Epidemiology of organomercury poisoning in Iraq

Epidemiology of organomercury poisoning in Iraq. I. Incidence in a defined area and relationship to the eating of contaminated bread A. W. AL-MUFTI,a J. F. COPPLESTONE, G. KAZANTZIS,C R. M. MAHMOUD,d & M. A. MAJID d A survey was carried out in a defined area in rural Iraq where there had been many cases of organomercury poisoning following the consumption of bread contaminated by mercury, in order to determine the true incidence of the disorder. The results were compared with those obtained from a similar rural area from which few cases had been reported. A questionnaire was used to determine the amount of contaminated bread eaten and the frequency of symptoms; a simple clinical examination was performed and blood and hair samples were collected for estimation of total mercury concentration. Of 700 people over the age of 5 years in the high-exposure area, 66% admitted to having eaten contaminated bread, while none of the 864 persons in the low exposure area had done so. The mean period during which contaminated bread was eaten was 32 days, but some people had eaten it for as long as 3 months. A mean of 121 loaves was eaten, the maximum being 480 loaves. For the mean number of loaves the intake of methylmercury was likely to have been between 80 mg and 250 mg, but the people who had consumed the largest amount of contaminated bread may have ingested up to 1 000 mg of methylmercury over a 3-month period. Of those with signs of alkylmercury poisoning at the time of the survey, 80% had eaten more than 100 loaves, and 53 (71%) out of 75 persons who had eaten more than 200 loaves showed some evidence of poisoning. The incidence rate for poisoning was estimated at 271 per 1 000; this figure includes a mortality rate of 59 per 1 000, 32 per 1 000 cases with severe disability, 41 per 1 000 cases with mild or moderate disability and 138 per 1 000 cases with only subjective evidence of poisoning at the time of the study. INTRODUCTION The outbreak of organomercury poisoning in Iraq during the winter of 1971-72 followed the distribution of alkylmercury-treated wheat and barley to farmers for sowing. In numerous households over the country alkylmercury-treated grain had been ground into flour and made into bread and, in addition, treated barley had been fed to livestock. The wheat had been treated predominantly with methylmercury compounds, the wheat samples analysed having a mean value of 7.9 Ag/g with a fairly wide range of 3.7-14.9 ug/g in 24 samples.1 Small amounts of ethylmercury and of other organomercury compounds were also detected. - Director-General of Scientific Affairs, Ministry of Health, Baghdad, Iraq. - Medical Officer, Vector Biology and Control, World Health Organization, Geneva, Switzerland. - Senior Lecturer, Middlesex Hospital Medical School, London, England. - Ministry of Health, Baghdad, Iraq. - 23 - 24 The pattern of organomercury poisoning over the whole country, as revealed by hospital admissions data, showed without exception that all the cases had occurred in rural areas and that no cases had been reported from towns. The hospital patients admitted to having eaten bread made from mercury-dressed wheat and no secondary cases had occurred as a result of eating animal products from herds fed with mercury-dressed grain. A true picture of the extent and severity of such an epidemic could not be obtained from hospital admissions data. It was known that fatalities occurred without admission to hospital and milder cases may well not have reported to a hospital at all. The extent of the outbreak could therefore be estimated only by a population-based field survey, which could also provide information on the clinical picture of the poisoning in those mildly affected, and on the relationships between the clinical picture, exposure to organomercury and the concentration of organomercury attained in the body. It was hoped that the survey could also be used as a baseline for continuing observations on the exposed population for the presence of long-term effects, including effects on growth and development in children. OBJECTIVES The objectives of the survey were: (1) to determine the extent, in terms of morbidity and mortality, of organomercury poisoning in a selected, severely affected area; (2) to investigate the degree of exposure in an area believed to be almost free from such poisoning; (3) to determine the relationships between the estimated intake of organomercury, the clinical picture, and the levels of organomercury observed in those exposed; (4) to delineate a population for a continuing study of possible long-term effects of organomercury, with particular reference to growth and development; (5) to assess the degree and nature of residual morbidity, which information