Papilledema in 15 renally compromised patients treated with growth hormone.

Intracranial hypertension with papilledema has been reported in renal patients, but a survey of the literature suggests that the incidence rate is low. We present reports of 15 of approximately 1,670 patients with renal disorders, who were treated with growth hormone for impaired growth and subsequently developed symptoms and/or signs of intracranial hypertension. The male:female ratio was 6.5:1, and the median age was 12 years. The median duration of growth hormone treatment before onset of symptoms or signs was 13 weeks. All but 2 patients were symptomatic. In the patients in whom growth hormone therapy is known to have been discontinued, the symptoms and signs of intracranial hypertension abated. At least 4 of these patients experienced a recurrence when re-exposed to growth hormone. Many of the affected patients presented with predisposing conditions, but growth hormone appears to have been the precipitating factor. Prospective funduscopic evaluation may be warranted in patients with renal disorders who are receiving growth hormone.

Benign intracranial hypertension in children with growth hormone deficiency treated with growth hormone.

We report 13 cases of benign intracranial hypertension (IH) in children with growth hormone (GH) deficiency treated with GH in the United States. The group consisted of eight boys and five girls, 3 to 16 years of age (median, 9 years). The interval from starting GH therapy to diagnosis of IH was 2 weeks or less in six patients, between 2 and 12 weeks in four, 8 months in one, 5 years in one, and unknown in one. Seven patients were not known to have previously described IH risk factors; the other six had at least one factor each. All patients but one had headache, nausea, vomiting, and visual changes. All had papilledema, and cerebrospinal fluid pressures were elevated (> 250 mm H2O) in all nine patients tested. The GH dosage range was 0.17 to 0.35 mg per kilogram body weight per week (median, 0.30 mg/kg per week) for the 11 patients with dosage data. After discontinuation of GH and treatment with lumbar punctures and/or medications, signs and symptoms resolved in eight children; in two of these children signs and symptoms reappeared when GH therapy was restarted. In four patients signs and symptoms resolved while GH therapy was continued; one child was treated with a ventriculoperitoneal shunt because of an arachnoid cyst, after which GH was restarted without subsequent IH. In the 12 patients with idiopathic GH deficiency the course of IH was benign, with complete resolution of all signs and symptoms. Because doses and scheduling of GH administration have changed since the introduction of recombinant GH, higher doses and increased frequency of administration may be contributing to the development of IH in some patients. We suggest beginning therapy at the lowest recommended dose, with gradual titration to higher doses, and the performance of routine funduscopic examinations during initiation of GH therapy and whenever signs or symptoms of IH develop.

Prepubertal gynecomastia during growth hormone therapy.

We report 22 cases of prepubertal gynecomastia diagnosed during growth hormone (GH) treatment. The age and dose range were 2 to 12 years and 0.18 [corrected] to 0.3 mg/kg per week, respectively. In most of the patients, gynecomastia appeared between 0.5 and 7 months after GH was started. Three boys were using drugs other than GH. This condition appears to be self-limited and benign.

The risk of epithelial ovarian cancer in short-term users of oral contraceptives.

Short-term use (less than 1 year) or oral contraceptives has been associated with increased to slightly decreased risks of epithelial ovarian cancer in several studies. To determine what might account for a statistically significant 40% reduction in risk associated with as little as 3 to 6 months of use, a finding previously reported from the Cancer and Steroid Hormone Study, and to consider the implications for mechanisms of pathogenesis, the authors compared numerous characteristics of short-term users of oral contraceptives (41 cases, 412 controls) with those of never users (242 cases, 1,517 controls). The reduced risk among short-term users was consistently restricted to women who stopped using oral contraceptives for medical reasons, which were essentially side effects; there was little evidence of a protective effect among women who stopped for nonmedical reasons. Factors such as age, parity, family history of ovarian cancer, estrogen dose, history of sterilization, and latency (interval from first use) could not account for the finding. These analyses suggest that short-term use of oral contraceptives has little to no effect per se on reducing the risk of epithelial ovarian cancer and that side effects resulting in cessation of oral contraceptive use shortly after it was begun may be indicative of factors that are protective against the disease.

Breast cancer detection in relation to oral contraception.

