The incidence rate was calculated as number of end-stage renal disease
cases per 1000 patient-years during 5-year follow-up periods after diagnosis
of type 1 diabetes. The last data point of each curve (asterisks) represents
a 7-year period (eg, 30-37 years for those diagnosed in 1965-1969).
The difference between the curves is the cumulative risk of dying without
end-stage renal disease (ESRD).
Finne P, Reunanen A, Stenman S, Groop P, Grönhagen-Riska C. Incidence of End-stage Renal Disease in Patients With Type 1 Diabetes. JAMA. 2005;294(14):1782-1787. doi:10.1001/jama.294.14.1782
Author Affiliations: Finnish Registry for Kidney
Diseases (Drs Finne and Stenman), National Public Health Institute (Dr Reunanen),
and Department of Medicine, Division of Nephrology, Helsinki University Central
Hospital (Drs Groop and Grönhagen-Riska), Helsinki, Finland.
Context End-stage renal disease (ESRD) is one of the most severe complications
of type 1 diabetes. Yet, data on patients’ risk of developing ESRD are
Objectives To estimate the long-term risk of developing ESRD and to assess how
age at diagnosis of diabetes, time period of diagnosis, and sex affect the
Design, Setting, and Patients A cohort of all patients younger than 30 years diagnosed as having type
1 diabetes in Finland in 1965-1999 (n = 20 005) was identified
from the Finnish Diabetes Register. The cohort was followed up from diagnosis
of diabetes until development of ESRD (dialysis or kidney transplantation
as identified from the Finnish Registry for Kidney Diseases), death, or end
of follow-up on December 31, 2001.
Main Outcome Measure Cumulative incidence of ESRD, accounting for death as a competing risk.
Results The cohort was followed up for maximally 37 years, with a median of
16.7 years. During 346 851 person-years, 632 patients developed ESRD.
The cumulative incidence of ESRD was 2.2% at 20 years and 7.8% at 30 years
after diagnosis. The risk of developing ESRD was lowest in patients whose
diagnosis occurred at younger than 5 years. The risk of ESRD was lower for
patients diagnosed as having type 1 diabetes in later years. The risk did
not differ significantly between sexes.
Conclusions With regard to ESRD, the prognosis of type 1 diabetes has improved during
the past 4 decades. Children diagnosed as having diabetes before age 5 years
have the most favorable prognosis. Overall, incidence of ESRD appears to be
lower than previously estimated.
Diabetic nephropathy is one of the most devastating complications in
patients with type 1 diabetes, being the major predictor of premature death.1,2 One third of patients develop microalbuminuria
and about 15% to 25% develop persistent proteinuria during the first 20 years
of living with the disease.1,3- 5 Notably,
the cumulative incidence of hospitalization due to diabetic nephropathy has
been reported to be 20% at 24 years after diagnosis of type 1 diabetes.6 In general, the incidence of diabetic nephropathy
seems to be lower in patients whose diagnosis occurred in more recent years.1,4,7
Although its incidence has decreased, diabetes is still the most important
cause of end-stage renal disease (ESRD) in industrialized countries.8 In Finland, type 1 diabetes accounts for two thirds
of diabetic ESRD cases.9 The cumulative incidence
of ESRD has varied from 4% to 17% at 20 years after diagnosis, and after 30
years it has been shown to be 16%.3,10,11 It
is of note that earlier studies on risk of ESRD have been rather small and
based on patients from single centers3,11 or
on prevalence cohorts instead of incidence cohorts.10 Thus,
large population-based studies with long-term follow-up have not yet been
performed; therefore, the true incidence and age- and sex-stratified risk
estimates of ESRD among patients with type 1 diabetes are not known.
In type 1 diabetes, proteinuria is associated with cardiovascular complications
that account for the majority of early mortality,12 but
in earlier years, the main cause of death among these patients was uremia.1,2 However, the cumulative survival of
patients diagnosed as having type 1 diabetes before 18 years of age has improved
over time and was reported to be 95% to 97% at 20 years after diagnosis.13 Our objectives for the current study were to estimate
the long-term risk of ESRD and death in patients with type 1 diabetes and
to study how age at diagnosis of diabetes, time period of diagnosis, and sex
affect these risks. We were able to do so by combining information from 3
Finnish nationwide population-based registries.