would be valuable for the planning of remedial services. METHOD Logistic considerations in rural Iraq made it necessary to choose for the survey a limited area, within easy reach of Baghdad, where heavy exposure had occurred. In the absence of a frame for picking a random sample, the whole population in this area, or as near to this as possible, was studied. The Greater Mussayeb Project area, which is accessible by a good road, lies 60 km south of Baghdad. About 600 hospital admissions with a diagnosis of organomercury poisoning had originated from this area and approximately 500 of these came from two villages, Hillali and Zubaidi. Hospital returns of the 16 provinces showed that Babylon had the third highest admission rate for organomercury poisoning (3.2 per 1 000 inhabitants), but that the concentration of cases in the villages of Hillali and Zubaidi was one of the highest in the country. These villages had a total population of 5 300 and the ratio of hospital admissions to population was almost 1:10. People in rural Iraq live in widely dispersed villages that are often difficult to define geographically. The village of Hillali could be clearly defined by the system of waterways that form part of the Greater Musseyeb Project, and as such was very suitable for a population-based study of a clearly defined area. Most of the inhabitants of this area were farmers with their families working in the cooperative of Hillali, close to their homes. This was important from the point of view of completeness of the population study, 25 for it meant that the whole family could be seen by the examining teams. Access to the villages was by car along unmade roads, which became impassable during the rainy season and difficult during the heat of summer. This heavily exposed area was matched with a control area where the population led a similar rural existence but had little or no history of exposure to organomercury. There were 3 provinces in Iraq where hospital admissions showed a low incidence of organomercury poisoning, one being Anbar, which had a rate of 0.09 per 1 000 rural popula- tion. Less mercury-dressed grain was distributed in Anbar and it appeared that very little had been ground into flour and made into bread. A suitable "control" area within this province was located at Saglawiya, some 60 km west of Baghdad, which was sufficiently distant from the affected area to make the direct spread of mercury unlikely. However, Saglawiya could not be as well defined geographically as Hillali. Two doctors and 6 medical auxiliaries were formed into 2 teams with a vehicle and driver for each. One team operated in each village, although there was some interchange. Information was collected by a precoded questionnaire which had been tested on patients with organomercury poisoning both in hospital and in the community. Data collected included basic identification and demographic information, together with information on the dietary pattern, the extent to which mercury-contaminated bread had been eaten, and informa- tion on symptoms and signs indicative of organomercury poisoning. Time relationships were kept as simple as possible, all events being related to the end of Ramadan (19 November 1971) and to the birthday of the Prophet (26 April 1972). The relationship of the onset of each symptom to these times was recorded and subjects were also asked whether the symptom had improved, deteriorated, or remained constant up to the time of the interview. Clinical examination was performed where possible on all persons aged 5 years and over, and on some children below this age. The physical signs of organomercury poisoning relate mainly to the central nervous system, and their elicitation, which requires detailed neurological examination, can be difficult. In a field survey, where examination often had to be performed on the floor of rural homes with members of the family standing around, examination had to be simple and short. The doctors recorded the presence of speech disturbances and any grossly abnormal behaviour. Auditory acuity was assessed by the subjects' response to the quietly spoken word as in normal conversation. Tests for visual acuity were restricted to the elicitation of the outline of the field of vision by means of confrontation, using a moving index finger as the stimulus. It was appreciated that this was a difficult test to apply under field conditions, but it is crucial in diagnosis of severe organomercury poisoning. The results of this test were recorded in 4 broad categories: normal, moderate constriction, tunnel vision, and blindness. Ataxia was assessed by means of the finger-to-nose test and by observing the gait while the person was walking in a straight line. Astereognosis was recorded in terms of normal or impaired