Abstract
Objective. To determine whether the risk of death from type 1 insulin-dependent diabetes mellitus (IDDM) was similar among young non-Hispanic black, non-Hispanic white, and Hispanic patients.
Design. Retrospective study of death certificates for Chicago residents between 1 and 24 years of age with any mention of diabetes during 1987 through 1994. Prevalence was estimated by an ongoing incidence registry in the city, the 1990 US Census, and published studies. Autopsy reports and/or medical records were examined to determine more clearly the circumstances of death. Case-fatality rates for IDDM in non-Hispanic black, non-Hispanic white, and Hispanic patients were calculated. Deaths in those with diabetes were compared with the mortality experience of the underlying population using race-specific standardized mortality ratios.
Results. A total of 30 diabetes-related deaths occurred in the 8-year interval: 23 among non-Hispanic black, 5 among Hispanic, and 2 among non-Hispanic white paients. The average annual case-fatality rate for all ethnic groups combined was 247.2/105 (95% CI: 166.9–353.5). Race-specific rates were 447.8/105(283.9–671.7) for non-Hispanic black patients, 175.6/105(56.9–409.2) for Hispanic patients, and 48.2/105(5.8–174.0) for non-Hispanic white patients; there were no gender differences in risk. A total of 8 individuals died at the onset of disease (7 non-Hispanic black patients and 1 Hispanic patient). Compared with the underlying population, ethnic-specific standardized mortality ratios were elevated significantly for non-Hispanic black and Hispanic patients but not for non-Hispanic white patients.
Conclusions. Short-term mortality is elevated substantially among non-Hispanic black and Hispanic youth with IDDM. The ninefold greater risk of death for non-Hispanic black compared with non-Hispanic white youth with diabetes may indicate gaps in access to comprehensive diabetes care.
- IDDM =
- type 1 insulin-dependent diabetes mellitus •
- CDPH =
- Chicago Department of Public Health •
- CMTDR =
- Chicago Minority Type 1 Diabetes Registry •
- SMR =
- standardized mortality ratio
Individuals with type 1 insulin-dependent diabetes mellitus (IDDM) are at a markedly elevated risk for short- and long-term complications of the disease. The acute complications of diabetes are related to either very high or very low blood glucose levels; both extremes can lead to severe metabolic and physiologic disruption, coma, and death. Although these occurrences are preventable if they are identified early and treated appropriately, they account for the majority of deaths among those with diabetes of short duration.1 Recent work confirms that the long-term complications of IDDM, particularly renal disease, blindness, and neuropathy, can be forestalled by intensive blood glucose control2; however, this costly option may not be available to those with limited financial or social resources.
The impact of diabetes-related mortality among young people can be expressed in several ways. Panzram3 and others calculated the age-specific reduction in life expectancy after diagnosis (years of potential life lost). Some of the most valuable data are generated when clinic or population-based cohorts of patients are identified at diagnosis and studied over many decades to determine their mortality experience.1 ,4 Other investigators have relied on vital statistics or other government datasets to document the rates and causes of death.5 These diverse approaches have established that there has been a marked improvement in survival of young people with diabetes through 1980, that the causes of death change as the duration of diabetes increases, and that specific factors such as smoking, age at onset, and sociodemographic characteristics can be identified that affect significantly the risk of death for those with IDDM.4
We investigated patterns of early mortality related to IDDM using the resources of the City of Chicago Department of Public Health (CDPH), the Cook County Medical Examiner's Office, and the Chicago Minority Type 1 Diabetes Registry (CMTDR), a population-based registry of all non-Hispanic black and Hispanic Chicago residents developing IDDM before the age of 18. Inclusion criteria and methods for conducting the CMTDR were described previously.6
METHODS
Death record tapes obtained from the Illinois Department of Public Health for the years 1987 to 1994 listed all Chicago resident deaths irrespective of the location of death. Ascertainment was conducted periodically during 1995 and 1996 as computerized records were finalized. We included all deaths with diabetes listed either as an underlying or as any one of four contributing causes of death (ICD-9 code 250); certificates with a trauma code were excluded.7Deaths before 1 year of age were excluded based on the rarity of a diagnosis of diabetes in that age group, as well as the potential for misclassification of a death related to perinatal complications or congenital anomaly. Hospital records, autopsy reports, and medical examiners' investigation reports were obtained and reviewed; deaths associated with type 2 diabetes mellitus or diabetes secondary to another condition were excluded. Race is listed on all death certificates; Hispanic ethnicity, listed on death certificates since 1989, was defined either from the ethnicity field on the death certificate or from a Spanish surname as defined by the US Census Bureau.8 All seven deaths in minority IDDM patients that were ascertained previously through the CMTDR were retrieved using these methods. All-cause mortality data were obtained from the CDPH; race-specific data were unusable for 1987 and 1988 because separate information on Hispanic ethnicity was not recorded consistently on the death certificate before 1989. Therefore, comparisons with the population at large (standardized mortality ratio [SMR] calculations) are restricted to the years 1989 to 1994.
