Article Text

Download PDFPDF

Criteria for the selection of paediatric patients susceptible to reconciliation error
  1. Dolores Pilar Iturgoyen Fuentes1,
  2. Clara Meneses Mangas2,
  3. Margarita Cuervas Mons Vendrell1
  1. 1 Pharmacy Service, Hospital Infantil Universitario Nino Jesus, Madrid, Spain
  2. 2 Pharmacy Service, Hospital San Juan Grande, Jeréz de la Frontera (Cádiz), Spain
  1. Correspondence to Dr Dolores Pilar Iturgoyen Fuentes, Hospital Infantil Universitario Nino Jesus, Madrid 28009, Spain; dipifuentes{at}


Objectives Many medication errors occur during care transitions, which are critical points for patient safety. There is strong evidence in favour of medication reconciliation as a strategy to avoid errors in adults, though few studies have been made in the paediatric setting. Likewise, no recommendations have been established for the selection and/or prioritisation of paediatric patients amenable to reconciliation.

Methods A retrospective study was conducted involving patients subjected to reconciliation by a pharmacist on admission to hospital and who experienced at least one reconciliation error between January and November 2018. Univariable and multivariable analyses were performed to identify possible factors associated with reconciliation error, using a logistic regression model to determine the odds ratio (OR) with the corresponding 95% confidence interval (95% CI).

Results The group of patients with at least one reconciliation error included 334 patients, compared with the group of patients without reconciliation errors, which included 1426 patients. It was determined that schoolchildren and adolescent patients had a risk of presenting a reconciliation error on hospital admission that was more than double for younger patients (OR 2.32, 95% CI 1.26 to 4.25, and OR 2.68, 95% CI 1.44 to 4.99, respectively). This risk was multiplied by five if we compared polymedicated patients versus non-polymedicated patients (OR 4.48, 95% CI 3.35 to 5.99). Patients with a neurological or onco-haematological underlying disease had a 12 and 10 times higher risk of presenting a reconciliation error compared with patients with other types of underlying diseases (OR 11.97, 95% CI 7.57 to 18.92, and OR 9.96, 95% CI 6.09 to 16.28, respectively). Finally, patients with narrow therapeutic index medicines in their usual treatment had an almost three times greater risk of presenting a reconciliation error when admitted to the hospital, although this last factor was not determined as an independent risk factor as for the others (OR 2.98, 95% CI 2.22 to 3.99).

Conclusions The paediatric population is characterised by a number of risk factors for reconciliation error. Knowledge of these factors can allow the prioritisation of medication reconciliation in a concrete group of patients. In order to generalise the results obtained in this study, they must be confirmed in other paediatric care settings involving larger samples and different types of patients.

  • Safety
  • Education, Pharmacy, Continuing

Data availability statement

Data sharing not applicable as no datasets were generated and/or analysed for this study. Not applicable.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • The benefits of medication reconciliation are well established in adult patients, but not in paediatric patients, being a population not included in the guidelines for medication reconciliation published so far. However, it is known that a significant number of children have chronic illnesses leading to complex pharmacological treatment.


  • There are a series of specific factors that cause a greater risk of medication errors in children.


  • Once these risk factors have been identified, we can make a selection of patients to prioritise the available healthcare resources and develop specific conciliation protocols for paediatric centres, like those in adult hospitals.


Medication error (ME) is one of the leading causes of avoidable damage in the healthcare setting. Ensuring patient safety is a crucial aspect of clinical practice, and in this regard multiple interventions have been developed with the aim of reducing the impact of ME. In 2017, in the setting of the Third Global Patient Safety Challenge, the WHO proposed different solutions to guarantee safe practices, under the principle of ‘Medication without harm’.

