Epidemiologic studies utilizing surveys: accounting for the sampling design

Dietary iron as a risk factor for myocardial infarction. Public health considerations for Nova Scotia

BACKGROUND

Epidemiological and experimental studies have suggested that high levels of dietary iron and hemeiron can lead to myocardial injury. Lean meat, a primary source of iron and hemeiron, is promoted because it is lower in fat and cholesterol. Does lean meat put us at risk for myocardial infarction, and should we reconsider its promotion?

METHODS

We analyzed the importance of dietary iron and hemeiron as a risk for myocardial infarction among 2,198 Nova Scotians who participated in a nutrition survey and who were followed for eight years, using logistic regression.

RESULTS

Acute myocardial infarction incidents occurred in 94 (4.3%) participants. We found no increased risk for myocardial infarction associated with high intake of iron and hemeiron.

CONCLUSIONS

Based on Nova Scotian data showing no increased risk for myocardial infarction with high intake of iron and hemeiron, there is no need for immediate reconsideration of promotion of lean meat.

Rural-urban home health care differences before the Balanced Budget Act of 1997

This study arose from concerns that home health care may be more difficult to provide to rural than urban elderly patients (because of geographic barriers, personnel shortages, and other factors) and may therefore be less effective in terms of patient outcomes. Case mix, home health care service use, and outcomes (primarily discharge status) were analyzed for a national random sample of 3,869 rural and urban elderly home health patients. Longitudinal data covered the period from home health admission to discharge or 120 days (whichever occurred first). Primary data collection instruments were designed to obtain longitudinal patient-level health status data; agency records and Medicare data provided service use information. (The study did not address access but focused on services and outcomes after admission to home health care.) Two-group statistical tests and multivariate analyses were employed to assess rural-urban differences. The major findings were that, after adjustment for rural-urban case mix and agency differences, rural compared to urban patients received fewer home health services and attained less favorable discharge outcomes. For example, the rural patients had a higher case mix adjusted hospitalization rate. Because the study data pertain to 1995 through 1996, the results provide a baseline for future analyses of possibly different rural compared to urban effects of the Balanced Budget Act of 1997, which resulted in major changes in Medicare payment for home health care.

Individual and neighbourhood determinants of health care utilization. Implications for health policy and resource allocation

OBJECTIVES

To investigate the importance of both individual and neighbourhood socioeconomic characteristics for health care utilization.

METHODS

Various linkage procedures generated a longitudinal dataset with information on 2,116 Nova Scotians, their residential neighbourhoods, 8 years of health care utilization and vital status. Unilevel and multilevel regression analyses were employed to examine the effects of both individual and neighbourhood characteristics on health care use.

RESULTS

Individual income and education determined physician and hospital use. Also, neighbourhood characteristics, specifically average income and percentage of single mother families, were found to determine health care use. When considering individual and neighbourhood characteristics simultaneously, individual income and education determined physician and hospital use independently, while neighbourhood income determined physician use independently.

CONCLUSIONS

Both individual and neighbourhood socioeconomic characteristics determine health care use. Acknowledging this allows better targeting of health policy and planning, and enables more accurate needs-based resource allocation.

Use of out-of-hospital variables to predict severity of injury in pediatric patients involved in motor vehicle crashes

STUDY OBJECTIVE

We sought to create a clinical decision rule, on the basis of variables available to out-of-hospital personnel, that could be used to accurately predict severe injury in pediatric patients involved in motor vehicle crashes as occupants.

METHODS

We analyzed the National Automotive Sampling System database, a national probability sample, using pediatric patients up to 15 years old (occupants only) involved in motor vehicle crashes from January 1993 to December 1999. The National Automotive Sampling System database includes patients from regions throughout the country, weighted to represent a nationwide sample. Twelve out-of-hospital variables were used in classification and regression tree analysis to create a decision rule separating children with severe injuries (Injury Severity Score [ISS] > or =16) from those with minor injuries (ISS < 16). Misclassification costs and complexity parameters were selected to yield a decision tree with appropriate sensitivity and specificity for the identification of severely injured patients, while also being simple and practical for out-of-hospital use. Probability weights were used throughout the analysis to account for the sampling design and sampling weights.

