A cross-sectional study based on an online survey was conducted in Spain. The field study was conducted between April to June 2022. This manuscript has been built following the CHERRIES checklist (28).

This project obtained approval from the Ethics Committee board of Sant Joan d’Alacant University Hospital in December 2021 (project CODE: 21/063) and was registered by ClinicalTrials (ref. NCT05247801). The information provided to study participants emphasized that their responses would be kept confidential, participation was voluntary, and informed consent was requested upon recruitment. Given the sensitivity of information regarding errors made in the home, all personal data was excluded from the responses, and only aggregated data was used to ensure confidentiality.

An informal caregiver was defined as any person responsible for administering care and prescribed medication to the adult at home, including family members or individuals hired by the recipient or relatives.

Errors were defined as all avoidable and involuntary events related to actions/omissions of medication that may imply or not harm (29).

Training encompasses various methods and channels for information dissemination, such as written instructions, instructional videos, and observing demonstrations by healthcare professionals. We define training broadly to include any form of passive or active learning provided to informal caregivers. Notably, hours of caregiving experience alone were not considered formal training.

All participants were aged 18 years or older and were actively caring for a person at that moment and for at least during the previous year, providing care and medications. Participants were stratified into two groups: qualified and non-qualified caregivers based on their self-reported training. We set a threshold of 20 h of training on techniques, strategies, and resources for caregiving duties. If caregivers reported at least this amount of training, they were considered qualified. This threshold aligns with the minimum limit for training courses in health schools. Health and care qualified professionals were not included. The time taken to respond was not an exclusion criterion since the respondents’ age might influence this aspect. The participants’ wish to withhold demographic data to ensure their anonymity was respected.

A required sample size of 384 caregivers in each group (a total of 768) was calculated based on the formula for estimating the proportion of occurrence of a specific event in an infinite population. Parameters were established to achieve a precision of 5% in the estimation of a proportion using a 95% two-sided normal approximation confidence interval. It was assumed, based on previous studies conducted in Spain (13), that the expected proportion is 50%. Supplementary File 1 provides more information about the calculation of the sample size performed.

The survey was announced, and recruitment was encouraged through invitations sent to all subscribers of a national database of caregivers registered with companies in the care economy sector and national caregiver associations. Given that males represent a small proportion of caregivers, their representation was artificially increased to 30% by inviting 3 males for every 10 females. This adjustment was implemented to ensure the validity of the data and to enable meaningful comparisons between genders, ensuring that male caregivers were sufficiently represented in the analysis. During the two-month recruitment frame, all received three times a message with study purpose and the invitation to respond.

The survey content was formulated based on the researchers’ prior experience (19, 30), a review of relevant literature, and online discussions with three experts in health psychology, pharmacy, and public health. These experts contributed to a better understanding of informal caregivers’ perspectives during group sessions, helping to identify relevant issues that needed to be explored and providing their clinical experience. The readability of the questions and the suitability of the response scales were ensured with the collaboration of informal caregivers and by incorporating insights from previous studies.

These discussions aimed to synthesize their collective expertise on patient safety, medication, and caregiving errors. To refine the survey, two members of caregivers’ associations and five additional experts conducted an online pre-reading, providing feedback on its clarity, completeness, structure, and relevance. After incorporating their insights, the survey underwent evaluation for readability, completion time, and content appropriateness before dissemination to respondents.

This online form was hosted on a plain website created to allocate surveys, owned by the research group¹ (31), where participants’ responses were automatically entered into the study database. The platform prevented multiple responses from the same IP address, but no cookies were used. Participants did not receive any incentives for participating. We did not track data regarding survey link email opening or clicks from the central server.

It comprised fixed multiple-choice and open-ended questions designed to assess: (a) care recipients’ health status, (b) safety measures to prevent errors and description of medication errors at home, and (c) caregivers’ experiences. The survey included a total of 49 items: 11 yes/no questions, 19 questions with a Likert scale structure (never during last year, once per year, once per month, twice a month, once a week, more than twice a week), 16 multiple choice questions, and three open-ended questions.

In the first part of the survey, caregivers were asked about their knowledge regarding medication indications, dosages, or storing conditions, primarily using yes/no questions. Regarding medication errors, we examined with Likert scale statements the probability of: (a) administering an incorrect dose of different medication formulations (e.g., eye drops); (b) administering a wrong medication due to a similar appearance; (c) duplicating or missing a dose and the underlying causes; (d) not following providers’ indications regarding treatment schedule or duration; (e) giving expired medication; and (f) not following medication manipulation or conservation recommendations. Each category was displayed on a single page. Some questions were conditionally displayed according to participants’ responses and respondents were allowed to change answers before submitting the responses (see Supplementary File 2 for survey content).

