Review of Drug Utilization Studies in Neonatal Units: A Global Perspective

Rational prescribing is challenging in neonatology. Drug utilization studies help identify and define the problem. We performed a review of the literature on drug use in neonatal units and describe global variations. We searched databases (EMBASE, CINAHL and Medline) from inception to July 2020, screened studies and extracted relevant data (two reviewers). The search revealed 573 studies of which 84 were included. India (n = 14) and the USA (n = 13) reported the most. Data collection was prospective (n = 56) and retrospective (n = 26), mostly (n = 52) from one center only. Sixty studies described general drug use in 34 to 450,386 infants (median (IQR) 190 (91–767)) over a median (IQR) of 6 (3–18) months. Of the participants, 20–87% were preterm. The mean number of drugs per infant (range 11.1 to 1.7, pooled mean (SD) 4 (2.4)) was high with some reporting very high burden (≥30 drugs per infant in 8 studies). This was not associated with the proportion of preterm infants included. Antibiotics were the most frequently used drug. Drug use patterns were generally uniform with some variation in antibiotic use and more use of phenobarbitone in Asia. This study provides a global perspective on drug utilization in neonates and highlights the need for better quality information to assess rational prescribing.


Introduction
Prescribing drugs to newborn infants, particularly those born preterm, is a challenge fraught with complexities including lack of evidence-based information about pharmacokinetics and pharmacodynamics of drugs, efficacy and side-effect profiles for some of the most frequently used drugs. Despite this, infants in neonatal care are exposed to many drugs, often off-label, unlicensed and without clear guidance on dosing. The large gaps in knowledge translate into large differences in interpretation of the sparse evidence that is available, leading to wide variations in practice on one hand and the perpetuation of incorrect practices on the other.
Drug-utilization research provides an insight into the pattern of prescribing and is the essential first step towards rational drug use and evidence-based pharmacotherapy [1]. Physicians prescribe drugs not necessarily based on the available evidence but also under influence from psychosocial and circumstantial aspects that impact their decisions [2]. Investigation into the trends and variability of drug use in the neonatal population can provide information that could guide effective strategies to improve prescribing practices and highlight areas for research. Observational studies describing patterns of drug use provide preliminary evidence to support this agenda. Although evidence for Sixty studies, conducted between 1983 and 2020, reported drug use in all therapeutic categories. Most (43 of 60) collected data prospectively while 17 retrieved retrospective data. The studies were conducted in 26 countries (Figure 2) with India and the United States of America (USA) accounting for the largest number of reports, 14 and 13 respectively. There was one study that involved several European countries (21 participated) [65] and one study conducted in Germany and Brazil [66].

