OUP user menu

★ Editor's choice ★

The incidence of adverse events among home care patients

Nancy Sears , G. Ross Baker , Jan Barnsley , Sam Shortt
DOI: http://dx.doi.org/10.1093/intqhc/mzs075 16-28 First published online: 2 January 2013


Objective Incidence of adverse events (AEs) among home care patients and preventability ratings were estimated. Risk factors, AE types and factors associated with AEs were identified.

Design This study used a stratified, randomized sample of home care patients discharged in the fiscal year 2004/05. Trained nurse reviewers completed retrospective chart abstractions; charts for cases that were positive for screening criteria suggesting the presence of AEs were reviewed by trained physicians to determine the presence of and preventability of AEs.

Setting Three publicly funded home care programs in Ontario, Canada.

Main outcome measures Prevalence and types of AEs; ratings of preventability.

Results At least one screening criterion was positively identified in 286 (66.5%) of 430 cases. Physician reviewers identified 61 AEs in 55 (19.2%) of the 286 (12.8% of the 430) cases. The AE rate was 13.2 per 100 home care cases [95% confidence interval (CI): 10.4–16.6%, standard error 1.6%]. 32.7% (20 of 61 AEs) of the AEs were rated as having >50% probability of preventability; 6 deaths (10.9% of patients with an AE; 1.4% of all patients) occurred in AE-positive patients. The most common AEs were falls and adverse drug events.

Conclusions Providing health care through home care programs creates unintended harm to patients. The incidence rate of AEs of 13.2% suggests a significant number of home care patients experience AEs, one-third of which were considered preventable. Improvements in patient and informal caregiver education, skill development and clinical planning may be useful interventions to reduce AEs.

  • adverse events
  • patient safety
  • home care
  • community care
  • evaluation


Research in a number of countries has demonstrated that between 3 and 17% of hospital patients experience an adverse event (AE) [17]. Despite the growing use and complexity of services outside hospitals, there has been limited research on harm to patients in other health-care sectors. This study adapts the hospital research methods to assess the incidence of AEs within the home care population and the factors influencing exposure to these AEs. The three criteria definition of an AE for home care patients was adapted from that used in the US [3], Australian [6], UK [1], New Zealand [7] and Canadian [2], hospital studies. An adverse event is (i) ‘an unintended injury or complication (ii) which results in disability, death or increased use of health-care resources and (iii) is caused by health-care management’. Injuries included physical harm such as lacerations, bruises and fractures. Psycho-social injuries are more challenging to identify and evaluate, and were not included in this study. Since home care includes supervision and care provided by the patient's family and friends who assume responsibility to deliver health-care interventions that in hospital would be carried out by staff, the definition expands health-care management to also include health-care services provided by the patient and informal caregivers as well as those by health-care professionals.

Home care as a substitute for, or following, hospitalization is a growing component of Canadian health care [810]. There have been a few studies describing AEs in home care and the factors that contribute to them in Canada or elsewhere [1124]. Others studies have identified patients readmitted to hospital but these studies did not differentiate between patients that were and were not receiving home care [2527].

This paper reports the results of a study examining the prevalence of AEs within a population of home care patients in Ontario, Canada. Ontario has a regionalized home care program. In each region a community care access center (CCAC) case manages home care services including the provision of nursing care, physical and occupational therapies, speech-language pathology, dietetic and social work counseling, pharmaceuticals, medical equipment and supplies. Individualized care plans incorporate services from pharmacists, specialist and family medicine physicians, and community support services such as meals-on-wheels. The program aims to keep patients healthy and safe at home, avoid unnecessary hospitalizations and prevent premature admission to long-term care institutions [28].

The goals of the study were to identify the prevalence of AEs, the risk factors for home care patients associated with AEs, the types of events that occurred and the extent to which these events might be reduced by greater attention to patient safety in home care settings.


