Abstract
Kidney supportive care (KSC) is a patient-centered model of multidisciplinary care designed for patients with advanced chronic kidney disease (CKD) and end-stage kidney disease (ESKD). Our goal was to characterize the types, frequencies, and costs of services accessed by patients enrolled in a KSC program. We analyzed health care utilization data prospectively collected from 102 patients who enrolled in the KSC program during the first 52 weeks of its existence. The data comprised program appointments, emergency department presentations, ambulance service use, outpatient visits, inpatient episodes, and dialysis treatments made within the Brisbane area of Metro North. Costs of resource use were estimated using Queensland Health funding principles and guidelines. Analyses included descriptive statistics, correlations, and multivariate regressions. During the median program participation of 22 weeks, patients had 3975 contacts with health care, with the total value of services amounting to nearly A$3 million. Dialysis treatments accounted for 70% of visits and 49% of costs. Patients receiving dialysis had higher utilization of outpatient services and associated cost, compared to patients who were not dialyzed. The presence of diabetes and the choice of conservative pathway were both predictors of higher frequency and cost of services. Longer program participation was associated with lower weekly utilization and cost. The program attracted patients representing various characteristics, pathways, needs, and outcomes. Exploring these patterns will enable better understanding of the patient population and improved service planning, in KSC and similar programs that aim to comprehensively address the needs of patients with advanced CKD and ESKD.
Keywords
Introduction
The substantial burden that renal replacement therapy puts on a person and their family is increasingly recognized. 1 Problems include a multidimensional burden of physical and psychological symptoms 2,3 and their effect on health-related quality of life. 4,5 Research into preferences and priorities of patients and their families indicates that chief concerns are a peaceful transition to death and having their decisions supported and respected. 6 Contrasting with this is the appreciation that renal replacement therapy pathways do not always correspond to these wishes. 7 Moreover, a growing body of evidence shows that, in certain end-stage kidney disease (ESKD) groups, dialysis offers little benefit in terms of improved survival and reduced morbidity. 8
With the limitations of conventional dialysis pathways well-documented, there is a growing interest in health care models focused on the broader needs of the patient and their family. Such models are grounded in health services that holistically address multiple interlinked aspects of health including symptom management, psychological well-being, social living, and social care, in an integrated fashion. 9 Still, despite our understanding of these needs, and the providers’ willingness to align the delivery of health care to patients’ expectations, little is known about the optimal organization of health care for these groups of people.
Kidney supportive care (KSC) is a novel, person-centered model of care designed to align with the preferences and priorities of people with advanced chronic kidney disease (CKD) or ESKD and their families. Its purpose is educating people living with advanced CKD in managing its symptoms and making complex health care decisions, such as withdrawal from dialysis or end-of-life care. It provides a structured care pathway without excluding people already on predialysis, hemodialysis (inpatient, home, and satellite center), peritoneal dialysis, transplant, and conservative pathways. Care is provided by an interdisciplinary team jointly led by a palliative medicine specialist and a specialist kidney clinical nurse consultant who case manages the patients, complemented by an advanced medical trainee in nephrology, a kidney pharmacist, and a social worker specializing in end-of-life care. Kidney supportive care can be accessed upon referral from the time of CKD diagnosis until death, with or without dialysis; its key feature is the personalization of care to match each person’s individual circumstances.
Results reported in this article are from the first implementation of this model of care in a public hospital service in Brisbane, Queensland, Australia; this particular implementation is henceforth referred to as the KSC program. Program details including the organization of the clinic, the referral process, patient flows, funding arrangements, and baseline results have been presented elsewhere. 10 The study was carried out with the approval from the Human Research Ethics Committee of the Royal Brisbane and Women’s Hospital (HREC/16/QRBW/208).
Materials and Methods
Data
For the purposes of this study, we defined the analysis set as comprising those people who attended at least 1 KSC appointment in the 12-month period between February 2016 and January 2017 and remained in the program for at least 7 days.
Data were prospectively collected from the clinical record generated during usual clinical care and included demographics, symptoms (Integrated Palliative Care Outcome Scale–Renal [IPOS–Renal]), comorbidities (Charlson Comorbidity Index [CCI]), and quality of life (Short Form 36 [SF-36]). 10 Medical records were retrieved containing information on the use of health care services within the study period. This information encompassed health care accessed within the public health care system of Metro North area of Queensland, Australia, and included KSC clinic appointments, emergency department (ED) presentations, ambulance services, outpatient visits, inpatient episodes, and episodes of dialysis treatment.
