Patient-Generated Health Data in Surgical Site Infection: Changing Clinical Workflow and Care Delivery

Patient-Generated Health Data Use Requires Changes to Clinical Workflow

Whether in the form of subjective narrative of a present illness or a structured diary of symptoms over time, data presented to the clinician by the patient represent an integral element in the diagnosis and treatment strategy. However, relatively recent developments in wearable and mobile technology have changed the nature of the recording and communication of patient data. New streams of serial—even continuous—objective data (e.g., body temperature, heart rate, wound photographs) pose new challenges to provider data review and decision making, both because of their nature and their frequency. We posit in this article that the process of care delivery must evolve to accommodate the integration of new patient-generated health data forms (PGHD).

In our own efforts to modify clinical workflows to support the introduction of PGHD into post-operative, post-discharge care, we have encountered three areas that impact our implementations. These are (1) the divergent needs of patients and providers, (2) the need for incremental changes that account for the roles of nurses as well as those of patients and surgeons, and (3) the importance of using clinical utility to drive successful workflow change. Herein we share observations on the impact of those areas across several projects and state general principles useful in the introduction of PGHD in clinical care.

Patient-generated health data have been defined as “all health-related data created by patients outside the clinical setting to help address a health concern” [1], an overarching category inclusive of both objective patient-collected data and subjective patient-reported outcomes (PROs). Examples of PROs include data obtained through questionnaires, and examples of PGHD include sensor, image, and self-reported clinical observations such as blood pressure or temperature. However, PGHD has sometimes been defined to exclude PROs, especially when PROs are perceived as primarily being collected in a clinical setting [2]. The distinction based on the point of collection may be misleading, because we have observed the data commonly collected from patients about surgical site infection (SSI), which is certainly a valid health concern, to include a mix of both objective and subjective data. For example, while surgical site images and body temperature are objective data, the documentation of patient observations of the surgical site such as erythema, presence and color of discharge, and degree of relative warmth compared with surrounding tissue are subjective. We use PGHD as an overarching term for these data.

Changes to clinical workflows that support the incorporation of PGHD into post-surgical, post-discharge monitoring for SSI take place in the context of broader and ongoing changes to the delivery of healthcare. Increasing patient engagement is part of cultural shift to patient-centered care, which is outlined in an accompanying article in this issue [3]. This shift, combined with the broad goals of improving both clinical care and public/population health will drive a change in role for many patients from passive recipients of healthcare to active providers of health data and preferences that inform their care. And that change in the role of the patient, including the need to consider new data types and new communication channels, will require changes in the surgical care team's workflow, particularly after the patient has left the hospital after surgery. Offering an on-call phone number on discharge paperwork will no longer be sufficient to provide optimal post-operative, post-discharge surgical care.

Listening to the patient's “voice” (their experience of their illness and their concerns) is essential to providing patient-centered evidence-based care [4]. In cancer care, Berry et al. [5] have extended the concept of voice to include patient priorities and values, which are important for shared decision-making and effective engagement. Eliciting that voice and ensuring that the patient's voice is accurately and consistently conveyed to a diverse, multispecialty, interprofessional care team is facilitated by technologies that support both the organization of information from patients, and its rapid and consistent communication. Increasingly, patients use mobile, or smartphone-based apps through which data are either transmitted to their providers or integrated with the electronic medical record. However, despite a growing number of apps, as well as specialized sensor devices (e.g., Bluetooth-enabled scales, “smart” wound dressings, and blood pressure cuffs) used to collect PGHD, it remains unclear how best to incorporate this information into clinical care [6]. The converging trends of patient engagement and PGHD suggest that patients may increasingly choose to retain and manage those data over time, just as they increasingly track other elements of personal wellness and health, introducing further complexities to clinical workflows.

Health Information Technology and Clinical Workflow: A Complex History

Health information technology (HIT) implementations have long been recognized as creating substantial disturbances to clinical workflow, sometimes uncovering ambiguities in practice, and sometimes creating threats to patient safety as a result of unintended consequences to established clinical processes. To understand the challenges of HIT interventions for SSI, it may be useful to examine analogous challenges encountered with the implementation of Computerized Provider Order Entry (CPOE), a frequently studied HIT intervention that inherently requires changes in physician workflows.

A 1993 report on an early and problematic CPOE implementation described objections by resident physicians sufficient to cause a temporary halt to CPOE after its implementation at a large academic center [7]. Substantial safety issues with CPOE were documented as early as 2005, with reports highlighting the impacts of new health IT clinical order entry systems, attributing errors specifically to usability and training issues [8], and to inflexible medication logistics combined with a need for emergent treatment [9]. To highlight the lack of “guardrails” in CPOE systems, one team categorized 63,040 self-reported, CPOE-attributed medication errors to build a set of 21 scenarios reflecting classes of inpatient and outpatient errors. These scenarios were tested against 16 electronic medical records, and 375 deliberately entered erroneous orders, 79.5% of errors were entered easily or with only minor workarounds [10]. These broad issues—usability, training, inadequate decision support, and inadvertent inhibition of urgent clinical intervention—all are relevant to the development and implementation of new HIT tools to address SSIs.