This study used the two phase retrospective chart review methods previously used in hospital AE studies in the USA, Australia, New Zealand, Britain, Denmark, Canada and elsewhere [1, 2, 4, 6, 7, 20]. Data were abstracted from patient-specific clinical case management charts compiled by regulated health and social work professionals responsible for the patient. These charts contain clinical case manager generated narratives describing all face-to-face, telephone, and paper-based and electronic correspondence with patients and their families, health-care professionals and services involved in each patient's care; weekly to monthly and ad hoc narrative summaries of direct care provision and patient progress/outcomes by direct care providers and services; and laboratory and other test results. Many regard AE detection through chart review as the gold standard [29], although methods such as direct observation for errors can yield larger numbers of AEs [30]. Self-reporting has recently been used to estimate the incidence of errors in general medical practice [31]. Nurse reviewers, using AE-sensitive criteria, screened charts abstracting data from the date of discharge back to the date of admission or 12 months earlier, whichever was shorter. Physician reviewers subsequently assessed whether AEs were present in the flagged charts and judged the degree of preventability of identified AEs. This method has moderate-to-good reliability [32].

Home care in Ontario is organized into regional programs that contract for services from multiple agencies. Three home care programs in South-eastern Ontario covering a geographically contiguous urban, suburban and rural population of 442 800 over 19 473 km2 [33] were selected on a convenience basis as the study site. These sites use the Minimum Data Set Resident Assessment Instrument for Home Care (MDS RAI-HC) for patient documentation. The MDS-RAI covers multiple patient health domains including function, cognition, mood, behavior, vision, communication, nutrition and symptoms and has been used in prior research studies [34]. The sampling frame included 7467 discharged home care patients who received care from nurses between 1 April 2004 and 31 March 2005. Home care programs in Ontario admit patients from all age groups (birth to 100+ years) and include patients with care requirements due to acute and chronic medical conditions, surgical recuperations, pre- and perinatal care, palliative care, disability/rehabilitation needs and trauma. Typically, 46% of Ontario's home care service recipients have chronic conditions or health-care needs requiring care over a long period of time. The remaining 54% require shorter term care while recovering from hospital-based care or injury [28]. No care groups or age groups were excluded from the study population.

Based on an estimate of a 5% error margin and 95% confidence level, with an estimated AE incidence of 12%, a sample size of 159 cases (Roasoft, Inc. Sample Size Calculator) would be sufficient to establish the incidence of AEs. However, oversampling was used to allow the identification of a sufficient number of AEs to permit classification by types. A stratified (by site), randomized sample of 450 patients [α: 0.05, confidence interval (CI) 95%, margin of error 2.91%] was selected. No substitutions were made. The sampling cascade is summarized in Fig. 1.

Nurses familiar with home care and specifically trained to systematically collect standardized data reviewed all charts for the presence of one or more screening criteria (Table 1). This review was done to identify those patients at risk of an AE and to enable physician reviewers to focus on these charts. This approach has also been used in the hospital studies.

View this table:
Table 1

Phase I screening

The screening criteria were adapted from those applied in the Canadian hospital study [2], modified and validated through a Delphi technique by an expert panel with expertise in home care case management, home care nursing, family medicine and AE research. Screening criteria were pre-tested for content validity and data availability on a sample (n = 40) of home care charts.

Cases positive for one or more of the screening criteria were reviewed by physicians who examined chart entries from the time of admission to home care or for a period starting 12 months prior to the discharge date, whichever was shorter. Three criteria were necessary to determine if an AE was present. These were (i) the presence of injury to a patient, (ii) that required at least the use of additional health-care resources and (iii) where the injury was assessed as resulting from the care received rather than from the underlying disease or condition. The physician reviewers were asked to identify the extent to which AEs were the result of health-care management (or its omission) rather than disease progression. Preventability of AEs was also scored by physicians on a 1–6 scale as in the hospital AE studies [2, 3, 6]. Key steps taken to enhance the reliability of these assessments included limiting the number of physician data collectors to two, training reviewers on the use of the tools and encouraging discussion of complex events at the point of data collection. Computerized web-based data collection forms, direct electronic data entry and electronic transfer of data reduced the potential for missing or lost data. Inter-rater reliability was assessed by comparing the results with the assessment of a third physician reviewer on 58 (20%) randomly selected charts, 50% originally rated as positive and 50% negative of the 286 charts reviewed by the original physician reviewers.

