Dialysis Outcomes and Practice Patterns Study (DOPPS)

Introduction The Dialysis Outcomes and Practice Patterns Study (DOPPS) is a leading source of up-to-date, representative, and comprehensive data on hemodialysis practice and patient outcomes worldwide. Since its inception, the study’s primary aim has been to identify practices that extend survival and improve health-related quality of life (HR-QOL) for hemodialysis patients. The DOPPS has yielded research findings directly relevant to patients, health care providers, and policy makers. The following provides an overview of the study’s design and analytic approach, selected research findings and relevance to policy, new directions, and ways that the study complements other sources of dialysis data. DOPPS Overview DOPPS 1 began in 1996 in the United States and in 1998–1999 in Japan, France, Germany, Italy, Spain, and the United Kingdom. In DOPPS 2 (2002–2005), 3 (2006–2008), and 4 (2009–2011), these countries were joined by Australia, New Zealand, Canada, Belgium, and Sweden. The launch of DOPPS 5 is underway in 2012, with new directions and additional countries as detailed later in this article. DOPPS 1–4 followed 300–350 facilities in each study phase, having now collected basic demographic and mortality (census) data for >200,000 patient-years and detailed data for >75,000 patient-years. Additional details are available at http://www.dopps.org. Funding for the DOPPS was initiated by Amgen in 1996 in support of a “longevity initiative.” Funding for Japan DOPPS has been provided solely by Kyowa Hakko Kirin since 1999. Outside of Japan, funding was provided solely by Amgen through 2008, but since 2009, has been provided by a consortium of sponsors including Amgen as well as Abbott, Baxter, Fresenius Medical Care, Sanofi Renal, and Vifor Fresenius Renal Pharma (current sponsors). Recent expansion of the DOPPS programs (into new countries, CKD, and peritoneal dialysis PD) has been made possible by directed sponsor support. All funding support is provided without restrictions on publications and preserves the independence of our scientific research. DOPPS Goals, Design, and Data DOPPS research goals are centered on the hypothesis that measurable differences in dialysis facility practices influence patient longevity, morbidity, and HR-QOL. Identifying opportunities for improvement is a primary motivation and guides research priorities (1,2). DOPPS research findings derive logically from key features of the study design. The DOPPS is a prospective cohort study with data from a random sample of patients within a random sample of hemodialysis units in each participating country. We use a stratified selection process to represent the composition of facility types (e.g., free-standing, hospital based, satellite, etc.) and regions within each country. Descriptions of DOPPS sampling and methods have been published (1,3,4). Random sampling to enroll a representative sample of dialysis units was quite novel among epidemiologic studies at study launch. This approach supports the accurate description of actual practice in national hemodialysis populations and ensures that findings can be generalized to these populations. It also maximizes variation in practices and outcomes in order to enhance the analytic ability to identify important, potentially causal associations. We have learned much from the often surprising differences in practice among participating countries. The DOPPS captures detailed longitudinal patient- and facility-level information using a common protocol and standardized data collection instruments. With each new phase of the DOPPS, operational flexibility allows questionnaire refinement and new substudies to test new hypotheses and keep the study current. The study collects census data, including demographic and survival data, for all patients in each DOPPS facility. For enrolled patients, detailed longitudinal data are collected that include demographics, numerous comorbidities, medications and dosing, monthly laboratory values, vascular access type and procedures, cause-specific hospitalizations, date and cause of death, and others. The reliability of clinical data abstraction in the DOPPS was confirmed most recently by independent reabstraction from a random selection of participants in our Chinese study, as well as similar data abstraction from European study sites previously. The DOPPS collects rich data on patient-reported outcomes, with participants completing questionnaires annually on diverse topics such as HR-QOL, recovery time after a dialysis session, satisfaction with care, medication adherence, and others. To provide facility practice information, the unit’s medical director and nurse manager complete questionnaires annually, and surveys can also be collected from other staff. Proposals for new study modules, with collection of novel facility practice data or patient-reported data, preferably with validated survey instruments, are encouraged. With respect to clinical data, the DOPPS has chosen explicitly not to collect additional biomedical tests (laboratory, radiographic, or other) beyond those ordered for routine