Analyses
Utilization outcomes were first summarized using descriptive statistics and then tested for equality of means between dialysis and nondialysis groups. We further used Pearson correlation coefficients to explore the associations between measures of symptoms (IPOS-Renal), comorbidities (CCI), and health-related quality of life (SF-36), and health care utilization and costs.
Given that no control group was available for us to perform comparative analyses, we statistically analyzed the observational data to gain insights into participant outcomes. Specifically, we employed 2 multivariate regression analyses: Model (1) used a Poisson formulation 11,12 with the purpose of explaining variation in the weekly number of contacts with health care using a set of participant characteristics. Model (2) was a linear regression 13,14 of log-transformed average weekly health care costs on the common set of participant characteristics. In both models, the respective outcome variables accounted for all health care except for dialysis. We excluded dialysis to control for the variation in the data that is attributable to this single major factor and thus to focus on the areas where utilization and cost are less understood. Postestimation diagnostics were performed to ensure model assumptions were met and results were valid.
The set of independent variables used in the regressions included age, gender, diabetes, cardiovascular disease, the use of dialysis (none, for part of the study, throughout the study), conservative care status to represent any changes in health care use patterns associated with the choice of conservative pathway, and log-transformed time in the program, allowing for determination of changes in utilization patterns associated with longer program participation. The “dialysis” variable controlled for the use of renal dialysis of any modality during the study, reflecting the needs of 3 groups of patients: (1) kidney replacement therapy (n = 59), (2) advanced but pre-end-stage CKD (n = 39), and (3) conservative care (n = 4). The “conservative” variable captured transitions from the initial group to conservative management, which was the case in 10 (18%) of the kidney replacement therapy patients and 15 (38%) of the patients with advanced CKD (Table 1).
Characteristics of the Study Population and Health Services Use.
Abbreviations: CKD, chronic kidney disease; ED, emergency department; HHD, home hemodialysis; HD, hemodialysis; IQR, interquartile range; KSC, kidney supportive care; PD, peritoneal dialysis.
Costs of health care were approximated using Queensland Health clinical costing guidelines and principles 16 for inpatient, outpatient, and ED care. The use of Queensland Ambulance Service was costed using the published average cost specific to Metro North. 17 Foreign currency equivalents to A$1 were US$0.80, Can$1.03, GBP 0.62, and EUR 0.71. 18
Results
Patient Population and Health Services Use
Out of the 129 people who were referred to the KSC program in the 12-month study period, 102 attended at least 1 KSC appointment and remained in the program for at least 7 days (Table 1). The median time of program participation was 22.1 weeks. Fifty-two (51%) participants were female, and 5 (5%) were Indigenous Australians. The median age at the first visit was 74.5 (interquartile range: 63.6-83.8) years. Forty-seven (46%) participants received 1 or more session of dialyses of any modality. In-center hemodialysis was the dominant modality (43, 42%); a small number of patients received peritoneal dialysis (3, 3%) or home hemodialysis (4, 4%).
Presentations to ED were in triage category 3 in 48% of cases, category 4 in 24%, and category 2 in 20% (Table 1). The most common major diagnostic categories (MDCs) recorded in the ED were urology (23%), circulatory (19%), injury (14%), and respiratory (11%). Among participants presenting to the ED, 77% were admitted. The most common MDCs recorded in participants admitted to hospital were kidney and urinary tract problems (28%), circulatory (19%), and respiratory (17%) conditions.
Instances of Health Care Utilization
During the 12-month study period, the 102 participants recorded 3975 contacts with the hospital health care system. The vast majority were dialysis treatments (2761, 69.5%), followed by outpatient visits (683, 17.2%). Kidney supportive care program appointments (223, 5.6%), ED presentations (118, 3.0%), in-patient episodes (109, 2.7%), and ambulance services (81, 2%) collectively represented the remaining 13.3% of health care instances (Table 2).
Health Care Utilization—Number of Contacts With Health Care.
Abbreviations: KSC, kidney supportive care; ED, emergency department; QAS, Queensland Ambulance Service.
Bold, statistical significance of 0.05.
a No dialysis record within the study period.
b ≥1 dialysis of any type.
Participants accessed the services 1.64 times per week on average. However, this average was skewed by the 47 patients attending 2761 dialysis sessions. Even after excluding dialysis treatments, these participants still used health care more frequently, at the rate of 0.75 times per week versus 0.40 in participants not receiving dialysis. This difference was attributable to outpatient care which they used at 0.42 times per week, compared to 0.12 in participants who were not on dialysis. In the remaining health care categories, nominal differences in means were not supported by statistically significant P values (Table 2).