ME may occur at any point in the medication utilisation process, though its impact is greater under certain clinical circumstances—particularly in the hospital setting. According to the WHO, the priority concerns in this respect are polymedication, care transition and paediatric patients, since these constitute a particularly vulnerable population group.1 In effect, paediatric patients are particularly susceptible to ME, and it has been reported that the probability of ME leading to an adverse event (AE) in such patients is multiplied by three with respect to the adult population.2–4

In order to avoid or reduce errors in care transition, use can be made of strategies such as medication reconciliation (MR).5 6 This is defined as the formal and standardised process of obtaining the full medications list of the patient before care transition, comparing it with the active prescriptions, and analysing and resolving the discrepancies found in order to guarantee that the patient receives all the medications of his or her chronic treatment, adjusted to the current clinical condition and to the new prescriptions made following care transition. MR can be considered complete when each medication the patient is receiving has been intentionally continued, discontinued or modified at each care transition point.7

The MR process should be based on the following steps8:

  1. Compilation of the full and exact list of the regular medication used by the patient

  2. Review of the medication prescribed on admission

  3. Comparison of both lists and detection of possible discrepancies

  4. Reporting of discrepancies to the supervising physician and resolution of the problem

  5. Registry of the resolved discrepancies

  6. Generation of a list of reconciled medication.

The Canadian Patient Safety Institute, in the document ‘Safer Healthcare now’, classified the discrepancies as follows9:

  1. Justified intentional discrepancy: the change is clearly documented or justified

  2. Unjustified intentional discrepancy: the change is not ME, but may cause confusion and could ultimately result in ME

  3. Unintentional discrepancy or reconciliation error (RE): a change was made in a treatment previously received by the patient, without justification; this represents potential ME that can give rise to AE.

MR has been shown to significantly reduce ME in adults,10 11 and many studies on this subject have been published in this population.12–15 In contrast, few studies have focused on the paediatric population, and the sample sizes are moreover small.16–19

Strategies are needed to be able to address a larger number of patients, with more efficient use of the available resources. This requires the selection of those patients who may obtain greater benefit. Such selection in turn can be based on the existence of certain characteristics inherent to the patient or to the treatment received, and which increase the probability of error.

The REDFASTER group of the Spanish Society of Hospital Pharmacy (Sociedad Española de Farmacia Hospitalaria (SEFH)) published a number of criteria to be met by a patient in the emergency room in order to undergo reconciliation, such as an emergency room stay of over 24 hours, home treatment, and an estimated stay in the emergency room of under 24 hours but receiving regular treatment with drugs to be reconciled in less than 4 hours.20

These guidelines do not contemplate the paediatric population, and there are no publications on the criteria or characteristics of paediatric patients amenable to reconciliation.

It is necessary to determine those factors associated with the existence of RE in the paediatric population in order to develop an algorithm for the selection of patients at an increased risk of having RE, and which may serve as a tool for implementing MR in other paediatric care centres.

In 2018, the working group of Meneses et al 21 developed an MR on admission programme for paediatric patients, designed to be implemented by a hospital pharmacist. The study reconciled 1760 patients on admission to hospital. Sixty percent presented with some background disease, with neurological problems being the most common type of disorder. In turn, 14% were receiving regular treatment with four or more medications at home, including some medications with a narrow therapeutic index (MNTI). Discrepancies were detected in 34% of the patients. Of all the identified discrepancies, 63% were not justified and were classified as RE—the main type of error being the omission of some necessary medication. Such REs were detected in 334 patients (56%).

Based on the results published by our working group,21 and in the same way as has been done in the adult population, it would be useful to develop an algorithm capable of identifying those paediatric patients most likely to have RE, and which consequently could benefit most from the MR process. This would allow us to ensure MR in those paediatric patients most at risk of having RE during care transitions, with optimisation of the available resources. By analysing the characteristics of those patients who experience some RE, it could be possible to develop an algorithm for selecting those paediatric patients amenable to reconciliation on admission to hospital.


Taking the results obtained by Meneses et al 21 as a starting point, we conducted a retrospective descriptive study of the characteristics of all the patients subjected to reconciliation by a pharmacist on admission to hospital and who experienced at least one RE during the period between January and November 2018. Those patients with any missing study data were excluded from the analysis. Likewise, an analysis was made of the characteristics of the patients that did not present RE, followed by a comparison of the two groups and an analysis of the possible risk factors associated with the occurrence of RE.

The following parameters were recorded for all patients: age, gender, background disease, number of home medications, polymedication (defined as the use of four or more medications on a regular basis), MNTI in the regular treatment of the patient, residency in the community of Madrid (Spain), updating of the regular treatment in HORUS (electronic medical records for primary care), and department of admission.

The background disease conditions were classified as neurological, onco-haematological, respiratory, psychiatric, gastrointestinal, common paediatric disorders, or others.