clinical care, to avoid biasing the study toward selection of highly motivated (i.e., nonrepresentative) patients or facilities, or to influence practice among participating facilities. Because the DOPPS does not measure bioassays centrally, measurement error may be introduced by nonstandardized measurement. However, the effect is likely small. Even in the case of parathyroid hormone (PTH) assays, most DOPPS facilities use a single assay type (e.g., in the most recent survey: 88% intact PTH, 8% bio-intact PTH, 4% unsure), and we collect data on each laboratory’s reference range, allowing for standardization to this range in sensitivity analyses. One example of the value of a multinational study is that we have a large amount of data on bioassays measured frequently in some, but not other, countries. For example, serum C-reactive protein (CRP) levels are measured routinely (e.g., monthly) in most DOPPS countries but rarely in the United States. Postdialysis electrolyte levels are measured often in many countries; serum fibrinogen, B-type natriuretic peptide, and troponin levels are common in a few countries. Our large longitudinal database of CRP levels formed the basis for a recent publication describing the range of CRP levels in routine dialysis practice, as well as the added prognostic information gained by CRP even when measured along with inflammatory markers relied on in the United States such as serum albumin and ferritin levels (5). Future analysis will evaluate whether physician responses to elevated CRP levels help to improve clinical outcomes. Reporting and Evaluating Practice Trends The DOPPS is a unique resource to understand temporal trends in dialysis care that may occur for many reasons, including reimbursement changes, regulatory shifts, publication of key research findings, release of new practice guidelines, changes in availability or promotion of products, and so forth (6,7). In recent years, these trend data have been made readily accessible to the public, in a user-friendly format, to provide the most up-to-date and detailed source of dialysis practice data and for international comparisons. Detailed trends in dialysis care across all DOPPS countries are reported in the DOPPS Annual Report, initiated in 2009 (http://www.dopps.org/AnnualReport). In late 2010, we launched the DOPPS Practice Monitor (DPM), our flagship product to monitor US dialysis trends (http://www.dopps.org/DPM) (Figure 1).Figure 1: DPM home page. The DPM home page (http://www.dopps.org/DPM) is updated every 2 months with the most recent trends in numerous US hemodialysis practices. DPM results are based on a nationally representative sample, as described in the text. DOPPS, Dialysis Outcomes and Practice Patterns Study; DPM, DOPPS Practice Monitor.DPM for US Trends The DPM reports detailed, contemporary trends in US dialysis care from the national sample of 120–140 DOPPS facilities. Findings are updated every 4 months, with data lagged by about 4 months. Downloadable slides are also on the website. Publications include analyses of key findings after each major website update, which are accompanied by editorial comment (4,8,9). The DPM was originally developed to report trends in dialysis care before, during, and after implementation of the new Centers for Medicare and Medicaid Services (CMS) bundled dialysis payment system, the End-Stage Renal Disease Prospective Payment System (ESRD PPS), over January 2011 to January 2014 (10,11). Now in place, the DPM provides a mechanism for public, detailed, and up-to-date reporting of trends in care regardless of cause. An illustrative example is the June 2011 revised US Food and Drug Administration label for prescription of erythropoiesis-stimulating agents (ESAs) to CKD patients, followed in July 2011 by the proposed Quality Incentive Program (QIP) update to remove the hemoglobin floor of 10 g/dl. DPM data indicate dramatic changes in anemia care soon thereafter, with declines in average ESA dose and hemoglobin levels that were larger and more abrupt than after PPS implementation in January 2011. Events now or in the near future also likely to shape clinical practice include roll-out and updates of the QIP in 2012 (based on data from 2010 forward), the anticipated addition to the PPS of oral dialysis-related medications in 2014, and the release of clinical practice recommendations (e.g., Kidney Disease: Improving Global Outcomes anemia in 2012), clinical trial findings (e.g., Evaluation of Cinacalcet HCI Therapy to Lower Cardiovascular Events in 2012), and competitor products (e.g., peginesatide in 2012) (12,13). In addition to ESA dosing and hemoglobin level, other notable US trends we have reported from 2010 to 2012 include sizeable rises in intravenous iron dosing, ferritin levels, and PTH levels. At the same time, phosphorus levels above recommended targets remain commonplace. By raising public awareness of these findings, the DPM can help inform decisions surrounding calls to introduce QIP measures for these metrics or for consequences such as red blood cell transfusions. This helps assure that financial incentives with the PPS do not remain