GASTROENTEROLOGY
1991;101:368-372
Long-Term Survival Analysis in Hereditary Hemochromatosis PAUL
C. ADAMS,
MARK SPEECHLEY,
Departments of Medicine and Epidemiology London, Ontario, Canada
This study investigated the long-term survival rates hemochromatosis. of 85 patients with hereditary Eighty-five patients with documented hereditary hemochromatosis diagnosed between 1958 and 1989 and followed up at the University Hospital (University of Western Ontario) medical center were retrospectively reviewed for this analysis. The current status of the patient was assessed by interview or written questionnaire completed by the patient or the family physician. Estimates of differences in survival rates were obtained using Kaplan-Meier life-table and Cox regression analysis. Liver histology, clinical features of the disease, and number of venesections were analyzed to determine their relationship to survival. In the course of a mean follow-up interval of 8.1 + 8.8 years (range, O-31 years), there were 17 deaths among the 85 hemochromatosis patients. Patients with cirrhosis at the time of diagnosis were 5.5 times more likely to die than noncirrhotic patients. Patients who were noncirrhotic at the time of diagnosis had an estimated survival that was not significantly different from age- and sex-matched members of the normal population. Diabetes did not increase the risk of death after data were controlled for the presence of cirrhosis. Early diagnosis and treatment of hemochromatosis in the precirrhotic stage can lead to long-term survival similar to that in the general population. The presence of cirrhosis significantly increases mortality and is the major clinical factor affecting survival.
H
and ANN E. KERTESZ
and Biostatistics,
emochromatosis is now recognized as one of the most common autosomal recessive diseases in the white population, with an estimated gene frequency of 1:20 (1). There is evidence to suggest that depletion of body iron stores by venesection therapy will prolong life in patients with hemochromatosis, although this effect has never been shown in a controlled trial (z-5). This study analyzes the vari-
University
of Western Ontario,
ables affecting long-term survival in 85 patients with hemochromatosis who have undergone venesection therapy. Patients and Methods Patient Population The diagnosis of hemochromatosis was based on the clinical history, physical examination, and serum ferritin and transferrin saturation and was confirmed by liver biopsy in all patients. Liver biopsy specimens were classified as normal, fibrotic or cirrhotic without knowledge of the clinical outcome of the patients. The proband case was the first diagnosed case in the family, and discovered cases were other family members subsequently found to have hemochromatosis. Supporting data such as hepatic iron concentration, radioiron absorption studies and HLA typing, and presenting clinical features were available in many of the patients and have been previously reported (6-9). Patients with other conditions associated with iron overload such as iron-loading anemias, transfusional iron overload, and dietary iron overload were excluded. Patients were considered diabetic if they required long-term treatment with insulin or oral hypoglycemic agents to control hyperglycemia. The presence of arthritis was established by history and physical examination. Patients were treated by the removal of approximately 500 mL of blood on a weekly basis until the serum ferritin level was 14 years. Pa-
100
Figure 1. Survival analysis curves comparing cumulative survival (96) in 85 hemochromatosis patients with a hypothetical age- and sexmatched population from the same region. Numbers on the graph refer to the number of patients in each group at risk for each follow-up interval after deaths and censoring. A 99% confidence interval was calculated by using 1.98 x SE. The survival was sig nificantly reduced in most periods, with the expected survival rates clearly lying outside the confidence interval calculated for survival.
80
60
-
40
-
20
-
I
0’ 0
I
10
5
Follow
I
up
interval
15
(years)
20
370
ADAMS ET AL.
