at Hiroshima and Nagasaki.” Studies Danae et of exposures to iodine 131 in humans have been limited largely to ™I therapy for patients with Graves’ hyper- thyroidism. It is unclear from these studies whether “I alone results in an excess of thyroid nodules.**“' Much less is known about the health risks of exposure to short-lived radioiodines other than ™I. This information may be important in assessing the impact of radioiodine exposure from nuclear reactor accidents. While the people from Rongelap and Utrik have been exhaustively studied during the last 33 years, these previous studies of thyroid neoplasia did not include the total geographical extent of the Republic of the Marshall Islands. To define more carefully the risk of thyroid neoplasia from nuclear fallout containing radioactive iodines, we conducted a retrospective cohort study of thyroid nodules in 7266 Marshallese people from 14 atolls, including several southern atolls, which served as the source of the best available unexposed compari- son group. METHODS Study Hypothesis The objectives of this study were as follows: (1) to determine the prevalence of thyroid nodules in people who were living on 14 northern and southern atolls at the time of the 1954 BRAVO detonation; (2) to test the null hypothesis that no difference existed in the prevalence of thyroid nodules amongthe12 atolls of this study previously thought unexposed to radioactive fallout; and (3)if the null hypothesis is rejected, to deter- mine which factors might explain the variation in rates of thyroid nodules. Study Location The Marshall Islands are located 2400 miles southwest of Hawaii; approxi- mately 35000 people (1985) live on 24 atolls spread among 375000 square miles in the central Pacific Ocean. This population is distributed roughly in thirds on the following atolls: Majuro Atoll, the administrative district of the government of the Marshall Islands; Kwajalein Atoll; and the remaining 22 atolls, known collectively as the “outer istands.” This study took place between June 1983 and March 1985 on 14 of the 24 inhabited atolls in the Marshall Islands (Fig 1). For this study, northern atolls were defined as those north of Majuro (Rongelap, Utrik, Mejit Island, Ailuk, Likiep, Wotje, Maloelap, Kwajalein, Lae, Ujae, and Wotho), and southern atolls were defined as those south of Majuro WJaluit, Ebon, and Mili). These 630 JAMA, Aug 7, 1987—Vol 258, No. 5 14 study atolls were selected to include all northern atolls that could have possibly been in the path of fallout and as manysouthern atolls as logistically feasible. Atolls that were not studied in- cluded five central atolls, two currently uninhabited northern atolls (Rongerik and Ailingnae), one southern atoll, and two atolls west of Bikini. Study Design and Sample A population-based retrospective cohort design was employed. Among the 7266 Marshallese people screened in this study, 2273 persons were alive in March 1954 and were residing on one of the 14 study atolls; they were, therefore, potentially exposed to the shortlived radioiodines. Since only these people were at risk for radioiodine-induced thyroid neoplasia,it is this group of 2273 persons that makes up the sample in this study. During the course of this study, all residents (age 6 years and older)of each island selected for screening were in- vited and encouraged to receive thyroid examinations. Extensive discussions with traditional leaders of each atoll were conducted prior to each trip to ensure maximal communication to residents of each island. One to two weeks were spent on each atoll performing the screening examinations. The population of each atoll at the time of screening was estimated from the 1980 Marshall Islands census data.* To offset the effect of self-selection by islanders of each atoll population, we attempted to screen the entire popula- tion of 13 primary atolls. Since migra- tion out of the country is rare, the primary problem was capturing those members of the population, especially the population of 1954, who had moved to either of the two population centers, Majuro or Kwajalein. Screening pro- grams were therefore conducted on Majuro and Kwajalein for those individuals and their families who had lived on any of the 13 northern or southern atolls in 1954. However, since we screened nearly a third of the population of Kwajalein Atoll for thyroid nodules, we also included Kwajalein as a primary atoll, makinga total of 14 studyatolls. Exposure Criteria Since the short-lived radioiodines (I, I, ™I, ™) all have half-lives of less than eight days, the bulk of the radioiodine exposure from the BRAVO event occurred during the month of March1954. Therefore, the most important historical information concerning the radioiodine dose was the location of residence in March 1954. Because most individuals can provide vivid descrip- Table 1.—Distance and Direction of Study Atolls From Bikini Atoll es Atolt Rongeiap - Utrik Mejit Island+ Ailuk Likiep Wotje Maloelap Lae Ujae Wotho Kwajalein Jaluit Distance, Miles* 6, “t 120 20 321 398 342 306 376 460 198 187 1t2 192 6 15 18 26 25 28 71 80 64 51 538 589 64 42 500 Ebon Mili 54 nS "Distance from atoll to BRAVO testsite on Bikini Atoll in statute miles. tAngle of atoll from 0° latitude line drawn through Bikini Atoll, tMejit island is classified as an atoil for the purposes of this study. tions ofwhat they were doing during the dramatic BRAVOtest, the question was posed in the following manner: “Where were you living when the ‘bomb’ caused the Rongelap and Utrik people to be moved from their homeland?” Individuals born after March 1, 1954, but before Dec 31, 1954, were classified as in utero at the timeof the blast, and their atoll of residence in 1954 was classified according to their atoll of birth. Becauseof the relative proximity of all the islands within each atoll and the long distance between any atoll and the blast site, all individuals from different islands within an atoll were classified by the atoll name for the purposes of 1954 residence status. Since the people live on small land masses, the atolls represent discrete points in the vast ocean area of the Marshall Islands. The distance from each atoll to the site of the 1954 BRAVO test (Bikini Atoll) was therefore selected as a proxy for the radioiodine dose received in 1954. A second variable was developed to better characterize the exposure status of the Marshallese people. A directional variable, 6, was selected as a proxy for meteorologic conditions, such as wind and precipitation, that may have influenced the distribution of the fallout cloud. We defined 6 as the angle in degrees, measured clockwise, of each of the 14 atolls from a 0° latitude line drawn through Bikini Atoll, using Bikini as the vertex. Table 1 shows the distance of each atoll from the BRAVOtestsite as well as the angle 9 of each atoll from an east-west line drawn through Bikini. Diagnostic Criteria We defined a thyroid nodule as one that was palpable, discrete, and estimated to be 1.0 cm or greater. Findings of indiscrete or uncertain lesions and Thyroid Neoplasia— Hamilton et al