Tele-emergency utilization: In what clinical situations is tele-emergency activated?

Introduction

Tele-emergency (tele-ED) is an electronic, two-way, audio/visual communication service between a central emergency care centre (tele-ED hub) and a distant hospital emergency department (remote ED) designed to provide real-time emergency care consultation.^1,2 The hub ED clinical team often consists of a board certified emergency department (ED) physician and ED-trained nurses. In contrast, the remote ED provider is usually a primary care physician, nurse practitioner or physician assistant. Tele-ED can provide timely and high-quality emergency department services that improve rural patient care and support rural hospital staff during critical medical events.^1,2

In emerging technologies, early publications often focus on the feasibility of a specific technology application or on user satisfaction with it. This is certainly the case in the telemedicine literature. ³ We recently conducted a systematic review of the tele-ED literature and found only a handful of published studies evaluating its use to connect a tele-ED hub with remote EDs for assistance with the broad spectrum of patients who present there. ¹ Studies indicated that both remote ED and hub providers view consultations as improving patient care ⁴ and that this model of tele-ED care is comparable to traditional care in terms of patient outcomes.^5,6 Compared to usual care, tele-ED has been found to reduce unnecessary transfers from remote EDs to hub EDs.^4,7 Thus, the limited available research has demonstrated its feasibility and acceptance. However, we found no studies describing how often it is activated in usual practice and in what circumstances it is used. We became interested in the research question ‘under what conditions are remote ED personnel most likely to activate tele-ED?’ Data available for this study help to answer this question to further understand the value of tele-ED for rural hospitals.

Methods

Avera Health is an integrated health system, comprised of 31 owned, leased or managed hospitals, as well as clinics and other entities. The hospitals maintain separate medical staffs, bylaws and policies, but come together to develop evidence-based practices. These hospitals also share in an enterprise electronic medical record (EMR), Meditech, which links all Avera affiliated entities. Avera Health’s eEmergency programme is a tele-ED service with a hub staffed 24/7 by emergency medicine specialty physicians and registered nurses from Avera McKennan Hospital, a tertiary care hospital located in Sioux Falls, SD. This hub provides tele-ED services to remote EDs in South Dakota and seven surrounding states. The hub creates records in a tele-ED log every time the tele-ED system is activated and the log contains information specific to the tele-ED encounter (e.g. technical problems, patient disposition). The database for the current analyses was created by merging the tele-ED log data with the Meditech EMR system data. Meditech EMR data provided for these analyses involved all cases presenting with an ED encounter. These analyses focus on tele-ED cases from the period 1 November 2009 (the date of the earliest full month of tele-ED implementation) until 30 June 2012 (the date of the latest data provided for analysis). Five urban Avera Health hospitals have tele-ED implemented but use it in limited situations, often to facilitate transfers to Avera McKennan Hospital, and thus cases from these larger hospitals were excluded. Of the 51 critical access hospitals (CAHs) that had used the tele-ED service for at least six months, 21 CAHs also had the Meditech EMR system operational during this time. In these 21 CAHs, there were 60,193 total ED encounters. Of these, 1512 were ED encounters where tele-ED was activated, and 58,681 were ED encounters where tele-ED was not activated.

Nonparametric correlation analyses were used to compare hospital characteristics and rates of tele-ED activation. Logistic regression was used to model the likelihood of activation of tele-ED based on hospital characteristics, patient major diagnosis categories and patient age. Hospital characteristics examined included total ED encounters per year and total inpatient admissions per year. Patient major diagnosis categories were defined by ICD-9 CM codes. ⁸ Major diagnosis categories with fewer than 40 tele-ED activated cases were all aggregated into the ‘other’ category, which was used as the reference category in analyses. Major diagnosis categories with more than 40 tele-ED activated cases were circulatory, injury, mental, respiratory, symptoms and other. Chi-square test of proportions was used to assess if there was a difference in discharge destination between patients where tele-ED was activated and those where it was not activated. Logistic regression was used to model the likelihood of transfer based on activation of tele-ED and patient major diagnosis category.

