Presented at the Society for Academic Emergency Medicine Annual Meeting, Boston, MA, June 2011; and the American Geriatrics Society Annual Scientific Meeting, Washington, DC, May 2011.
Funded by the Duke Endowment and the Community Connection for Seniors.
The authors have no relevant financial information or potential conflicts of interest to disclose.
Older patients discharged from the emergency department (ED) have difficulty comprehending discharge plans and are at high risk of adverse outcomes. The authors investigated whether a postdischarge telephone call–mediated intervention by a nurse would improve discharge care plan adherence, specifically by expediting post–ED visit physician follow-up appointments and/or compliance with medication changes. The second objectives were to determine if this telephone call intervention would reduce return ED visits and/or hospitalizations within 35 days of the index ED visit and to determine potential cost savings of this intervention.
This was a 10-week randomized, controlled trial among patients aged 65 and older discharged to home from an academic ED. At 1 to 3 days after each patient's index ED visit, a trained nurse called intervention group patients to review discharge instructions and assist with discharge plan compliance; placebo call group patients received a patient satisfaction survey call, while the control group patients were not called. Data collection calls occurred at 5 to 8 days and 30 to 35 days after the index ED visits for all three groups. Chi-square or Fisher's exact tests were performed for categorical data and the Kruskal-Wallis test examined group differences in time to follow-up.
A total of 120 patients completed the study. Patients were 60% female and 72% white, with a mean age of 75 years (standard deviation [SD] ± 7.58 years). Intervention patients were more likely to follow up with medical providers within 5 days of their ED visits than either the placebo or the control group patients (54, 20, and 37%, respectively; p = 0.04). All groups performed well in medication acquisition and comprehension of medication indications and dosage. There were no differences in return visits to the ED or hospital within 35 days of the index ED visit for intervention patients, compared to placebo or control group patients (22, 33, and 27%, respectively; p = 0.41). An economic analysis showed an estimated 70% chance that this intervention would reduce total costs.
Telephone call follow-up of older patients discharged from the ED resulted in expedited follow-up for patients with their primary care physicians. Further study is warranted to determine if these results translate into improved patient outcomes, decreased return ED visits or hospital admissions, and cost savings resulting from this intervention.
Los pacientes mayores dados de alta desde el servicio de urgencias (SU) tienen la dificultad para comprender los planes de alta y poseen alto riesgo de resultados adversos. Se investigó si una intervención mediante una llamada telefónica por un enfermero tras el alta mejoraría la adherencia al plan de cuidados al alta, específicamente para las citas de seguimiento al médico tras la visita al SU y/o el cumplimiento con los cambios de medicación. Los objetivos secundarios fueron determinar si esta intervención telefónica reduciría las revisitas al SU y/o los reingresos hospitalarios en los primeros 35 días desde la visita índice al SU, y determinar el ahorro potencial de costes de esta intervención.
Ensayo clínico aleatorizado de 10 semanas de duración en pacientes de 65 años o más dados de alta a su domicilio desde un SU universitario. Entre el día 1 y 3 de la visita índice al SU, un enfermero con formación realizó una llamada telefónica a los pacientes del grupo de intervención para revisar las instrucciones al alta y ayudar con el cumplimiento del plan al alta; los pacientes del grupo de llamadas placebo recibieron una encuesta telefónica de satisfacción; mientras que los pacientes del grupo control no recibieron la llamada telefónica. La recogida de datos de las llamadas telefónicas tuvo lugar entre los días 5 a 8 y 30 a 35 tras la visita índice al SU para los tres grupos. Se realizaron test de ji-cuadrado y test exacto de Fisher para las variables cualitativas y el test de Kruskal Wallis para analizar las diferencias entre los grupos en el tiempo de seguimiento.