Patients with type 1 diabetes were identified within the Finnish Diabetes
Register, which covers almost 100% of all patients with this type of diabetes
in Finland.14 The cohort comprised all patients
diagnosed as having type 1 diabetes in 1965-1999 before age 30 years. The
register is based on the Central Drug Registry’s records of approvals
of free-of-charge medications. In Finland, insulin treatment for diabetes
has been free since 1964. To qualify as a type 1 diabetes diagnosis, insulin
treatment had to have been started at the time of diagnosis and continued
for at least 1 year and patients had to have been younger than 30 years at
the time of diagnosis. These criteria have remained unchanged over the course
of the study. Information on ESRD through 2001 was obtained from the Finnish
Registry for Kidney Diseases.9 This registry
has an estimated 97% to 99% coverage of all patients accepted for renal replacement
therapy (dialysis or kidney transplantation) in Finland since 1965. Information
on deaths through 2000 was acquired by database linkage with the Population
Register Centre in Finland. The database linkage was possible because of the
Finnish system of unique personal identification numbers for all citizens.
The study is purely based on information derived from national registries.
The Finnish Diabetes Register is maintained by the National Public Health
Institute, a governmental institution. The Finnish Registry for Kidney Diseases
is maintained by the Finnish Association for Organ-Transplant and Kidney Patients
and is also fully financed by the Finnish government. All data were obtained
with written informed consent from the patients, including the understanding
that such data were to be analyzed and included anonymously in registry reports
and scientific publications.
Patients were followed up from diagnosis of type 1 diabetes until occurrence
of ESRD (dialysis or kidney transplantation), death, or end of follow-up on
December 31, 2001. Cumulative incidence of ESRD was calculated with a method
that takes into account the effect of death as a competing risk event, as
described by Kalbfleisch and Prentice.15 The
method allows for the fact that patients who die are no longer at risk of
ESRD. This differs from the cumulative incidence estimated by the Kaplan-Meier
method, which in this case would introduce bias because it would erroneously
assume that those who die remain at risk in the future. Patients were censored
on December 31, 2001. Incidence of all-cause mortality was estimated using
Kaplan-Meier survival probabilities (1 − survival probability).
In the analysis of mortality, death due to any cause was the event and patients
were censored on December 31, 2001.
The incidence rate of ESRD was calculated as number of ESRD cases divided
by number of patient-years (in thousands) during 5-year periods of follow-up.
Adjusted relative risks of ESRD associated with sex, age at diagnosis, and
time period of diagnosis were estimated by fitting a proportional subdistribution
hazards regression model that takes death into account as a competing risk
event.16 Adjusted relative risks of death were
calculated using Cox regression analysis. All possible interactions between
the explanatory variables were tested and found to be statistically nonsignificant.
The effect of ESRD on risk of death was studied by including ESRD as a time-dependent
variable in Cox regression.
The R statistical software, version 1.7.0 (The R Foundation for Statistical
Computing, Vienna, Austria; available at http://www.r-project.org)
with the “cmprsk” package was used for analysis of cumulative
incidence (“cuminc” function) and for proportional subdistribution
hazards regression (“crr” function). An extended Cox regression
model with a time-dependent variable (ESRD) was constructed using Stata statistical
software, release 8.0 (Stata Corp, College Station, Tex). All other analyses
were performed using SPSS 12.0.1 for Windows (SPSS Inc, Chicago, Ill).
In 1965-1999, 20 005 patients in Finland younger than 30 years
were diagnosed as having type 1 diabetes (Table
1). The median follow-up time after diagnosis was 16.7 (range, 0-37.0)
years. During 346 851 person-years of follow-up, 632 cases of ESRD and
1417 deaths were identified (Table 2).
Figure 1 depicts the cumulative
incidence of ESRD according to sex and age at diagnosis of type 1 diabetes.
The cumulative incidence among all type 1 diabetic patients was 2.2% (95%
confidence interval [CI], 1.9%-2.5%) after 20 years and 7.8% (95% CI, 7.1%-8.5%)
after 30 years of follow-up. Correspondingly, the cumulative incidence was
2.1% (1.7%-2.5%) and 8.3% (7.3%-9.3%) for males and 2.2% (1.7%-2.6%) and 7.8%
(6.7%-8.8%) for females. Patients of both sexes diagnosed as having type 1
diabetes before age 5 years had a smaller risk of developing ESRD (3.3% after
30 years; 95% CI, 1.7%-4.9%) than did other patients (8.4%; 95% CI, 7.6%-9.2%).