Our primary research question was ‘in what clinical situations does the button get pushed?’ In this case, tele-ED service is literally activated by the push of a large red button located under a system video screen in the remote ED. To further understand factors related to tele-ED service activation, phone interviews and site visits were conducted with 46 clinicians and 35 administrators at 26 rural hospitals that had used these tele-ED services for at least six months. The telemedicine hub provided a list of hospitals that used the tele-ED service, and CEOs/administrators at 12 CAHs were interviewed over the phone in autumn 2011. In spring 2012, site visits were completed at eight hospitals, chosen to minimize travel time. In autumn 2013, on-site interviews were conducted at 13 additional hospitals, again chosen to minimize travel time in a different geographic region from the previous interviews. We used semi-structured interview guides developed after considerable conversation with Avera Health telemedicine service leaders. As shown in Table 1, some questions were asked of CEOs/administrators, some were asked of clinicians and some questions were asked of all interviewees. Interviews were conversational and probes were used for follow-up if responses warranted the need for additional information. Interviewees also had opportunities to add information. The interview guide and protocol were approved by the University of Iowa Institutional Review Board. Three co-authors, senior faculty experienced in interviewing personnel in rural hospitals, conducted the interviews. One co-author, a rural ED physician, conducted nearly all interviews with physicians. Interviews were recorded, transcribed and anonymized. Content and themes relevant to perceptions of tele-ED utilization were extracted by one co-author using a qualitative inductive approach and verified by a second co-author. The themes reported here were approved by the other two co-authors who conducted the interviews, and confirmed by the two co-authors who work daily with the tele-ED service. The six co-authors discussed patterns of use during multiple conference calls and reviewed the quantitative data and interview findings presented in the manuscript until agreement was met.

OPEN IN VIEWER

Table 1.

The questions asked in semi-structured interviews about tele-ED.

Interview guide – questions asked of CEOs/administrators

Question 1: tele-ED implementation decision

a. Why did the hospital decide to participate in Avera’s tele-ED programme? Probe: Were there particular motivating factors?

b. Does the hospital take the view that tele-ED benefits outweigh its cost? Probe: How so?

c. If you were speaking with another hospital about potentially using tele-ED services, how would you explain the value and what would you add as cautionary notes?

Question 2: tele-ED effect on physicians

a. What impact has tele-ED had on the physician work environment?

b. How does the organization receive feedback from physicians about tele-ED? Probe: What are some of the specific comments from physicians?

Question 3: tele-ED effect on staff satisfaction

a. How has tele-ED availability affected the nursing staff’s level of comfort in completing their tasks and caring for patients?

b. Has nursing staff satisfaction changed since tele-ED implementation? Probe: If so, in what way?

Interview guide – questions asked of both administrators and clinicians

Question 4: tele-ED implementation effects

a. Does tele-ED facilitate care coordination? Probe: If so, how?

b. Has the use of tele-ED affected the process of health care in the hospital? Probe: If so, how?

c. Has tele-ED affected the quality of care rendered? Probe: If so, in what ways?

d. Has tele-ED affected patient clinical outcomes? Probe: If so, in what ways?

Interview guide – questions asked of clinicians

Question 5: tele-ED use

a. Describe some specific instances of using the tele-ED system.

b. How does the tele-ED system affect your work?

c. When you have used the system, how satisfied have you been with the process and outcome? Probe: Has this changed over time? If so, how?

d. How is the decision made to activate tele-ED? Probe: Who makes the decision? Does this differ across situations? Does this differ across clinicians?

e. Are any protocols used for tele-ED activation? Probe: If yes, describe? For which conditions?

