Definitions, Data Sources and Search Strategy

We define a MAE as ‘any deviation between the medication prescribed and that administered’[26, 27]. We take the term ‘prescribed’ to include verbal or written prescriptions and instructions. Our definition of a home carer is any person that provides care and assists in the living activities of patients within their home. This includes both formal paid caregivers working for a healthcare or social service organisation (e.g. agency carers or community nurses) and informal caregivers who are not paid (e.g. relatives or friends of the care recipient).

A review of literature published between 1 Jan 1946 and 23 Sep 2013 was carried out in the online databases EMBASE, MEDLINE, PSYCHINFO, COCHRANE and the Cumulative Index to Nursing and Allied Health Literature (CINAHL). The search strategy involved three conceptual facets: Medication errors/safety, the domiciliary setting, and carer involvement. Care recipients could be adults or children.

To achieve a balance of specificity and sensitivity, multiple iterations and combinations of search terms were tested. Terms were omitted if the sensitivity/specificity were compromised, e.g. the term ‘outpatient’ resulted in a very high number of irrelevant articles. Included terms were based on their use in the literature; e.g. ‘adverse drug event’ is commonly used to include drug errors[7]. The most relevant Medical Subject Heading (MeSH®) term ‘Medication Errors’ was included. For carer involvement, we included healthcare professionals that are most likely to manage medications in the home. S1–S3 Tables present the final search strategies used. The references of the included articles were cross-referenced for missed articles.

Study Selection

Titles and abstracts were screened against a set of inclusion criteria (S4 Table). Studies selected had to meet all of the following criteria: (a) errors occurred in the home (b) carers were responsible for the delivery of medication and (c) empirical data were provided. We excluded papers describing a single case study, such as an MAE legal case. However, we included papers describing multiple case studies and qualitative studies where there was more than one care-recipient participant. Papers that did not report data for carer-caused MAEs separately to other medication errors or from other administrators (e.g. patients themselves) were excluded, unless over 80% of combined data related to carers, in which case we contacted the author to provide additional details[28–30].

One reviewer (AP) screened titles and abstracts of all articles while a second reviewer (SG) blind-screened a random sample of 10%. Discrepancies were discussed with a third reviewer (BDF) until consensus reached. There was a high percentage of inter-reviewer agreement (90%) between AP and SG and Cohen's kappa suggested good inter-rater reliability (K = 0.73, P = 0.00)[31]. The remaining full text papers were screened by one reviewer (AP), and another randomly assigned 10% was blind-reviewed (SG). There was a good inter-rater reliability (K = 0.95, P = 0.00) between the reviewers, and AP, SG and BDF discussed all articles identified as borderline inclusion/exclusion. All included articles were checked by at least two of the three reviewers to confirm that they met the inclusion criteria. All non-English language papers were reviewed by native language speakers who also all (apart from one) had experience of systematic reviews. If they met the inclusion criteria, the articles were translated by professional translators.

Data Extraction and Synthesis, and Quality Assessment

Relevant data were extracted from the included articles using a standardised template comprising study details (e.g. study design), medication-specific information (e.g. drug/administration type), outcome information (causative/preventative), sample details (e.g. sample size), and MAE details (e.g. error types). One reviewer (AP) extracted the data and assessed the quality of all articles, while another (SG) extracted and assessed a 10% random sample, with moderate agreement (K = 0.65, P = 0.00).

A quality appraisal tool designed for a range of study designs was used to assess study quality[32]. Each article was scored (on a range 0–2) on each of 10 criteria for qualitative studies, 14 for quantitative studies and all 24 for mixed methods. The scores for each article were added together then divided by the possible total score to give an overall percentage. Additional MAE-specific quality assessment questions were also used, based on similar systematic review assessments[33, 34]. These focused on whether MAEs and their sub-categories were specified/defined, method of MAE detection appropriateness of data collection on MAE causes, and any error validation used.

Since we anticipated that studies would be too heterogeneous to allow for meta-analysis, we used a narrative synthesis to retain original meanings of findings[35]. We inductively identified and grouped contributory factors into a new framework. The review followed PRISMA guidelines[36]. Since we did not conduct a quantitative meta-analysis, we did not formally assess risk of bias but instead described possible sources of bias as part of our quality appraisal of included studies.

Study Samples, Methods and Settings

The majority of studies were carried out in the US[18, 28, 29, 37, 39, 40, 42, 43, 48, 49, 51, 52, 54, 57, 59, 61–65, 68]. The remainder were from Saudi Arabia[55], Spain[38], UK[50, 53], Canada[56], Israel[44, 58, 60], France[41, 47, 66], Australia[30, 45], China[46] and South Korea[67]. Only one studied formal carers, i.e. home health agency nurses[37]; all others investigated informal carers. Of the 35 that included informal carers, 15 included parents among other carers[18, 28–30, 39, 41, 42, 44, 52, 53, 55, 61, 62, 67, 68] and 18 studied only parents[38, 40, 43, 45–47, 49, 50, 54, 56–60, 63–66]. Moreover, the parent samples were mostly mothers, for example with over 90% mothers in 7 studies[28, 29, 53, 57, 61, 64, 65] and 78.3% in another[67]. The remaining two studies focused on older adult care-recipients who were mostly cared for by their adult children carers or by their spouses[48, 51]. Although many studies did not explicitly exclude informal carers who were friends of the care-recipient, and included them in their definition of caregiver[18], no studies focused solely or significantly on this type of carer.