Characteristics of Included Studies
Most of the included studies (60/84) evaluated drugs in all drug groups or categories. These 84 included 8 studies that also reported separate analyses of antibiotic use and 20 studies that reported use of off-label medications. In addition, 11 studies reported antibiotic usage only, 6 reported off-label or unlicensed drug use and 7 reported pharmacological groups that were frequently used rather than listing individual drugs. The studies were all observational with 56 prospective and 26 retrospective data collection over a varied time period. Two studies collected both retrospective and prospective data [5,6]. Studies were largely based in a single center (52/84)  . Thirty-two studies were based in more than one neonatal unit, ranging from 2 centers (7 studies) [57][58][59][60][61][62][63] to 341 centers (one study) [64].
Sixty studies, conducted between 1983 and 2020, reported drug use in all therapeutic categories. Most (43 of 60) collected data prospectively while 17 retrieved retrospective data. The studies were conducted in 26 countries (Figure 2) with India and the United States of America (USA) accounting for the largest number of reports, 14 and 13 respectively. There was one study that involved several European countries (21 participated) [65] and one study conducted in Germany and Brazil [66]. drugs received by infants born at 22-24 weeks gestational age only.
Participants were infants admitted to neonatal units who received at least one drug during their stay. Several studies excluded certain drugs and infusions such as vitamin K, intravenous fluids, parenteral nutrition and fluids used to maintain patency of vascular access. The details of inclusion and exclusion for each included study is given in the tables in Appendix B.
Thirty-four of 60 studies reported the proportion of preterm infants (born at <37 weeks gestational age) among their cohort (range 20% [24] to 87% [71]) in addition to the two studies (34), (31) that included preterm infants only. In addition, one study Puia-Dumitrescu 2020 [72] reported drugs received by infants born at 22-24 weeks gestational age only.
Participants were infants admitted to neonatal units who received at least one drug during their stay. Several studies excluded certain drugs and infusions such as vitamin K, intravenous fluids, parenteral nutrition and fluids used to maintain patency of vascular access. The details of inclusion and exclusion for each included study is given in the tables in Appendix B.
Among the studies that used the WHO-ATC system, anti-infectives for systemic use were the most frequently prescribed pharmacological group in the majority (14 studies) [13,[23][24][25]30,34,42,43,50,51,59,71,73,76]. This was followed by agents for the alimentary tract and metabolism (4 studies) [7,63,65,74] and agents for the central nervous system (1 study) (32). Among the four studies that listed the pharmacological groups according to their pharmacological class, three studies reported that antimicrobials were the most frequently prescribed group [19,39,47] and one study by Ashwin et al. (2018) identified that penicillins were the most frequently prescribed [10]. Kumar et al. (2008) reported that the gastrointestinal agents were the most frequently prescribed pharmacological group [38].

Most Frequently Prescribed Drugs
Forty-eight studies reported the most frequently prescribed drugs. Figure 5 shows the drugs and the number of studies that reported it among its list of most frequently prescribed drugs and Table 1 gives a summary of the data by geographic region.

drugs.
Most studies did not report the indications of use, dose, frequency or duration of use or adverse effects of the frequently used drugs.
Most studies did not report the indications of use, dose, frequency or duration of use or adverse effects of the frequently used drugs.

Most Frequently Prescribed Antibiotics
Seven studies solely reported the most frequently prescribed antibiotics. In addition, several antibiotics appeared in the list of the most frequently prescribed drugs in studies that did not focus only on antibiotics. In total, 59 studies reported the most frequently used antibiotics. Figure 6 shows the antibiotics and the number of studies that reported it among its most frequently prescribed antibiotic/drug by geographical region. In addition to the data in Figure 6, two studies from Israel [11,45] reported gentamicin, ampicillin and amoxicillin as the most frequently prescribed antibiotics, and one of these [45] also included meropenem among the most frequently prescribed. The two Australasian studies [46,71] included gentamicin, vancomycin, ampicillin and benzylpenicillin in both lists. One African study [6] reported gentamicin, amoxicillin and ceftriaxone as the top three most frequently prescribed antibiotics. The single study from China [55] reported use of cefoperazone-sulbactum, and piperacillin-tazobactum as the most frequently used for all gestational age groups.

Discussion
This review presents a comprehensive global perspective of neonatal drug utilization research. Over 15 million infants are "born too soon" every year and provision of essential newborn care is imperative for meeting the United Nations' target to reduce neonatal mortality rates, a key component of the Sustainable Development Goals. Pharmacotherapy plays a large role in neonatal

