Several recent studies (1-6) support a high prevalence

Transcription

1 Prevalence of Hereditary Hemochromatosis in Primary Care Patients Pradyumna D. Phatak, MD; Ronald L. Sham, MD; Richard F. Raubertas, PhD; Karin Dunnigan, MD; Mary Theresa O'Leary, RN, MS; Caroline Braggins, MBA; and Joseph D. Cappuccio, MD Background: Despite evidence from screening studies in northern European populations, the prevalence of hemochromatosis in primary care populations in the United States remains speculative. Objective: To establish the feasibility of screening for hemochromatosis and to estimate the prevalence of hemochromatosis in a large primary care population. Design: Cross-sectional prevalence study. Setting: 22 primary care practices in the Rochester, New York, area. Patients: ambulatory patients without a previous diagnosis of hemochromatosis. Intervention: Serum transferrin saturation screening tests were offered to all adult patients in participating primary care practices. Measurements: Patients with a serum transferrin saturation of 45% or more on initial testing had a serum transferrin saturation test done under fasting conditions and had serum ferritin levels measured. Those who had a fasting serum transferrin saturation of 55% or more and a serum ferritin level of 200 /utg/l or more with no other apparent cause were presumed to have hemochromatosis and were offered liver biopsy to confirm the diagnosis. Results: 25 patients had biopsy-proven hemochromatosis; 22 patients met the clinical criteria for hemochromatosis but declined liver biopsy and were classified as having clinically proven hemochromatosis; and 23 patients had a serum transferrin saturation of 55% or more with no identifiable cause, indicating probable hemochromatosis. The prevalence of clinically proven and biopsy-proven hemochromatosis combined was 4.5 per 1000 (95% CI, 3.3 to 5.8 per 1000) in the total sample and 5.4 per 1000 (CI, 4.0 to 7.1 per 1000) in white persons. The prevalence was higher in men than in women (ratio, 1.8:1). Conclusions: Hemochromatosis is relatively common among white persons. Routine screening of white persons for hemochromatosis should be considered by primary care physicians. Several recent studies (1-6) support a high prevalence of hereditary hemochromatosis (2 to 5 per 1000) in the U.S. population. Nationally representative data are available on the proportion of persons in the United States who have elevated serum iron measures (serum ferritin level and serum transferrin saturation) on initial testing (7), but the proportion of persons with hemochromatosis can be determined only through further diagnostic follow-up, such as that provided by screening studies. A meta-analysis of screening studies done throughout the world and a recent study done in a health maintenance organization (5, 6) suggest that hemochromatosis is relatively common among white and Hispanic persons. A gene for hemochromatosis was discovered in 1996 (8), but many questions remain unanswered about the accuracy of genetic testing for hemochromatosis. Experts currently do not advocate widespread population screening done by using genetic methods (9). Serum transferrin saturation is thought to be a sensitive marker for hemochromatosis (1). The threshold transferrin saturation value used for screening has ranged from 45% to 70% (4, 10-18). Some authorities suggest that a persistently elevated serum transferrin saturation with no other explanation is enough to predict homozygosity for the hemochromatosis allele. It has been suggested that the threshold value for serum transferrin saturation screening should be lower for women than for men because a threshold value of 62% failed to detect 40% of suspected female homozygotes in one study (1). We report our experience with serum transferrin saturation screening for hemochromatosis in primary care patients. Our goal was to estimate the prevalence of hemochromatosis and to establish the feasibility of screening for hemochromatosis in a large primary care population. Ann Intern Med. 1998;129: Methods From the Mary M. Goolcy Hemophilia Center, Inc., Rochester General Hospital, and the University of Rochester School of Medicine and Dentistry, Rochester, New York. For current author addresses, see end of text American College of Physicians-American Society of Internal Medicine We enrolled 22 primary care practices from the Rochester, New York, area. These urban and suburban practices included 63 physicians, nurse prac-

