Tailoring Clinical Decision Support for Rural Hospitals: a qualitative pilot study of accessing infectious disease expertise for Staphylococcus aureus Bacteremia

Abstract

Background:

Clinical decision support tools (i.e. electronic alerts) can improve outcomes for patients with Staphylococcus aureus bacteremia (SAB) in hospitals with on-site infectious disease (ID) specialists. However, many rural hospitals lack on-site ID consultants and/or electronic health record (EHR) interoperability, presenting challenges for SAB patient care.

Methods:

We conducted qualitative interviews with rural hospital employees about SAB management processes and needs, and elicited feedback on an EHR alert used at a large urban hospital with on-site ID specialists. We used a rapid qualitative inquiry process to identify interview themes.

Results:

Rural hospitals had wide variation in existing availability of ID consultations and processes for seeking consultations. Participants provided suggestions for tailoring and implementing an electronic alert encouraging ID consultation within rural hospital work systems.

Discussion:

Increasing ID consultations including through EHR alerts in rural hospitals has many barriers. An EHR alert providing first-line guidance on urgency, therapy, and communication may support guideline-concordant care for SAB within the unique workflows of smaller and more rural hospitals.

Conclusions:

Electronic alerts, while useful in hospitals with on-site ID specialists to encourage ID consultations, will need to be tailored and implemented in rural hospitals to achieve similar clinical decision support results.

INTRODUCTION

Staphylococcus aureus bacteremia (SAB) is a serious infection, with estimated 30-day mortality rates of up to 29%.¹ SAB can cause life-threatening complications such as infective endocarditis.²

Infectious diseases (ID) consults are associated with reduced mortality, morbidity/sequalae, relapses, and readmissions in patients with SAB.^3–5 However, many hospitals (particularly smaller and/or rural hospitals) lack on-site ID specialists, which poses a problem to optimizing care.^{6, 7} In these hospitals, laboratory staff, nursing staff, infection preventionists, and pharmacists play critical roles filling this void,⁸ including highlighting the criticality of laboratory results indicating SAB, guiding appropriate antibiotic choices, and prompting clinically appropriate follow-up diagnostic testing (such as echocardiography to detect endocarditis and repeating blood cultures to document clearance of infection).

Yet, ID consultation remains important for SAB management, especially when hospital staff have infrequent experience treating SAB due to low patient volume. ID consults can provide expertise regarding the more complex aspects of SAB management: identifying the source of infection, ensuring metastatic sites of infection are fully evaluated, identifying and treating sequelae (e.g. endocarditis), and ensuring appropriate longitudinal care (e.g., treatment and duration).^{4, 9}

Electronic health record alerts to prompt ID consultation have been shown to be effective interventions to improve timely consultation and maintain good patient outcomes compared with more time intensive interventions.^{10, 11} At our academic medical center, we have developed and implemented the S. aureus Bacteremia Electronic Reminder (SABER) EHR alert to promote ID consultations for SAB.¹⁰ In this case, the alert automates manual reviews and outreach previously conducted by the antimicrobial stewardship team to identify patients with S. aureus-positive blood cultures and encourage providers to seek ID consultation. SABER provides a pop-up EHR alert when a hospitalized patient has had a positive blood culture for 36 or more hours without an ID note placed or an ID consult ordered. The 36-hour window was added to allow time for provider autonomy in considering patient needs and care goals as well as seeking consultations, and to reduce unnecessary alerts for providers who planned to seek consults. The alert notes that the patient has SAB, that mortality and complication rates are decreased with ID consultations, and recommends the provider obtain an ID consult. Notably, the alert provides an option to directly place the consultation, as our institution has full-time on-site ID specialists available for these consults. This alert successfully automated the previous manual process, maintaining high rates of ID consultations and decreasing time to consult after implementation of the alert.¹⁰

Despite the success of SABER at our institution, it is unclear if and how it could be adapted to hospitals without full-time on-site ID specialists available for consult on patients with SAB. In particular, there remain knowledge gaps around the current work systems of in-person and telehealth ID consultation availability, processes, and EHR operability in rural hospitals.^{12, 13} Identifying needs and barriers is a critical step in improving ID consultation availability and use in rural hospitals.¹³ In addition, electronic alerts to encourage ID consultations that are effective in urban hospitals need to be tailored to rural hospital settings, including determining how it might integrate into existing ID consultation (or clinical decision support) formats when traditional, in-person consults are not available. To address these knowledge gaps, we conducted a qualitative pilot study to understand the SAB management work system and opportunities to integrate electronic alerts to encourage clinical decision support for ID consultations in hospitals in rural Wisconsin.

