Southern Cooking and Lung Cancer

Abstract: Dietary associations were examined as part of a case-control study exploring reasons for exceptionally high rates of lung cancer in northeast Florida. Interviews, which included a nationally standardized food frequency questionnaire, were conducted with 507 patients diagnosed with lung cancer during 1993-1996 or their next of kin and 1,007 persons of similar age, race, and gender randomly selected from the general population. A substantial reduction in risk was associated with high consumption of nutrient-dense fruits and vegetables. Risk was nearly doubled among men and women in the highest quartile of fat intake. The effects were most prominent for saturated and monounsaturated fats and not apparent for polyunsaturated fat consumption. Increased risk was linked to consumption of several individual high-fat foods, including some traditional Southern foods or methods of cooking, such as cooking vegetables with lard/fatback/bacon fat. Reported use of vitamin/mineral supplements was associated with decreased risk of lung cancer as well as dietary consumption of vitamins A, C, and E and some carotenoids. The findings are consistent with emerging evidence that risk of lung cancer rises with increasing dietary fat consumption. They indicate the need for further research to determine whether the association between fat intake and lung cancer is causal and, if so, to clarify the relationships with individual fat fractions.

Lung cancer accounted for 160,000 deaths in the United States in 1997, second only to heart disease as the leading cause of death ( 1). Age-adjusted rates of lung cancer mortality in Jacksonville (Duval County), FL, are among the highest in the country, ranking in the top 2% of all state economic areas.[ 1] Studies by the National Cancer Institute in the 1970s indicated that occupational factors, especially work in shipbuilding and associated asbestos exposure during World War II, contributed to part of the excess rates among men during the 1960s and 1970s ( 2). Since then, rates of lung cancer have become elevated among women as well as men. Our studies in the 1990s, commissioned by the City of Jacksonville, showed that excessive cigarette smoking intensity, more than prevalence, provides further explanation of the area's high lung cancer mortality rates ( 3, 4).

Diet is another element of the lifestyle now hypothesized as a risk factor for cancer, including lung cancer ( 5, 6). The South has a long-recognized history of consumption of fat at levels higher than the rest of the United States that has been documented by studies such as the 1977-78 Nationwide Food Consumption Survey. The reported mean consumption of the fat/oil food group in grams per day per person of 45.6 in the South was substantially higher than the total US mean of 36.3 ( 7, 8). As part of a large case-control study of lung cancer risk factors in Jacksonville, FL, we compared dietary habits of cases and controls to provide clues to dietary determinants of risk. This report details the results of the diet portion of this study.

A population-based case-control study was conducted in which patients newly diagnosed with lung cancer and controls of similar age, race, and gender randomly selected from the population of Duval County, FL, were enrolled. All county residents aged 30-84 years newly diagnosed with primary lung cancer (International Classification of Diseases, 9th revision, Codes 162 and 231.0-231.2) between 1 September 1993 and 29 February 1996 comprised the potential case group. Potential cases were reported to the study team by the tumor registries at every hospital in Duval County at the time of identification of a diagnosis to be reported to the Florida Cancer Data System. Medical record review was performed by the study epidemiologist on 829 cases. Histological confirmation by the study pathologist found 749 cases eligible for study. Interviews were completed for 507 of these or 68% of the potential cases, with nonresponse nearly equally divided between refusal and loss to follow-up. Nonresponders did not differ significantly from responders by zipcode, age, race, or gender.

Controls were selected from the county's population of households with telephones; they were screened to obtain controls with the same expected distribution as cases with respect to age, gender, and race. A random-digit-dialing procedure was used to generate >6,600 phone numbers, which were called up to 8 times each. Interviews were completed with 1,007 residents of the county with no diagnosis of lung cancer. Counts of nonworking telephone numbers, unanswered calls, out-of-range subjects, and refusals were commingled, precluding calculation of nonresponse rates among controls. A detailed description of the sample selection and data collection is provided elsewhere ( 4).

In addition to sections describing subject's demographics and tobacco use, residential, occupational, and medical history, the interview questionnaire contained a detailed section on dietary habits. This portion of the survey instrument was adapted from the Food Frequency Questionnaire (FFQ) portion of the Health Habits and History Questionnaire and its corresponding "Dietary Analysis System," developed by the National Cancer Institute ( 9) and used in the 1987 National Health Interview Survey ( 10). Additional questions were taken from the Continuing Population Survey of the American Cancer Society ( 11). In addition to questions on typical frequency and serving size for a list of 104 common foods and beverages, the section asked questions on supplementary vitamins, frequency of restaurant visits, and use of cooking fats and table fats. Participants were interviewed face-to-face and asked to answer by thinking back over the past, excluding the current year. Interviewers were trained to present questions in a neutral manner.

The Dietary Analysis System computed consumption of total dietary kilocalories, percent kilocalories from fat, percent kilocalories from saturated fat, fiber, micronutrients, and other descriptors of diet. For each dietary variable, quartiles were determined for the overall sample, and each individual was categorized by quartile. These ordinal data formed the basis for most comparisons of the dietary habits of cases vs. controls. Additionally, the frequencies from the individual foods were used to compute average servings per day for a variety of food groups, including nutrient-dense fruits and vegetables, "Southern" foods, and sweets.

Data collection for cases included telephone interviews if the subject was too sick to agree to (or to complete) a full face-to-face interview. This interview was shortened, particularly in the diet section, which was limited to 12 foods and 5 beverages. Although the data for these 29 cases were used to check whether cases that were difficult to follow up differed from other cases, these subjects could not be included in the dietary analysis. In addition, 10 other cases and 5 controls who completed the full interview reported unusually low food consumption (dietary kilocalories <400/day or <3.5 servings of solid food per day) or an unusually large number of foods on the food list with missing data (>10) were eliminated from the analyses. After this screening, 1,470 subjects (468 cases and 1,002 controls) remained for the dietary study.

