reported drugs and MedDRA high‐level terms (HLT) pairs (suspect and interacting drugs only)

Adverse drug reaction	Drug	ATC	Number of reports	% of the reports (n = 1688)*
Anaemias	Zidovudine	J05AF01	113	6.7
Neutropenias	Zidovudine	J05AF01	89	5.3
Medication errors	Phytomenadione	B02BA01	84	5.0
Maladministration	Phytomenadione	B02BA01	71	4.2
Blood gas and acid base analyses	Zidovudine	J05AF01	51	3.0
Adverse effect absent	Phytomenadione	B02BA01	39	2.3
Neutropenias	Nevirapine	J05AG01	29	1.7
Renal failure and impairment	Ibuprofen	C01EB16	24	1.4
Thrombocytoses	Zidovudine	J05AF01	24	1.4
Liver function analyses	Zidovudine	J05AF01	23	1.4

^* Because individual reports may contain more than one active substance – HLT pair, the sum of all percents would be >100%

Intrinsic causality was ‘doubtful’ for 1437 (35%) of drug–ADR pairs, ‘possible’ for 1327 (32%), ‘probable’ for 845 (21%), ‘certain’ for 297 (7%) and not attributed for 200 drug–ADR pairs (5%). Drug–ADR pairs that were given a score of ‘certain’ were mainly phytomenadione (vitamin K1) with maladministration or medication error or no adverse effect (n = 118) and fluconazole with maladministration (n = 8). The drug–ADR pair was ‘never described before’ for 90 (2%), ‘not described’ for 459 (11%), ‘not well known’ for 437 (11%), ‘well known’ for 2103 (51%) and not attributed for 1017 (25%). The substance most frequently implicated in pairs ‘never described before’ was chlorhexidine for ADRs such as erythema, inflammation or non‐therapeutic responses (n = 10).

The three most frequently reported substances by MedDRA SOCs are given in Table 5.

Table 5.

The three most frequently reported drugs per type of ADRs according to MedDRA system organ class (SOCs) (suspect and interacting drugs only)

SOCs	Total n	Drug	n	Drug	n	Drug	n
Blood and lymphatic system disorders	587	Zidovudine	239	Nevirapine	62	Lamivudine	48
Injury, poisoning and procedural complications	527	Phytomenadione	173	Chlorhexidine	31	Zidovudine	22
General disorders and administration site conditions	478	Phytomenadione	50	Chlorhexidine	24	Solutions for parenteral nutrition	17
Investigations	346	Zidovudine	107	Stavudine	24	Nevirapine	24
Nervous system disorders	326	Zidovudine	19	Midazolam	15	Ergocalciferol	11
Gastrointestinal disorders	321	Ibuprofen	55	Zidovudine	54	Ergocalciferol	20
Cardiac disorders	279	Diphemanil methylsulfate	17	Caffeine	17	Combinations of vitamins	15
Respiratory, thoracic and mediastinal disorders	268	Ergocalciferol	22	Carbocisteine	13	Combinations of vitamins	11
Metabolism and nutrition disorders	224	Zidovudine	33	Stavudine	16	Nevirapine	16
Skin and subcutaneous tissue disorders	190	Chlorhexidine	9	Amoxicillin	9	Cefotaxime	9
Renal and urinary disorders	136	Ibuprofen	24	Amikacin	9	Fentanyl	9
Vascular disorders	96	Midazolam	6	Doxapram	6	Labetalol	5
Musculoskeletal and connective tissue disorders	72	Sufentanil	5	Potassium clorazepate	5	Diazepam	3
Eye disorder	56	Oxytetracycline	12	Zidovudine		Chlorhexidine	5
Hepatobiliary disorders	54	Methylthioninium chloride	4	Ceftriaxone	3	Paracetamol	3
Infections and infestations	52	BCG vaccine	16	i.v. solution additives	6	Amino acids IV solutions	3
Psychiatric disorders	51	Caffeine	5	Zidovudine	4	Phytomenadione	3

Remaining SOCs account for less than 50/4106 drug‐ADR pairs

In ADRs coded as maladministration, medication errors or overdose (n = 464 pairs), the most frequently reported drugs were phytomenadione (n = 164, 35%), chlorhexidine (n = 31, 7%), paracetamol (n = 15, 3%), ergocalciferol (n = 14, 3%) and gentamycin (n = 13, 3%). Three neonatal deaths in this category of ADRs were related to digoxin (n = 2) and lopinavir/ritonavir (n = 1) overdoses.

Secondary analysis

The results of the analysis excluding 158 reports (year 2001) were overall comparable with those of the full dataset with regard to the characteristics of the reports, types and frequency of ADRs, therapeutic drug classes and frequency of suspect/interacting drugs reported (data not shown). However, some differences were noted in the nature of the drug–ADR pairs reported. Notably, the association zidovudine–thrombocytoses, zidovudine–blood gas and acid base analyses and nevirapine–neutropenias were less frequent compared with the full dataset analysis. On the contrary, the following associations: ibuprofen intestinal ulcers and perforation, chlorhexidine maladministration, chlorhexidine medical errors, BCG vaccine infections and zidovudine colitis became more frequent (details are given in Appendix).