Ciento veinte pacientes completaron el estudio. Los pacientes fueron en un 60% mujeres y en un 72% blancos, con una edad media de 75 años (DE 7.58 años). Los pacientes del grupo de intervención tuvieron mayor probabilidad de seguimiento con un médico en los 5 primeros días de la visita al SU que los pacientes de los grupos de control y placebo (54%, 20% y 37% respectivamente, p = 0,04). Todos los grupos realizaron bien la adquisición de la medicación y la comprensión de las indicaciones y dosis de la medicación. Hubo una diferencia numérica pero no a nivel estadístico de las revisitas al SU o el reingreso hospitalario en los primeros 35 días de la visita al SU índice para los pacientes de los grupos de intervención, comparado con los pacientes del grupo control y placebo (22%, 33%, y 27% respectivamente, p = 0,41). Un análisis de costes mostró una oportunidad estimada de reducir los costes totales con esta intervención en un 70%.
El seguimiento telefónico de los pacientes mayores dados de alta desde el SU resultó en una mayor solicitud de seguimiento de los pacientes a los médicos de atención primaria. Se requieren futuros estudios para determinar si estos resultados se traducen en la mejoría de los resultados del paciente, y confirman una disminución de la revisita al SU o los reingresos hospitalarios y una reducción de costes con esta intervención.
The number of older adults receiving health care in emergency departments (EDs) is increasing rapidly, with an estimated 19.8 million ED visits made by adults aged 65 years and older in 2009. Older patients discharged to home from the ED are at high risk of experiencing adverse outcomes, including receiving inappropriate medications, return to the ED, and death. Current ED care processes, and in particular ED discharge processes, are not optimized to meet the complex care needs of older adults or to ensure coordinated care beyond the ED.
When patients are discharged from the inpatient hospital setting, it is believed that the quality of their transitional care can be improved, in part, if expedited follow-up with a primary care physician and medication management services are provided.[5, 6] Increased attention to these pillars of high-quality transitional care may help mitigate the risks experienced by older adults on discharge from the ED as well.
Prior studies have evaluated the effect of care coordination interventions for older adult ED patients, but have largely targeted patients with services prior to their discharge from the ED. These investigations have included assessments by geriatric specialists and referrals to community and home services. These interventions are resource-intensive and have had only intermittent success in decreasing hospitalizations, reducing ED visits, increasing primary care visits, and improving functional outcomes.[7-14] Several studies have found that comprehension of discharge instructions is a significant barrier to care compliance[15-18] and that many older patients discharged from the ED do not understand their discharge instructions, nor do they attend follow-up appointments.[18, 19] Studies conducted in other patient populations have shown that telephone follow-up encourages compliance with discharge instructions and increases the likelihood of physician follow-up.[20-23] Prior work has also shown that telephone follow-up interviews are feasible among older patients discharged from the ED.
Telephone follow-up with discharged older adults or their caregivers may be a low-cost and practical strategy to increase care plan compliance. Many EDs already use similar services to contact patients regarding urgent test results and to arrange indicated follow-up care after discharge. To our knowledge, no published studies have evaluated whether a follow-up telephone call performed by a nurse after an older adult's discharge from the ED can improve patient compliance with the discharge medical treatment plan. If a telephone call conducted after discharge can facilitate medical follow-up and/or improve medication adherence, it may be an effective and inexpensive intervention with potential to improve older adults' post-ED outcomes.
Our primary outcome was to evaluate whether a telephone call intervention conducted by a trained nurse 1 to 3 days after an older adult's index ED visit would achieve enhanced care plan compliance as measured by expedited outpatient follow-up within 5 days of the index ED visit and/or increased compliance with medication changes. We had two secondary outcomes. We sought to determine if the telephone call follow-up in the intervention group was associated with a lower likelihood of a return visit by the patient to the ED or inpatient hospitalization within 35 days of the index ED visit. We also conducted an economic analysis of the potential for cost savings from this intervention.
We conducted a prospective randomized controlled trial. Institutional review board approval and telephone consent from each patient were obtained, and the trial was registered at ClinicalTrials.gov (NCT01207180).