The drug classes implicated in the discrepancies were recorded according to the Anatomical Therapeutic Chemical classification of the Spanish Agency of Medicines and Medical Devices (Agencia Española de Medicamentos y Productos Sanitarios (AEMPS)).

Regular or chronic treatment in turn was defined as treatment with a duration of at least 6 months. Polymedication was taken to constitute regular treatment with four or more drugs, based on the ‘Model of selection and pharmaceutical care of chronic paediatric patients’ of the SEFH.22

Medications with a narrow therapeutic index were those listed as MNTI by the AEMPS,23 along with other drugs not included in this list but considered by some working groups as being medications of high perceived risk in paediatrics.24

For the analysis of factors associated with RE, we generated a binomial logistic regression model with RE as a dichotomic dependent variable and those variables found to be significant in the bivariate analysis as independent variables. A stepwise method was used for the selection of variables, presenting a model only with those variables found to be significant.

The multivariable analysis was performed based on a logistic regression model to calculate the odds ratio (OR) with its corresponding 95% confidence interval (95% CI) and p value.


Of the 1760 patients included in the initial study,21 a total of 334 (18.98%) presented at least one RE (range 1–8). More than 50% were males, and most of the subjects were between 5–12 years of age. The most frequent background disease conditions were of a neurological and onco-haematological nature. Almost all the patients received at least one medication on a regular basis, and 32% were polymedicated. In turn, over a quarter of the patients presented some MNTI in their regular treatment, and more than 75% resided in the community of Madrid. The patient characteristics are described in table 1.

Table 1

Characteristics of the patients in the reconciliation error group

Table 2 shows the main study parameters corresponding to the patients without RE and the patients with at least one RE. The age distribution differed between the two groups, with a larger proportion of schoolchildren and adolescents in the RE group versus the patients without RE. Background disease was also more prevalent in the RE group, since only 26 patients in this group had no background disease (7.78%) versus 47.83% of the patients in the group without RE.

Table 2

Comparison of patients with and without reconciliation error

With regard to treatment, statistically significant differences were recorded for practically all the variables, as can be seen in table 2, where polymedication and the number of regular medications were seen to be greater among the patients with at least one RE. Lastly, differences were also observed regarding the percentage of MNTI in the group of individuals with at least one RE. Of note is the observation that a lack of updating of regular treatment in HORUS was also more frequent among the patients with RE.

Table 3 describes the risks for each study variable obtained from the univariable logistic regression analysis. In this regard, schoolchildren were seen to be at greater risk of having at least one RE than newborn or nursing infants. With regard to the number of medications involved in regular treatment, polymedication and the inclusion of MNTI were seen to be associated with the presence of RE. As can be seen in table 3, the existence of background disease was significantly associated with the presence of RE.

Table 3

Univariable analysis of the study variables

Only those variables found to be significant in the univariable analysis were entered as independent variables in the final model (table 4). Age was identified as a risk factor for RE in all age groups except the preschool children. Other factors associated with the occurrence of RE were polymedication, allergies, MNTI, and neurological and onco-haematological diseases. The risks associated with these factors are shown in tables 3 and 4, with statistical significance being observed in all cases.

Table 4

Multivariable analysis of the study variables

In order to validate the model, we analysed its discriminating and predictive capacity. Goodness of fit was confirmed using the Hosmer and Lemeshow test (p=0124), and the predictive capacity of the model was seen to be good (77.32%) (figure 1).

Figure 1

Receiver operating characteristic (ROC) curve of the logistic regression model.


Paediatric patients constitute a vulnerable population, with a high risk of having ME, and strategies must be adopted in order to avoid such errors. In this regard, we consider MR on admission to be one of the measures needed to guarantee the safety of the paediatric population.

In the adult population the MR prioritisation criteria have been well established, though few studies offering criteria for the optimisation of this activity in the paediatric population can be found in the literature. Nolt et al 19 considered the priority paediatric patients for MR to be those admitted to the paediatric intensive care unit or with cardiological or endocrinological diagnoses, since they are more likely to receive a larger number of home medications.