RESULTS

Using a sample size of 8,392 children, we constructed a decision rule using 3 out-of-hospital variables (Glasgow Coma Scale score, passenger space intrusion > or =6 in [> or =15 cm], and restraint use) to predict those patients with an ISS of 16 or more. Internal cross-validation was used to determine the sensitivity and specificity, yielding values of 92% and 73%, respectively, for the prediction of patients with an ISS of 16 or more.

CONCLUSION

Out-of-hospital variables available to field personnel could be used to effectively triage pediatric motor vehicle crash patients using the decision rule developed here. Prospective trials would be needed to test this decision rule in actual use.

Moderate renal insufficiency and the risk of cardiovascular mortality: results from the NHANES I

BACKGROUND

Conflicting evidence exists concerning whether renal insufficiency is an independent risk factor for cardiovascular disease in the general population. The objective of this study was to determine whether moderate renal insufficiency was associated with total and cardiovascular mortality, independent of traditional cardiovascular risk factors, in a community sample representative of the U.S. general non-institutionalized population.

METHODS

Participants in the U.S. First National Health and Nutrition Examination Survey (NHANES I, 1974-1975) and NHANES I Epidemiologic Follow-up Study (NHEFS, 1992, 18 year follow-up) were evaluated. The primary analysis was limited to 2352 adults with complete data, and no baseline cardiovascular disease. A creatinine of 104 to 146 micromol/L in women, and 122 to 177 micromol/L in men (approximate glomerular filtration rate of 30 to 60 mL/min/1.73 m2) was defined as moderate renal insufficiency. Supplementary analyses included participants with marked renal impairment and baseline cardiovascular disease.

RESULTS

The unadjusted hazard ratio for moderate renal insufficiency compared to preserved renal function was significant for total mortality (hazard ratio 1.7; 95% confidence interval 1.3 to 2.2), and for cardiovascular mortality (2.2; 1.5 to 3.1). After adjustment for traditional cardiovascular risk factors, there was no independent association between moderate renal insufficiency and total mortality (1.0; 0.8 to 1.4), or cardiovascular mortality (1.2; 0.8 to 1.8). These results were consistent in supplementary analyses.

CONCLUSIONS

These results do not support moderate renal insufficiency as an independent risk factor for cardiovascular disease in the general population. The association between moderate renal insufficiency and cardiovascular disease, demonstrated in other epidemiologic studies, appears to be due to co-occurrence of renal insufficiency with traditional cardiovascular risk factors.

Association of awareness, intrapersonal and interpersonal factors, and stage of dietary change with fruit and vegetable consumption: a national survey

PURPOSE

To examine associations of awareness, intrapersonal and interpersonal factors, and stage of change with consumption of fruits and vegetables.

DESIGN

Nationally representative, random digit dial survey conducted in 1997 with a response rate of 44.5%. Psychosocial correlates of fruit and vegetable consumption were assessed using regression analyses.

SETTING

United States.

SUBJECTS

A total of 2605 adults who were 18 years and older.

MEASURES

Awareness of the "5 A Day for Better Health" program and its message, along with stage of change; taste preferences; self-efficacy; and perceived benefits, barriers, threats, social support, and norms related to fruit and vegetable consumption.

RESULTS

Awareness and intrapersonal and interpersonal factors explained 24% of the variance in fruit and vegetable consumption beyond the 9% explained by demographic characteristics. Knowledge of the 5 A Day message was associated with a 22% increase in fruit and vegetable consumption. Self-efficacy for eating fruits and vegetables and taste preferences (affect) were the factors most consistently and strongly associated with both higher consumption and higher likelihood of being in action or maintenance stages of change. Affect and perceived barriers were more strongly associated with increased vegetables and salad than fruit.