GASTROENTEROLOGY Vol. 101, No. 2
tients with cirrhosis had a significantly decreased survival, with cumulative survival of 72% at 5 years, 62% at 10 years and 46% at 20 years. The noncirrhotic patients had cumulative survival of 96% at 5 years, 93% at 10 years, and 93% at 20 years (Figure 2). There was no significant difference in survival between patients in this latter group and a hypothetical cohort of age- and sex-matched patients from Ontario (12). Effects of Clinical Variables on Survival
The effects of clinical variables such as cirrhosis, diabetes, arthritis, age, and venesection therapy on survival were analyzed, and the percent dying and mean survival times for each independeat prognostic factor are listed in Table 1. Differences in survival times for cirrhosis, diabetes, and arthritis are statistically significant using the log rank test. Differences in survival between cirrhotic and noncirrhotic patients are plotted in Figure 2. Patients with fibrosis and normal liver are combined in the noncirrhotic group for several reasons. While the simple relative risk for fibrosis relative to normal liver histology is 4.25, the mean survival times are similar. Cirrhosis, on the other hand, confers a simple relative risk of 11 and is associated on average with 5 fewer years of survival. Because there were only 12 cases of fibrosis, estimates of its true effect on mortality were less precise. Finally, fibrosis was not significantly different from normal liver in a stepwise survival model. The multivariate model used stepwise survival analysis and included disease covariates (diabetes, arthritis, and cirrhosis) and age at diagnosis. The results of this analysis showing the order of entry for each prognostic factor, the improvement x2 statistic and associated
100
Table 2. Stepwise Survival Analysis (Cox Regression) of 85 Hemochromatosis Patients, Showing Order of Entq and Covariates
Variable
Entry order
Cirrhosis“ Arthriti? Age (Y#
1 2 3
Improvement
x2: “P < 0.001; bP < 0.025.
Coefficient
SE
1.712
0.586
1.448 0.0483
0.622 0.0216
Coefficient/SE
Adjusted relative risk
2.922
-2.326 2.230
5.54
0.24 1.05
probability value, and the adjusted relative risk are shown in Table 2. Liver histology was the first variable entered, followed by arthritis and age. Diabetes, which was marginally significant in bivariate analysis (P = 0.04), failed to pass the entry criterion of P < 0.15. After adjustment for the effects of age and arthritis, death occurred in patients with cirrhosis more than 5.5 times as frequently as in those with normal livers or fibrosis. Death occurred in patients with arthritis only one quarter as often as in those without arthritis after adjustment for cirrhosis and age at diagnosis. Effects of Venesection
Therapy on Survival
The mean number of venesections to deplete iron stores was 43 -+ 51. Estimates of the therapeutic effects of venesection therapy are complicated because mortality is curvilinear across quartiles of number of venesections, as shown in Figure 3. Those receiving fewer than 8 or more than 64 venesections died at approximately 6 times the rate of those receiving intermediate numbers of venesections. As
L--, 56 ------------, 48 _____________________---______-------------. 32
II 60
-
40
-
20
-
0' 0
6
I
I
I
5
10
15
Follow
up
interval
(years)
Figure 2. Cumulative survival (94~)in 27 cirrhotic and 58 noncirrhotic patients. Numbers on the graph refer to the number of patients in each group at risk for each follow-up interval after deaths and censoring. Survival was significantly reduced in cirrhotic patients as compared with noncirrhotic patients (P < 0.004;log-rank test).
SURVIVAL ANALYSIS IN HEMOCHROMATOSIS
August 1991
n
=
20
n =
21
-
8
9
25
n =
19
n =
371
20
45 40 35 Figure 3. Percent of patients dying by quartile of number of venesections in 80 hemochromatosis patients. None of the patients who died in the first quartile completed venesection therapy for iron depletion before death. There were 8 deaths in the first quartile (6 cirrhotic, 2 noncirrhotic), I death in the second quartile (cirrhotic), 2 deaths in the third quartile (1 cirrhotic, 1 noncirrhotic) and 6 deaths in the fourth quartile (5 cirrhotic, 1 noncirrhotic).
