On the basis of information from case series, hemochromatosis has long been associated with several chronic medical disorders. However, no population-based studies have been done to determine the degree to which these disorders are more prevalent in persons with hemochromatosis than in the general population. We used mortality data derived from death certificates to examine national trends and patterns of hemochromatosis-associated death in the United States from 1979 through 1992. We also identified medical conditions associated with hemochromatosis-associated death and determined how often these conditions occurred in persons with hemochromatosis compared with the general population.

Methods

Top Methods Results Discussion Author & Article Info References

We used data from Multiple-Cause Mortality Files compiled by the National Center for Health Statistics for the years 1979 to 1992 [12]. Multiple-Cause Mortality Files include demographic and geographic information on each decedent; International Classification of Diseases [ICD], Ninth Revision, codes for the underlying cause of death; and a list of up to 20 conditions listed on the death certificate. The Multiple-Cause Mortality Files exist in two formats: entity axis and record axis. The entity axis format provides a separate code for each disease listed on the death certificate, whether it is an underlying cause or a contributory condition. The record axis format uses linkage rules to combine some listings of conditions and to determine the underlying cause of death, conditions contributing to death, and the positions of these conditions as listed on the death certificate [12]. We selected all records that contained code 275.0 (disorders of iron metabolism, which includes hemochromatosis) anywhere in the record axis portion. From these records we excluded records that contained codes 273.8 (other disorders of plasma protein metabolism, which includes hereditary atransferrinemia), 277.1 (disorders of porphyrin metabolism, which includes porphyria cutanea tarda), 282.0 to 282.9 (hereditary hemolytic anemias), 285.0 (sideroblastic anemia), and 964.0 (poisoning by iron and its compounds). We called these remaining records "hemochromatosis-associated deaths" and determined the proportion of these deaths for which hemochromatosis (code 275.0) was the underlying cause.

We calculated age-adjusted annual mortality rates per million U.S. residents by using the 1980 U.S. standard population. We used U.S. census data from 1970, 1980, and 1990 to estimate yearly population distribution by age, sex, and ethnic group for the study period (1979 to 1992). We calculated age-specific mortality rates by using the following age groups: less than 1 year, 1 to 4 years, 5 to 9 years, 10 to 14 years, 15 to 19 years, 20 to 24 years, 25 to 29 years, 30 to 34 years, 35 to 39 years, 40 to 44 years, 45 to 49 years, 50 to 54 years, 55 to 59 years, 60 to 64 years, and 65 years and older. Because 92% of decedents who had hemochromatosis were white, we categorized patients as "white" or "nonwhite." We also analyzed the proportion of hemochromatosis-associated deaths according to autopsy status.

To study the association between hemochromatosis and other medical conditions, we evaluated 1) the likelihood that decedents who had hemochromatosis listed on their death certificates also had other medical conditions listed and 2) the likelihood that decedents who had another medical condition listed on the death certificate also had hemochromatosis listed. For the first evaluation, we reviewed all hemochromatosis-associated deaths and identified any other condition that was present on at least 2% of the records. For the second evaluation, we reviewed all Multiple-Cause Mortality File records that listed any of the medical conditions identified in the first evaluation. If at least 0.02% of the records listing a medical condition identified in the first evaluation also listed hemochromatosis, we selected that condition for further analysis. For both evaluations, we stratified deaths into three groups according to the age of the patient at the time of death (<30 years, 30 to 49 years, and 50 years of age). We calculated proportionate mortality ratios [PMRs] to assess 1) whether the presence of a medical condition is more likely (PMR> 1) or less likely (PMR < 1) in deaths in which hemochromatosis is present than in all deaths (PMR₁ = the proportion of deaths in which hemochromatosis is present that are associated with a particular medical condition [exposed] divided by the proportion of all deaths in which hemochromatosis is present [comparison population]) and 2) whether the presence of hemochromatosis is more likely (PMR> 1) or less likely (PMR < 1) in deaths in which a particular medical condition is present than in all deaths [13] (PMR₂ = the proportion of deaths in which a particular medical condition is present that are associated with hemochromatosis [exposed] divided by the proportion of all deaths in which a particular medical condition is present [comparison population]).

In calculating PMRs, we tested for the potential confounding effects of age, sex, and ethnic group by calculating a stratum-specific PMR and its 95% CI for each age-specific, sex-specific, and ethnic group-specific stratum and comparing the stratum-specific PMR to the summary PMR for all deaths. We calculated a 95% CI for each PMR by assuming that the number of deaths associated with hemochromatosis and another medical condition is a Poisson variable [14]. To calculate the denominator of the PMRs, we randomly selected 30% of U.S. deaths from 1979 to 1992.