METHODS

For this pilot study we used a qualitative approach, guided by naturalistic inquiry, because little is known about this phenomenon.¹⁴ Our study met criteria for exempt human subjects research by the University of Wisconsin-Madison Institutional Review Board in April 2024 and participants provided verbal informed consent before interviews.

Participants

Healthcare personnel were eligible to participate if they worked at rural hospitals in Wisconsin and had firsthand experience in the surveillance, diagnosis, treatment, or management of patients with SAB. Our eligibility criteria were inclusive to capture the breadth of possible roles engaged in SAB management across diverse work systems, such as physicians, bedside nurses, physician assistants, pharmacists, and infection preventionists. We considered rural hospitals to be outside of major population centers in Wisconsin and Minnesota (e.g., Madison, Milwaukee, and Minneapolis-St. Paul), and in particular sought participants from critical access hospitals. Critical access hospitals meet certain criteria from the Centers for Medicare & Medicaid Services, in particular, they have fewer than 25 beds and are located in rural areas.¹⁵ Wisconsin has over 50 critical access hospitals to support its population living far from these major metropolitan areas.

We purposively sampled to obtain an array of roles across Wisconsin’s state-wide networks of rural hospitals, infection prevention, and antimicrobial stewardship personnel. We used direct invitations to contacts, email outreach across relevant mailing lists, as well as brief presentations at network meetings to share information about the study and seek volunteers. We also used snowball sampling, asking participants or other contacts within our networks to share information about the study with their contacts who may be eligible.

Data Collection

We developed a semi-structured interview guide using the Systems Engineering Initiative for Patient Safety (SEIPS) framework to understand healthcare system factors impacting the tailoring and implementation of an EHR alert to promote ID consultation for patients hospitalized with SAB at rural hospitals.¹⁶ In addition to asking about general SAB management and specialty consultation workflows, participants were asked for feedback on the SABER alert that was developed and implemented at our large urban academic medical center with on-site ID specialists [supplement 1].¹⁰ Participants were shown an image of the pop-alert as well as the algorithm for the alert to fire (or activate). The interviewer described this existing alert and then asked about options for tailoring and integration of this alert into local rural settings.¹⁰

Interviews were conducted via video conferencing between July 2024 and January 2025 by two qualitative health services researchers (JK and LL). Interviews lasted approximately 30 minutes, and were audio recorded and transcribed verbatim.

Data Analysis

Interview transcripts were analyzed using a rapid qualitative inquiry approach as previously described.¹⁷ Three analysts (interviewers JK, LL, and qualitative researcher with experience as an infection preventionist LM) independently summarized transcripts and met to discuss emerging themes. The team then mapped the themes to the SEIPS framework to evaluate barriers and implementation needs across the work system. Group discussions by the multidisciplinary team to identify and interpret emerging themes supported accurate interpretation of interviews. The team maintained an audit trail to describe decision-making and alignment on themes.

RESULTS

Participant Characteristics

Six participants working at six different hospitals (four critical access hospitals and two flagship hospitals for rural healthcare systems) volunteered to participate in interviews. The critical access hospitals had less than 25 beds and were located in rural counties in south central, southwestern, and western Wisconsin. The flagship hospitals were located in northern Wisconsin. While these hospitals themselves were located in small cities (population <150,000), the hospitals served as the referral centers for health systems with large geographic footprints across rural northern Wisconsin, including multiple critical access hospitals within each system, with no other major metropolitan areas (>500,000 population) within approximately 150 miles. Patients seen at these flagship hospitals are thus generally from rural areas within this part of the state, and are often admitted as referrals from the critical access hospitals within these health networks. These hospitals include services related to SAB management for smaller hospitals within these networks, for example, through providing laboratory services to identify S. aureus positive blood cultures. None of the critical access hospitals in our study were part of these larger networks covered by these flagship hospitals.

