Movement Disorders Vol. 5 , No. I , 1990, pp. 66-70 0 1990 Movement Disorder Society
Follow-up Study of Early-Life Protective and Risk Factors in Parkinson’s Disease Lawrence I. Golbe, Timothy M. Farrell, and “Patricia H. Davis Department of Neurology, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, New Brunswick, New Jersey, and *Department of Neurology, Louisiana State University Medical School, Shreveport, Louisiana, U.S.A.
Summary: Previous studies suggest that Parkinson’s disease (PD) is negatively associated with early-life intake of vitamin E-rich foods and positively associated with rural experience. Using a new survey design, we attempted to confirm and extend these results. We gave a telephone questionnaire to 106 patients with PD and to their spouses as controls. It assessed premarital consumption of 31 foods of various vitamin E content, vitamin supplements, and exposure to rural living. Respondents rated food consumption with respect to what they perceived as the average for their sex and age at that time. We found female patients with PD less likely than spouses to have eaten “peanuts and peanut butter” (p < .05), which are high in vitamin E . “Salad with dressing,” also high in vitamin E, gave a similar result (p < .05) for a male-predominant patient group. Separate comparison of male controls with female controls ruled out sex-related preferences as the explanation of our findings. Patients had more extensive rural experience and were more likely to have frequently sprayed pesticides (p < .05) than had controls. Our results justify further investigations into early-life vitamin E intake, pesticides, and neurotoxins associated with rural life. Key Words: Parkinson’s disease-Diet-Vitamin ERural-Case control-Risk factors.
Therefore, we attempted to 1. confirm our earlier results using a different survey design; 2. extend those results by inquiring in more detail into specific vitamin E-rich foods and foods commonly eaten with them; 3. investigate a relationship between PD and early-life use of vitamin E and other oral antioxidant supplements; 4. confirm in a PD population of typical-age onset the greater rural experience that has been reported among persons with young-onset PD; 5. explore specific factors suspected of explaining that rural predilection; 6. investigate the epidemiological relationship between PD and early-life smallpox vaccination (7); and 7. repeat and extend our inquiry into foods that, by analogy with the cycad suspected of causing the amyotrophic lateral sclerosis-PD-dementia com-
In a case-control study of dietary risk factors in Parkinson’s disease (PD) we found that patients with PD were less likely than controls to have consumed foods rich in vitamin E during early life (1). Similarly, Tanner et al. (2) found early-life use of vitamin E supplements to be more common among controls than among those with PD. Experimental data in animal models of PD support the notion of a protective role of vitamin E, presumably via an antioxidative mechanism (3). A large double-blind study of the role of vitamin E in slowing the progression of early PD is in progress (4). Association of rural living with young-onset PD has been reported by Tanner et al. (5) and by Rajput et al. (6) using different survey designs. Address correspondence and reprint requests to Dr. L. I. Golbe at Department of Neurology, CN-19, UMDNJ-Robert Wood Johnson Medical School, New Brunswick, NJ 08903, U.S.A.
66
PROTECTIVE AND RISK FACTORS IN P D plex of Guam (8), may be positively associated with subsequent development of PD. SUBJECTS AND METHODS Rationale for Survey Technique Retrospective quantitative assessment of dietary habits is notoriously unreliable. Our previous survey (1) attempted to avoid this difficulty by assessing patients’ early-adult dietary preferences among patients and same-sex sibling controls relative to those of their respective spouses at that time. Although we pretested this technique for reliability, its validity as a solution to the problem of obtaining retrospective dietary data remains unsupported. Therefore, the present survey technique was designed along more orthodox lines, with spouses serving as controls. Again, however, we did not ask respondents to rate food consumption on an absolute scale. Rather, we asked them to compare their consumption of each food item with the level of consumption that they perceived a s average for persons of their age and sex at that time. Questions were expressed in terms of premarital exposure to minimize the “overmatching” that use of spouse controls may produce. Inasmuch as the cause of PD is hypothesized to act during childhood or early adult life, the date of marriage seemed a scientifically sensible boundary as well as one that would be relatively easy to recall in multiple-decade retrospect. The survey was conducted by telephone, because mail surveys in PD have been found insufficiently reliable for research purposes (9). Subject Qualifications Patients qualified if they had typical idiopathic PD, had been seen between January 1986 and July 1988 at the Movement Disorder Center at UMDNJRobert Wood Johnson Medical School, were not demented (per the medical record), were married before PD onset, and had no first-degree relatives with a parkinsonian disorder. Such PD kindreds (10) are the result more of genetic than environmental factors and therefore would confound our present inquiry into environmental factors. Spouse controls were excluded as demented if they could not give their zip code and the patient’s date of birth. These data were checked against the patient’s medical record.