Multivariate analysis on the RAI-HC data was carried out to determine the factors associated with the occurrence of AEs. Five potential confounding factors were identified from review of the literature as associated with both the occurrence of AEs and common risk factors. Three factors (age, living alone or with others, and communication difficulties due to cognitive causes) were used in a logistical model to determine the impact of other risk factors. The other two (discharge status and language) were removed from the model due to high association with other factors or low frequency in the database.


Of the 430 charts in the sample, 286 (66.5%) were rated positive for one or more of the standardized screening criteria. The most common screening criteria were unplanned admission to acute care hospital (33.9%), unplanned visit to hospital emergency department (33.7%), patient injury, harm, trauma or complication arising during the index admission (21.3%); new problem/diagnosis noted during the index admission (20.4%) and recognized actual or potential environmental risks, including patient behavioral and physical environmental factors (17.6%) (Table 1).

Physicians reviewed the 286 positively screened charts and identified 61 AEs across 55 patients (12.8% of patients sampled). Four (7.2% of 55) patients experienced 2 AEs and 1 (1.8% of 55) experienced 3 AEs. After weighting the three-site stratified sample frame to account for the number of patients at each site, the overall AE rate was 13.2% [95% CI: 10.4–16.6%, standard error (SE) 1.6%].

There were a number of patient factors significantly associated with AEs. Logistical analysis was done in two stages. In the first stage three factors (age >65, living alone and communication difficulties due to cognitive causes) were assessed in relationship with the occurrence of AEs and found to be significantly associated. These factors are likely to increase the vulnerability of patients to AEs. Controlling for these three factors in a logistical model, the remaining risk factors were assessed. Based on this analysis, a number of risk factors were identified including history of falls, use of psychotropic medications, depression, anxiety and anger, social isolation, care for several instrumental activities of daily living (IADL) and activities of daily living (ADL), and a number of primary diagnoses or co-morbidities (cerebral vascular accidents, peripheral vascular disease, multiple sclerosis, cancer, Parkinson's disease and fractures other than hip) which had significant risks for AEs (Table 2).

View this table:
Table 2

Association of patient factors with AEs controlling for age, living alone and communication difficulties due to cognitive causes

Since home care includes care provided by family and friends and the patients themselves, as well as home care staff, physician reviewers assessed whether the AEs were associated with care provided by health professionals and home care workers, informal caregivers and the patients themselves. Decisions and care provided by health professionals/home care workers were rated as being likely (>50%), certain or virtually certain to have contributed to the AE in 29.5% (95% CI: 19.0–40.6%) of patients with AEs. Informal caregiving by family members or friends was rated as being likely (>50%) to virtually certain to have contributed to the AE in 27.9% (95% CI: 17.8–40.8%) of patients with AEs. Self-care by patients was rated as having contributed to an AE in 52.6% (95% CI: 40.1–64.8%) of patients with AEs. In several instances, multiple contributing sources (health-care professionals, informal caregivers and self-care) co-existed for individual AEs. There was no significant association between the type of caregiver and the level of patient impairment (harm) resulting from the AE. Relationships between the source of contribution and injury (categorized by the level of patient impairment) are shown in Table 3.

View this table:
Table 3

Sources contributing to AEs and the resulting level of patient impairment

Approximately one-quarter of AEs resulted in only slight impairment, but half (31 of 61 AEs) resulted in moderate [recovery expected in 1+ months to permanent impairment with ≤50% disability) months] or serious (permanent impairment with >50% disability or death). There were six deaths (two related to medication errors; one due to sepsis; one related to treatment of congestive heart failure; one gastrointestinal bleed and one due to general, prolonged untreated deterioration not otherwise specified). Overall, 32.7% of AEs (20 AEs) were rated as preventable [that is, where preventability was rated as being >50% (Table 4)]. Physician reviewers found no difference in the preventability of AEs that were linked to health care provided by formal caregivers, informal caregivers or by clients themselves.

View this table:
Table 4

AE preventability rating by physician reviewers

A comprehensive taxonomy of AEs in home care has not been developed. AEs in this study were categorized into 11 types; 12 types if falls are subdivided by type of injury (Table 5) representing general descriptions of AEs from previous studies [1115, 2224, 35]. Falls with injuries accounted for 24.6% of AEs; 40% of reported falls (9.8% of AEs) resulted in fractures and 60% (14.8% of AEs) in lacerations or tissue injuries. 16.4% of AEs were medication errors, two resulting in death. In total, six AE-associated deaths (1.4% of all patients) occurred; self- or informal caregiving contributed to four of these deaths, including the two which were medication related. Of the 61 AEs, 11 required that the client be transferred to a hospital emergency department and another 15 required that the client be admitted to hospital.