All the information presented is derived from data self-reported by the participant caregivers. The ELIXIR Research Data Management Kit (RDMkit) has been used as a guide to assure lawfulness, fairness and transparency, limitation and minimization, accuracy, storage limitation and integrity, and confidentiality on the website where this survey is hosted. The data preservation and curation were covered by FISABIO using the free and safe data storage, operated by the institution. All data from this study will be made available upon request.

First, the quality of data was analyzed for its suitability for the intended purpose. Records of caregivers who did not respond to at least 80% of the questions, who did not adequately complete the mandatory sections, or who did not provide care or medication were excluded.

After data curation, quantitative results were reported as mean and standard deviation or median and interquartile range for medication data. The total number of unintentional errors was calculated based on the self-reported frequency by caregivers over the past year. Caregivers reported the number of errors made in the past year, which were extrapolated to an annual scale using the following conversions: 1 error per year was kept as 1 error, 1 error per month was converted to 12 annual errors, 2 errors per month to 24 annual errors, 1 error per week to 54 annual errors, and more than 1 error per week was estimated at 104 annual errors. This extrapolation allowed for the standardization of responses into an annualized metric. This calculation was also performed for each type of medication administration route (e.g., inhalers, eye drops). Comparisons were made using either Fischer’s Exact Test or Chi-Square Test, with adjustments for variable dependency and significance. Bivariate statistics were used to explore differences in the impact of training, sex, relationship to the care recipient, full-time dedication, and the use of external aids for medication management. A Poisson regression model was utilized to analyze factors associated with the frequency of self-reported avoidable medication errors over the past year. Cases with missing values were excluded from the regression analysis to ensure data integrity. To mitigate confounding bias, interactions between caregiver age and gender were considered, as these factors can influence the likelihood of errors. This approach allows to determine whether the observed differences are truly attributable to caregiver training or rather to their experience. Additionally, comparisons were conducted by segmenting the sample on the reported level of training to examine the error frequency according to variables such as caregiver sex, relationship to the care recipient, full-time dedication, and use of external aids for medication management.

Qualitative variables were presented as percentages after grouping responses into categories based on their similarity. Two researchers (EGH and PB) performed the grouping, with a third researcher (JM) participating in case of doubts regarding the correct classification of participants’ responses. Open-ended questions were analyzed considering the frequency of mentioned mistakes and calculating the percentage of responses for each type of error. Responses were condensed, and the most frequently mentioned topics were summarized.

Data statistical analysis was performed using SPSS es 28.0.0.0 and RStudio Desktop 2022.02.3. A significant difference was considered when the p value was <0.05.

A total of 747 caregivers responded to the online survey. Sixty-two were excluded due to an insufficient number of questions answered. Consequently, 685 participants were included, achieving a response rate of 89.2% (685 out of 768 expected). Among the included participants, 530 were women (87.5%; Table 1). Qualified caregivers administered a mean of 8.7 different medications (95% CI 4.4–12.9) per day, whereas non-qualified caregivers administered 6.4 different medications (95% CI 6.0–6.8) per day. When comparing qualified and non-qualified caregivers’ groups, no significant differences were found in terms of sex (p = 0.999), age (p = 0.546), nationality (p = 0.258), co-existence with other caregivers (p = 0.073), full-time dedication (p = 0.303), or use of external aids for medication management (p = 0.504). However, a greater proportion of non-qualified caregivers were care recipients’ relatives compared to the qualified caregivers (p < 0.001), and more qualified caregivers were found to care for multiple recipients simultaneously (professionals 43.2%) and non 40.5% (p = 0.585). There was no significant association between recipients taking more than 10 medications per day and being primarily cared for by women (18.1% of care provided by women and 21.1% by men, p = 0.528).

Pills, capsules, or tablets were the most common form of medication administration, followed by injectables (Table 1). The frequency of presentations administered by both caregivers’ groups (qualified and non-qualified) was similar (p = 0.504).

A total of 9,236 medication errors were self-reported by the participants over the last year, representing an average of 13.5 (SD 38.2, 95% CI 10.5–16.5) errors per caregiver per year. The most frequent errors were related to dosage administration, followed by errors related to the administration schedule (Table 2). The mean of self-reported errors with injectables per informal caregiver per year was higher than with pills, inhalers, syrups, or eye drops (Table 2).