Interview participants included one laboratory manager, two infection preventionists, and three pharmacists with roles in identification, surveillance, and management of SAB. Themes regarding current SAB and consultation practices and possible use of the EHR alert within the participants’ rural hospital work systems were organized into three domains: current SAB management, tailoring the alert, and implementation of the alert (Table 1).

Table 1.

Themes and Exemplar Quotes from Interviews

Domain	Theme [SEIPS element16]	Quote
Current SAB Management	Variability in ID consultant process and formality [Organization]	“Initially it was more what we consider curbside consults… it was telephone. They would just tell you and you’d never see a note. Over time it became more formal. The notes came in. They were looking through things and accessed Epic. They had privileges. But it was not what I would normally expect with an ID consult group on-site, not as thorough of a workup.” [Interview (I)3-Pharmacist] “For SAB they really wouldn’t need to see the patient, so it wouldn’t be an e-consult. It would be more of a chart review, here’s my recommendations, and then they would write a note and put it in Epic for us.” [I1-Pharmacist] “We have this little mobile cart that goes from room to room… You can see [the ID specialist’s] face right on the screen… A lot of it is chart review. They do look at all the labs and the cultures and stuff. And then they actually visit with the patient, if they are able to visit with a coherent patient.” [I5-Infection Preventionist]
	Differences in Access to ID Expertise: Dependent on patient’s insurance [Organization] Variable access to same EHR as consultant [Tools & Technology] How recommendations are communicated [Tools & Technology, Tasks]	“For [Insurance 1] infectious disease consults, they’re able to see all of our imaging, all of the complete EMR picture. If it’s a [Insurance 2] patient, they can see some of the information, but where they probably lack is some of the imaging. They can’t see as well or in details. So if you get a [Insurance 2] consultation, it’s a little more difficult for them remotely to [see] an MRI or a CT scan, they might have difficulty.” [I2-Pharmacist] “Physician either calls [Healthcare System 1] direct, or they go through the [Healthcare System 2] number, based on where the patient would go or where their primary care is because we see a lot of patients that have both insurances.” [I4-Infection Preventionist]
	Variation in SAB experience and sense of urgency [People]	“You had mentioned the outcomes being much better for patients when there’s an infectious disease consultation. I think there’s probably some lack of awareness of that.” [I2-Pharmacist] “Depending on the time of day when they came in, you wouldn’t necessarily have that urgency. You know, we need to get them on primary, first-line antibiotics for this MSSA… there might be a delay..” [I4-Infection Preventionist] “A lot of frontline clinicians tend to presume these are uncomplicated…it was not recognizing that you need to kind of work backwards and see, where did this really start? Was it the central line they have, the ventricular device that was just put in a month before?” [I3-Pharmacist]
	Internal communication mechanisms [Organization]	“Getting the information sooner, as quickly as possible, because the onus is on the lab to call the provider. But the provider doesn’t necessarily come up to us [pharmacy] and say, hey, this person has got MRSA bacteremia, or anything like that. So we have to find it later or wait until rounds or something.” [I1-Pharmacist] “The lab is calling [the physician/care team]… so it will say on the lab reports, such and such tech called the nurse or the physician at such and such time. But I was not added to that. If it’s bad enough that you’re going to tell the doctor, I’d like to know too.” [I5-Infection Preventionist]
Tailoring the alert	Alert fatigue [People]	“People are just tired of clicking on things. And even if you make a compelling case, it’s hard to get buy-in from those folks.” [I6-Laboratory Manager] “We go through with all these pop-ups in Epic, and people just get fatigue. And that’s why it’s hard to pick and choose which things do you have a pop-up for or do you not.” [I4-Infection Preventionist]
