Kidney International, Vol. 57, Suppl. 74 (2000), pp. S-60–S-65

MORTALITY

Why has the gross mortality of dialysis patients increasedin Japan?

TAKASHI AKIBA, SHIGERU NAKAI, TORU SHINZATO, CHIKAO YAMAZAKI, TATEKI KITAOKA,KAZUO KUBO, and KENJI MAEDA

Patient Survey Committee, Japanese Society for Dialysis Therapy, Nagoya, Japan

Why has the gross mortality of dialysis patients increased in ments came from continuous efforts to increase dialysisJapan? doses in hemodialysis and peritoneal dialysis patients,

Background. The gross mortality of Japanese chronic dial- as well as from implementation of a comprehensive ap-ysis patients gradually increased from 8.1% in 1983 to 9.4%proach to prevent multi-systemic complications in dial-in 1997. We describe the demographic changes in dialysis pa-ysis patients. In contrast, gross mortality of Japanesetients and clarify why the gross mortality has increased.

Methods. Data used in our analysis were taken from annual chronic dialysis therapy has been reportedly increasing.reports by the Patient Registration Committee of the Japanese Our goals in this paper are to describe the demographicSociety for Dialysis Therapy. The relative risk of death was changes in dialysis patients and to clarify problems incalculated by a logistic regression procedure and Cox propor-

improving chronic dialysis therapy.tional hazard model.Results. The first-year survival starting from the initiation

of dialysis therapy slowly improved from 0.836 in 1983 to 0.857METHODSin 1996, but 5-year and 10-year survival consistently decreased

in the same patient groups. The patients’ mean age of beginning Data used in this analysis were taken from the annualdialysis steadily increased from 51.9 in 1983 to 62.2 in 1997.

reports by the Patient Registration Committee of theThe incidence of patients with diabetes as the original diseaseJapanese Society for Dialysis Therapy [4–9]. The Patientshowed a tremendous increase from 15.6% in 1983 to 33.9% in

1997, and patients with nephrosclerosis also showed a twofold Registration Committee’s patient survey method hasincrease in our population. Heart failure accounted for 30–35% been described elsewhere [5–7]. Briefly, at the end ofof mortality until 1991, but this started to decrease after 1992, November from 1968 onward, the committee mailed thewhen recombinant human erythropoietin became clinically

questionnaire form to the head doctors of all facilitiesavailable, and the level fell to 23.9% by 1997. The relative risklisted in the society. The recipients were asked to returnof death in the first year of initiating dialysis therapy decreased

to almost half of the reference year 1983 (P , 0.0001). the survey replies by the end of the following March.Conclusion. The increase in the gross mortality of Japanese The items surveyed were facility characteristics and de-

dialysis patients might not be due to a deteriorating quality of mographics. A longitudinal survey of first-time dialysisdialysis therapy, but rather is a reflection of accepting patientspatients and subsequent follow-up of their clinical andwith older age and comorbid factors contributing to higherlaboratory data was started in 1983. In 1994, we askedrisks of mortality with dialysis therapy.facilities to respond using a Lotus 1-2-3 or MicrosoftExcel worksheet. This collection form was used by 986facilities in 1998, making it possible to increase itemsIn 1984, we described huge differences in the mortalitysurveyed to include factors such as doses of erythropoie-of dialysis patients in the United States, Europe, and

Japan [1]. Life expectancy of dialysis patients age 45–64 tin, cardiothoracic ratio, and predialysis systolic and dia-years in the United States was almost one-third of that stolic blood pressure. Response rates in 1997 were 99.7%for Japanese dialysis patients. Recent publications on for the items concerning facility status and 96.4% forthe mortality of dialysis patients in the United States the patient longitudinal surveys. As of the end of 1997,have consistently reported significant improvements in 340,462 patients’ records had been registered in the pa-gross mortality of hemodialysis and peritoneal dialysis tient longitudinal survey. Data analysis was executedpatients [2, 3]. These reports suggested that the improve- using SAS/STAT software (version 6.03, SAS Institute

Japan, Tokyo). The relative risk of death was calculatedby a logistic regression procedure and Cox proportionalKey words: survival, incidences, dialysis, aging, diabetic nephropathy,

ESRD hazard model as case-mix variables of gender, age, dura-tion on dialysis, and original diseases. Reports from the 2000 by the International Society of Nephrology

S-60

Akiba et al: Mortality of dialysis patients in Japan S-61

Fig. 2. Trends in survival of dialysis patients at first (j), fifth (r), andtenth year (m), by year.

