Article Text


Original articles
Medication errors in the homes of children with chronic conditions
  1. Kathleen E Walsh1,2,
  2. Kathleen M Mazor2,3,
  3. Christopher J Stille1,2,
  4. Irma Torres4,
  5. Joann L Wagner2,
  6. Juliet Moretti2,
  7. Kevin Chysna2,
  8. Christy D Stine1,
  9. G Naheed Usmani1,
  10. Jerry H Gurwitz2,3
  1. 1Department of Pediatrics, University of Massachusetts School of Medicine, Worcester, Massachusetts, USA
  2. 2Meyers Primary Care Institute, Worcester, Massachusetts, USA
  3. 3Department of Medicine, University of Massachusetts School of Medicine, Worcester, Massachusetts, USA
  4. 4UMass Memorial Medical Center, Worcester, Massachusetts, USA
  1. Correspondence to Dr Kathleen E Walsh, Benedict Second Floor, UMass Medical Center, 55 North Lake Avenue, Worcester, MA 01655, USA; walshk02{at}


Background Children with chronic conditions often have complex medication regimens, usually administered at home by their parents.

Objective To describe the types of medication errors in the homes of children with chronic conditions.

Methods Our home visit methods include direct observation of administration, medication review and prescription dose checking. Parents of children with sickle cell disease and seizure disorders taking daily medications were recruited from paediatric subspecialty clinics from November 2007 to April 2009. Potential errors were reviewed by two physicians who made judgements about whether an error had occurred or not, and its severity.

Results On 52 home visits, the authors reviewed 280 medications and found 61 medication errors (95% CI 46 to 123), including 31 with a potential to injure the child and 9 which did injure the child. Injuries often occurred when parents failed to fill prescriptions or to change doses due to communication problems, leading to further testing or continued pain, inflammation, seizures, vitamin deficiencies or other injuries. Errors not previously reported in the literature included communication failures between two parents at home leading to administration errors and difficulty preparing the medication for administration. 95% of parents not using support tools (eg, alarms, reminders) for medication use at home had an error compared to 44% of those using supports (χ2=13.9, p=0.0002).

Conclusions Home visits detected previously undescribed types of outpatient errors which were common among children with sickle cell disease and seizure disorders. These should be targeted in future intervention development.

Statistics from


While substantial attention has been paid to inpatient medication errors, there is little information about errors in the home, where the vast majority of medicines are taken. The limited literature that exists indicates that errors may be common and parents are often unaware of them.1,,3 In a multisite study of outpatient medication use among adults and children with cancer, we found rates of medication errors among children to be more than double those of adult patients, resulting primarily from a higher rate of errors in home medications.4 To inform intervention development, a description of the depth and breadth of the problem of medication errors in children at home is required.

What is already known on this topic

  • Children with chronic conditions take complex, life-saving medication regimens in the outpatient and home setting.

  • While substantial attention has been paid to inpatient medication errors, there is little information about medication errors in the outpatient setting.

What this study adds

  • This study used home visits to discover the range of types of medication errors, and associated injuries, in the homes of children with chronic conditions.

  • Previously undescribed errors include complex communication failures between parents at home, misunderstanding physician instructions to change doses and labelling issues.

  • Physicians were unaware of 80% of the errors we detected.

Existing methods are not well designed to detect home medication errors. Because patients are often unaware of their errors, they are unable to report them during an interview and they will not be recorded in the medical record.5 Direct observation of actual medication administration in the home is the best method to determine outpatient administration errors.1 The objective of this study was to describe the range of types of medication errors in the homes of children with chronic conditions so as to inform future intervention development, using sickle cell disease and seizure disorders as model conditions because the medications used have a narrow therapeutic index and missed doses are important. These conditions also have different levels of coordination between families, primary care providers and specialists.


