Article Text

Original article
Impact of clinical pharmacist interventions in reducing paediatric prescribing errors
  1. Cecilia M Fernández-Llamazares1,
  2. Miguel A Calleja-Hernandez2,
  3. Silvia Manrique-Rodriguez1,
  4. Cristina Pérez-Sanz1,
  5. Esther Duran-García1,
  6. Maria Sanjurjo-Saez1
  1. 1Pharmacy Service, Gregorio Marañón General University Hospital, Madrid, Spain
  2. 2Pharmacy Service, Hospital Virgen de las Nieves, Granada, Spain
  1. Correspondence to Cecilia M Fernández-Llamazares, Pharmacy Service, Hospital General Universitario Gregorio Marañón, C/ Doctor Esquerdo 46, 28007 Madrid, Spain; cmartinezf.hgugm{at}


Objective To assess the impact of pharmacist intervention in reducing prescribing errors in paediatrics, and to analyse the clinical significance and reasons behind the errors detected.

Methods Cross-sectional epidemiological study analysing the activities of the paediatric pharmacist in a maternity and children's hospital with 180 paediatric beds, between January 2007 and December 2009. The following variables were analysed: impact of the pharmacist's recommendation on patient care, reason for the intervention, clinical significance, type of negative outcome associated with the medication, acceptance rate, medication involved, intervention detection date and observations.

Results A total of 1475 interventions in medical orders for 14 713 paediatric patients were recorded (40 (2.9%) extremely significant interventions and 155 (11.1%) very significant interventions). There were 1357 prescribing errors, 833 of which were dosing errors. 2.2% of the errors detected were potentially fatal (30 cases) and 14.3% (194 cases) were clinically serious. The main reason for interventions was detection of a dosage between 1.5 and 10 times higher than that recommended. The overall rate of acceptance of the pharmacist's suggestions was 94.3%. The pharmacist carried out an average of 0.019 interventions per patient day throughout the study period.

Conclusion Interventions by a clinical pharmacist had a major impact on reducing prescribing errors in the study period, thus improving the quality and safety of care provided.

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.


Over recent decades, the pharmacist's role has evolved with the development of pharmaceutical care, defined as the active participation of the pharmacist in patient care, in collaboration with the doctor and other healthcare professionals in order to achieve results which improve the patient's quality of life.1

According to Dean et al, “a prescribing error occurs when, as a result of a prescribing decision or prescription writing process, there is an unintentional significant reduction in the probability of treatment being timely and effective or increase in the risk of harm when compared with generally accepted practice”.2

In 1999, Overhage et al3 examined the problems posed by the lack of a validated scale to evaluate pharmacists' influence on patient care, by gathering data from studies assessing pharmacist intervention and from studies analysing the severity of errors at the prescription stage (when most errors occur among both adult patients4 5 and children).6 Using these data, they constructed and validated a simple scale with two dimensions: ‘severity of prescription errors’, based on the classification system proposed by Folli et al,7 and ‘the value of pharmacists’ clinical interventions', based on the categories set out by Hatoum et al.8

What is already known on this topic

  • The clinical pharmacist model is relatively new in Spain.

  • Dosing errors are one of the most frequent prescription errors in paediatrics.

What this study adds

  • A clinical pharmacist was effective in detecting prescribing errors.

  • A small but significant number of the prescribing errors detected were potentially fatal.

Hepler and Strand described the concept of medication-related problems (MRP) and developed a system for categorising them.9 Over time, however, their system was found to produce highly heterogeneous results. Subsequently, both the concept and classification of MRPs gradually changed until 2004, when the Spanish government produced the Consensus Document on Pharmaceutical Care.10 The members of the group which drew up the document felt it was necessary to make a clear distinction between problems related to the process of medication use and patient health problems arising when medication use leads to an unexpected or undesired effect. At the Third Consensus of Granada on Drug-Related Problems,11 the two concepts were defined as follows:

  • MRP: situations which, throughout the process of the use of medications, cause or may cause the appearance of a negative outcome associated with medication (NOM);

  • NOM: patient health outcomes that are not consistent with the pharmacotherapy objectives and are associated with the use of medicines.

The later development of clinical pharmacy as a specialty in Spain compared to other countries has meant that there are few clinical pharmacists, although great efforts are being made to establish clinical academic and practical training.12 13 However, there have been many advances in automation in our country, especially in our setting, with regard to the development and implementation of computerised prescription order entry systems, automated dispensing cabinets and even in the use of smart infusion pumps. In this situation, clinical pharmacists sometimes take part in the rounds, but their main contribution to improving patients' pharmacotherapy is by validating medical orders. Therefore, we need to record and analyse clinical interventions to confirm the value and clinical importance of clinical pharmacists. Our validation focuses on confirming the correct drug dose for the particular patient, the suitability of the drug prescribed and its indication, the dose adjustments recommended in light of the patient's characteristics, the absence of contraindications and clinically significant interactions, and proposals for alternative drugs when they are not included in the hospital's pharmacotherapeutic guidelines.


