Adele L. Franks

Hospitalizations during pregnancy among managed care enrollees.

OBJECTIVE To describe the prevalence of hospitalizations during pregnancy, the reason for hospitalization, the length of stay, and the associated costs. METHODS We analyzed data from a national managed care organization and determined the occurrence of hospitalizations for 46,179 women who had a live birth or a pregnancy loss in 1997. RESULTS Overall, 8.7% of women were hospitalized during their pregnancy. Of these, 5.7% were hospitalized and discharged while pregnant, 0.8% experienced extended stays before a live birth or pregnancy loss, and 2.1% experienced pregnancy loss. Hospitalizations were more common among younger women, women with multiple gestations, and women in the northeastern United States. Women who had a live birth were primarily hospitalized for preterm labor (24%), hyperemesis (9%), hypertension (9%), kidney disorders (6%), and prolonged premature rupture of membranes (6%). Charges totaled over

Iron Overload, Public Health, and Genetics: Evaluating the Evidence for Hemochromatosis Screening

36 million. CONCLUSION Antenatal hospitalizations are common.

Contraception and ectopic pregnancy risk.

As many as 1 million persons in the United States are affected by hemochromatosis, a genetic condition characterized by excess iron absorption and pathologic iron deposition in tissue [1]. If undetected and untreated, hemochromatosis can result in illness (such as cirrhosis, hepatoma, diabetes, cardiomyopathy, arthritis, arthropathy, and hypopituitarism with hypogonadism) and death. The identification and treatment of asymptomatic persons in whom iron measures are elevated but hemochromatosis is not clinically apparent have been recommended as a potentially cost-effective strategy for preventing hemochromatosis-associated illness and death [1-10]. Nonetheless, some experts argue that before universal screening can be recommended, the clinical expression and natural history of hemochromatosis must be clarified and the infrastructure necessary to support a universal screening program (including laboratory standardization and physician education) must be established [11]. The recent discovery of a gene [12, 13] associated with hemochromatosis has made it possible to use DNA testing along with, or instead of, iron measures in screening. Although this discovery has increased interest in hemochromatosis, it has also raised new questions about screening for and diagnosis of the disease. One objective of the meeting on Iron Overload, Public Health, and Genetics, sponsored by the Centers for Disease Control and Prevention and the National Institutes of Health in March 1997 [14], was to review the scientific information available on population screening for hemochromatosis. Our assessment of the evidence and recommendations for action are presented here. Methods for Evaluating the Evidence for Population Screening for Hemochromatosis We used U.S. Preventive Services Task Force criteria [15] to evaluate evidence related to population screening for hemochromatosis that was presented at the meeting on Iron Overload, Public Health and Genetics or was published before August 1997. In this paper, we examine six assumptions that have been used to support the case for population screening for hemochromatosis in the United States. 1. Prevalence: Hemochromatosis is relatively common. 2. Burden of suffering: Hemochromatosis is associated with significant morbidity and mortality. 3. Efficacy of treatment: Phlebotomy can prevent complications by reducing the iron burden in persons with hemochromatosis. 4. Accuracy of screening tests: The available screening tests detect hemochromatosis earlier than no screening and do so with sufficient accuracy. 5. Effectiveness of screening: Screening and early treatment of persons with hemochromatosis improves health compared with treatment of patients after the development of clinical signs and symptoms. 6. Safety and adverse effects: The potential benefits of screening and treatment outweigh the adverse effects. The studies of effectiveness that we reviewed were each assigned one of five grades of evidence on the basis of study design (Table 1). Table 1. U.S. Preventive Services Task Force Criteria for Judging Quality of Evidence* Prevalence Determination of the prevalence of hemochromatosis is complicated by variation in case definitions of the disease and by uncertainty about the progression from genetic susceptibility through iron overload to clinical disease. Case definitions of hemochromatosis may include one or more of the following: genetic mutations, abnormal iron measures, and clinical signs and symptoms [16]. It is difficult to reach agreement on a standard because each case definition has strengths and weaknesses. Clinical case definitions have two chief weaknesses. First, such definitions (for example, bronze diabetes and cirrhosis) often represent the end stages of disease and have limited usefulness for prevention. Second, many clinical signs and symptoms that occur early in the course of hemochromatosis are nonspecific (for example, fatigue, abdominal pain, joint pain, and elevated liver enzyme concentrations) and may be attributed to other causes. Case definitions more specific to hemochromatosis, such as persistently elevated serum transferrin saturation or liver iron deposition without cirrhosis, allow for the detection of hemochromatosis before clinical signs and symptoms occur, but the rate and degree of progression from abnormal iron measures to clinical symptoms are uncertain. Even