2 titioners, and physician assistants and ranged in type from solo practices to large clinics. Practice size varied from 1500 to patients. Participating physicians were asked to offer screening to all adult patients (>18 years of age) seen during their enrollment period. Patients with previously diagnosed hemochromatosis were excluded. Patient enrollment rates varied from 19% to 94%. After each participant had given signed informed consent, about 10 ml of venous blood was obtained. Serum was separated, and serum transferrin saturations were analyzed with a Hitachi 747 analyzer (Hitachi, Tokyo, Japan). A standard Boehringer-Mannheim system (Boehringer-Mannheim, Indianapolis, Indiana) was used with FerroZine (Boehringer-Mannheim) as the reactive chromogen read at 570 nm. Serum ferritin levels were measured on the Ciba-Corning Automated Chemiluminescence System (Ciba-Corning, Medfield, Massachusetts) by using a two-site chemiluminometric sandwich immunoassay with acridinium ester as the luminescent tag. The subsequent screening protocol is outlined in Figure 1. Patients with a screening serum transferrin saturation of 45% or more were recalled so that a second transferrin saturation test could be given under fasting conditions and the serum ferritin level could be measured. All patients with a fasting serum transferrin saturation of 45% or more were evaluated by the investigators. Patients with a fasting serum transferrin saturation of 55% or more and a serum ferritin level of 200 /ig/l or more who had no apparent secondary cause of abnormal iron Figure 1. Screening protocol. In practice, liver biopsy was recommended for some patients who did not meet the strict study criteria for liver biopsy (see Figure 3). 1 December 1998 Annals of Internal Medicine Volume 129 Number

3 Figure 2. Screened patients. The number of patients who had positive results on an initial screening test for hemochromatosis and the results of subsequent testing are shown. Some patients dropped out at each stage. TS = transferrin saturation. steps of the protocol (measurement of fasting serum transferrin saturation and evaluation by investigators [Figure 2]), some participants dropped out; these participants provided incomplete information about their hemochromatosis status and represent "right-censored" observations, in the terminology of life-table analysis. Therefore, the Kaplan-Meier method (19) was used to estimate the prevalence of hemochromatosis and CIs were determined by using the likelihood ratio method (20). In the analysis, we assumed that at these two screening steps, the likelihood that a participant dropped out was unrelated to his or her true hemochromatosis status. The demographic characteristics of study patients were compared by using chi-square tests. status, such as iron-loading anemias or other causes of chronic liver disease, were offered liver biopsy with quantitative iron estimation to confirm the diagnosis of hereditary hemochromatosis. First-degree relatives of these patients were offered screening for hemochromatosis. Patients with suspected hemochromatosis who declined liver biopsy were offered therapeutic phlebotomy. Mobilizable iron stores were determined on the basis of the blood volume removed before iron depletion (serum ferritin level < 25 /Jtg/L) was achieved. The following formula was used: g of iron = (hematocrit/3) X weight of blood X The diagnosis of hemochromatosis was based on liver biopsy findings when possible. Patients with quantitative hepatic iron concentrations of 50 jutmol per g or more or a hepatic iron index score (quantitative hepatic iron concentration in /xmol per g dry weight divided by age) of 1.9 or more were classified as having biopsy-proven hemochromatosis. Among patients in whom biopsy could not be done, those with a serum transferrin saturation of 55% or more on initial testing and on repeated testing done under fasting conditions, a serum ferritin level of 200 jltg/l or more, and no identifiable secondary cause of iron overload were classified as having clinically proven hemochromatosis. These patients were advised to undergo therapeutic phlebotomy and to have their first-degree relatives screened for hemochromatosis. Patients with initial and fasting serum transferrin saturations of 55% or more and serum ferritin levels less than 200 jutg/l were classified as having probable hemochromatosis. They were advised to have serial follow-up of iron studies and to have their first-degree relatives screened for hemochromatosis. The prevalence of hemochromatosis was defined as the probability that a primary care patient screened according to the above protocol would receive a diagnosis of hemochromatosis. At two Results Demographic Characteristics of the Study Sample A total of patients was enrolled in the study over a 26-month period. The demographic characteristics of these patients are shown in Table 1. The proportion of men in the sample was greater among the white patients than among the nonwhite patients (44% compared with 37% to 39%; P< 0.001), and the white patients were older than the nonwhite patients (median age, 54 years compared with 43 to 45 years; P < 0.001). Of the enrolled Table 1. Demographic Characteristics of Enrolled and Screened Patients Characteristic Enrolled Patients Patients Who Completed Initial Screening Test White patients, n (%) (77) (78) Male, % Age, % years years years >70 years African-American patients, n (%) 2682(14) 2268(14) Male, % Age, % years years years >70 years 8 8 Patients of other or unknown ethnicity, n (%) 1628(9) 1308(8) Male, % Age, % years years years >70 years 9 8 Total sample, n (%) (100) (100) Male, % Age, % years years years >70 years December 1998 Annals of Internal Medicine Volume 129 Number 11

