Development and evaluation of an assessment of the age-appropriateness/inappropriateness of formulations used in children

To develop and test a tool to assess whether a DF (formulation) is appropriate for children and estimate the proportion of formulations considered ‘inappropriate’ in a cohort of hospitalised paediatric patients with a chronic illness.

UK NHS Research Ethics Committee (North-West Liverpool East, REC Reference:14/NW/1437, Approval Date:08/12/2014) and the Health Research Authority (HRA), Integrated Research Application System (IRAS) no:164741.

The tool was developed using a selection of cases (purposive sampling) gathered in the AaF study. Accordingly, data collection is described first. Data collection provided both a broad and representative corpus of medicine administration episodes (MAEs) for tool development and a separate, but much larger, dataset for the evaluation of MAEs.

A prospective, observational study collected MAE data from a UK tertiary paediatric hospital (Alder Hey Children’s NHS FT, (AH)) and neonatal unit (Liverpool Women’s NHS FT, (LWH)) across several recruitment periods, between January 2015 and March 2016 (18 weeks in total). On admission, patients were tracked (until discharged from admission ward) and assigned a unique study ID code. Targeted paediatric speciality wards included cardiology, nephrology, rheumatology, gastroenterology, neonatology, and neurology (ward based “sample of convenience”).

Written informed consent was obtained from parent/guardian for children under 16 years; 16–17 years provided own consent; written assent was requested from children over 6 years. Consent allowed the research team to access patient clinical/prescription records needed to assess study eligibility.

Participants aged 0–17 years, diagnosed with a long-term medical condition (LTMC) and taking at least one medicine at the time of review. Prematurity was included as a LTMC.

Data were gathered by review of clinical records and in person. Details of MAEs were recorded on a case report form (CRF), noting how the intended dose was prepared and administered, including any physical modification of the DF and/or if any adaptation was needed to aid administration; for example, mixing with a drink due to a palatability issue. Product brand/manufacturer details were recorded. The researchers did not directly observe the medicines being prepared or administered.

Each medication review provided a “snapshot” of all medicines to be administered on that day to the study participant. Further consent was obtained for patients readmitted during a subsequent data collection period and a new study ID code assigned.

All pharmaceutical preparations of medicines prescribed were included for assessment including tablets, capsules, oral liquids, inhalers, nebules, dermatologicals, suppositories and injections. Medical devices (e.g. sodium chloride 7% nebules (Nebusal®)) were also screened for suitability. Intravenous fluids, parenteral nutrition, dialysis products, dressings, cleansing agents, cosmetics and food supplements were excluded.

Data were transferred to a spreadsheet (Microsoft Excel 2016, Version 16.0) for each hospital site.

The AaF tool is a flowchart designed to aid categorisation of medicines in terms of their age-appropriateness. As shown in Fig. 1, the development of the AaF tool occurred over two phases—(I) learning (test stage) and (II) validation.

During the learning phase, three pharmacist investigators independently assessed 20 ‘test cases’ (AH 16/LWH 4). The ‘test cases’ represented a total of 112 MAEs from the AaF study. This was a purposive sample selected to include a variety of DFs, different routes of administration (RoAs) and included MAEs across the full age spectrum (neonates, infants, young child, older child, teenagers, and adolescents) [25, 26]. A comparison of the ‘test case’ age-appropriateness outcomes was made. Any discrepancies were discussed between the three investigators (“The panel”) until consensus agreement was reached on the ‘age-appropriate’ outcome for each of the test case reviews.

An adapted version of the tool was developed through consensus between the three investigators. Each question (included in the AaF tool) and resulting ‘pathways’ were reviewed by the research team to ensure relevance and that all possible MAE scenarios had been covered. Any questions leading to disagreement between users were adapted appropriately by consensus. The final version of the tool (Fig. 2) consists of five questions, with dichotomous responses to each question.

Using the tool, the assessor assigned one of seven age-appropriate outcome result categories. A further detailed sub-category could also be selected to aid data analysis. The assessor noted any specific reasoning for their decision; for example, if a medicine had been added to food to aid administration. Further detail is provided in Online Resource 1.

To test for differences between users of the tool, inter-rater reliability (IRR) was measured. For a case to be marked as ‘matched agreement’, the detailed sub-category outcome assigned by each assessor matched exactly (Online Resource 1).

As seen in Fig. 3, the 20 ‘test cases’ previously reviewed during the initial learning phase were not included in the validation stage. Remaining cases were divided equally between two assessors for independent assessment. Random validation checks between assessors and an external senior pharmacist were undertaken.

Any medicines prescribed and intended to be given on the day of review (24-h) were recorded for each patient. Patient characteristics included age, gender, weight, attending consultant and LTMC diagnosis. Product information included drug name, strength of active pharmaceutical ingredient (API), DF, manufacturer (including brand name), how the medicines was used/taken, administration device, prescribed dose, frequency, and route. Drug history was verified to determine if the item prescribed was new or taken prior to admission.