	Agreement in value of automation, simple straightforward alert, brevity, and provider autonomy in treating SAB [Tools & Technology]	Automation: “I like it, and there’s a lot of automation involved. Right now our techs have to spend a lot of time being like, did I put the correct comment in this culture? Did I click all the right buttons? And that becomes tedious for them.” [I6-Laboratory Manager] Provider autonomy: “I like the 36-hour thing… I like the 96-hour termination thing. That’s pretty cool.” [I5-Infection Preventionist]*
	Guidelines for first steps/first-line therapies [Tasks]	“Things that might be useful include a link to guidelines, like the IDSA guidelines or similar, a list of basics – hey, we see this result, get an ID consult. In the meantime, here’s some empiric therapy options.” [I2-Pharmacist] “I’d probably start with the therapy. We got to get them on our first-line therapies.” [I3-Pharmacist]
	Encourage repeat blood cultures [Tasks]	“Something that the guidelines would recommend doing, repeating [blood cultures], until they’re negative every 24–48 hours. That could be another BPA [best practice alert], where you have a BPA fire or a reflex test of some kind where you have a reminder to the provider to reorder them again.” [I2-Pharmacist]
	Encourage identifying source of infection or considering complicating factors [Tasks]	“I think we still need to be thinking about location of infection or source. Sometimes just being aware that you need to know where this is coming from… If you’re not going to get an ID consult, that’s kind of how you got to be thinking.” [I2-Pharmacist] “Then you start thinking about, is there some source we’d like to find? Are we going to be seeding a new source? Do we have central lines in? And then start going down the list of, let’s rule out endocarditis, their ECHO.” [I3-Pharmacist]
	Integrating alert within current consult and patient care workflows [Tasks]	“So I see ID note in the chart, so that shuts off the BPA. For a critical access hospital with our caseload, there’s probably not going to be an ID note in there, so that probably won’t suppress the note from firing.” [I2-Pharmacist] “Does it hit them in their workflow appropriately? Like if I’m just trying to handle something real quick, I’m just going to click through this and miss it. Can you have it fire back to me at some time?” [I3-Pharmacist] “36 hours seems like a long time for me, but I get it. Obviously we’d like to intervene sooner than that.” [I1 – Pharmacist]
	Identify ways to streamline communication; engage antimicrobial stewardship team/pharmacist [Organization]	“Once you’d respond to this alert, is there a way for other people to see what they said?” [I1-Pharmacist] “Antimicrobial stewardship people should definitely be getting notified. That way at least, if the doctor is ignoring the popup, you got someone else that’s getting it, and they can hound the doctor and try to get them to get a consult.” [I2-Pharmacist] “Pharmacy is kind of the back wall where they catch it and they bring it back to the physician and say, you know, you really should address this.” [I4-Infection Preventionist]
Implementing the alert	Approval processes/bureaucracy [Organization] In house ownership of EHR (more rapidly changeable) or Large health system ownership of EHR (multi-step approval process)	“We have a whole committee here that prevents you from putting these sorts of pop-ups into the kind of immediate view of providers because people got grumpy about having pop-ups. And so if you try to do that you have to go in front of the pop-up committee and make your case as to why this has to occur..” [I6-Laboratory Manager] “We’re a critical access hospital, and anything we want to implement, it usually has to be approved by a big hospital. And so a lot of times, they’re not going to want to create something just unique for us. It has to be something they want to do for themselves.” [I2-Pharmacist]
	Tailor to local context and communication [Organization, People]	“Not just simply coming in with an alert but establishing it socially with clinicians and experts and then handing off into this. Like now we’re going to go to this [alert] so that we can focus on other things.” [I3-Pharmacist] “You can build a beautiful decision tree, but if the people involved in it go, I don’t really want to consult… that makes it hard to pull off.” [I6-Laboratory Manager]
	Awareness of need for consult [People]	“They either were always set up and knew the value of a consult, or something else came up that compelled them [such as penicillin allergy or patient not improving].” [I3-Pharmacist] “I would say 50% of the hospitalists really appreciate it [consults]. The other 50% think it’s very expensive. Doctors can be a little bit frustrated that somebody is in their territory.” [I5-Infection Preventionist]