4 Figure 3. Evaluated patients. The outcome, based on transferrin saturation and serum ferritin level, of the 255 evaluated patients is shown. Three patients who met the study criteria for liver biopsy had biopsy deferred because of comorbid conditions and received a diagnosis of clinically proven hemochromatosis. Of 21 patients who met the study criteria for liver biopsy and underwent biopsy, 18 had biopsy-proven hemochromatosis and 1 had clinically proven hemochromatosis (see text for details). Thirteen patients with a transferrin saturation of 45% to 55% had liver biopsy because of unexplained elevations in the serum ferritin level. Seven of the 13 had biopsy-proven hemochromatosis. patients, 2739 (15%) did not complete the screening protocol; 2499 of these dropouts occurred before the initial serum transferrin saturation test was done. Dropout rates were significantly higher for men than for women (17% compared with 15%; P < 0.001), for nonwhite patients than for white patients (17% for African-American patients, 21% for patients of other or unknown ethnicity, and 15% for white patients; P < 0.001), and for younger patients than for older patients (21% for patients 18 to 39 years of age, 16% for patients 40 to 54 years of age, 11% for patients 55 to 69 years of age, and 9% for patients > 70 years of age; P < 0.001). However, because of the large sample sizes, these statistically significant differences may not have had clinical significance. The dropouts did not substantially change the demographic characteristics of the study sample (Table 1). Evaluation of Patients with Serum Transferrin Saturations of 45% or More A total of 932 patients had an initial serum transferrin saturation of 45% or more (Figure 2). Of 311 patients who subsequently had a fasting serum transferrin saturation of 45% or more, we could evaluate 255. The disposition of these patients is shown in Figure 3. Eighty-two patients persistently had a serum transferrin saturation of 55% or more; of these, 50 had a serum ferritin level of 200 /xg/l or more and 32 had a serum ferritin level less than 200 /xg/l. Of the 50 patients whose serum ferritin level was 200 jutg/l or more, 35 had no secondary explanation for their iron status and no contraindications for liver biopsy. Thus, they met our protocol criteria for liver biopsy. Of these 35 patients, 14 declined biopsy and 21 underwent biopsy. Of the 21 who had biopsy, 18 had biopsy-proven hemochromatosis (defined as a hepatic iron index score > 1.9 or a quantitative hepatic iron concentration > 50 Ltmol per g dry weight). All biopsy specimens that met these criteria had a grade of at least 2+ on Prussian blue staining. One patient's specimen had a grade of 2+ but was inadequate for quantitative iron determination. This patient also had a sister with biopsy-proven hemochromatosis and was classified as having clinically proven hemochromatosis. Of the remaining 2 patients who underwent biopsy, 1 had chronic hepatitis C and the other had druginduced hepatitis. An additional 13 patients had liver biopsy at the clinical discretion of the investigators but were not eligible for biopsy according to the protocol because one or both of their serum transferrin saturations were less than 55% (Figure 3). Seven of the 13 had biopsy-proven hemochromatosis (of the other 6, 3 had steatohepatitis, 1 had chronic hepatitis C, 1 had drug-induced hepatitis, 1 December 1998 Annals of Internal Medicine Volume 129 Number

5 and 1 had nonspecific hepatitis). Thus, a total of 25 study patients had biopsy-proven hemochromatosis (Table 2). Seventeen patients met our study criteria for liver biopsy and had no identifiable secondary cause of iron overload but did not have biopsy, either because they refused it (n = 14) or because the investigators did not recommend it because of comorbidity (n = 3). These 17 patients, as well as the 1 patient with an inadequate liver biopsy specimen, were classified as having clinically proven hemochromatosis and were advised to have empirical therapeutic phlebotomy. Four other patients were judged by the investigators to have clinically proven hemochromatosis even though one or both of their serum transferrin saturation values were less than 55%. Thus, 22 study patients had clinically proven hemochromatosis; 7 of the 22 have achieved iron depletion (Table 3). Twenty-three patients repeatedly had serum transferrin saturations of 55% or more and serum ferritin levels less than 