30 25 20 15 10 5 0 5 0
-
Number
might be expected, the average follow-up time increases from the lowest to the highest quartile (4, 7, 7, and 13 years, respectively), suggesting that the increased mortality in the group receiving the fewest treatments was caused by advanced disease at the time of diagnosis. The average follow-up time among the four patients with cirrhosis in the lowest quartile of venesections was 1 year (all died): in the fourth quartile, the average follow-up time among cirrhotic patients was 11 years, during which time 5 of 12 died. Discussion
This retrospective study followed the clinical course of 85 patients with hemochromatosis for up to 31 years, with an average follow-up period of 8.1 years. The data suggest that patients whose conditions are diagnosed and treated before the development of cirrhosis have a survival time similar to that of the normal population (Figure 1). Cirrhosis is the major predictive clinical factor affecting survival, with cirrhotic patients faring considerably worse than noncirrhotic patients despite therapy. The apparent improved survival time in patients with arthritis compared with those without arthritis is not readily explainable, although our patients who died of liver failure within a year of diagnosis may not have survived long enough to develop arthritis, which has a high prevalence in the aging general population. Another possibility is that the presence of arthritis led to earlier diagnosis and treatment of the disease before the development of cirrhosis. With the advent of insulin therapy and the widespread use of venesection therapy, the cause of death in hemochromatosis has changed from complications of diabetes and heart
of
26
-
63
64
+
venesections
failure to liver failure and hepatocellular carcinoma. The cause of death in our patients was clearly related to hemochromatosis in only 8 of 17 cases, with liver disease and/or hepatocellular carcinoma accounting for 7 of these deaths. The beneficial effect of iron depletion by venesection therapy has never been subjected to controlled studies, but several studies have suggested a beneficial effect of therapy, particularly in the precirrhotic stage of the disease (2-5). In the absence of a control group of similar patients who did not receive therapy, the effect of venesection therapy on long-term survival is difficult to evaluate. Patients in our study who had advanced disease at the time of diagnosis died in the hospital before initiation or completion of venesection therapy. Patients with severe iron loading, as assessed by the number of venesections required to deplete the iron stores, also had an increased mortality rate: this finding is consistent with the observations of Niederau et al. (2), who observed an increased mortality in patients who required more than 18 months of venesection therapy. In 1935, Sheldon reported that patients with hemochromatosis had a life expectancy of l-5 years from the time of diagnosis (13). Following the institution of insulin therapy, Finch and Finch reported that the principal causes of death in patients with hemochromatosis were hepatic failure, portal hypertension, cardiac failure, and infection, with an average survival of 4.4 years after signs of clinical disease became manifest (14). The cumulative survival of treated hemochromatosis patients in previous studies is shown in Table 3. The improved survival rates in the present study and in the study by Niederau et al. (2) may reflect a higher percentage of discovered and likely precirrhotic patients because of
372
ADAMS ET AL.
GASTROENTEROLOGY Vol. 101, No. 2
Table 3. Cumulative Survival Reported in Treated Hemochromatosis Patients
5. Milder MS, Cook JD, Stray S, Finch CA. Idiopathic
Survival (%) Author
Year
Powell (3) 1970 Bomford and Williams (4) 1976 Milder et al. (5) 1980 Niederau et al. (2) 1985 Adams et al. 1991
6.
n
5yr
10yr
15yr
20yr
51 85 34
78.2 66 70
32 70 76 81
30 59 71
30 49 71
163 85
92 87
7.
8.
9.
more widespread use of HLA typing and family screening. In conclusion, this study shows that noncirrhotic patients treated for hemochromatosis have an excellent long-term prognosis, with survival rates comparable to those of the general population. The increased mortality associated with the presence of cirrhosis emphasizes the need for diagnosis in the precirrhotic stage of the disease.
10.
11.
12.
13. 14.