67
Reduction of Sex-Related Artifact Because some dietary habits may be sex-related and we used spouses as controls, there was a risk that the male predominance in PD would produce artifactual dietary “predilections.” Our stipulation that the activity or dietary intake be compared with others “of your age and sex at that time” sought to eliminate this artifact. Nevertheless, we further reduced the sex-related artifact by performing the analysis for three groups. The first was the entire group of 106 patients (71 men and 35 women, with spouses as controls). The second consisted of only the 71 male patients (with wives as controls) and the third consisted of the 35 female patients (with husbands as controls). As additional vigilance against sex-related artifact, we tabulated the preferences of male and female controls, respectively, for the food items that gave statistically significant differences between cases and controls. In this way, we sought a m a l e female difference that might explain the casecontrol difference. Only 28 of the 106 patients (26%, 18 men, 10 women) had been respondents in our previous survey (1). Questionnaire Administration One trained and experienced questioner (T.M.F.) administered all questionnaires using a standardized format. Case-control pairs were surveyed only when both were home together. Spouse controls were questioned first and were asked not to repeat questions aloud lest the patient offer influence or formulate his or her own answers in advance. After the spouse completed the survey, he or she handed the telephone to the patient, who was instructed not to solicit advice from the spouse during questioning. Data Format and Analysis Cases and controls were asked whether their consumption of 31 foods was “much less than average,” “a little less than average,” “about average,” “a little more than average,” or “much more than average” for persons of their age and sex during the years between age 18 and the year of the present marriage. The food items are listed in Table 1 . Case-control pairs were considered “discordant” if their answers on this five-point scale differed by at least two levels. For example, if for a given food item the patient responded “much less
Movemenl Disorders, Vol. 5 , No. I , 1990
L . I . GOLBE ET AL. TABLE 1. Odds ratios for 31 food items Sex of cases
71M 35F
71M OF
OM 35F
(controls were their spouses) Salad with dressing Salad without dreGing Spicy foods Fried foods Raw tomatoes Cooked tomatoes or tomato sauce Tomato juice Catsup Raw onions Cooked onions Lettuce Spinach Cucumbers Green peppers Radishes Olives Olive oil Salad oil or dressing Peanuts or peanut butter Nuts other than peanuts Walnuts Cashews Almonds Apples Apple juice Bananas Peaches Plums or prunes Grapes Strawberries Other bemes
TABLE 2. Smallpox vaccination as a Parkinson’s disease risk factor (all differences not significant)
kappa 0.6“ 0.6 1.6“ 1.3 0.8
0.5“ 0.7 2.4“ I .9“ 0.8
0.9 0.6 0.8 0.6 0.7
.33 .46 .48 .73 .38
1.3 0.9 0.9
1.6 1.2 1.2 1.5 0.8 0.7 1.1 0.7 0.7 2.0“ I .2 1.4 0.7 1.1 1.5 1.4 1.4 1.3 1.7 1.9” 1.8
0.7 0.5 0.7 0.4 1.1 0.9 0.6 0.5
.45 .52 .82 .36 .31 .39 .42 .45 .36 .48 .61 .46 .28 .40 .53 .23 .43 .48 .39 .48 .24 .29 .30 .24 .36 .54
1.o
0.9 0.8 0.9 0.6 0.8 1.5 0.9 1.3 0.8 0.7 0.8 1.O
.o
1
I .o 1.5 1.3 1.1 0.9 0.9 1 .o 0.9 1 .o
1.o
1.1 1.3 1.2 1.2
1.o 0.9
0.5 1.1 0.5 0.3“ 0.3“ 0.6 0.5 0.5 1.2 0.6 0.4 0.8 0.7 0.5 0.6 0.6
Three separate analyses were made to examine the effect of sex on food preferences. Odds ratios (OR) and kappa are defined in the text. For OR < 1, the cases were less likely than controls to have eaten the item. a p < .05 (Mantel-Haenszel x ‘).
than average” and the spouse the same or “a little less than average,” the pair was considered concordant. If that spouse had answered “about average,” the pair would have been scored as “patient < spouse.” This method was designed to maximize specificity at the possible expense of sensitivity. For the questions on vitamin supplementation, respondents were asked whether they had or had not “used the supplement tablet for at least 6 months per year for most of the years from childhood until your present marriage.” Cases and controls were also asked whether they had been vaccinated against smallpox as a child, as an adult, or both, and if as an adult, at what age (Table 2). They were reminded that the vaccination leaves a “round scar usually on the left shoulder.” They were also asked whether they had Lived in a
Movement Disorders, Vol. 5 , N o . I , 1990
Smallpox vaccination as child only Smallpox vaccination as an adult only Smallpox vaccination as both Never vaccinated against smallpox If vaccinated as adult, age of earliest vaccination (mean 2 SD)
Cases
Controls
56 4 42 I 27.2 (k11.4)
52 3 42 3 31.4 (214.6)
rural area, been exposed to farm animals (excluding dogs and cats), and used well water. Each exposure was defined as “for at least 2 months per year for most of the years from birth until the start of your present marriage.” They were also asked whether they had “sprayed pesticides or insect spray at least once a year for 5 years (not necessarily consecutively) before your present marriage.” Reliability Assay and Sex-Related Bias Test After all cases and controls had been surveyed, we resurveyed the first 12 patients and 12 controls as a test-retest reliability study for the 31 dietary items. All resurveys were performed at least 1 month after the initial survey of that respondent. We calculated the kappa statistic for each questionnaire item as a measure of its test-retest reliability (11). We calculated the odds ratio (OR) and MantelHaenszel x2 for the total group (N = 106) and, to detect sex-related dietary preferences, for the male patient group (N = 71) and for the female patient group (N = 35). OR is calculated by dividing the number of case-control pairs discordant with the case more likely to eat the item by the number of pairs discordant with the control more likely to eat the item. The comparison of male with female controls was based on the total of “much less than average” plus “a little less than average” with the total of “much more than average” plus “a little more than average .” Chi-square was performed on the resulting 2 x 2 table for each food item that gave statistically significant results in the case-control analysis.