Tobacco smoking was controlled for in all statistical analyses. Individuals were categorized as never smokers; former smokers (quit >/=10 yr previously) who had smoked <15 cigarettes/day, former smokers who had smoked 15-24 cigarettes/day, former smokers who had smoked >/=25 cigarettes/day, current/recent (quit <10 yr ago) smokers who smoke <15 cigarettes/day, current/recent smokers who smoke 1524 cigarettes/day, current/recent smokers who smoke >/=25 cigarettes/day, and pipe/cigar smokers who never smoked cigarettes.

Education was summarized with three levels: <12 years, high school or high school and trade school, or at least some college.

Statistical analyses were adjusted for consumption level by expressing the dietary variable as a percentage of kilocalories or by including quartile of kilocalories as an independent variable in the statistical model. Thus a reported effect for increased consumption of a dietary component is an effect relative to other people who reported about the same caloric amount. This was necessary to adjust for possibly large differences in individual participants' reporting level. This mechanism should control for bias in recall that affected the frequency or serving size reported for all (or a wide variety of) foods in the questionnaire. To further examine possible bias arising from the relative responders, analyses were repeated using only self-responders.

Odds ratios (ORs) were used as a measure of association between dietary factors and the risk of lung cancer, as computed from unconditional logistic regression adjusting for gender, age, race, education, smoking, and (where so noted in the tables) quartile of kilocalories ( 12). Tests for significance of trends in ORs with increasing dietary intake were computed from the regression analyses by assigning ordinal scores to the quartile of consumption level.

Cases and controls had similar distributions by age, race, and gender, as shown in Table 1. Controls tended to be better educated than cases (48.9% of controls with at least some college compared with 28.0% of cases), possibly because of a greater willingness of educated controls to participate in the lengthy interviews.

Although most interviews were completed by the subject, some were completed by a surviving relative, usually a spouse. Among cases, 160 were completed by a relative (34.2%); among controls, only 2 (0.2%) were completed by a relative. (Among the 162 "relative responders," in 40 the subject was present during the interview, but a relative answered most questions.) Interviews completed by relatives typically showed higher consumptions than those completed by the subject. Mean number of kilocalories for men whose relative responded was 2,151 compared with 1,767 when the subject responded; for women the corresponding values were 1,684 and 1,501. Such variables as percent kilocalories from fat, however, showed no relationship with respondent, which indicated that the pattern of preferred foods was less affected than the overall quantity.

The 29 cases given the shortened phone interview did not differ with respect to age, race, gender, zipcode, or smoking from the other cases. The two groups did not differ with respect to total number of servings for the 12 foods the two interviews had in common, although the cases with the shortened interview reported significantly higher total beverage servings.

Table 2 compares mean intakes of recommended dietary indexes frequently cited as markers of good diets ( 13) for cases with controls. For men and woman the controls reported significantly lower kilo-calorie consumption but showed no significant difference in mean servings of vegetables and fruits, cereals and grains, or fiber. Not only did cases apparently consume more kilocalories than controls, a larger fraction of those calories were from fat, as shown by a comparison of percent kilocalories from fat for men and women. In men and women the cases consumed a greater fraction of saturated and monounsaturated fat but did not differ from controls in the fraction of kilocalories from polyunsaturated fat. Neither group has an average profile that matches the National Research Council's recommended 30% of calories from fat, with 10% saturated, 10% polyunsaturated, and 10% monounsaturated fat ( 13).

Table 3 presents ORs for quartile level of intake of calories and of the recommended indexes. High consumption of total fruits and vegetables, cereals, and fiber was not associated with any significant protective effect against lung cancer. However, risk tended to rise with increasing intake of calories and fat. Those in the highest quartile of consumption of kilocalories and percent kilocalories from fat showed significantly increased risk. This risk was most strongly associated with saturated and monounsaturated fat and was not related to polyunsaturated fat.

Table 4 presents ORs for level of consumption of individual foods in the FFQ. (For brevity, only those foods with average servings >0.05/day are displayed.) A substantial reduction of risk was noted for those individuals in the highest category of consumption for "nutrient-dense" fruits and vegetables [a group of 18 fruits and vegetables providing the greatest ratio of nutrients to energy ( 14)]. Among the individual foods in this category, a number were associated with significantly reduced risk. On the other hand, consumption of "Southern cooking" foods, a broad group of 20 regional favorites from the exchange list by the American Diabetes Association ( 15), tended to be associated with increased risk. Among the individual Southern foods, several were associated with significantly elevated risk, whereas only coleslaw/cabbage and sweet potatoes/yams (also nutrient-dense vegetables) and beans were associated with decreased risk. The Southern and other foods associated with increased risk in Table 4 often tended to be foods with high percent kilocalories from fat.

The total servings per day in the Southern foods group is dominated by foods high in fat or frequently prepared with fat. The most frequently consumed items are "fat used in cooking" (0.51 time/day), followed by "butter/fat added to vegetables" (0.44 serving/day).Within this group, the most commonly consumed nutrient-dense fruit or vegetable is coleslaw (0.11 serving/day). Hence, although the Southern diet contains many healthful foods, the ORs reflect the association of lung cancer with the more frequently consumed high-fat foods.

Another component of traditional Southern cooking is the addition of lard or bacon fat to vegetables ( 7, 8). Among men, 41% of cases reported using lard or bacon fat in this way compared with 20% of controls. In women, the corresponding percentages were 38% compared with 21%. After adjustment for control variables, there was nearly a doubling of risk among users of lard and bacon fat [95% confidence interval for OR = 1.1-1.7 (men) and 1.2-1.9 (women)]. A similar kind of trend was noted for "eat skin on chicken," although statistical significance was weaker [95% confidence interval for OR = 1.0-1.4 (men) and 1.01.5 (women)].