The definition of a MAE was often not reported; in some cases this was because the study focus was wider and included any type of medication error. Those that defined a MAE classified it as deviation from prescriptions and/or instructions/accepted guidelines. For those investigating paracetamol dosing errors, a tolerance of 10-15mg per dose was most commonly used.

The majority of studies (28/36) were quantitative studies that used retrospective record review, questionnaires or structured closed question interviews, quantitative observations, or a combination of these. Three studies reported interventions[39, 45, 65], three were qualitative[48, 50, 51], and two used mixed-methods (qualitative and quantitative)[18, 37]. Most (25/36) studies included self-reports, either alone (15 studies) or together with other data collection methods (10 studies). The remaining studies did not use self-report and instead used more objective data collection methods, such as record review and observations.

Quality assessment scores ranged between 50–100%, with most (26/36) scoring above 80%. Common methodological issues were undefined error categories and convenience sampling. In some cases, validation of errors would have strengthened the study and those that used retrospective study or self-reported errors as their main outcome measure may suffer from additional bias.

While all studies focused on MAEs in the home, some (24) of the data collection took place elsewhere: in hospital (5)[39, 51, 54, 64, 68], specifically in emergency departments (9) [38, 40, 49, 55–57, 61, 62, 66], as well as medical centres (4)[44, 58, 63, 65], outpatient clinics (2)[29, 42], a home healthcare agency[37], a poisons centre[30] and other clinics/practices (2)[28, 43].

Types and Prevalence of MAE

The most common type of MAE was dosage errors, [28–30, 38–41, 44, 46, 47, 49, 51, 53–63, 65–68]. Omitted administration[28, 29, 37, 39, 50, 51, 66, 68] and wrong medication[30, 37, 43, 44, 46, 49, 61] were also common. Other types of MAEs were wrong time of administration[30, 46], wrong route of administration[30, 41, 49] wrong patient[41], inappropriate combination with other drugs[47], wrong preparation technique[37, 47], giving expired medication[29, 49], wrong formulation[41, 49], age-inappropriate drugs[41]; not washing equipment[47] and either not completing treatment[42, 43, 52] or not stopping treatment[48].

It is difficult to compare error rates across studies due to the differing methods used to identify errors. For errors not related to a specific medication group, the carer administration error rate ranged from 1.9 to 33% of medications administered[29, 42, 47, 52], 12% of carers administering medication[18], and 70% of patients receiving medication[28]. In addition, 7.8% (26) of carers reported that they had given an insufficient dose, 6.6% (22) reported that they had given an overdose and 5.4% (18) reported that they had given the wrong medicine[46]. A separate study evaluated the relative percentage of administration versus non-administration errors and showed 1.7% of medication errors and 22% of the near misses occurred at the administration stage[43].

A small number of studies endeavoured to assess the impact of medication errors. Three studies used clinical record review to determine the percentage of cases of poisoning caused by medication errors. One study found that 10% of cases of medication poisoning in children under seven were caused by dosage errors, although it is unclear whether all occurred at administration[38]. Another study found that in 15.9% cases of acute poisoning, including medication and other agents, there was an error in dosage of a medication and in 8.8% cases an incorrect drug was given by a member of the child's family[44]. A similar study found that therapeutic errors in children (<16yrs) reported to a poison centre were mostly administration errors by carers (80.9%, 397 out of 491), rather than the patients themselves[30]. A fourth study found only 1.6% of unintentional overdoses in 1–4 year olds had documentation of dosing errors on the part of the caregiver[40].

Causes of MAEs and Potential Contributory Factors

There were recurring factors cited across the studies, Table 2 presents the findings on how these affected MAEs. The most common risk factors identified were equipment issues and miscommunication or poor understanding of illness, instructions or calculations. Polypharmacy, storage, type of medication, carer education level, language, and dosage change were identified as factors across several studies. The factor with the most contradictory evidence was age of children care-recipient, with opposing results on whether older or younger children are at more risk of MAEs. The only other factor for which not all studies showed consistent findings was healthcare professional communication and carers’ understanding of instructions. Although nine of 12 studies found this to be significantly associated with MAEs, the others did not.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Causes and potential contributory factors of informal carers’ MAEs

https://doi.org/10.1371/journal.pone.0167204.t002

Interventions and Carers’ Activities to Prevent MAEs

Preventative activities by carers included planning medication schedules around meal/bedtime routines for caregivers with predictable schedules[51] and using a multi-compartment compliance aid[39]. Carers identified the need to be constantly vigilant in monitoring care-recipients’ conditions[51], and often detected errors; in one study 85% of MAEs were detected by a family member (compared with 12% by healthcare professionals)[41]. However, in 16% of 152 ADE cases, parents did not disclose their errors[42].