RESULTS
Foods with Apparent Protective Effect Table 1 lists the food questionnaire items with OR for each of the three groups of cases. The total group of patients with PD was less likely to have eaten “salad with dressing” than
69
PROTECTIVE AND RISK FACTORS IN PD
were their spouses (OR = 0.6, p < .05). The allmale group gave a similar result (OR = 0.5, p < .05), but male controls did not differ from female controls in their preference for this item. This suggests that sex-related preferences cannot fully explain the findings in the (male-predominant) total group or in the all-male group. The all-female patient group was less likely than their husbands to have eaten “peanuts and peanut butter” (OR = 0.3, p < .05) and “nuts other than peanuts” (OR = 0.3, p < .05). Female controls were as likely as male controls to have eaten “peanuts and peanut butter” (p = .35, x2).That finding, therefore, appears not to be merely sex related. However, female controls were less likely than male controls to have eaten “nuts other than peanuts” (p < .01, x’), a finding compatible with a sex-related explanation for the finding among the female patients. Foods with Apparent Causative Effect
Patients in the total group (N = 106) were more likely than controls to have eaten “spicy foods” (OR = 1.6, p < .05) and this finding was also true of the all-male group (OR = 2.4, p < .05). For this food item there was no sex-related effect among controls. Patients in the all-male group also were more likely than their wives to have eaten “fried foods” (OR = 1.9), “radishes” (OR = 2.0), and “apple juice” (OR = 1.9), p < .05 for each. For none of these food items was there a sex-related effect among controls. None of the other food items gave statistically significant findings for any of the three subject groups. Vitamin Supplements
The vitamin and mineral supplements in the questionnaire were multivitamins, “B vitamins,” C, E, calcium, selenium, and zinc. Fewer than seven patients or controls had used any of the vitamin supplements and fewer than two had used any of the mineral supplements. None of the differences was significant. Other Risk Factors (Tables 2 and 3)
Only one patient and three controls denied having received smallpox vaccine, which was not a significant difference. The percentage of patients who had received the vaccine as adults (x2) and the age
TABLE 3 . Other putative risk factors for Parkinson’s disease
Rural experience Exposure to farm animals Use of well water Sprayed pesticides
Case yes, control no
Case no, control yes
Both yes
Both
no
Odds ratio
30
15
17
43
2.0”
20
15
5
64
1.3
16
14
5
64
1.1
14
2
0
89
7.0“
~~
See text for definition of exposures. Odds ratio > 1 indicates positive (i.e., potentially causal) association with Parkinson’s disease. “ p < .05.
at the first adult vaccination (Student’s unpaired I test) did not differ between the two groups. Exposure to rural living and history of pesticide use were each associated with PD (p < .05). The male/female ratio of cases who had sprayed pesticides (10/4) was representative of the entire group (71/35), eliminating sex bias as the explanation of that potentially occupation-related observation. A sex-related bias was further ruled out by companson of male with female controls. The proportions of persons with rural experience (males: 9 yes, 26 no; females: 23 yes, 48 no) and with a history of pesticide use (males: 0 yes, 35 no; females: 2 yes, 69 no) did not differ between sexes. Exposure to farm animals and well water was not associated with PD among our subjects. Reliability Assay Kappa values from the test-retest procedure for each of the 31 food items ranged from .23 to .82 (Table I). Kappa = 0 indicates no test-retest agreement, .6 considerable agreement, and 1.0 perfect agreement (12). DISCUSSION Our finding a negative association with PD of premorbid consumption of “salad with dressing” and “peanuts and peanut butter” is consistent with the hypothesis that early-life intake of foods high in vitamin E may have prophylactic value against PD. This data supports, in part, that of our previous survey (1). Plant-derived oils (such as those used in salad dressings) and oily plant foods (such as peanuts and other nuts) have high vitamin E content (13). We also confirmed the finding of others that
Movement Disorders, Vol. 5 , No. 1, 1990
70
L . I . GOLBE ET AL.