Costs Associated With Health Care Use
The total cost of health care generated by KSC participants was nearly A$3 million, 49% of which was the costs of dialysis (A$1.5 million; Table 3). The largest categories contributing to the remaining 51% of costs were inpatient (31.7%), outpatient (9.6%), and KSC program appointments (4.8%). Emergency department and Queensland Ambulance Service amounted to 3.6% and 1.3% of the total cost, respectively. Participants on dialysis generated 77% of the costs, with the care provided to participants who were not on dialysis (ie, predialysis or conservative care), accounting for the remaining 23%.
Cost of Health Care Recorded in the KSC Program.
Abbreviations: KSC, kidney supportive care; ED, emergency department; QAS, Queensland Ambulance Service.
Bold, statistical significance of 0.05.
a No dialysis record within the study period.
b ≥1 dialysis of any type.
The mean weekly cost of health care recorded for program participants was A$1546. Those on dialysis generated 3.6 times (A$2524) the weekly cost of participants not on dialysis (A$710), although the difference between the groups was not statistically significant when the costs of dialysis treatment were excluded. Inpatient care was the largest contributing item (A$712, 46%) followed by dialysis (A$569, 37%) and outpatient care (A$108, 7%). Outpatient costs were statistically significantly different between the dialysis and nondialysis groups (A$180 and A$47, respectively).
Figure 1 is a box plot presenting the distribution of the mean weekly costs of health care used by KSC participants. The minimum was A$13, the first quartile A$162, the median A$1180, and the third quartile A$1897. The maximum recorded for a single participant was A$20 989.
Distribution of mean weekly health care cost. Five outliers not shown (A$4781, A$5250, A$8273, A$10 637 and A$20 989). Outliers defined as Q3 + 1.5 × interquartile range. 15
Figure 2 presents the top MDCs contributing to total inpatients costs. Admissions associated with kidney and urinary tract conditions accounted for nearly 30% of the inpatient cost, followed by respiratory (25%) and circulatory (18%). The cost per admission was the highest in the respiratory category (A$12,962), followed by kidney and urinary tract (A$9397) and nervous system (A$8392). Admissions due to all other MDCs accounted for 20% of all inpatient costs, with an average cost per episode of A$5962.
Inpatient cost per major diagnostic category. Labels indicate category name, number of cases, average cost per case (A$), and the percent contributing to total inpatient costs (A$942 243).
Correlation With Symptoms, Comorbidities, and Health-Related Quality of Life
Using explorative coefficients of correlation, we did not find evidence of associations between symptoms (as measured by IPOS-Renal), comorbidities (CCI) or health-related quality of life (SF-36 index score), and the utilization and cost of health care (Table A1). The coefficients were for:
– Weekly frequency of health care use: CCI ρ = 0.2 (P = .052); IPOS–Renal ρ = 0.08 (P = .47); and SF-36 ρ = −0.06 (P = .65);
– Weekly costs of health care: CCI ρ = 0.12 (P = .26); IPOS–Renal ρ = 0.09 (P = .42); and SF-36 ρ = −0.17 (P = .17).
Multivariate Regression Analyses
The Poisson model for the count of contacts with health care explained approximately 12% variation in health care utilization (Table 4). The variables of age, gender, and cardiovascular disease comorbidity were not statistically significant. The presence of diabetes, on the other hand, was associated with an increase in health care utilization of 38% (e0.32-1). Participants receiving dialysis for a part of or throughout the study period had, respectively, 170% and 49% higher utilization than participants not receiving dialysis (the reference category). Those choosing conservative care were associated with a 38% increase in health care utilization relative to participants who did not choose a conservative pathway. Longer program participation was associated with a lower incidence of weekly health care use, with a 10% additional time in program resulting in a 3.8% decrease in the weekly number of contacts with health care.
Results of Multivariate Regression Analyses.
Abbreviations: CI, confidence interval; CVD, cardiovascular disease.
Bold, statistical significance of 0.05.
(1) Poisson model (contacts with health care for reasons other than dialysis).
(2) OLS with log-transformed outcome variable (costs of health care other than dialysis).
* Wald chi2(8) for model (1) and F(8, 93) for model (2).
^ In the case of model (1) reported is pseudo-R2.