Discussion

We provide a unique overview of PV reports after direct drug exposure in neonates using national data over a period of 26 years. The number of neonatal reports is steadily increasing parallel to the global increase in PV reporting observed over the years worldwide 9, 10, 13. Although few PV studies have been published in children, all show that a substantial number of ADRs occurs in the first years of life 5, 6, 7, 8, 14, 15, 16, 17, 18. Yet, reporting rates in our study were relatively low compared with previously published data with estimations as high as 0.5% in children 6, 19. Possible explanations comprise either a marginal drug exposure or a more frequent under‐reporting in neonates compared witho older age groups. Most reports concerned male neonates and occurred in the first days of life. Births and hospitalizations in the neonatal intensive care unit (NICU) of male neonates generally exceed those of female ones 20, 21. In our study, fatal outcomes were also more frequent in males, further pointing towards an effect of the gender on the susceptibility to sickness and by extension to ADRs early in life 9, 10.

As sick neonates are mainly cared for in specialized units, the majority of reports were issued by medical specialists from university hospitals. However, the paucity of ADRs reports provided by nurses is striking. Nurses are key caregivers in NICUs and thus their role in neonatal ADR identification and reporting should be enhanced. Conversely, non‐health professionals were very recently endorsed to report in France which explains the paucity of reports provided by them.

Reports were addressed mainly to two out of the 31 regional PV centres. Although clinicians are expected to report to the PV centre of their region, some may forward reports to centres that have a specific expertise in paediatric PV and thereby seek expert guidance. Another explanation may be the exhaustive ADR reporting in the context of a research study. In fact, a substantial number of reports in 2001 were associated with the retrospective assessment of ADRs related to antiretroviral therapy in a single PV centre. This centre has a recognized expertise in antiretroviral drug safety in children 22 and actively participates in a national prospective epidemiological study on the risk and prevention of mother‐to‐child HIV transmission 23.

Neonatal ADRs were mainly classified as serious (59%) compared with previous PV studies with any paediatric age (13%–37%) 6, 8, 19, 24. However, the seriousness was essentially related to hospitalizations (admission or prolonged stay) rather than the occurrence of disability or fatalities. Despite the physiological vulnerability of neonates, fatalities were less frequent in our study (3.7%) compared with data reported in older children (5.6%–7%)8, 15. Even though reassuring, this finding may actually reflect the difficulty of neonatologists in diagnosing serious ADRs in the context of multiple comorbidities such as extreme prematurity. In line with previous studies on medication use in NICUs 20, the high number of drugs reported should also be highlighted especially since many are not licensed for use in neonates 25.

Yet, the importance of our description lies in the fact that it enables comparison with the reaction–drug patterns observed in other age groups to identify safety issues specific to neonates. First, we observed a clear predominance of ADRs related to the use of anti‐infectives. Antiretrovirals have already been incriminated in the occurrence of serious and fatal ADRs in neonates mainly through in utero exposure 9, 10. Our study confirms that direct exposure in early life entails the same risks namely that of zidovudine, which is the recommended first line antiretroviral to prevent mother‐to‐child HIV transmission in France 26. Antibiotics were also frequently reported in various neonatal ADRs. Although, anti‐infectives are prescribed in NICUs to treat or prevent severe, life‐threatening conditions, their toxicity profile should urge neonatologists to evaluate systematically the risk–benefit ratio of their use or alternatively, to define adequate conditions of exposure and monitoring.

Second, one out of five ADRs was reported in terms of incorrect drug administration under the ‘injury, poisoning and procedural complications’ SOC. Drug formulations are often not adapted to the youngest,.Subsequently administration and dosage errors are frequent in NICUs 27, 28, 29. On the one hand, these errors occurred in the hospital setting and led to serious consequences as in the case of electrolyte solutions 30. On the other hand, a significant number of errors occurred in the outpatient environment related to the administration of vitamins or antiseptic preparations such as chlorhexidine. In France, prescription of oral phytomenadione is common in exclusively breastfed babies and errors are frequent as some parents tend to give it once daily instead of once weekly 31. These ADRs led to mild adverse outcomes or had no clinical consequences. However, they draw attention to the difficulties that parents might face when administering a drug to their newborn and, therefore, the importance of appropriate education and training. In France, two ‘dear doctor’ letters were issued to limit administration errors for phytomenadione in 2008 31 and for ergocalciferol in 2011 32.

