Research and Reviews: Dairy

Special Circular 163-99

Food Frequency Questionnaires as a Measure of Exposure to Organochlorines in Epidemiological Studies

A. Wicklund Glynn1
1Toxicology Division, The Swedish National Food Administration, Uppsala, Sweden
Department of Environmental Toxicology, Uppsala University, Uppsala, Sweden

L. B. Willett
2The Ohio State University
Department of Animal Sciences

2 For more information, contact at: The Ohio State University, Ohio Agricultural Research and Development Center, 128 Gerlaugh Hall, 1680 Madison Avenue, Wooster, OH 44691; 330-263-3792; willett.2@osu.edu

Abstract

Introduction

In the United States and many other industrialized countries, food is currently the major source of human exposure to PCB and chlorinated pesticides, such as DDT. The severe restrictions in the use and production of these chemicals for more than two decades have minimized the risk of direct occupational exposure. Moreover, since the chemicals are not available for the consumer, direct exposure during use does not happen any more. As a result, the exposure of humans to organochlorines has decreased significantly since the early 1970s (Ahlborg et al., 1995; Noren, 1993). However, in large areas of the world, the general population is still exposed to elevated amounts of organochlorines in the food because of the persistence of the compounds in the environment.

In epidemiological studies of the effects of organochlorines on human health, it is of vital importance to perform good and reliable exposure estimations. Analysis of the compounds in human tissues is probably the most accurate measure of exposure. However, this approach is both time-consuming and expensive. As an example, when doing PCB and dioxin analysis, it is usually only possible to analyze a few samples per day, and the analysis may cost more than $1,000 per sample. Moreover, it is sometimes difficult to get people to participate in studies that include blood sampling or fat biopsies.

Thus, in order to avoid analysis of biological tissues, several studies have based their exposure assessment on food consumption questionnaires (Dar et al., 1992; Fein et al., 1984; Lonky et al., 1996; Mendola et al., 1995; Rylander and Hagmar, 1995). The rationale behind this is that the rate of intake of contaminated food is a good predictor of the amounts of organochlorines in the body. In this review, we highlight some of the problems with the use of food frequency questionnaires as a tool for organochlorine exposure assessment.

Food Frequency Questionnaires

Based on the findings of high concentrations of organochlorines (mainly PCB) in certain fish species from the Great Lakes area in the United States and along the Baltic Sea coast of Sweden, several studies have estimated the exposure of the study participants from their rate of fish consumption (Dar et al., 1992; Hagmar et al., 1992; Jacobson and Jacobson, 1997; Lonky et al., 1996, Mendola et al., 1995; Rylander and Hagmar, 1995). In the U.S. studies, detailed information about Great Lakes fish consumption rates was obtained from the participating individuals. In order to account for the highly variable concentrations of PCB between fish species, a weighted value was assigned to each fish species consumed, based on the average PCB level reported for that species (Dar et al., 1992; Fein et al., 1984; Lonky et al., 1996; Mendola et al., 1995). The weighted sum of contaminated fish consumption was then used as the exposure estimate in the final study of associations between PCB exposure and health effects.

In Sweden, another approach has been used. The researchers identified a population (fisherman families) with a high consumption of organochlorine-contaminated fish from the east coast of Sweden (the Baltic Sea) (Hagmar et al., 1992; Rylander and Hagmar, 1995; Svensson et al., 1995a). Fishermen families from the west coast of Sweden (the Atlantic Ocean) with a similar socioeconomic status as the Baltic Sea fishermen were identified as a group with lower exposures to organochlorines. This is due to a lower level of the compounds in west coast fish than in east coast fish (Atuma et al., 1996).

Food questionnaires were used in order to confirm that these family groups had a high consumption of fish in comparison to the general population in Sweden. It was also confirmed that fishermen from the Baltic Sea coast had higher organochlorine concentrations in their blood than the fishermen living on the Atlantic coast of Sweden (Svensson et al., 1995b).

In the final analysis of the associations between PCB exposure and health effects, the incidence of disease in the Baltic Sea population was compared with that of the Atlantic coast population or the general population of Sweden. In these comparisons, the assumption was made that the organochlorine exposure was highest among the Baltic Sea population, as predicted by the high rate of contaminated fish consumption in this family group.

A major problem with these studies was that the actual concentrations of organochlorines in the bodies of the study participants were not known. The rate of contaminated food consumption will only be a valid measure of exposure if it is a good predictor of the actual organochlorine concentration in the bodies of the study participants. However, in several of the studies no effort was made to confirm this by studying the correlation between consumption of contaminated food and the body burden of organochlorines.

Questionnaires - Concentrations in Biological Samples

In cases of a very high consumption rate of contaminated food among the studied subjects (several meals per week - daily consumption), as in the case of fishermen, the consumption rate is a fairly good predictor of the actual organo-chlorine concentrations in the body (Johansen et al., 1996; Svensson et al., 1991). This is due to a high contribution of contaminated foods to the exposure of these consumers. However, in the general population, extreme consumption rates of contaminated foodstuffs are rare. In this case, highly contaminated foodstuffs usually give a low contribution to the exposure of the individuals. It may, consequently, not be feasible to use the rate of contaminated food consumption as a measure of exposure in epidemiological studies on the general population.