Discussion
This review presents a comprehensive global perspective of neonatal drug utilization research. Over 15 million infants are "born too soon" every year and provision of essential newborn care is imperative for meeting the United Nations' target to reduce neonatal mortality rates, a key component of the Sustainable Development Goals. Pharmacotherapy plays a large role in neonatal care, particularly intensive care. This role is complicated by several factors including the developmental immaturity of newborn infants, paucity of evidence-base for efficacy, dosing and adverse effects information and the lack of licensed formulations. It is therefore not unsurprising that there is an explosion of interest in this area as reported by Allegaert et al., who found an increasing number of studies investigating drug utilization in newborns [3]. We found drug utilization studies from most parts of the world. Some regions are however sparsely represented-we found only one study from China and one from Africa, both published in the year 2020. India, which has the largest number of preterm births, contributed the largest number of studies, closely followed by the USA. The heightened interest in this area in India is interesting in view of the WHO-led concern that the WHO South-East Asia Region, which includes India, is likely the most at-risk part of the world for the emergence of resistance to microorganisms [80]. Although an increasing number of studies from Europe, and from South America and Australasia, also add to the volume of publications suggesting a world-wide interest, there remains a distinct lack of any collaborative international effort to explore the problem.
Methodologically, the studies remain limited to assessing the most common prescribed drugs either in general, or those that are off-label or unlicensed. Details required to assess the rational use of medications such as indication, dose or duration of use are lacking. Most studies were restricted to single centers and included a limited sample size. Larger studies such as those from the Peadiatrix Medical group in the USA [69,70] are powered by electronic patient records. It is plausible that the use of electronic patient records may enable further large-scale evaluations of drug utilization. This requires efforts to improve electronic patient records such as use of standardized nomenclature and categorization of drugs, collection of data on indications, dosage, adverse effects and medication errors which empower unraveling the yarn of rational prescribing (or the lack of rational prescribing) in neonatal medicine.
The populations included in the studies within this review are quite heterogeneous. Most studies include all neonatal unit admissions with a varied proportion of premature infants. However, it is likely that the composition of the premature cohort is not uniform as studies from high-income countries are likely to include a much more immature population compared to the preterm cohorts in the more resource-limited settings. We found wide variation in the number of medications used per infant ranging from 1.7 drugs per infants reported by Bonati et al. [75] to 11.1 per patient as reported by Neubert et al. [43]. However, we did not see a relationship between the proportion of premature infants included in the study with the average number of drugs prescribed per patient. This is likely to be because of the heterogeneity in the populations and variations in which drugs were excluded from the study. The burden of medication exposure in newborn infants was also well demonstrated by the maximum number of drugs per patient reported in some studies-62 in the most extreme example [38] with several others reporting use of more than 30 drugs in some infants.
We found that the drug utilization pattern is similar across most regions and nations, with a predominance of antibiotics use in all reports. Few studies reported drugs other than an antibiotic as the one in most common usage e.g., caffeine featured at the top of the list in 2 studies. This could be because of the high proportion of premature infants in the study, however we could only confirm this is one study [71] where 87% of included infants were born preterm. Variations in which drugs were excluded from analysis in each study accounts for some other drugs which were not antibiotics appearing as the most frequently prescribed, such as parenteral nutrition, vitamin K and multivitamins which, due to their ubiquitous use, were excluded from most studies. We saw some regional variations: in studies from Asia, specifically India, phenobarbitone was frequently reported. This may reflect the high prevalence of birth asphyxia which, along with prematurity and infections, is one of the three causes reported to account for 0.79 million of 1.01 million neonatal deaths in India in the Million Death Study [81].
