Potential life-years gained over a 5-year period by correcting DOPPS-identified modifiable practices in haemodialysis: results from the European MONITOR-CKD5 study

Background DOPPS reported that thousands of life-years could be gained in the US and Europe over 5 years by correcting six modifiable haemodialysis practices. We estimated potential life-years gained across 10 European countries using MONITOR-CKD5 study data. Methods The DOPPS-based target ranges were used, except for haemoglobin due to label changes, as well as DOPPS-derived relative mortality risks. Percentages of MONITOR-CKD5 patients outside targets were calculated. Consistent with the DOPPS-based analyses, we extrapolated life-years gained for the MONITOR-CKD5 population over 5 years if all patients were within targets. Results Bringing the 10 MONITOR-CKD5 countries’ dialysis populations into compliance on the six practices results in a 5-year gain of 97,428 patient-years. In descending order, survival impact was the highest for albumin levels, followed by phosphate levels, vascular access, haemoglobin, dialysis adequacy, and interdialytic weight gain. Conclusions Optimal management of the six modifiable haemodialysis practices may achieve 6.2% increase in 5-year survival. Trial Registration NCT01121237. Clinicaltrials.gov registration May 12, 2010 (retrospectively registered). Electronic supplementary material The online version of this article (10.1186/s12882-019-1251-z) contains supplementary material, which is available to authorized users.


Background
The Dialysis Outcomes and Practice Patterns Study (DOPPS) [1] reported [2] that an additional 143,617 patient life-years could be gained in the United States over a 5-year period if six modifiable haemodialysis practices were brought into best practice compliance. Four of the six modifiable haemodialysis practices (dialysis dose, phosphate control, improved anaemia, and partial correction of serum albumin) are supported by published guidelines [3][4][5][6] while reduced inter-dialytic weight gain [7,8] and reduced use of catheters for vascular access [9][10][11][12] have been linked to mortality. These analyses were replicated for Belgium [13], Spain [14], France [15] and Sweden [16], except that albumincorrected serum calcium was substituted for intradialytic weight gain. Here, we present a simulation of potential life-years gained by correcting these six practices across 10 European countries using data from the MONITOR-CKD5 study.

Definition and Selection of Target Ranges for Modifiable Practices
We used the same target ranges set forth in the 2004 DOPPS analysis [2] for dialysis dose (Kt/V ≥ 1.2), phosphate control (serum phosphate ≤5.5 mg/dL), and fluid management (interdialytic weight gain ≤5.7%). For partial correction of serum albumin, we applied the range used in the Belgian [13], Spanish [14], and French [15] replications (≥4.0 g/dL). Best practice guidelines for anaemia management have changed since 2004 with more conservative haemoglobin targeting and individualized risk-based dosing of erythropoiesis-stimulating agents [5,18,19]. Hence, the original [2] haemoglobin target of ≥11 g/dL was reduced to ≥10 g/dL in our analysis. For the evaluation of catheter use for vascular access, all DOPPS-based analyses [2,[13][14][15][16] used a facility-level variable (percentage of catheter use at the centre-level).
In the MONITOR-CKD5 study, vascular access was not collected as a facility-level but as a dichotomous patientlevel variable. Some centres contributed small numbers of patients, hence determining centre percentages would have been subject to bias.
Estimation of "Out of Range" Table 2 presents the recommended targets for each modifiable factor for the original DOPPS study [2] and the replication studies [13][14][15][16] and the percentage of each study sample outside the target range. We calculated the valid percentages of patients outside the indicated range for the MONITOR-CKD5 evaluable sample (n = 2023).

Selection of Mortality Relative Risk Values
In MONITOR-CKD5 patients were followed for up to 24 months, which is shorter than in DOPPS. Per expert opinion, except for haemoglobin and catheter use, we adopted the mortality RR values generated by the five country-specific analyses (Table 2) [2,[13][14][15][16] based on congruence in definition of target range between the MONITOR-CKD5 and these studies, and similarities in population. Due to guideline-recommended changes in Hb targets since 2004, the RR of 1.22 associated with Hb < 10 g/dL reported by Locatelli et al. was utilized [20]. This is an estimate adjusted for age, gender, comorbidities, mobility, malnourishment, ability to eat independently, years with end-stage renal disease (ESRD), country, and facility. Since MONITOR-CKD5 study measured catheter use at the patient level, the mortality RR adopted for this parameter was based on findings from a patient-level analysis of mortality risk associated with vascular access from DOPPS [21]; this estimate was adjusted for age, gender, race, years with ESRD, body weight, and comorbidities to reduce treatment-by-indication bias.