In a recent study, the correlation between blood concentrations and intake of organochlorines from food was studied in Swedish men from the general population (Table 1) (Wicklund Glynn et al., 1997). The chemicals included in the study were a group of industrial chemicals called polychlorinated biphenyls (PCB), the fungicide hexachlorobenzene (HCB), and the insecticides DDT (p,p=-DDT and p,p=-DDE), hexachlorocyclohexane (HCH), and chlordane (transnonachlor and oxychlordane). These chemicals have not been used in Sweden for almost two decades, but they are still present in the Swedish environment and food. The intake of organochlorines was estimated from a detailed food frequency questionnaire, covering foods contributing most to the intake of the pollutants (dairy products, meat products, eggs, and fish). The participants in the study (n=81, 40 to 75 years of age) were asked about both their current and past consumption frequencies to account for possible changes in food habits during the exposure period. The intake was calculated using data on organochlorine concentrations in food obtained from the monitoring program for organochlorines performed by the Swedish National Food Administration.

The intake of organochlorines, as estimated from the questionnaire, was in all cases positively correlated to the concentrations of organo-chlorines in blood serum (Table 1). Hence, the concentration of organochlorines in serum of the men increased with increasing intake of the substances. However, the association was not strong since the correlation coefficients (r) were generally less than 0.3. The correlation did not improve markedly if intake from fish was used in the correlation analysis instead of total intake from animal foods in general. Moreover, there were no significant differences in correlation coefficients if the current intake or the intake 20 years ago was used (Table 1). Age and body-mass index adjustment of the correlation or exclusion of individuals with suspected history of occupational exposure did not improve the results (Wicklund Glynn et al., 1997).

Thus, in this study, the intake of organochlorines, as calculated from a food frequency questionnaire, was not a good predictor of the organochlorine concentrations in the body of men from the general population (Wicklund Glynn et al., 1997). Several factors contribute to the discrepancy between the calculated intake and the serum concentrations of organochlorines. It may be difficult to give good estimates of the real consumption rates of contaminated foods when answering a food frequency questionnaire. Moreover, the individual variation in absorption, metabolism, and excretion of the substances among the study subjects adds to the uncertainty between intake estimate and serum concentration.

Similarly, in a study of possible health effects in infants and children born to mothers with high consumption of PCB-contaminated Lake Michigan fish, it was found that the rate of maternal fish consumption did not predict cord serum PCB concentrations in the studied newborns (Fein et al., 1984). Consequently, in the follow-up studies, the statistical analysis of the results were mainly based on the cord serum PCB measurements (Jacobson and Jacobson, 1997). Fiore et al. (1989) concluded that "Great Lakes fish consumption" is a relatively nonspecific indicator of PCB exposure. In this case, the researchers found that the reported consumption rate of Great Lakes fish among Wisconsin anglers explained only 10 to 15% of the variation seen in serum PCB concentration (Fiore et al., 1989).

From these studies, it is evident that food frequency questionnaires are not a good tool for exposure assessment of organochlorines in studies on the general human population. Another problem with exposure measurements using food frequency questionnaires is that food contains many compounds with the potential to affect human health. It is thus difficult to link an observed health effect to a specific substance. For instance, in several of the Great Lakes studies, the consumption of Great Lakes fish was used as an estimate of PCB exposure (Dar et al., 1992; Fein et al., 1984; Lonky et al., 1996; Mendola et al., 1995). However, it is well known that Great Lakes fish contain a complex mixture of environmental pollutants. Consequently, in these cases, it can not with certainty be concluded that PCB is the cause of an observed association between the consumption rate of Great Lakes fish and a health effect. This is further illustrated by the association between blood concentration of different organochlorines and the intake of the most abundant and persistent PCB compound CB 153 by Swedish men (Table 2). In this case, the intake of CB 153 predicted the serum concentrations of oxychlordane, transnonachlor, and p,p=-DDT to the same degree as the serum concentration of CB 153.

Biological Samples

It may thus be concluded that, in epidemiological studies on the general human population, analysis of organochlorines in biological samples is necessary for an accurate exposure assessment. However, even this approach may cause problems, since the concentrations of different organochlorines in the body are often related to each other.

For instance, in the serum of the Swedish men a high correlation was found between the concentrations of oxychlordane and transnonachlor (r=0.85) and CB 153 and p,p=-DDE (r=0.79) (Table 3). A correlation coefficient close to 0.5 or higher was found in all cases. Furthermore, in a study of fishermen from the Baltic Sea coast it was found that the blood plasma concentration of the highly toxic PCB compound, CB 126, was strongly correlated to the concentration of the dominating dioxin-like compound pentachloro-dibenzofuran (r=0.78) (Skerfving et al., 1994).

