Impact of event notification services on timely follow-up and rehospitalization among primary care patients at two Veterans Affairs Medical Centers

Abstract

Objective

To examine the effectiveness of event notification service (ENS) alerts on health care delivery processes and outcomes for older adults.

Materials and methods

We deployed ENS alerts in 2 Veterans Affairs (VA) medical centers using regional health information exchange (HIE) networks from March 2016 to December 2019. Alerts targeted VA-based primary care teams when older patients (aged 65+ years) were hospitalized or attended emergency departments (ED) outside the VA system. We employed a concurrent cohort study to compare postdischarge outcomes between patients whose providers received ENS alerts and those that did not (usual care). Outcome measures included: timely follow-up postdischarge (actual phone call within 7 days or an in-person primary care visit within 30 days) and all-cause inpatient or ED readmission within 30 days. Generalized linear mixed models, accounting for clustering by primary care team, were used to compare outcomes between groups.

Results

Compared to usual care, veterans whose primary care team received notification of non-VA acute care encounters were 4 times more likely to have phone contact within 7 days (AOR = 4.10, P < .001) and 2 times more likely to have an in-person visit within 30 days (AOR = 1.98, P = .007). There were no significant differences between groups in hospital or ED utilization within 30 days of index discharge (P = .057).

Discussion

ENS was associated with increased timely follow-up following non-VA acute care events, but there was no associated change in 30-day readmission rates. Optimization of ENS processes may be required to scale use and impact across health systems.

Conclusion

Given the importance of ENS to the VA and other health systems, this study provides guidance for future research on ENS for improving care coordination and population outcomes.

Trial Registration

ClinicalTrials.gov NCT02689076. “Regional Data Exchange to Improve Care for Veterans After Non-VA Hospitalization.” Registered February 23, 2016.

INTRODUCTION

Delivering high quality, coordinated care requires that providers access, manage, and share information efficiently. Retrieving, managing, and sharing health information in the conduct of care coordination, however, is challenging. In a study of internal medical residents, providers were observed to spend between 5% and 9% of their time looking for information on a patient.¹ In a care coordination survey of US patients, 27% of respondents reported that their test results were not available or that duplicate tests were ordered during a medical appointment; and 17% reported that information was not shared among their multiple care providers.² Similar gaps in care coordination were observed in Canada, Australia, France, the Netherlands, and Norway.² Additional evidence from a systematic review of 33 studies in 14 nations confirms that primary care providers often perceive care delivery to be disjointed, with little flow of information or continuity of care between settings.³ Coordination is especially important for older adults who possess greater medical comorbidities and experience disproportionate rates of postoperative morbidity and mortality.^4,5

To facilitate and enhance care coordination, providers seek to leverage health information technologies to access information on their patients, regardless of its source. Health information exchange (HIE) is the electronic transfer of clinical, administrative, or other information necessary for the delivery of health care across diverse systems or organizations.⁶ Most often HIE takes the form of an information system transaction mediated by technical standards, such as the electronic reporting (or pushing) of a result from a laboratory information management system to an electronic health record (EHR) system using the Health Level Seven (HL7) messaging standard. Other times HIE involves querying (or pulling) information from another system, such as the transmission of a summary of care record using the HL7 Consolidated Document Architecture (CDA) standard outlined in meaningful use (now referred to as Promoting Interoperability) criteria for Medicare and Medicaid providers.^7–9

Event notification is a form of HIE involving the electronic reporting (or pushing) of information pertaining to a clinical event from 1 provider to another facilitated by a messaging standard. Notification usually pertains to acute care events (eg, hospitalization, emergency care), and the notifications are typically sent to primary care providers responsible for coordination of care.¹⁰ Recent systematic reviews of HIE do not include any studies on event notification.^11,12 To date there has been just 1 quantitative study of event notifications sent to ambulatory providers for acute care events over a 3.5 year period using a regional HIE network.¹³ The study found a statistically significant 2.9% reduction in 30-day readmissions observed during the intervention period. While promising, the evidence represents a single before–after cohort study. More robust evidence on the benefits of event notification services (ENS) as a form of HIE for care coordination is needed.