US census data from 1990 for the City of Chicago9 were used to define the population at risk, Chicago residents from 1 to 24 years of age, numbering 443 987 non-Hispanic black individuals, 270 617 Hispanic individuals, and 254 725 non-Hispanic white individuals. The prevalence of IDDM was estimated using a method based on published incidence rates, age-specific survival rates, and the classic epidemiologic relationship of prevalence to average disease duration and incidence.10 This method, which is described in detail in the “Appendix”, produced annual estimates of 642 non-Hispanic black patients, 356 Hispanic patients, and 519 non-Hispanic white patients from 1 to 24 years of age with IDDM.
Mortality rates were estimated directly using events per population at risk per year; 95% CI around these rates were calculated with the Poisson distribution. The microcomputer program Excel11was used to manage and analyze the data.
RESULTS
Sociodemographic Characteristics and Duration
There were 30 diabetes-related deaths in the 8-year interval from 1987 to 1994; 23 deaths were in non-Hispanic black patients, 5 were in Hispanic patients, and 2 were in non-Hispanic white young people. The resulting average annual case-fatality rates (95% CI) were 447.8/105 (283.9 to 671.7) for non-Hispanic black patients with diabetes, 175.6/105 (56.9 to 409.2) for Hispanics, and 48.2/105 (5.8 to 174.0) for non-Hispanic white patients (Fig 1). The 95% CI around the rate for Hispanic patients overlapped those for non-Hispanic black and white patients, indicating no significant differences. However, there was a distinct and statistically significant discrepancy between the rates for non-Hispanic white and black patients, with a black:white rate ratio of 9.3.
Mean annual diabetes case-fatality rate by ethnicity, Chicago, 1987 to 1994
The mean age at death in these young people was 17.1 years (median 19.3; range 7.3–24.7), and the male:female ratio was 1.1 (Fig 2). Of the 13 patients who were known to have died after a specific duration of disease, the average interval from diagnosis to death was 7.1 years (range 1.4–20.0).
Age at death by sex, Chicago, 1987 to 1994.
Comparison With All-Cause Mortality
Overall, non-Hispanic black individuals made up 46% of the 969 329 people who lived in Chicago and were between the ages of 1 and 24 in 1990, but they accounted for 77% of diabetes deaths in this study. Hispanic individuals, comprising 28% of the total population, accounted for 17% of diabetes deaths, whereas non-Hispanic white individuals made up 26% of the population and 7% of diabetes deaths. An important consideration in interpreting these data are that the age-adjusted all cause mortality rates for non-Hispanic black individuals in the general population of Chicago are substantially higher than for non-Hispanic white individuals. Therefore, we calculated ethnic-specific SMRs for IDDM to permit comparisons with the population at large (Table 1). Black and Hispanic IDDM patients, aged 1 to 24 years, had a threefold increased risk of death. In contrast, non-Hispanic white patients with diabetes had approximately the same risk of death as the general population of non-Hispanic white individuals who lived in Chicago and were between the ages of 1 and 24 years. For non-Hispanic black and Hispanic patients with diabetes, this difference was statistically significant, whereas the mortality risk for diabetic non-Hispanic white patients was not significantly different from that of the general population, as the 95% CI around the point estimate of SMR included unity (Table 1).
Standardized Mortality Ratios for Young Chicagoans With IDDM by Ethnicity, 1989–1994
Causes of Death
Examination of the specific circumstances of death shows that 8 deaths (26.7%) occurred at the onset of disease, with diabetes recognized too late to treat successfully, or in 5 cases, with diagnosis made on autopsy (Table 2). There were 9 additional deaths attributable to acute diabetic complications in the 17 persons with known disease. Among 5 young people whose duration of disease was unknown, the cause of 4 deaths was listed only as diabetes or diabetic ketoacidosis.