unchecked. Advantages over Other Data Sources Detailed DOPPS data, from a nationally representative sample of dialysis centers, complement data sources that may include more patients but are not representative nationally (e.g., large dialysis organization and other electronic health record data) or are limited in breadth of data (e.g., most registries and administrative data). CMS data also have limited information on incident dialysis patients and the rising percentage of patients covered primarily by private insurance or Medicare Advantage programs, because their care often does not generate Medicare claims. The DOPPS is well suited to study whether differing financial incentives among these patients will translate into differences in care. Prior DOPPS work comparing national financing for dialysis and evaluating links between incentives and patient care across the DOPPS countries has already been detailed and informative (14). Some Caveats DOPPS data are reported in aggregate. Facilities and facility groups are not identified individually, and the data are not intended to provide oversight of performance. Because the DOPPS is based on a sample, findings may differ slightly from national census data, and trends reported may merit confirmation with national data, in cases in which those are eventually available. In particular, estimated event rates (e.g., mortality, hospitalizations, transfusions) should be confirmed with national data. Specification of calculated variables also differs slightly from other data sources. For example, our calculation of the percentage of fistula use differs slightly from the Fistula First Initiative using CMS data (e.g., use of a new or marginally functional fistula is incompletely standardized across data sources), as does calculation of hemoglobin levels over time compared with CMS specifications for QIP measures (15,16). Despite these considerations, we have documented that DOPPS descriptive findings are generally reflective of national dialysis data, because of attention paid to representative cross-sectional sampling and use of sampling weights to report aggregated data. Table 1 demonstrates that data from the DPM sample closely correspond to the national statistics reported by the Elab Project for 97% of US dialysis patients for the fourth quarter of 2010 (17). Similarly, we have shown close correspondence of DOPPS data for the United Kingdom with clinical practice data reported by the UK Renal Registry (18). Such comparisons continue on a regular basis.Table 1: Comparisons of hemodialysis patients in the US DOPPS DPM national sample with CMS Elab census dataIdentifying and Understanding International Trends The DOPPS serves as a resource to monitor and understand temporal trends in dialysis care in each participating country, and to disentangle effects of a policy change in any one country from secular trends internationally. Some policy examples are as follows. In Japan, trends in anemia management from before to after changed ESA reimbursement in 2006, from separately billable (on a per-dose basis) to bundled within the overall dialysis reimbursement (19). This policy change was associated with reduced erythropoietin doses, increased intravenous iron use, and stable hemoglobin levels (19). In Germany, trends in care after the introduction of a weekly dialysis reimbursement rate in 2002–2003, were followed by a quality monitoring system in 2009 based on four parameters (Kt/V ≥1.2, dialysis frequency ≥3 sessions/week, dialysis session length ≥4 hours, hemoglobin ≥10 g/L). Improvements in these metrics occurred before the quality monitoring system; however, the percentage of patients with hemoglobin 90 days at study entry, weighted by facility sampling fraction, in DOPPS 2 (2002–2004), DOPPS 3 (2005–2008), and DOPPS 4 DOPPS, Dialysis Outcomes and Practice Patterns Study; and New Distribution of vascular access type (2002–2011) among countries with stable or catheter burden over time. Cross-sections of patients on dialysis >90 days at study entry, weighted by facility sampling fraction, in DOPPS 2 (2002–2004), DOPPS 3 (2005–2008), and DOPPS 4 DOPPS, Dialysis Outcomes and Practice Patterns Study; of Outcomes The DOPPS has been a key resource to directly mortality across countries because of representative sampling and data collection methods One motivation for the study was the of mortality in the United States compared with Japan and Europe based on data DOPPS analyses confirmed that this even after detailed for patient updated with data from to in Data from the DOPPS, as well as national registries and the indicate that international survival differences among dialysis patients are in part by survival differences in the populations More a DOPPS analysis a dialysis practice as a key cause of international in outcomes survival differences for the United States compared with were by differences in facility vascular access US and European facilities with similar of and catheter use on similar In recent years, with the Fistula First has been a in fistula use in the United in other countries, fistula use has catheter use has (Figures 2 and In the of these the most