References
hemochromatosis, an interim report. Medicine [Baltimore) 1980;59:3449. Adams PC. Hepatic iron in hemochromatosis. Dig Dis Sci 1990;35:690-692. Adams PC, Frei JV, Bradley C, Lam D. Hepatic iron and iron absorption in hemochromatosis. Clin Invest Med 1990;13:256259. Borwein ST, Ghent CN, Flanagan PR, Chamberlain MJ, Valberg LS. Genetic and phenotypic expression of hemochromatosis in Canadians. Clin Invest Med 1983;6:171-179. Adams PC, Kertesz AE, Valberg LS. Clinical presentation of hemochromatosis: a changing scene. Am J Med 1991;90:445449. Benedetti J, Yuen K, Young L. PlL-Life tables and survival functions. In: Dixon WJ, ed. BMDP Statistical Software. Berkeley, CA: University of California Press, 1983:557-575. Hopkins A. PZL-survival analysis with covariates--Cox models. In: Dixon WJ, ed. BMDP Statistical Software. Berkeley, CA: University of California Press, 1983:576-594. Statistics Canada. Life tables, Canada and provinces 19801982. Catalogue 84-532. Ottawa: Ministry of Supply and Services, 1984. Sheldon JH. Haemochromatosis. New York: Oxford University, 1935. Finch SC, Finch CA. Idiopathic hemochromatosis, an iron storage disease. A. Iron metabolism in hemochromatosis. Medicine (Baltimore) 1955;34:381-430.
Adams PC, Halliday JW, Powell LW. Early diagnosis and treatment of hemochromatosis. Adv Intern Med 1989;34:111126.
Niederau C, Fischer R, Sonnenberg A, Stremmel W, Trampisch HJ, Strohmeyer G. Survival and causes of death in cirrhotic and in noncirrhotic patients with primary hemochromatosis. N Engl J Med 1985;313:1256-1262. Powell LW. Changing concepts in haemochromatosis. Postgrad Med J 1970;46:200-209. Bomford A, Williams R. Long term results of venesection therapy in idiopathic hemochromatosis. Q J Med 1976;45:611623.
Received July 30,199O. Accepted December 15,199O. Address requests for reprints to: Paul C. Adams, M.D., Department of Medicine, University Hospital, P.O. Box 5339, London, Ontario, Canada N6A 5A5. Dr. Adams acknowledges the grant support of the Medical Research Council of Canada and is a Career Scientist of the Ministry of Health of Ontario. The authors acknowledge the advice and assistance of Dr. L. S. Valberg.
1991;101:368-372
Long-Term Survival Analysis in Hereditary Hemochromatosis PAUL
C. ADAMS,
MARK SPEECHLEY,
Departments of Medicine and Epidemiology London, Ontario, Canada
This study investigated the long-term survival rates hemochromatosis. of 85 patients with hereditary Eighty-five patients with documented hereditary hemochromatosis diagnosed between 1958 and 1989 and followed up at the University Hospital (University of Western Ontario) medical center were retrospectively reviewed for this analysis. The current status of the patient was assessed by interview or written questionnaire completed by the patient or the family physician. Estimates of differences in survival rates were obtained using Kaplan-Meier life-table and Cox regression analysis. Liver histology, clinical features of the disease, and number of venesections were analyzed to determine their relationship to survival. In the course of a mean follow-up interval of 8.1 + 8.8 years (range, O-31 years), there were 17 deaths among the 85 hemochromatosis patients. Patients with cirrhosis at the time of diagnosis were 5.5 times more likely to die than noncirrhotic patients. Patients who were noncirrhotic at the time of diagnosis had an estimated survival that was not significantly different from age- and sex-matched members of the normal population. Diabetes did not increase the risk of death after data were controlled for the presence of cirrhosis. Early diagnosis and treatment of hemochromatosis in the precirrhotic stage can lead to long-term survival similar to that in the general population. The presence of cirrhosis significantly increases mortality and is the major clinical factor affecting survival.