early-life rural experience and pesticide use are associated with subsequent development of PD (5,6,14). Suspicion was first directed at pesticides and herbicides as etiologic agents by the chemical resemblance of some such compounds to methylphenyl-tetrahydropyridine (MPTP), a known nigral toxin (15). We found a positive association of consumption of spicy foods, fried foods, radishes and apple juice with subsequent development of PD among men. We reduced the possibility that this finding is merely a reflection of sex-related food preferences by (a) requesting that respondents answer the questions by comparing themselves with others of the same sex and (b) comparing male controls with female controls in a search for sex-related preferences. The latter strategem revealed no sex relation for any of these four items. An explanation for the appearance of significant results only for male patients, then, may be an interaction of a causative exogenous agent with an endogenous factor related to male sex. There are several reasons to interpret our results with caution. We used two comparisons for each food item-total group and either the all-male group or the all-female group. This means that at least some of the statistically significant results may be due to chance alone (type I error). Retrospective recall of dietary habits is especially suspect (16). Reliability (the consistency of the respondents’ recall) in our survey was suboptimal for some items, and validity (the degree to which the respondents’ recall reflects reality) could not be measured at all. Although we disqualified subjects with dementia or a possible genetic component, standardized the survey procedure as carefully as possible, and measured the effect of sex bias, we could not eliminate or measure recall bias. For example, it is likely that many patients with PD were aware that an etiologic role of pesticides has been suspected and may have scrutinized their past for such exposure more carefully than had controls. The four food items that were positively associated with development of PD have not been imp& cated in the pathogenesis of the disease to date. Attempts to confirm these epidemiologic results and to seek a potentially nigrotoxic chemical component common to these foods are warranted.
Movement Disorders, Vol. 5 , N o . J , 1990
We conclude that attention should continue to be directed at vitamin E, especially when consumed lifelong, as a potential PD-protective agent. Further studies should attempt to clarify the role of pesticides or other garden- or farm-associated agents in the cause of PD. Further epidemiologic studies are also necessary to identify other rural factors that may explain the rural-PD association. REFERENCES 1. Golbe LI, Farrell TM, Davis PH. Case-control study of early-life dietary factors in Parkinson’s disease. Arch Neurol 1988;45:1350-3. 2. Tanner CM, Cohen JA, Summerville BC, Goetz CG. Vitamin use and Parkinson’s disease. Ann Neurol 1988;23:182. 3. Cadet JL, Klotz A, Jackson-Lewis V , Brin M, Fahn S. Vitamin E or vitamin C treatment attenuate apomorphineinduced rotations in rats after intrastriatal injection of 6hydroxydopamine. Arch Neurol 1988;45:810. 4. Shoulson I. Experimental therapeutics directed at the pathogenesis of Parkinson’s disease. In: Calne D, ed. Handbook of experimental pharmacology: drugs for the treatment of Parkinson’s disease. New York: Springer-Verlag 1989:289305. 5. Tanner CM, Chen B, Wang W-Z, et al. Environmental factors in the etiology of Parkinson’s disease. Can J Neurol Sci 1987;14~419-23. 6. Rajput AH, Uitti RJ, Laverty W. Early onset Parkinson’s disease in Saskatchewan-environmental considerations for etiology. Can J Neurol Sci 1986;13:312-6. 7. Golbe LI, Duvoisin RC, Holowczak JA. Smallpox vaccination and Parkinson’s disease: hypothesis and pilot study. Arch Neurol (in press). 8 . Spencer PS, Nunn PB, Hugon J, et al. Guam amyotrophic lateral sclerosis-parkinsonism-dementia linked to a plant excitant neurotoxin. Science 1987;237:517-22. 9. Golbe LI, Pae J. Validity of a mailed epidemiologic survey and physical self-assessment in Parkinson’s disease. Movement Disorders 1988;3:245-54. 10. Golbe LI, Duvoisin RC, Miller DC. A large kindred with autosomal-dominant Parkinson’s disease. Ann Neurol 1988;24:151-2. 11. Fleiss JL. Statistical methods for rates and proportions. New York John Wiley, 1973:143-7. 12. Heyman A, Wilkinson WF, Stafford JA, et al. Alzheimer’s disease: a study of epidemiological aspects. Ann Neurol 1 984 ;15:335-4 1. 13. Machlin LJ. Vitamin E. In: Machlin LJ, ed. Handbook of vitamins: nutritional, biochemical and clinical aspects. New York: Marcel Dekker, 1984. 14. Tanner CM. Influence of environmental factors on the onset of Parkinson’s disease. Neurology 1986;36(suppl 1):215. 15. Langston JW, Ballard P, Tetrud JW, Irwin I. Chronic parkinsonism in humans due to a product of mependine-analog synthesis. Science 1983;219:979-80. 16. Marshall J, Priore R, Haughey B, Rzepka T, Graham S . Spouse-subject interviews and the reliability of diet studies. Am J Epidemiol 1980;112:675-83.
