Reduced incidence of end stage renal disease among the elderly in Denmark: an observational study

During the nineties, a number of studies were published, showing that antihypertensive therapy in patients with chronic kidney disease (CKD) delayed the progression of uremia. Whether intensive antihypertensive therapy, with the aim of reducing blood pressure to below 130/80 (as opposed to conventional therapy) is per se effective, is still controversial. The original MDRD study [1], comparing low and high intensity antihypertensive therapy, found that beneficial effects were limited to patients with proteinuria >1 g/day, and other large studies, such as the REIN-2 [2] and AASK [3] trials, have failed to demonstrate an overall effect of intensive treatment. However, the ESCAPE trial of 385 children with CKD, demonstrated a 35% reduction in uraemia progression. Anithypertensive therapy, regardless of type, reduces proteinuria [1, 3]. In contrast, there is no doubt that renin-angiotensin system (RAS) blockade has a specific protective effect over and above other antihypertensive agents, presumably because of its extra anti-inflammatory and anti-proteinuric effects. Early studies suggested that uraemia progression could be delayed by 35% [4, 5]. Since then a number of large studies have shown that RAS blockade prevents the development of diabetic nephropathy [6] and reduces uraemia progression by 30-40% [7‐10]. These studies have recently been reviewed [11]. The effect is present both in proteinuric and non-proteinuric diseases, but the effect is greater, the greater the degree of proteinuria and the achieved reduction in proteinuria. We hypothesized that these papers will have led to national antihypertensive prescription changes which would in turn result in a reduction in the incidence of end stage renal disease (ESRD).

While the national population only rose by 8.3% during the study period, considerable changes in the age structure were seen. The population aged 60–69 years increased from 492,000 to 683,000 (38.9%), 70–79 years from 367,000 to 386,000 (5.3%), and over 80 from 188,000 to 227,000 (21.5%).

The age of the oldest incident ESRD patient in the registry rose almost linearly from 75.3 years in 1985 to 81.8 in 1990 and 95.1 in 2007, after which it stabilized, an average increase of 7.6 months/year. The average age rose from 51.6 ±15.9 to 61.9 ± 16.1 in 2001 and to 64.5 ±18.2 years in 2010. The Charlson Comorbidity Index remained steady for patients <50 years, but increased significantly for older age groups (Table1, Figure1).

Data concerning national prescription rates for the four most common antihypertensive drugs were available for 1995–2010 (Figure2). During this period consumption rose from 24.5 to 101.3 DDD/capita/year, a four-fold increase. The proportion of consumption attributed to renin-angiotensin system (RAS) blockers rose from 34 to 59%.

The incidence of ESRD is shown in Table2, and the absolute patient numbers in Table3. The incidence of patients <60 years remained approximately constant during the entire period of observation (Table2). A fall in incidence of 19% between 2001 and 2011 for this group was not significant. Since 2001 the ESRD incidence has fallen by 7% among 40–49 year olds, and 26% among 50-59-year-olds. These changes were not significant. The incidence of patients 60–70 years (Figure3) rose from 167 patients per million of population (ppm) in 1990 to 400 ppm in 2001, and then fell steadily to 234 ppm, a 42% fall (p < 0.01). The incidence of patients 70–80 years rose from 106 ppm in 1990 to 593 in 2002, and then fell to 398, a 33% fall (p < 0.03). Active treatment of patients over 80 years was virtually nonexistent at the beginning of the study period, rose slightly to 58 ppm in 1997, and then rose rapidly to 557 ppm in 2007. It has since fallen to 341 ppm, a 39% fall (NS). The secular trend for incidence for the years 2000–2011 was significant for those aged 60–70 (r = −0.91, p < 0.001), but not for patients 70–80 years (r = −0.45, NS).

The contributions of risk reduction population structure changes and population number changes is shown in Table4. For all groups, a fall in incidence was noted, which was independent of changes in population number and structure. The fall was greatest for patients 60–80 years.

In order to analyze which renal diagnoses had experienced the highest fall in incidence, incidence rates for the years 2000–01 were compared with the years 2009–2010 for common renal diagnoses. The results are shown in Table5. Improvements were seen over a wide range of diagnoses, with no clear distinction between proteinuric (e.g. diabetic nephropathy and glomerulonephritis) and non-proteinuric (e.g. CIN and polycystic renal disease). A post-hoc analysis of sub-diagnoses in the diagnosis group “Other”, showed falls in most subgroups, but the number of patients in each subgroup did not permit statistical analysis. However, the incidence of patients with vasculitis (ICD-10 codes M30.x and M31.x) among patients aged 60–79 fell 58% from 20.8 ppm/year to 8.7 (p < 0.03).