ability, or total inability, to identify common objects. Current disability was assessed according to whether the person was (i) nonr-ambulant; (ii) ambulant but unable to wash, dress, or eat without difficulty; (iii) ambulant and able to perform the above but unable to follow usual occupation or to go school; and (iv) able to follow his usual occupation or to go to school. At the end of the survey the records of symptoms and signs for all subjects in the survey were assessed by the coordinator (G.K.) and classified in 5 diagnostic categories without reference to the eating of contaminated bread. The categories, which referred to the state at the time of the survey, were: 26 (1) no evidence of mercury poisoning; (2) mercury poisoning - subjective evidence only; (3) mercury poisoning - mild or moderate disability; (4) mercury poisoning - severe disability; (5) unable to determine. Venous blood and scalp hair samples for mercury analysis were collected from as many of the survey population as possible. A number of lactating women also provided samples of breast milk for analysis. Chemical analysis of whole venous blood and of hair processed in 1-in (2.54 cm) segments was performed for total mercury concentration by the atomic absorption method described by Magos2 and subsequently, for blood, by Magos & Clarkson.3 A small pilot survey was performed in October 1972 to see whether at this stage blood and hair samples would still be useful discriminators of exposure. Twenty subjects were chosen from the heavily exposed area and a similar number from the lightly exposed area. In the latter, 17 of the 20 values for blood mercury fell below 20 ng/ml and all except one value for hair mercury fell below 2.5 ,g/g. The results did not suggest that significant exposure to mercury had occurred, although serial examination of hair segments showed a small rise in mercury concentration in the third and fourth inches, which corresponded to the time of the outbreak. The results from the highly exposed group showed thatboth blood and hair mercury levels were sufficiently high to act as powerful discriminants in an epidemiological study even some months after the apparent end of the outbreak. The survey commenced in October 1972 and was completed in May 1973, although those who had eaten mercury-contaminated bread were seen by the end of April. Heavy rains in January added to the difficulty of field visits and the bulk of these were performed during February, March and April 1973. The greater part of the survey was, therefore, performed after an interval of about one year from the time of the outbreak as indicated by hospital admissions. While the defined area of Hillali was surveyed as fully as possible, a number of persons were also seen from the neighbouring village of Zubaidi, and from other adjacent farming communities. These persons, who made long journeys to request that they should be examined, were excluded from the defined population used for the determination of incidence rates and other epidemiological variables, but the data relating to clinical variables have been included in some of the tables. The questionnaire could not be fully completed for all persons who were included in the survey, although basic demographic data were almost complete. Information on age was missing for only 3 persons out of the total 2 147 surveyed. Some parts of the question- naire were not applicable to all persons. Questions on the amount of contaminated bread eaten and for how long, and the detailed questions on symptoms were only applied to those aged 5 years and over. Of the total of 1 743 persons aged 5 years and over, in all but 3 information on whether they had eaten contaminated grain was available. Information on symptoms and signs in relation to blood and hair mercury levels was available for 82% and 89% respectively of those over 5 years old. RESULTS Of the total of 2 147 persons surveyed, 926 persons were examined in Hillali, 207 persons from the neighbourhood of Hillali, and 1 014 persons in the lightly exposed area of Saglawiya. At a later date 348 persons from the two areas were re-examined and further blood and hair specimens were taken for a continuing study to be reported on later. 