Results

Analyses were first conducted to examine the characteristics of the 21 CAHs using the tele-ED services. The number of staffed beds ranged from 15 to 25, with over half of the institutions having 25 staffed beds (the maximum permitted in CAHs). The number of full-time equivalent (FTE) staff ranged from 24 to 316 and averaged 113, annual admissions ranged from 112 to 1963 and averaged 502 and annual ED encounters ranged from 417 to 7633 and averaged 1957. On average, hospitals had 2.6 tele-ED encounters per month and 72.9 total ED encounters per month for an average 3.5% rate of tele-ED activation. Nonparametric (Spearman) correlation analyses found that there was no relationship between the rate of tele-ED activations and number of hospital beds or FTE staff. There was a statistically significant inverse relationship between the rate of tele-ED activation and total hospital admissions (r = –0.49, p = 0.02), total number of ED encounters (i.e. combined tele-ED and non-tele-ED encounters, r = –0.44, p = 0.04) and total ED encounters per month (r = –0.53, p = 0.01). In other words, higher tele-ED activation rates were associated with lower total admissions, total ED encounters and ED encounters/month. The rate of tele-ED activation in these hospitals did not change over time (r = 0.03, p = 0.87).

The second set of analyses was conducted to examine diagnosis characteristics and compare patients for whom tele-ED was activated to patients for whom tele-ED was not activated. Table 2 displays counts of the major principal discharge diagnosis ICD-9 CM categories. The table displays the number of tele-ED activations (with percent of total), total number of ED encounters where tele-ED was not activated (with percent of total) and the rate of tele-ED activation within each diagnosis group. Chi-square analysis was used to test if there was a difference in tele-ED activation between the major diagnosis categories. Tele-ED was more likely to be activated for patients with major diagnosis categories of circulatory, injury, mental and symptoms. It was less likely to be activated for patients with major diagnosis categories of digestive, endocrine/metabolic, genitourinary, musculoskeletal, nervous system, pregnancy/childbirth, respiratory and skin/subcutaneous tissue. Looking at more specific diagnosis categories with sufficient n, the highest rates of tele-ED activation were seen in cases with a principal discharge diagnosis of ischemic heart disease (22 activations/100 cases), poisoning (18 activations/100 cases) and cerebrovascular disease (17 activations/100 cases).

OPEN IN VIEWER

Table 2.

Principal discharge diagnoses and tele-ED activation.

Major category	Tele-ED encounters (n = 1512) count percent		Non-tele-ED encounters (n = 58,389) count percent		Tele-ED activation rate	p-value**
<not specified>	*	*	292
Circulatory	304	20.1%	2564	4.4%	0.11	<0.0001
Congenital anomalies	*	*	48	0.1%	0.02	*
Digestive	37	2.4%	3851	6.6%	0.01	<0.0001
Diseases of the blood and blood-forming	*	*	264	0.5%	0.03	*
Endocrine/metabolic	21	1.4%	1152	2.0%	0.02	0.02
Genitourinary	23	1.5%	2448	4.2%	0.01	<0.0001
Infectious	*	*	1075	1.8%	0.01	*
Injury	447	29.6%	14,103	24.2%	0.03	<0.0001
Mental	95	6.3%	1980	3.4%	0.05	<0.0001
Musculoskeletal	31	2.1%	4178	7.2%	0.01	<0.0001
Neoplasms	*	*	195	0.3%	0.00	*
Nervous system	21	1.4%	3668	6.3%	0.01	<0.0001
Perinatal	*	*	37	0.1%	0.05	*
Pregnancy/childbirth	11	0.7%	1043	1.8%	0.01	0.0003
Respiratory	63	4.2%	5803	9.9%	0.01	<0.0001
Skin/subcutaneous tissue	13	0.9%	1508	2.6%	0.01	<0.0001
Symptoms	400	26.5%	13,405	23.0%	0.03	0.0001
V code	23	1.5%	1063	1.8%	0.02	0.06

Logistic regression was used to model the likelihood of activation of tele-ED based on annual hospital admissions and ED encounters, major patient diagnosis category and patient age. The diagnosis categories included in the model were circulatory, injury, mental, respiratory, symptoms and other. Diagnosis categories with fewer than 40 tele-ED activated cases were all aggregated into the ‘other’ category, which was used as the reference category. The results of that analysis are shown in Table 3. Patients with circulatory diagnosis were 11 times more likely to have tele-ED activated than those with ‘other’ diagnosis (AOR 11.71, 95% CI: 9.67–14.17). Likewise, the injury, mental and symptoms diagnosis categories were also more likely to have tele-ED activated than those with ‘other’ diagnoses. Patient age, total hospital admissions and total hospital ED encounters were not statistically significant. The c statistic was 0.754, indicating good predictive accuracy of the logistic regression model. ⁹

OPEN IN VIEWER

Table 3.