Current SAB Management and ID Consultations

Participants described barriers to SAB management, including difficulties consulting ID with varying formats of consultation (Figure 1). Notably, Wisconsin has several health systems with different affiliated insurance partners, further complicating processes for seeking ID consultations and receiving recommendations even for different patients within the same hospital. The flagship hospitals had some ID support on-site, however, resources were limited and shared with smaller hospitals within their networks (for example, via telehealth consults). No critical access hospitals in our study had full-time, on-site ID specialists. However, all hospitals had pharmacist support, arguing for including them in the alert and/or clinical decision support workflow as advocates of ID consultation and to offer initial treatment recommendations.

Figure 1.

Variability in infectious disease (ID) consultant process and access across rural hospitals, including in formality (such as access to electronic health record) and presence of ID specialist (such as via phone or in-person)

Tailoring the Alert

Participants agreed that design features which minimized alert activation were highly desirable due to alert fatigue. These included a 36-hour delay between a positive blood culture result triggering the alert, protecting clinician autonomy to consult ID on their own, and automatic termination after 96 hours. Because of variability in provider knowledge of SAB and urgency as well as different processes and thoroughness of ID consultations at rural hospitals (Figure 1), participants recommended including preliminary advice in the alert regarding repeat blood cultures and first-line antibiotics. If clinicians opted not to consult ID, one participant recommended the alert prompt consideration of the bacteremia’s source and potential seeding of other sites. Participants also wanted to ensure they were alerted, in addition to the treating clinician, so that they could advocate for ID consultation. Finally, because some critical access hospitals only had access to curbside ID consultation, they needed a way to turn off the alert in the absence of a formal ID note, which would typically terminate it.

Implementing the Alert

Two implementation themes emerged: local consulting culture and administrative workflows. ID consultations were more likely to be seen as “somebody in their territory” (I5-Infection Preventionist) when specialists were off- rather than on-site. Participants encouraged strong plans to engage clinicians and promote a culture of consultation prior to implementing the alert. Implementing would also require working with highly variable administrative procedures. Only one critical access hospital was able to develop and implement EHR alerts fully in-house. Others were dependent on their larger healthcare system and felt they had limited input. Some required local committee approval due to excessive alert fatigue.

DISCUSSION

In this pilot study we aimed to understand current work systems in place for SAB management and seeking ID consultations for SAB in hospitals in rural Wisconsin. Our study identified variability in access to and process for seeking ID consultation in rural hospitals, as well as barriers and facilitators of EHR-based alerts for clinical decision support to improve guideline-concordant SAB management.

Individuals from rural areas are at increased risk of poorer health outcomes than those from urban areas, with higher morbidity and mortality from a wide range of conditions in rural communities.^{18, 19} Disparities are also seen in SAB outcomes. Treatment failure was significantly more likely in rural pediatric patients with SAB than urban patients, and rural pediatric patients had higher rates of infection with methicillin-resistant S. aureus as well as endovascular infections.²⁰ Studies have shown higher socioeconomic status patients and patients seen at higher complexity hospitals with SAB are more likely to receive an ID consultation.^{20, 21} Increasing ID consultation rates for SAB in rural hospitals is thus a major area for improvement and presents an opportunity to address possible disparities for patients at these hospitals.¹²

However, while ID consultations are an evidence-based practice to improve guideline-concordant care and outcomes for patients with SAB,^{3, 22, 23} we have described the challenges of seeking ID consultations in rural hospitals particularly when there is no on-site ID specialist (Table 1, Figure 1). Telehealth and remote ID consults are options to connect patients in rural hospitals with ID expertise, however, as we found here and others have also reported there are barriers with these processes as well.²⁴ Notably for the participants in our study, there were wide ranges in consulting knowledge, options, and processes even for patients within the same hospitals (Figure 1). Thus, our study provides necessary first-hand perspectives of the challenges facing rural hospitals without on-site ID specialists in managing and treating patients with SAB.

Notably, while the effect of ID consultations has been demonstrated with bedside consultations, the effectiveness of other (e.g., telehealth) consultants on mortality has been mixed.²⁵ Additional research into the most effective formats and logistics for ID consultations is needed.²⁵ However, many sites here lack in-person options for consultants, and thus telehealth or automated informative alerts may serve as a viable option to promote guideline-concordant care in the absence of bedside consultations. Our work identified barriers that exist for a range of consultation formats in rural hospitals that will need to be considered in developing similar interventions to encourage ID consultations.

Our study is also unique in its evaluation of an EHR alert that had already been successfully developed and implemented at a large, urban hospital – thus our results provide specific insights into how work conducted in larger hospitals with on-site ID specialists to develop, implement, and test similar alerts may not be sufficient to overcome barriers at rural hospitals. Disparities in the adoption and utilization of EHRs between urban and rural healthcare settings have been well documented.^{26, 27} Barriers to EHR use have been reported on multiple levels including the healthcare system, hospital, and provider level.^{26, 28} While EHR alerts to notify providers of the need for ID consultations for SAB have been used successfully to decrease time to consultation,^{10, 11} we identified barriers impacting both general EHR use as well as use of EHR alerts specifically to encourage clinical decision support for ID consultation. These barriers may further complicate efforts to use EHRs to reduce disparities in SAB care and outcomes between rural and urban patients – particularly in regards to connecting patients in rural hospitals with ID consultations. It is critical to understand the current practices, needs, and options to adapt EHR interventions to rural hospital settings in order to effectively use the EHR as a tool to improve SAB care for patients at these hospitals.