Results

Top Methods Results Discussion Author & Article Info References

The Multiple-Cause Mortality Files for 1979 to 1992 contain records for 29 million deaths, 4858 (0.017%) of which had hemochromatosis listed in the record axis. Of these, 44.2% (n = 2148) (43.1% of white persons and 50.6% of nonwhite persons) had hemochromatosis listed as the underlying cause of death.

The age-adjusted rate of hemochromatosis-associated death increased from about 1.2 per million in 1979 to 1.8 per million in 1992, a 60% increase (Figure 1). Rates were higher for male than for female persons, but the pattern of increase was similar for the two sexes. The mean rate for 1979 to 1992 was more than twice as high for white persons (1.5 per million) as for nonwhite persons (0.7 per million).

View larger version (14K): [in this window] [in a new window]   Figure 1. Age-adjusted rate of hemochromatosis-associated death by sex (left) and ethnic group (right). Multiple-Cause Mortality Files, United States, 1979 to 1992.

The age-specific rates of hemochromatosis-associated death were higher for infants than for persons 1 to 34 years of age (0.9 per million compared with 0.1 per million; P < 0.01) (Figure 2). The rate was seven times higher for persons 50 years of age and older than for persons younger than 50 years of age (5.6 per million compared with 0.8 per million; P < 0.001). The rate increased dramatically in men at 45 years of age and in women at 55 years of age. Among persons 35 years of age and older, the rate was significantly higher in men than in women. Among persons 49 years of age and younger, the rate was slightly higher for white persons than for nonwhite persons (0.7 per million compared with 0.5 per million; P < 0.01); among persons 50 years of age and older, it was twice as high for white persons than for nonwhite persons (5.1 per million compared with 2.5 per million; P < 0.001).

View larger version (12K): [in this window] [in a new window]   Figure 2. Age-specific rate of hemochromatosis-associated death by sex (left) and ethnic group (right). Multiple-Cause Mortality Files, United States, 1979 to 1992.

We noted a significantly greater proportion of hemochromatosis among deaths with autopsy than among deaths without autopsy (0.032% compared with 0.015%; P < 0.001). The largest difference was in persons 50 years of age and older (0.042% with autopsy compared with 0.015% without autopsy).

Medical conditions listed on at least 2% of Multiple-Cause Mortality File records of hemochromatosis-associated deaths fell into six diagnostic categories (Table 1). Hemochromatosis was more often associated with liver disease (excluding neoplasms) than with cardiac disorders, nonhepatic neoplasms, diabetes mellitus, liver neoplasms, and infectious diseases. Male patients were more likely than female patients to have these comorbid conditions (data not shown). For example, 14.3% of male patients and 2.8% of female patients had both hemochromatosis and liver neoplasms listed on their death certificates. The comorbid conditions liver disease, liver neoplasms, and nonhepatic neoplasms were reported significantly less frequently on the death certificates of nonwhite persons than on the death certificates of white persons (data not shown).

Liver disease, liver neoplasms, cardiomyopathy, diabetes mellitus, and viral hepatitis were more likely to occur among hemochromatosis-associated deaths than among all deaths (all PMR₁ s> 1.0) (Table 2). Combining certain medical conditions, notably liver disease or liver neoplasms with diabetes mellitus, resulted in extremely high PMR₁ s. The PMR₁ for nonhepatic neoplasms was less than 1. The PMR₁ for cardiac disorders was also less than 1, with the exception of that for deaths among persons younger than 40 years of age (PMR₁, 3.1 [95% CI, 2.6 to 3.4]) and that for deaths in persons with cardiomyopathy (PMR₁, 4.8 [CI, 4.3 to 5.3]) (data not shown).

Hemochromatosis was more likely to be present in persons who died with liver disease, liver neoplasms, cardiomyopathy, diabetes mellitus, or viral hepatitis (all PMR₂ s> 1.0) (Table 3). The PMR₂ s were even higher when liver neoplasms or liver disease was combined with diabetes mellitus.

