Participant eligibility and recruitment

Mayo Clinic IRB approved this research (#18–000292). Written consent was obtained from participants. Participants were recruited from an academic medical center in the Midwestern United States with its own not-for-profit dialysis facilities, including two in-center dialysis units that serve 170 total patients. We invited participation from English-speaking patients prescribed maintenance dialysis at the medical center who had no major barriers to consent (e.g. cognitive impairment). Patients receiving in-center hemodialysis or receiving regular follow-up care for home dialysis modalities (either hemodialysis or peritoneal dialysis) were eligible.

During the initial recruitment, patients were approached at the two in-center dialysis units and the center’s outpatient clinic where patients receiving home dialysis underwent monthly follow-up. We approached all eligible patients for voluntary participation in these locations between April 3–24, 2018. Patients who consented completed their survey in person and were told they would be contacted at a future date if selected for interviews. At the time of interview, some participants had received a transplant; we did not exclude these patients from interview or observation. No compensation was provided for the survey, but patients were compensated for their participation in interviews and observations.

Quantitative data collection

Quantitative data analysis

We first summarized demographics and IIS scores. Participants scoring in the highest IIS quartile were defined as having high burden, and participants scoring in the lowest IIS quartile were defined as having low burden. We then examined differences in demographic characteristics between high versus low burden groups using Fisher’s Exact Test for categorical variables and the Kruskal Wallis test for continuous variables. Statistical analysis was performed using SAS version 9.4 (Cary, NC) and p-values ≤0.05 were considered to be statistically significant.

Qualitative participant selection

Following survey analysis, we purposefully invited all patients in the high (n = 16) and low (n = 17) burden groups, unless deceased, to return for an interview. In addition to these purposefully selected patients, four within the sample were recruited for ethnographic observations, described below. These were selected based on willingness to participate, IIS quartile, and richness of interview data.

Qualitative data collection

Our research team (NE, SW, CML, AB) interviewed patients asking questions about patients’ day-to-day (dialysis and non-dialysis) schedule, activity management, and experiences with the healthcare system. The interview team was made up of two doctoral-level researchers and two bachelors-level researchers, all with training in qualitative interviewing and analysis. KRB oversaw interview training and progress. The interviews were conducted primary in the dialysis center, where patients have semi-private dialysis stations. Participants were also given the option to do the interview in in a private room at the healthcare center or in their home if they preferred. Semi-structured interviews were recorded and lasted 15 to 76 minutes (mean 40 minutes) [22]. All interviews were voice recorded, transcribed verbatim, and documented with field notes, which are notes taken by the qualitative researcher regarding observation/interview contextual details. The interview guide was iteratively revised throughout the first five interviews, and the final semi-structured interview guide is attached as S1 Interview guide.

Ethnographic observations were intended to supplement interview data by providing a view of patients’ day-to-day lives managing their ESKD. There is evidence that patients normalize many of the tasks that they do to manage their health [23], and therefore, these minutia may be omitted during experiential interviews but be visible during observation. KHP, KRB, and CML observed these participants’ practices in their daily living environments, including their in-center treatment location, homes, and workplaces. Participants were observed during two to four individual sessions, with up to eight hours of total time spent with each. Researchers took detailed ethnographic field notes, which were typed or voice-recorded and transcribed immediately after concluding each observation day.

Qualitative data analysis

Our team analyzed 33 semi-structured patient interviews and 13 patient observation notes. Field notes from observations and interview transcripts were imported into Nvivo 12, a qualitative data management software. Data analysis generally followed an adapted Grounded Theory approach [24] without a set hypothesis a priori. We first began by reading observation notes and interviews. Five interviews were inductively coded by four coders (SW, NE, AT, PO) using line-by-line coding to create a codebook and calibrate usage of the codebook. The creation of the codebook was overseen by KRB, BT, AJP, and KP. After coders were calibrated, pairs of coders coded the remaining transcripts in duplicate. KRB then synthesized data using the Nvivo matrices feature. As patterns of coding clusters emerged, the query function was used to examine multi-code overlap. Finally, we summarized information into themes across patients as a whole, then looked more specifically at the high and low burden quartiles in comparison. All themes were derived inductively from the data.