200 jxg/l with no secondary explanation. They did not qualify for liver biopsy and were classified as having probable hemochromatosis. They were advised to have their blood iron status checked every 2 years and will be candidates for genetic testing in the future. They were also advised to recommend screening to their firstdegree relatives, as were patients with biopsy-proven and clinically proven hemochromatosis. (Six relatives with proven hemochromatosis were detected in this manner. These additional cases of hemochromatosis are not included in our calculations of prevalence because the patients were not part of the screened sample.) Prevalence of Hemochromatosis We calculated three estimates of prevalence by varying the strictness of our criteria for deciding which patients had hemochromatosis. Biopsy-Proven Hemochromatosis The prevalence of biopsy-proven hemochromatosis was calculated by 1) using only cases of hemochromatosis identified by biopsy and 2) assuming that no clinically proven or probable cases of hemochromatosis actually were cases of hemochromatosis. This method of determining prevalence is likely to result in an underestimate of the true prevalence. Biopsy-Proven and Clinically Proven Hemochromatosis The prevalence of biopsy-proven and clinically proven hemochromatosis combined was calculated by assuming that 1) all cases of clinically proven hemochromatosis actually were cases of hemochromatosis and 2) no probable cases of hemochromatosis actually were cases of hemochromatosis. The first assumption is not consistent with our observation that only 25 of 33 patients who had biopsy had hemochromatosis, but the resulting overestimate may be offset by the second assumption. This is our preferred estimate. Table 2. Demographic and Clinical Characteristics of Patients with Biopsy-Proven Hemochromatosis Sex Ethnicity Age Fasting Serum Serum Cirrhosis Hepatic Iron Hepatic Iron Mobilizable Transferrin Saturation Ferritin Level Concentration Index Score Iron y % lig/l IxmollL per g dry weight 9 Female White No Female White Yes Male White Yes Not treated Male White No Male White No Female White No * Male White No Male White No Not treated Male White No Female White No Male White No Female White No * Male White Yes Male White No Not treated Male White No Male White No Female White No Female White No * Male White No Male White No * Female White No Male White No Female White No Male White No * Male White No * Patient has not yet achieved iron depletion December 1998 Annals of Internal Medicine Volume 129 Number 11

6 Table 3. Demographic and Clinical Characteristics of Patients with Clinically Proven Hemochromatosis Sex Ethnicity Age Fasting Fasting Mobilizable Serum Serum Iron Transferrin Ferritin Saturation Level y % ixgli 9 Female White Male White Not treated Male White Female White Not treated Male White Not treated Female White Not treated Male White Female White Not treated Male White Not treated Male White Male White Female White Not treated Female White Not treated Female White Not treated Female White Male White * Male White Male African-American Not treated Male White Female White Male White * Male White Not treated * Patient has not yet achieved iron depletion. Biopsy-Proven, Clinically Proven, and Probable Hemochromatosis The prevalence of biopsy-proven, clinically proven, and probable hemochromatosis combined was calculated by assuming that all clinically proven and probable cases of hemochromatosis actually were cases of hemochromatosis. The result is likely to be an overestimate of true prevalence. The three estimates are shown in Table 4. The overall prevalence of biopsy-proven and clinically proven hemochromatosis combined was 4.5 per 1000 (95% CI, 3.3 to 5.8 per 1000). The prevalence in white patients was significantly higher than that in African-American patients (5.4 compared with 0.9; P < 0.05), and the prevalence in white patients was significantly higher among men than among women (7.4 compared with 4.0; P < 0.05). Prevalence rates among practices with high enrollment rates did not differ significantly from those among practices with low enrollment rates. Discussion Our study shows that screening with serum transferrin saturation testing in a large primary care population will identify cases of hemochromatosis at a frequency matching that suggested by previous prevalence estimates. Possible deficiencies in our study design include reliance on the sensitivity of transferrin saturation testing and substantial patient dropout at every stage of the protocol. In addition, variation in recruitment rates at the study sites may have introduced bias in our prevalence estimates, although we did not find substantial differences between prevalences in practices with low enrollment rates and practices with high enrollment rates. Despite these weaknesses, we can draw useful conclusions about the prevalence of hemochromatosis