Figure 2 shows the crude incidence
rate of ESRD. Within 15 years of diagnosis of diabetes, ESRD was rare. Thereafter,
the incidence rate increased rapidly. After 20 years of follow-up, the incidence
rate reached a plateau, which was higher for those diagnosed as having diabetes
during an earlier time period. Between 20 and 30 years after the diagnosis
of type 1 diabetes, the average incidence rate was 7.3 ESRD cases (95% CI,
6.4-8.3) per 1000 patient-years for males and 5.5 cases (95% CI, 4.6-6.5)
per 1000 patient-years for females (rate difference, 1.8; 95% CI, 0.5-3.2).
In multivariate analysis, age at onset and time period of diagnosis
affected the risk of developing ESRD, whereas sex did not (Table 3). Patients aged 0 to 4 years had a lower risk than patients
in other age groups. Patients diagnosed as having type 1 diabetes between
1965 and 1969 had the highest risk of ESRD; thereafter, the prognosis has
The cumulative mortality was 6.8% (95% CI, 6.3%-7.2%) at 20 years and
15.0% (14.1%-15.9%) at 30 years after diagnosis of type 1 diabetes (Figure 3). The most deaths (in absolute numbers)
occurred among patients who had not developed ESRD. The cumulative risk of
dying with ESRD was 0.7% (95% CI, 0.5%-0.8%) at 20 years and 3.3% (2.9%-3.8%)
at 30 years after diagnosis of type 1 diabetes. However, it is noteworthy
that patients with ESRD had a relative risk of 13.1 (95% CI, 11.1-15.3) compared
with other patients with type 1 diabetes when adjusting for age, sex, and
time period of diabetes diagnosis. Notably, male patients had a 66% higher
risk of death due to any cause (Table 3).
The risk of death increased with age at diagnosis. The time period for the
diagnosis of diabetes strongly affected survival: patients with diagnosis
in 1975-1979 had 48% lower risk of dying than those with diagnosis in 1965-1969.
We found that the cumulative incidence (risk) among patients with type
1 diabetes for development of ESRD within 30 years was approximately 7.8%,
which is lower than previously reported.3,11 In
patients whose diagnosis of diabetes occurred before age 5 years, the risk
was found to be considerably lower. The risk was also lower for patients whose
diagnosis occurred in more recent years.
Our estimates are based on data linkage among 3 nationwide Finnish registry
databases, all of which are almost 100% complete. Therefore, virtually all
Finnish inhabitants diagnosed as having type 1 diabetes before age 30 years
were followed up until start of dialysis, kidney transplantation, death, or
end of follow-up in 2001. Our study includes more than 20 000 patients
with type 1 diabetes, of whom 632 cases developed ESRD, and the study is,
to date, the largest one estimating risk of ESRD in type 1 diabetes. Two earlier
studies on incidence cohorts have been published, one including 292 patients
with type 1 diabetes and 44 cases of ESRD3 and
the other comprising 142 patients, of whom 25 developed ESRD.11 Another
study estimated risk of ESRD in 2 populations with type 1 diabetes.10 Altogether, 2369 patients were included and 109 ESRD
cases were identified. However, the patients were not followed up from the
time of diagnosis; therefore, the cumulative risk of ESRD after diagnosis
of type 1 diabetes could not be estimated appropriately. Our study is the
first population-based study in which cumulative risk of ESRD in patients
with type 1 diabetes has been estimated. The large number of patients in our
study enables group comparisons that have not been feasible in earlier studies.
Our study shows for the first time a reduced risk of ESRD in patients diagnosed
as having type 1 diabetes before age 5 years. The finding that risk of ESRD
has decreased over time is also new. Notably, our study population represents
a country with a fairly homogeneous health care system. Furthermore, virtually
all patients are white. Thus, the data may not be reflective of other ethnic
or racial groups or populations with different access to care and health care
The risk of ESRD in Finland was lower than observed in previous studies.
