and 80°, which bounds the area of this study. As shown in Fig 5, these probability contours illustrate that the chance of developing thyroid nodules is influ- enced by both distance and 6 in a variable manner. For example, for a fixed distance of 300 miles from Bikini, the probability decreases as 6 increases. However, for most fixed distances greater than 400 miles from the test site, the probability increases with 6. These results are consistent with previ- ously published fallout patterns showing an initial eastern pathway of the BRAVOfallout cloud.* They are also consistent with a computer simulation pattern that suggested that after a@ predominantly eastern direction, toward Utrik, the fallout cloud moved south and west from Utrik.* Absolute Risk Assessment The absolute risk coefficient has been used to compare the risk for thyroid nodules among exposed populations“ and can be expressed as follows: absolute risk coefficient = number of excess cases/Gy/years at risk/1x10° persons (number of excess cases/rad/years at risk/1 million persons), where number of excess cases is the number of observed nodules minus the number expected. Using a prevalence of nodules of 2.45% determined in this study for unexposed Marshallese, we determined a new absolute risk coefficient for the Rongelap and Utrik people exposed to radioactive iodines. Since estimates of the thyroid dose and years at risk for these populations were known from pre- Otherrisk factors for thyroid neoplasia, however, do not appear to be present. There is no evidence for iodine deficiency in this population: the diet of the Marshallese population is well known to have ampleiodine content, especially on the outer islands, where the dietis high in fresh fish.“ In contrast to the United States, no head and neck irradiation of Marshallese children was employed as therapy for benign diseases ofchildhood such as acne, presumed tonsilar or thyfungal infections of the scalp. There are no known dietary or environmental goitrogens that are used in the Marshall Islands. If other unknown risk factors for thyroid disease are present in this population, it must be postulated that they exert their effects in a pattern such that the risk from exposure decreases with distance from Bikini Atoll. Thus, that 12 of the 14 atolls in this study were unexposed, other environmental assessment studies reported evidence that suggests that fallout contamination was not limited to Rongelap and Utrik. Robison and colleagues“ reported that several inhabited atolls other than Rongelap and Utrik contained low lev- els of long-lived radionuclides that were though the dose extrapolations from 1978 to 1954 were not done for these 137 and strontium 90, would not have contributed significantly to the thyroid dose during these years. An additional report documented a gamma dose at Ailuk Atoll to be 0.01 Gy/h (1.0 rad/h) one hour after the BRAVOdetonation;® such data suggest that.this atoll, previously thought to be shall Islanders were not limited to the utilizing all available meteorologic data predicted that after an initial eastern direction, the maximal point of radiation 16 hours after the detonation would have been midway between Rongelap 634 JAMA, Aug 7, 1987—Vo! 258, No. 5 cause of continued exposure for the entire decay process ofthe radioiodines, their cumulative thyroid doses may have been as high as or higher than those on Utrik. One methodologic advantage af- forded by the Marshall Islands is that the geographyof these islands has pro- vided considerable variation on our not easily move from atoll to atoll in short periods of time, especially in 1954, makingit possible to ascertain on which atoll relative toBikini was also an im- two northern atolls of Rongelap and Utrik but occurred throughout many of the Marshall Islands. These findings suggest that the geographic extent of radioiodine exposure from the 1954 BRAVO test was much broader than previously assumed. Without thyroid dose estimates for peopleliving on 12 of the 14 atolls in this study, radiation exposure cannot be proved as the cause of these neoplasms. after detonation, no such evacuation took place on other atolls.“ Thus, peo- Although authors of previous clinical studies of Marshall Islanders assumed mostlikely cause of these neoplasms. atolls, the low doses received from the longer-lived isotopes, such as cesium portant risk factor. Our results indicate that excess thyroid nodules in Mar may have been affected by long exposure times. While people on Rongelap and Utrik were evacuated 48 to 72 hours proxy for exposure. While the thyroid sons). This study demonstrates a strong inverse linear relationship between the probability of thyroid nodules developing in Marshall Islanders and the distance of their 1954 home atoll from the Bikini test site. The direction of each atolls previously assumed to be free from exposure to radioactive fallout the absence of other known risk factors for thyroid nedularity and the presence of a strong inverse linear relationship between thyroid nodularity and the distance ofeach atoll from the BRAVOtest site suggest radioactive fallout as the likely residual from intermediate-range fallout in the Marshall Islands. Al- COMMENT (Fig 2) The thyroid doses of people living on ple on those atolls may have had lower mic enlargement, cervical adenitis, or . peak exposures than on Utrik, but be- vious studies (see “Methods” section), we calculated a new risk coefficient of 1100 excess cases/Gy/y/1x 10° persons (1.0 excess cases/rad/y/ million per nodules was highestin this region on the atolls of Lae, Ujae, Wotho, and Likiep unexposed, received fallout. A com- puter simulation of the fallout cloud and Kwajalein.“ This suggests that the fallout cloud may have shifted from an initial eastern path to a south or southwest direction. This simulation modelis consistent with the results of our study, which show that, except for Rongelap, the prevalence of thyroid doses for persons on these atolls is not known, the position of small land masses across thousands of square miles of ocean permits us to mow the distance from the blast site exactly. Second, residents of these atolls could atolls persons were living during the exposure period of the BRAVOtest. These factors may in part explain why the variables distance and @ appear to be such strong proxies for radiation dose or conditions that affected the dose. This study has several limitations that deserve mention. The ascertainment of exposure, which involves reports from participants, is subject to recall bias, especially in cultures such as the Marshall Islands that are not time oriented. Asking the question in terms of where onelived in March 1954 might yield answers of questionable accuracy. However, persons in this study were asked where they lived when the “bomb” exploded, causing the Rongelap and Utrik people to be evacuated from their homeland. The detonation of the BRAVO hydrogen test was a dramatic event: people on many atolls in the northern Marshalls could see the light, feel the blast, and see the fallout on vegetation hours after the blast. In much the same way as people recall clearly what they were doing at the time of Pearl Harboror the assassination of John F. Kennedy, this dramatic hydrogen bombaffected the Marshallese people in a way such that they could provide vivid descriptions of what they were doing and where they were living in 1954 at the time of the test. For individuals living on southern atolis who could not Thyroid Neoptasia—Hamilton et al