Article Text
Abstract
Introduction Patients aged 65 years or older account for a growing proportion of emergency department (ED) repeat attendances. This study aimed to identify health and non-health factors associated with repeat ED attendance, defined as one or more visits in the previous 6 months in patients aged 65 years or older, and to examine the interaction between social and health factors.
Methods 306 patients were interviewed. Demographic, socioeconomic, physical, mental health and post-ED referrals were examined. Logistic regression was used to identify factors independently associated with a repeat ED visit, OR and 95% CI are presented. Log likelihood ratio tests were used to test for interactions.
Results ED revisits were reported by 37% of this elderly population. Independent risk factors for a repeat ED visit were previous hospital admission OR 3.78 (95% CI 2.53 to 5.65), anxiety OR 1.13 (95% CI 1.04 to 1.22), being part of a vulnerable social network OR 2.32 (95% CI 1.12 to 4.81), whereas a unit increase in physical inability as measured by the Nottingham Health Profile had a week association OR 1.01 (95% CI 1.00 to 1.02). There were no significant interactions between social networks and the other health-related variables (p>0.05). In patients directly discharged from ED, 48% (71/148) had no documented referrals made to community services, of which 18% (27/148) were repeat ED attendees.
Conclusion ED act as an important safety net for older people regardless of economic or demographic backgrounds. Appropriate assessment and referral are an essential part of this safety role.
- Elderly
- emergency care systems
- emergency department
- environmental medicine
- imaging
- repeat visit
Statistics from Altmetric.com
The general population of countries in the developed world is experiencing an unprecedented increase in life expectancy. Ireland is no exception, with over 11% of the population aged 65 years and older and 2.7% aged 80 years and older (http://www.cso.ie). This reflects the tremendous gains in socioeconomic circumstances and healthcare provision within the country over the past 50 years.1 However, this increase in the older population brings with it complex interactions between social, economic and health circumstances that challenge the current health and social welfare infrastructures that developed to support a smaller and less aged population.
Internationally, it is recognised that elderly people are overrepresented in the emergency department (ED), with between 12% and 21% of all ED visits being in the over 65 years age group.2 A study of four Dublin teaching hospitals in 2004 found that 5% of attendees were aged 65–69 years and this increased to 10% for those aged 80 years and older.3
Re-attendance at the ED and hospital re-admission among older people may be indicators of the organisation, ability and responsiveness of health and social care services to meet the complex needs of this group.4 5 The definition of a repeat visit or hospital re-admission ranges from 28 days to 6 months. Studies that selected a 28 or 30-day cut-off reported ED re-attendance between 10% and 29%, with hospital re-admissions between 3% and 17% among older patients.2 6 7 However, studies that selected a 6-month cut-off suggest that earlier time points underestimate these outcomes and ED re-attendance can be as high as 44%, with re-admissions up to 25%.2 Reasons for attendance at ED are multifaceted phenomena only partly explained by the severity or acuity of illness.8 9 Other associated factors include availability of community services, especially diagnostics, out of hour GP services, accessibility to specialist hospital-based services as well as population socioeconomic and demographic factors.8 9 It is likely that such factors play an important role in the intercountry variability in hospital re-admission or ED re-attendance rates.5 10 Little is known about the characteristics and circumstances of this older population who experience repeat ED visits in a European context. In particular, the interaction between health and socioeconomic or demographic factors is poorly understood.
The aim of this study is to identify physical, cognitive, demographic and socioeconomic factors associated with repeat ED attendance, defined as a visit within the previous 6 months, in a population aged 65 years and older, and to examine the interaction between independent health and non-health factors in a predictive model of repeat ED attendance.
Methods
This study was carried out in the ED of two Dublin teaching hospitals. The study was approved by the ethics committee in each hospital. Patients were eligible for inclusion if they were aged 65 years or older, residing in the community, had no significant cogitative impairment that would preclude giving informed consent, and spoke English. In addition, permission of the nursing and medical staff was obtained before approaching patients to seek written informed consent.
Patients were recruited into the study over an 18-month-period. The target sample size was 300 patients, 150 from each study site. The population was further stratified into patients admitted or discharged directly from the ED. Four hundred and thirteen patients were approached, 26% (106/413) refused or were too ill. The final sample comprised 307 patients, 158 admitted and 149 discharged from the ED. In this analysis one patient was excluded because of missing outcome data.
Patients were interviewed in the ED or within 72 h of hospital admission. A structured questionnaire was used to obtain demographic, socioeconomic, current ED visit details, and health and social service contact in the previous 6 months. Patients' physical, cognitive and social network status were measured using the Nottingham health profile (NHP), mini mental status evaluation, hospital anxiety and depression scale and Wenger's network assessment tool.11–14 All questionnaires were administered by trained research assistants. Additional information on diagnosis, Manchester triage category and follow-up referral was obtained from the ED electronic records.