Study Setting and Population
This study took place over 10 weeks from September 5 through November 9, 2010, among patients aged 65 years and older discharged from an academic Level I trauma center ED. The ED has an annual census of 67,000 patients and is located in Southeastern United States. Inclusion criteria for patients in this study included age 65 years or older, living in a home setting (e.g., not a nursing home, assisted living, or rehabilitation center), returning to a home setting after ED discharge, and having been instructed to seek outpatient follow-up. Patients were enrolled after visits to the ED on Sunday, Monday, and Tuesday, to facilitate follow-up telephone calls during the week, as we did not have the staffing to make calls over the weekend. Patients were excluded if they were admitted to the hospital, discharged to a setting other than their home, not fluent in English, or not instructed to seek outpatient follow-up or neither the patient nor the caregiver were able to pass a mental cognition screening examination. Patients instructed to obtain follow-up appointments but not given new medications were included. Patients were excluded from the outcomes of outpatient follow-up and medication compliance if they returned to the ED or were admitted to inpatient settings within 5 days, as these events could decrease their likelihood of completing a follow-up appointment or acquiring medications, given the shorter time frame they had available to complete these actions. These patients were included, however, in our secondary outcomes of interest, return visits to the ED, or hospitalization within 35 days and economic analysis of the intervention, as the calls may have affected their likelihood to return to the ED.
We randomly selected up to nine older adults per enrollment day who were discharged home from the ED for participation in this 10-week study. Block randomization was created by having the study coordinator place marbles of three colors into a bag, one marble for each patient with the colors representing the study groups, and blindly drawing a marble for each patient to determine to which group the patient would be assigned. The three study groups were: 1) an intervention group that received a telephone call from a trained study nurse between 1 and 3 days after ED discharge to review and facilitate compliance with patient-specific discharge instructions (see Data Supplement S1, available as supporting information in the online version of this paper, for the script); 2) a placebo group that received a scripted five-question patient satisfaction survey telephone call made between 1 and 3 days after ED discharge to control for the effect that a telephone call from the hospital may have upon discharge instruction compliance (Data Supplement S2, available as supporting information in the online version of this paper); and 3) a control group that did not receive a telephone call during the 1- to 3-day post–ED discharge time period. Following a prewritten script, the trained study nurse provided intervention group patients with any necessary assistance in facilitating home services, scheduling follow-up appointments, managing medication(s), and receiving referrals to available community resources. Each study patient subsequently received a call between 5 and 8 days (Data Supplement S3, available as supporting information in the online version of this paper) and 30 and 35 days (Data Supplement S4, available as supporting information in the online version of this paper) after the ED visit to measure compliance with discharge instructions using a standardized survey (see Figure 1 for a flow diagram of the study groups).
Subsequent telephone interviews were conducted with the patient or his or her caregiver. If the patient was unavailable or physically or cognitively unable to participate in the telephone interview, the caregiver was interviewed. Three telephone call attempts were made to reach each participant. All interviewees needed a minimum score of 4 of 6 on the Six-item Screener, a previously validated instrument for measuring cognitive impairment (sensitivity = 88.7%, specificity = 88%), conducted over the telephone, to participate in the interview. If the patient did not pass this screening exam, a caregiver was asked to participate. The caregiver also had to obtain at least 4 of 6 on the Six-item Screener. If neither the patient nor the caregiver passed the Six-item Screener, they were excluded from the study. The Six-item Screener was done only on the first telephone call for each study participant and was not repeated on subsequent calls. Consent was obtained over the telephone at the beginning of each interview. Data collected included patient demographics, respondent's report of the existence and date of follow-up appointments, whether any newly prescribed medications from the ED visit had been obtained, information regarding the respondent's comprehension of medication indications and their dosing schedule, and whether the patient had had subsequent ED visits or hospitalizations since the index visit. To determine whether patients had acquired new medications, patients were allowed to read off of their medication bottles. When assessing comprehension of medication indications and dosing, patients were similarly allowed to refer to their discharge instructions and prescription bottles when asked. All intervention group calls were conducted by a trained study nurse and followed a prewritten script. Placebo group and control group telephone calls, as well as data acquisition calls for all three groups, were conducted by trained research assistants, with assistance from a trained nurse when necessary, and also followed a script (Data Supplement S1). Prior to contacting any study patients, research assistants were trained in the conduct of study procedures and were monitored in their learning of and adherence to the study scripts.