The incidence of actively treated ESRD has increased continuously since the start of maintenance dialysis therapy in the sixties. This has been largely driven by a steady increase in take-on rates. There are two reasons for this. Firstly, economic growth and a steady reduction in the costs of dialysis, mean that health services have been able to afford treating more patients. Secondly, the results of treating patients of increasingly high age and morbidity, in particular DM, have improved, such that active treatment of marginal groups is justified. This pattern can be seen in this national survey. During the period, a linear increase in the age of the oldest patient in the registry was seen, and the age-adjusted morbidity increased. This increase, approximating 6 months per year, was far greater than the increase in expected life expectancy of the background population. During the nineties, increased incidence rates for 60-80-year-olds and diabetics [12] were seen. After 2000, in response to encouraging treatment results [13], the incidence of 80-year-olds rapidly increased, while the incidence of diabetics stabilized [12]. Thus, it is highly likely that the initial rise in incidence in elderly age groups is an expression of these secular trends. The increase in incidence rates for Type 2 diabetics over the age of 70 after 2001 is probably also part of this trend. One would therefore expect that incidence rates at some point would stabilize at a higher level.

The present study shows that ESRD incidence among patients <40 years has remained stable for 20 years. Since 2001 the ESRD incidence has fallen by 7% among 40–49 year olds, and 26% among 50-59-year-olds. These changes were not significant. Since 2001 the ESRD incidence among 60-70-year-olds has fallen by 42%, and 70-80-year-olds since 2002 by 33%. The Riskdiff analysis (Table4) shows that these changes are independent of population structure. A recent fall in incidence since 2007 of 39% among patients older than 80 years is too recent to be meaningful. It could be due to a real increase in incidence similar to the 60–80 year-olds, or a waning enthusiasm by primary health carers for referring these often frail patients for treatment. This is in contrast with other national results [14, 15] which have merely showed a stabilization in these age groups. We believe this to be one of the first cases of a national reduction in ESRD incidence in the absence of social or economic unrest. As a direct result the number of prevalent dialysis patients in Denmark has now fallen by 5% since 2008. Taiwan has also noted a fall in ESRD incidence from 432 to 384 ppm between 2005 and 2008; this has not however yet resulted in a fall in prevalence [15].

During the nineties, a number of possibly modifiable factors in the progression of uraemia were identified. Aggressive treatment of hypertension is probably important [1, 16]. RAS blockade by ACE-I and A2A were shown to have specific nephroprotective properties, in particular in patients with diabetes and proteinuria [4, 5, 9, 10, 17, 18]. A protective effect of protein restriction has been suggested [19], as has a nephrotoxic effect of tobacco [20]. Increased use of these prophylactic measures would be expected to reduce ESRD incidence. Figure2 shows that the prevalence of antihypertensive therapy, and in particular RAS blockade in the general population has indeed increased substantially, to a level of 0.27 DDDs/capita/year. The percentage of non-smokers (or irregular) has increased from 58 to 77% between 1995 and 2008 [7]. There is no evidence that the incidence of hypertension has increased in Denmark; thus the increased drug use is probably an expression of more intensive individual therapy. This study has only documented an increase in general antihypertensive therapy, in particular RAS blockade, and a reduction in ESRD incidence among the elderly. This being an observational study, any discussion about causality must be purely speculative, but it is possible that the observed decrease in ESRD incidence is an expected consequence of the intensified prophylaxis. If this is so, there appears to be a lag time of at least 5 years between a change in antihypertensive therapy and a decrease in incidence. This is not surprising: in order to delay ESRD significantly, treatment has to be initiated while the patient still has a significant renal function. During the period of observation, there were no governmental changes in the organization or financing of ESRD treatment. All patients are treated at hospital-based, publicly financed nephrology centers. There has been an increased awareness of the importance of predialysis nephrology care, and today all patients with a GFR below 30 ml/min are recommended specialist care. This may have contributed to the fall in ESRD rates, independently of concurrent antihypertensive therapy and RAS blockade. In common with international trends, there has probably been a tendency to start dialysis at a higher level of GFR since 2000; this would a priori increase the number of ESRD patients slightly.