The aim of this study is to assess the impact of pharmacist intervention on reducing prescribing errors in paediatrics, and to analyse the clinical significance and causes of the errors detected.

A cross-sectional epidemiological study was carried out by a clinical pharmacist who analysed the activities of other paediatric clinical pharmacists between January 2007 and December 2009. The study was carried out at the maternity and children's hospital, part of the Gregorio Marañón General University Hospital in Madrid. The maternity and children's hospital has 180 paediatric beds and 138 obstetrics and gynaecology beds. It has an electronic prescription system associated with 18 automated dispensing machines. Patients were excluded if, in the 24 h following the intervention, they were transferred to a hospital unit without an electronic prescription system, they were discharged or they died. In these cases, the result of the intervention was deemed impossible to assess.

The clinical pharmacists systematically recorded their recommendations in a Microsoft Access computer database validated prior to the study, with acceptable inter-rater reliability κ values. This database stored the measurement variables described below (impact of pharmacist's recommendation on patient care, reason for intervention, clinical significance, NOM type, acceptance rate, medication involved (active principle and brand name), pharmacist responsible, date of detection of the intervention, clinical unit and observations).

The impact of the pharmacist's recommendation on patient care was assessed using a slightly modified version of the scale designed by Overhage et al3 (box 1). The severity of the prescribing error was also assessed using the same modified scale, generating a categorical variable with five possible categories: 1, potentially fatal; 2, serious; 3, significant; 4, minor; and 5, no error (see supplementary online table 1). The severity of the prescribing error was also examined using a logistical regression analysis to assess the influence of medication type and medical specialty.

Box 1 Impact of the pharmacy service and severity of the prescribing error

Impact of the pharmacy service.

  1. Extremely significant.

    • The recommendation resolves a situation that could have extremely serious consequences, or a situation that puts the patient's life in danger.

  2. Very significant.

    • The recommendation prevents real or potential damage to a vital organ.

    • The recommendation prevents a serious adverse effect resulting from a drug interaction or contraindication.

  3. Significant

    • The recommendation improves the quality of life of the patient (standard practice defined by the hospital)

  4. Somewhat significant

    • The recommendation has a neutral effect depending on how it is interpreted by the professional involved (to differentiate it from significant recommendations, where the hospital's standard practice supports the recommendation)

  5. Insignificant

    • For informative purposes only.

    • General recommendations, not specific to a certain patient.

  6. Harmful intervention

    • Inappropriate recommendations that could lead to a worsening of the patient's condition.

Acceptance rate

The pharmacist's recommendations were divided into one of three categories: accepted, not accepted or impossible to assess. Recommendations were considered to have been accepted if the physician implemented the change suggested by the pharmacist within 24 h of the recommendation. Analyses were then carried out to determine if there was any link between acceptance of the recommendation and the clinical significance of the intervention, using the χ2 test for linear trend.

Possible NOMs were separated into three groups: potential, real (appearance of NOM) and NOM-free. A ‘suspected NOM’ was recorded in situations where the patient ran the risk of a health problem associated with the use of medication, generally due to the existence of one or more MRPs considered to be risk factors for that NOM.

As with MRPs, NOMs were divided into three categories (necessity, effectiveness and safety) and each of these categories was further subdivided into two subcategories (see supplementary online table 2). Interventions where there was considered to be no possibility of any NOMs were classified as NOM-free.

Table 2

Reasons for pharmacist intervention

Both the brand name and the active principle of medications involved in the intervention were recorded. The database also included a free text observations field where precise pharmacotherapy follow-up information for each patient could be entered. In terms of drug activity indicators, the number of interventions per patient day was calculated, where the number of patient days was the number of active prescriptions per patient per day (so that all patients who were prescribed medication were taken into account).

The requirement for informed consent was waived and the authors were authorised by the Ethics and Investigation Committee to sign an undertaking to preserve confidentiality and to use the data collected solely for the purpose of scientific publication.


A total of 1475 interventions in 61 458 medical orders for 14 713 paediatric patients were recorded (2.4 interventions/100 medical orders).

The unaccepted interventions were excluded, leaving 1391 accepted interventions. The distribution of the different impacts of those interventions is shown in figure 1. There were 40 (2.9%) extremely significant interventions and 155 (11.1%) very significant interventions (fair inter-rater reliability of 0.21–0.40). No harmful interventions were detected.

Figure 1

Impact of pharmacist's recommendations on paediatric units.

Among the accepted interventions, the pharmacist detected 1077 potential or real NOMs (almost perfect inter-rater reliability of 0.81–1.00), 7.6% of which (82 cases) were related to necessity, 27.6% (297 cases) to effectiveness, and 64.8% (698 cases) to safety (supplementary figure 1).