In a cohort from Allegheny County, Pennsylvania, the cumulative incidence
of dialysis was reported to be 13% at 20 years after diagnosis of type 1 diabetes
compared with 2.4% in our cohort.10 The findings
of Krolewski et al11 were more similar to ours,
with a 2% to 8% cumulative incidence of ESRD at 20 years and 15% to 17% at
30 years after diagnosis, depending on age at diagnosis. The incidence of
type 1 diabetes in Finland is one of the highest in the world.17 Because
of this, much emphasis has been put on the Finnish diabetes care system, which
has been nationally organized since the 1960s. It has been speculated that
this may be the reason for the relatively good prognosis of Finnish patients
with type 1 diabetes.18 We further demonstrate
that the prognosis with regard to ESRD has improved during the past decades.
This is in line with earlier findings on persistent proteinuria,1,4,7 but
declining risk of ESRD has not previously been reported. This indicates that
the development of diabetes care has been beneficial. Major milestones include
the introduction of disposable syringes in the early 1970s, the development
of home glucose monitoring and of semisynthetic and synthetic human insulin
in the 1980s, and multiple insulin injection regimens, the effect of which
improved with the introduction of rapid-acting insulin in the 1990s.
The risk of persistent proteinuria and ESRD has been shown to be lower
among patients diagnosed as having diabetes at younger than 10 years than
among those whose diagnosis occurred during the ages of 10 and 20 years.3,4,11 But, increased risk
of ESRD among patients with diagnosis occurring at ages 5 to 9 years compared
with those whose diagnosis occurred before age 5 years has not been reported
previously. There are, however, some indications that patients diagnosed as
having diabetes before age 5 years have lower rates of complications such
as microalbuminuria, retinopathy, and hospitalization due to nephropathy.6,19,20 The reasons are not
known. One explanation may be that patients with early-onset diabetes are
genetically different from those who develop diabetes later,21,22 which
also could lead to differences in the susceptibility to develop ESRD. There
might be other explanations as well. Good self-management of diabetes correlates
with good metabolic control.23 Self-care is
a skill that can be compared with other skills—such as speaking—and
is better managed the earlier it is acquired. Children who become used to
dealing with diabetes at an early age might adhere to the insulin treatment
and diet better than those who face the requirements of diabetes at an older
We also analyzed the cumulative incidence of ESRD according to attained
age. Patients whose diabetes was diagnosed at age 0 to 4 years reached the
same risk of developing ESRD about 5 years later than those whose diagnosis
occurred at age 5 to 9 years and 10 years later than those whose diagnosis
occurred at age 10 to 14 years. Thus, patients diagnosed as having diabetes
before age 15 years reached a similar risk of ESRD at virtually the same age
(eg, 5% cumulative risk of ESRD at an approximate age of 33 years). This supports
the notion that the prepubertal duration of diabetes contributes less than
the postpubertal duration to the risk of diabetic complications.20 However,
patients diagnosed as having diabetes at older than 15 years reached the same
cumulative risk of ESRD at a considerably later stage in life.
Earlier studies have shown that only a subset of patients with diabetes
(approximately one third) develop persistent proteinuria and that the incidence
rate of this complication peaks between 15 and 20 years after diagnosis of
type 1 diabetes, after which time it starts to decline.1- 4 One
study also demonstrated decreasing incidence rates of ESRD after 20 years
of diabetes duration.11 In contrast, we found
that the incidence rate of ESRD, after a rapid increase between 15 and 20
years after diagnosis of type 1 diabetes, reached a plateau and thereafter
remained stable up to 35 years of follow-up. However, ESRD is an end point
of nephropathy occurring on average 10 years later than proteinuria.3 Decreases in the incidence rate of ESRD may become
apparent only after even longer follow-up. The incidence rate of ESRD was
virtually zero during the first 15 years after diagnosis of diabetes. Because
more than 40% of the patients (n = 8442) received diagnoses in 1987
or later and, thus, were followed up for less than 15 years, a substantial
proportion of the patients did not contribute to the long-term risk estimates
of ESRD. Notably, in these patients only 5 cases of ESRD occurred, showing
that the risk of developing ESRD within 15 years of diabetes debut is almost
nonexistent. This is probably a reflection of effective early prevention,
which reduces the risk of developing nephropathy.5
For the calculation of cumulative incidence, we used methods that correct
for death as a competing risk event. In contrast with the Kaplan-Meier method
and the standard Cox regression, the methods used herein take into account
that deceased patients no longer are at risk of developing ESRD. Thus, the
statistical methods used herein consider that a high mortality rate reduces
the risk of ESRD because the patients die before they develop the complication.