Analyses of tumor size and breast cancer stage were used to determine whether biased detection of breast cancer could have materially influenced estimates of risk associated with use of oral contraceptives. In a population-based case-control study conducted from 1980-1982, surveillance for breast cancer by breast exams, but not mammography, was found to be strongly linked to use of oral contraceptives. Tumors were slightly smaller and less likely to be late-stage (TNM stage III or IV) in patients who had used oral contraceptives. The net effect of any diagnostic bias on advancing the date of cancer diagnosis, whether from breast exams or other sources, was estimated to be less than 8 weeks. This corresponds to spuriously increasing the risk of early-occurring breast cancer in oral contraceptive users by at most 2.4% (relative risk = 1.024).

Clustering of adverse drug events: analysis of risk factors for cerebellar toxicity with high-dose cytarabine.

BACKGROUND: Cerebellar toxicity is a severe, therapy-limiting adverse reaction of cytarabine given in high doses. The Food and Drug Administration received a report of an increased frequency of cerebellar toxicity at the University of Wisconsin Hospital and Clinics after a switch from the product (Cytosar-U) manufactured by The Upjohn Co., Kalamazoo, Mich., to the generic form made by Quad Pharmaceuticals, Inc., Indianapolis, Ind. PURPOSE: To compare the incidence of cerebellar toxicity in Quad-treated patients with Upjohn-treated patients, a record-based cohort study was conducted at the University of Wisconsin Hospital and Clinics between January 1986 and August 1989. METHODS: The incidence of cerebellar toxicity was studied in 63 leukemia patients according to the manufacturer of the product received (34 Upjohn only, 25 Quad only, and four both manufacturers). The relative risk of cerebellar toxicity was adjusted for other known risk factors. RESULTS: Patients in the manufacturer-defined treatment groups did not differ significantly with respect to age, sex, type of leukemia, disease stage, calculated creatinine clearance, presence of abnormal liver function tests, or total dose received. The crude relative risk of cerebellar toxicity comparing the Quad product with the Upjohn product was 5.0 (95% confidence interval = 1.8-13.7). Adjustment for potential confounders did not alter the association. Other risk factors for cerebellar toxicity, independent of manufacturer, were age greater than 50 years, type of leukemia, disease stage, total dose greater than or equal to 20 g/m2, abnormal pretreatment liver function, and reduced creatinine clearance. CONCLUSION: This study found a significantly higher incidence of cerebellar toxicity with high-dose cytarabine manufactured by Quad Pharmaceuticals when compared with the incidence of cerebellar toxicity with the Upjohn product. Further research at independent institutions would be necessary to allow generalization of this finding. In addition, our findings suggest that a dose reduction in high-dose cytarabine therapy may be indicated for patients with reduced glomerular filtration rates.

Oral contraceptive estrogen dose and the risk of deep venous thromboembolic disease.

Despite the well-recognized association between oral contraceptives and deep venous thromboembolism, little is known about the risks associated with currently marketed formulations containing less than 50 micrograms of estrogen. To assess the venous thrombogenicity of low-estrogen oral contraceptives (those containing less than 50 micrograms of estrogen) relative to intermediate-dose (50 micrograms of estrogen) and high-dose (greater than 50 micrograms of estrogen) formulations, we conducted a cohort study of oral contraceptive users between the ages of 15 and 44 years in the Michigan Medicaid population. The period of the study was from 1980 through the third quarter of 1986. A total of 2,739,400 oral contraceptive prescriptions received by 234,218 women were analyzed. Using the low-estrogen cohort as the referent group, the age and calendar period adjusted relative risk of venous thromboembolism in users of intermediate-dose formulations was 1.5 (95% confidence interval (CI) 1.0-2.1, p = 0.04), and the relative risk in users of high-dose formulations was 1.7 (95% CI 0.9-3.0, p = 0.06). These data provide evidence that the dose-response relation between oral contraceptive estrogen and venous thromboembolism extends from 50 to 30 micrograms of estrogen, the dose range of currently marketed formulations.

Trends in the content and use of oral contraceptives in the United States, 1964-88.