more uncertainty exists about progression from genetic susceptibility to clinical disease. Although the autosomal recessive nature of hemochromatosis and its link to the HLA region on chromosome 6 have been recognized for 20 years [17], two mutations associated with hemochromatosis-C282Y and H63D-were found only in 1996 [12]. Several investigators have estimated the frequency of the mutations in small groups of persons without clinical evidence of hemochromatosis (45 to 381 persons) [12, 18-24]. In the largest study published to date [25], the prevalence of homozygosity for C282Y was 1 in 1000, the prevalence of heterozygosity for both C282Y and H63D (compound heterozygosity) was 16 in 1000, and the prevalence of homozygosity for H63D was 20 in 1000 among 1450 persons from northern Europe. None of the study participants from other regions (Africa, Asia, and Australia) carried two C282Y mutations. These studies [12, 19-25] were drawn from convenience samples, and none was designed to represent the general population. Therefore, these studies may overestimate or underestimate the prevalence of genetic susceptibility to hemochromatosis. On the basis of case definitions that use elevated iron measures (such as body iron stores) in screening studies, the prevalence of hemochromatosis is 2 to 5 per 1000 persons in white populations [11, 26-30]. The estimated prevalence of hemochromatosis in black populations is lower, less than 1 in 1000 [1, 26, 31, 32]. These estimates are higher than the estimated prevalence of homozygosity for the major mutation, C282Y, but they are lower than the prevalence of compound heterozygosity or homozygosity for H63D. This discrepancy suggests either the presence of as yet undiscovered mutations for hemochromatosis or reduced penetrance of compound heterozygosity and homozygosity for H63D. Burden of Suffering The prevalence of clinical disease due to hemochromatosis is uncertain. Hemochromatosis can lead to cirrhosis and other liver diseases, hepatocellular carcinoma, diabetes, cardiomyopathy, arthritis, hypopituitary hypogonadism, fatigue, joint pain, skin bronzing or graying, abdominal pain, impotence, amenorrhea, and cardiac arrhythmias [1]. The most common early symptom is weakness or fatigue [1]. Although diabetes and heart disease occur more often in cirrhotic patients, they are also seen in patients with hemochromatosis who do not have liver disease [33, 34]. The classic triad of liver disease (cirrhotic or noncirrhotic), diabetes, and skin bronzing occurs in a minority of patients (for example, 17% of patients in one case series [33] and 3% of patients identified through screening studies [11]). Deaths in persons with hemochromatosis are most often associated with liver disease, hepatocellular carcinoma, diabetes, or cardiomyopathy. Morbidity Estimates of morbidity have usually been derived from case series of patients with known hemochromatosis [33-35]. The proportion of patients with hemochromatosis who have associated illness is probably greater in case series than in the general population because a disproportionate number of patients in case series may be detected because of their symptoms (selection bias). A review of family-based screening studies was done to address this limitation [36]. In that review, 52% of 146 family members 15 to 72 years of age in whom hemochromatosis had been diagnosed by HLA haplotyping were asymptomatic. The other 48% had at least one clinical manifestation of disease, such as cirrhosis, other liver disease, diabetes, cardiomyopathy, arthropathy, skin bronzing, fatigue, weight loss, abdominal pain, or impotence. In addition, the risk for symptoms associated with hemochromatosis increased with age-73% of men and 44% of women older than 40 years of age had at least one clinical finding. Siblings and other family members identified through HLA testing may have a different risk for disease expression than do persons with hemochromatosis in the general population. Among persons with hemochromatosis (defined by elevated iron measures [16]) in population screening studies, 45% of men and 43% of women older than 40 years of age had at least one clinical finding [11]. Screening studies have not compared the prevalence of clinical findings in persons with hemochromatosis and persons without hemochromatosis. Many clinical findings associated with hemochromatosis (such as abdominal pain, fatigue, arthritis, and diabetes) are also common in persons without hemochromatosis. Therefore, some of the illness attributed to hemochromatosis may be due to other causes. The studies with the least biased methods to date [37-51] have estimated the proportion of persons with selected clinical conditions who have underlying hemochromatosis (Table 2). The prevalence of hemochromatosis ranges from 11.0% to 15.0% in patients with hepatocellular carcinoma to 0.0% to 1.5% in patients with diabetes. If the population prevalence is assumed to be 0.2% to 0.5% [11, 26, 27], these studies suggest that the risk for hemochromatosis may be elevated in persons with hepatocellular carcinoma, liver disease, hepatitis, cardiac arrhythmias, arthropathy, and diabetes. The risk for hemochromatosis may be further elevated in persons with combinations of these diseases (for example, diabetes and liver disease), as was suggested by an analysis of death certificates [52]. Table 2. Estimated Prevalence of Elevated Iron Status and Iron Overload Due to Hemochromatosis among Patients with Chronic Diseases* In the studies listed in Table 2, with the exception of one recent casecontrol