Follow-up Study of Early-Life Protective and Risk Factors in Parkinson’s Disease Lawrence I. Golbe, Timothy M. Farrell, and “Patricia H. Davis Department of Neurology, University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, New Brunswick, New Jersey, and *Department of Neurology, Louisiana State University Medical School, Shreveport, Louisiana, U.S.A.
Summary: Previous studies suggest that Parkinson’s disease (PD) is negatively associated with early-life intake of vitamin E-rich foods and positively associated with rural experience. Using a new survey design, we attempted to confirm and extend these results. We gave a telephone questionnaire to 106 patients with PD and to their spouses as controls. It assessed premarital consumption of 31 foods of various vitamin E content, vitamin supplements, and exposure to rural living. Respondents rated food consumption with respect to what they perceived as the average for their sex and age at that time. We found female patients with PD less likely than spouses to have eaten “peanuts and peanut butter” (p < .05), which are high in vitamin E . “Salad with dressing,” also high in vitamin E, gave a similar result (p < .05) for a male-predominant patient group. Separate comparison of male controls with female controls ruled out sex-related preferences as the explanation of our findings. Patients had more extensive rural experience and were more likely to have frequently sprayed pesticides (p < .05) than had controls. Our results justify further investigations into early-life vitamin E intake, pesticides, and neurotoxins associated with rural life. Key Words: Parkinson’s disease-Diet-Vitamin ERural-Case control-Risk factors.
Therefore, we attempted to 1. confirm our earlier results using a different survey design; 2. extend those results by inquiring in more detail into specific vitamin E-rich foods and foods commonly eaten with them; 3. investigate a relationship between PD and early-life use of vitamin E and other oral antioxidant supplements; 4. confirm in a PD population of typical-age onset the greater rural experience that has been reported among persons with young-onset PD; 5. explore specific factors suspected of explaining that rural predilection; 6. investigate the epidemiological relationship between PD and early-life smallpox vaccination (7); and 7. repeat and extend our inquiry into foods that, by analogy with the cycad suspected of causing the amyotrophic lateral sclerosis-PD-dementia com-
In a case-control study of dietary risk factors in Parkinson’s disease (PD) we found that patients with PD were less likely than controls to have consumed foods rich in vitamin E during early life (1). Similarly, Tanner et al. (2) found early-life use of vitamin E supplements to be more common among controls than among those with PD. Experimental data in animal models of PD support the notion of a protective role of vitamin E, presumably via an antioxidative mechanism (3). A large double-blind study of the role of vitamin E in slowing the progression of early PD is in progress (4). Association of rural living with young-onset PD has been reported by Tanner et al. (5) and by Rajput et al. (6) using different survey designs. Address correspondence and reprint requests to Dr. L. I. Golbe at Department of Neurology, CN-19, UMDNJ-Robert Wood Johnson Medical School, New Brunswick, NJ 08903, U.S.A.
66
PROTECTIVE AND RISK FACTORS IN P D plex of Guam (8), may be positively associated with subsequent development of PD. SUBJECTS AND METHODS Rationale for Survey Technique Retrospective quantitative assessment of dietary habits is notoriously unreliable. Our previous survey (1) attempted to avoid this difficulty by assessing patients’ early-adult dietary preferences among patients and same-sex sibling controls relative to those of their respective spouses at that time. Although we pretested this technique for reliability, its validity as a solution to the problem of obtaining retrospective dietary data remains unsupported. Therefore, the present survey technique was designed along more orthodox lines, with spouses serving as controls. Again, however, we did not ask respondents to rate food consumption on an absolute scale. Rather, we asked them to compare their consumption of each food item with the level of consumption that they perceived a s average for persons of their age and sex at that time. Questions were expressed in terms of premarital exposure to minimize the “overmatching” that use of spouse controls may produce. Inasmuch as the cause of PD is hypothesized to act during childhood or early adult life, the date of marriage seemed a scientifically sensible boundary as well as one that would be relatively easy to recall in multiple-decade retrospect. The survey was conducted by telephone, because mail surveys in PD have been found insufficiently reliable for research purposes (9). Subject Qualifications Patients qualified if they had typical idiopathic PD, had been seen between January 1986 and July 1988 at the Movement Disorder Center at UMDNJRobert Wood Johnson Medical School, were not demented (per the medical record), were married before PD onset, and had no first-degree relatives with a parkinsonian disorder. Such PD kindreds (10) are the result more of genetic than environmental factors and therefore would confound our present inquiry into environmental factors. Spouse controls were excluded as demented if they could not give their zip code and the patient’s date of birth. These data were checked against the patient’s medical record.