To determine differences between groups with regard to follow-up visits and medication changes, chi-square or Fisher's exact tests were performed for categorical data. To examine differences in time to follow-up appointment between groups, the Kruskal-Wallis test was used. Data are presented as frequencies and percentages for categorical data and medians with interquartile ranges (IQRs) for continuous or metric data. Patients were not enrolled in the study more than once during the 10-week study period. Sample size calculations determined that 39 patients were needed in each group to detect a 30% difference in time to follow-up appointment between groups, with alpha set at 0.05 and a power of 80%. As the authors were unaware of any literature to guide what difference in time to follow-up after ED visit would be meaningful, a 30% difference in time to follow-up was chosen in the belief that this magnitude of difference may be clinically significant. All analyses other than the economic analyses were performed using SAS statistical software (version 9.3, SAS Institute, Cary NC). The economic analysis was performed with Stata (version 12, StataCorp LP, College Station, TX).
To assess the potential economic implications of the intervention, we calculated the extent to which the current study might result in cost savings or be cost-effective based on intervention costs and the expected costs of ED visits or hospital use within 35 days. To address uncertainty, we created 10,000 bootstrap replicates of the incremental cost and incremental effect (return hospital or ED visits avoided) following standard methods.
Specifically, for estimating effects, we simulated a binomial distribution for the likelihood of having a return hospital or ED visit within 35 days using results from the treatment and control groups only. We did not consider the results for the placebo group in the economic analysis. We estimated the cost of the telephone intervention to be roughly $21 per completed intervention telephone call, given the nurse's annual salary and benefits, and an average of 30 minutes for each interventional call including any appointments that needed to be scheduled or other additional calls. We estimated a return hospital admission to cost $6,442 (hcupnet.ahrq.gov/) and an ED visit without hospitalization to cost $1,000. For costs, we also obtained an estimate from the hospital that 63.6% of patients in this age group who sought care in the ED were admitted to the hospital based on a sample of 4,000 ED visits. We used these estimates to simulate 10,000 pairs of incremental effects and incremental costs for the treatment versus control group; these pairs were graphed in a cost-effectiveness plane to determine the proportion of estimates that support the intervention being “cost-effective” (i.e., reduced visits at greater total cost), “cost saving” (i.e., reduced visits at lower total costs), or a “bad investment” (i.e., more visits at greater total costs).
During the study period, 180 eligible patients were identified, of whom 178 were randomized to the three study groups. The two remaining patients were not randomized for inclusion in the study as the number of eligible patients on the day of their index ED visit exceeded nine. It was necessary to randomize patients before enrolling them over the telephone, as what group they were in determined when they were called and enrolled. Of these 178 patients, 18 declined study participation, 19 could not be reached by telephone during follow-up, 12 (four in each group) were disqualified from primary outcome analysis secondary to return to an ED or other hospital setting within 5 days of their discharge, three were disqualified for other exclusion criteria, and six had incomplete surveys. No patients were disqualified secondary to failing the Six-item Screener, as in each case the caregiver was able to pass the Six-item Screener. Thus, the final analysis sample comprised 120 patients: 39 patients in the intervention group, 35 patients in the placebo group, and 46 patients in the control group (Figure 2). Eighty-nine percent of patients were reached by telephone within three tries, with 66.7% of patients being reached on the first try.
Overall, subjects completing the study were mostly female (60%) and white (72%), with a mean (±SD) age of 75 (±7.58) years (Table 1). In 77% of the cases, the patient was interviewed rather than a caregiver. Forty-nine percent of subjects were prescribed at least one medication. Of these, over 90% of the subjects or their caregivers stated that the prescription had been filled and knew the medication name, dosage, and indication. There were no significant differences between the groups in terms of likelihood of respondent report of filling the prescription, knowledge of name, dosage, or indication (Table 2).