27 The age and sex distribution of these persons is shown in Table 1, which also gives for comparison similar data for the rural population of Iraq, based on the 1970 estimate of the population, which was derived from the 1967 census. The 926 persons from Hillali came from 179 households with a mean of 5.2 persons per household, although some households contained up to 20 persons. In Saglawiya the mean number of persons per household in the survey was between 1 and 2, the maximum being 19. Of the persons seen in Hillali, 61% worked on the land, compared with 11% in Saglawiya. On the other hand, one-third of the Saglawiya group attended school, compared with only 1.3% in Hillali at the time of the examination. TABLE 1. AGE AND SEX DISTRIBUTION OF THE SURVEY POPULATION OF HILLALI AND SAGLAWIYA, COMPARED WITH THE RURAL POPULATION OF IRAQ (1970 estimate) Hillali Saglawiya Rural Iraq Age male female male female male female(years) No. % No. % No. % No. % % % 1 22 5.6 31 5.8 5 0.9 10 2.2 4.7 4.4 1-4 90 22.8 83 15.6 73 12.9 60 13.4 15.4 15.2 5-9 73 18.5 80 15.0 181 32.0 79 17.6 15.6 14.9 10-14 35 8.9 69 13.0 193 34.2 58 12.9 13.4 11.8 15-19 27 6.8 43 8.1 24 4.2 55 12.2 9.4 10.4 20-29 35 8.9 64 12.0 21 3.7 48 10.7 11.1 11.8 30-49 51 12.9 103 19.4 30 5.3 82 18.3 19.1 19.8 50+ 61 15.5 59 11.1 37 6.5 56 12.5 11.4 11.6 Unknown 0 - 0 - 1 0.2 1 0.2 - - Total 394 99.9 532 [100 565 99.9 449 100 100.1 99.9 Of 700 persons over the age of 5 years eaten contaminated bread, while none of the Saglawiya had done so. Of the 176 persons (8.6%) had eaten contaminated bread. in Hillali, 459, or 65.6%, admitted to having 864 persons in the low-exposure area of from the neighbourhood of Hillali, only 14 The bread eaten in rural Iraq is an unleavened flat disc of dough weighing about 200 g. A number of these small loaves may be eaten in a day. In the area of high exposure the average number of loaves eaten per day was 3.2, although the variation was large, some people having eaten up to 10 loaves per day. In the lightly exposed area, much less bread appeared to have been eaten, the average being 0.7 loaves per day. Fifty persons said that they had started eating the contaminated bread before the end of Ramadan (19 November) and had continued to do so for a mean period of 39 days; 420 people, who had started eating contaminated bread after that date, did so for a mean of 32 days, and some people continued for up to 3 months. Some people were unable to estimate the number of days on which they had eaten contaminated bread. There were altogether 427 persons for whom this information is available, and from these it was possible to estimate the mean number of loaves eaten over the entire period; this was 121, with a range of from 6 to 480 loaves. 28 Taking the average weight of a loaf as 200 g and assuming an average moisture content of 30%, the average methylmercury content of each loaf may be calculated. Bakir et al.1 found the mean concentration of methylmercury in their flour samples to be 9.1 JUg/g, with a range of 4.8-14.6 pg/g. On the basis of these figures, the intake of methylmercury from contaminated bread in Hillali averaged about 150 mg, with a range of 80-250 mg. If the higher figure for the methylmercury content of the flour is taken, then those people who consumed the greatest number of loaves, i.e., 480, would have ingested up to 1 000 mg of methylmercury over the 3-month period. In the highly exposed area bread was undoubtedly the most important article in the diet, which also consisted of milk and its products, dates, vegetables, and fats. Of the popula- tion surveyed, 95% ate meat less than once a week and 98% ate no fish. In the lightly exposed area meat was eaten a little more frequently; 75% of those surveyed ate meat less often than once a week and 93% did not eat fish at all. It is evident, therefore, that meat and fish, even if contaminated, would have made a negligible contribution to the intake of mercury in the surveyed population. Of the persons who had eaten contaminated bread, one-half reported symptoms that could have been due to alkylmercury poisoning. However, cannot be regarded as specific, and to evaluate their significance they to the frequency of their occurrence in those people who claimed not to contaminated bread, and who acted as the control group for this purpose a variety of these symptoms have to be related have eaten (Table 2). TABLE 2. FREQUENCY OF SYMPTOMS IN THOSE WHO HAD EATEN AND THOSE WHO HAD NOT EATEN CONTAMINATED BREAD Contaminated bread Symptom Eaten Not eaten No.- % of total No.- % of total Paraesthesia - circumoral 104 25 6 <1 - extremities 155 37 22 2 - elsewhere 118 28 9 <1 All paraesthesia 159 38 22 2 Persistent pain in limbs 107 26 12 1 Persistent headaches 91 22 12 1 Legs weak or unsteady 123 30 2 <1 Repeated falls 79 19 - - Inability to walk 61 15 - All gait disturbances 130 31 2 <1 Difficulty in feeding and dressing 42 10 - Change in speech 65 16 - Change in sight 92 22 11 1 Inability to see 16 4 - - Any sight disturbance 93 22 11 1 Change in hearing 32 8 2 <1 History of rash 59 14 2 <1 Total number of persons surveyed 415 1 012 - Of persons with symptom. 