OBJECTIVE

The objective of this study is to examine the effectiveness of ENS on health care delivery processes and outcomes for older adults who experience acute care events. Given the ability of HIE to facilitate access to patient information, especially after a handoff or transfer of care, we hypothesized that HIE would impact care coordination activities,¹⁴ including timely follow-up and reintegration into primary care following an acute care episode. We further hypothesized that HIE would reduce readmissions.

MATERIALS AND METHODS

Study design

To examine the impact of ENS alerts sent to primary care teams following acute care encounters, we conducted a concurrent cohort study embedded within a 2-site, prospective cluster randomized trial. Older patients (age ≥65 years) were assigned to ENS alerts in 2 study intervention arms: (1) one in which electronic notification was sent to the patient’s primary care team following an acute care event (ENS only), and (2) one in which electronic notification was sent to the patient’s primary care team and patients received a 30-day postevent care coordination intervention delivered by social workers (ENS Plus). For the trial, assignment to an arm was based on computer-generated random numbers. A third arm consisting of patients who were not enrolled in either of the intervention arms, but who fulfilled basic eligibility criteria and had an acute care event during the study period, was identified as a no notification (eg, usual care control) group. This control group, however, was not randomly assigned. For the purposes of this analysis, the intervention arms were collapsed into a single group since both arms involved ENS alerts. This allowed us to compare both ENS groups to the nonintervention usual care controls. In Figure 1, we depict the resulting design used for this analysis. The protocol for the trial is published on ClinicalTrials.org (NCT02689076) and available online.¹⁵

Figure 1.

CONSORT diagram depicting the selection and recruitment of subjects into the study.

The study was reviewed and approved by the Institutional Review Board (IRB) of Indiana University as well as the VA Research & Development Committee at both the Indianapolis VA Medical Center and the Bronx VA Medical Center.

Study setting

We included patients who receive primary care at 2 US Department of Veterans Affairs (VA) medical centers. The first medical center is the James J. Peters VA Medical Center (JJP VAMC) located in the Bronx, New York. The JJP VAMC cares for more than 26 000 patients annually via a tertiary care facility providing comprehensive inpatient as well as outpatient care services in addition to 4 community-based outpatient clinics. The second medical center is the Richard L. Roudebush VA Medical Center (RLR VAMC) located in Indianapolis, Indiana. The RLR VAMC serves more than 62 000 patients annually and consists of a tertiary care facility providing comprehensive inpatient as well as outpatient care services in addition to 3 community-based outpatient clinics. Both medical centers also serve as teaching hospitals and regional referral sites.

Event notification services for non-VHA acute care

Upon utilization of non-VHA acute care, such as an inpatient admission or emergency department (ED) visit, an HL7 admission-discharge-transfer (ADT) message was electronically sent from the non-VA acute care facility to an HIE network connected to the 2 participating VA sites. In New York, the JJP VAMC partnered with the Bronx Regional Health Information Organization (Bronx RHIO). The Bronx RHIO is an HIE that includes hospitals, health systems, ambulatory care centers, individual physician offices, long-term care and home care, as well as community and other organizations. Collectively, these providers deliver the vast majority of the health care received by the borough’s 1.4 million residents, including over 95% of the borough’s annual hospital discharges, over 600 000 annual ED visits and 4.5 million annual ambulatory care visits. Providers at JJP VAMC have accessed the Bronx RHIO since 2008.