ORs for dietary consumption of micronutrients are presented in Table 5. Increased intake of total carotenoids was associated with significantly reduced risk. This decrease was found among those with increased intake of alpha-carotene, beta-carotene, and all other provitamin A. Those with high intakes of vitamins A, C, and E also showed reduced risk.

Over 40% of subjects reported taking supplemental vitamins or minerals "fairly regularly." Hence, supplemental vitamins were an important potential source of micronutrients. ORs summarizing the association of supplemental vitamin intake with lung cancer are presented in Table 6. Regular use of supplemental vitamins or minerals was associated with significantly reduced risk. Although multivitamins showed no association with risk, supplemental intakes of vitamins C and E were associated with a significant protective effect.

ORs for consumption of beverages are reported in Table 7. No clear trend in ORs was seen for alcohol consumption; however, moderate (1 drink/day) consumption of alcoholic beverages showed a significant inverse association. Patterns with coffee or tea also were not consistent. Consumption of caffeinated coffee or tea was associated with higher risks for lung cancer. The effect, however, was not evident for coffee alone, nor did it appear among the highest consumers of tea. Furthermore, no increased risk was found for increasing intake of soft drinks, most of which are caffeinated.

All the statistical analyses for Tables 2-7 were repeated using only self-responders, in case observed effects might be due to systematic differences in the way self- and relative responders recall food consumption. In general, the conclusions remain unchanged for Tables 2, 3, and 5-7. In particular, the increased risks reported in Table 3 for those with high fat intake, high saturated fat intake, and high monounsaturated fat intake are stable even when relative responders are excluded. Also, the protective effects noted in Table 5 for those in the highest quartile of consumption for alpha-carotene, beta-carotene, and all other provitamin A are significant even in the smaller group. ORs for vitamins A and E for self-responders follow the same pattern as in the larger group but are not statistically significant in the smaller sample.

Results for individual foods (Table 4) are somewhat less stable when relative responders are excluded. Although nutrient-dense fruits and vegetables still show significant protective effects, the OR for Southern cooking is no longer significant. It appears that this is due to a lessening in the OR for certain frequently consumed foods, especially bacon, eggs, and coleslaw/cabbage. However, the most frequently consumed item in this group is "fat used in cooking," and this was associated with significantly increased risk, even in the smaller group.

Over the past decade, several epidemiological studies have reported an association between intake of foods high in fat and increased risk of lung cancer ( 5, 6). The results have not always been consistent, with some investigations finding no effect. The strongest finding, a six-fold increase in risk among those in the upper quintile of fat intake, was reported from a large case-control study of lung cancer among nonsmoking women in Missouri ( 16), but a recent subsequent study by the same authors involving smokers and nonsmokers failed to confirm the finding and raised the possibility of biased recall, including overreporting fat intake by the next of kin of the cases ( 17, 18). Nevertheless, the collective studies have raised the hypothesis that a high-fat diet or its components, including cholesterol and saturated fat, may influence the process of lung carcinogenesis. There is some biological rationale for the hypothesis, since fat has been described as a tumor promoter in experimental animals ( 19). Because of these and other findings linking diet and lung cancer, we included an extensive standardized dietary questionnaire in our study of the possible determinants of the exceptionally high rates of lung cancer in Jacksonville. In this study, we have estimated elsewhere ( 4) that smoking accounts for >90% of all lung cancers among Jacksonville residents. Although cigarette smoking is the dominant cause of lung cancer, we took the opportunity to evaluate whether the traditional Southern high-fat diet was associated with elevated risk in this population.

The dietary instrument employed in this study was the household FFQ developed by the National Cancer Institute and used in other surveys throughout the country ( 9). The questionnaire included >100 food and beverage items and covered nearly all the common foods consumed nationwide. Its use provided an assessment of fat and other macro- and micronutrient intake and included most (but not all) of the foods indigenous to Northeast Florida. Furthermore, some information was obtained on methods of food preparation common to the Southern diet, such as the addition of fat to vegetables during cooking. The questionnaires probably underestimated total calorie intake [among controls, caloric intake was 10-15% lower than national averages, which in turn are thought to be slight underestimates ( 20)], but our regression analyses were standardized for total calories to estimate the effects of fat and other specific food or nutrient consumption controlling for total energy intake.

The key dietary distinction between cases and controls was the higher consumption of high-fat foods by the cases. Various foods high in fat content were associated with significantly increased risk, and consumption in the upper quartile of the overall index percentage of total calories from fat was associated with an approximately doubled risk of lung cancer. Although the ability of the questionnaire to distinguish different types of fat (especially monounsaturated fats) was limited, the association held for saturated and monounsaturated fat, but not for polyunsaturated fat. The elevated ORs tended to be found for many of the high-fat foods that could be considered part of the Southern diet. Indeed, among the 20 items on our list of Southern foods, significantly elevated ORs were found for 7. The lung cancer patients in our study averaged 36% (men) and 34% (women) of calories from fat compared with national averages around 1990 of 33% in men and women, with much of the excess due to saturated fat (13% vs. 11% nationally) ( 21). Furthermore, the percentages in Jacksonville may be underestimates, because not all sources of fat, including Some in Southern cooking such as barbeque, are taken into account in the standardized FFQ methods for estimating fat intake.

The lung cancer patients tended to consume lower amounts of fruits and vegetables than the controls, but the differences were neither large nor significant. In contrast, most previous studies have shown reduced risks of lung cancer among persons in the upper quantiles of fruit and vegetable intake ( 5, 6, 22). However, when our analysis focused on nutrient-dense fruits and vegetables, reduced risk was found for persons in the highest quartile of consumption. In fact, 7 of the 18 nutrient-dense fruits and vegetables showed protective effects, whereas none of the 12 non-nutrient-dense fruits and vegetables was associated with reduced risk. Hence, one possible explanation for the lack of an overall difference in total fruits and vegetables between cases and controls is that benefit from nutrient-dense choices is obscured by the absence of effect for non-nutrient-dense choices. It is noteworthy that consumption of citrus fruits and juices, many of which are produced in Florida, was associated with a lower risk of lung cancer.