View this table:
Table 5

AEs by type

The kappa statistics for the measurement of agreement among physician reviewers were substantial: determination of whether an injury had occurred, 0.811 (95% CI: 0.689–0.933); determination of unexpected death, 0.772 (95% CI: 0.472–1.072); determination of whether health-care/self care/informal caregiving contributed to the injury, 0.791 (95% CI: 0.567–1.012); overall identification of an AE, 0.828 (95% CI: 0.683–0.972) and determination of preventability, 0.714 (95% CI: 0.458–0.971). These inter-rater reliability (kappa) kappa values are equal to or better than results using similar chart review methods in hospital settings [1, 2, 4, 6].


This study focused on a stratified, random sample of patients in Ontario, Canada who had received home care, including nursing services, and were discharged in 2004/05. At least one screening criterion for an AE was positively identified in 286 (66.5%) of 430 cases. Sixty-one AEs were identified in 55 (19.2%) of the 286 cases selected for review, which were 12.8% of the 430 cases. The AE rate was 13.2 per 100 home care cases (95%, CI: 10.4%–16.6%, SE 1.6%). Falls resulting in lacerations or soft tissue injuries or fractures represented 25% of AEs, while adverse drug events accounted for 16% of AEs. Almost one-third (32.7%) of the AEs had a high preventability rating (four or higher on a 6-point preventability scale). The type of caregiver did not influence the level of severity of impairment incurred from an AE. Most AEs resulted in impairment or disability that resolved in <1 year; 9.9% (CI: 95% 4.5%, 20.6%) of patients who experienced an AE (1.4% of all home care patients studied) later died.

Most previous home care AE research focuses on adverse medication events [1114]. One study reported that one-third of home care patients had evidence of a potential medication problem [14]; another described look-alike/sound-alike medications as a latent error source [12]. An Australian study of hospital-in-the-home patients reports 1.7% of these patients experienced AEs [15]. Studies of incident reports in a home care agency in the USA found the most frequently reported AEs to be pressure ulcers (1.9%), urinary tract infections (1.4%), falls (1.4%), wound infections (1.4%) and unexplained deaths (1.0%). These data need to be interpreted cautiously since incident reporting has frequently been found to underestimate the number of actual AEs [1621].

Studies of hospital readmissions have reported that 9–48% of these readmissions were associated with substandard hospital care or inadequate post-hospital discharge care [22]. One study of patients discharged from the general medical service of a US tertiary care hospital found that 19% (95% CI: 15–23%) experienced new or worsening symptoms, unanticipated visits to health facilities for tests or treatment or death [23]. A similar study found similar outcomes in 23% (CI: 19%–28%) of 328 patients discharged from an Ontario teaching hospital [27]. Neither study differentiated between patients who were and were not receiving home care.

Clinical databases have also been used to identify AEs in home care patients. US investigators found that 13% of homecare clients experienced an AE [23]. Using data from the RAI-HC assessments of home care clients in three Canadian provinces, Doran et al. [24] identified the safety risks and potential AEs in this population. The most prevalent events were new fall or new hospital visit; however, given the nature of the data it was uncertain the extent to which these events were linked to the care or clients or the result of other factors, including disease progression.

The AE rate found in this study is 2.4 times higher than that from a previous study of home care patients in Winnipeg, Canada which found an annual AE rate of 5.5% (CI: 95%: 3.3–7.7%) [22]. The higher rate may reflect the differences in program and/or patient characteristics across home care programs and/or methodological differences, including lower screening criteria sensitivity, the use of social workers to screen charts for potential AEs, exclusion of patients with short lengths stay and the use of a limited definition for AEs in the earlier study.

