A mixed-method exploratory evaluation was conducted with a convenience sample of 6 outpatient dialysis centers across the United States from August 2018 to April 2019. A separate initial pilot-facility visit was conducted to test the data collection tools and help better understand the logistical issues of a dialysis center (eg, where in a dialysis facility to stand to conduct observations but not interrupt patient care). The initial pilot facility data were not included in the analysis. The centers were recruited through ASN representatives (K.A.L., A.S.K., R.E.G., and others) via an online sign-up portal. The online sign-up asked about patient population, location, and willingness to host the team. Our goal was to include large dialysis organizations (LDOs), small dialysis organizations, for-profit and not-for-profit facilities, academic and nonacademic, and adult and pediatric units. If a center expressed interest, ASN representatives worked directly with leadership from each site and then with legal representatives and others to ensure that all concerns were addressed.

At all facility visits, the team included 3 human factors engineers, and a minimum of 1 expert in dialysis infection prevention, 1 nephrologist, and 1 nephrology nurse. The facilities included both pediatric and adult outpatient dialysis centers, those owned by large for-profit and not-for-profit dialysis organizations, and an academic center. For most of the participating sites, NTDS leaders volunteered their own facility; for some of the facilities, the NTDS member was also a dialysis organization leader and connected us with a facility within their organization. A connection was made with the nurse manager and medical director of the facility in advance of the visit, and ASN staff collaborated on details of the visit with the nurse manager. The facilities were notified in advance, and dates were planned with the facilities.

Structured observations were obtained using the SEIPS 1.0 model. We chose SEIPS because of its health care origins, and chose 1.0 as the simplest representation of the interconnected macroergonomic factors. Encounters were defined as a staff member entering the patient’s dialysis station to perform a task with the dialysis patient or with the associated equipment in the patient station. The “patient dialysis station” was defined as the immediate vicinity of the dialysis chair, chairside table, dialysis machine, and/or adjacent computer station. Definitions of the terms used in this approach are described in Table 1.

Several types of data were collected. First, structured and unstructured ethnographic observations were conducted on the treatment floor. Ethnography is a method used in social and behavioral sciences in which observations of individuals are gathered in their natural environment rather than in a laboratory. The goal of an ethnographic approach is to gain insight into how individuals interact in their natural environment. Second, qualitative information was gathered from staff conversations as a group or individually.

Finally, direct environmental data (eg, quantitative measurements of noise and lighting using consumer apps) were captured. The data collection focused on 4 infection prevention processes: environmental disinfection of the dialysis station, hand hygiene, injection safety (ie, injectable medication preparation and administration), and central venous catheter (CVC) and arteriovenous vascular access (fistulae/graft) care. These processes have been identified as central to infection prevention.²¹

During observations of these processes, we assessed and identified system components such as human factors, work system factors, and extrinsic factors (Table 2). Human factors are those issues that are intrinsic to humans and impact their performance. Here we highlight distractions (ie, interruptions, alarms), task stacking, and perceived differences between what items are considered “clean” and what are “dirty” and when. Work system factors are those that impact human performance but are not intrinsic to individuals. These factors are part of the design of the work system, including contributions of the physical environment, workflow scheduling, and policies and procedures. Extrinsic factors are those that are outside the control of the individual or work system.

Objective, quantitative observations, such as time on task, were recorded in the TimeCaT 3.9 application (HICAPPS SpA). Structured observations were also conducted specifically on the task of environmental disinfection using link analysis, which is a method to track the time and motion of staff. Unstructured ethnographic observations were manually recorded with paper and pencil.^11,16‘¹⁸ Routine care was observed at patient stations (such as catheter care). If observed, tasks not directly related to the 4 process areas of focus were also recorded (eg, adjusting dialysis machine settings, gathering supplies, completing routine patient assessments, documenting in the electronic medical record, silencing an alarm, or visualizing a patient’s blood lines). Observers identified system design factors that might influence staff performance on infection prevention tasks, thereby mitigating or escalating the risk of an avoidable infection. The observations were not intended to assess for “errors” per se but rather to gather objective and subjective information regarding the work-system of the facility.

To assess staff perceptions of safety and infection prevention, facility staff were engaged in semistructured conversations. Open-ended questions were posed to elicit system or work culture challenges and facilitators perceived by staff. During observations on the treatment floor, brief inquiries to the staff may have been made. These inquiries were semistructured and focused on staff-reported challenges or facilitators related to infection prevention.

The team conducted exit discussions with facility leadership and issued each a follow-up summary of our observations and the opportunity to discuss the report with the research team. These conversations were led by individuals with experience in quality improvement and patient safety (S.H.P., L.D.W.), with significant attention paid to how the feedback was framed and might he received. Facility reports were not binding in any way to participation in this study, did not include individual staff or patient identifying information, and were not shared publicly.

This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy (45 C.F.R. part 46, 21 C.F.R. part 56; 42 U.S.C. §241 (d); 5 U.S.C. §552a; 44 U.S.C. §3501 et seq.). Virginia Tech’s institutional review board determined this work to be not human subjects research (18–911).

Over the course of approximately 2.5 days per facility, a total of 157.5 hours of facility observations were conducted. Observations occurred across hours of facility operations, including early morning start and evening shift (Table 3).