Circumstances of IDDM Deaths in Young Chicagoans by Ethnicity and Duration of Disease
Of the 23 deaths in black patients, 7 occurred at the onset of diabetes. Overall, acute complications were responsible for 16 (69.6%) of the 23 deaths in black patients and 4 (80%) of the 5 deaths in Hispanics (Table 2). One of the two deaths of non-Hispanic white patients was attributed to diabetic ketoacidosis in a 24-year-old patient whose disease duration was unknown. The other death of a non-Hispanic white patient was a result of chronic renal failure after 20 years of diabetes and also occurred at age 24, the upper limit of the age range. Renal disease was a factor in the deaths of four non-Hispanic black young people, as well.
DISCUSSION
A marked excess of deaths attributable to diabetes was recorded among young non-Hispanic black and Hispanic individuals. If the non-Hispanic white rate applied to non-Hispanic black individuals in Chicago, only 3 of the 23 observed deaths would have occurred. Acute complications accounted for the majority of deaths in both non-Hispanic black and Hispanic individuals, corresponding with findings in other studies of those with short-term disease. In Chicago, non-Hispanic white young people with diabetes had an excellent mortality experience, comparable with those reported in other developed countries,12 demonstrating that a high standard of care can be achieved locally.
The current study is not the first to document the relationship of ethnicity to diabetes mortality and morbidity. Among persons <30 years of age with diabetes, Cowie et al13 found that black individuals developed end-stage renal disease at twice the rate of white patients. In a more general context, the relationship of ethnicity to gaps in medical care has been demonstrated repeatedly.14 Local survey data demonstrate that minority families are more likely to have no health insurance: 24% of non-Hispanic black households and 34% of Hispanic households contained at least one uninsured member in 1995, compared with 10% of non-Hispanic white households in the Chicago metropolitan area.15 The strong inverse correlation of all-cause mortality with socioeconomic status increased nationally between 1960 and 1986 for all race/sex groups,16 whether measured by income or by educational attainment. The observation that social class disparities in health care quality and outcomes exist and are increasing may help to explain the widely different mortality experience of non-Hispanic white and black children with diabetes in the current study.
There are clear limitations to this study. Rates are based on estimates of prevalent cases of diabetes that may or may not be accurate depending on the accuracy of the incidence assumptions and the 1990 census data17 (“Appendix”). However, the magnitude of the ethnic disparity in diabetes-related mortality is so great that small distortions may be unimportant. The death certificate alone was available for 8 (26.7%) of the 30 deaths, and it is certainly possible that additional information on the circumstances of these deaths may have been important. Some deaths may have been overlooked; however, we are reasonably certain that this is not the case for two reasons. First, all registered patients in the CMTDR who died were ascertained using the methods described. Second, the investigative efforts of the Cook County Medical Examiner's office are exceptionally thorough; all unexplained and out-of-hospital deaths in the age range of interest are investigated routinely. Most importantly, there is no plausible reason to suspect that there is a greater risk of missing diabetes-related deaths among non-Hispanic white young people, compared with non-Hispanic black or Hispanic young people.
It may be argued that virtually all the deaths recorded in this study could have been prevented by adequate preventive care or by proper emergency interventions. Metabolic decompensation is relatively easy to recognize in young people, and appropriate treatment for the acute complications of IDDM usually can be delivered in a clinic or hospital emergency department setting. Because highly sophisticated diagnostic and therapeutic procedures are almost never required, inadequacies in clinical care are suspected generally when death occurs from acute IDDM complications. To put these findings into perspective, the vast majority of all deaths in persons between the ages of 1 and 24 are totally preventable; homicide, other unintentional injury, and motor vehicle accidents are the leading causes of death for people who live in Chicago and are between the ages of 1 and 24 irrespective of ethnicity (unpublished data, CDPH). The Diabetes Epidemiology Research International group12 examined mortality in Japan, the United States, Finland, and Israel. These investigators demonstrated important international differences in the risk of dying within the first 25 years of IDDM, although deaths at onset were excluded from their analysis. Death rates were low in Finland and Israel, elevated in the United States, and extremely high in Japan, a nation in which IDDM is very rare. Presumably, the major determinants of international discrepancies in short-term diabetes-related mortality are related to differences in sociodemographic factors and standards of medical care.