29 The commonest symptom reported was paraesthesia, which was present in 38% of 415 people who had eaten contaminated bread and in 2.2% of 1 012 people who had not done so. This most frequently involved the extremities, but was reported also over the trunk (28%) and in the circumoral region (25%). Other symptoms of frequent occurrence in the exposed group as compared with the control group were of difficulty in walking - 31% and 0.2% respectively. These symptoms were subdivided into a feeling of weakness or unsteadiness of the legs (30% of the exposed group), repeated falls (19%), and transient or more permanent inability to walk (15%). A nonspecific symptom such as persistent pain in the limbs might be considered not to provide a clear differentiation between the exposed-and the control groups. However, this symptom was present in 26% of the exposed group, but in only 1.2% of the control group. Similarly, persistent headaches were reported in 22% of the exposed group and in only 1.2% of the control group. Visual disturbance was complained of in 22% of the exposed group and 1.1% of the control group. Other symptoms recorded in both groups are shown in Table 2. A large difference in the frequency of symptoms in the surveyed population could, therefore, be demonstrated by dividing all those seen into two groups, those who had eaten and those who had not eaten contaminated bread. This difference in frequency was even more marked as regards the physical signs, (Table 3). Relevant abnormalities of gait were found in 11%, visual field defects in 9%, astereognosis in 7%, and ataxia of the arms in 5% of those who had eaten contaminated bread. Visual field defect was found in 3 persons in the control group ( <1%) and the other abnormalities were not found at all. The least discriminating of the signs was deafness, which was recorded in 13 of the exposed and 6 of the control group. Rash, which appeared important as a symptom when related to organomercury ingestion, could not be evaluated as a clinical sign owing to the time interval that had elapsed between the onset of poisoning and the examination. TABLE 3. FREQUENCY OF ABNORMAL SIGNS IN RELATION TO ORGANOMERCURY POISONING, ELICITED IN THOSE WHO HAD EATEN AND THOSE WHO HAD NOT EATEN CONTAMINATED BREAD Contaminated bread Sign Eaten Not eaten No.! % of total No.2 % of total Dysarthria 29 7 4 <1 Deafness 13 3 6 <1 Emotional instability 20 5 2 <1 Apathy 19 5 5 <1 Visual field - moderate constriction 24 6 2 <1 - severe constriction 8 2 1 <1 Blindness 6 1 - - All visual field defects 38 9 3 -l Ataxia 22 5 - - Stereognosis - impaired 13 3 - - - absent 18 4 - - Gait impaired - able to walk without assistance 37 9 - - Able to walk with assistance 2 <1 - - Unable to walk 8 2 - - All gait disturbances 47 11 - - Total number of persons surveyed 415 1 012 - Of persons with sign. 30 Each symptom was related to the total consumption of contaminated bread (Table 4). While some people claimed to have eaten large quantities of contaminated bread without developing symptoms, the proportion of those who did develop symptoms increased with increasing bread consumption. For example, 4% of those who had eaten a total of less than 50 loaves developed paraesthesiae, but 68% of those eating more than 200 loaves had done so. Nobody eating less than 50 loaves had symptoms related to ataxia of the arms, but with increasing bread consumption the percentage with ataxia increased progressively to 17% in those who had eaten more than 200 loaves. TABLE 4. DISTRIBUTION OF SYMPTOMS RELATED TO THE CONSUMPTION OF CONTAMINATED BREAD Bread consumption (loaves x days) Symptom 1-49 50-99 100-199 200+ Paraesthesia 4 26 50 68 Persistent pain in limbs 2 18 32 40 Persistent headaches 3 14 25 32 Difficulty in walking 6 25 41 47 Difficulty in using arms 0 6 14 17 Change in speech 3 7 23 28 Change in sight 2 13 28 39 Change in hearing 2 5 6 12 History of rash 8 10 18 16 Total persons in bread consump- 66 165 121 75 tion category Note: The figures relating to each symptom indicate the percentage of affected persons in those who claimed to have eaten the amount of bread specified in the column headings. In this table, and in Tables 5 and 6, only those people for whom bread consumption is known are included. A similar progression was seen in relating physical signs to bread consumption (Table 5). No neurological signs of alkylmercury poisoning were found in persons who had eaten fewer than 50 loaves, but the percentage with signs increased with increasing bread consumption. For example, ataxia of the legs was recorded in 5% of those who had eaten between 50 and 100 loaves, in 15% of those who had eaten between 100 and 200 loaves, and in 23% of those who had eaten more than 200 loaves. The total number of loaves eaten (i.e., loaves x days) was then related to the clinical diagnostic category (Table 6). Again, clear progression was demonstrated; 11% of those people who had eaten less than 50 loaves were diagnosed as suffering from alkylmercury poisoning, all on subjective evidence only; 34% of those who had eaten between 50 and 100 loaves, 59% of those who had eaten between 100 and 200 loaves and 71% of those who had eaten more than 200 loaves had some degree of mercury poisoning. 