In Indianapolis, the RLR VAMC partnered with the Indiana Health Information Exchange (IHIE). IHIE is a nonprofit organization that operates the Indiana Network for Patient Care (INPC)—the nation’s largest interorganizational clinical data repository with participation from over 117 hospitals, 18 000 practices and over 50 000 providers and includes data on more than 17 million patients.¹⁶ IHIE captures data from >90% of hospitals in the geographic area served by the RLR VAMC. IHIE has exchanged data with the VA since 2011 when it began sending summary of care documents for veterans enrolled in the Veterans Lifetime Electronic Record program,¹⁷ now referred to as veterans HIE (VHIE).¹⁸

Each HIE network responded to the ADT message by notifying (or alerting) the VA that an enrolled patient visited a non-VA care facility. The intervention is summarized in Figure 2. Alerts were sent between March 2016 to December 2019.

Figure 2.

Information flow and technical architecture for the event notification service intervention.

As depicted in Figure 2, an ADT message within the HIE network triggered the ENS intervention. For RLR VAMC, IHIE transmitted the event notification to VA Direct Messaging, a component of the VA’s HIE program that utilizes the DirectTrust network. Study coordinators at RLR VAMC logged into VA Direct each day to retrieve the notifications. At JJP VAMC, study coordinators logged into the Bronx RHIO daily to retrieve new ENS notifications using that HIE network’s population health tools. Following notification of the non-VA acute care event, study coordinators at the 2 VA sites sent an internal electronic note within the VA’s EHR system to the veteran’s primary care medical home team (referred within the VA as a Patient-Aligned Care Team or PACT). This note, which becomes part of the veteran’s medical record, identifies the non-VA care facility and provides information on the reason for the visit or chief complaint. It further provides details on the external provider and how to contact them. Each note required acknowledgment by someone on the PACT team to signal it was read. The individual responsible for reading and acknowledging the note differed by PACT team. No notifications or notes were created for usual care arm patients.

Study participants

Veterans were included if they (1) received primary care at 1 of the participating VA medical centers; (2) were 65 years or older; (3) utilized any non-VA services (including nursing, lab, physician, pharmacy, and/or hospital services) within 2 years prior to enrollment; and (4) had a non-VA acute care event after enrollment. Participants were excluded if they were enrolled in the Geriatric Resources for Assessment and Care for Elders (GRACE) program at the RLR VAMC, a program that provides postdischarge services to veterans following a VHA inpatient encounter, or if they were enrolled in hospice.^19,20

Outcome measures or evaluation criteria

We hypothesized that ENS alerts to the PACT team would result in timelier follow-up with the veteran postdischarge from the non-VA facility. We further hypothesized that, due to timelier follow-up by VA primary care, all-cause readmissions and ED visits would decrease. Therefore, our main outcome measures included (1) timely PACT team follow-up as defined by a phone call when the PACT team spoke with the patient within 7 days and/or an in-person visit within 30 days, and (2) readmission as defined by a VA and/or non-VA hospital admission or ED visit within 90 days after the initial, non-VA hospitalization or ED visit discharge to the community. We expressed each variable dichotomously in the analysis, indicating whether timely follow-up or readmission occurred within the specified time periods.

Methods for data acquisition and measurement

Data on non-VA admissions and ED encounters were captured from the logs of HIE-based alerts sent to the medical centers during the study period. Patients’ baseline information was collected by trained research assistants when a patient was enrolled. Baseline information included age, sex (dichotomous, as recorded in the EHR), race/ethnicity, co-insurance (Medicare, Medicaid), presence of chronic conditions in the medical record 1 year prior to enrollment, Charlson comorbidity index measured during the year prior to enrollment, and site of enrollment (Bronx vs Indianapolis). All veterans received benefits from the VA, and many patients were eligible for co-insurance through Medicare. Furthermore, some veterans met eligibility criteria for Medicaid co-insurance. Outcome data were captured 30 days following a non-VA acute care event by trained research assistants using both the VA’s EHR system and data from both HIE networks.

Using EHR and HIE records, study team members extracted details of all VA and non-VA admissions and ED visits within 30 days of each eligible non-VA encounter during the study period. They further examined the notes entered by PACT team members to ascertain whether the veteran was contacted by the VAMC following the non-VA discharge. Outcomes at 30 days were used, because any impact of the care coordination intervention that some in the ENS group received would not influence or bias readmissions during that time window.