When death was considered as a competing risk event, there was no statistically
significant difference in risk of ESRD between male and female patients in
multivariate analysis, whereas if the competing risk was not accounted for
(Cox regression), males had 18% higher risk of ESRD than females. This is
explained by the fact that male patients had considerably higher mortality
and died more frequently before they developed ESRD. Thus, male patients could
have had a greater risk of ESRD than female patients, if they had lived long
enough to develop this complication.
Type 1 diabetes is associated with a considerable premature mortality.18 The 20-year cumulative mortality of patients diagnosed
as having diabetes before age 18 years was previously reported to be 3.1%
in Finland, 4.6% in Israel, and 5.5% in a cohort in Allegheny County, Pennsylvania.13 In this study, the same cumulative risk of death
was 4.0%, which is comparable with the earlier findings. The risk of death
was higher among males and patients with diagnosis occurring at an older age.
Patients diagnosed as having diabetes in more recent years showed a considerably
lower risk of dying. In earlier years, uremia was considered the most important
cause of death among proteinuric diabetes patients.1,2 In
this study, approximately every fifth death during the first 30 years of follow-up
occurred among patients who had developed ESRD. Among 222 deceased ESRD patients
in this study, uremia was identified as the cause of death in only 1 patient,
whereas cardiovascular complications caused two thirds of the deaths, which
is in line with earlier findings on patients with diabetes who were successfully
treated for renal failure.24 However, because
cardiovascular complications are known to be more common among patients with
diabetes who have proteinuria,2,12 nephropathy
can serve as an indirect cause of mortality. This was evident in our study,
which showed a 13-fold risk of death among patients with ESRD compared with
other patients with type 1 diabetes. This emphasizes the severity of ESRD
as a complication of diabetes.
In conclusion, our data indicate improved prognosis of type 1 diabetes
with regard to both ESRD and death. Patients younger than 5 years at diabetes
onset have the most favorable prognosis. The overall incidence of ESRD appears
to be lower than previously reported.
Corresponding Author: Patrik Finne, MD,
PhD, Department of Clinical Chemistry, University of Helsinki, Biomedicum
Helsinki A424b, PO Box 700, FIN-00029 Helsinki, Finland (email@example.com).
Author Contributions: Dr Finne had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Finne, Reunanen,
Acquisition of data: Finne, Reunanen, Stenman,
Analysis and interpretation of data: Finne,
Reunanen, Groop, Grönhagen-Riska.
Drafting of the manuscript: Finne, Reunanen,
Stenman, Groop, Grönhagen-Riska.
Critical revision of the manuscript for important
intellectual content: Finne, Reunanen, Groop, Grönhagen-Riska.
Statistical analysis: Finne, Reunanen, Grönhagen-Riska.
Obtained funding: Finne, Grönhagen-Riska.
Administrative, technical, or material support:
Finne, Reunanen, Stenman, Grönhagen-Riska.
Study supervision: Finne, Reunanen, Grönhagen-Riska.
Financial Disclosures: None reported.
Funding/Support: Dr Finne was supported with
a grant from the Liv och Hälsa Foundation.
Role of the Sponsor: The sponsor had no role
in the design and conduct of the study; collection, management, analysis,
and interpretation of the data; or preparation, review, and approval of the
Acknowledgment: We acknowledge the nephrologists
and staff in all central hospitals in Finland, which have reported to the
Finnish Registry of Kidney Diseases: Helsinki University Central Hospital,
Turku University Central Hospital, Satakunta Central Hospital, Kanta-Häme
Central Hospital, Tampere University Hospital, Päijät-Häme
Central Hospital, Kymenlaakso Central Hospital, Etelä-Karjala Central
Hospital, Mikkeli Central Hospital, Itä-Savo Central Hospital, Pohjois-Karjala
Central Hospital, Kuopio University Central Hospital, Keski-Suomi Central
Hospital, Etelä-Pohjanmaa Central Hospital, Vaasa Central Hospital, Keski-Pohjanmaa
Central Hospital, Oulu University Hospital, Kainuu Central Hospital, Länsi-Pohja
Central Hospital, Lappi Central Hospital, and Åland Central Hospital.