Overview and study site

In this prospective cohort study we reviewed medical records and medication bottles and performed in-home direct observation of medication use for children with sickle cell disease or seizure disorders from November 2007 to April 2009 recruited from subspecialty ambulatory clinics at one academic health centre. The word parent is used to refer to the parent, guardian or primary caretaker. This study was approved by the University of Massachusetts Institutional Review Board.


We included children taking daily home medicines for sickle cell disease or seizure disorders whose parents spoke English or Spanish. We recruited subjects less than 21 years of age (consistent with the National Institutes of Health definition of a child)6 as these subjects were part of the general population cared for in our paediatric subspecialty clinics. We approached families during clinic visits to obtain informed consent. For Spanish speaking families, we used a medical interpreter and Spanish consent forms during recruitment, and home visits were performed by a nurse (IT) who is a native Spanish speaker. Home visits were scheduled when the person who normally administered the medicines was giving them to the child and when the child took the most medicines during the day.

Demographic information

We collected parent self-reports on sociodemographic information (gender, race/ethnicity, educational status, marital status). Because health literacy may be associated with medical error,7 8 we administered the English or Spanish Short Test of Functional Health Literacy in Adults9 and a single item question about how often (s)he needed help reading written materials.10 Parents were asked what support tools, such as calendars or pill boxes, they used to help them give their child medicines at home.

Home visit methods and ordering errors

We developed home visit methods from established methods used to quantify nurse administration errors in the hospital11; this development is described elsewhere.1 The method includes: (1) medication review, (2) observation of medication administration and (3) event classification. The 1 h home visit was carried out by a paediatrician or a study nurse. The paediatrician trained the study nurse in non-intrusive, non-judgemental observational methods using a structured didactic training session and 10 joint home visits. Interobserver reliability for the detection of errors during the home visit, as determined after each had been independently visiting homes for several months, was good (κ=0.89, 95% CI 0.67 to 1.0).

Prior to the home visit, information on the child's age, allergies, weight, height, diagnoses and all medication names, doses and frequencies were obtained from the child's medical record, including the dictated subspecialty clinic notes and the medication list. Prescribed or administered doses more than 10% above or below the recommended dose were considered errors, as in previous studies.12 If there were differences between guidelines and formularies, we chose the lowest minimum and the largest maximum dose and considered anything within these as the recommended dose.12 We did not consider cases where doses outside guideline doses were clinically reasonable (eg, based on drug levels) as errors.

Medication review and direct observation

A medication was defined as any pharmaceutical, biological or intravenous fluid which is prescribed or available over the counter. The study paediatrician or nurse reviewed how each medication was used with the parent and photographed the label. The medication review included double checking the dose, frequency and duration on the bottle label, reviewing the bottle contents and the expiration date. The parent was asked to prepare and administer medications exactly as (s)he would normally do, including any as needed (PRN) medicines that the parent believed should be given at the time of the home visit. In order to describe problems caused by regularly missing medication doses and also avoid over counting minor mistakes, missed doses were counted as an error if 50% or more of the doses were missed in the week prior to the visit, given average adherence rates of 48–72% in published studies.13 Missed dose errors were detected by parent report using a four-item scale14 and medication bottle review. The research nurse was instructed to intervene as needed to prevent an error. A protocol written for the study described who to contact for a dangerous medical error, cases of child abuse or neglect, or parental anxiety or depression. While the nurse did intervene to prevent errors from occurring, no further urgent medical care was needed beyond the nurse's advice.

Potential errors were detected in several ways, including but not limited to: reviewing prescribed doses from the medical record; reviewing the dispensed dose from the bottle label; comparing the prescribed dose to the bottle label with parent reports of how the medication is given, and how the medication is actually given; and by parent report about how each medication was administered. If parents reported administering a dose which was different from that noted in the medical record because the doctor had told them to change the dose, this was not counted as an error.