Statistical Package for the Social Science (SPSS), Version 24 for descriptive analysis of data. Each individual MAE was used as the unit of analysis (UoA).

A total of 610 participants (including 15 readmitted at AH) were consented into the AaF study. 2,199 MAEs were recorded as being successfully administered to 427 children. Infants (28 days to 11 months) were the most common age group evaluated (104/427, 24.4%). Demographic analysis is shown in Table 1.

Assessor 1 checked 111 MAE assessments completed by Assessor 2. Assessor 2 checked 109 MAE assessments completed by Assessor 1. Percentage exact agreement (%EA) of MAE assessments was found to be 91.7% (99/108) and 93.1% (95/102) for Assessor 1 and Assessor 2, respectively as 10 MAEs were referred to panel. Overall %EA of 92.4% (194/210) for all 40 ‘dual-check’ cases was found and therefore no further duplicate checks were undertaken to test the reproducibility of assessment outcomes by different reviewers. Assessor 3 (senior pharmacist) checked 234 MAE assessments completed by Assessor 1/2 (or both if duplicated case). 29 MAEs were referred to panel. Assessor 3 had %EA of 83.9% (174/205) against either assessor. Any disagreements found (including all ‘refer to panel’ cases) were adjudicated and resolved by discussion at the panel consensus meeting thus 100% agreement was reached for all 40 ‘dual check’ and 40 ‘validation’ cases reviewed.

Results of the MAE assessments are shown in Fig. 4. In total, 259/2,199 (11.8%) MAEs were found to use AiFs, of which 201/259 (77.6%) were at AH site. At AH 34.0% (597/1,758) MAEs episodes and 20.4% (90/441) MAEs at LWH were classified as AaFs. The prevalence of AiFs detected was inversely proportional to age at each site (Online Resource 2). At AH the main reason for manipulating a medicine was DF issue (39% of 201 MAEs) as in general, a smaller dose was required than was available; followed by need to facilitate administration (35%, 70/201). At LWH, this was found to be 67% for DF issue and 26% to facilitate administration of the 58 MAEs recorded as AiFs.

The greatest proportion of MAEs recorded at both sites involved a licensed medicine being used in an OL manner (AH 46.4% (800/1,723), LWH 67.4% (296/439)). At AH, 7.7% (107/1,397) of all the medicines given did not have a paediatric indication; at LWH, this figure was 4.4% (17/386). Only 20.5% (90/439) of MAEs captured at LWH were recorded as being given in a licensed way, according to the products MA. The number of authorised medicine administrations captured was higher at AH with 597/1,723 (34.6%) being given as intended and in accordance with the products MA. This included any permitted manipulations needed. UL medicine use was found to be 18.9% (326/1,723) and 12.1% (53/439) of all MAEs documented at AH and LWH sites, respectively.

Medicines categorised as AiFs more than once in the study are shown in Table 2. The most common AiFs identified were Melatonin 2 mg Capsules and Phosphate 500 mg Effervescent Tablets at AH and LWH sites, respectively. At AH, a third (65/201, 32.3%) of MAEs subsequently categorised as AiFs involved a solid oral dosage form (SODF) such as tablets/capsules (24 distinct APIs). The greatest proportion of AiFs identified were administered via the oral route (29.4%, 59/201) closely followed by Percutaneous Endoscopic Gastrostomy (PEG) administration (27.4%, 55/201) at AH. At LWH most AiFs were administered parenterally (60.3%, 35/58) followed by administration via orogastric tube (OGT) (17.2%, 10/58).

Age-appropriate assessment of DF administration has “context-independent” and “context-dependent” components. The AaF tool offers a structured approach to both components, recognises a multitude of reasons (including preference) and helps to minimise differences between assessors. Internal validation of the tool has demonstrated good IRR, albeit by the people who developed it.

The tool uses reports of direct observation of attempted/successful administration to confirm and thus expand upon the ages of children and reasons why a medicine may either be classified as AaF or AiF [23, 27‐29]. The classification ‘Maybe age-appropriate’ indicates that the DF was either authorised but used OL or was an UL medicine. Further research and development by manufacturers and scrutiny by the regulator, or research and publication by scientists/clinicians, would give reassurance that such use is appropriate. When formulations are identified as being age-inappropriate, understanding the ‘real-life’ alternate administration strategies used by caregivers to improve acceptability, swallowability and/or dose adaptability of the product to facilitate administration to a child can be a vital source of information for the pharmaceutical industry and medicines regulators [30].