Discussion

Top Methods Results Discussion Author & Article Info References

Hemochromatosis is underdiagnosed, often misdiagnosed, and probably underreported [1, 11, 15]. Because it can lead to numerous chronic conditions, its symptoms are likely to be confused with those of more common diseases, such as alcoholic liver disease, diabetes, and osteoarthritis [1, 4, 15, 16]. Although hemochromatosis is a relatively common genetic disorder, physicians are still not widely aware of it and are often unfamiliar with methods for recognizing and treating it [17]. Most available information on hemochromatosis-associated death has been based on screening trials and case series; our study offers one of the few broad population-based descriptions of this association.

Prevalence

From 1979 through 1992, 1.7 per 10 000 death certificates completed in the United States mentioned hemochromatosis. This prevalence of recognized hemochromatosis at death is similar to the clinically observed prevalence of hemochromatosis, 1.4 per 10 000 deaths, seen in a hospital setting and in the general U.S. population [17, 18]. Recent studies done by using biochemical and genetic measures have estimated the prevalence of hemochromatosis among white blood donors [1, 6] and Hispanic persons [9] to be 50 to 80 cases per 10 000 persons, a considerably higher rate.

Autopsy may help identify cases of hemochromatosis, especially among decedents with cirrhosis of the liver [19]. A survey of pathologists-in-chief of selected hospitals in eight U.S. states in 1965 reported an estimated average prevalence of hemochromatosis of about 18 per 10 000 autopsies [20]. In our study, the prevalence of hemochromatosis was 3.2 per 10 000 deaths with autopsy; this was twice the rate seen among deaths without autopsy. There are at least two possible reasons for the higher frequency of hemochromatosis among deaths with autopsy: Persons with hemochromatosis are more likely to die in institutions with higher autopsy rates (such as academic centers), and autopsy provides an opportunity for gross and microscopic examination of internal organs commonly affected by hemochromatosis.

Nevertheless, rates such as 3.2 per 10 000 and even 18 per 10 000 deaths with autopsy are substantially lower than would be expected based on the frequency in the general population as measured by biochemical and genetic methods. Routine autopsy does not include biochemical testing (such as tests for transferrin saturation) or histologic testing (such as iron staining of liver biopsy specimens) that would identify hemochromatosis; this may account for this low prevalence of hemochromatosis found on autopsy. Another possibility is that many persons with the hemochromatosis genotype died of other causes without manifesting clinical and pathologic evidence of hemochromatosis.

Temporal Trend

The temporal increase in hemochromatosis seen in our study was noted previously. In 1968, when hemochromatosis was first assigned a unique ICD code, the crude rate of death with hemochromatosis as the underlying cause was 0.5 per million; by 1992, this rate had increased to 0.9 per million [11]. This trend could be the result of an increased frequency of the hemochromatosis gene, increased penetrance of the gene, increased recognition of hemochromatosis, increased misdiagnosis of other conditions as hemochromatosis, or a combination of these factors. The frequency of the hemochromatosis gene could have increased if heterozygotes for hemochromatosis bore children more often with other heterozygotes or homozygotes than did persons without the gene for hemochromatosis, but no evidence supports this hypothesis. Because the gene for hemochromatosis was only recently identified, no studies of penetrance of the gene's expression over time have been done. Until evidence for or against temporal changes in the frequency or penetrance of the hemochromatosis gene is available, we must assume that the upward trend in reported hemochromatosis is related primarily to improved recognition of the disease.

Demographic Characteristics

Our findings on the distribution of hemochromatosis by age, sex, and ethnic group and the association of hemochromatosis with other medical conditions generally agree with those of previous studies [1, 3, 5, 16]. In our study, cases of hemochromatosis listed on the death certificates of persons younger than 1 year of age were probably cases of neonatal hemochromatosis, which is considered to be a different disease than hemochromatosis in adults. Neonatal hemochromatosis has not been shown to be HFE-associated and is more common in nonwhite persons [21]. Many studies have found major differences between the sexes in the frequency of diagnosis of hemochromatosis [1, 3, 6, 16]. This clinical discrepancy is attributed to physiologic sources of iron loss in women (for example, through menstruation and pregnancy) and the effect of estrogen on serum iron concentration [8, 22]. In addition, it seems that the clinical disease manifests differently in men than in women; the classic symptoms occur more often in men [23].