Table 1. Patient Demographics
Data are reported as n (%).
Patient interviewed (rather than caregiver)
Table 2. Prescribed Medications
Data are reported as n (%).
Prescribed at least one medication
Knew medication name and dosage
Knew reason for medication
Overall, 103 (86%) participants had scheduled follow-up appointments to see physicians at some point after their ED visits by the time they were contacted for their 5- to 8-day post–ED visit telephone interviews. The majority of patients in the intervention group made their own follow-up appointments, but for the minority of intervention group patients or caregivers who requested help making the appointments, the nurse making the intervention calls helped schedule appointments for them when possible. The study nurse was able to determine before calling if the patient had an appointment scheduled within the health care system of the ED that this study was conducted within. The median time from the ED visit to the date of the follow-up appointment was 7 days (IQR = 2 to 10 days). Within the groups, the median number of days from the ED visit to the follow-up appointment was 4 days for the intervention group, 9 days for the placebo group, and 6 days for the control group. The intervention group was 1.8 times more likely than the control and placebo groups combined to see a physician within 5 days (p = 0.04; Figure 3). Fifty-four percent of the intervention group had already seen a physician within 5 days of their ED visit, compared with 37% of the control group and 20% of the placebo group (p = 0.05; Figure 4). Ten of 45 (22%) intervention group subjects as compared to 13 of 39 (33%) placebo call subjects and 13 of 48 (27%) control group subjects had ED visits or hospital admissions within 35 days of discharge from the ED (p = 0.41).
For the economic analysis, we conducted simulations with data on 35-day return rates to the ED or hospitalizations for the intervention and control groups. We assessed the distribution of the estimates in a cost-effectiveness plane, which showed that most of the variation came from the estimated rates of return visits using the small study sample. The only variation in costs came from the relatively precise estimate of the likelihood of hospital admission based on a large sample of 4,000 ED visits, as the estimates of ED and hospitalization costs were fixed values. The analysis showed that the intervention had an estimated 70% chance of being cost saving (i.e., reduced return visits at lower total cost), and a 3% chance of being cost-effective (i.e., reduced return visits but at higher total cost). We estimated that the intervention had a 27% chance of resulting in more return visits at higher total costs (i.e., being a bad investment), although this likelihood may be reduced substantially with a larger sample and more precise estimates of the effect of the intervention on subsequent ED visits and hospitalizations.
In patients discharged from the hospital to home, timely follow-up with primary or specialty care providers and medication management are two of the four pillars of discharge planning that together have been demonstrated to decrease subsequent hospitalizations.[5, 6] Although these two pillars have not been demonstrated to decrease return visits to the ED or hospitalizations for patients discharged from the ED, it is reasonable to posit that, especially for older patients or patients with multiple comorbidities, expedited follow-up and improved medication acquisition and comprehension of indication and dosing may facilitate safer transitions of care and reduce both return visits to the ED and future hospitalizations. Indeed, some ED-based care coordination interventions have been shown to benefit older patients, while others have not. Transitions programs like those developed by Coleman et al.,[5, 6] which includes the creation of a personal health record and the initiation of a series of visits and telephone calls by a transitions coach, require a significant investment in infrastructure and may be beyond the ability of most EDs to implement. However, many EDs already have a staff member responsible for calling discharged patients with urgent test results. As a result, a system relying on follow-up telephone calls may be feasible to implement in many EDs.