There were 1357 prescribing errors, 833 of which were dosing errors. Overall, 30 errors (2.2%) were potentially fatal (table 1), 194 (14.3%) were clinically serious, 874 (64.4%) were significant and 259 (19.1%) were of minor significance.

Table 1

Potentially fatal prescription errors

The most frequent interventions (substantial inter-rater reliability of 0.61–0.80) by the pharmacists were in relation to dosage errors where there was a significant overdose (table 2). There were also a significant number of underdosing errors.

Antibiotics (16.3%), antiemetics (12.1%) and gastroprotective agents (11.8%) were the therapeutic groups with the highest number of interventions. Ondansetron (12.1%), ranitidine (11.8%), amoxicillin/clavulanic acid (6.3%), paracetamol (6.1%) and metamizole (4.7%) were the individual medicines with the highest number of interventions.

Once the recommendations where no prescribing error had been made were excluded, it was found that 94.3% of the 1357 interventions detected were accepted. There was also a linear relationship between the severity of the error and the rate of acceptance, so the more serious the error, the more likely it was that the recommendation would be accepted (χ2=3.476, p<0.05).

There were 84 instances where the pharmacists made a recommendation which was not accepted by the prescriber. In 14 cases there was a negative outcome. The most frequent negative outcome was quantitative ineffectiveness (underdosing).

The pharmacist carried out an average of 0.019 interventions per patient day throughout the study period.


A number of authors have confirmed the lack of uniformity of studies into the prevalence of errors and the differing definitions proposed for those errors,14,,16 resulting in data varying from one study to the next. Therefore, before proceeding with the data analysis, it was important to confirm that the considered errors were consistent with the most uniform definition of a paediatric prescribing error.17 In our study, of the 30 reasons found for prescribing errors, only two of the 21 assessed in the literature consulted were not considered to be errors.17 However, seven additional reasons not assessed by Ghaleb et al were considered by other authors to be causes of errors.3

In the vast majority of cases, the pharmacist's interventions had a significant impact on the patient's health. Similar studies have found that 93%18 of all interventions had a positive impact. The percentage found here (78.6%) is somewhat lower than but comparable to that found by Virani et al (86%).19

The majority of possible harmful implications were related to safety followed by effectiveness, with a minority related to necessity of treatment. These findings are similar to those of other studies in paediatric settings, where 11% of errors were found to be related to necessity, 32.5% to effectiveness and 52.5% to safety.20

Potentially fatal errors accounted for just over 2% of the prescribing errors. This is higher than reported in most studies which found an incidence ranging from 0.2% to 1.28%,21,,23 with one study describing a higher incidence of 5.6%.7 None of the studies mention any harmful effects on the patients, probably because the errors were corrected prior to administration.15

Dosing errors were the most frequent prescription error and in most cases involved an overdose. This is in keeping with the majority of previous studies,7 14 18 21,,33 although some have found that dosage errors are not the most frequent prescribing error.34

Antimicrobials are often the most frequently prescribed group of medicines and it is therefore unsurprising that they are the therapeutic group with the greatest number of interventions. These results are consistent with those of other studies, which have found that antibiotics21 25 26 29 and sedatives24 26 35 are most frequently involved.

Furthermore, the drugs involved in the most serious errors, as we have seen, include high-risk medications, such as potassium, digoxin and insulin.31 36 37 That is consistent with the high-risk concept: prescribing errors in these drugs are not the most frequent, but their consequences are clinically more serious. Most of the drugs involved in errors have a narrow therapeutic index, and so errors have more serious repercussions.15 36,,38

With regard to acceptance rates, it is important to point out that the rate found here, which is over 94%, is very similar to those found in other, similar studies, which reported figures ranging from 90.4%,21 91% and 95.8%,18 to the highest, 98%.19

Quantifying the activities of the pharmaceutical service is perhaps one of the most complicated aspects of this study because of the heterogeneity of indicators and the lack of standard measurements.39 Our findings were similar to those of others,7 although some have found a much higher rate of interventions.32 40

The main limitation of this study is the fair inter-rater reliability of the intervention impact between pharmacists.41 In order to improve the coding process, we wrote a standard operating procedure which explains the meaning of each of categorisation variable, and provides practical examples of how to categorise complex cases.

In other countries, clinical pharmacists are very much involved in multi-disciplinary teams, attending the rounds with the clinicians and participating actively in decisions on drug therapy and averting prescribing errors.

Interventions by a clinical pharmacist had a major impact on reducing prescribing errors, thus improving the quality and safety of care provided.


The authors thank all paediatricians working in Hospital Materno-Infantil Gregorio Marañón for their help, patience and support for the clinical pharmacist. The authors would like also to thank Jose María Bellón for his help with the statistical analysis.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

    • Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
    • Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
    • Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Competing interests None.

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