Consequently, there will always be some degree of uncertainty left in an association between the exposure to a certain compound and an observed health effect, since all environmental pollutants can not be measured in an epidemiological study. A causal association between an observed health effect and the exposure to an environmental pollutant can only be settled with reasonable certainty by the combined results from experimental and epidemiological studies of high quality, performed in different parts of the world. It is of great importance for future epidemiological studies that efforts are made to develop inexpensive, fast, and noninvasive analytical techniques for analysis of chemicals in the human body.

References

Ahlborg, U. G., L. Lipworth, L. Titus-Ernstoff, C.-C. Hsieh, A. Hanberg, J. Baron, D. Trich-opoulus, and H.-O. Adami. 1995. Organo-chlorine compounds in relation to breast cancer, endometrial cancer, and endometriosis: An assessment of the biological and epidemiological evidence. Crit. Rev. Toxicol. 25:463.

Atuma, S. S., C.-E. Linder, A. Wicklund Glynn, Ö. Andersson, and L. Larsson. 1996. Survey of consumption of fish from Swedish waters for chlorinated pesticides and polychlorinated biphenyls. Chemosphere 33: 791.

Dar, E., M. S. Kanarek, H. E. Anderson, and W. C. Sonzogni. 1992. Fish consumption and reproductive outcomes in Green Bay, Wisconsin. Environ. Res. 59:189.

Fein, G. G., J. L. Jacobson, S. W. Jacobson, P. M. Schwartz, and J. K. Dowler. 1984. Prenatal exposure to polychlorinated biphenyls: Effects on birth size and gestational age. J. Pedriatr. 105:315.

Fiore, B. J., H. A. Anderson, L. P. Hanrahan, L. J. Olson, and W. C. Sonzogni. 1989. Sport fish consumption and body burden levels of chlorinated hydrocarbons: A study of Wisconsin anglers. Arch. Environ. Health 44:82.

Hagmar, L., K. Lindén, A. Nilsson, B. Norrving, B. Åkesson, A. Schytz, and T. Möller. 1992. Cancer incidence and mortality among Swedish Baltic Sea fishermen. Scand. J. Work Environ. Health 18:217.

Jacobson, J. L. and S. W. Jacobson. 1997. Evidence for PCBs as neurodevelopmental toxicants in humans. Neurotoxicol. 2:415.

Johansen, H. R., J. Alexander, O. J. Rossland, S. Planting, M. Lövik, P. I. Gaarder, W. Gdynia, K. S. Bjerve, and G. Becher. 1996. PCDDs, PCDFs, and PCBs in human blood in relation to consumption of crabs from a contaminated fjord area in Norway. Environ. Health Perspect. 104:756.

Lonky, E., J. Reihman, T. Darville, J. Mather Sr., and H. Daly. 1996. Neonatal behavioral assessment scale performance in humans influenced by maternal consumption of environmentally contaminated Lake Ontario fish. J. Great Lakes Res. 22:198.

Mendola, P., G. M. Buck, J. E. Vena, M. Zielenzy, and L. E. Sever. 1995. Consumption of PCB-contaminated sport fish and risk of spontaneous fetal death. Environ. Health Perspect. 103:498.

Norén, K. 1993. Contemporary and retrospective investigations of human milk in the trend studies of organochlorine contaminants in Sweden. Sci. Total Environ. 139/140:347.

Rylander, L. and L. Hagmar. 1995. Mortality and cancer incidence among women with a high consumption of fatty fish contaminated with persistent organochlorine compounds. Scand. J. Work Environ. Health 21:419.

Skerfving, S., B.-G. Svensson, L. Asplund, and L. Hagmar. 1994. Exposure to mixtures and congeners of polychlorinated biphenyls. Clin. Chem. 40:1409.

Svensson, B.-G., A. Nilsson, M. Hansson, C. Rappe, B. Åkesson, and S. Skerfving. 1991. Exposure to dioxins and dibenzofurans through the consumption of fish. New Engl. J. Med. 324:8.

Svensson, B.-G., Z. Mikoczy, U. Strömberg, and L. Hagmar. 1995a. Mortality and cancer incidence among Swedish fishermen with a high dietary intake of persistent organochlorine compounds. Scand. J. Work Envirn. Health 21:106.

Svensson, B.-G., A. Nilsson, E. Jonsson, A. Schytz, B. Åkesson, and L. Hagmar. 1995b. Fish consumption and exposure to persistent organochlorine compounds, mercury, selenium, and methylamines in Swedish fishermen. Scand. J. Work Environ. Health 21:96.

Wicklund Glynn, A., S. Atuma, M. Aune, P. O. Darnerud, A. Wolk, B. Lundkvist, and H.-O. Adami. 1997. Human exposure to organochlorines. Food frequency questionnaires as a tool for exposure assessment in the general population. Report to the Swedish EPA. The Swedish National Food Administration, Uppsala, Sweden.