Previous studies have suggested that the frequency of HLA-A3 is associated with the high frequency of hemochromatosis in white populations; this HLA type is much less common among black Americans, Native Americans, and Japanese persons, which may explain some of the differences between ethnic groups in frequency of hemochromatosis [1, 10, 24]. Of course, we cannot rule out the possibility of a lack of recognition of hemochromatosis among nonwhite populations because the disease has traditionally been described as prevalent among persons of northern European descent. A recent study [9] suggests that the prevalence of hemochromatosis among Hispanic persons of Mexican or Filipino descent is similar to that among non-Hispanic white persons in the United States. On the other hand, the higher hemochromatosis-associated mortality rates for black infants, children, and young adults than for their white, age-matched peers may be due to the misdiagnosis as hemochromatosis of other iron overload-associated conditions that have a high prevalence among black Americans, such as sickle-cell disease [25, 26]. The recent discovery of the hemochromatosis gene and increased genetic testing in persons with iron overload disease will probably contribute substantially to the understanding of these conditions in all ethnic groups [27, 28].

Association of Hemochromatosis with Other Conditions

In our study, liver disease was the medical condition most commonly associated with hemochromatosis. In most other studies [29, 30], liver disease has been the most common complication of hemochromatosis, occurring in 30% to 94% of persons with hemochromatosis. Our PMR analysis showed that the prevalence of liver disease was 13 times higher among persons who died with hemochromatosis than among those who died without hemochromatosis. Because liver disease is one of the 10 leading causes of death and years of potential life lost in the United States [31] and because the clinical manifestations of hemochromatosis can be prevented, overall mortality rates due to liver disease might be substantially reduced if hemochromatosis were identified more readily and treated early.

Diabetes mellitus was listed on 17% of the death certificates of persons with hemochromatosis-associated deaths-2.4 times more frequently than would be expected for all deaths. This finding is compatible with the lower ranges of results from previous studies [1, 3, 4], which have shown diabetes to be present in 12% to 82% of patients with hemochromatosis. The very high PMRs for hemochromatosis in decedents with the combination of liver disease and diabetes or the combination of liver neoplasms and diabetes suggest that the presence of either combination should prompt immediate evaluation for hemochromatosis.

The likelihood of liver neoplasms alone in decedents with hemochromatosis was 23 times greater than would be expected for all decedents. This markedly elevated risk, however, is much lower than that found in another study, in which patients with hemochromatosis were 200 times more likely than patients without hemochromatosis to die of liver neoplasms. In another study [5], liver neoplasms accounted for nearly one third of deaths in persons with hemochromatosis. In our study, only 10% of the death certificates of persons with hemochromatosis-associated deaths listed liver neoplasms; this suggests that liver neoplasms are underdiagnosed in persons with hemochromatosis-associated death. Rates of liver neoplasms were markedly different among male (14%) and female patients (3%) with hemochromatosis.

Cardiac disorders were reported in 52% of persons who died with hemochromatosis, but the PMR₁ was only 0.9. Whether elevated iron stores are a cause of coronary artery disease has long been debated [32-37]; our finding does not support the contention that elevated iron stores increase risk for coronary artery disease. The PMR₁ for cardiomyopathy among hemochromatosis-associated deaths was 4.7, which is consistent with previous findings [4].

In persons with hemochromatosis-associated deaths, we found no significant association between nonhepatic neoplasms and hemochromatosis (PMR, 0.8). Previous studies have suggested that the risk for nonhepatic neoplasms is increased among persons with elevated iron stores, particularly those with hemochromatosis [5, 38, 39]. However, our study and another recent study [5] failed to show such an increase.

Limitations

Our study was based on data derived from death certificates, and it suffers from the shortcomings of these data [40, 41]. Lists of causes of death on death certificates are frequently incomplete or inaccurate, especially with respect to medical conditions that do not usually result in death or deaths that occur outside of hospitals [40, 42].

The estimated PMR for a particular cause of death is affected by the relative frequency of other causes of death. A high PMR for one cause of death among hemochromatosis-associated deaths may represent a true risk but may also represent a deficit of reported deaths from another cause. In addition, some high PMRs may be due to differential ascertainment by physicians who knew that their patients had hemochromatosis and knew which conditions were likely to be associated with this disorder.

Conclusions

Because of the wide discrepancy between the prevalence of hemochromatosis (on genetic or biochemical testing) in the general population ( 50 per 10 000) and the frequency of mention of hemochromatosis on death certificates (1.7 per 10 000), further studies are needed to assess the magnitude of morbidity and mortality associated with hemochromatosis. We also need a better understanding of the natural history of hemochromatosis before large-scale screening programs for this disorder can be justified. Fortunately, the cloning of the hemochromatosis gene and the discovery of the most common mutations associated with hemochromatosis now make it possible to match phenotypic, genotypic, and clinical data [27, 28].