Odds ratios of principal discharge diagnosis on activation of tele-ED, adjusted for patient age and hospital characteristics.

Factor	AOR*	95% CI
Circulatory disease	11.71	9.67–14.17
Injury disease	3.08	2.60–3.65
Mental disease	6.82	5.30–8.77
Respiratory disease	1.06	0.80–1.40
Symptoms disease	2.89	2.44–3.43
Patient age	1.00	0.99–1.00
Total hospital emergency department encounters	1.00	1.00–1.00
Total hospital admissions	1.00	1.00–1.00

The third set of analyses was conducted to examine discharge disposition. Chi-square analysis indicated a statistically significant difference in disposition proportions between cases where tele-ED was activated and those where it was not (chi-square = 18,262.4, p < 0.0001). Table 4 shows that tele-ED patients were more likely to be admitted to the local hospital, and much more likely to be transferred than patients where the tele-ED system was not activated. Tele-ED encounters were also more likely to be pronounced dead than non-tele-ED encounters. In contrast, non-tele-ED encounters were far more likely to be discharged from the ED than tele-ED encounters.

OPEN IN VIEWER

Table 4.

Discharge disposition and tele-ED utilization.

Discharge disposition	Tele-ED encounters (n = 1512) count percent		Non-tele-ED encounters (n = 55,713) count percent
<Not specified>	0		2968
Discharged	278	18.4%	45,131	81.0%
Admitted to local hospital	363	24.0%	9267	16.6%
Transferred to another hospital	819	54.2%	598	1.1%
Pronounced dead	51	3.4%	73	0.1%
Behavioural health	*	*	443	0.8%
Other	*	*	201	0.4%

Table 5 shows the three most frequent discharge dispositions (discharged, admitted to the local hospital or transferred) by major diagnosis category. Many of the major diagnosis categories have too few tele-ED encounters to draw meaningful inferences so those have been eliminated from the table. In the five major diagnosis categories (circulatory, injury, mental, respiratory, symptoms) with sufficient cases for each disposition type, a higher proportion of non-tele-ED encounters were discharged than were tele-ED encounters. The pattern of admission to the local hospital was mixed; it was more likely in cases with circulatory and mental diagnoses in non-tele-ED encounters, but more likely in cases with injury, respiratory and symptoms diagnoses in tele-ED encounters. A clear pattern emerged for transfers: in all five of these major diagnosis categories (circulatory, injury, mental, respiratory, symptoms), patients where tele-ED was activated were much more likely to be transferred than cases where tele-ED was not activated. In all five groups, the tele-ED rate of transfer was at least 45% and the non-tele-ED rate of transfer was never greater than 5%.

OPEN IN VIEWER

Table 5.

Encounter characteristics – discharge disposition, by diagnosis group.

	Tele-ED encounters			Non-tele-ED encounters
Major diagnosis group	Discharged	Admit local	Transferred	Discharged	Admit local	Transferred
Circulatory	4.3%	23.4%	61.2%	53.6%	39.3%	1.2%
Digestive	*	40.5%	48.6%	71.4%	24.6%	0.5%
Endocrine/metabolic	*	*	47.6%	54.3%	39.7%	*
Genitourinary	*	*	52.2%	75.0%	20.1%	0.8%
Injury	24.4%	12.5%	60.9%	87.9%	6.7%	1.0%
Mental	24.2%	15.8%	60.0%	52.9%	25.5%	4.8%
Musculoskeletal	38.7%	*	32.3%	85.1%	8.3%	0.9%
Respiratory	17.5%	30.2%	52.4%	73.9%	21.0%	0.3%
Symptoms	16.5%	35.3%	46.8%	74.2%	18.1%	1.3%
V code	56.5%	*	*	67.7%	6.9%	2.0%