Considerable changes have occurred in the background population between 1990 and 2011. The average longevity has increased from 72.2 to 77.8 years for males, and 77.3 to 81.6 for females [21]. While increased longevity will of course be expected to increase the absolute numbers of elderly patients, it will not in itself affect the incidence, expressed as a fraction of the population at risk. The Riskdiff analysis shows that the observed changes are real and independent of any change in population structure. It shows that the evolution of the population 60–80 years would have lead to a rise of incidence of 16% while the observed incidence was −21%. The underlying risk fell by 36%. Ischaemic heart disease as a cause of death fell from 25.6% of all deaths to 9.2%, and cerebrovascular disease from 9.1% to 6.9%. These changes could also partly be related to more intensive antihypertensive therapy. It is difficult to predict how these changes might affect ESRD incidence: on the one hand, since cardiac and renal disease are often related, with common etiologies such as diabetes, atherosclerosis and hypertension, a better cardiac survival might lead to more patients surviving to renal failure; on the other hand, the prophylactic treatments that reduce the incidence of heart disease might also reduce the incidence of renal disease.

Two findings were surprising. The fall in incidence was distributed between different renal diagnoses, without any clear distinction between proteinuric and non-proteinuric diseases. A change in coding practices during the period of observation cannot be excluded, but since the diagnoses were made by a small group of nephrologists, using standard ERA-EDTA definitions, we consider this unlikely. Also, the fall in incidence was mainly confined to patients over the age of 60. While ESRD incidence was lower for patients aged 30–59 was lower in 2011 than 2001, the difference was smaller and non-significant, partly because of the small number of patients in these age groups. There are several possible explanations for this apparent difference: some diagnoses, common among younger patients, such as polycystic renal disease and hereditary disorders, may be less amenable to prophylaxis; early diagnosis and prophylaxis may be rarer among younger patients; it is possible that long-term therapy is required to make a noticeable difference. Antihypertensive therapy has been shown in the ESCAPE study to also be effective in children [16]. This is a recent study, which cannot have affected previous therapy; no data is available concerning antihypertensive therapy among children in this population. A further disadvantage of this study is that data concerning antihypertensive use was only available after 1995, and only as DDDs, rather than number of patients being treated.

Not all health indicators have moved in the right direction. The number of obese adults (body mass index >30 kg/m²) has risen from 5.5% in 1987 to 7.6% in 1994 and to 13.4% in 2010. It is thus all the more remarkable that the expected epidemic in diabetic nephropathy has not occurred, and that the incidence of type 2 diabetic nephropathy is stabilised. Thus, the theory that intensive prophylactic intervention can reduce the incidence of diabetic nephropathy seems to have been justified in practice.

It is possible that unidentified factors could have contributed to the fall, e.g. a reduction in consumption of nephrotoxic drugs or an improvement in the urological treatment of patients with post-renal uremia. Improvements in immunosuppressive therapy may have contributed to the fall in vasculitis incidence. It is even possible that the initial increase in incidence seen among patients over 70 years is partly caused by a postponement of ESRD among patients 60–70 years to a later age, the real fall in incidence thereby being exaggerated.

If one assumes that there is a causal connection and that, without intervention, incidence among 60-70-year-olds would have remained at 400 ppm and among 70-80-year-olds at 592, a rough estimate of the possible economic benefits of prophylaxis can be made. Further assumptions are required for this calculation: the mean survival after ESRD is 4.3 and 2.5 years respectively (DNR average 2000–2010); the average cost is €50,000/year. It then follows that ESRD incidence has been reduced by 33.8 ppm/year, ESRD prevalence by 121 ppm and ESRD expenditure by approximately €6 per capita/year. Drug expenditure needs to be subtracted from this to calculate the net economic benefit. For commonly used ACE inhibitors this is however less than 10 cents/day.

During the period of observation, a significant fall in ESRD incidence in the population between 60–80 years age was seen, and also an increase in prescription rates for antihypertensive drugs, particularly RAS blockade. It is possible that these two phenomena are connected. The findings suggest that ESRD is a preventable disease.

All patients resident in Denmark, and thus possessing a national identity number, starting active treatment for ESRD between 1.1.1990 and 31.12.2011 were included in the study. Their data were extracted from the following databases:

ESRD incidence rates were calculated for different age groups and renal diagnoses. Renal diagnoses up to 31.12.2010 were categorized as shrunken kidneys (ICD-10 code N18.x, Q60.5), chronic glomerulonephritis (N02.x-N07.x), chronic interstitial nephritis (CIN) (N11.x-N15.x, N20.x, N31.9, Q62.x), polycystic renal disease (Q61.x) hypertensive (I12.x), type 1 diabetic (DM) (E10.x), type 2 diabetic (E11.x), other.

¹vHerlev Hospital, University of Copenhagen, Graevlingestien 9, 2880, Bagsvaerd, Denmark. ²Research Unit of Clinical Epidemiology, Centre for National Clinical Databases – South, University of Southern Denmark and Odense University Hospital, Odense, Denmark.