67
Reduction of Sex-Related Artifact Because some dietary habits may be sex-related and we used spouses as controls, there was a risk that the male predominance in PD would produce artifactual dietary “predilections.” Our stipulation that the activity or dietary intake be compared with others “of your age and sex at that time” sought to eliminate this artifact. Nevertheless, we further reduced the sex-related artifact by performing the analysis for three groups. The first was the entire group of 106 patients (71 men and 35 women, with spouses as controls). The second consisted of only the 71 male patients (with wives as controls) and the third consisted of the 35 female patients (with husbands as controls). As additional vigilance against sex-related artifact, we tabulated the preferences of male and female controls, respectively, for the food items that gave statistically significant differences between cases and controls. In this way, we sought a m a l e female difference that might explain the casecontrol difference. Only 28 of the 106 patients (26%, 18 men, 10 women) had been respondents in our previous survey (1). Questionnaire Administration One trained and experienced questioner (T.M.F.) administered all questionnaires using a standardized format. Case-control pairs were surveyed only when both were home together. Spouse controls were questioned first and were asked not to repeat questions aloud lest the patient offer influence or formulate his or her own answers in advance. After the spouse completed the survey, he or she handed the telephone to the patient, who was instructed not to solicit advice from the spouse during questioning. Data Format and Analysis Cases and controls were asked whether their consumption of 31 foods was “much less than average,” “a little less than average,” “about average,” “a little more than average,” or “much more than average” for persons of their age and sex during the years between age 18 and the year of the present marriage. The food items are listed in Table 1 . Case-control pairs were considered “discordant” if their answers on this five-point scale differed by at least two levels. For example, if for a given food item the patient responded “much less
Movemenl Disorders, Vol. 5 , No. I , 1990
L . I . GOLBE ET AL. TABLE 1. Odds ratios for 31 food items Sex of cases
71M 35F
71M OF
OM 35F
(controls were their spouses) Salad with dressing Salad without dreGing Spicy foods Fried foods Raw tomatoes Cooked tomatoes or tomato sauce Tomato juice Catsup Raw onions Cooked onions Lettuce Spinach Cucumbers Green peppers Radishes Olives Olive oil Salad oil or dressing Peanuts or peanut butter Nuts other than peanuts Walnuts Cashews Almonds Apples Apple juice Bananas Peaches Plums or prunes Grapes Strawberries Other bemes
TABLE 2. Smallpox vaccination as a Parkinson’s disease risk factor (all differences not significant)
kappa 0.6“ 0.6 1.6“ 1.3 0.8
0.5“ 0.7 2.4“ I .9“ 0.8
0.9 0.6 0.8 0.6 0.7
.33 .46 .48 .73 .38
1.3 0.9 0.9
1.6 1.2 1.2 1.5 0.8 0.7 1.1 0.7 0.7 2.0“ I .2 1.4 0.7 1.1 1.5 1.4 1.4 1.3 1.7 1.9” 1.8
0.7 0.5 0.7 0.4 1.1 0.9 0.6 0.5
.45 .52 .82 .36 .31 .39 .42 .45 .36 .48 .61 .46 .28 .40 .53 .23 .43 .48 .39 .48 .24 .29 .30 .24 .36 .54
1.o
0.9 0.8 0.9 0.6 0.8 1.5 0.9 1.3 0.8 0.7 0.8 1.O
.o
1
I .o 1.5 1.3 1.1 0.9 0.9 1 .o 0.9 1 .o
1.o
1.1 1.3 1.2 1.2
1.o 0.9
0.5 1.1 0.5 0.3“ 0.3“ 0.6 0.5 0.5 1.2 0.6 0.4 0.8 0.7 0.5 0.6 0.6
Three separate analyses were made to examine the effect of sex on food preferences. Odds ratios (OR) and kappa are defined in the text. For OR < 1, the cases were less likely than controls to have eaten the item. a p < .05 (Mantel-Haenszel x ‘).
than average” and the spouse the same or “a little less than average,” the pair was considered concordant. If that spouse had answered “about average,” the pair would have been scored as “patient < spouse.” This method was designed to maximize specificity at the possible expense of sensitivity. For the questions on vitamin supplementation, respondents were asked whether they had or had not “used the supplement tablet for at least 6 months per year for most of the years from childhood until your present marriage.” Cases and controls were also asked whether they had been vaccinated against smallpox as a child, as an adult, or both, and if as an adult, at what age (Table 2). They were reminded that the vaccination leaves a “round scar usually on the left shoulder.” They were also asked whether they had Lived in a
Movement Disorders, Vol. 5 , N o . I , 1990
Smallpox vaccination as child only Smallpox vaccination as an adult only Smallpox vaccination as both Never vaccinated against smallpox If vaccinated as adult, age of earliest vaccination (mean 2 SD)
Cases
Controls
56 4 42 I 27.2 (k11.4)
52 3 42 3 31.4 (214.6)
rural area, been exposed to farm animals (excluding dogs and cats), and used well water. Each exposure was defined as “for at least 2 months per year for most of the years from birth until the start of your present marriage.” They were also asked whether they had “sprayed pesticides or insect spray at least once a year for 5 years (not necessarily consecutively) before your present marriage.” Reliability Assay and Sex-Related Bias Test After all cases and controls had been surveyed, we resurveyed the first 12 patients and 12 controls as a test-retest reliability study for the 31 dietary items. All resurveys were performed at least 1 month after the initial survey of that respondent. We calculated the kappa statistic for each questionnaire item as a measure of its test-retest reliability (11). We calculated the odds ratio (OR) and MantelHaenszel x2 for the total group (N = 106) and, to detect sex-related dietary preferences, for the male patient group (N = 71) and for the female patient group (N = 35). OR is calculated by dividing the number of case-control pairs discordant with the case more likely to eat the item by the number of pairs discordant with the control more likely to eat the item. The comparison of male with female controls was based on the total of “much less than average” plus “a little less than average” with the total of “much more than average” plus “a little more than average .” Chi-square was performed on the resulting 2 x 2 table for each food item that gave statistically significant results in the case-control analysis.