The results of this review clearly demonstrate that antibiotics remain the most frequently used drug in neonatal medicine. This is not unexpected as the burden of infections remains high; neonatal sepsis or meningitis accounted for 16% neonatal deaths globally in 2015 [82]. High risk of death and poor outcomes in survivors warrants the reliance on empirical antibiotic usage based on the sensitive but nonspecific clinical diagnosis of possible infections, particularly in preterm infants, and the antibiotics given to clinically well infants born with risk-factors for early-onset sepsis. Unfortunately, the selective pressure exerted by this widespread use is driving antimicrobial antibiotic resistance. Although this is a global problem it is unequally spread, with data from high-income countries such as the UK showing that 95% of pathogens were susceptible to the most commonly used empirical antibiotic regimens, while in low-and middle-income countries up to 70% of pathogens isolated in neonatal sepsis may not be susceptible to the recommended first-line regimens [83]. Many neonates in hospitals in south Asia are now treated with carbapenems as a first-line therapy for sepsis or presumed sepsis [84]. This is reflected in our findings with the more frequent appearance of antibiotics such as third generation cephalosporins and meropenem, and tazobactum in studies from Asia and Latin America. Data from South Asia reflect a high burden on neonatal sepsis and a distinct pathogen profile with predominance of Gram-negative organisms and lower prevalence of group B streptococci as compared to high income countries [85]. In a review of neonatal sepsis in South Asia, Chaurasia et al. reported that 50-88% of common isolates from health facilities are resistant to first-line antibiotics ampicillin and gentamicin and often to third-generation cephalosporins such as cefotaxime. However, most remain susceptible to meropenem and vancomycin, antibiotics that are on the WHO-specified "watch group" [85]. The choice of antibiotics in China as reported by Yue et al. [55] is also unusual when compared to most other countries. Authors suggest that this is driven by the high levels of ampicillin resistance and prohibition of gentamicin use due to the high risk of hearing loss in the population. Against this backdrop, the widespread availability and antimicrobial use in neonates and the contribution of antimicrobial resistance as a complicating factor in neonatal sepsis becomes extremely important and rather than increasing use of antibiotics, infection prevention measures such as hand hygiene, surveillance cultures, contact precautions and antibiotic stewardship should be implemented [86].
Our findings are in keeping with previous reviews. Allegart et al.(2019) [3] which updated the review by Rosli et al. 2017 [87] focused on research objectives, methodology and patterns of drug use across neonatal units. This review also highlighted that antimicrobials such as penicillins and aminoglycosides are amongst the most frequently prescribed drugs to hospitalized infants which is consistent with our findings. Krzyzaniak et al. (2016) also highlighted the frequent report of antibiotics in their included studies [88]. They concluded that patterns of drug utilization were similar across the globe. Our findings, although broadly consistent with this, do demonstrate some variations which may be explained by the difference in disease burden and pattern of antibiotic use in different regions of the world. This difference may be explained by the limited number of studies included in Krzyzaniak et al. In addition, although individually several studies do report this plausible relationship [18,23,32,34,38,43], we did not see a consistent relationship between the proportion of premature infants included in the studies with the number of drugs prescribed per infant as reported by Krzyzaniak et al.. This variation may be because the relationship between prematurity and drug utilization is not straightforward. Moderate to late preterm infants are often well with minimal medical needs while some term infants suffer significant morbidities requiring multiple drugs and prolonged intensive care. The large proportion of term infants who do not require any medications are not admitted to neonatal units and hence are not included in studies where the population is restricted to those in the neonatal unit. In this population, the number of drugs per infant may be more affected by the criteria for admission, range of gestational ages admitted and morbidities in those infants.
Although the included studies have all reported use of medicines prescribed to infants admitted to neonatal units, the studies do not report the admission criteria for their units. Variations such as those in types of neonatal units (for example those providing high levels of intensive care or surgical units vs. special care nurseries) and difference in survival of extremely preterm infants (who form a large part of the work in high-income countries but may not survive beyond a few hours in low-income settings) could account for variations that make any cohesive analysis difficult. The analysis of data extracted from the included studies is limited by the heterogeneity of the included populations, variations in study designs and different methods of reporting the findings. In addition, our review is limited by exclusion of non-English-language studies which may be the reason for missing data or very few reports from some parts of the world.

Conclusions
We found that the pattern of drug utilization in neonatal units is largely similar across global regions. A few exceptions reflect the patient population included in the study, differences in the burden of neonatal pathologies and the variations in antibiotic usage reflect the global burden of antimicrobial resistance. The review also highlights the lack of details such as paucity on information indication, dose and duration of use or adverse effects, calling for improved collection and analysis of drug utilization data in neonatal medicine. Such research, particularly when conducted collaboratively across national and continental boundaries, is imperative to promote rational use of medicine in neonates.

Patents
Not Applicable.   Table A1. Search strategy for drug utilization review (from inception of database to February 2019 using OR, AND).

Database
Search Terms     Inclusion: all admitted neonates Exclusion: blood products, IV fluids, TPN Not reported Not reported Not reported Table A2. Cont.