Yet, as these reports of problems with CPOE and computerized clinical decision support were published, others described substantial benefits of CPOE. In 1998, Bates et al. [11] showed a 55% decrease in “non-intercepted” (administered) serious medication errors, and an 85% decrease in preventable adverse drug events. Specific to surgical patients, a study of CPOE showed no difference in medication error rates in a six-month period prior to, and in the two six-month periods after CPOE implementation [12]. However, the study did show improved efficiency, with a decrease from 41 minutes for the entry and transcription/transmission of a manual order to approximately 0.5 minutes for CPOE entry and transmission. This highlights the benefit of electronic capture of information from the primary source, and the potential for its rapid dissemination. It may be that the electronic collection and review of PGHD in the setting of post-discharge incision monitoring will similarly benefit from similar efficiencies as a result of the rapid electronic transmission and dissemination of information.

Although it is important to be aware of the potential pitfalls of HIT, our belief is that the problems discovered in its implementation are not inevitable but are the results of suboptimal HIT design and implementation. Substantial benefits can accrue from the introduction of digital methods of information storage and exchange that engage both patients and providers. Moreover, using these technologies, health care organizations may improve efficiency, communication, satisfaction, and outcomes after discharge to home after surgery.

Experience with the Incorporation of PGHD in Clinical Care

Our personal experience with PGHD in surgery comes from the mobile Post-Operative Wound Evaluator (mPOWEr) system, the design and implementation of which we have previously described [13–15]. We have also implemented a variety of systems outside of surgery, in the domains of cancer [16,17], human immunodeficiency virus (HIV) [18], and chronic pain [19]. These similarly incorporate the structured collection of health information and organize and communicate the outpatient's experience to their providers. Moreover, we have participated in activities and initiatives at the health system level designed to create a more comprehensive and systematic approach to engaging patients and incorporating their out-of-clinic experiences into their care [20]. This has led us to the following observations, based on experience with workflows in several domains.

Divergent Needs of Patients and Providers: Discordant Expectations Matter

In developing a post-discharge app for SSI detection and monitoring, we began by engaging stakeholders in order to understand the needs and preferences of the intended users, both patients and providers [13]. Sanger et al. [14] interviewed both patients and providers (half of whom were surgeons), identifying discord and agreement in 10 thematic areas including, for example, communication and expressivity of information. Although all stakeholders expressed an interest in the prospect of enhancing information for clinical decision-making with PGHD (images in particular), patient and provider user expectations of frequency of contact and means of communication regarding data interpretation differed and were sometimes in conflict. Patients generally expected rapid feedback and preferred engagement via electronic communications over telephone, which they viewed as more disruptive. Providers and surgeons in particular were wary of the burden of reviewing a lot of “normal” data and the expectation of responding to text communications, even when inconvenient or disruptive to other clinical activities. Expectations also differed on structuring information: patients wanted the expressive power of free text, whereas providers wanted uniform, constrained responses. Interestingly, nurses preferred giving patients the option of a brief, “stage-setting” free text comment.

The published results of this work on divergent expectations include a set of 10 thematic design recommendations for mobile health (mHealth) in SSI designed to decrease discord and amplify agreement between the expectations of providers and patients. These design recommendations must be evaluated in the context not only of addressing expectations, but also for the impact they have on the work of providers, and on the tasks inherent in being a post-discharge patient. An important observation from our work has been that clinic staff—and nurses in particular—may serve to bridge the gap between patient's and surgeon's expectations regarding technology for SSI monitoring, much as they already deliver traditional therapies and support patients.

Building on Existing Processes: Incremental Change Matters

Sanger's work taught us to think more critically about the importance of alignment between patients and surgeons and also about impacts on healthcare provider roles. Although mHealth tools in SSI offer an entirely new workflow for patients, provider workflow may be best approached as adding enhancements and improved efficiencies to existing processes. Acceptance by the entire provider team is important for successful adoption. In implementing patient-centered tools for shared decision support in cancer, Berry et al. [21] has shown that the two strongest predictors of an mHealth tool by patients were reminders to use the tool and endorsement of the tool by their providers. For providers to endorse new types of health information in monitoring SSI and for patients to feel their effort is valued by the clinician, eliciting PGHD must include clearly conveyed expectations for timeliness and methods of communication. In three clinics that piloted use of our app (mPOWEr), we found that most patients who uploaded the app used it only once [22,23]. Serial use of the app for surgical wound monitoring over time was rare, but extended to as long as 30 days after surgery for those patients with surgical site complications. This pattern of app use was fostered by a clinic nurse champion who engaged directly with patients and encouraged surgical wound photo uploads to facilitate remote triage, rather than mandatory in-person clinical follow-up visits. Her role, in what has been our most successful site, was both to remind patients that their input was heard and valued and also to bridge patient and provider expectations around communication by managing the review and messaging for the surgeons. These roles are not a radical departure from traditional clinic nursing, but rather an extension of these tasks to include information and communication mediated by technology.