Data analysis
The network assessment tool describes five types of support networks, family dependent, local integrated, wider community focused, local self-contained and private restricted. The ‘local self-contained’ and ‘private restricted’ are characterised by low levels of family or community support and are recognised as the most vulnerable network types. In this analysis the two vulnerable networks were compared with the three more robust networks. The NHP measures six factors related to physical and emotional wellbeing on a continuous scale from 0 to 100. The lower the score the better the function on each item. These individual items were divided into tertiles to identify the proportion of people in low (0–33), medium (34–66) and high-risk (67–100) categories.
In this analysis attendance at ED in the past 6 months was treated as a dichotomous outcome, the population was stratified into those patients who had one or more ED visits in the past 6 months or patients for whom this was their first ED visit within that time period.
All variables were treated as categorical in the initial data description; frequencies and percentages are presented. Mantel–Haenszel χ2 was used to test for significant differences between the groups and evidence of trend was tested using the Cochran–Armitage test for trend. Univariate logistic regression was used to identify factors independently associated with a repeat ED visit. A multivariate logistic regression model was constructed to identify which variables remained independent predictors of a repeat ED visit adjusting for other factors in the model. All variables with a p value of 0.1 or less were entered into the model, and using backward stepwise elimination were retained in the model if the p value was less than 0.05. In the final model the simpler linear rather than categorical variable was used if there was no significant difference to model fit.
A receiver operator curve was used to examine the predictive value of the model. Interactions between social factors and other independent variables were tested using logistic likelihood ratio tests to assess if social factors had a modifying effect on the other predictive variables. Analysis was performed using SAS version 9.
Results
In this study population of 306 patients, 37% (114/306) reported a repeat ED visit in the previous 6 months. A single previous visit was reported by 21% (63/306), 9% (29/306) reported two visits and 7% (22/306) reported three or more visits.
Demographic and socioeconomic variables
There was no significant difference between the repeat ED visit group and the non-repeat ED visit group in terms of demographic or socioeconomic variables (table 1). In both groups, over 30% of patients were aged 80 years or older and over 40% lived alone. The majority of patients owned their own home and 50–57% of patients were in social class 4–7 (skilled manual, unskilled or unknown).
There was no significant difference in ED re-attendance between hospital sites. Over 50% of patients self-referred, but it was not associated with repeat attendance (OR 0.85, 95% CI 0.53 to 1.36). Patients in the repeat ED visit group were significantly more likely to have had previous hospital admissions and were also nearly twice as likely to be admitted to hospital on this occasion (OR 1.99, 95% CI 1.24 to 3.19).
In the previous 6 months only 7% (7/114) of patients in the repeat ED visit group did not see their GP, whereas 49% (56/114) did not have contact with community-based nursing or other social services (table 1).
Medical related conditions accounted for nearly 80% of presenting complaints (figure 1). The highest proportion of repeat attendees presented with cardiac (14%, 14/114) or respiratory (14%) conditions. Falls, other infections, genitourinary and gastrointestinal complaints each accounted for 9–10% of presentations. Genitourinary and respiratory conditions were significantly more likely to be associated with a repeat visit, whereas falls were more common in the non-repeat visit group.
Physical, cognitive and social network assessment
Patient acuity was indicated by the Manchester triage score, 11% (34/306) were in triage category 1–2 (highest risk), but there was no significant difference between the groups. Patients in the repeat ED visit group were over twice as likely to report moderate or high levels of reduced physical ability compared with the non-repeat visit group (table 2). There was no significant difference between the groups with regard to the other variables in the NHP. Significantly higher levels of anxiety and depression were reported by the repeat ED visit group, 28–34% recorded scores greater than seven (indicating a potential case) compared with 12–22% of the non-repeat visit group. There was a higher proportion of repeat ED visit patients in the vulnerable network category, but this did not reach statistical significances in the univariate analysis.
Predictors of re-attendance
In the multiple regression model four variables remained independent predictors of a repeat ED visit (table 3). Previous hospital admission in the past 6 months was the strongest predictor variable, with over a tripling of the risk per previous admission (OR 3.78, 95% CI 2.53 to 5.65). There was a 13% increase in the risk of re-attendance per unit increase in anxiety score, while being in a vulnerable network doubled the likelihood of a repeat ED visit. A one unit increase in the NHP physical ability score had a weak association with a 1% increase in the risk of re-attendance. There was no significant interaction between social networks and repeat hospital admission, physical ability or anxiety; this indicates that these variables were not modified by social networks and their influence was independent of each other.
This model accounted for only 25–34% (R squared) of the variation in the data. The area under the receiver operator curve was 0.8, with a sensitivity of 53% and specificity of 86%. This suggests the model did not accurately reflect the complex circumstances that led to repeat attendance in this population.