Only three intervention studies identified strategies to prevent carers’ MAEs. One was a randomised convenience sample of parents administering antibiotics that tested instructions, plus administration demonstration and (marked/non-marked) equipment for medication administration[65]. The medication administration demonstration along with a marked syringe at the correct dosage level improved the accuracy of dosage (100% accuracy) compared with prescriptions, syringe and demonstration alone (83% accuracy) or compared with prescription alongside verbal instructions alone (37% dose accuracy)[65]. The second study was a randomised controlled trial of 10 weekly lessons carried out in the parent’s home on child health and home safety[45]. The lessons significantly increased the parents’ health behaviours on how to use medicines alongside a significant increase in parent knowledge and skills in using medicines safely (compared with visits and current services alone)[45]. The third study implemented a hospital-to-home transitional care nurse coaching intervention for people managing complex medication regimens at home. The intervention comprised of interviews and medication reconciliation at the hospital and following hospital discharge, followed by a home visit to observe medication use. The study identified that 62.5% (10 out of 16 home visits) more discrepancies between medications prescribed and administered were detected at home visits than by telephone, and that coaching the caregiver responsible for the medication regimen was important in resolving such discrepancies[39].

Review and Research Strengths and Limitations

This is the first review to explore informal and formal carers’ MAEs at home. A strength is that there were no restrictions on article language, allowing for broad inclusion of the international literature. Additionally, the reviewers held a relevant mix of expertise on patient and medication-related safety, caregivers, and homecare. While the included studies were heterogeneous and did not allow for substantial quantitative comparison, there were clearly identifiable groupings of error types, prevalence and causes as well as clear research gaps that emerged from the narrative nature of the review.

Prevention of MAE was less easily identifiable than error causation, due to the potentially more ambiguous and broader range of preventative activities. This difficulty may have compromised the extent to which we identified evidence on carers preventing errors. The quality scores are subjective and limited by the quality assessment scale used and its restricted criteria. A major limitation was that the systematic review was only up to September 2013.

This review emphasises the need for more research on home care medication safety, particularly as medication-related models from the hospital setting have been deemed unsuitable for application in the home[77]. Our review has additionally discovered the gaps within the current literature. Most studies were based in the USA with mother carers, despite statistics showing that there are only 30% more females than males in caring roles in the USA[78]. More research is required on formal carers who frequently administer medications in the home, and informal carers who are friends, spouses or adult children of the care-recipient. Currently, we know very little about these carers or their medication-related activities and any additional risk factors. For instance, some carers may make more errors due to sensory deficits or cognitive problems[79, 80]. More research with formal caregivers would allow for a useful comparison and distinction between formal and informal carers, particularly in regards to potential differences in causes of MAEs and their implications for tailored interventions. Further investigation is also required to clarify whether younger or older children are at higher risk of MAEs and whether this is context-dependent. Studies would additionally benefit from random sampling strategies and validation of errors. The latter, in particular, would help minimise subjective biases resulting from self-reports, especially as parents do not always disclose their MAEs[42]. Despite not restricting the articles by the English language, there is not a full representation of countries researching this topic. Other countries may have their own specific set of barriers contributing to errors, such as poor access to up-to-date medicine information[81]. Preventative interventions are presently scarce.

The data collected from home and clinical settings were analysed together. However, observed demonstrations of administration in clinical settings may be different from the home setting, for example, there may be more distractions in either the home or clinical environment and equipment might have been available in clinics that is not available in the home. As found in an earlier systematic review on MAEs in hospitals[82], the MAE definitions and error categories are rarely reported, which hinders study comparison.

Implications for Clinical Practice

Specific actions for clinical practice could include interventions to improve communication between healthcare professionals and carers. This would involve consideration of carers’ understanding of the information, particularly on the prescription/medication labels, administration techniques, and on what to consider if the prescription changes, and the importance of storage. Carers could be encouraged to keep notes of this guidance to aid recall of instructions. Similarly, more detailed advice can be provided to carers on how they communicate with one another (e.g. inter-parental communication). This could also form part of more formal carer training and home visits would allow for more objective verification of appropriate administration. Carer calculations can be checked for accuracy, including those dependent on the weight of the care-recipient. Similarly, appropriate and inappropriate equipment (e.g. teaspoons) should be identified. Where possible, tailored equipment (e.g. with marked dosage lines) could be provided. Such consultations and home visits could further incorporate more questions regarding carer concerns (on medication management) and their mental well-being.

These actions should be considered particularly when the care-recipient is taking multiple medications or when there is likelihood of frequent dosage changes. Those administering ‘taken as needed’ medications should be particularly made aware of the importance of expiry dates. Extra care could also be taken for more complicated routes of administration and potentially with certain types of medication that may be more susceptible to error (e.g. cold/cough medicines), as well as for informal carers who are older, single, have lower educational levels, poorer health, or are non-native speakers.