Hospitalization for pregnancy complications, United States, 1986 and 1987

Studies of the association of ectopic pregnancy with contraception have generated a conflicting array of results because of methodologic differences between studies. We estimated the absolute incidence rates of ectopic pregnancy for various contraceptives by multiplying the pregnancy rate by the proportion of pregnancies with ectopic implantation for each method. Our results indicated a more than 500-fold difference in ectopic pregnancy incidence, from a low of 0.005 ectopic pregnancies per 1000 women years of oral contraception or vasectomy to a high of 2.6 per 1000 women years of no contraception. These estimated incidence rates should be useful for clinicians and patients seeking to better understand the risks and benefits of contraceptives.

Screening for hemochromatosis

OBJECTIVE The purpose of our analysis was to provide a national overview of the magnitude of the public health burden associated with inpatient care for pregnancy complications. STUDY DESIGN We analyzed data from the National Hospital Discharge Survey for 1986 and 1987. We calculated ratios of hospitalizations for pregnancy complications for every 100 hospitalizations involving a birth. Standard errors for these ratios were calculated with RATIOEST, and relative ratios with 95% confidence intervals were calculated for subgroups of interest. RESULTS We found that for every 100 hospitalizations involving a birth, there were 22.2 nondelivery hospitalizations for pregnancy complications (14.6 antenatal complications, 7.6 pregnancy loss complications). These ratios were higher for black than for white women (relative ratio 1.4, 95% confidence interval 1.2 to 1.6). The effects of marital status, age, and insurance coverage differed between black and white women, and mean length of stay was longer for black than for white women. CONCLUSION Hospitalization for pregnancy complications is far more common than is widely appreciated and is more frequent among black than white women.