Drug marketing and physician survey data were used to examine trends in the use and hormonal content of oral contraceptives in the United States between 1964 and 1988. Retail prescriptions for oral contraceptives peaked at approximately 68 million in 1973 and have remained between 50 million and 60 million since 1981. Despite this relative consistency in the number of prescriptions, physician "mentions" of oral contraceptives have increased by approximately 75 percent. This increase may reflect closer monitoring of women on oral contraceptives. Use of multiphasic formulations has steadily risen, accounting for 37 percent of the oral contraceptive prescriptions in 1988. Mean estrogen and progestin doses in all types of formulations have steadily declined. A change in the type of estrogen and progestin used in preparations has coincided with this decline in dose. The association between age and use of high-dose formulations seen in the past was no longer evident in 1988. The data demonstrate that oral contraceptive formulations in wide use today differ in hormone content from those of the past, when most of the major studies addressing the risks associated with oral contraceptive use were completed. There is therefore a need to determine the risks and long-term effects associated with these newer formulations.

A method of pharmacoepidemiologic analysis that uses computerized Medicaid.

A method of pharmacoepidemiologic data analysis that utilizes computerized Medicaid data is presented. A cohort design in which Medicaid enrollees receiving drugs that are normally used to treat similar underlying conditions is described. A period of time in which Medicaid service transactions are evident is required before an individual is eligible for selection into a cohort. Selection of study subjects and descriptions of cohorts are based on Medicaid service histories occurring during the preliminary, prerequisite period. Time at risk is considered to begin after a prescription for a study drug is dispensed and continues until either a refill is dispensed, a prescription for an alternative drug within the same therapeutic class is dispensed, or a predetermined number of days has passed. Subjects are followed forward in time and relevant health care transactions that are suggestive of suspected adverse drug reactions are noted. Incidence densities associated with sequentially ranked prescriptions within sequential courses of therapy are compared. Methods to increase the accuracy of case ascertainment are briefly discussed. Separate validation studies may be used to evaluate the validity of computerized case ascertainment methods and to compensate for misclassification of outcome. The proposed method is intended to provide timely estimates of risk for selected outcomes. For outcomes that cannot be accurately ascertained from computerized data, this method may be useful in determining the feasibility of more customized studies.

Oral contraceptive use in women with a family history of breast cancer.

To evaluate the effect of oral contraceptive use on the risk of breast cancer from 20-54 years of age in women with a family history of the disease, we analyzed data from the Cancer and Steroid Hormone Study of the Centers for Disease Control and the National Institute of Child Health and Human Development. For 2 years, beginning December 1980, the study enrolled from eight geographical areas in the United States 4730 women with breast cancer and 4646 controls who were breast cancer-free. For women with a first-degree family history of breast cancer, 554 cases and 280 controls, there was no evidence that use of oral contraceptives, even long-term, contributed to their risk of the disease. Neither total duration of use nor duration of use before first term pregnancy bore any relationship with breast cancer risk. Analyses designed to reveal a potential latent effect also showed no evidence of an adverse effect. For women with a second-degree family history of breast cancer, 777 cases and 595 controls, some isolated elevations in risk were observed for selected subgroups of oral contraceptive users. Detailed analyses of oral contraceptive formulation, the characteristics of the women involved, and the patterns of risk observed by latent period and duration of use suggest that these results, most within the limits of chance variation, are not likely to be a consequence of oral contraception.

Oral contraceptives and premenopausal breast cancer in nulliparous women.

Characteristics of women with potential for modifying the relationship between use of oral contraceptives (OCs) and the risk of premenopausal breast cancer were investigated using data from the Cancer and Steroid Hormone study, a population-based, case-control study conducted in eight geographic areas of the United States. Cases consisted of 2945 women who were premenopausal and 20-54 years of age when breast cancer was diagnosed between December 1, 1980, and December 31, 1982; controls consisted of 2646 women with no history of breast cancer who were also premenopausal and 20-54 years of age when selected during the same period. Results are presented with the cases and controls divided into eight groups on the basis of age at diagnosis on selection (20-44, 45-54), parity (0, greater than or equal to 1), and age at menarche (less than 13, greater than or equal to 13). Among nulliparous women who experienced menarche before age 13, the relative risk of developing breast cancer in the age interval 20-44 years is estimated to be 1.0 for never-users of OCs (reference), 1.3 for greater than 0-3 years of use (95% confidence intervals 0.7-2.4), 1.3 for 4-7 years (95% CI 0.7-2.6), 2.7 for 8-11 years (95% CI 1.2-6.3), and 11.8 for 12 years or longer (95% CI 1.4-95.7). OC use is not significantly related to the risk of breast cancer among women in any of the other seven groups. These findings suggest that prolonged OC use may accelerate the onset of breast cancer for a small group of susceptible women while having no appreciable impact on overall risk. The findings should be regarded as tentative, however, since they are based upon numerous comparisons and because age of menarche was stratified at 13 years to highlight the concentration of breast cancer risk apparent in our data.