H
and ANN E. KERTESZ
and Biostatistics,
emochromatosis is now recognized as one of the most common autosomal recessive diseases in the white population, with an estimated gene frequency of 1:20 (1). There is evidence to suggest that depletion of body iron stores by venesection therapy will prolong life in patients with hemochromatosis, although this effect has never been shown in a controlled trial (z-5). This study analyzes the vari-
University
of Western Ontario,
ables affecting long-term survival in 85 patients with hemochromatosis who have undergone venesection therapy. Patients and Methods Patient Population The diagnosis of hemochromatosis was based on the clinical history, physical examination, and serum ferritin and transferrin saturation and was confirmed by liver biopsy in all patients. Liver biopsy specimens were classified as normal, fibrotic or cirrhotic without knowledge of the clinical outcome of the patients. The proband case was the first diagnosed case in the family, and discovered cases were other family members subsequently found to have hemochromatosis. Supporting data such as hepatic iron concentration, radioiron absorption studies and HLA typing, and presenting clinical features were available in many of the patients and have been previously reported (6-9). Patients with other conditions associated with iron overload such as iron-loading anemias, transfusional iron overload, and dietary iron overload were excluded. Patients were considered diabetic if they required long-term treatment with insulin or oral hypoglycemic agents to control hyperglycemia. The presence of arthritis was established by history and physical examination. Patients were treated by the removal of approximately 500 mL of blood on a weekly basis until the serum ferritin level was 14 years. Pa-
100
Figure 1. Survival analysis curves comparing cumulative survival (96) in 85 hemochromatosis patients with a hypothetical age- and sexmatched population from the same region. Numbers on the graph refer to the number of patients in each group at risk for each follow-up interval after deaths and censoring. A 99% confidence interval was calculated by using 1.98 x SE. The survival was sig nificantly reduced in most periods, with the expected survival rates clearly lying outside the confidence interval calculated for survival.
80
60
-
40
-
20
-
I
0’ 0
I
10
5
Follow
I
up
interval
15
(years)
20
370
ADAMS ET AL.
GASTROENTEROLOGY Vol. 101, No. 2
tients with cirrhosis had a significantly decreased survival, with cumulative survival of 72% at 5 years, 62% at 10 years and 46% at 20 years. The noncirrhotic patients had cumulative survival of 96% at 5 years, 93% at 10 years, and 93% at 20 years (Figure 2). There was no significant difference in survival between patients in this latter group and a hypothetical cohort of age- and sex-matched patients from Ontario (12). Effects of Clinical Variables on Survival
The effects of clinical variables such as cirrhosis, diabetes, arthritis, age, and venesection therapy on survival were analyzed, and the percent dying and mean survival times for each independeat prognostic factor are listed in Table 1. Differences in survival times for cirrhosis, diabetes, and arthritis are statistically significant using the log rank test. Differences in survival between cirrhotic and noncirrhotic patients are plotted in Figure 2. Patients with fibrosis and normal liver are combined in the noncirrhotic group for several reasons. While the simple relative risk for fibrosis relative to normal liver histology is 4.25, the mean survival times are similar. Cirrhosis, on the other hand, confers a simple relative risk of 11 and is associated on average with 5 fewer years of survival. Because there were only 12 cases of fibrosis, estimates of its true effect on mortality were less precise. Finally, fibrosis was not significantly different from normal liver in a stepwise survival model. The multivariate model used stepwise survival analysis and included disease covariates (diabetes, arthritis, and cirrhosis) and age at diagnosis. The results of this analysis showing the order of entry for each prognostic factor, the improvement x2 statistic and associated
100
Table 2. Stepwise Survival Analysis (Cox Regression) of 85 Hemochromatosis Patients, Showing Order of Entq and Covariates
Variable
Entry order
Cirrhosis“ Arthriti? Age (Y#
1 2 3
Improvement
x2: “P < 0.001; bP < 0.025.