Another possible explanation for the lack of a clear association with total fruits and vegetables in Jacksonville may be the tendency in traditional Southern cooking for these foods to be prepared with fat or for fat to be added at the table, as shown by the use of "lard or bacon" discussed earlier. In this way, a beneficial effect may be masked by the adverse effects of increased fat intake.

This study yielded ORs of lung cancer associated with estimated high intakes of several micronutrients found in fruits and vegetables that tended be <1.0. Such was the situation for dietary intake of certain carotenoids, vitamins A, C, and E, and supplemental vitamins/minerals, especially vitamins C and E. Interest in beta-carotene as a potential cancerpreventive agent had been high, because multiple case-control and cohort studies assessing risk of cancer, particularly lung cancer, had found, as we had, that rates of cancer were lower among those consuming the largest quantities of foods rich in carotenoids ( 5, 6). The enthusiasm dropped precipitously, however, after reports in the mid-1990s of increased rather than decreased risk of lung cancer among participants in randomized clinical trials in Finland and the United States who had taken beta-carotene supplements ( 23, 24). The adverse effect seems limited to heavy smokers, especially since a trial of beta-carotene supplementation among US physicians, few of whom smoke, showed no concomitant increase in lung cancer ( 25). By contrast, in our study the protective effect reported in Table 5 for dietary beta-carotene remained when the analysis used only current/recent smokers. It is noteworthy that, in the randomized trials, incidence of lung cancer was significantly higher in persons who at baseline had low serum beta-carotene levels, so the low blood levels were effective predictors of subsequent risk ( 26). The Finnish ( 23) and US ( 24) trials nevertheless provided a warning that the attention to supplemental beta-carotene, to the exclusion of other carotenoids or to other constituents in foods high in beta-carotene, had been overdone and that nutritional findings from observational epidemiological studies must be interpreted cautiously. Our findings of lower risk among consumers of dietary and/or supplemental carotenoids and vitamins C and E should also be treated cautiously, although they are consistent with the overall literature suggesting protective effects of some components of fruit and vegetable intake. Persons who took any types of vitamin or mineral supplements were less common among the cases, and it is possible that these vitamin consumers may have other healthful behaviors, in addition to their supplement intake, that are associated with reduced risk of lung cancer.

Lung cancer is not considered to be a cancer related to alcohol consumption, although some studies have reported increased risk among heavy users, especially for beer drinking ( 5, 27). We found no clear link between drinking alcohol and risk of lung cancer after adjusting for tobacco intake. Neither tea nor coffee drinking individually showed consistent relations to lung cancer risk, although ORs tended to rise with increasing intake of caffeinated tea/coffee. The trend may be unlikely to represent a real effect of caffeine, however, since no association was seen with increasing numbers of sodas consumed. Tea contains polyphenolic and other constituents that have been shown to have fairly strong cancer inhibition properties in experimental animals, but large-scale cohort studies of tea drinkers have generally shown no association between tea drinking and risk of lung cancer ( 28). Caffeine has demonstrated in vitro anticarcinogenic activity and the ability to enhance the cytocidal effect of certain anticancer chemotherapeutic agents, but epidemiological evidence of any potential beneficial effect of caffeine on cancer risk is lacking ( 29).

The number of cases dropped from the analysis due to incomplete dietary information (39 subjects) is much larger than the number of controls (5 subjects). Because individuals with unrealistically low kilocalories were excluded, it is possible that this could have contributed to the higher average kilocalories among cases. However, most of the missing data were due to the 29 cases who were too ill to participate in the full interview. A comparison of their limited data to that for cases with full interviews showed no selective factors that might bias the results. Among those subjects with full interviews, only 2.1% of cases and 0.5% of controls had incomplete dietary information. Although these incompletion rates are different, they are only a small portion of the data. Hence, it is unlikely that differential rates for completing the survey could explain our results.

The potential for biased recall is a concern in any case-control study relying on recollection of past habits and events by the study subjects. It is possible that the cases, thinking about the causes of their diseases, may differentially recall and report certain exposures, especially those considered to be unhealthful. However, dietary factors are not commonly perceived as risk factors for lung cancer, and we have no evidence that the cancer patients may have overreported intake of fat. Persons agreeing to participate as controls tended to be of higher socioeconomic status, but we adjusted for education in all analyses, so socioeconomic status differentials in diet would be controlled. Nevertheless, a cautious interpretation of our findings is prudent.

In summary, this case-control study in an area with one of the highest rates of lung cancer in the United States (and the world) suggests that diet and nutrition influence risk of this cancer. Intake of foods high in fat, typical of the traditional Southern diet, appeared as the key dietary factor linked to increased risk, although low intake of nutrient-dense fruits and vegetables also contributed. Cigarette smoking is the dominant cause of lung cancer and contributes to the area's high lung cancer burden, so the main prevention activities should focus on smoking cessation. The results of this study, however, suggest the need for further clarification of dietary fat intake as a risk factor for lung cancer and its possible contribution to the elevated rates of lung cancer across broad stretches of the South.