in primary care settings and about the practical outcome of our particular screening strategy. The high prevalence of hemochromatosis in asymptomatic persons has been described previously. Some large prevalence studies have screened blood donors (5) or inpatients (12). Blood donors tend to be young and healthy, and some may have falsely low iron stores as a result of regular donation. Persons with elevated liver enzyme concentrations (a known early manifestation of hemochromatosis) are excluded from the blood donor pool. Hospital inpatients are also, by nature, a select group. Neither blood donors nor inpatients are therefore truly representative of the community at large or of a typical primary care practice. Other studies have screened a less selective population, but all of these studies were relatively small and had little power to detect differences between subgroups. Of note, a study that screened only older men (16) found a prevalence of almost 10 per 1000, which may reflect the fact that many persons with hemochromatosis accumulate significant iron stores later in life. Our large screening study done in primary care practices confirms the suspected prevalence of previously undiagnosed hemochromatosis in this setting. Although several previous studies have suggested the same high prevalence of hemochromatosis, screening for hemochromatosis is not currently recommended as a routine practice in primary care. We previously constructed a decision model that shows the cost-effectiveness of screening for hemochromatosis (21). Our current study validates the prevalence assumptions of our model and strengthens the conclusion that screening should be recommended. Our screening strategy differs somewhat from Table 4. Group Prevalence of Hemochromatosis Prevalence per 1000 (95% CI) Biopsy-Proven Biopsy-Proven Proven and Hemochromatosis and Clinically Probable Proven Hemochromatosis Hemochromatosis Total sample 2.4( ) 4.5( ) 6.6( ) White patients 3.0( ) 5.4( ) 8.1 ( ) Women 2.0( ) 4.0( ) 6.6( ) Men 4.2( ) 7.4( ) 10.3( ) African-American patients 0.0 ( ) 0.9 ( ) 0.9 ( ) Patients of other or unknown ethnicity 0.0( ) 0.0( ) 0.0( ) 1 December 1998 Annals of Internal Medicine Volume 129 Number

7 those of previous screening studies. Because the serum transferrin saturation test is thought to be the most sensitive screening test for hemochromatosis, we used it as our initial test (5, 22, 23). In previous studies, the cut-point for serum transferrin saturation on initial testing has varied from 45% to 70%. Some authors have contended that a persistently elevated serum transferrin saturation (>62%) in the absence of identifiable secondary causes is sufficient to diagnose hemochromatosis (1). However, body iron burden estimated by using liver biopsy with quantitative iron measurement remains the accepted principal criterion for diagnosis (24). We used conservative criteria to determine eligibility for liver biopsy in order to minimize the performance of unnecessary invasive tests. Our strategy was effective: Twenty-five of 33 patients who underwent liver biopsy had proven hemochromatosis. The other 8 had other causes of chronic liver disease, and none had a normal liver biopsy result. When only patients who met the strict criteria for liver biopsy (serum transferrin saturation > 55% on more than one test and serum ferritin level > 200 /xg/l with no identifiable secondary cause) were considered, 18 of 20 patients with adequate liver biopsy specimens had proven hemochromatosis. Lowering the threshold value for screening will increase the false-positive rate, but we show that some biopsy-proven cases of hemochromatosis would be missed if strict screening criteria (initial and fasting serum transferrin saturation > 55%) were used. The amount of iron required on liver biopsy to confirm the diagnosis of hemochromatosis is controversial. The hepatic iron index was found to effectively discriminate between persons who are homozygous for the hemochromatosis allele and those who are heterozygous (24). The most stringent diagnostic criteria require a hepatic iron index score of 1.9 or more to confirm the diagnosis of hemochromatosis. In our experience, this index is less useful in older persons because much higher quantitative iron stores are required to achieve a score greater than 1.9. We have found several persons with quantitative hepatic iron concentrations of 50 fimol per g dry weight or more, no secondary cause for iron loading, and high body iron stores (as evidenced by mobilizable iron stores) who have hepatic iron index scores less than 1.9. Many of these persons also have family members with high body iron stores; this supports our conclusion that these persons have hereditary hemochromatosis. Our criteria for biopsy-proven hemochromatosis therefore include a hepatic iron index score of 1.9 or more or a hepatic iron concentration of at