Fig. 1. Recent trends in the gross mortality (%/year) of dialysis patientsin Japan, 1983–1997.

Patient Registry Committee of the Japanese Society forDialysis Therapy were published annually [8]. The re-ports were distributed to all members and participatingfacilities at the annual meetings of the Japanese Societyfor Dialysis Therapy. Excerpts from the reports werepublished in the December issue of the Journal of theJapanese Society for Dialysis Therapy each year [9].

RESULTS

The gross mortality rate of dialysis patients was 8.1%in 1983, gradually increasing to 9.4% in 1997 (Fig. 1).Survival of dialysis patients after the first year followingthe initiation of dialysis therapy slowly improved from0.836 in 1983 to 0.857 in 1997 (Fig. 2). In contrast, five-year survival consistently decreased from 0.626 in pa-tients who started dialysis in 1983 to 0.578 in patients whostarted dialysis in 1992. Ten-year survival also showedconsistent declines.

The mean age of patients beginning dialysis was 51.9in 1983, steadily increasing to 62.2 in 1997 (Fig. 3). Themean age of patients receiving on-going dialysis treat-ment also showed an increase, from 48.3 in 1983 to 59.2 in1997. The average life span of the Japanese population,expressed as a mean of males and females, was 76.99 in1983 and gradually increased to 79.62 in 1994.

Fig. 3. Trends in the average life span of Japanese population (m),Percent changes in the five main original diseases of the mean age of incident patients (r), and the mean age of the prevalent

patients (j), by year.patients initiating dialysis by the year in which dialysis

Akiba et al: Mortality of dialysis patients in JapanS-62

Fig. 5. Trends in the distribution of main causes of death of dialysispatients, 1983–1997. Recombinant erythropoietin became extensivelyadministered in 1992. Rare and unknown causes are not shown.Fig. 4. Distribution of five main original diseases of incident patients

by the year of the initiation of dialysis therapy.

therapy was begun are given in Fig. 4, which shows atremendous increase in diabetic patients from 15.6% in1983 to 33.9% in 1997. Patients beginning dialysis whohad nephrosclerosis also increased twofold, from 3.0%in 1983 to 6.8% in 1997. Patients starting dialysis whohad chronic glomerulonephritis concomitantly decreasedfrom 58.3% in 1983 to 36.6% in 1997, though the numberof new dialysis patients increased from 5,750 in 1983 to10,703 in 1997.

The five main causes of death in dialysis patients wereheart failure, malignancy, infection, myocardial in-farction, and cerebrovascular accident (Fig. 5). Heartfailure accounted for around 30–35% of mortality in1991, but its incidence started to decrease from 1992,when recombinant human erythropoietin (rHuEPO) be-came extensively administered, and the level fell to23.9% by 1997.

Twelve month survival of long-term dialysis patientsdiffered significantly by hematocrit (Hct) levels at theend of 1994 (P , 0.0001) (Fig. 6). Survival rates at thetwelfth month were 0.761 for Hct ,15%, 0.853 for Hct15–20%, 0.897 for Hct 20–25%, 0.939 for Hct 25–30%,0.950 for Hct 30–35%, 0.943 for Hct 35–40%, and 0.945for Hct .40%. Fig. 6. Survival of prevalent patients during 12 months of 1995 by

hematocrit (Hct) levels at the end of 1994.Relative risk of death after 1 year on dialysis was

Akiba et al: Mortality of dialysis patients in Japan S-63

Fig. 7. Relative risk of 1-year death adjustedby gender, age, duration on dialysis, and origi-nal diseases. (A) original diseases (r, non-diabetic; m, diabetes; j, total), (B) age at initi-ation of dialysis therapy (j, 30-44 years; r,45-59 years; m, 60–74 years).