The second model explained 35% of the variation in costs of health care other than dialysis. Age and gender, as well as cardiovascular disease comorbidity, were not statistically significant determinants. Participants who had diabetes had 131% higher cost than those without diabetes. Costs recorded in those receiving dialysis for some but not all of the study period were 329% higher than in the reference category of nondialysis. No such relationship was found for patients receiving dialysis throughout the study period. Conservative care was associated with 77% higher costs other than dialysis, compared to not entering this pathway. Patients who remained in the program for longer had lower costs of health care, with a 10% increase in program participation time associated with a 6% decrease in the mean weekly cost.
Regression methods based on the normal distribution were confirmed appropriate for our data, with both the outcome variable and error terms being normally distributed according to visual inspection (plots of error terms against fitted values) and statistical tests (Shapiro-Wilk, Shapiro-Francia, and the skewness/kurtosis test for normality; Table A2).
Discussion
This is the first study to comprehensively capture, describe, and analyze the utilization and costs of health care in patients participating in a KSC program. In doing so, it supplements the scarce literature of health utilization outcomes 19,20 and responds to previously identified knowledge gaps 21 -23 in this heterogeneous patient population. It comes at a time when the limitations and consequences of dialysis, 8,24 the materiality of health care resources, 25 and equity of access 26 are reinvigorating the search for better ways of delivering care. 9,27
Five key findings emerge from this study. First, the KSC population was balanced between participants not on dialysis (who had discontinued, withdrawn, or never started) and those receiving dialysis with varying vintages (from numerically small numbers of acute episodes to regular continuing weekly appointments). There was a clear distinction in the utilization of nondialysis health care between the 2 groups. In particular, dialysis users sought health care more frequently (notably outpatient care) and accessed health care that was more costly. Interestingly, participants who were on dialysis for some but not all of the study period (in particular those who switched from an active to conservative pathway during the study) ended up using more (nondialysis) health care compared to both those not on dialysis and regular dialysis users.
Second, the participants who had diabetes were considerably more likely to access health care, and the health care they received was more costly, compared to nondiabetics. This finding stems from the diabetes coefficient in model 2 being greater than that in model 1, which implies that the increase in frequency was insufficient to fully explain the increase in cost. We found no such relationship for cardiovascular disease, the other key comorbidity accounted for in our model.
Third, there is a clear association between the active choice of conservative pathway and health care utilization, with conservative care associated with increases in the frequency and cost of health care other than dialysis. This finding should be interpreted with a degree of caution, however. While the choice of conservative pathway may have important implications for both the patient and the health care provider, there is also a possibility of confounding by the health status beyond the control variables included in our model. On the one hand, conservative care is likely to require more frequent contacts with the interdisciplinary team aimed at controlling symptoms and improving health-related quality of life. On the other, it is plausible that the participants who use health care frequently, due to their poorer condition, are more likely to choose the conservative pathway. While the statistically significant association is a notable finding from our model, further investigation is required to fully explain the reasons for, and consequences of, choosing conservative care.
Fourth, longer KSC program participation was associated with less frequent use of health care and a diminished cost of health care services. The effect was maintained in an alternative model specification that controlled for the death event (Table A3). Despite a somewhat reduced effect size of time in the cost model, this corroborated the conclusion that the higher observed utilization by patients with shorter program participation was not due to their impending death. The implication of this finding is that KSC not only has the capacity to address the complex needs of patients with advanced CKD and ESKD, 10 but in so doing it can also reduce the pressure on health care system resources. This outcome has been previously demonstrated in a similar setting involving integrated, person-centered health care and palliative care for chronic and severe heart failure. 28 Further studies with controlled design, longer follow-up, and concurrent health outcome measures will reaffirm this favorable effect and determine whether KSC is a dominant (better outcomes, lower costs) model of care vis-à-vis standard practice.
Finally, we did not find statistically significant relationships between symptoms, comorbidities, health-related quality of life, and health care utilization. While the monitoring and management of symptoms are of utmost importance in this patient population, 29 -31 it would appear that the number, severity, and frequency of symptoms, assessed using index measures, is not a good candidate predictor for the use of health care and its costs. Still, our study was not designed to detect such relationships and thus the outcome should be interpreted as indicative, and inviting further exploration, rather than conclusive. The fact that these are complex and multifactorial concepts, the problems of their measurement using composite indexes, the powering of the analysis, and the heterogeneity of study participants may have all contributed to this statistical outcome.