The need to develop new authorised medicines acceptable for children and their carers is important, as is development of paediatric versions of older medicines, some of which are UL [31‐34]. Observational data collected in 2016 was used to develop and test the AaF tool in 2019. Since the AaF study patient level data was gathered, three medicinal products have been exclusively designed for use in children during 2016–2018 and approved under the EU Paediatric-Use Marketing Authorisation (PUMA); namely Glycopyrronium bromide (Sialanar®), Melatonin (Slenyto®) and Vigabatrin (Kigabeq®) [35]. Oral suspension of omeprazole was licensed in 2019 for children and adults [36, 37]. Inspection of our results shows that melatonin capsules and omeprazole MUPS frequently caused problems and were considered AiFs so these new authorised DFs might be expected to improve acceptability. Repeating observations of administration with the tool can confirm this. Anecdotally (NPPG online discussion forum https://​nppg.​org.​uk/​) suggests that issues still exist and that the additional cost of changing practice is a significant factor (e.g. Melatonin tablets continue to be crushed and administered with liquid rather than using Slenyto®).

The use of AiFs was identified across all age groups at both hospital sites, demonstrating a partial lack of suitable AaFs available for use in children at that time. The proportion of authorised medicines used was found to be higher at AH site, likely due to the age of the patient population as very few medicines have been specifically designed for neonatal use, thus younger children are more likely to encounter AiFs. The prevalence of ULOL medicine use in children, found to be 68.2% of MAEs (1,475/2,162) within this study, across both hospital sites, shows a significant unmet clinical need continues [38‐41].

We have previously shown that DF modification at the point of administration to achieve the required dose is common in paediatric practice [42]. However, the safety and risks in terms of dose accuracy and handling associated with this practice are often unknown [11, 12]. This study demonstrated that manipulations of medicines used in children are performed for many reasons, including patient preference. The most common reason recorded in this study was due to an issue with the DF available, in terms of its strength and/or DF type.

Enteral tube administration (ETA) was identified as a common way to give medicines to children in an inpatient setting. The Summaries of Product Characteristics (SmPCs) for all oral medicines included in the study were reviewed for information in 2019, in relation to their use via any form of ETA. At AH, only 9% of 1,106 oral MAEs included information regarding administrations via a type of ETA. This figure was lower at LWH site, with only 5% of 206 oral MAEs mentioning use of ETA within the SmPC. Administering medicines through enteral tubes without appropriate information can potentially lead to inaccurate dosing and/or blockage of the feeding tube, especially when non-liquid medications are used [43]. Advice is provided on conducting appropriate studies [13, 44].

Other studies have reported assessment of AaFs [22, 23, 38, 45, 46] by screening lists of medicinal products according to set criteria or making assumptions for the expected ‘user’; in terms of generic age/weight band profiles and the anticipated physical capability of a child, consistent with paediatric developmental milestones. One such assumption is that children under 6 years are unable to swallow SODFs such as tablets, which is not the case [47]. Children, as young as 6 months [48] and even neonates [49], are capable of swallowing mini-tablets and for standard tablets this can be as young as 2 to 4 years, depending on the DF size [47, 50]. Participation in ‘pill-school’ education and training programmes is particularly helpful in enabling children to learn the skills necessary to swallow SODFs [51].

This study is one of few [39, 52] that includes individual patient/user information (including overall acceptability and palatability) which formed part of each medicine assessment; limiting the number of assumptions needed. In development it has expanded the detail in which AiFs can be categorised and subsequently, in application to the remaining MAEs it has been demonstrated to be capable of consistent application.

A limitation of this study could be the age of the observational data used to develop and validate the tool and to report on the use of AiFs at that time (2015–2016). The age of the data does not affect the internal validation. The components of the data collection were decided on before the tool was designed [53]. It is possible that other elements of data collection could lead to an improved tool. If practice changes it is simple to add additional reasons for manipulation of medicines if not covered adequately by the tool. Discussion with clinical practitioners, review of drugs in current practice and of current literature confirms that practice has changed little in the intervening time period and many of the useful improvements in paediatric formulation (e.g. mini-tablets; paediatric strength injections) have not yet been placed before regulators or brought to market [54].

Not all aspects of a formulation contributing to age-appropriateness were assessed (e.g. excipient content) [55‐57]. The proportion of medicines found to be age-appropriate may be an overestimate particularly for oral liquids [55]. A simple question can be added to the tool to accommodate information on the suitability of excipients. No ‘unsuccessful’ administrations were recorded as part of this study. This may relate to the specialist hospital settings and the “one-day snapshot” study design as problematic medicines may have been resolved, prior to review.

There is a need to quantify the impact of using AiFs for carers/patients, in terms of their safety, inconvenience and cost (using a risk–benefit analysis) [58, 59]. External validation should be undertaken in differing settings and the tool adapted for those prescribing and selecting appropriate medicines to administer.