Nurse-led remote triage of surgical site problems with mHealth-mediated PGHD collection and communication has been demonstrated previously in outpatient pediatric burn wound care and post-cardiac procedures [24,25]. Workloads required for remote monitoring of cardiac implantable devices in heart failure have been estimated at less than 45 minutes per month for every 100 patients monitored [26], suggesting that task shifting from patient-initiated event-based outpatient care to nurse-led serial remote triage is feasible and practical. Rather than redirecting primary triage of surgical sites to surgeons, a focus on enhancing the data available to nurses for remote surgical site triage improves the quality of engagement with patients and reinforces sustainability. Whether via conventional telephone communications or newer means such as electronic health record patient portals, nurses are already on the frontline of post-discharge communications and are responsible for escalating information to physicians when appropriate. Recognizing and potentiating the fundamental work of nurses through leveraging technology is key to remote patient monitoring integration and may be one viable way to reconcile the discordant expectations of patients and surgeons.

A Story of Successful Clinical Use of PGHD: Clinical Utility and Efficiency Matter

Patient reported outcomes in HIV have had an 11-year history of successful use in the Centers for AIDS Research Network of Integrated Clinical Systems (CNICS), and that success has been driven in large part by a strong emphasis on the clinical utility of the information collected, rather than a focus on its research value. Figure 1 illustrates a representative, patient-specific, provider summary of PRO data used in a clinical visit. Funded by the National Institutes of Health, CNICS is an R24 research resource, and its value depends on it being not only an accurate and complete source of HIV data, but also on its comprehensiveness [27]. Yet, one of the authors (W.B.L.) has led CNICS' technical development since its inception, and during that time has observed the CNICS network leadership consistently reject requests to collect data that are not clinically useful, and focus on minimizing the number of questionnaire items, with a site-specific session goal of between four and 12 minutes. One important measure of the success of CNICS with PROs is that across eight academic healthcare organizations, from 2007 through March 2019, more than 80,000 computerized PRO sessions have been completed by more than 20,600 patients [28]. A second measure comes from qualitative research on provider perceptions PRO use across the network, which found that “Providers value same-day, electronic patient-reported measures for use in clinical HIV care with the condition that PROs are (1) tailored to be the most clinically relevant to their population; (2) well integrated into clinic flow; and (3) easy to interpret, highlighting chief patient concerns and changes over time” [29]. Integration into clinic flow is highly variable, with different sites electing different target times for sessions, locations for survey administration, device choices, levels of integration with the electronic medical record, methods of providing results to clinicians prior to or during the clinical visit, and roles of clinic and research staff. Our experience has been found to be congruent with that of PRO projects at four other centers [30].

OPEN IN VIEWER

FIG. 1.

Depicts the summarized, clinically relevant patient-reported outcome (PRO) measures used in the Centers for AIDS Research Network of Integrated Clinical Systems HIV PROs. This example shows a first-visit summary with critical values emphasized. Subsequent visit data would add repeat longitudinal measures.

The dissemination of this PRO initiative and its successful implementation across all clinics contributing to the research repository has been voluntary and has relied on site level enthusiasm to uncover and implement local strategies to integrate the system into diverse outpatient clinical care workflows for HIV. The system is flexible to accommodate different strategies at different sites. However, if providers did not find the data to be valuable and clinic management did not find some specific local path to acceptable efficiency, the PROs would not have adopted across the entire CNICS network.

New Technologies, New Opportunities

It is easy to forget that medical care, without HIT, is also suboptimal in many ways, with unacceptable rates of errors, challenges to provider efficiency, suboptimal patient follow-up and detection of complications, and structural barriers to patient–provider communication. New technologies and new perspectives on the relationship between providers and patients together are driving substantial changes in clinical workflow, patient engagement, and in what information is available for diagnosis, treatment, and improving the understanding of diseases. These new technologies support new modes of information gathering and new information types, including PGHD and PROs. They allow both patients and providers to enter and access information in the locations, and at the times that are most convenient to them—just as all of us now expect to conduct the many other transactions and communications in our daily lives.

We have focused on the importance of understanding the interaction between mHealth tools and clinical workflows, as provider dissatisfaction is an important and legitimate barrier to changes in practice. However, using technology to support patient–provider exchange of information in the post-surgical, post-discharge period has implications beyond the individual. The benefits of rapid dissemination of information from the primary source may have a significant impact on public health and population health workflows as well. Currently, both surveillance and infection control quality improvement activities within the healthcare setting depend upon infection preventionists investigating specific cases. Because of cost and human resource constraints, this process is typically conducted retrospectively using administrative data. Claims data are insufficient in detail to detect most infectious complications, and especially because more than 50% of SSIs arise after hospital discharge [31], this method may result in an underestimation of the true SSI rate. While participation in the National Safety Quality Improvement Project (NSQIP) can yield more precise tracking of surgical outcomes, the data include only 30-day outcomes of a subset of all surgical cases performed. Widespread, routine collection of primary data from patients who have left the hospital and are recovering at home has the potential to provide a new data source with greater detail to help understand the impact of SSI both on the healthcare organizations, and on the general public. Ensuring that patients are motivated to provide accurate information on SSI and its absence depends on patients perceiving that providing that information improves their care. Without acceptable workflows, and provider acceptance, the secondary use of these data to improve public health cannot be realized.