Our findings suggest possible areas of investigation to reduce disparities in providing guideline-concordant care and improving patient outcomes for SAB in rural hospitals. An EHR alert providing first-line guidance on urgency, therapy, and communication may be an initial step to promote guideline-concordant care for SAB within the unique workflows of rural hospitals and other hospitals without full-time on-site ID specialists. For example, an alert for SAB could include brief points to notify providers of urgency (i.e., need for ID consultation and seeking the source of infection), need to begin appropriate antimicrobial therapy, and communication (i.e., need to notify relevant personnel (particularly AMS, pharmacy and infection prevention) for follow-up). Participants also noted the importance of engaging other personnel, including pharmacists and infection preventionists, in SAB management in rural hospitals. Many hospitals already have collaborative antimicrobial stewardship teams including pharmacy and infection control members, providing infrastructure for additional SAB support.²⁹ These personnel can emphasize the urgency of SAB, provide first-line treatment guidance, and/or encourage ID consultations. EHR alerts and work system modifications can leverage the critical role that these teams play in rural hospitals to bridge gaps in the absence of on-site ID specialists.

A theme throughout interviews was variability between sites, with no one-size-fits-all option and the need to tailor any alert to the local context. However, many of these rural hospitals were part of healthcare networks and thus impacted by larger hospitals’ activities and oversight, such as through access to ID expertise dependent on patients’ insurance and through EHR development and implementation being conducted through larger networks. Because of this bureaucracy, it may be difficult to tailor the alert itself to each facility. However, an opportunity to engage on-the-ground relationships should be a consideration in future work of adapting, implementing, and evaluating interventions to improve SAB management across hospital settings. For example, identifying a local SAB champion, such as a member of infection control, pharmacy, and/or antimicrobial stewardship teams, could serve as a point of contact for support to complement an EHR alert within local work systems.

This work was limited by the small sample size and focus on hospitals in rural Wisconsin. As a pilot study, this project provides insight into future research needs and directions. For example, the breadth of ID consultation options we found just among this small subset of hospitals (Figure 1) suggests that a large survey or needs assessment of a broader range of diverse hospitals would be useful to understand the most common arrangements and develop tools that fit these work systems.

In addition, the work is limited by the lack of physician participants who would be most likely to request ID consultations (such as hospitalists). While we specifically attempted to reach physicians to invite them to participate, we were unable to recruit any physician participants. Likewise, we were not able to recruit frontline staff nurse participants, despite their potential role in facilitating specialty care as a patient advocate. Additional research is needed to include participants from these roles given their criticality in SAB diagnosis, treatment, and management. Future development of any tools and processes to support ID consultations will also need to engage these stakeholders. Despite this limitation, our findings demonstrate challenges observed by other members of the work system in promoting ID consultations and the need for additional strategies to improve access to specialist expertise and guideline-concordant care for patients with SAB. Thus, this report will complement future work to engage physicians and frontline nursing staff to ensure alerts or programs to encourage ID consultation represent the wide range of roles in this work system. Given the existing presence and engagement of pharmacists in SAB management, formally engaging pharmacy and AMS teams in SAB management may be key for intervention.³⁰ Participants noted current pharmacist roles in identifying potentially complex SAB cases as well as need for additional testing to prevent sequalae, and streamlining this communication may improve SAB decision support even in the absence of on-site and readily available ID consultations.

CONCLUSIONS

Overall, participants reported that an EHR alert implemented at a large academic hospital to optimize ID consultations for SAB was useful and could support SAB management. However, because of the unique work systems in rural hospitals without full-time on-site ID specialists available for consultations, participants identified adaptations that would be needed to tailor and implement a similar alert in rural hospitals. Disparities seen in SAB management and outcomes in rural hospitals as well as in EHR implementations at rural hospitals compared to urban hospitals require attention to these areas to ensure alerts are effectively adapted to these work systems.