Logistic regression was used to model the likelihood of being transferred among cases that were either locally admitted or transferred based on tele-ED activation and major diagnosis category. Transfer destination was modelled (with ‘local admission’ as the reference group) on tele-ED activation (‘not activated’ as the reference group), and major diagnosis category (circulatory, injury, mental, respiratory, symptoms and other, with ‘other’ as the reference group). The model suggests that (holding other variables constant) the odds of transfer for tele-ED activation cases are 32 times that of non-tele-ED activation cases. As shown in Table 6, the odds of transfer (holding tele-ED activation constant) are higher in those cases with injury, mental and symptoms diagnoses; the odds of transfer are lower in patients with respiratory diagnoses. The c statistic was 0.840, indicating strong predictive accuracy of the logistic regression model. ⁹

OPEN IN VIEWER

Table 6.

Adjusted odds ratios of tele-ED activation and principal diagnosis on discharge disposition.

Factor	AOR*	95% CI
Tele-ED activation	32.37	27.64–37.91
Circulatory disease	1.19	0.93–1.52
Injury disease	3.25	2.63–4.03
Mental disease	3.98	3.08–5.16
Respiratory disease	0.54	0.38–0.75
Symptoms disease	1.30	1.06–1.59

To further understand factors related to tele-ED service activation, we conducted phone interviews and site visits with 81 clinicians and administrators at 26 rural hospitals that used these tele-ED services. A qualitative inductive approach was used to identify in what situations it is used most, when and why clinicians use it or do not use it, barriers to clinicians using it, changes over time in the pattern of use and key incidents that affected its use (e.g. a particularly compelling outcome). Several themes related to tele-ED use were particularly uniformly expressed. First, interviewees were almost universal in describing how beneficial tele-ED services were to facilitate transfer when needed. Second, nurses particularly valued tele-ED hub personnel documenting care during emergent cases, relieving them to focus on patient care. Third, tele-ED hub involvement was highly valued during time-urgent critical cases such as chest pain, stroke symptoms, trauma, multiple patients needing simultaneous attention for serious injuries and while local physicians were en route to the ED. A fourth theme emerged in interviews at some but not all hospitals: that the tele-ED hub involvement was perceived as drastically improving outcomes for at least one local patient. Themes related to changes in use over time varied, with some hospital interviewees expressing immediate and continued enthusiasm and interviewees at other hospitals describing less resistance as familiarity grew and user satisfaction increased.

Discussion

The findings from the current analyses point to several situations in answer to our research question of ‘when does the tele-ED service get activated?’ First it is clear that the patient’s diagnosis is an important factor triggering tele-ED use. In particular, patients presenting at rural EDs with circulatory, injury, mental and symptoms diagnoses were significantly more likely to have tele-ED services activated. This was especially the case for patients with certain diagnoses. Specifically, tele-ED was activated for 22% of patients diagnosed with ischemic heart disease, 17% of those diagnosed with cerebrovascular disease and 18% of patients diagnosed with poisoning.

The second situation related to tele-ED activation involves the patient’s disposition. Patients for whom tele-ED was activated were much more likely to be transferred to another hospital than patients where the tele-ED system was not activated. Tele-ED encounters were also much more likely to be admitted to the local hospital than non-tele-ED encounters, who were much more likely to be discharged.

These findings suggest a conundrum involving tele-ED activation and discharge disposition. Namely, is tele-ED activated more when it appears a patient will be transferred? Or once tele-ED is activated is it more likely that the clinicians involved will determine that the patient needs to be transferred? Our quantitative data cannot definitely answer this question. However, our interview data with users of these tele-ED services illuminate some patterns. Interviewees uniformly described how beneficial tele-ED services were to facilitating transfer when needed. They described the tele-ED hub personnel coordinating all arrangements for transfer, permitting the remote ED clinical staff to maintain ‘hands-on’ care for the injured or seriously ill patient. This was especially valued because remote EDs frequently have limited clinical staff to handle critical situations such as serious trauma or multiple critically ill patients. Thus, when the remote ED clinical staff suspect that a patient needs to be transferred, they appear to activate the tele-ED service to help speed those arrangements.