RESULTS
Foods with Apparent Protective Effect Table 1 lists the food questionnaire items with OR for each of the three groups of cases. The total group of patients with PD was less likely to have eaten “salad with dressing” than
69
PROTECTIVE AND RISK FACTORS IN PD
were their spouses (OR = 0.6, p < .05). The allmale group gave a similar result (OR = 0.5, p < .05), but male controls did not differ from female controls in their preference for this item. This suggests that sex-related preferences cannot fully explain the findings in the (male-predominant) total group or in the all-male group. The all-female patient group was less likely than their husbands to have eaten “peanuts and peanut butter” (OR = 0.3, p < .05) and “nuts other than peanuts” (OR = 0.3, p < .05). Female controls were as likely as male controls to have eaten “peanuts and peanut butter” (p = .35, x2).That finding, therefore, appears not to be merely sex related. However, female controls were less likely than male controls to have eaten “nuts other than peanuts” (p < .01, x’), a finding compatible with a sex-related explanation for the finding among the female patients. Foods with Apparent Causative Effect
Patients in the total group (N = 106) were more likely than controls to have eaten “spicy foods” (OR = 1.6, p < .05) and this finding was also true of the all-male group (OR = 2.4, p < .05). For this food item there was no sex-related effect among controls. Patients in the all-male group also were more likely than their wives to have eaten “fried foods” (OR = 1.9), “radishes” (OR = 2.0), and “apple juice” (OR = 1.9), p < .05 for each. For none of these food items was there a sex-related effect among controls. None of the other food items gave statistically significant findings for any of the three subject groups. Vitamin Supplements
The vitamin and mineral supplements in the questionnaire were multivitamins, “B vitamins,” C, E, calcium, selenium, and zinc. Fewer than seven patients or controls had used any of the vitamin supplements and fewer than two had used any of the mineral supplements. None of the differences was significant. Other Risk Factors (Tables 2 and 3)
Only one patient and three controls denied having received smallpox vaccine, which was not a significant difference. The percentage of patients who had received the vaccine as adults (x2) and the age
TABLE 3 . Other putative risk factors for Parkinson’s disease
Rural experience Exposure to farm animals Use of well water Sprayed pesticides
Case yes, control no
Case no, control yes
Both yes
Both
no
Odds ratio
30
15
17
43
2.0”
20
15
5
64
1.3
16
14
5
64
1.1
14
2
0
89
7.0“
~~
See text for definition of exposures. Odds ratio > 1 indicates positive (i.e., potentially causal) association with Parkinson’s disease. “ p < .05.
at the first adult vaccination (Student’s unpaired I test) did not differ between the two groups. Exposure to rural living and history of pesticide use were each associated with PD (p < .05). The male/female ratio of cases who had sprayed pesticides (10/4) was representative of the entire group (71/35), eliminating sex bias as the explanation of that potentially occupation-related observation. A sex-related bias was further ruled out by companson of male with female controls. The proportions of persons with rural experience (males: 9 yes, 26 no; females: 23 yes, 48 no) and with a history of pesticide use (males: 0 yes, 35 no; females: 2 yes, 69 no) did not differ between sexes. Exposure to farm animals and well water was not associated with PD among our subjects. Reliability Assay Kappa values from the test-retest procedure for each of the 31 food items ranged from .23 to .82 (Table I). Kappa = 0 indicates no test-retest agreement, .6 considerable agreement, and 1.0 perfect agreement (12). DISCUSSION Our finding a negative association with PD of premorbid consumption of “salad with dressing” and “peanuts and peanut butter” is consistent with the hypothesis that early-life intake of foods high in vitamin E may have prophylactic value against PD. This data supports, in part, that of our previous survey (1). Plant-derived oils (such as those used in salad dressings) and oily plant foods (such as peanuts and other nuts) have high vitamin E content (13). We also confirmed the finding of others that
Movement Disorders, Vol. 5 , No. 1, 1990
70
L . I . GOLBE ET AL.