recent comparisons of survival will be of among hemodialysis patients by in DOPPS 3 (2005–2008), with and without for patient for facility patients. also for patient years on and DOPPS, Dialysis Outcomes and Practice Patterns Study; DOPPS is not intended to provide event which is a of registries national census data in most sites to in the DOPPS may be more and have on than other facilities. Despite this the mortality rates reported by the DOPPS have been similar or slightly those reported by national registries of Practice its a primary aim of the DOPPS has been to identify facility practices associated with the patient outcomes, with on recent years, we have chosen that use in facility practice as a to by differences in health include in which a (e.g., dialysis session may to survival because is often to and in which a clinical (e.g., hemoglobin may because is more in patients One approach to these sources of is for in and now in clinical studies DOPPS analyses using have been central to the study’s effect on hemodialysis care our analyses which as publications which is not an analysis but not results The is that patients are by dialysis facilities, but with the same clinical are to levels of to or For example, facility to provide dialysis by a catheter or to a or dialysis session length are based in part on such as policy, In DOPPS data, we a range of facility practice variation within and between countries (Figures and all this analytic approach can be to and we the approach with other when Distribution of facility time, by DOPPS and to patients not on with >90 days in facilities with patients with time data in DOPPS 1 DOPPS 2 (2002–2004), DOPPS 3 (2005–2008), and DOPPS 4 DOPPS, Dialysis Outcomes and Practice Patterns Study; and New Recent DOPPS analyses of are illustrative of the study of practice variation to help inform We have that many facilities use a single of or even for all patients, and that this single between facilities (i.e., facilities one or Recent DOPPS publications evaluating by analysis and other have the that levels are and have the that the of 2 may the of compared with levels Other analyses of composition are Dialysis In recent years, we have that dialysis session length has in the United has in most other DOPPS countries (Figure By and the DOPPS has that time is associated with mortality in for and other (Figure Facilities with time also have phosphorus and of dialysis session length with mortality and for time on blood and catheter use, stratified by country and phase of study, and for facility In the the time is the of In the the is the time for a patient based on the time practice and facility the United facility measures are not to dialysis session other than measures that are via blood rate and large even with session average is in the United States than in Europe and is much than in Japan a quality monitoring system based on dialysis session length ≥4 was in in 2009, and average session length in that country is now one of the in DOPPS countries (20). In our dialysis session length is now one of the key practice differences between the United States and other DOPPS countries, and its merit attention from research and policy New The DOPPS programs have in recent years with the launch of DOPPS in in 2010 a random sample of dialysis facilities in major in in 2011 a random sample of facilities selected from the national and in the other countries in late 2012, as well as a CKD in in late 2012 major of international expansion is to provide a detailed description of practice variation in participating which we will inform of to practice variation and improve patient outcomes. Other to the DOPPS programs include the Kidney Disease Outcomes and Practice Patterns Study and the Dialysis Outcomes and Practice Patterns Study is launched in countries, and data collection will launch in to of and of care for CKD patients, including and the to dialysis care, which have been to study to We will be relevant because of care is likely to improve outcomes and may soon be to reimbursement from the of care in the United States launched in with the International of will on the major of PD to identify to time on The has for this research because are differences in access to outcomes on this studies to about practice are At the same time, new financial incentives or will likely PD use in the years The DOPPS to research findings directly relevant to patients, health care providers, and policy makers. DOPPS have been published in to and DOPPS are at major national and international DOPPS findings have been to support and clinical practice and have been or reported to health policy and regulatory the The time, facility vascular anemia dialysis dose outcomes, etc.) have been but have our international data and analyses to identify and practice a on findings With the study’s new directions and we to expand toward and quality of life for our patients. is for the DOPPS, and and are with the All are of for The DOPPS studies are by for and by scientific research from Amgen Kyowa Hakko Kirin in Sanofi Renal Vifor Fresenius Renal Pharma and Fresenius Medical 2012), without restrictions on