The excess mortality recorded among young people who live in Chicago and have diabetes corresponds roughly with findings from other studies; a unique finding in this study is the marked ethnic disparity in risk of death. It is interesting to speculate whether the higher mortality among young persons with IDDM reported elsewhere is also confined to an ethnic or socioeconomic subgroup of the diabetic population. The current data show that a segment of the pediatric population in Chicago is enjoying the recently demonstrated improvement in diabetes survival; both deaths in non-Hispanic white individuals were in the oldest possible age group included, and at least one of these patients had survived to develop long-standing chronic complications. Pathophysiologic differences in the presentation of IDDM could exist, allowing the disease to go unrecognized among non-Hispanic black and Hispanic young people until tragedy occurs; however, there is no plausible scientific evidence to support this possibility. Alternatively, ethnicity is likely to serve as a marker for social and economic inequities that could explain the ninefold difference in risk of death between non-Hispanic black and white children with diabetes.
These findings have several implications. First, in view of the large number of deaths at the onset of diabetes, it is critically important to raise the threshold of awareness of the disease among practitioners treating minority children. Minority parents and teachers need to recognize that IDDM is almost as common among non-Hispanic black and Hispanic children as among non-Hispanic white children.6Second, this study joins earlier papers in documenting continuing ethnic gaps in health outcomes in the United States. Non-Hispanic black and Hispanic individuals are more often uninsured or underinsured, but it is unclear whether this can explain fully the current findings. Treatment standards for the most severe acute and chronic complications of IDDM are accepted widely, and treatment of acute IDDM complications requires interventions that are readily available in any accredited clinical setting. Mortality among non-Hispanic white young people with diabetes was comparable with that in other industrialized countries,12 demonstrating that the resources to prevent unnecessary diabetes-related mortality are available to at least one segment of Chicago's population. These resources must be made accessible to all young people with diabetes, irrespective of their ethnic group or social class.
ACKNOWLEDGMENTS
This work was supported in part by National Institutes of Health Grant R29-DK44751.
The efforts and expertise of Steve Whitman, CDPH; Merwyn Nelson and Dean Lahue, Illinois Department of Public Health; and Kanika Ghai, Cook County Hospital, contributed to the project. Cecilia McKenzie, Sha Li, Henry Choi, Sharon Casillas, and Maria L. Roces served diligently as research assistants.
Members of the Chicago Minority Type I Diabetes Study Group who lent valuable assistance to this project include R. Kendrick and E. Sciammarella, Chicago Department of Health; M. Bousefield, City of Chicago Department of Planning; A. Dekker, and K. Suarez, Chicago Osteopathic Hospital; J. Daboul, D. Edidin, S. Goodman, A. Kohrman, B. Pries, G. Richards, and B. Silverman, Children's Memorial Hospital; P. Mueller, Christ Hospital; H. Weiss, Columbus-Cabrini Medical Center; J. Duff, R. Pildes, H.-J. Rhim Choi, and M. Weigensberg, Cook County Hospital; S. Duck, Evanston Hospital; H. Friedman, F. Medelli, and D. Tibbs, Gottlieb Hospital; S. Drunga and K. Sawyer, Grant Hospital; C. Abraira, Hines VA Hospital; A. Schneider and E. Shelton, Humana-Michael Reese Hospital; M. Winters, Illinois Association of School Nurses; C. Nandi and V. Williams, Illinois Dept of Public Health; M. Martin, Ingalls Memorial Hospital; A. Sion and P. Friedell, Jackson Park Hospital; A. Hagger and K. Reed, LaGrange Memorial Hospital; J. Baron, D. Nyland, and K. Vellodi, Little Company of Mary Hospital; G. Callahan and P. Krautwald, Loretto Hospital; M. Gottschalk and W.P. Zeller, Loyola University Medical Center; E. Baumann, M. Chertack, and W. Maurer, Lutheran General Hospital; D. Pierzchala, Macneal Hospital; C. Menendez, Mercy Hospital and Illinois Masonic Hospital; R. Levin and P. Mukundan, Mount Sinai Hospital; R. Ramsay-Goldman, L. Pachman, and T. Pitts, Northwestern University; G. Northrup, Norwegian-American Hospital; E. Constein, Oak Park Hospital and Westlake Hospital; T. Kenar, Our Lady of Resurrection Hospital; M. Jacob, Ravenswood Hospital; A. Davis, N. Dumbovik, M. Kreiter, and R. Levy, Rush-Presbyterian-St. Luke's Hospital; L. Yu, St Anthony's Hospital; S. Lambert and T. Cardona, St Elizabeth's Hospital; L. Czarkowski, St Francis Hospital, Evanston; B. Pulliam, St James Hospital; B. Danielczyk and M. Lang, South Suburban Hospital; M. Oviedo, Swedish Covenant Hospital; R. Cunningham, B. Price, and J. Schwartz, Evangelical Health Systems Trinity Hospital; R. Briers, J. Cara, L. Cole, F. Cogan, E. Ekwo, K. Herold, A. Kohrman, C. Ober, B. Rich, I. Rosenthal, and F. Thorp, University of Chicago Hospitals; and F. Ziai, University of Illinois and Illinois Masonic Hospitals.