During the process of synthesizing the data, KRB met with team members involved in the project to discuss the results and maintain trustworthiness of the interpretations. Throughout the process, the team responsible for analyzing the data, discussed each individual’s various perspectives and how these lenses influenced their perceptions of the data. Specifically, BT, AJP, and NE are clinicians, whereas SW, AT, PO, KP, and KRB were non-clinical researchers at the time of analysis. KRB is also a person living with chronic illnesses but does not live with CKD. During the analysis process, KRB also took detailed notes about her interpretations of the data and shared syntheses of those thoughts with members of the team during meetings to ensure additional perspectives were considered.

Quantitative findings

Patient characteristics are depicted in Table 1, including comparisons between high and low burden groups. Table 2 shows patient characteristics of the in-center versus home dialysis cohorts. The only statistically significant quantitative difference between the high and low burden groups was that participants experiencing high burden were significantly younger than those reporting low burden (48.4 vs 68.6, p = <0.001). In-center and home dialysis patients had no statistically significant differences except for sex.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Patient demographic characteristics by patients reporting high versus low burden.

https://doi.org/10.1371/journal.pone.0260914.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Patient demographics by in-center versus home dialysis.

https://doi.org/10.1371/journal.pone.0260914.t002

Qualitative findings

We found travel as universally problematic amongst both patient groups. Additionally, we found five themes for which the illness experience of ‘high burden’ patients was significantly different than those reporting low burden. Demonstrative quotes from the themes that emerged from the comparative analysis are illustrated in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Patient quotes from high- versus low-burden groups.

https://doi.org/10.1371/journal.pone.0260914.t003

Summary of findings

This study utilized a mixed-methods design to gain an in-depth understanding of ESKD patients on dialysis reporting high versus low treatment and illness burden. Patients reporting high burden had more difficulty establishing a new rhythm of life on dialysis, more disrupted biographies, fewer appraisal-focused coping strategies, less supportive social networks, and more negative past healthcare interactions, which may have impacted how they experienced healthcare in the present. Quantitatively, we noted patients with high treatment and illness burden were significantly younger. While this initially may seem paradoxical, since chronic conditions accumulate with age, younger patients may be in different phases of individual and family development compared to their older peers. Younger patients may face greater personal or familial pressures that compete with their attention to healthcare such as paid employment and caregiving activities. This finding aligns well with the qualitative differences reported in the theme “biographical disruption,” attributed to the term first coined in Bury’s 1982. In his exploration of chronic illness as a disruptive event, he noted that the emergence of chronic illness disrupted both the perception of the self and the self in relationship with others [25]. Further, in studying the concept of hope in older patients living with chronic illness, it has been noted that older adults may reach for hope in different ways than younger individuals, namely through the processes of transcendence and positive reappraisal [26]. This is consistent not only with our finding that older patients report less burden, but also that those reporting reduced burden used positive reframing related to their situation. Our finding is also consistent with existing treatment burden literature, which has shown a greater percentage of patients under 60 years of age report unacceptable levels of treatment burden compared to those over 60 (51% versus 22%) [27].

We also found a significant difference in the number of males and females receiving in-center versus home hemodialysis or peritoneal dialysis. In examining the US Renal Data System 2018 incident cases, approximately 7 females begin any type dialysis for every 10 males [28]. Similar numbers persist across modality type, whereas in our sample, the ratios of females to males currently participating in-center dialysis versus a home modality was 2:5 versus 2:1. Therefore, we attribute this finding to sampling bias in our small sample.

Comparison to existing literature

While we conducted our analyses inductively, the themes which emerged fit well with existing theory. The Theory of Patient Capacity (TPC) expands on the concept of capacity named in the CuCoM and states that patient capacity results from a dynamic interaction between patients and their Biographies, Resources, Environments, patient Work, and Social Networks. The theme of “rhythm” fits well with the “patient work” construct, which proposes that when patients are given too many tasks at once, their capacity to act may be overwhelmed [29]. Furthermore, this theme suggests patients are tapping into their own personal “resource” of self-efficacy. The theme of “biographical disruption” overlaps with the construct of “biography,” which describes patients’ ability to overcome the disruptions in social roles and self-perception. The theme of “social network support” is in alignment with the “social network” construct, which suggests that patient capacity for self-care is influenced by robust, supportive social networks. Finally, the theme of “coloring experiences” is in alignment with the TPC’s “environment” construct, focused on patients’ interactions with their healthcare environments as principally responsible for their capacity to enact self-care.