Population Estimation
Haemodialysis prevalent and incident counts for the 10 countries in MONITOR-CKD5 are presented in Table 3. Data for 7 of the 10 countries were obtained from the 2014 ERA-EDTA Registry Annual Report [22] Slovenia was included in the 2013 [23] but not the 2014 ERA-EDTA Report; hence, 2013 data were utilized. Neither Germany nor Italy are included in the ERA-EDTA Registry. The most recent and sufficiently detailed data from their respective national databases were used instead, which were 2006 data for Germany [24] and 2013 data for Italy [25]. Country-level prevalent and incident counts were extrapolated from year of data to 2017 using an annual growth rate of 5%. However, this growth rate was not included in the calculation of projected life-years over the 5 years, thus providing a conservative, lower-bound estimation of life-years gained.* Patients younger than 20 years of age are not reported 1

Calculation of Attributable Patient-Years
We replicated the Port et al. [2] methodology for estimating potential life-years gained using the MONITOR-CKD5  proportions of patients out of range for each practice, the current population statistics for the 10 countries in the study, and the published relative risks associated with each practice. Based on the estimated percentages of patients falling outside the target ranges of the six modifiable practices and the mortality risk associated with each, Port el al.
extrapolated to the US haemodialysis population for a 5-year period to quantify potential life-years gained [2]. Projections were obtained by computing the difference in the area under the 5-year survival curve between the general survival of haemodialysis patients based on the actual mortality rate and the estimated survival if patients outside targets were brought into target range on the six modifiable factors based. The projected total haemodialysis patient-years summed over the 5 years of the analysis period PY is defined as.
where t is the total time period in years, FD the fraction still on dialysis at t, L the annual rate of haemodialysis patients lost from the haemodialysis population due to death or transfer to transplantation or peritoneal dialysis, N 0 the number of patients prevalent at the start of the 5-year time interval, and N 1 the number of incident patients entering the haemodialysis population during each year. PY is calculated for status quo (PY 0 ), where none of the patients outside the target range are brought into compliance; for each modifiable factor i assuming all patients are brought into compliance (PY i where i = Kt/V, phosphate, Hb, interdialytic weight gain, albumin, AVF/graft); and for the condition that all patients are brought into compliance on all six factors (PY a ).
From Eq. 1, we estimated the potential patient-years gained over the 5-year time interval ΔPY 5 , for a given modifiable factor and aggregated across all six factors, as the difference between PY 0 (the projected total patient-years summed over 5 years) and the corresponding PY i for the ith factor or PY a for the six factors aggregated; or.

Ethics
The MONITOR-CKD5 study protocol was approved by the ethical review committees of participating centres in accordance with national laws and regulations. Written informed consent was obtained from all participating patients.

Consort
N/A; this is not a clinical trial but a simulation study.  Table 4 summarizes the patient life-years gained over five years, estimated per Eqs. 1 through 3 above, if respectively an optimal 100% (or suboptimal 50%) of patients falling outside of the recommended target range on the six modifiable practices were brought into compliance. Results are reported for each modifiable factor separately, as well as for all 6 factors combined. The baseline for the 10 MONITOR-CKD5 countries is the estimated 303,517 haemodialysis patients for 2017. This baseline is adjusted upward for the estimated 66,597 incident cases added over the 5-year period, adjusted downward for the variable annual death rate in general and associated with compliance with each factor, and further adjusted downward for the fixed annual loss rate to peritoneal dialysis or transplant.

Estimated Patient Life-Years Gained
The estimated potential patient-years gained over the 5-year period at status quo (PY 0 ) is 1,563,221. At the singular modifiable factor level, bringing all patients into Kt/V compliance yields 9319 additional patient-years over PY 0 (for a total of 1,572,540); into phosphate compliance an additional 22,052 patient-years (1,585,273); into Hb compliance an additional 12,826 patient-years (1,576,047); into interdialytic weight gain compliance an additional 7222 patientyears (1,570,443); into albumin compliance an additional 92,344 patient-years (1,655,565); and into vascular access compliance an additional 21,746 patient-years (1,584,967). On the aggregate, bringing all patients into compliance on all six modifiable factors results in an incremental gain of 97,428 patient-years for a total of 1,660,650 patient-years (note that years gained if all 6 factors in target range is less than the sum of the years gained on the 6 individual factors since the modifiable practices are not mutually exclusive; i.e. patients can be outside target range on ≥1 factor).