Follow-up after direct ED discharge
The ED records of patients discharged directly from the ED (n=148) were examined for evidence of referral for community follow-up. There was no documented evidence of referral in 48% (71/148) of patients' records, of which 18% (27/148) were repeat ED attendees. The most frequent referral was for GP follow-up 44% (66/148), an additional 3% (5/148) were referred to the care of the elderly community team or community nursing and 3% (5/48) to other services.
Discussion
Over a third of elderly patients (37%) had a repeat ED visit in the previous 6 months. Factors predicting re-attendance at ED were predominantly related to physical and psychological health. In particular, each previous hospital admission was likely to triple the risk of re-attendance. Hospital admissions, especially repeat admissions, are likely to be a surrogate marker for poor and deteriorating health and thus may not be independent of measures of physical ability using the NHP. Anxiety and being part of a vulnerable network were also independent predictors of repeat ED visits.
However, despite measuring multiple factors, the final model only explained approximately 30% of the variation in the data, with 53% sensitivity in accurately identifying patients who experienced repeat ED visits. Also in this study, with the exception of GP contact, there were low levels of contact and follow-up by community nursing and other health and social services in the 6 months before the ED visit or for patients discharged directly home following an ED visit.
The incidence of repeat ED visits is similar to that reported in other studies that used longer follow-ups, and ranged between 32% and 43%.15–18 Studies examining predictive factors associated with an initial or repeat ED visit in elderly individuals all identify a physical health component.15–17 19 A recent systematic review of ‘determinants of ED visits’ applied ‘Andersen's behavioural model’ to identify the homogeneity in predictive factors across studies despite the diversity of measures used.9 Patient ‘need’, either perceived or evaluated, was primarily and consistently identified. Two other factors, ‘predisposing’ (sociodemographic or health beliefs) and ‘enabling’ factors (alternative services or support), were also present to a greater or lesser extent in many of the studies. In this study ‘need’ as indicated by previous hospitalisation, physical ability and anxiety as well as ‘enabling’ factors in terms of social networks were identified. Predisposing factors relating to economic and demographic factors were not significant in this model.
This study did not specifically measure health beliefs, but patients were asked to identify alternatives to the ED, only 24% (59/306) were able to consider an alternative treatment setting to the ED. Their recommendations centred round the availability of a wider range of treatments, supports and diagnostics in primary care, also better relaying of information between a patient's GP, the ‘out of hours’ GP services and hospitals. In this study, acute injury or episodes of acute deterioration in a chronic health condition that patients felt were beyond the scope of their normal GP service drove contact with the ED.
The high incidence of repeat hospitalisation (41%) or ED visits (37%), coupled with the low level of contact with community resources either before or after the ED visit, suggests that transitional care between hospital and the community may not have been in place for a significant number of patients. It is also noteworthy that 90% of this population had contact with their GP in the previous 6 months, and on this occasion 44% of patients were referred by a healthcare professional. This suggests that the current level of primary care provision within the community is inadequate to deal with the healthcare needs of this population.20 In this study the ED appears to act as an important safety net for this population regardless of economic status or social support networks (only 17% were in a vulnerable network).
The aim of health and social care is to optimise health potential and support independent community living for as long as possible.21 22 These supports need to be responsive and directly accessible within the person's community. In this study an ED visit or repeat visit often did not act as a catalyst for referral to multi-professional geriatric expertise or the instigation of preventive or community services. ED assessment and intervention for older individuals needs to shift from a simple condition focused model to a broader assessment of risk factors, including an individual's current formal and informal community supports.23 24 ED also need to develop alert systems to identify elderly re-attendees and actively monitor reasons for re-attendance and changes in health status between visits.
Study limitations
This was a convenience sample and a self-selecting population, only those patients who felt well enough to participate in the study did so, this may account for the low number of patients in triage category 1–2. Contact with medical services including hospital admission and ED visits relied on patient recall, there is evidence to suggest that while recall of contact with healthcare services was accurate, the volume of contact may be inaccurately recalled by this population.25 There was no measure of physical or cogitative deterioration between one ED visit and the next. Diagnosis, co-morbidities and community service referral relied on ED documentation, which was not always complete and may result in misclassification and underreporting of service referral.
Conclusions
The elderly population in this study regardless of sociodemographic background relied on the ED as an important safety net. Healthcare needs drove contact with the ED, this was associated with a perception by older people of a lack of alternative resources in primary care. The ED should be proactive in identifying and monitoring both health and non-health risk factors with appropriate referral to specialist and community supports.
Acknowledgments
The authors would like to acknowledge the cooperation and help of the patients who gave freely of their time, the nursing, medical and IT staff within the ED and the three research assistants who worked on this study: I Lyons, L O'Connell and N Brennan.
References
Footnotes
Funding This project was funded by a joint project grant from the participating hospitals and the Irish Health Research Board.
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the The Mater Misericordiae University Hospital and St Vincent's University Hospital Dublin.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Primary survey