least 50 jutmol per g dry weight in the absence of an identifiable secondary cause of iron overload December 1998 Annals of Internal Medicine Volume 129 Number 11 With the recent discovery of the HFE gene (8, 25), a genetic test for the C282Y and H63D mutations is now available. The place of this test in the screening protocol remains to be defined. It is likely that the test will abrogate the need for liver biopsy in at least a subset of patients, making routine screening even more effective. The prevalence of hemochromatosis in our study sample was higher among men than among women (ratio, approximately 1.8:1), a finding analogous to those of other large screening studies (15). With an autosomal recessive disease, the ratio of affected men to affected women should be 1:1. This discrepancy is important in light of the similar prevalence of initially elevated serum transferrin saturations at lower thresholds found among white women compared with white men in the United States (7) and is probably explained by increased disease expression in affected men. Women are at least partly protected from the development of iron overload by menstrual blood loss and pregnancy-related demands for iron. Presumably, a direct test for the defective gene would show an equal prevalence in men and women. Our screening strategy requires the presence of significant accumulated iron stores before a diagnosis of hemochromatosis is considered proven. The variable penetrance of the hemochromatosis allele is well illustrated by the difference in prevalence in men and women. This difference implies that about half of female homozygotes are missed by a screening strategy that requires accumulation of body iron stores for the satisfaction of diagnostic criteria. It remains possible that the serum transferrin saturation cut-point of 55% that we used for determining eligibility for liver biopsy is not low enough, particularly for women. In our practice, we have seen persons with hemochromatosis confirmed on liver biopsy who have serum transferrin saturations consistently less than 45%. These persons would be missed by our screening strategy. Thus, our study probably underestimates the true prevalence of homozygosity for the hemochromatosis allele. This can be directly addressed only by largescale screening directed at the defective gene. Previous screening studies have been performed primarily in white populations (6). Although nationally representative data have not shown a substantially lower prevalence of elevated initial serum transferrin saturations on screening in black persons (7), the true prevalence of hemochromatosis cannot be estimated on the basis of these data. Our study included a substantial number of persons of African descent. Only one case of clinically proven hemochromatosis was found in this group, and the prevalence of hemochromatosis was significantly lower in this group than in white patients. The occurrence

8 of iron overload in African-American persons is well documented, and a non-hla-linked genetic mechanism for this has been proposed (26). Our study does not exclude the occurrence of such a syndrome because the 95% CI for prevalence in this group still ranges from 0 to 3.7 per Moreover, it has been suggested that serum transferrin saturation tends to be lower in African-American persons with iron overload than in white persons with iron overload (27); our screening strategy may therefore have missed these persons. No cases of hemochromatosis were found among study participants who belonged to other ethnic groups, but the number of such participants was relatively small. The number of homozygotes for the hemochromatosis allele who will develop clinically relevant iron overload in their lifetime remains unclear. The risk is clearly lower in women. We identified several older patients with hemochromatosis who did not yet have evidence of cirrhosis on liver biopsy. Arguably, these patients may not have benefited from diagnosis. However, the degree of iron overload present in many of these patients strongly suggests that irreversible damage would have occurred if the disease had gone unrecognized and tissue iron stores had continued to accumulate. The only way to definitively address the rate of development of clinically significant iron overload would be to follow a cohort of homozygous persons without treatment. However, the standard treatment for hemochromatosis therapeutic phlebotomy is so simple and safe that withholding it from persons known to be affected would be unethical. Thus, such a study will never be performed. A diagnosis of hemochromatosis is routinely followed by the screening of affected first-degree relatives. Our study led to the diagnosis of hemochromatosis in six first-degree relatives of study patients, an outcome that further increases the value of screening. In conclusion, our study confirms a high prevalence of