Coefficient
SE
1.712
0.586
1.448 0.0483
0.622 0.0216
Coefficient/SE
Adjusted relative risk
2.922
-2.326 2.230
5.54
0.24 1.05
probability value, and the adjusted relative risk are shown in Table 2. Liver histology was the first variable entered, followed by arthritis and age. Diabetes, which was marginally significant in bivariate analysis (P = 0.04), failed to pass the entry criterion of P < 0.15. After adjustment for the effects of age and arthritis, death occurred in patients with cirrhosis more than 5.5 times as frequently as in those with normal livers or fibrosis. Death occurred in patients with arthritis only one quarter as often as in those without arthritis after adjustment for cirrhosis and age at diagnosis. Effects of Venesection
Therapy on Survival
The mean number of venesections to deplete iron stores was 43 -+ 51. Estimates of the therapeutic effects of venesection therapy are complicated because mortality is curvilinear across quartiles of number of venesections, as shown in Figure 3. Those receiving fewer than 8 or more than 64 venesections died at approximately 6 times the rate of those receiving intermediate numbers of venesections. As
L--, 56 ------------, 48 _____________________---______-------------. 32
II 60
-
40
-
20
-
0' 0
6
I
I
I
5
10
15
Follow
up
interval
(years)
Figure 2. Cumulative survival (94~)in 27 cirrhotic and 58 noncirrhotic patients. Numbers on the graph refer to the number of patients in each group at risk for each follow-up interval after deaths and censoring. Survival was significantly reduced in cirrhotic patients as compared with noncirrhotic patients (P < 0.004;log-rank test).
SURVIVAL ANALYSIS IN HEMOCHROMATOSIS
August 1991
n
=
20
n =
21
-
8
9
25
n =
19
n =
371
20
45 40 35 Figure 3. Percent of patients dying by quartile of number of venesections in 80 hemochromatosis patients. None of the patients who died in the first quartile completed venesection therapy for iron depletion before death. There were 8 deaths in the first quartile (6 cirrhotic, 2 noncirrhotic), I death in the second quartile (cirrhotic), 2 deaths in the third quartile (1 cirrhotic, 1 noncirrhotic) and 6 deaths in the fourth quartile (5 cirrhotic, 1 noncirrhotic).
30 25 20 15 10 5 0 5 0
-
Number
might be expected, the average follow-up time increases from the lowest to the highest quartile (4, 7, 7, and 13 years, respectively), suggesting that the increased mortality in the group receiving the fewest treatments was caused by advanced disease at the time of diagnosis. The average follow-up time among the four patients with cirrhosis in the lowest quartile of venesections was 1 year (all died): in the fourth quartile, the average follow-up time among cirrhotic patients was 11 years, during which time 5 of 12 died. Discussion
This retrospective study followed the clinical course of 85 patients with hemochromatosis for up to 31 years, with an average follow-up period of 8.1 years. The data suggest that patients whose conditions are diagnosed and treated before the development of cirrhosis have a survival time similar to that of the normal population (Figure 1). Cirrhosis is the major predictive clinical factor affecting survival, with cirrhotic patients faring considerably worse than noncirrhotic patients despite therapy. The apparent improved survival time in patients with arthritis compared with those without arthritis is not readily explainable, although our patients who died of liver failure within a year of diagnosis may not have survived long enough to develop arthritis, which has a high prevalence in the aging general population. Another possibility is that the presence of arthritis led to earlier diagnosis and treatment of the disease before the development of cirrhosis. With the advent of insulin therapy and the widespread use of venesection therapy, the cause of death in hemochromatosis has changed from complications of diabetes and heart
of
26
-
63
64
+
venesections
failure to liver failure and hepatocellular carcinoma. The cause of death in our patients was clearly related to hemochromatosis in only 8 of 17 cases, with liver disease and/or hepatocellular carcinoma accounting for 7 of these deaths. The beneficial effect of iron depletion by venesection therapy has never been subjected to controlled studies, but several studies have suggested a beneficial effect of therapy, particularly in the precirrhotic stage of the disease (2-5). In the absence of a control group of similar patients who did not receive therapy, the effect of venesection therapy on long-term survival is difficult to evaluate. Patients in our study who had advanced disease at the time of diagnosis died in the hospital before initiation or completion of venesection therapy. Patients with severe iron loading, as assessed by the number of venesections required to deplete the iron stores, also had an increased mortality rate: this finding is consistent with the observations of Niederau et al. (2), who observed an increased mortality in patients who required more than 18 months of venesection therapy. In 1935, Sheldon reported that patients with hemochromatosis had a life expectancy of l-5 years from the time of diagnosis (13). Following the institution of insulin therapy, Finch and Finch reported that the principal causes of death in patients with hemochromatosis were hepatic failure, portal hypertension, cardiac failure, and infection, with an average survival of 4.4 years after signs of clinical disease became manifest (14). The cumulative survival of treated hemochromatosis patients in previous studies is shown in Table 3. The improved survival rates in the present study and in the study by Niederau et al. (2) may reflect a higher percentage of discovered and likely precirrhotic patients because of
372
ADAMS ET AL.