The primary outcome of this study was to evaluate whether we could achieve enhanced care plan compliance after discharge to home from the ED as measured by expedited outpatient follow-up and/or increased compliance with medication changes in adults age 65 years and older. Patients who were called by a trained nurse 1 to 3 days after discharge from the ED were nearly twice as likely as the placebo and control groups to follow up with their primary or specialized care physicians within 5 days. Five days was chosen as indicating appropriately rapid follow-up time as in this study the intervention group received the telephone call from the study nurse up to 3 days after discharge, so any change in follow-up time would need to allow for appointments a couple days after the call. The nurse was able to help patients and their caregivers navigate the health care system, arrange for appointments with doctors, review discharge instructions, and obtain needed referrals to community services. A visit to a physician soon after an ED visit has several potential advantages that include reexamination of the patient by a medical provider, review of medications, an assessment of the efficacy of the treatment initiated in the ED, and an evaluation to determine if further interventions are needed to optimize the possibilities of patient recovery.
This study did not demonstrate any significant difference between the groups in percentage of patients who obtained prescribed medications or who knew the name, dosage, or indications for these medications. This is likely due to the high rate of medication acquisition and measured comprehension across all groups. There are a number of reasons why these rates may be so high. Patients were allowed to read from their printed discharge instructions and prescription bottles when being interviewed by the research assistants. As a result, we may not have adequately assessed how well patients or their caregivers actually understood their medication indications and dosages. Although we believe that the majority of patients acquired the new medications prescribed in the ED, we have relied on respondent self-report of this endpoint. It is possible that direct observation of the patients' medications may produce different results.
It is interesting to note that in this study the placebo group had delayed follow-up with physicians compared to the control group. This finding, if reproduced, may suggest that there are unintended consequences of patient satisfaction calls. It is possible that when patients are contacted by hospital personnel with no mention of need for further evaluation, patients may gain a false sense of security that additional follow-up is not urgent.
Eighty-nine percent of patients were reached by telephone call follow-up. The high rate of success in reaching patients and the 70% likelihood that this intervention reduces total cost suggests that telephone call follow-up of older patients discharged from the ED may be a feasible and sustainable intervention. This study was not powered to identify a reduction in return visits to the ED or admissions to the hospital. However, the results of this study suggest that a larger study could test the hypothesis that an intervention telephone call as described here may decrease ED visits and hospital admissions within 35 days of initial ED discharge.
First, the study was performed at a single center that is part of a large health care system with multiple providers. As a result, this system may be better able to arrange timely follow-up appointments than more geographically isolated EDs.
Second, the need to randomize and consent patients after they left the ED presented some potential sources of bias. Control group patients received their first calls from the research assistant 5 to 8 days after their index ED visits, while intervention and placebo group participants received their first calls 1 to 3 days after their index ED visits. This required the research assistants to perform cognitive screening at day 5 through day 8 for control patients with the Six-item Screener and prevented blinding of the research assistants. We attempted to mitigate any potential bias this could introduce by scripting all of our research assistants' interviews. Also, the need to enroll patients after they left the ED resulted in needing to randomize patients before enrolling them and hence to exclude some patients after enrolling them. Third, our primary endpoint of time to outpatient follow-up and compliance with medications is a multifaceted endpoint examining aspects of care plan compliance. Although two of the outcomes meet the threshold for significance of p ≤ 0.05, these do not meet significance after adjustment for multiple comparisons. Fourth, follow-up information was self-reported, and although the information was able to be verified if the patient followed up within the same health care system in which the ED is located, there was no attempt made to verify subjects' answers as correct or erroneous if the patient followed up outside of this health care system. Fifth, we do not have exact numbers for how often the nurse helped make intervention group patients' appointments, as she was not asked to record this and hence know only that it was a small percentage of patients based on her recollection of making these calls. This may be important as the nurse making the appointments may contribute to the shorter follow-up time for the intervention group patients. Finally, a single nurse made all the intervention calls; hence it is unclear if other trained research nurses would obtain the same results.
Telephone call follow-up of older patients discharged from the ED with assistance complying with discharge instructions resulted in patients having expedited follow-up appointments with primary or specialty care. More study is warranted to determine if these results can be applied to multiple sites and in more diverse patient populations, as well as if this intervention can reduce return visits to the ED and/or hospitalizations and demonstrate cost savings.
The authors acknowledge the many research assistants for contacting participating subjects and Heather Altman, MPH, for her support and efforts to make the study possible.