The different factors that could be associated with the frequency of medication errors by informal caregivers are shown in Table 3, along with the average frequency of errors per caregiver per year. Receiving a higher number of training hours (qualified caregivers) also tended to decrease the probability of errors by 21 times. Being aware of the number of medications the recipient takes contributed to a 65% reduction in the probability of medication errors occurring. Similarly, being a full-time caregiver reduced the probability by 40%, and being the sole caregiver for the individual by 20%. In the same way, not using external aid for medication management was associated with a 20% higher probability of making medication errors. There is also an interaction between age and training (p ≤ 0.001), showing that for qualified caregivers, the probability of making errors decreases by 10% as they age. In contrast, being a direct relative of the care recipient was associated with a 30% higher probability of errors. Being male or female has no direct relation to making medication errors, but the interaction between sex and being a qualified caregiver does show a decrease (p ≤ 0.001), reducing the probability by 30% if the caregiver is a qualified woman.

Table 4 presents the types and frequency of medication errors reported, categorized by caregiver qualification concerning sex, direct relative relationship, full-time caregiving status, and use of external aids for medication management.

Both groups of caregivers predominantly faced challenges in dealing with care recipients’ swallowing problems, dysphagia, and providing an adequate texture accordingly (Table 5). Secondly, coordinating with relatives, sharing information, and agreeing on care procedures were reported as significant sources of stress and difficult emotions. Additionally, disorder or lack of specific instructions about medication and care was identified as a source of errors.

This study adds to the existing knowledge by examining caregiver factors that influence medication and caregiving errors in home settings. Factors such as training, sex, use of external aids, being a relative of the recipient, or dedicating full-time caregiving were considered. This is the first European study analyzing patient safety information from informal caregivers. The study involved middle-aged women caregivers attending to recipients of polypharmacy, mirroring the typical profile of informal caregivers in Spain, where 84–89% are mostly women, average age of 53, with 44% being housewives and 60% having a basic education level (32).

The results indicate that more training positively impacts medication safety. Medication errors increased with less than 20 h of training, when caregivers were relatives, especially males, and when no external aids for medication management were used. Informal caregivers with other occupations committed more medication errors, while full-time dedication to care led to fewer errors. Sharing caregiving responsibilities across multiple households also posed a higher risk of errors. Since the sample in this study mainly consisted of individuals whose primary dedication is caring for others at home, it is possible that the caregiving experience also contributed to safer medication use.

These data confirm that the use of medications at home is not free from unintentional errors that can have consequences for patients (33). This study shows that medication errors rise with more complex treatment regimens (15). Injected medication errors were more frequent among non-qualified caregivers, who also had higher confusion rates with pill medications due to a lack of strategies to avoid confusion. As it has been described elsewhere, similarities in naming or spelling (sound-alike), and/or physical appearance or packaging (look-alike) are sources of confusion (34). Increasing awareness is everybody’s responsibility to decrease potential harms derived from these errors has been confirmed.

The findings align with those from other Western countries, where family members provide the majority of care. However, this study emphasizes that poor caregiver training is strongly linked to higher medication error rates (35). Our results are in line with previous publications where caregivers with limited language proficiency or health literacy tend to make more dosing errors (up to 83.1%), with the greatest odds in the group of non-qualified caregivers (36). Unqualified male caregivers made more errors in medication administration compared to females, contrasting with studies on nurses, where medication error rates were higher in the female group (37). Non-qualified male caregivers made more errors, including dosage and schedule adherence issues (38). Few authors have addressed these problems using a gender perspective. In studies on children’s informal caregivers, being male was associated with appropriate medication preparation (39). However, in light of these data, studies reporting medication errors should incorporate this novel analytic approach, where the caregiver’s gender is considered (30), or even specifically reported (40). This seems relevant if we consider that in Europe, data suggest that the life expectancy gap between men and women has narrowed, leading to a growing number of men taking on the caregiver role for their partners (41). This trend should be considered alongside studies indicating that many men feel a sense of diminished masculinity when assuming caregiving roles (42, 43). This perception, combined with a general lack of caregiving experience over the years, may result in less attention to available informational resources, potentially impacting the quality of care provided and influencing these outcomes. Further research could focus on these aspects to elucidate the reasons behind these differences, which may pose a greater risk for female care recipients than for male ones.

The literature reports that informal caregivers (for both the elderly and infants) make avoidable errors such as misdosing injected medicine and administering medication out of schedule (44, 45). This study confirms these initial findings and provides new data. Being relatives of the recipients, qualified caregivers showed increased nebulizer dosing errors, eye drop dose errors, and decreased compliance with treatment duration or storing conditions. Qualified caregivers with another occupation tended to commit more dosing errors with injected medication, confusion with oral medication, and duplicate administration. Being a full-time informal caregiver reduced errors with syrups, scheduling, and medication manipulation. Previous descriptions indicate that a person’s working status negatively impacts their error rates, especially missed doses (46).