Methods for data analysis

Baseline characteristics of patients were assessed using descriptive statistics. Comparisons in baseline characteristics between the 2 arms were examined using χ² test or Fisher exact test for categorical variables and t-test for continuous variables. Since the units of randomization were primary care teams rather than veterans, there was potential for imbalance between groups in terms of baseline covariates due to clustering. Group comparisons were analyzed using PROC GLIMMIX for outcomes assuming a random intercept and binary distribution while accounting for clustering within primary care teams. P values and odds ratios were calculated for comparisons between groups, adjusting for baseline group differences in age, race/ethnicity, and Medicaid status. A significance level of P < .05 was adopted for all analyses. Based on projections for total patients (N = 310) at the start of the study, our power to detect a primary outcome effect size of 0.33 was 80% with 2-sided alpha 0.05. Analyses were performed by using SAS statistical software (version 9.4; SAS Institute Inc, Cary, NC).

RESULTS

In Table 1 we present the baseline characteristics of study participants, stratified by group assignment. Overall the mean age of participants was 74.8 years, and the cohort was almost exclusively male (98.0%) which is typical in VA-based studies. Enrollment site and the type of non-VA acute care received during the study period was balanced, with no differences between arms. Age did vary by arm, with the usual care group mean age younger (72.5 years) than the notification group (76.9 years, P < .01). More non-Hispanic black veterans were in the usual care group (30.4%) compared to the notification group (17.1%, P < .05). The notification group contained more veterans with Medicaid coinsurance (9.0%) than the usual care group (3.7%, P < .05).

Table 1.

Baseline characteristics of study participants by group

	Total		No Notification		Notification
	(n = 393)		(n = 191)		(n = 202)
Characteristic	No. (%) or Mean ±SD		No. (%) or Mean ±SD		No. (%) or Mean ±SD		P value
Age, years	74.8	±7.8	72.5	±7.1	76.9	±7.9	.002
Male gender	385	(98.0)	186	(97.4)	199	(98.5)	.430
Race/ethnicity							.036
Non-Hispanic White	236	(61.6)	101	(54.9)	135	(67.8)
Non-Hispanic Black	90	(23.5)	56	(30.4)	34	(17.1)
Other (including Hispanic)	57	(14.9)	27	(14.7)	30	(15.1)
Insurance type other than VA benefits
Medicare	357	(91.3)	180	(95.2)	177	(87.6)	.003
Medicaid	25	(6.4)	7	(3.7)	18	(9.0)	.033
Chronic conditions
Charlson Comorbidity Index	1.5	±1.9	1.7	±2.2	1.3	±1.6	.145
Chronic pulmonary disease	81	(20.6)	44	(23.0)	37	(18.3)	.232
Congestive heart failure	50	(12.7)	30	(15.7)	20	(9.9)	.049
Diabetes	156	(39.7)	73	(38.2)	83	(41.1)	.588
VA use in year prior to enrollment
Total # VA ED visits	0.7	±1.8	0.8	±2.1	0.6	±1.5	.447
VA hospitalization	58	(14.8)	33	(17.3)	25	(12.4)	.181
Site of enrollment							.465
Bronx, NY	190	(48.3)	101	(52.9)	89	(44.1)
Indianapolis, IN	203	(51.7)	90	(47.1)	113	(55.9)
Index non-VA acute care type							.739
Hospital admission	176	(44.8)	87	(45.5)	89	(44.1)
Emergency department visit	217	(55.2)	104	(54.5)	113	(55.9)
Average length of stay for index non-VA encounter, days	2.1	±3.6	2.0	±3.3	2.2	±4.0	.623

Note: χ² or t tests results are reported with P values adjusted for clustering within primary care teams, the unit of randomization used in the trial.