This study demonstrates that home care patients are at risk from injury as a result of their care. Patient safety studies in hospitals have focused on harm resulting from the actions of professional and other staff. However, home care often includes care provided by patients' family members and friends, and the patients themselves. This raises new challenges for improving safety. Self-care was rated as a contributing factor in over half (52.6%) of AEs and is more important than the contribution of care provided by health-care staff. Self- or informal caregiving together were rated as a contributing factor in two-thirds of AE-associated deaths. These findings suggest there may be benefit in enhanced patient and informal caregiver education, skill development and clinical planning as interventions to reduce AEs. However, only one-third of AEs were assessed as being likely preventable suggesting that improvements to patient safety in home care will be challenging.

In addition to their impact on patients, this study identifies that AEs lead to increased health-care utilization, including emergency department visits and hospital readmissions, reinforcing the concerns raised from studies that have focused on these problems from a hospital perspective [36, 37]. Efforts to improve the safety of home care will help to reduce the incidence of hospital readmissions and the related, subsequent use of healthcare resources.

There were several limitations in this study. Since the research was an exploratory study, efforts were made to identify which patient factors were not correlated with AEs as well as those that were significantly correlated. For this reason a large selection of factors of interest to home care practitioners were included in the analysis. However, the potential for type II error increases with the inclusion of a high number of variables examined within a limited sample. The rate of AEs found in this study may be an underestimate since those patient charts not identified as being positive for one or more screening criteria were not reviewed for AEs by physicians. The use of a broader list of screening criteria also might have yielded a larger number of positively screened charts, possibly leading to the discovery of more AEs. As well, some home care practitioners may document more extensively and increase the likelihood that AEs are identified [38] Since the data were abstracted by health-care professionals from patient charts that have been compiled by other professionals, there may be biases inherent in the abstracted data that could affect the reliability of ratings of type of caregiver contributing to the AEs. Variance in measured rates of AEs may be due to differences in methods of identifying and documenting AEs, as well as reviewer biases and the inclusion or exclusion of AEs of lesser severity [39, 40]. When compared with chart review, direct observation for errors can yield larger numbers of AEs [30]. Retrospective AEs analyses may be prone to hindsight and attribute bias [30, 41, 42].

The discovery of AEs using chart review is prone to hindsight bias [41, 43, 44]. The preventability measure is a six-level subjective measure. While this measure has been used internationally in published reports of hospital AE studies, it has not been commonly used in home care studies. Inter-rater reliability of preventability ratings for this study had moderate kappa measures [0.714 (95% CI: 0.458–0.971]; however, given the subjectivity of the measure, some caution should be taken in interpreting this result. Generalizability may be limited if home care programs elsewhere differ substantively from those in this study.

The increasing use and complexity of home care services presents patient safety challenges [35]. Home care is an intricate combination of the provision of basic to complex health-care interventions in a residential setting that is not designed for the delivery of health care. Home care patients control the environment in which care is provided and their personal histories contribute to the selection of who, other than paid home care providers and health-care professionals, are involved in the direct provision of interventions. These circumstances expose home care patients to a different set of risk factors than they might experience in institutionalized settings [24, 35, 37]. This study is one of the first to report on the type and incidence of AEs in a home care population using trained nurses and physicians to undertake detailed chart audits. Additional research is needed to further explore and report the relationships between patient and disease characteristics and the specific nature of AEs in order to suggest ways of identifying patients at risk. Development of taxonomies that describe patient safety events could strengthen attempts to locate and describe risks to patient safety within home care and across health-care settings [45]. There are indications, such as the recent Canadian Patient Safety Institute funding of a national home care patient safety study that interest in patient safety in home care settings is growing.


This work was supported by funding from the Ministry of Health and Long-Term Care in Ontario, Canada through the University of Toronto, Department of Health Policy, Management and Evaluation, Healthcare Management Research Unit; and Comcare Health Services Ltd.

Conflict of interest

No author has any financial interest or connections, direct or indirect, or other situations that might raise the question of bias in this work or the conclusions, implications or opinions stated, including pertinent commercial or other sources of funding for the individual authors or for the associated departments or organizations, personal relationships or direct academic competition.

Overlapping publications

This paper has not been published nor submitted for publication elsewhere.


All authors fulfill the criteria of authorship as specified in the ‘Uniform Requirement for Manuscripts submitted to Biomedical Journals’ (http://www.cmje.org/).


View this table:
Appendix 1

Brief description of AEs occurring in 55 patients, by corresponding rating of preventabilitya


View Abstract