Event classification

A medication error was defined as an error in drug ordering, dispensing, administering or monitoring.15,,17 Two physician reviewers, trained in event classification methods as in previous studies,15 16 18 classified events as an error with injury, an error with potential for injury that did not injure the patient, or an error with little or no potential for harm. The severity of the injury or potential injury was classified as fatal, life-threatening, serious or significant. Differences in judgements were resolved through discussion and consensus. The inter-rater reliability for independent ratings before discussion showed moderate agreement for event classification (κ=0.5, 95% CI 0.4 to 0.6) and for severity (κ=0.4, 95% CI 0.2 to 0.6).19 The severity ratings here are comparable to other studies in this field.20 One author with experience in patient safety research judged whether the error could be detected using medical record review, parent interview, or by parents bringing medications to clinic.


The primary outcome was a description of the range of types of medication errors detected. Rates of errors per home visit and the percentage of medications with an error were determined and Poisson exact confidence intervals were estimated. We performed bivariate analysis with child, parent and home environment characteristics and their association with the child having at least one medication error. For power calculations we assumed that each child took four medications at home, and that we would find an error in 15% of these medications.21 We determined that we would need to review 200 medications at home to have a 95% CI for error rates per medication of less than 5%.


We approached all 71 eligible parents; 56 (79%) consented to the study and we visited 52. Among the children we visited, 24 had sickle cell disease and 28 had seizure disorders (table 1). The mother was most often the person who had primary responsibility for administering medications to the child (table 2).

Table 1

Demographic characteristics of home visit study subjects and their parents

Table 2

Home environment description

Overall error rates

On 52 home visits with 280 medications reviewed, we found 61 medication errors (95% CI 46 to 123), detected in 22% of medications reviewed (95% CI 17% to 28%) (table 3). Of these, 31 errors had potential to injure the patient and an additional nine resulted in injury. Several errors with injury occurred when parents missed doses or failed to fill prescriptions; as a result the clinical problem being treated persisted (table 4). For example, a child was prescribed ranitidine for gastritis. The child was not taking any ranitidine on the home visit. For several subsequent visits the child continued to report chest and abdominal pain. Occasionally parents did not inform the clinician of the missed doses, the clinician assumed that the medication failed at that dose and increased the dose, performed additional tests or started a new medication. Ten per cent of errors could have been detected by medical record review, 52% by parent interview, 47% by parents bringing medications into the office and 73% by all three. Forty-one per cent of children had no errors, while two children had three errors and one child had four errors. In bivariate analyses, 95% of children whose parents used no support tool at home to help with medication use had at least one medication error compared to 44% of children whose parents used some support tool at home (χ2=13.9, p=0.0002). No other patient or parent risk factors were identified.

Table 3

Number and rates of medication errors for children with sickle cell and seizure disorders visited at home

Table 4

Clinical details of the 61 medication errors detected in children with sickle cell disease or seizure disorders

Errors in children with sickle cell disease

The most common types of errors among children with sickle cell were missed doses and previously undescribed types of administration errors. Administration errors included problems dissolving oral chelation therapy in water and consuming the entire 8 ounces of liquid. Injuries resulted when some children with evidence of liver damage rarely took their chelation therapy at all (eg, a child with liver failure due to chronic transfusion therapy who missed more than half his chelation doses). Of the 16 homes we visited which had paracetamol and/or ibuprofen for their child, nine had errors with these medicines. Many of these errors were under dose errors, where parents reported administering a dose of medication which was too low for the child's weight. These children suffered when pain was undertreated.

Errors in children with seizure disorders

The most common types of errors among children with seizure disorders were administration errors, usually caused by confusion about changes in dose, and the use of expired medicines. For example, a parent was told to increase the dose of an antiepileptic medicine, due to increased frequency of seizures. On the home visit, the parents were still administering the lower previous dose. Expired medicines were usually PRN (as needed) medicines where the parent did not notice that the medicine had expired.