The interviews also help elucidate the frequent activation of tele-ED related to certain diagnoses. ED staff are well aware of the need to urgently treat and/or transfer patients with ischemic heart disease and cerebrovascular disease. Interviewees described protocols in many remote EDs that included activation of tele-ED for all patients presenting with certain conditions. In fact, Avera Health provided guidance to remote ED personnel recommending tele-ED activation for all patients presenting with: acute head injury; advanced airway management – intubation; all resuscitations; advanced burn management; dizziness with unstable vital signs; multi system trauma evaluation and resuscitation; serious or complex medical emergencies; and shock of any etiology. While not specifically called out in this policy, it is likely that patients presenting with chest pain or possible stroke symptoms had tele-ED activated at many CAHs. Our quantitative data found that all patients diagnosed with these two conditions where tele-ED was activated were either admitted locally or transferred to another hospital, indicating that tele-ED activation occurred for patients needing serious, timely attention.

Another situation with a high rate of tele-ED activation was for patients with a mental diagnosis. When tele-ED was activated, 60% of patients with a mental diagnosis were transferred to another hospital. Multiple interviewees described valued assistance provided by tele-ED hub personnel to conduct assessments and find placements for patients needing hospital admission. This pattern illustrates the interaction between the patient’s diagnosis and discharge disposition in explaining use of tele-ED services by remote EDs.

The largest limitation of the current study is that the data do not include disease severity. Our data show that 81% of ED encounters when tele-ED is not activated were discharged from the ED rather than being admitted to a hospital. Clearly, tele-ED activation is not needed for the vast majority of ED encounters and the fact that so many were discharged is indicative of the frequent low-acuity diagnoses in ED cases. In contrast to this data limitation, a strength of the data is its generalizability to CAHs and rural hospitals nationally. The characteristics of the CAHs in this study are very similar to the national average for CAHs in terms of number of beds and staff, although the CAHs in this study average fewer ED encounters annually than the national average.^10,11 The pattern of diagnoses and discharge dispositions in the current data are also similar to those nationally in the National Hospital Ambulatory Medical Care Survey-ED (NHAMCS-ED) database for rural hospitals. ¹² Specifically, we compared the data in Table 2 on diagnosis categories between the non-tele-ED percentages in these CAHs and the non-metropolitan statistical area (MSA) percentages in the NHAMCS-ED data and found similar percentages. In addition, we compared the data in Table 5 on disposition categories in these two datasets and found similar percentages, with the exception that more patients presenting in Avera CAH EDs were admitted locally (16.6% vs 10.0%), with concomitantly fewer discharged or transferred (82.1% vs 88.4%). Thus, in comparing these two key variables in the current dataset with a nationally representative dataset, we believe there is evidence that the current findings on data from Avera CAHs are generalizable to patterns in EDs in rural hospitals nationally. In addition to the generalizability of the study setting, the qualitative findings are consistent with earlier studies showing that remote ED staff are satisfied with the process and interaction with the hub,^4,6,13 that remote ED providers view consultations as improving patient care,^4,6,14–17 that this model of tele-ED care is comparable to traditional care in terms of patient outcomes,^5,6 and that it enhances patient throughput and transfers^7,13 and reduces avoidable transfers. ⁴

Medical emergencies challenge any health care system, but more so in rural areas where time and distance to definitive health care are usually longer.^18,19 Rural hospitals, especially CAHs, have few clinical staff available during emergent conditions. ²⁰ Tele-ED services can provide invaluable support during times of emergency. ² Although only activated in 3.5% of ED encounters on average, our findings provide evidence that tele-ED activation is well reasoned and related to those situations when extra expert assistance is particularly beneficial.