Discussion
Using as baseline the estimated 303,517 haemodialysis patients estimated to be alive in 2017 in the 10 European countries that participated in the MONITOR-CKD5 study, and adjusting this upward for growth in this patient population and downward for mortality and transition to peritoneal dialysis or transplantation, we projected that the 5-year period starting in 2017 would yield a net total of 1,563,221 patient-years under the haemodialysis practice patterns observed in the study. However, as the DOPPS investigators have shown for several countries [2,[13][14][15][16], mortality in haemodialysis patients could potentially be reduced and life expectancy increased by focusing on six modifiable factors: dialysis quality (Kt/V), managing phosphate, haemoglobin, and albumin levels, reducing interdialytic weight gain (or, alternately managing calcium), and minimizing catheter use for vascular access.
If all haemodialysis patients in the 10 MONITOR-CKD5 countries were brought into the target range of these six modifiable practice patterns, an additional 97,429 patient life-years would hypothetically be gained over the 5-year period starting in 2017. This corresponds to a 6.2% increase in 5-year survival. In comparison, the corresponding rates from the country-specific DOPPS analyses of the modifiable practices are, in ascending order, 8.2% for the US, 13.1% for France, 14.7% for Spain, 17.8% for Sweden, and 22.5% for Belgium [2,[13][14][15][16]. The differences in rates may be due to sample size. However, considering that these simulation analyses cover a time span of approximately 15 years, this variation in rates may also suggest significant improvements having been achieved in haemodialysis care over time [28]. Though we chose not to perform countryspecific analyses, with collectively 25.3% of MONITOR-CKD5 patients being treated in Central and Eastern European countries, our lower rate in life-years gained underscores the advances in the quality of haemodialysis care made across all European regions. This is also evident from the loss rates used in the various analyses. Consistent with their time period and country, Port et al. [2] assumed a 24.0% annual death rate and 6.5% annual rate of other loss (transition or peritoneal analysis or transplantation), for a total loss rate of 30.5%. In the four country-specific DOPPS analyses [13][14][15][16], the annual death rates ranged from 17.1 to 24.3%, the annual other loss rates from a gain of 2.1% to a loss of 5.3%, for total loss rates between 17.0 and 27.4% In our ten-country European analysis, these rates were 15.0% (death-related), 5.3% (other), and 20.3% (total).
With the historical trends of fewer patients outside of the recommended range on each of the modifiable practices and decreasing mortality rates in dialysis populations, there is a diminishing opportunity for further improvement and, hence, smaller gains in potential life years. Yet, there is still room for improvement. While the percentage of patients out of range have decreased across all six modifiable practices, only 15% of patients in the MONITOR-CKD5 study were in range on all six. The historical differences are also evident at the level of the individual practice patterns. In the Port et al. [2] analysis, reducing catheter use was associated with the greatest gain in patient life-years (with the highest potential gains in the USA), followed by increasing serum albumin levels (a measure of nutritional status and inflammation), and phosphate control. In our analysis, the greatest gains in theoretical patient life-years were achieved by targeting patient's serum albumin level followed, remotely, by phosphate control and vascular access. In the country-specific DOPPS analyses (which did not include interdialytic weight gain but serum calcium level instead), albumin levels and catheter use were the modifiable factors that prevailed in impact, followed by calcium levels and, variably, phosphate and Hb levels. Despite these differences in rates, all analyses converged on the importance of bringing albumin levels into target range, assuring vascular access by means of fistula or graft, and mineral control. That dialysis quality contributed only modestly to survival in all analyses may reflect the longstanding acceptance of the Kt/V ≥ 1.2 standard. One notable challenge for clinicians is that while serum albumin exerts the most influence on mortality in these models in terms of relative risk, it is perhaps the most resistant of the six practices to therapeutic intervention, since it is linked to different factors, mainly inflammation and nutritional status [29].
Mortality risks attributable to the six modifiable practice patterns have not been systematically evaluated in randomized controlled trials (RCT); though a RCT of an established practice such as dialysis quality of Kt/V ≥ 1.2 versus < 1.2 would be ethically untestable. Hence, estimating preventable mortality and associated life-years gained as a function of modifiable practices may be appropriate as secondary endpoints in randomized trials. Further, the role of patient adherence, in particular to dietary restrictions, phosphate binders, and dialysis regimen, should be examined in future studies [30][31][32].
Our analyses have limitations. Like DOPPS, MONITOR-CKD5 is a non-controlled study. The use of patient-level versus facility-level data for catheter use may introduce treatment-by-indication bias especially in terms of patient age, comorbid conditions, and disease severity. The adjusted mortality risk associated with patient-level vascular access data utilized in our study [21] reduced such confounding but some residual bias cannot be excluded. Countries in MONITOR-CKD5 differed in some of the modifiable practices MONITOR-CKD5 [28]. The estimation of life-years gained using the MONITOR-CKD5 aggregate sample may conceal specific countries where opportunities for larger gains exist. Not being part of the ERA-EDTA Registry, national reports were used for Germany and Italy. The MONITOR-CKD5 patient population was virtually Caucasian with minimal racial/ethnic variation. manuscript. IA was responsible for the study design, the statistical analysis, interpretation of results, the writing, and final approval of the manuscript. KM was responsible for the study design, the statistical analysis, interpretation of results, the writing, and final approval of the manuscript. All authors agree to be accountable for all aspects of the work.

Ethics approval and consent to participate
The MONITOR-CKD5 study protocol was approved by the ethical review committees of participating centres in accordance with national laws and regula-

Consent for publication
Not applicable because this report does not contain data for individual patients.
Competing interests CC, JM, DG, FD, PZ, and GL received compensation from Hexal AG for their participation in the work reported here. AK is an employee of Hexal AG. KD, KM, and IA are affiliated with Matrix45; by company policy, they cannot hold equity in sponsor organizations and cannot receive direct personal benefits, financial or other, from sponsor organizations. Matrix45 provides similar services to other biopharmaceutical companies without exclusivity constraints.

Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.