early-life rural experience and pesticide use are associated with subsequent development of PD (5,6,14). Suspicion was first directed at pesticides and herbicides as etiologic agents by the chemical resemblance of some such compounds to methylphenyl-tetrahydropyridine (MPTP), a known nigral toxin (15). We found a positive association of consumption of spicy foods, fried foods, radishes and apple juice with subsequent development of PD among men. We reduced the possibility that this finding is merely a reflection of sex-related food preferences by (a) requesting that respondents answer the questions by comparing themselves with others of the same sex and (b) comparing male controls with female controls in a search for sex-related preferences. The latter strategem revealed no sex relation for any of these four items. An explanation for the appearance of significant results only for male patients, then, may be an interaction of a causative exogenous agent with an endogenous factor related to male sex. There are several reasons to interpret our results with caution. We used two comparisons for each food item-total group and either the all-male group or the all-female group. This means that at least some of the statistically significant results may be due to chance alone (type I error). Retrospective recall of dietary habits is especially suspect (16). Reliability (the consistency of the respondents’ recall) in our survey was suboptimal for some items, and validity (the degree to which the respondents’ recall reflects reality) could not be measured at all. Although we disqualified subjects with dementia or a possible genetic component, standardized the survey procedure as carefully as possible, and measured the effect of sex bias, we could not eliminate or measure recall bias. For example, it is likely that many patients with PD were aware that an etiologic role of pesticides has been suspected and may have scrutinized their past for such exposure more carefully than had controls. The four food items that were positively associated with development of PD have not been imp& cated in the pathogenesis of the disease to date. Attempts to confirm these epidemiologic results and to seek a potentially nigrotoxic chemical component common to these foods are warranted.
Movement Disorders, Vol. 5 , N o . J , 1990
We conclude that attention should continue to be directed at vitamin E, especially when consumed lifelong, as a potential PD-protective agent. Further studies should attempt to clarify the role of pesticides or other garden- or farm-associated agents in the cause of PD. Further epidemiologic studies are also necessary to identify other rural factors that may explain the rural-PD association. REFERENCES 1. Golbe LI, Farrell TM, Davis PH. Case-control study of early-life dietary factors in Parkinson’s disease. Arch Neurol 1988;45:1350-3. 2. Tanner CM, Cohen JA, Summerville BC, Goetz CG. Vitamin use and Parkinson’s disease. Ann Neurol 1988;23:182. 3. Cadet JL, Klotz A, Jackson-Lewis V , Brin M, Fahn S. Vitamin E or vitamin C treatment attenuate apomorphineinduced rotations in rats after intrastriatal injection of 6hydroxydopamine. Arch Neurol 1988;45:810. 4. Shoulson I. Experimental therapeutics directed at the pathogenesis of Parkinson’s disease. In: Calne D, ed. Handbook of experimental pharmacology: drugs for the treatment of Parkinson’s disease. New York: Springer-Verlag 1989:289305. 5. Tanner CM, Chen B, Wang W-Z, et al. Environmental factors in the etiology of Parkinson’s disease. Can J Neurol Sci 1987;14~419-23. 6. Rajput AH, Uitti RJ, Laverty W. Early onset Parkinson’s disease in Saskatchewan-environmental considerations for etiology. Can J Neurol Sci 1986;13:312-6. 7. Golbe LI, Duvoisin RC, Holowczak JA. Smallpox vaccination and Parkinson’s disease: hypothesis and pilot study. Arch Neurol (in press). 8 . Spencer PS, Nunn PB, Hugon J, et al. Guam amyotrophic lateral sclerosis-parkinsonism-dementia linked to a plant excitant neurotoxin. Science 1987;237:517-22. 9. Golbe LI, Pae J. Validity of a mailed epidemiologic survey and physical self-assessment in Parkinson’s disease. Movement Disorders 1988;3:245-54. 10. Golbe LI, Duvoisin RC, Miller DC. A large kindred with autosomal-dominant Parkinson’s disease. Ann Neurol 1988;24:151-2. 11. Fleiss JL. Statistical methods for rates and proportions. New York John Wiley, 1973:143-7. 12. Heyman A, Wilkinson WF, Stafford JA, et al. Alzheimer’s disease: a study of epidemiological aspects. Ann Neurol 1 984 ;15:335-4 1. 13. Machlin LJ. Vitamin E. In: Machlin LJ, ed. Handbook of vitamins: nutritional, biochemical and clinical aspects. New York: Marcel Dekker, 1984. 14. Tanner CM. Influence of environmental factors on the onset of Parkinson’s disease. Neurology 1986;36(suppl 1):215. 15. Langston JW, Ballard P, Tetrud JW, Irwin I. Chronic parkinsonism in humans due to a product of mependine-analog synthesis. Science 1983;219:979-80. 16. Marshall J, Priore R, Haughey B, Rzepka T, Graham S . Spouse-subject interviews and the reliability of diet studies. Am J Epidemiol 1980;112:675-83.