Acknowledgments and Notes
This work was conducted under the leadership of the Heart and Lung Institute of Baptist/St. Vincent's Health System (Jacksonville, FL). Cooperating local hospitals include St. Vincent's, Riverside, University, Orange Park Humana, St. Luke's, Methodist, Baptist, Baptist-Beaches, Columbia Memorial, Jacksonville Naval Station Regional Medical Facility, and Veteran's Administration Hospitals (Lake City and Gainesville, FL). The cooperation of the Duval County Health Department was essential to the reporting system. Financial support was provided by City of Jacksonville Environmental Protection Board, Independent Life Insurance Company, Kirbo Charitable Trust, American Lung Association of Florida, American Cancer Society, Veterans of Foreign Wars Ladies Auxiliary, St. Vincent's Medical Center of Baptist/St. Vincent's Health System. Address reprint requests to Donna L. Mohr, University of North Florida, 4567 St. Johns Bluff Rd. S, Jacksonville, FL 32224.

Submitted 23 October 1998; accepted in final form 16 June 1999.

1 Data provided by the National Cancer Institute (Bethesda, MD).

Table 1. Demographic Comparison of Cases and Controls Among Subjects With Complete Dietary Questionnaires[a]

Legend for Chart:

A - Cases
B - Controls
C - P Value (for homogeneity)



35-54 yr 13.0 14.4
55-64 yr 27.8 25.3
65-74 yr 42.3 38.6
75-84 yr 16.9 21.7 0.11


Male 59.6 56.3
Female 40.4 43.7 0.23


White 83.6 82.8
Nonwhite 16.4 17.2 0.73


High school or high school
and trade school 38.6 32.0
High school and at least
some college 28.0 49.0 0.001
a: Values are percentages; n = 1,470 subjects.

Table 2. Comparison of Cases and Controls for Overall Dietary Indexes[a]

Legend for Chart:

A - Mean Daily Intake
B - Men: Cases (n = 279)
C - Men: Controls (n = 564)
D - Women: Cases (n = 189)
E - Women: Controls (n = 438)


Total kilocalories

2,132 1,769 1,555 1,347
(t = 6.7[*]) - (t = 4.3[*])

Vegetables and fruits, servings

4.3 4.5 4.4 4.7
(t = -1.3) (t = -1.7)

Cereals and grains, servings

2.9 2.8 2.7 2.6
(t = 1.3) (t = 1.0)

Dietary fiber, g

13.6 14.1 11.3 12.0
(t = -1.2) (t = -1.8)

% Kilocalories from fat (total)

35.8 32.9 34.9 31.5
(t = 5.9[*]) (t = 5.1[*])

%Kilocalories from saturated fat

13.4 11.8 12.9 11.4
(t = 7.1[*]) (t = 5.1[*])

%Kilocalories from monounsaturated fat

13.2 11.9 12.7 11.1
(t = 6.4[*]) (t = 6.2[*])

%Kilocalories from polyunsaturated fat

5.6 5.7 5.8 5.6
(t = -0.9) (t = 0.8)
a: Statistical significance is as follows: *, significant difference, with 2-tailed p < 0.01.

Table 3. Odds Ratios for Quartiles of Major Dietary Indexes[a]

Legend for Chart:

A - Quartile: 1[b]
B - Quartile: 2
C - Quartile: 3
D - Quartile: 4
E - P Value (for trend)


Kilocalories[c] 1.0 0.84 1.37
2.33[1] <0.0001

% Kilocalories[c]

From fat 1.0 1.38 1.78[1]
2.11[1] <0.0001

From saturated fat 1.0 1.45 1.49
2.15[1] 0.0003

From monounsaturated fat 1.0 1.24 1.56[*]
2.97[1] 1.0001

From polyunsaturated fat 1.0 1.23 1.44
1.00 0.78

Fruits and vegetables[c,d] 1.0 1.02 1.13
0.83 0.53

Fruits[c,d] 1.0 0.96 1.19
0.84 0.66

Vegetables[c,d] 1.0 1.02 1.04
1.00 0.99

Fiber[c,d] 1.0 0.94 0.95
0.72 0.19

Cereals and grains[c,d] 1.0 1.09 0.83
0.85 0.31
a: Statistical significance is as follows: *, significantly different from 1.0, with 2-tailed p < 0.05; 1, significantly different from 1.0, with 2-tailed p < 0.01.

b: In all cases, baseline group is Quartile 1, or lowest level of consumption. c: Adjusted for age, gender, race, education, and smoking. d: Uses quartile of kilocalories as an additional control variable.

Table 4. Odds Ratios for Consumption of Foods[a-c]

Legend for Chart:

A - Quartile: 1[b]
B - Quartile: 2
C - Quartile: 3
D - Quartile: 4
E - P Value (for trend)[d]


Food groups

Nutrient dense 1.00 0.81 0.85
0.481 0.002[2]

Sweets 1.00 1.15 0.97
1.61[*] 0.070[3]

Southern cooking 1.00 0.78 1.49
1.58 0.006[4]

Southern cooking

Grits 1.00 0.63[1] 0.98
1.21 0.13

Eggs 1.00 1.19 1.50[*]
1.78[1] 0.003[4]

Bacon 1.00 0.84 1.25
1.63[*] 0.004[4]

Sausage 1.00 0.91 1.04
1.43 0.18

Other beans 1.00 1.01 0.89
0.51[1] 0.010[**]

Mustard and other greens 1.00 1.48[*] 1.03
1.08 0.92

Coleslaw, cabbage,
sauerkraut 1.00 1.29 0.74
0.66 0.032[**]

Sweet potatoes, yams 1.00 0.97 0.89
0.55[1] 0.009[**]

Pork 1.00 0.97 1.21
0.88 0.61

Fried chicken 1.00 1.56[*] 1.02
1.00 0.75

Fried fish 1.00 1.09 0.93
1.12 0.72

Hot dogs 1.00 0.92 1.53[*]
1.09 0.17

Ham, lunch meats 1.00 1.36 1.24
2.01[1] 0.010[5]

Biscuits, muffins 1.00 1.36 1.44[*]
1.01 0.50

Cornbread 1.00 1.16 1.66[*]
1.46 0.025[5]

Gravies 1.00 1.65[*] 2.05[1]
1.64[*] 0.003[4]