31 TABLE 5. DISTRIBUTION OF CLINICAL SIGNS RELATED TO THE CONSUMPTION OF CONTAMINATED BREAD Bread consumption (loaves x days) Sign 1-49 50-99 100-199 200+ Dysarthria 0 3 9 13 Deafness 0 2 2 4 Emotional instability/apathy 0 5 11 9 Constriction of visual fields 0 5 13 17 Ataxia of arms 0 2 10 8 Astereognosis 0 4 11 11 Ataxia of legs 0 5 15 23 Total persons in bread consump- 66 165 121 75 tion category Note: The tigures relating to each sign indicate the percentage of affected persons in those who claimed to have eaten the amount of bread specified in the column headings. TABLE 6. DISTRIBUTION OF CASES BY DIAGNOSTIC CATEGORY RELATED TO THE TOTAL CONSUMPTION OF CONTAMINATED BREAD Estimated grain Total Diagnostic category Total,Estimated grain p any eaten consumption Subjective Mild/moderate Severe diagnosis category No. % of total No. % of total No. % of total No. % 1-49 66 7 11 0 0 0 0 7 11 50-99 165 45 27 8 5 4 2 57 34 100-199 121 46 38 13 11 12 10 71 59 200-499 75 33 44 14 19 6 8 53 71 Total 427 131 35 22 188 The consumption of contaminated bread was examined in relation to age of persons placed in the 3 diagnostic categories and of a random sample of 59 persons with no evidence of mercury poisoning (Table 7). The mean consumption of contaminated bread rises in each age group, with increasing severity of mercury poisoning. While in each category the total amount of bread consumed increases with age, if bread consumption is expressed on an average body weight basis, the children in each of the affected groups appear to have consumed more contaminated bread than the adults. Taking only clinical criteria for the diagnosis of organomercury poisoning, and using the categories mentioned above, the incidence rate in the population of Hillali was calculated at 271 per 1 000, this figure including all those persons who had symptoms suggestive of poisoning but no abnormal signs on examination, and also all the reported deaths (Table 8). The total figure was subdivided as follows: mortality rate, 59.4 per (a 00o CU U). - C$ > >. 0 CU 0 CUCU 0 z cn o 1- N- N ,-4 ,-4 -4 U, a% 4 eJ U, ,-- ON U, 0 C.N - ,_ U) -- 0 CU U)>>-4 0U) '0 0x >'o -t N- 0 - 0 U, N4 0 CY)o*N CYN o U, z cq1 0L 0 ~4 -t 00 0 4 C o C -A U) - - o -t -< c' NJ -'- - >>4)e- N c r0U z' -4 4 CUt 0) 09:: 000%Y 0) '% N li %O C' 0 ) co co > U, 000% c'j '0 L4 C-4 r f 4 z 0% U, a% 40% 00104) % 4-i 144 U) co 0 0"i 4Ji ,0I 0 CU 4 000 UA 1-. 00 0 0 U -HO)r 0 4 rq r-400 0) '0 - 00 CU -H 0 -H4 0 0.r4 w .0 0g0 X ~ ~ ~~~~~~~~~~E-1E 32 14 co 04 000 ow a 0 0 00 H Eq z H z0 0 0 z 0 H z 00 0 0 H 00 H C) H H 00 -4 H Ev m E0 vo Eq C] co U1)*,0 co -H 03 -.) 0co CUbo 0 0 4-i CU 14- U) 0 0 ,4 0 '0 0 U) CU 4~i r0 -H 0q 44 0 0 CU '0 0 CU la, 0 CO 0)0 4w 0 U) 0%0 U, 0 4 CU0 0 U)CU -Hn - U) -' U - ' 0- 1- 4-o 0 CU 014 0 H0 V CU 0 6800i U,0 0a ul) 33 1 000; patients with severe disability, 32.4 per 1 000; patients with mild or moderate disability, 41.0 per 1 000, and patients with only subjective evidence of poisoning, 138.2 per 1 000. The incidence of cases of organomercury poisoning with any degree of disability at the time of examination was 73 per 1 000 of the population of Hillali. The case fatality rate was 22% if account is taken of all possible cases, but if those with subjective symptoms are excluded the case fatality rate is very much higher. TABLE 8. INCIDENCE RATES FOR DIFFERENT CATEGORIES OF ORGANOMERCURY POISONING FOR THE POPULATION OF HILLALI Number of Incidence per 1 000Category cases population 1. Subjective evidence only 128 138.2 2. Mild or moderate disability 38 41.0 3. Severe disability 30 32.4 4. Death 55 59.4 Total 251 271.0 DISCUSSION There are obvious limitations to any field survey. For a population-based study on the incidence of organomercury poisoning it is necessary to have an accurate estimate of the population at risk and an accurate estimate of the number of cases. In the absence of a population count it is difficult to assess the completeness of the survey. The fact that at Hillali people at work were likely to be found in fields close to their homes favoured a high coverage rate. The survey teams reported great interest in their activities on the part of the population. Nobody refused to supply basic information. A few people refused clinical examination and a few refused to give specimens of blood and hair. However, these were more likely to be unaffected than affected people. The diagnosis on which the incidence rate was based was made on the clinical picture and not on the laboratory results. The authors