undiagnosed hemochromatosis in a primary care population. Our findings add to the growing evidence that supports routine screening with serum transferrin saturation testing for hemochromatosis in white persons. Acknowledgments: The authors thank the primary care physicians at Rochester General Hospital, Rochester, New York, without whose participation this study could not have been successfully conducted. Grant Support: By grant ROl HS07616 from the Agency for Health Care Policy and Research. Requests for Reprints: Pradyumna D. Phatak, MD, Hematology Unit, Rochester General Hospital, 1425 Portland Avenue, Rochester, NY Current Author Addresses: Drs. Phatak, Sham, Dunnigan, and Cappuccio and Ms. O'Leary and Ms. Braggins: Rochester General Hospital, 1425 Portland Avenue, Rochester, NY Dr. Raubertas: Department of Biostatistics, University of Rochester, 601 Elmwood Avenue, Rochester, NY References 1. Edwards CQ, Kushner JP. Screening for hemochromatosis. N Engl J Med. 1993;328: Baer DM, Simons JL, Staples RL, Rumore GJ, Morton CI. Hemochromatosis screening in asymptomatic ambulatory men 30 years of age and older. Am J Med. 1995;98: Balan V, Baldus W, Fairbanks V, Michels V, Burritt M, Klee G. Screening for hemochromatosis: a cost-effectiveness study based on 12,258 patients. Gastroenterology. 1994;107: Edwards CQ, Griffin LM, Goldgar D, Drummond C, Skolnick MH, Kushner JP. Prevalence of hemochromatosis among 11,065 presumably healthy blood donors. N Engl J Med. 1988;318: Iron overload disorders among Hispanics San Diego, California, MMWR Morb Mortal Wkly Rep. 1996;45: Bradley LA, Haddow JE, Palomaki GE. Population screening for haemochromatosis: a unifying analysis of published intervention trials. J Med Screen. 1996;3: Looker AC, Johnson CL. Prevalence of elevated serum transferrin saturation in adults in the United States. Ann Intern Med. 1998;129: Feder JN, Gnirke A, Thomas W, Tsuchihashi Z, Ruddy DA, Basava A, et al. A novel MHC class l-like gene is mutated in patients with hereditary haemochromatosis. Nat Genet. 1996;13: Burke W, Thomson E, Khoury MJ, McDonnell SM, Press N, Adams PC, et al. Hereditary hemochromatosis: gene discovery and its implications for population-based screening. JAMA. 1998;280: Borwein ST, Ghent CN, Flanagan PR, Chamberlain MJ, Valberg LS. Genetic and phenotypic expression of hemochromatosis in Canadians. Clin Invest Med. 1983;6: Olsson KS, Ritter B, Rosen U, Heedman PA, Staugard F. Prevalence of iron overload in Central Sweden. Acta Med Scand. 1983;213: Hallberg L, Bjorn-Rasmussen E, Jungner I. Prevalence of hereditary hemochromatosis in two Swedish urban areas. J Intern Med. 1989;225: Elliott R, Lin BP, Dent OF, Tait A, Smith CI. Prevalence of hemochromatosis in a random sample of asymptomatic men. Aust N Z J Med. 1986;16: Tanner AR, Desai S, Lu W, Wright R. Screening for hemochromatosis in the UK: preliminary results [Abstract]. Gut. 1985;26:A Summary of a report on assessment of the iron nutritional status of the United States population. Expert Scientific Working Group. Am J Clin Nutr. 1985;42: Meyer T, Baynes R, Bothwell T, Jenkins T, Jooste P, Du Toit E, et al. Phenotypic expression of the HLA linked iron-loading gene in males over the age of 40 years: a population study using serial serum ferritin estimations. J Intern Med. 1990;227: Leggett BA, Halliday JW, Brown NN, Bryant S, Powell LW. Prevalence of haemochromatosis amongst asymptomatic Australians. Br J Haematol. 1990;74: Lindmark B, Eriksson S. Prevalence of hemochromatosis [Letter]. N Engl J Med. 1988;319: Cox DR, Oakes D. Analysis of Survival Data. New York: Chapman and Hall; Thomas DR, Grunkemeier GL. Confidence interval estimation of survival probabilities for censored data. Journal of the American Statistical Association. 1975;70: Phatak PD, Guzman G, Woll JE, Robeson A, Phelps CE. Cost-effectiveness of screening for hereditary hemochromatosis. Arch Intern Med. 1994; 154: Bassett ML, Halliday JW, Ferris RA, Powell LW. Diagnosis of hemochromatosis in young subjects: predictive accuracy of biochemical screening tests. Gastroenterology. 1984;87: Borwein S, Ghent CN, Valberg LS. Diagnostic efficacy of screening tests for hereditary hemochromatosis. Can Med Assoc J. 1984;131: Bassett ML, Halliday JW, Powell LW. Value of hepatic iron measurements in early hemochromatosis and determination of the critical iron level associated with fibrosis. Hepatology. 1986;6: Beutler E, Gelbart T, West C, Lee P, Adams M, Blackstone R, et al. Mutation analysis in hereditary hemochromatosis. Blood Cells Mol Dis. 1996; 22: Gordeuk V, Mukiibi J, Hasstedt SJ, Samowitz W, Edwards CQ. West G, et al. Iron overload in Africa. Interaction between a gene and dietary iron content. N Engl J Med. 1992;326: Barton JC, Edwards CQ, Bertoli LF, Shroyer TW, Hudson SL. Iron overload in African Americans. Am J Med. 1995;99: December 1998 Annals of Internal Medicine Volume 129 Number