On the other hand, based on the experience and results of Meneses et al,21 other types of paediatric patients should also be prioritised for MR, classified according to their background disease conditions and the type of home medication received. In this sense, the patients most susceptible to experiencing RE would be those with neurological disorders, those of older age, polymedicated individuals with some MNTI in their regular treatment, and patients living outside the community of Madrid.

The univariable and multivariable analyses identified a number of possible risk factors for RE, specifically older age, polymedication, the presence of some MNTI as part of regular treatment, and the existence of background neurological or onco-haematological disease.

With regard to patient age as a factor associated with the risk of RE, our findings are similar to those reported by DeCourcey et al,17 who identified older age as a risk factor for a greater number of adverse events.

In relation to polymedication, our results are consistent with those obtained in many studies in adults and in studies conducted in the paediatric population, such as those of Coffey et al 25 or Nolt et al,19 where patients with four or more home medications were seen to be more likely to have RE. In the paediatric study carried out by Condren et al,26 the number of medications was seen to be a stronger predictor of the appearance of discrepancies than the complexity of the background disease condition. The number of medications or the degree of polymedication would be more precise for stratifying those patients in which MR would be of greatest benefit.

In the study published by Iturgoyen et al,27 compiling the preliminary data of the study of Meneses et al,21 the increase in risk of RE according to the number of medications was evaluated by calculating the odds ratio, and in this regard the use of two or three medications as regular treatment at home, together with premedication, were seen to increase the risk of RE. This tendency has also been indicated in other studies of MR conducted in the paediatric population,19 25 with the number of medications as regular treatment being identified as a risk factor for RE. Thus, polymedication is a criterion that should be taken into account when selecting patients for reconciliation, independently of whether they are adults or paediatric patients.

The presence of MNTI in regular treatment was identified in our study as a factor associated with the appearance of RE, in coincidence with the observations of Unroe et al 28 in adults. On the other hand, antiepileptic drugs were seen to be one of the main medication groups implicated in RE. This may be explained by the frequent dose adjustments needed with these drugs, and which may make it difficult to have up to date information on the treatment dosing status at the time of patient admission. This situation, together with the fact that many antiepileptics are MNTI—with the potentially fatal consequences of omission or overdose—implies that these medications can give rise to serious RE. The results of the paediatric study published by Coffey et al 25 coincide with this, since the authors found antiepileptic drugs to be implicated in five of the six discrepancies classified as serious.

Condren et al 26 found the difference between the number of discrepancies in paediatric patients with complex chronic diseases and the number of discrepancies in uncomplicated patients to be statistically significant. These data are in line with those of our own study, where neurological disorders—regarded as complex chronic diseases or conditions related to complex pharmacological treatment—were identified as constituting a risk factor for RE.

With the selection of these patients it would be possible to conduct more efficient MR, prioritising those individuals found in other centres to be susceptible to RE capable of resulting in ME during hospital admission.

A multicentre study would be needed, involving a larger population with other types of disease conditions, in order to compare the results obtained with those published by Nolt et al 19 and thus draw more extrapolatable findings.

This is the first study to describe criteria for the selection of patients to be prioritised for MR in paediatric centres wishing to implement this care activity in their clinical practice.

Given the important proportion of RE detected in our sample of paediatric patients in the study of Meneses et al 21 and the scarce literature found in relation to MR in the paediatric population, as well as the omission of this population in the existing MR guides, criteria have been established to select those paediatric patients in our setting that are most likely to have RE during care transitions, and who therefore would stand to benefit most from MR. More efficient MR could be carried out in centres with paediatric patients. In this regard, a study applying and validating these criteria would be advisable.

Data availability statement

Data sharing not applicable as no datasets were generated and/or analysed for this study. Not applicable.

Ethics statements

Patient consent for publication


The authors would like to thank all the pharmacists and members of the Pharmacy Service and paediatricians from the hospital. They made this study feasible by their dedication to improving patient safety and safe practice in the healthcare chain. In addition, the authors would like to thank all the paediatric patients and their caregivers for their great contribution to this study.



  • EAHP Statement 4: Clinical Pharmacy Services.

  • Contributors DPIF and MCMV planned the study. DPIF and MCMV performed the study. DPIF and CMM collected the data. DPIF, CMM and MCMV analysed the data. DPIF and MCMV wrote the manuscript. CMM critically reviewed the manuscript. DPIF acting as guarantor

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.