Postmenopausal smoking, estrogen replacement therapy, and the risk of endometrial cancer

CONTEXT The discovery of the HFE gene in 1996 has introduced DNA testing as a possible tool for screening and diagnosis of hemochromatosis and increased interest in the disorder. Population screening using transferrin saturation has been advocated by experts to permit early detection and treatment with phlebotomy before the onset of clinical disease. METHODS Based on a literature review, we consider the relative risks and merits of two screening tests as part of a broader look at the evidence required for the recommendation of universal screening for hemochromatosis. RESULTS Several questions must be answered before universal screening can be recommended. Uncertainties remain about the penetrance and preventable disease burden, laboratory standardization, and optimal strategies to minimize potential risks of screening for hemochromatosis. CONCLUSIONS As a common genetic disorder with simple, effective therapy, hemochromatosis offers a model for other genetically influenced chronic diseases that some day may have interventions to improve prognosis. Resolution of questions related to prevention of chronic diseases from hemochromatosis, therefore, will have broad usefulness in the future.

Case-fatality rates for tubal sterilization in U.S. hospitals, 1979 to 1980

Previous studies of the association between cigarette smoking and endometrial cancer have yielded inconsistent results. There is some evidence that this discrepancy may be explained by differences in menopausal status between the groups of women studied. We addressed the issue of the effects of postmenopausal smoking on endometrial cancer risk using data from the Cancer and Steroid Hormone Study, a multicenter, population-based, case-control study of gynecologic cancers in the United States. We found that smoking after natural menopause is associated with a 70% reduced risk of endometrial cancer among estrogen users and a 50% reduced risk among nonusers of estrogen. These findings are consistent with previously proposed biologic effects of smoking on estrogen metabolism, which may have important clinical implications.

HFE genotype and transferrin saturation in the United States.

To update a 1977 to 1978 case-fatality estimate for tubal sterilization in U.S. hospitals, we reviewed the medical records of women reported by the Commission on Professional and Hospital Activities to have died after tubal sterilization procedures in 1979 or 1980. We project that the most reasonable case-fatality rate estimate is slightly greater than 9 per 100,000 sterilizations if all deaths associated with the procedure are considered. Rate estimates that assume minimum and maximum numbers of all associated deaths in our sample are approximately 6 per 100,000 and 10 per 100,000 sterilizations, respectively. However, when only deaths that can be attributed to sterilization per se are considered, the most reasonable case-fatality rate is estimated at between 1 and 2 per 100,000 procedures, a lower rate than previously reported. Rate estimates that assume minimum and maximum numbers of attributable deaths in our sample are approximately 1 per 100,000 and 5 per 100,000 sterilizations, respectively. These results further indicate that death attributable to tubal sterilization is rare.

Fatal pulmonary embolism during legal induced abortion in the United States from 1972 to 1985

Purpose: Examine the penetrance (defined by high transferrin saturation [TS]) of C282Y and H63D in the U.S. population.Methods: 5171 participants from the Third National Health and Nutrition Examination Survey, 1992 to 1994.Results: 77.1% (95% confidence interval [CI], 2.3, 95.1) of men and 51.9% (95% CI, 0, 84.2) of women with C282Y homozygosity had high TS. The associations of H63D homozygosity with high TS were stronger in people aged 50 years or older than in younger persons. Among Mexican-Americans, simple H63D heterozygosity was associated with high TS.Conclusions: The associations between HFE genotype and high TS may vary by sex, age, and ethnic group.

Research priorities in hereditary hemochromatosis.

To determine the risk factors for abortion-related deaths caused by pulmonary embolism, we investigated all deaths from legal abortions in the United States from 1972 through 1985. Of 213 deaths, 45 (21%) were due to air, blood clot, or amniotic fluid embolism. The risk of embolism death was higher among minority women and older women (34 to 44 years). Our analysis revealed that curettage at less than or equal to 21 weeks and abortions at less than or equal to 12 weeks, regardless of method, were both associated with the least risk of embolism death. In comparing 1972 to 1978 and 1979 to 1985, we found that the embolism mortality rate decreased 79%. During 1979 to 1985, the number of abortions performed by noncurettage methods decreased 58%, possibly as a result of earlier abortion morbidity studies, which showed that these methods carried a greater risk of complications. Although a decrease in mortality rates may be partially attributable to the declining use of these methods, our analysis suggests that changes in methods over time have not been universally applied to all racial groups.