believe that, where relevant physical signs were demonstrated, organomercury poisoning undoubtedly existed. We were on less sure ground, however, in those cases where the diagnosis was based on symptoms alone. Data are presented later to show why we believe that the majority of these were in fact true milder cases of mercury poisoning. To what extent reported symptoms were inaccurate would not be easy to determine, for a repeatability study to test the accuracy of the answers given was not feasible. If confabulation of symptoms occurred, the most likely reason would have been the psychological impact of seeing other members of the family suffering or dying from organomercury poisoning. It cannot be discounted that some of the symptoms could have been due to such factors. The age distribution of the Hillali sample was not unrepresentative of the rural population of the country as a whole, although males were under-represented at 42.5% of the total, compared with 49.8% males for rural Iraq. It is likely that some adult males were living and working outside the area. However, in the low-exposure area of Saglawiya, male children of school age were considerably over-represented at the expense of the adult working population. This was due to the way in which this population was chosen. In Saglawiya, people were less interested in the survey and some households refused entry to the survey team. One boys' school was included in the study, and a few people who were visiting the village dispensary were included. This method of selection is reflected in the low figure for the mean number of persons per household in Saglawiya, as many individuals rather than family units were seen, and in the occupational status, including the higher number of schoolboys. On the other hand, the number of children attending school in Hillali 34 was exceptionally low. It is possible that the epidemic was responsible for children being taken away from school to help in the home. The poor match between the demographic characteristics of the heavily exposed and lightly exposed groups does not in fact create great difficulties for comparative purposes. The household unit in rural Iraq is homogeneous as regards diet. As there is little variation in diet within an area and little choice of eating-places, most people return home from work or school for their meals. The schoolchildren who were included in the sample may therefore be considered to be fairly representative of the families to which they belonged. Accuracy in recall of details of diet in a field survey is not to be expected, especially where quantitation is required. A reasonably correct figure for the number of loaves of bread eaten daily may well be given, and it is unlikely that this pattern was varied when the bread had been made from mercury-dressed wheat. However, to recall the number of days on which the contaminated bread had been eaten might be expected to be more difficult, especially after a time lapse of up to one year. During the survey, the answer to the question asked was often discussed, first among the members of the family; where there was a discrepancy of opinion, the interviewer took what he thought to be the most reliable answer, often that given by the head of the household. The persons seen in Saglawiya appeared to eat much less bread per day than those in Hillali. The diet in Saglawiya appeared to be more varied, and meat was eaten a little more frequently than in Hillali, so that there were more alternatives to bread in the diet. However, underestima- tion of the figure for Saglawiya is more likely to have occurred than overestimation of that for Hillali. Again, the high proportion of children seen in Saglawiya biased the sample, the number of loaves eaten daily being smaller. It may also be that the children seen at school without their parents, as happened in Saglawiya, gave a less reliable estimate. The mercury content of a loaf of bread must again have been a widely variable figure,depending on the weight of the loaf, its moisture content and the mercury concentration in the flour. The procedure for making the loaves was fairly standard in Hillali. Most of the earthen ovens in which these loaves were baked were made in the same way, and a loaf of about 200 g was the most convenient size for the ovens, so that it is unlikely that the weight of the loaves varied widely. What is likely to have been more variable was the mercury content of the flour. According to Bakir et al.1 the methylmercury content of 19 samples of flour varied from 4.8 to 14.6 Ag/g, that is, by a factor of 3. The estimate given of the total number of loaves of contaminated bread consumed (i.e., loaves x days) bore a clear relationship to the symptoms given, to the signs elicited and to the clinical