adjusted by gender, age, duration on dialysis, and original coincided with the increase in the age of new dialysispatients (Fig. 3), the increase in patients with diabeticdisease. Relative risk of death by the year when dialysis

therapy started decreased to almost half of that in the nephropathy as an original disease (Fig. 4), and the 7.7%and 11.5% decreases of 5-year and 10-year survival, re-reference year 1983 (P , 0.0001) (Fig. 7A). Relative riskspectively (Fig. 2).of death of diabetes patients decreased to 0.346, though

Better survival with higher hematocrit (Fig. 6) and thethat of non-diabetes patients moderately decreased topercent reduction in deaths from congestive heart failure0.614.coinciding with the introduction of the clinical use ofRelative risk of death was compared by the patientrHuEPO (Fig. 5) might offset the degree of the increaseage at the initiation of dialysis therapy (Fig. 7B). Therein gross mortality to some extent. The relative risk of 1-did not appear to be any significant age-related differ-year death in dialysis patients by the year when dialysisence in risk at the start of the therapy. Relative risk oftherapy was initiated showed significant and continuousdeath was 0.531 for patients aged 30–44 years, 0.528 forimprovements when that of 1983 was used as a referencepatients aged 45–59 years, and 0.429 for patients moreyear (Fig. 7A). The improvement in the relative risk ofthan 60 years in 1995.death was greater in diabetic patients than in non-dia-The changes in the addition of new patients and thebetic patients. In contrast, the risks of death in threedeath of dialysis patients in Japan were graphically com-groups based on age at the initiation of dialysis therapypared (Fig. 8). The incidence of initiation of therapy andshowed recent improvement, although the degree of im-mortality rapidly increased from 1983 to 1997, and theprovement was not different among them (Fig. 7B).annual difference, which stood for the annual increase

The incidence of end-stage renal disease (ESRD) wasof long-term dialysis patients, continuously increasedsuggested to show an exponential increase [2, 10]. Theover the same period. When the initiation incidence andUnited States Renal Data System (USRDS) reportedthe number of deaths were plotted on a semilogarithmicthat incident patients in the United States doubled inscale, they both fitted to straight lines (r2 5 0.990 and8.5 years [2]. Doubling occurred at a faster pace among0.993, respectively). The slope of the mortality waspatients with diabetes (6 years) and hypertension (7steeper than that of initiation incidence.years) as causes of ESRD, as well as in elderly patients (6years for age 75 to 84). Our observations in the Japanese

DISCUSSION population also showed that the aging process of thepopulation and demographic changes in original diseasesOur analysis of data from the Patient Registry over

the fifteen-year period from 1983 to 1997 disclosed that could be the main reasons for these exponential increasesin new dialysis patients [11].the death rate of dialysis patients showed a definite in-

crease from around 8% to 9.5% (Fig. 1). This increase The annual death rates of dialysis patients in the

Akiba et al: Mortality of dialysis patients in JapanS-64

Fig. 8. Trends in incidence (j) and mortality(r) of dialysis patients in Japan. (A) standardscale; (B) semilogarithmic scale.

United States recently showed clear decreases, from 26% rather to the changes in aging, original diseases, or un-in 1988 to 23% in 1995 [3]. Reports have suggested that measured factor(s). More aggressive approaches for thethese decreases in annual mortality might be attributed improvement of dialysis therapy might be needed toto the increased delivery of dialysis dose, shift from cellu- overcome the high risk of patients to be treated in thelosic to synthetic hemodialysis membrane, and the use future.of erythropoietin [3, 11]. The increases in gross mortalityfrom 8–9.5% we observed over the recent 15-year period ACKNOWLEDGMENTSin Japan suggest that Japanese physicians might be pro-

We gratefully acknowledge the staffs of all dialysis facilities in Japanviding poor-quality dialysis therapy to their patients. Our for the generous cooperation with our survey. We also thank I. Matsu-analysis clearly indicated that the risk of death after 1 moto for her secretarial work.year adjusted by age, gender, original disease, and years

Reprint requests to Takashi Akiba, M.D., Ph.D., Department ofon dialysis therapy significantly and continuously de-Internal Medicine and Blood Purification, Tokyo Medical and Dental

creased with changes in the causes of death. The relative University, 1-5-45 Yushima Bunkyoku, Tokyo 113-8519, Japan.E-mail: takiba.med2@med.tmd.ac.jprisk of death for patients with diabetes as a cause of