Breast cancer in relation to early use of oral contraceptives. No evidence of a latent effect.

A long-term effect of oral contraceptives (OCs) on breast cancer risk has been suggested as an explanation for some studies' failure to detect an association between OCs and breast cancer. To address this latency hypothesis, we analyzed data on 4714 case subjects and 4540 control subjects from the population-based Cancer and Steroid Hormone Study. No support was evident for a latent effect of OCs on breast cancer risk through age 54 years: among parous women who had cumulated more than six years of OC use before their first term pregnancy, the risk of breast cancer, relative to nonusers before first term pregnancy, was 0.6 at zero to four years after first term pregnancy (95% confidence interval [Cl], 0.2 to 1.8), 0.7 at five to nine years (95% Cl, 0.3 to 1.7), and 1.1 at ten to 14 years (95% Cl, 0.3 to 3.9). Among nulliparous women with more than six years of OC use in total, the relative risk of breast cancer, by interval from last use of OCs, was 1.3 at zero to four years (95% Cl, 0.8 to 2.0), 1.1 at five to nine years (95% Cl, 0.5 to 2.0), and 0.6 at ten to 14 years (95% Cl, 0.1 to 3.7).

Breast cancer risk in relation to type of estrogen contained in oral contraceptives.

We report analyses designed to address the recent hypothesis that women who use combination-type OCs containing ethinylestradiol (EE) are at increased risk of breast cancer before age 45, if use of such OCs occurs prior to first term pregnancy (FTP). Our findings, based on data from 1679 cases and 1689 controls, 20-44 years of age, from the population-based Cancer and Steroid Hormone Study, are against the hypothesis. The relative risk of breast cancer by duration of exclusive use prior to FTP of OCs containing EE is estimated to be 1.0 (1-12 months EE use), 1.2 (13-48 months EE use), and 0.9 (49+ months EE use). There was no evidence of a latent effect. Among parous women with 49+ months exclusive use prior to FTP of EE-containing OCs, the relative risk of breast cancer was estimated as 0.9 (0-4 years after FTP) and 0.6 (5-9 years after FTP). Among nulliparous women with 49+ months exclusive use of EE-containing OCs, the relative risk of breast cancer was estimated as 1.0 (0-4 years since first use), 0.7 (5-9 years since first use), and 1.1 (10-14 years since first use). With regard to exclusive use of OCs containing mestranol, parous women who used such preparations long-term before their FTP showed no alteration of breast cancer risk, even 15 years or more after pregnancy. Nulliparous women with exclusive use of ME-containing OCs did show elevations in breast cancer risk, but the magnitude of risk in relation to duration of use and latent interval shows no pattern that suggests a cause-effect relationship.

PIP: This report present analyses addressing the recent hypothesis that women who use combination-type oral contraceptives (OC) containing ethinylestradiol (EE) are at increased risk of breast cancer before age 45, if they use OCs before the 1st term of pregnancy (FTP). Our findings, based on data from 1679 controls, 20-44 years of age, from the population-based Cancer and Steroid Hormone Study, are against the hypothesis. The relative risk of breast cancer by duration of exclusive use prior to FTP of OCs containing EE is estimated to be 1.0 (1-12 months EE use), 1.2 (13-48 months EE use), and 0.9 (49+ months EE use). There was no evidence of a latent effect. Among parous women with the 49+ months exclusive use prior to FTP of EE-containing OCs, the relative risk of breast cancer was estimated as 0.9 (0-4 years after FTP) and 0.6 (5-9 years after FTP). Among nulliparous women with 49+ months exclusive use of EE-containing OCs, the relative risk of breast cancer was estimated as 1.0 (0-4) years since 1st use), 0.7 (5-9 years since 1st use), and 1.1 (0-4 years since 1st use). With regard to exclusive use of OCs containing mestranol (ME); parous women who used such preparations long-term before their FTP showed no alteration of breast cancer risk, even 15 years or more after pregnancy. Nulliparous women with exclusive use of ME containing OCs did show elevations in breast cancer risk, but the magnitude of risk in relation to duration of use and latent interval show no pattern that suggests a cause-effect relationship.