Errors caused by complex communication failures

Parents were not aware of 29 (47%) of the errors we detected on home visits. Physicians were not aware of 51 (83%) of the errors we detected. Parents would sometimes alter the technique for administration from what the physician told them to do because they did not have the proper equipment at home or because the technique the physician told them to follow did not work at home. For example, a parent used a 60 ml syringe to measure and administer 5 ml of a seizure medicine. This alteration would result in errors that parents were not aware of and so did not mention to the physician. Most children were administered medicines by multiple people; communication between caregivers at home was prone to error. For example, a father was told to increase his child's seizure medicines due to a high frequency of seizures. The father never told the mother to do this. She administered the medicine at the previous lower dose. In another example, a mother gave the once daily dose of a medication to a baby then went to work. The father did not realise that the mother had already given the dose and gave it again himself before bringing the baby to day care.

Labels which do not reflect the current dosing schedule

During home visits, we reviewed 11 medications where the label did not reflect how the medication was being used in the home because the dose was increased or decreased. This put the patient at risk for two types of errors: not following the label when it was correct and following the label when it was not correct.


We found previously undescribed types of medication errors in the homes of children with chronic conditions taking daily medicines. Home visits highlight the complexity of the home medication use process and numerous ways things can go wrong, including medication preparation problems, communication failures in the doctor's office and at home, and doctors largely unaware of the problem. Approaches to prevention should target systems-based failures which contribute to home medication errors and improvements in communication.

Comparing our findings to existing literature is challenging because this is the first study, to our knowledge, designed to systematically describe the breadth and depth of medication errors which occur in the home. Several studies have focused on mistakes measuring liquid medicines and interventions to prevent these mistakes.2 21,,24 None of the administration errors we found on home visits were simple mistakes measuring liquid medicines.

Many types of errors we found were previously undescribed, including complex communication failures between parents and providers. Communication failures also occurred between multiple caregivers at home and there were problems handing over information, such as the fact that the medication was already given, leading to double doses. In the inpatient setting, support tools have been used to prevent errors in hand-overs between nurses and residents at change of shift.25 26 Similar interventions may support communication at home.

We chose to visit children with sickle cell disease or seizures, but many of the errors described here may occur in children with other chronic conditions. For example, labelling issues related to frequent changes in medication doses, and resulting errors may occur in children or in adults with conditions where doses change frequently. Some children with chronic conditions may experience additional types of errors not found in our study.

While this is the first study to systematically describe the range of types of home medication errors, combining medical record review with in-home direct observation, any study design has limitations. Some clinically serious errors may not be detected during the home visit, because the error would result in the child leaving the home to go to the hospital. However, we did discover several insidious errors associated with patient injury which other methods may have missed. Second, our presence in the home may have altered the patient or parent's behaviour (Hawthorne effect).27 In the home, parents may have been more careful to avoid making mistakes or may not have reported missed doses, which would cause us to underestimate rates of errors. The timing of the visits (a quarter in the morning, the rest in the afternoon and evening) may have affected the types of errors detected. Third, generalisability of this single site study, especially with the experienced parents we visited, may be an issue. Future multisite research may improve the generalisability of our findings.

What are the implications of these findings for clinical practice? Parents should be made aware of the importance of following medication use instructions exactly as given. Having written instructions and other support materials at home may prevent errors, particularly when medication doses are adjusted frequently and the bottle label may not have current dosing instructions. Provider inquiry about problems in home medication use should be routine for children on daily medicines.

Our findings indicate a need for research on interventions to prevent mistakes at home. More efficient error detection methods will be needed to evaluate interventions. If interventions target specific types of errors, error detection methods could target that specific type of error without visiting the home.

We found previously undescribed types of medication errors occurring at a high rate in the homes of children with seizure disorders and sickle cell disease. Systems for medication use at home are complex and error prone, perhaps even more so than systems in the hospital. While tremendous time and resources are spent preventing inpatient medication errors, our findings indicate that many dangerous errors occur in the home, the setting where most children take their medicines and that these errors are largely unnoticed.


View Abstract


  • Funding KEW was supported by a Robert Wood Johnson Physician Faculty Scholar award.

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the University of Massachusetts Medical School Human Subjects IRB.

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

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.