There are strategies to reduce medication errors, such as using dispensing devices or labeling the dosage on the package (47). In this study, external aids (e.g., alarms, mobile apps, or medicine organizers) helped reduce errors with inhalers but did not always prevent confusion or dosage errors. Those cautious with medication use and aware of potential risks are more likely to use these aids, acknowledging that errors can still occur. This is consistent with previous findings (30). While mobile apps generally meet caregivers’ needs, increasing awareness and providing external strategies to reduce caregiving issues is crucial.

Regarding caregiving error causes and challenges, caregivers experienced problems with food textures, recipients with swallowing problems and dysphagia, and poor communication among caregivers and family members (36). Regardless of their training, both groups reported these issues. To reduce care errors in chronic patients and assess individual needs, some tools have been developed in Spain (48). Researchers have raised concerns about caregivers’ limited knowledge of dysphagia and related care actions (49), which is still an active research field for medication errors in hospital settings, where providers are struggling with dose (50) and formula (51) adaptation to provide a suitable treatment in patients with dysphagia. Enhancing effective communication among everyone involved in recipients’ care is essential, as many errors stem from unfamiliarity with medication or care processes (52).

This study provides insights for designing interventions to prevent avoidable mistakes at home and create safer environments. For example, implement tailored training for informal caregivers, especially focused on medication administration (oral pills and injections), recognizing common errors, and understanding medication schedules. Managing food textures and dysphagia or avoid falling has proven to be a common challenge, and access to visual information on how to address these issues could be one of the priorities.

The training approach for informal caregivers should be specifically tailored to address the gaps among male elderly caregivers and those with external occupations who dedicated a limited time for caregiving, as the study highlights their increased likelihood of committing errors.

Create public resources addressing the emotional impact of caregiving and teaching caregivers coping mechanisms to target and reduce stress and design gender-sensitive support programs recognizing the specific challenges faced by male caregivers. Policy makers, social workers and clinicians should consider the factors analyzed in this study when designing more personalized information and instructions that informal caregivers need to carry out their tasks at home. This need for personalized training aligns with what has been stated in recent publications, where the increasing decline in the caregiver-to-patient ratio is already supported by artificial intelligence, helping to detect hazards or abnormal patterns in the living ambient (53).

The findings can be extended to other European countries, contributing to a detailed map of home care needs and helping policymakers develop resources and support materials to enhance safety at home.

The study relied on self-reported data, which may include biases and inaccuracies. Since we relied on participants’ recollections, their responses may not fully reflect reality and may also be subject to social desirability bias. The study is not free from the risk of declarative bias, as some respondents might be more honest than others (e.g., struggling to admit their errors). Furthermore, past caregivers’ errors were self-reported, so the extent to which these self-reported errors correlated with actual errors is unknown. The recall of error occurrences may differ between recent episodes and those further in the past. Additionally, caregiver-to-caregiver communication has been pointed out as a contributory factor to errors (23). When interpreting the results, it is important to consider that the group of non-qualified caregivers had higher rates of care recipients’ relatives than the qualified group. Also, since men are underrepresented among informal caregivers, we oversampled male participants to facilitate gender comparisons. However, this may limit the study’s ability to generalize the findings, but will contribute to expanding knowledge in the remaining gaps of the field.

The sample size was not large enough for high statistical power, and there was a bias in the information shared by multiple caregivers. The study’s cross-sectional nature limits the analysis of cause-and-effect relationships. Social desirability bias and declarative bias may have affected the accuracy of reported errors. The study also did not account for whether support devices were used correctly. The methodology used does not allow for causal relationships to be established, as it is lacking a longitudinal follow-up.

Future studies should explore the frequency of medication mistakes at home beyond self-reporting limitations. Analyzing the impact of training content, intensity, and duration on caregiving safety is necessary. Research should also address the emotional impact of avoidable errors on caregivers, particularly family members, and the direct or indirect costs of these errors on the healthcare system. Future efforts should aim to gain a more realistic view of caregivers’ situations, as there may be selection bias in this study, especially when caregivers share their duties with others.

Further research could investigate if mistakes at home have a similar impact on informal caregivers as adverse events in professional healthcare settings. Collaborating with caregiver and patient associations and employing citizen science approaches could overcome barriers in future studies.