Infection prevention in outpatient hemodialysis is highly related to staff behavior, specifically adherence to recommended hand hygiene, environmental disinfection, injection safety, and catheter care practices.⁶ This study does not offer tested solutions but a better understanding of how the system design influences normal infection prevention-related work.

Our findings suggest that there are many macroergonomic factors influencing human performance. This type of assessment can help to identify barriers and facilitators of infection prevention practices within dialysis facilities. System level observations demonstrated that interruptions, alarms, and task stacking occur commonly and can be barriers to the timely and thorough execution of infection control processes. The impact of a disruption is determined by the duration of the interruption, the complexity of the task being interrupted, the complexity of the distracting task, and the moment of interruption.^23–25 Distractions related to machine alarms and those related to task interruptions, which are pauses in the original task, have been previously described. These types of interruptions are known to increase errors.²⁶ Nuisance alarms can increase staff stress, unnecessarily increase the need for machine interaction and hand hygiene tasks, lead to alarm fatigue, and can harm patients by delaying the time to alarm response.²⁷ Task stacking may require substantial cognitive resources (visual, auditory, tactile, verbal, or spatial), which compounds the difficulty of the focused clinical task, especially if the task is being performed under pressure of time.²⁸

Dialysis-related infection prevention procedures require rigorous attention to complex tasks, manual dexterity, and attention to patient-related factors. Patient and dialysis shift turnover time was observed to be particularly busy, and time pressure was a common theme among staff comments. In this instance, the sheer volume of tasks and the perception of feeling rushed could influence individual caregivers’ ability to adhere to infection prevention tasks. Our observations suggest that there may be a mismatch between the time available and expectations for performance. One expert told us, “[This is] an ICU-type procedure, that would ideally be conducted in an inpatient sterile setting, being done outpatient with providers who often have much less training and expertise.” Human/system factors may be inadvertently aligned to test human limitations and thereby challenge compliance with the processes of infection prevention.

The study also showed differences in the design of the physical layout and the steps taken during our link analysis data collection. The difference could be the result of facility policies, layout, individual staff performing the procedures or other features. We caution over-interpretation of this finding as “optimal” or “suboptimal.” Rather, the result is that there are simply observed differences in practices across facilities.

System level observations and staff conversations also highlighted that external patient-related factors such as transportation issues and late arrivals (“Patients are slow to leave because of clotting or other things, and others arrive early. Some have to get taken off [dialysis] early … for transportation”), and internal in-center factors, such as scheduling, workflow, and facility design can exert pressures on strict compliance with infection control procedures. For example, although the time allotments between patients may appear to be adequate, they may be insufficient when viewed in the context of facility workflow, which in turn can inadvertently lead to staff workarounds that bypass infection prevention procedures. Future studies should further consider the benefits of patient and clinician stories, which might provide further insight into how work is performed on a daily basis.

Several of the facilitators for infection prevention identified in this study were related to teamwork and leadership. Although not measured here, previous work has highlighted the importance of frontline leadership’s engagement in safety-related behaviors and the resulting increase in follower safety behavior.²⁸ In addition, a culture of safety is highly related to daily engagement and communication, and leaders asking the frontline staff to discuss ongoing issues.^6–16 We did not observe these behaviors occurring consistently across all facilities in our exploratory sample. All units were made aware of the dates of the site visits. Although this is imperative for planning purposes, it also may have resulted in changes to staffing or additional training before our visit.

The data here represent a very small subset of hemodialysis centers across the United States. Our design was descriptive, so statistical comparisons are not available or appropriate. The results presented here should be interpreted with caution. Further validation of use of the SEIPS model in a follow-up study is necessary. We chose to use the most basic SEIPS model, version 1.0, but there are other, more nuanced versions of the model available in the literature, which could and should be considered for future studies. Although our coding was overall consistent and we used a structured coding system, we did not measure interrater reliability, and unintentional bias was not assessed.

Our study is exploratory and our conclusions meant to be hypothesis-generating. Nonetheless, the consistency of our findings across several geographies and types of dialysis facilities does raise important considerations about the relationship of macroergonomic factors for infection prevention. One major limitation of our work is the Hawthorne effect, which is likely inevitable when an unfamiliar group comes to visit. In addition, despite significant effort to ensure the sites and frontline workers were aware that the visits were not “audits,” some of the comments from the staff made it very clear that they were concerned about our presence and doing “the correct procedure.” This is an inherent limitation of observation methods, but we attempted to mitigate it by showing our notes to the staff if they asked to see them and by conducting briefings and out-briefings with leadership. Scheduled visits (rather than unannounced) were necessary for this study. It is difficult to interpret, as a short-term visitor, whether a unit is doing work as usual, but our assumption is that it is unlikely. Even so, we observed challenges, particularly germane if the unit was overstaffed for our benefit. Future studies should attempt to alleviate the obtrusiveness and disruptiveness of these kinds of visits; however, this would take significant organizational engagement and trust at a high level.

The data from our evaluation suggest that the complex constellation of human skills required for the optimal completion of infection prevention tasks within dialysis may be significantly affected (both positively and negatively) by the design of our systems of care. Dialysis operators and facility medical directors may find that an assessment of human factors in their dialysis unit will shed light on facilitators and barriers to infection prevention and other safety procedures, and lead to better facility workflow design. Further studies should explore how to incorporate these concepts into our risk-reduction strategies to help mitigate infection risks to patients and staff.