In Table 2, we present timely follow-up outcomes after the index non-VA hospitalization or ED visit. Compared to usual care, veterans whose primary care team received notification of their non-VA acute care encounter were 4 times more likely to have phone contact within 7 days (AOR = 4.10, P < .001) and 2-times more likely to have an in-person visit within 30 days (AOR = 1.98, P = .007) with a primary care provider (PCP, physician or nurse practitioner) after discharge. When follow-up was expanded to include registered nurses on the primary care team, veterans whose team received acute care notifications were almost 3 times more likely to have phone contact or in-person visit (AOR = 2.65, P < .001) within 30 days of the acute care event.

Table 2.

Timely follow-up by the patient’s VA primary care team following a non-VA acute care encounter, stratified by group

	Total		No Notification		Notification
	(n = 393)		(n = 191)		(n = 202)
Contact Outcome	No. (%)		No. (%)		No. (%)		AOR (95% CI)*		P value*
Follow-up with primary care physician or nurse practitioner
Phone contact within 7 days	81	(20.6)	19	(10.0)	62	(30.7)	4.10	(2.23, 7.55)	<.001
In-person visit within 30 days	111	(28.2)	39	(20.4)	72	(35.6)	1.98	(1.21, 3.24)	.007
Follow-up with primary care physician, nurse practitioner, or registered nurse
Phone contact within 30 days	134	(34.1)	47	(24.6)	87	(43.1)	2.39	(1.45, 3.94)	<.001
In-person visit within 30 days	117	(29.8)	42	(22.0)	75	(37.1)	1.98	(1.21, 3.23)	.007
Phone contact or in-person visit within 30 days	183	(46.6)	64	(33.5)	119	(58.9)	2.65	(1.68, 4.19)	<.001

^*Adjusted for age, race/ethnicity, and Medicaid status at baseline, and accounting for within primary care team clustering. Reference group = No Notification.

In Table 3, we present the findings for all-cause rehospitalization or ED utilization within 30 days after the index non-VA hospitalization or ED visit. With respect to hospitalizations, there is no significant difference between veterans in the usual care and the ENS groups. This is consistent across VA and/or non-VA hospital admissions within 30 days. With respect to ED encounters that did not result in hospitalization, veterans whose primary care teams received ENS notifications appeared to be twice as likely to have an ED encounter within 30 days. However, these differences are not statistically significant (AOR = 2.10, P = .068). When all secondary utilization events, VA or non-VA, are pooled within 30 days, we did not find a significant difference between the 2 groups (P = .057).

Table 3.

Hospital and emergency department utilization within 30 days following a non-VA acute care encounter, stratified by group

	Total		No Notification		Notification
	(n = 393)		(n = 191)		(n = 202)
Utilization Outcome	No. (%)		No. (%)		No. (%)		AOR (95% CI)*		P value*
Hospitalization within 30 days
VA admissions	24	(6.1)	14	(7.3)	10	(5.0)	0.93	(0.36, 2.39)	.875
Non-VA admissions	31	(7.9)	13	(6.8)	18	(8.9)	1.47	(0.65, 3.34)	.358
VA or Non-VA admissions	50	(12.7)	24	(12.6)	26	(12.9)	1.27	(0.65, 2.46)	.481
ED visit (without hospital admission) within 30 days
VA ED visits	20	(5.1)	8	(4.2)	12	(5.9)	1.42	(0.50, 4.05)	.511
Non-VA ED visits	17	(4.3)	5	(2.6)	12	(5.9)	3.11	(0.93, 10.43)	.066
VA or Non-VA ED visits	37	(9.4)	13	(6.8)	24	(11.9)	2.10	(0.95, 4.67)	.068
Hospitalization or ED visit within 30 days, VA or non-VA	75	(19.1)	31	(16.2)	44	(21.8)	1.74	(0.98, 3.09)	.057

^*Adjusted for age, race/ethnicity, and Medicaid status at baseline, and accounting for within primary care team clustering. Reference group = No Notification.