Postcoital test abnormalities in relation to contraceptive use.

Abnormalities of cervical mucus can have a bearing on a woman's fertility. One means of detecting the presence of such abnormalities is the postcoital test (PCT). As part of a population-based case control study of risk factors for infertility, the reproductive, contraceptive, medical, and sexual histories of women seeking treatment for infertility who had abnormal PCT results were compared with those of fertile controls. A greater proportion of infertile women with an abnormal PCT had previously used a diaphragm than had control women (relative risk (RR) = 3.5, 95% CI = 1.1-11.3). The excess risk associated with use of a diaphragm was particularly high for women who had used one for longer than one year (RR = 7.3, 95% CI = 1.4-37.8), or within one year of attempting to conceive (RR = 5.5, 95% CT = 1.4-22.1). No increased risk was associated with the use of other barrier methods, oral contraceptives, or the intrauterine device.

Exposure opportunity in epidemiologic studies.

We have shown that consideration of exposure opportunity is crucial to planning epidemiologic investigations because it focuses studies on questions of relevance, it helps to identify links to variables that can distort the assessment of an exposure's effect, and the cost of investigation can be reduced by the use of appropriate eligibility criteria. The reasons given by Poole for ignoring exposure opportunity are based on misleading claims of increased precision, disregard for probable confounding by factors related to exposure opportunity, and failure to consider realistically the cost of obtaining data on persons whose experience is irrelevant to the issues under investigation.

PIP: Poole's examples in his article on the relationship of exposure opportunity to validity and efficiency of case-control studies clearly demonstrate that the validity of a case-control study is not affected by inclusion or exclusion of persons who had no opportunity for exposure, provided that opportunity for exposure is not associated with other factors that alter the risk of the study disease. With regard to validity, in other words, exposure opportunity is important only insofar as it indicates confounding by some other factor. On the basis of Poole's reasoning, one might be led to think that opportunity for exposure is irrelevant to planning case-control studies. It will be shown, to the contrary, that consideration of exposure opportunity is important not only for the design of case-control studies but also for epidemiologic studies generally. The rationale for restricting a study to persons with an opportunity for exposure is principally a matter of focus, i.e., designing studies to answer questions of relevance and importance. Thus, despite Poole's suggestion to the contrary, one does not study the effects of oral contraceptive (OC) use in men when the real issue is the risk of disease in women who choose this contraceptive method. Poole's assertion that "exposure opportunity does not seem to concern researchers when they conduct follow-up studies" is clearly wrong. The prototypic followup study is a study that employs randomization to ensure that opportunity for exposure is identical for all participants. Observational followup studies, when appropriately designed, apply enrollment criteria based upon exposure opportunity to approximate the effects of randomization. Examples are cited to show that the target population for epidemiologic studies consists of persons at risk of both the study exposure(s) and disease(s). As a consequence, opportunity for exposure should be a criterion for enrollment of subjects. In case-control studies, opportunity for exposure, like other eligibility criteria, must be applied equally to cases and controls in order that they be sampled from the same target population. In practice, confounding by factors related to exposure opportunity is common. Any exposure that involves an element of choice should be considered a candidate for confounding. Poole's arguments for studying sterile women, based on considerations of "precision" are misleading. Even if a study's precision were "improved" by including subjects with no opportunity for exposure, investigation still should be restricted to persons with exposure opportunity, first, by reason of properly focusing the study, and second, to avoid the possibility that the assumption of no confounding by exposure opportunity might be wrong. Poole exaggerates the practical difficulties of determining exposure opportunity and underestimates its value in avoiding wasted effort. Consideration of exposure opportunity is crucial to planning epidemiologic investigations because it focuses studies on questions of relevants, helps to identify links to variable that can distort the assessment of an exposure's effect, and it reduces the cost of investigation.