Our study highlights some themes that are similar to others’ work in this area. For example, in a recent systematic review of 260 papers by Roberti et al. regarding the work associated with living on dialysis, it was regularly noted amongst studies that patients felt a lack of self-control around where and when they could travel [8]. Additionally, this review noted similar sources of social support, as well as the loss of social support that sometimes accompanied life on dialysis [8]. However, our work adds important information not previously highlighted. For example, while the review by Roberti et al. describes the work that is undertaken by patients on dialysis and the time management strategies required to organize that work [8], it does not discuss that the mastery of those task can be associated with less life disruption and a more positive treatment experience. The concept of biographical disruption in illness is not new, and has been found in patient narratives of those living with ESKD choosing to forego dialysis [30, 31]. However, our study appears to be the first to highlight this concept specifically amongst patients on dialysis.

Most importantly this study’s contribution to the literature is that the experience of dialysis is nuanced for each patient and burden is not universal. Some patients may feel a great deal of burden placed on their lives by dialysis whereas others experience less burden, and these two patient groups are different in their management of the disease and treatment. These differences may be amenable to interventions. For example, healthcare culture change and shared decision-making interventions may improve patients’ early dialysis journeys. Social support interventions may bolster patient networks where they are lacking, and capacity coaching [32, 33] interventions may improve patients’ abilities to establish a rhythm of dialysis and more positive framing of life with dialysis.

Limitations and strengths

We first must acknowledge the limitations of this work. Principally, this research was conducted in a single healthcare system of an academic medical center within the upper Midwest of the U.S, limiting our ability to see themes that would arise in settings where structural inequity is more pronounced, as described by Roberti et al. [8]. We did not note these inequities, with the exception of occasional insurance complications and confusion. Furthermore, we had a slightly lower response rate than our previous survey only study (61% versus 70%). We suspect that this is due to the fact that participants at the time of consent were asked to participate in both the survey and the interview, if selected. Although they consented at the time of survey completion, when contacted for interviews, not all patients who were selected chose to participate in that portion of the study. Those who declined may have been feeling overly unwell, tired, or dealing with competing priorities. However, it is worth noting the demographic characteristics of our responders appears to be well representative of our population based on previous studies [17, 34, 35] and that response rates between 37% and 56% have been established under similar data collection conditions [36].

Despite these limitations, the strengths of this research are in its mixed-methods design and its ability to present nuances between patients reporting high versus low treatment and illness burden. Whereas other work has looked quantitatively or qualitatively at these issues in independence, our design allowed us to truly understand the differences between these patient populations, and therefore hypothesize interventions to improve the experience of patients reporting high burden based on findings from their low-burden counterparts. Furthermore, while our themes were inductively derived, they fit well with existing theory from a larger body of work across chronic condition populations. This suggests findings are likely transferable to other dialysis populations and reproducible in other illness populations as well.

Practice and research implications

Hospitalization rates, emergency room utilization and readmission rates among dialysis patients are several standard deviations above the general United States Medicare cohort and strongly affected by patient adherence [37–40]. Decreasing these rates is of crucial importance directly to our dialysis-requiring population to improve quality of life and value. Currently, it is not standard practice in caring for patients on dialysis to measure their level of treatment and illness burden and respond accordingly. Furthermore, it is not current practice to adapt clinical actions based upon patients’ level of reported burden. However, this study highlights that this actually may be helpful to impact clinical outcomes, especially for patients with noted non-adherence.

This work provides important information to care teams as they design strategies to decrease suboptimal outcomes and further emphasize the importance of engaging our patients in developing care that fits. This work provides the foundational “preclinical phase” or “theory” portion of the design process [41], which is now translatable to intervention development. The work presented here is deeply steeped in the patients’ experiences of living on dialysis and navigating it in their own lives. This is a fundamental component to our team’s design process, which is in alignment with others in noting that integrating patient expertise into intervention design can offer valuable perspectives [42]. Managing health is an on-going task that goes beyond the fraction of their lives patients spend in the medical office or dialysis unit. There are many “on your own” hours that could be filled with patient peer support or peer coaching [43]. Based upon our data, this center’s teams are now in the development phase for an intervention that combines peer support and capacity coaching to patients specifically reporting high burden [32, 33]. Replication of this study’s concepts and designs among other demographic, socioeconomic, and geographic populations is necessary.