DISCUSSION

In this study, we found that automated event notifications following an acute care encounter improved timely follow-up by primary care providers 2- to 4-fold, suggesting that ENS is a relatively low-cost, low-burden way to improve follow-up after a hospitalization or ED visit, and possibly improve care coordination during this high-risk period. Yet the ENS alerts did not translate into a reduction in hospital or ED admissions within 30 days of the acute care event. Given the importance of ENS to the VA and other health systems, this study provides guidance for future research on ENS needed for improving coordination and outcomes.

Given the ability of HIE to facilitate access to patient information, especially after a handoff or transfer of care, there exists a strong theoretical case for HIE to impact care coordination activities,¹⁴ such as reintegration into primary care following an acute care episode. A unique contribution of this study is our analysis of timely follow-up by primary care. We found that when primary care teams are notified of out-of-network acute care episodes, they respond by initiating contact with the patient via phone or by scheduling an in-person visit. In preimplementation conversations with primary teams, clinicians said they often learn about non-VA encounters weeks or months after the acute care encounter, if they learn about them at all. Our findings provide evidence that care teams will use ENS notifications to coordinate follow-up care. Furthermore, these findings are in alignment with a recent survey of providers who receive ENS alerts and report providing follow-up care for ED and hospital admissions.²¹

While promising, event notifications may not improve patient outcomes. We found that veterans in the notification group did not have significantly fewer secondary hospitalizations or ED visits within 30 days after an index event. These findings contradict a prior study examining ENS,¹³ which observed a 2.9% reduction in 30-day readmissions. The prior study used a pre-/post- panel design with Medicare fee-for-service beneficiaries, whereas we employed a cohort design with veterans at 2 VA medical centers. Our cohort possessed unique characteristics from the Medicare panel. For example, our cohort was almost exclusively male, whereas in the panel study only 31% of patients were male. Moreover, we included individuals whose index event was either hospitalization or ED visit as opposed to the prior study in which all patients were hospitalized at index. Population and design differences might explain some of the disparate results.

ENS is a form of directed HIE where information is “pushed” to clinicians. A longitudinal analysis of Medicare fee-for-service claims from a different New York HIE found that directed HIE was neither associated with a reduction in ambulatory care sensitive hospitalization nor unplanned readmission.²² Similarly, a large Danish trial involving older adults (65+) found that a single follow-up visit postdischarge was not associated with a reduction in all-cause or ambulatory sensitive condition readmissions.²³ Therefore, timely follow-up alone may not be sufficient to prevent readmissions. There may also be confounders in our trial as well as the prior research, such as how well event notifications are integrated into the clinical workflow, that necessitate further study to better understand when ENS might be useful to reduce secondary acute care utilization. We did not distinguish readmissions based on primary diagnosis. Given a population of frail elderly patients, it is possible some readmissions were not related to the index acute care event. Furthermore, one-half to two-thirds of patients did not receive follow-up within 30 days, even with ENS, which reduced the likelihood of impacting readmissions. This underscores the importance of designing workflows that ensure follow-up after ENS by primary care team members. It further suggests that ENS alerts might need to target acute care encounters with higher likelihood of readmission (eg, coronary artery disease, stroke, diabetic ketoacidosis). There are other potential confounders, such as provider caseloads and patient characteristics not measured in this study, that require further examination in future research.