ESRD in 1996 also decreased to one-third of that in1983. The increase in the gross mortality of Japanese REFERENCESdialysis patients might not be due to deteriorating quality 1. Held PJ, Akiba T, Stearns NS, Murumo F, Turrene MN, Maedaof the dialysis therapy supplied, but rather a reflection K, Port FK: Survival of middle-aged dialysis patients in Japan and

the US, 1988–89, in Developments in Nephrology, Volume 35, Deathof accepting older, higher risk patients for dialysis.on Dialysis, edited by Friedman EA, Dordrecht, Kluwer AcademicThe slope of the mortality plot was steeper than thatPublishers, 1994, pp 23–23.

of the incidence plot on the semilogarithmic scale, sug- 2. U.S. Renal Data System: USRDS 1994 Annual Data Reportgesting that the improvement of dialysis therapy could Bethesda, MD, National Institute of Health, National Institute of

Diabetes and Digestive and Kidney Diseases, 1994.not overcome the increased risk in more elderly patients3. U.S. Renal Data System: USRDS 1998 Annual Data Report,and those with a higher frequency of diabetic nephropa- Bethesda, MD, National Institute of Health, National Institute of

thy. The prevalence of dialysis patients might decrease Diabetes and Digestive and Kidney Diseases, 19984. Marumo F, Maeda K, Koshikawa S: ESRD registry statistics ondue to the future increase in mortality if the background

dialysis mortality, in Developments in Nephrology, Volume 35,features of dialysis therapy such as the aging of the popu-Death on Dialysis edited by Friedman EA, Dordrecht, Kluwer

lation and changes in original diseases stay in the present Academic Publishers, 1994, pp 45–54trend. The chronic dialysis therapy supplied in Japan 5. Shinzato T, Nakai S, Akiba T, Yamazaki C, Sasaki T, Kitaoka

T, Kubo K, Shinoda T, Kurokawa K, Marumo F, Maeda K:recently might face socioeconomic restrictions due toSurvival in long-term hemodialysis patients: results from the annual“the prohibitive cost” of treating “a relatively small num- survey of the Japanese Society for Dialysis Therapy. Nephrol Dial

ber” of patients with this therapy and due to the fear of Transplant 11:2139–2142, 19966. Shinzato T, Nakai S, Akiba T, Yamazaki C, Sasaki T, Kitaoka“endless increases” in the prevalence patients on dialysis

T, Kubo K, Shinoda T, Kurokawa K, Murumo F, Maeda K:therapy. We conclude that the increase in the gross mor-Current status of renal replacement therapy in Japan: results of

tality of Japanese dialysis patients could not be attributed the annual survey of the Japanese Society for Dialysis Therapy.Nephrol Dial Transplant 11:2143–2150, 1996to the changes in the quality of dialysis therapy, but

Akiba et al: Mortality of dialysis patients in Japan S-65

7. Shinzato T, Nakai S, Akiba T, Yamagami S, Yamazaki C, Kitaoka Japan (as of Dec 31, 1995). J Jap Soc Dialy Ther 29:1–22, 1996 (inJapanese)T, Kubo K, Maeda K, Morii H: Report on the annual statistical

10. Port FK: End-stage renal disease: Magnitude of the problem,survey of the Japanese Society for Dialysis Therapy in 1996. Kidneyprognosis of future trends and possible solution. Kidney Int 48Int 55:700–712, 1999(Suppl. 50): S3–S6, 19958. Japanese Society for Dialysis Therapy: An overview of regular 11. Levine A: Anemia patients with renal insufficiency: documenting

dialysis therapy treatment in Japan (as of Dec 31, 1996) Tokyo, the impact and reviewing treatment strategies. Nephrol Dial Trans-Japanese Society for Dialysis Therapy, 1997 (in Japanese) plant 14:292–295, 1999

9. Patient survey committee of Japanese Society for Dialysis 12. Mallick NM, Marasi AE: Dialysis in the elderly, to treat or notto treat? Nephrol Dial Transplant 14:37–39, 1999Therapy: An overview of regular dialysis therapy treatment in