APPENDIX: A Method for Estimating Prevalence From Age-Specific Incidence and Survival With Application to IDDM
The problem addressed here is that of estimating prevalence of a disease or condition based on age-specific incidence rates, age-specific survival rates after diagnosis, and the current age distribution in the population. This method is applied here to estimate the prevalence of IDDM. We assume that cases are not cured, therefore duration of the disease is equivalent to survival with the disease. In the case of IDDM, this is a reasonable assumption.
Methods
Estimation Formula
Prevalence is defined as the proportion of the population that has ever been diagnosed with IDDM. Define random variables:D = age at diagnosis and T = age at death. Age-specific prevalence, defined as:
can be written as:
where I(k) = Pr(D = k‖D ≥ k), incidence at age = k; S * (t) = Pr(T ≥ t‖D ≥ t), survival before diagnosis; S ** (k,t) = Pr(T ≥ k + t‖D = k) or survival after diagnosis at age k; and S(t) = Pr(T ≥ t), survival in the population. Overall prevalence in the population is then a weighted average of age-specific prevalence rates or:
Data for Application to IDDM Prevalence
Age-specific incidence rates for IDDM were obtained from the CMTDR for non-Hispanic black and Hispanic individuals6; between the ages of 0 and 17 years. Because the CMTDR excludes non-Hispanic white individuals, the incidence for that ethnic group was derived from the Jefferson County, AL registry of patients between the ages of 0 and 19 years.18 Age-specific survival rates after diagnosis of IDDM were obtained from the Diabetes Epidemiology Research International mortality study12 and the Swedish Registry.19 Population survival rates and age distributions are obtained from the 1990 US Census.9
Discussion
A standard method from epidemiologic literature for estimating prevalence is to use the simple formula, P =ID × T in which P is prevalence,ID is incidence density, and T is mean duration of disease. This method depends on two assumptions: that the disease of interest is a rare disease and that there is a constant rate of termination that can be described completely by mean duration. The method proposed here is an elaboration of the same basic idea, accounting more accurately for age. Incidence is modeled to account for different incidence rates at different ages. If there is no cure for the disease of interest, the rate of termination becomes the survival rate. Again, we are allowing for different rates of survival depending on age. In both methods, we are making the rare disease assumption.
Footnotes
- Received September 7, 1998.
- Accepted November 17, 1998.
Reprint requests to (R.L.) Division Of Epidemiology-Biostatistics, University of Illinois at Chicago School of Public Health, 2121 W Taylor St (M/C 922), Chicago, IL 60612.
REFERENCES
- Copyright © 1999 American Academy of Pediatrics
In this issue
Jump to section
Related Articles
Cited By...
- Psychosocial Factors in Children and Youth With Special Health Care Needs and Their Families
- Racial-Ethnic Disparities in Management and Outcomes Among Children With Type 1 Diabetes
- Cardiovascular Disease Risk Factors in Youth With Diabetes Mellitus: A Scientific Statement From the American Heart Association
- Increasing Incidence of Type 1 Diabetes in Youth: Twenty years of the Philadelphia Pediatric Diabetes Registry
- Cause-Specific Mortality Trends in a Large Population-Based Cohort With Long-Standing Childhood-Onset Type 1 Diabetes
- Health Inequity in Children and Youth With Chronic Health Conditions
- Risk Factors for Mortality in a Diverse Cohort of Patients With Childhood-Onset Diabetes in Chicago
- Incidence of Type 1 Diabetes in Philadelphia Is Higher in Black Than White Children From 1995 to 1999: Epidemic or misclassification?
- Diabetes as a Tracer Condition in International Benchmarking of Health Systems
- Characteristics of autoimmune hepatitis in patients who are not of European Caucasoid ethnic origin
- Long-Term Trends in Childhood Diabetes Mortality: 1968-1998
- Other Articles Noted