Third, our data show that intake of vitamins or solutions of parenteral nutrition such as amino acids, was not devoid of risk. Their use was related to a variety of ADRs, including infections and cardiac disorders and this should warrant their cautious use and administration in neonates especially in preterms. Likewise, use of nervous system drugs such as caffeine, midazolam or morphine was frequently incriminated. These drugs are part of the essential bulk of NICU medications although there is limited evidence on their adequate use which probably exposes neonates to substantial safety risks.

Finally, toxicity of ibuprofen, in particular renal and gastrointestinal, in neonates has been clearly illustrated. The number, type and severity of ibuprofen‐related ADRs described in our study are not predominant in older children 9. The main clinical indication of ibuprofen in neonates is the closure of patent ductus arteriosus and, currently, there is no validated therapeutic alternative 33. In this context, the only possible improvement in terms of safety would be to optimize ibuprofen dosing schemes through pharmacokinetic/pharmacodynamic modelling but very few studies have undertaken this task 34.

Our results should be viewed in light of the known limitations of spontaneous reporting databases 35. Important information such as the term of birth or medical history may not have been exhaustively reported because this information often lies in free‐text fields and is not captured when analyzing standard fields of generic PV forms. Several data were often missing such as date of treatment initiation and/or discontinuation or drug dose and frequency of administration, which may be related to incomplete reporting but also errors during data capture and transfer. Yet, compared with previous paediatric PV studies 5, rates of missing data were very low in the FPVDB, only 2.1% for drugs and 0.01% for ADRs.

Notwithstanding these limitations, spontaneous reporting databases are currently the most valuable source of drug safety information in neonates 30, 36, 37. Development of a paediatric PV expertise and interactions with specialised pharmacologists in PV centres should be further encouraged in order to address the need to increase physicians' and nurses' awareness and to train them to diagnose and report better ADRs in neonates. To enhance further collection of informative reporting, another proposition would be the elaboration of a specific neonatal PV report comprising fields of information which are currently absent in the generic PV form but are essential to the drug–event causality assessment with respect to neonatal specificities.

In conclusion, this is a unique overview of neonatal ADRs after direct drug exposure. Reporting of ADRs in neonates is increasing. The majority of reports are classified as serious and a subsequent proportion is associated with drug administration errors. Anti‐infective medicines, mainly antiretroviral therapy, account for the majority of reported ADRs and ADR fatalities. Although based on national data, our results inform clinicians and stakeholders about the safety of drug use in neonates and the challenges of neonatal pharmacovigilance.

Competing Interests

All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf and declare no support from any organization for the submitted work, no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years and no other relationships or activities that could appear to have influenced the submitted work.

The French Medicines Agency (ANSM) provided the data and approved the conduct of the study but was not involved in data analysis and elaboration of this publication. The sole responsibility for the content of this publication lies with the authors.

This analysis of the French pharmacovigilance database was made possible thanks to the work of the 31 regional pharmacovigilance centres and the French Medicines Agency (ANSM). The authors acknowledge the support and assistance of Professor Elisabeth Autret‐Leca, Department of Clinical Pharmacology Regional Centre of Pharmacovigilance of Tours and Mrs Geneviève Brizion, Centre de Documentation Médico‐Pharmaceutique, AGEPS.

Funding

No sources of funding were used to assist in the preparation of this study and the drafting of the article.

Contributors

FK designed the study, conducted data analysis and interpretation of results. FB‐S, APJ‐B and EJ‐A contributed to data analysis and interpretation. The manuscript was initially drafted by FK and was critically reviewed and subsequently approved by each co‐author in its final form.

Most frequently reported drugs and MedDRA high‐level terms (HLT) pairs (suspect and interacting drugs only) after exclusion of calendar year 2001

Adverse drug reaction	Drug	ATC	Number of reports	% of the reports (n = 1688)*
Medication errors NEC	Phytomenadione	B02BA01	84	5.0
Maladministration	Phytomenadione	B02BA01	71	4.2
Anaemias NEC	Zidovudine	J05AF01	60	3.6
Adverse effect absent	Phytomenadione	B02BA01	39	2.3
Neutropenias	Zidovudine	J05AF01	31	1.8
Renal failure and impairment	Ibuprofen	C01EB16	24	1.4
Intestinal ulcers and perforation NEC	Ibuprofen	C01EB16	20	1.2
Maladministration	Chlorhexidine	D08AC02	17	1.0
Infections NEC	BCG vaccine	L03AX03	16	0.9
Colitis (excluding infective)	Zidovudine	J05AF01	15	0.9

^* Because individual reports may contain more than one active substance – HLT pair, the sum of all percents would be >100%

NEC not elsewhere classified.

Kaguelidou, F. , Beau‐Salinas, F. , Jonville‐Bera, A. P. , and Jacqz‐Aigrain, E. (2016) Neonatal adverse drug reactions: an analysis of reports to the French pharmacovigilance database. Br J Clin Pharmacol, 82: 1058–1068. doi: 10.1111/bcp.13034.