Butter/fat added to vegetable 1.00 1.12 1.02
1.01 0.93

Fat used in cooking 1.00 1.87[1] 1.61[*]
1.70[1] 0.005[5]

Nutrient-dense fruits
and vegetables

Peaches, apricots 1.00 0.89 1.23
1.24 0.19

Cantaloupe in season 1.00 0.88 0.99
0.78 0.41

Strawberries in season 1.00 1.06 0.61[*]
0.72 0.027[**]

Oranges 1.00 1.02 1.03
0.56[*] 0.044[**]

Grapefruit 1.00 0.66[*] 0.73
0.43[1] <0.0001[2]

Orange/grapefruit juice 1.00 0.90 0.96
0.63 0.69

Winter squash 1.00 0.95 1.36
1.21 0.14

Tomatoes, tomato juice 1.00 1.11 1.37
0.99 0.42

Broccoli 1.00 0.87 0.92
0.49[1] 0.011[**]

Cauliflower, brussels
sprouts 1.00 1.18 0.78
0.40[1] 0.007[**]

Spinach, cooked 1.00 1.13 1.35
1.05 0.46

Mustard and other greens 1.00 1.48[*] 1.03
1.08 0.92

Coleslaw, cabbage,
sauerkraut 1.00 1.29 0.74
0.66 0.032[**]

Carrots 1.00 1.15 1.00
0.46[1] 0.020[**]

Sweet potatoes, yams 1.00 0.97 0.89
0.55[1] 0.009[**]


Ice cream 1.00 1.48[*] 1.52[*]
2.59[1] <0.0001[4]

Doughnuts, cake, pastry 1.00 1.08 1.21
1.40 0.072[3]

Chocolate candy 1.00 1.02 0.98
1.18 0.51

Other candy, jelly, honey 1.00 1.66[1] 1.44
1.57[*] 0.023[5]

Other foods

Apples, applesauce, pears 1.00 0.72 1.56[*]
0.94 0.17

Bananas 1.00 1.21 1.25
0.70 0.26

Watermelon 1.00 1.58[*] 1.36
1.37 0.103

Other fruits 1.00 0.85 1.11
1.16 0.35

High-fiber cereals 1.00 0.47[1] 0.55[1]
-- 0.0001[2]

Fortified cereals 1.00 0.53[1] --
-- 0.0003[2]

Other cold cereals 1.00 0.80 0.80
0.96 0.49

Milk on cereals 1.00 0.83 0.78
0.71 0.083[6]

Sugar on cereals 1.00 1.19 1.23
-- 0.19

String/green beans 1.00 1.12 0.73
-- 0.61

Peas 1.00 1.18 0.86
-- 0.73

Corn 1.00 1.32 0.62[*]
-- 0.33

Green salad 1.00 0.80 0.92
0.71 0.21

Salad dressing/mayonnaise 1.00 1.35 0.94
0.80 0.21

Fried potatoes 1.00 0.85 1.18
1.00 0.69

Other potatoes 1.00 0.90 1.27
1.85[*] 0.009[4]

Rice 1.00 1.08 0.94
0.91 0.55

Other vegetables 1.00 1.10 1.13
-- 0.46

Mixed vegetables 1.00 1.22 1.08
0.62 0.30

Hamburgers, meatloaf, tacos 1.00 1.03 1.09
0.72 0.65

Beef (steaks, roasts,
sandwich) 1.00 1.61[1] 1.16
-- 0.20

Chicken/turkey (not fried) 1.00 1.39 0.86
0.75 0.101
Tuna, tuna salad 1.00 0.96 0.92
0.87 0.55

Other fish (broiled, baked) 1.00 0.81 0.77
0.35[1] 0.003[2]

Pasta with tomato sauce 1.00 1.06 1.26
0.62 0.69

Mixed dishes with cheese 1.00 1.18 1.19
1.65[*] 0.024[5]

Vegetable/tomato soups 1.00 1.16 1.03
1.66[*] 0.032[5]

Other soups 1.00 1.30 1.32
1.35 0.13

White bread 1.00 1.11 1.10
1.58[*] 0.024[5]

Dark bread 1.00 0.61[1] 0.73
0.53 0.040[**]

Salty snacks 1.00 1.41 1.47
1.06 0.77

Peanuts, peanut butter 1.00 0.86 1.14
1.19 0.27

Margarine on bread/rolls 1.00 0.76 0.68[*]
1.22 0.67

Butter on bread/rolls 1.00 1.03 0.94
-- 0.74

Cottage cheese 1.00 0.85 0.72
1.16 0.78

Other cheeses 1.00 0.88 1.34
1.23 0.12

Flavored/frozen yogurt 1.00 0.84 0.64[*]
-- 0.022[**]


Whole milk 1.00 1.55 1.80[1]
-- 0.0003[4]

2% Milk 1.00 -- --
0.82 0.24

Skim milk 1.00 0.62 0.73
-- 0.050[**]

Nondiet soft drinks 1.00 1.08 1.36
1.08 0.37

Diet soft drinks 1.00 0.60[*] 1.00
-- 0.58

Beer 1.00 0.77 1.17
-- 0.50

Wine, wine coolers 1.00 0.66 0.77
-- 0.096[6]

Liquor 1.00 0.77 0.73
-- 0.06[6]

Regular coffee 1.00 1.02 1.24
1.27 0.17

Decaffeinated coffee 1.00 0.54[*] 0.61[1]
-- 0.003[2]

Regular tea 1.00 0.96 1.37
1.64[*] 0.005[4]

Decaffeinated tea 1.00 0.76 --
-- 0.27

Lemon in tea 1.00 0.99 0.69[*]
-- 0.061[6]

Nondairy creamer in
coffee/tea 1.00 0.67[*] --
-- 0.019[**]