diagnosis, so that, on a group basis, in each age group studied, those who had eaten the largest amount of bread suffered the most severe effects. However, the numbers were small, 22 of the 30 cases of severe disability that were seen in the survey providing information on which the mean figures were based. The considerable difference in each age group between the amount of contaminated bread eaten by those with subjective symptoms and by those with no evidence of poisoning provides a characteristic which helps to validate the former as a category of mercury poisoning with a history of symptoms but with no abnormal signs at the time of examination. Further evidence in favour of defining a category of alkylmercury poisoning with a history of symptoms but not gross neurological signs at the time of the survey was provided by the distribution of symptoms between those who had and those who had not eaten con- taminated bread. The difference in distribution was very marked, even for symptoms that can be considered nonspecific, such as headache and persistent pains in the limbs. The effect of bread consumption on the clinical picture showed that a few people who had eaten a total of less than 50 loaves of contaminated bread had symptoms but no signs 35 of mercury poisoning at the time of the survey. A few who had eaten between 50 and 100 loaves had signs of alkylmercury poisoning, but about 80% of those with signs had eaten more than 100 loaves. Fifty-three out of 75, i.e., 71% of all persons who claimed to have eaten more than 200 loaves, had some evidence of alkylmercury poisoning, with or without signs at the time of the survey. Although Table 7 shows an increased total consumption of contaminated bread with increasing age in each diagnostic category, young children consumed considerably more bread in relation to body weight than did adults. The data in this table also suggest that adults were more sensitive than children to the effects of mercury on a comparable dose per body weight basis. Assuming an average weight of 20 kg for the children aged 5-9 years and 60 kg for those 15 years old, at an average "dose level" of 3.6 loaf-days per kg, the young children seen had no evidence of poisoning, while adults (aged 15 and above) had severe poisoning. RESUME EPIDEMIOLOGIE DES INTOXICATIONS PAR ORGANOMERCURIELS EN IRAK I. INCIDENCE DANS UNE DES ZONES AFFECTEES ET RAPPORT AVEC LA CONSOMMATION DE PAIN CONTAMINE Une enqugte a ete menee dans une zone rurale d'Irak o'u s'etaient produits de nombreux cas d'intoxication par derives organomercuriels A la suite de la consommation de pain contamine; cette enqugte avait pour objet d'6tablir l'incidence rdelle de l'affection. Ses resultats ont dtd compares avec les donndes concernant une zone rurale analogue oui un petit nombre de cas avaient 6t6 signalds. On a utilise un questionnaire pour determiner la consommation de pain contamine ainsi que la frequence des symptomes et l'on a procdd6 A des examens cliniques simples ainsi qu'A une collecte d1'chantillons de sang et de cheveux pour 6valuer la concentration de mercure. Sur un total de 700 personnes Ag6es de plus de cinq ans interrogdes dans la zone de forte exposition, 66% ont d6clar6 avoir consomme du pain contamine, alors qu'aucune des 864 personnes interrog6es dans la zone de faible exposition n'en avait consomm6. La p6riode moyenne de consommation de pain contamind etait de 32 jours, mais pour certains subjets elle etait de trois mois. La consommation moyennea 6te de 121 pains, avec un maximum de 480. Il est probable que pour le nombre moyen de pains consommes I'ingestion de composes methylmercuriels se situait entre 80 et 250 mg, mais les personnes qui ont consomme les plus grandes quantit6s de pain contamine peuvent avoir ing6rd en trois mois jusqu'A 1 000 ml de composes methylmercuriels. Parmi les personnes presentant des signes d'intoxication alkylmercurielle au moment de l'enqugte, 80% avaient consomme plus de 100 pains et 53 (71%) des 75 personnes qui avaient consomme plus de 200 pains presentaient des signes d'intoxication. L'incidence de I'intoxication a ete estimee A 271 pour 1 000; il y a eu 59 cas mortels pour 1 000, 32 cas d'incapacite grave pour 1 000, 41 cas d'incapacite legere ou mod6r6e pour 1 000 et, toujours pour 1 000, 138 cas privsentant, au moment de l'enqu^te, des symptOmes d'intoxication uniquement subjectifs. 36 REFERENCES 1. BAKIR, F. ET AL. Methylmercury poisoning in Iraq. Science, 181: 230-241 (1973) 2. MAGOS, L. Selective atomic absorption determination of inorganic mercury and methyl mercury in undigested biological samples. Analyst, 96: 847-853 (1971) 3. MAGOS, L. & CLARKSON, T. W. Atomic absorption determination of total, inorganic and organic mercury in blood. J. Am. Off. anal. Chem., 55: 966-971 (1972)