Generalizability and transferability of our results may be limited due to the idiosyncratic nature of supportive care (see Purtell et al for a comparison of design characteristics) 10 and owing to international differences in reimbursement system incentives 32 and preferences. 33 In particular, costs identified and measured in the Australian system may not readily translate to other countries. In the United States, for example, dialysis is commonly provided in an outpatient for-profit setting, contrastively to the predominantly publicly funded inpatient provision in Australia. Consequently, the patterns of costs and cost savings associated with supportive care may be only applicable locally.
Still, our results serve to inform clinical and cost-effectiveness evaluations of supportive care for patients with advanced kidney disease. Despite analogous models of care implemented in the United States, 34,35 the United Kingdom, 27 Canada, 9 Taiwan, 36 and elsewhere in Australia, 37 evidence of the economic performance of such models remains a major knowledge gap and a hindrance on their implementation. 22 This is a concern especially in North America where CKD prevalence 38 and dialysis rates per million population 21 are higher than in Australia. Considering also the high and growing level of public spending on ESKD, in the United States accounting for 7.2% of all Medicare claims in 2016 with 4.6% annual growth, 39 these data imply a larger than in Australia role for conservative and supportive care in managing the burden of EKSD. 40
Rather than specific cost estimates, the key findings of this study are concerning broader patterns of health care use and factor–outcome associations. As our analyses suggest, longer program participation can be associated with lower health service use, a finding that is potentially generalizable (cf. Sahlen et al) 28 and one that may foster the development of similar models of care globally. Furthermore, utilization not only informs about the burden on health system resources but also reflects health care needs in this population. For this reason, outcomes such as rates (eg, frequency of contacts with health care, ED presentations) and proportions (eg, shares of expenditures attributable to particular health care settings, MDCs, triage categories) may be of interest to foreign jurisdictions where local data are missing.
The main limitation of our study is that it did not account for the use of medicines, primary health care visits, or specialist and allied health services provided outside of the program. Such data are stored in federal information systems and not linked to state-level information systems accessible by the researchers. Consequently, these categories of health care were considered to be out of scope for our analyses. It has to be emphasized, however, that our study did account for the services used in the acute care sector which are likely to be accessed by people with advanced limitations in daily living.
Moreover, the data included in this study did not capture health care episodes that may have taken place in Brisbane outside of the Metro North area. Given the reduced mobility of the study population, we assumed that participants were unlikely to repeatedly and systematically present in other health service areas. Consequently, our data are likely to reflect a very high proportion of the health care used by these participants, with only incidental instances recorded outside of the Metro North jurisdiction.
Conclusion
Kidney supportive care is an interdisciplinary model of care designed to address the complex needs of patients with advanced CKD and ESKD. The pilot study spanning the initial 12 months of its first implementation demonstrates characteristics of the target population and indicates that program participation may lead to reduced reliance on resource-intensive health care. The work provides an evidence base for implementation of similar models of care and a platform for future research. As it stands today, KSC has the potential to become the new standard of care, offering better outcomes for selected patients and a cost-effective use of resources when compared to current practice. The latter statement remains a hypothesis to be tested in controlled studies.
Supplemental Material
Supplementary_material - Utilization and Costs of Health Care in a Kidney Supportive Care Program
Supplementary_material for Utilization and Costs of Health Care in a Kidney Supportive Care Program by P. Marcin Sowa, Louise Purtell, Wendy E. Hoy, Helen G. Healy, Ann Bonner and Luke B. Connelly in Journal of Palliative Care
Footnotes
Authors’ Note
The study was carried out under the approval of the Human Research Ethics Committee of the Royal Brisbane and Women’s Hospital, reference number HREC/16/QRBW/208. Written informed consent was obtained from all study participants or their guardians.
Acknowledgments
The researchers would like to thank the patients and acknowledge clinical staff involved in the organization and operation of the KSC program. We would also like to thank the Editor-in-Chief Dr Keith Mark Swetz and an anonymous reviewer for their valuable suggestions which led to an improved discussion.
Author Contributions
All authors contributed to the study design, interpretation of results, and preparation of the manuscript. P. Marcin Sowa and Louise Purtell processed the data. P. Marcin Sowa and Luke B. Connelly analyzed the data. P. Marcin Sowa drafted the manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The researchers received support from the National Health and Medical Research Council (NHMRC) Centre of Research Excellence (CRE). The research component of the program received financial assistance from the Australian Centre for Health Services Innovation (AusHSI).
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