Cream or half-and-half
in coffee/tea 1.00 0.60[*] --
-- 0.032[**]

Milk in coffee/tea 1.00 1.37[*] --
-- 0.050[3]

Sugar in coffee/tea 1.00 0.89 1.41
-- 0.087[3]

Artificial sweetener in
coffee/tea 1.00 1.00 0.73
-- 0.069[6]

Water (glasses) 1.00 1.03 0.98
-- 0.96
a: All odds ratios are adjusted for age, gender, race, education, smoking, and quartile of kilocalories.

b: In all cases, baseline group is that having the lowest consumption. Participants were divided into 4 categories, except in the case of infrequently consumed foods, where only 2 or 3 classes were possible.

c: Statistical significance is as follows: *, significantly different from 1.0, with 2-tailed p < 0.05; 1, significantly different from 1.0, with 2-tailed p < 0.01.

d: All P values are based on 2-tailed tests. For cases higher than controls: 3, 0.05 < p

Table 5. Odds Ratios for Quartiles of Dietary Consumption of Micronutrients[a,b]

Legend for Chart:

A - Quartile: 1[c]
B - Quartile: 2
C - Quartile: 3
D - Quartile: 4
E - All Subjects P Value (for trend)


Carotenoids (total) 1.0 0.67[*] 0.85
0.60[*] 0.07

alpha-Carotene 1.0 0.94 1.36
0.48[1] 0.064

beta-Carotene 1.0 0.65[*] 0.91
0.52[1] 0.030

Cryptoxanthin 1.0 1.11 1.30
0.86 0.70

Lycopene 1.0 1.44 1.20
1.12 0.81

Lutein 1.0 1.00 1.22
0.97 0.87

All other provitamin A 1.0 0.80 0.92
0.46[1] 0.004

Retinol 1.0 0.81 1.00
1.26 0.18

Vitamin A 1.0 1.01 1.09
0.58[*] 0.044

Vitamin C 1.0 0.99 0.71
0.68 0.026

Vitamin E 1.0 0.74 0.55[*]
0.54[*] 0.015
a: All odds ratios are adjusted for age, gender, race, smoking, education, and quartiles of kilocalories.

b: Statistical significance is as follows: *, significantly different from 1.0, with 2-tailed p < 0.05, 1, significantly different from 1.0, with 2-tailed p < 0.01.

c: In all cases, baseline group is Quartile 1, or lowest level of consumption.

Table 6. Odds Ratios for Supplemental Vitamin Consumption[a,b]

Legend for Chart:

A - No
B - Yes, but Not Regularly
C - Yes, Regularly
D - P Value (for trend)


Ever taken any
vitamins or minerals? 1.0 0.68 0.47[1] 0.0001

Legend for Chart:

A - No
B - Yes
C - P Value (for trend)


Take multivitamin fairly regularly? 1.0 0.71[*] 0.025

Legend for Chart:

A - None
B - Some
C - P Value (for trend)

Vitamin A 1.0 0.64 0.33
Vitamin C 1.0 0.58[*] 0.016
Vitamin E 1.0 0.51[1] 0.005
a: All odds ratios are adjusted for age, gender, race, education, and smoking.

b: Statistical significance is as follows: *, significantly different from 1.0, with 2-tailed p < 0.05, 1, significantly different from 1.0, with 2-tailed p < 0.01.

Table 7. Odds Ratios for Beverage Consumption[a,b]

Legend for Chart:

A - Amount Consumed[c]: 0/day
B - Amount Consumed[c]: 1/day
C - Amount Consumed[c]: 2-3/day
D - Amount Consumed[c]: >/=4/day
E - P Value (for trend)


Alcoholic beverages 1 0.63[*] 1.23
1.02 0.87

Beer[d] 1 0.80 --
-- 0.27

Wine[d] 1 1.12 --
-- 0.67

Liquor[d] 1 1.17 --
-- 0.46

Regular coffee, tea, soda 1 0.91 1.25
1.54 0.013

Regular coffee 1 1.06 1.26
1.29 0.15

Regular tea 1 1.39 1.85[1]
0.77 0.024

Regular coffee or tea 1 1.24 1.65[*]
1.88[1] 0.001

Sodas 1 1.02 1.00
0.82 0.81

All coffees, including decaffeinated 1 0.83 0.85
0.91 0.75

All teas, including decaffeinated 1 1.26 1.59[*]
0.77 0.12
a: All odds ratios are adjusted for age, gender, race, smoking, education, and quartile of kilocalories.

b: Statistical significance is as follows: *, significant difference at 0.01

c: Categories for grouped variables: 0/day = 0 /=4/day = 3.5

d: 0.0 /=0.5 servings per day.

1. Landis, SH, Murray, T, Bolden, S, and Wingo, PA: Cancer statistics 1998. CA Cancer J Clin 48, 6-30, 1998.

2. Blot, WJ, Davies, JE, Brown, LM, Nordwall, CW, Buiatti, E, et al.: Occupation and the high risk of lung cancer in Northeast Florida. Cancer 50, 364-371, 1982.

3. Sa, P, Tousey, PM, Blot, WJ, and Wolfe, KW: Tobacco smoking: a survey in a community with excess lung cancer. South Med J 90, 601-605, 1997.

4. Tousey, PM, Wolfe, KW, Mozeleski, A, Mohr, D, Cantrell, BB, et al.: Determinants of the excessive rates of lung cancer in Northeast Florida. South Med J 92, 493-501, 1999.

5. Blot, W J, and Fraumeni, JF, Jr: Cancers of the lung and pleura. In Cancer Epidemiology and Prevention, 2nd ed, D Shottenfeld and J Fraumeni, Jr (eds). New York: Oxford University Press, 1996, pp 637-665.