Moving forward, event notification should be further studied as timely follow-up and reintegration into primary care following acute care events are an organizational priority for the VA and other health systems. In a survey of non-VA providers in a practice-based research network,²⁴ respondents reported poor communication with VA colleagues, and their interactions were perceived to be with a “system” rather than a colleague. Moreover, the VA is under pressure to expand access to non-VA care for veterans following legislation from Congress.²⁵ Similarly, hospitals who treat Medicare and Medicaid patients are under pressure to implement event notifications to primary care following discharge. In its Interoperability and Patient Access final rule (CMS-9115-F), the US Centers for Medicare and Medicaid Services (CMS) requires hospitals, including psychiatric hospitals and critical access hospitals, to send electronic ADT notifications “to another health care facility or to another community provider or practitioner.”²⁶ As ENS is implemented across health systems, it will be important to further understand its impact on care coordination and patient outcomes. For example, the quantitative analysis in this article is not sufficient to explain why ENS reduced readmissions in Unruh et al¹³ but did not impact outcomes in this trial. Moreover, future research should consider additional processes as well as proximal outcomes for patients that may be more closely linked to transitions of care. Qualitative methods should examine the implementation of ENS,²⁷ and future trials or quantitative studies should examine ENS used for coordination of care across a variety of populations beyond older adults. Future research should also help tease out under what conditions ENS best meets care team and patient needs to impact outcomes. Our study is designed to compare outcomes between those who received only ENS alerts and those who received ENS alerts plus the care transitions intervention, since care coordination services might influence downstream care processes and utilization. This is the next planned analysis for our team.

LIMITATIONS

This study has several limitations. First, the ENS intervention group included veterans who provided face-to-face consent to participate in a prospective trial aimed at improving care coordination, resulting in a selection of veterans for the ENS group who were older and had different race and insurance profiles than the usual care group. However, there were no differences in VA ED and hospital use between the groups in the year prior, suggesting that the groups had similar VA care-seeking patterns at baseline. Similarly, there were no significant differences in chronic disease burden between the groups, suggesting that neither group was “sicker” than the other. In addition, we controlled for significant baseline differences in our adjusted models. Second, veterans are unique from nonveteran populations, as they tend to be primarily male and possess VA health benefits beyond traditional insurance plans, among other factors. In an analysis performed before the end of the trial, we found that non-VA events were associated with the veteran’s income, residential distance from the VA medical center, and self-reported regular use of non-VA providers,²⁸ factors that were not ascertained in the usual care control group and therefore not controlled for in this study. These factors, as well as other unmeasured factors may have influenced non-VA usage and outcomes. Third, it is possible that the degree of PACT responsiveness to ENS could have been impacted by the number of study patients in that PACT. However, if anything, this would have resulted in a diluted observed effect because, when a PACT had multiple patients in the study, it was often the case that it had patients in both the ENS and the usual care groups. Finally, we examined all-cause readmissions, which did not discriminate based on the index admission diagnosis. Some readmissions were likely independent of the original acute care episode, such as trauma after a fall resulting in an ED visit even though the original inpatient encounter was for pneumonia. Despite these limitations, the study employed a robust design and captured data from 90%–95% of non-VA providers in the markets where the veterans lived.

CONCLUSION

Event notifications sent to primary care teams following acute care encounters increase timely follow-up for older patients postdischarge, improving care coordination. Yet some veterans still did not receive timely postacute follow-up, and timely follow-up did not directly translate into reductions in secondary hospitalizations or ED visits. While promising, additional study of the impact and implementation of ENS alerts are necessary to optimize their use in support of care coordination and population outcomes.

FUNDING

The project described is supported by Merit Review Award Number I01 HX001563 from the US Department of Veterans Affairs (VA) Health Services Research & Development Service of the VA Office of Research and Development. The study protocol was initiated by the investigators; the funder did not play a role in drafting or editing the protocol. The protocol and opinions are solely the responsibility of the authors and do not necessarily represent the official views of the VA.

AUTHOR CONTRIBUTIONS

BED and KSB conceived of and designed the study, and BED wrote the initial draft of the manuscript. KMJ supported data management and analysis. The following authors reviewed, commented and participated in revisions of the manuscript: ALS, VMG, KMJ, NSK, JM, CCS, and KSB. BED synthesized feedback and finalized the manuscript as well as the figures. All authors read and approved the final manuscript.

DATA AVAILABILITY

The data underlying this article cannot be shared publicly to protect the privacy of individuals included in the study. A limited data set could be created and shared pursuant to a Data Use Agreement (DUA) appropriately limiting use of the dataset and prohibiting the recipient from taking steps to identify or re-identify any individual whose data are included on reasonable request to the corresponding author.