GASTROENTEROLOGY Vol. 101, No. 2
Table 3. Cumulative Survival Reported in Treated Hemochromatosis Patients
5. Milder MS, Cook JD, Stray S, Finch CA. Idiopathic
Survival (%) Author
Year
Powell (3) 1970 Bomford and Williams (4) 1976 Milder et al. (5) 1980 Niederau et al. (2) 1985 Adams et al. 1991
6.
n
5yr
10yr
15yr
20yr
51 85 34
78.2 66 70
32 70 76 81
30 59 71
30 49 71
163 85
92 87
7.
8.
9.
more widespread use of HLA typing and family screening. In conclusion, this study shows that noncirrhotic patients treated for hemochromatosis have an excellent long-term prognosis, with survival rates comparable to those of the general population. The increased mortality associated with the presence of cirrhosis emphasizes the need for diagnosis in the precirrhotic stage of the disease.
10.
11.
12.
13. 14.
References
hemochromatosis, an interim report. Medicine [Baltimore) 1980;59:3449. Adams PC. Hepatic iron in hemochromatosis. Dig Dis Sci 1990;35:690-692. Adams PC, Frei JV, Bradley C, Lam D. Hepatic iron and iron absorption in hemochromatosis. Clin Invest Med 1990;13:256259. Borwein ST, Ghent CN, Flanagan PR, Chamberlain MJ, Valberg LS. Genetic and phenotypic expression of hemochromatosis in Canadians. Clin Invest Med 1983;6:171-179. Adams PC, Kertesz AE, Valberg LS. Clinical presentation of hemochromatosis: a changing scene. Am J Med 1991;90:445449. Benedetti J, Yuen K, Young L. PlL-Life tables and survival functions. In: Dixon WJ, ed. BMDP Statistical Software. Berkeley, CA: University of California Press, 1983:557-575. Hopkins A. PZL-survival analysis with covariates--Cox models. In: Dixon WJ, ed. BMDP Statistical Software. Berkeley, CA: University of California Press, 1983:576-594. Statistics Canada. Life tables, Canada and provinces 19801982. Catalogue 84-532. Ottawa: Ministry of Supply and Services, 1984. Sheldon JH. Haemochromatosis. New York: Oxford University, 1935. Finch SC, Finch CA. Idiopathic hemochromatosis, an iron storage disease. A. Iron metabolism in hemochromatosis. Medicine (Baltimore) 1955;34:381-430.
Adams PC, Halliday JW, Powell LW. Early diagnosis and treatment of hemochromatosis. Adv Intern Med 1989;34:111126.
Niederau C, Fischer R, Sonnenberg A, Stremmel W, Trampisch HJ, Strohmeyer G. Survival and causes of death in cirrhotic and in noncirrhotic patients with primary hemochromatosis. N Engl J Med 1985;313:1256-1262. Powell LW. Changing concepts in haemochromatosis. Postgrad Med J 1970;46:200-209. Bomford A, Williams R. Long term results of venesection therapy in idiopathic hemochromatosis. Q J Med 1976;45:611623.
Received July 30,199O. Accepted December 15,199O. Address requests for reprints to: Paul C. Adams, M.D., Department of Medicine, University Hospital, P.O. Box 5339, London, Ontario, Canada N6A 5A5. Dr. Adams acknowledges the grant support of the Medical Research Council of Canada and is a Career Scientist of the Ministry of Health of Ontario. The authors acknowledge the advice and assistance of Dr. L. S. Valberg.