6. Ziegler, RG, Mayne, ST, and Swanson, CA: Nutrition and lung cancer. Cancer Causes Control 7, 157-178, 1996.

7. US Department of Agriculture: Dietary levels: households in the United States, Spring 1977. In The 1977-78 Nationwide Food Consumption Survey (NFCS). Washington, DC: US Govt Printing Office, 1985. (Rep. H-11; NTIS Accession No. PB91-103515)

8. US Department of Agriculture: Dietary levels: households in the South, Spring 1977. In The 1977-78 Nationwide Food Consumption Survey (NFCS). Washington, DC: US Govt Printing Office, 1985. (Rep H-14; NTIS Accession No. PB91-103663)

9. Block, G, Hartman, AM, and Naughton, D: A reduced dietary questionnaire: development and validation. Epidemiology 1, 58-64, 1990.

10. US Public Health Service: National Health Interview Survey. Washington, DC: US Dept of Health and Human Services and Bureau of the Census, US Dept of Commerce, 1987. (Form HIS-10)

11. Thun, MJ, Day-Lally, CA, Calle, EE, Flanders, WD, and Heath, CW, Jr: Excess mortality among cigarette smokers: changes in a 20 year interval. Am J Public Health 85, 1223-1230, 1995.

12. Breslow, NE, and Day, NE. Statistical Methods in Cancer Research. Lyon, France: Int Agency Res Cancer, 1980, vol 32, pp 5-338.

13. National Research Council. Committee on Diet, Nutrition, and Cancer. Assembly of Life Sciences: Diet, Nutrition, and Cancer. Washington, DC: National Academy Press, 1982.

14. Guthrie, H, and Picciurro, MF: Human Nutrition. St. Louis, MO: Mosby, 1995.

15. Shils, ME, Olson, JA, and Shike, M: Modern Nutrition in Health and Disease, 8th ed. Philadelphia, PA: Lea & Febiger, 1994, pp A-122A-123.

16. Alavanja, M, Brown, C, Swanson, C, and Brownson, R: Saturated fat intake and lung cancer risk among nonsmoking women in Missouri. JNCI 85, 1906-1916, 1993.

17. Swanson, C, Brown, C, Brownson, R, and Alavanja, M: Re: Saturated fat intake and lung cancer risk among nonsmoking women in Missouri. JNCI 89, 1724-1725, 1997.

18. Swanson, C, Brown, C, Sinha, R, Kulldorff, M, Brownson, R, et al.: Dietary fats and lung cancer risk among women: the Missouri Women's Health Study. Cancer Causes Control 8, 883-893, 1997.

19. Wynder, EL, Hebert, JR, and Kabat, GC: Association of dietary fat and lung cancer. JNCI 79, 631-637, 1987.

20. Briefel, RR, Sempos, CT, McDowell, MA, Chien, S, and Alaimo, K: Dietary methods research in the Third National Health and Nutrition Examination Survey: underreporting of energy intake. Am J Clin Nutr 65 Suppl, 1203S-1209S, 1997.

21. McDowell, MA, Briefel, RR, Alaimo, K, Bischof, AM, Coughman, CR, et al.: Energy and Macronutrient Intakes of Persons Ages 2 Months and Over in the United States: Third National Health and Nutrition Examination Survey, Phase 1, 1988-91. Washington, DC: Div of Health Examination Statistics, National Center for Health Statistics, 1994, pp 6-21. (Advance Data No. 255)

22. Pillow, PC, Hursting, SD, Duphorne, CM, Jiang, H, Honn, SE, et al.: Case-control assessment of diet and lung cancer risk in African Americans and Mexican Americans. Nutr Cancer 29, 169-173, 1997.

23. ATBC Cancer Prevention Study Group: The effect of vitamin E and beta-carotene on the incidence of lung cancer and other cancers in male smokers. N Engl J Med 330, 1029-1035, 1994.

24. Omenn, GS, Goodman, GE, Thornquist, MD, Balmes, J, Cullen, MR, et al.: Effects of a combination of beta-carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med 334, 1150-1155, 1996.

25. Hennekins, CH, Buring, JE, Manson, JE, Stampfer, M, Rosner, B, et al.: Lack of effect of long-term supplementation with beta-carotene on the incidence of malignant neoplasms and cardiovascular disease. N Engl J Med 334, 1145-1149, 1996.

26. Albanes, D, Heinonen, OP, Taylor, PR, Virtamo, J, Edwards, BK, et al.: alpha-Tocopherol and beta-carotene supplements and lung cancer incidence in the alpha-Tocopherol, beta-Carotene Cancer Prevention Study: effects of baseline characteristics and study compliance. JNCI 88, 1560-1570, 1996.

27. Potter, JD, Sellers, TA, Folson, AR, and McGovern, PG: Alcohol, beer and lung cancer in postmenopausal women: the Iowa Women's Health Study. Ann Epidemiol 2, 557-595, 1992.

28. Blot, WJ, Chow, WH, and McLaughlin, JK: Tea and cancer: a review of the epidemiologic evidence. Eur J Cancer Prev 5, 425-438, 1996.

29. Nehlig, A, and Debry, G: Coffee and cancer: a review of human and animal data. World Rev Nutr Diet 79, 185-221, 1996.


By Donna L. Mohr; William J. Blot; Phyllis M. Tousey; Michele L. Van Doren and Kevin W. Wolfe

D. L. Mohr is affiliated with the University of North Florida, Jacksonville, FL 32224. W. J. Blot is affiliated with the International Epidemiology Institute, Rockville, MD 20850. P. M. Tousey and M. L. Van Doren are affiliated with The Heart and Lung Institute at St. Vincent's, Inc., Baptist/St. Vincent's Health System, Jacksonville, FL 32204. K. W. Wolfe is affiliated with St. Vincent's Medical Center, Jacksonville, FL 32204

Share this with your friends
Southern Cooking and Lung Cancer.pdf758.3 KB