CONCLUSIONS

Dietary intervention programs to increase fruit and vegetable consumption should emphasize the 5 A Day message, increased self-efficacy, and ways to make vegetables more palatable and easily accessible. Understanding the factors that influence dietary choices should be used when designing dietary interventions.

Complex sampling: implications for data analysis

Investigators in dental public health often use strategies other than simple random sampling to identify potential subjects; however, their statistical analyses do not always take into account the complex sampling mechanism. Often it is not clear whether a given strategy requires adjustment for stratification and/or cluster sampling of observations. We propose that the need for such adjustment depends on the primary study objective. As a general rule, we recommend that if the study goal is to estimate the magnitude of either a population value of interest (e.g., prevalence), or an established exposure-outcome association, adjustment of variances to reflect complex sampling is essential because obtaining appropriate variance estimates is a priority. However, if the study goal is to establish the presence of an association, especially in a preliminary investigation of novel conditions or understudied populations, obtaining appropriate variance estimates may not be of primary importance; hence, adjustment of variances for complex sampling is not always required, but often is recommended. This paper describes several types of complex sampling designs, methods of adjusting for complex sampling strategies, examples illustrating the effect of adjustment, and alternative approaches for analysis of complex samples.

Consumption of alcoholic beverages and subjective health in Spain

STUDY OBJECTIVE

To examine the relation between alcohol and main alcoholic beverage consumption and subjective health in Spain.

DESIGN

Logistic regression analysis using a cross sectional survey based on self reported data on alcohol and alcoholic beverage consumption, subjective health and the principal confounding factors (age, sex, civil status, educational level, job status, social support, region of residence, size of town or city, tobacco consumption, physical activity during leisure time and work hours, and chronic disease).

SETTING

The 1993 Spanish National Health Survey.

PARTICIPANTS

A 19 573 person sample, representative of the non-institutionalised Spanish population aged 16 years and over.

MAIN RESULTS

Among Spaniards, 31.4% reported their health as suboptimal (fair, poor or very poor) and 56.9% consumed alcohol regularly, with the majority having a preference for wine. Light (1-2 drinks per day) or moderate consumption (3-4 drinks per day) was the most frequent pattern. After adjusting for confounding factors, a negative dose-response relation was observed between consumption of total alcohol, wine and beer, and prevalence of suboptimal health (linear trend: p<0.001 for total alcohol, p=0.023 for wine, and p=0.030 for beer). In contrast, for consumption of spirits the prevalence of ill health in moderate drinkers was lower than in non-drinkers, with no clear relation at higher consumption. While persons reporting a preference for wine had a lower frequency of suboptimal health than did abstainers, they showed no difference in frequency of subjective ill health with respect to persons with preference for other types of drink or no preference whatsoever.

CONCLUSIONS

The higher the consumption of total alcohol, wine and beer, the lower the prevalence of suboptimal health. These results differ from those obtained in several Nordic countries, where a "J shaped" relation has been observed for total alcohol and wine, and suggest that the relation between alcohol consumption and subjective health may be different in Mediterranean countries.

Examining the link between coronary heart disease and the elimination of chronic dental infections

BACKGROUND

While it has been suggested that periodontal disease may be associated with coronary heart disease, or CHD, there are no data to suggest that the elimination of chronic dental infections actually lowers the risk of developing chronic CHD. The goal of this study was to determine whether people with a definitive elimination of all potential dental infections--edentulous people, who are at the optimum endpoint of dental infection elimination from a CHD perspective--lower their CHD risk over time when compared with people who have a specific dental infection, periodontitis.

METHODS

The authors examined data from a prospective cohort of 4,027 people who participated in the First National Health and Nutrition Examination Survey, or NHANES I, Epidemiologic Follow-up Study. The primary outcome measure was the first CHD event.

RESULTS

During a mean follow-up of 17 years, there were 1,238 CHD events (538 fatal). The confirmed elimination of chronic dental infections did not lead to a decreased risk of experiencing a CHD event (relative risk, 1.02; 95 percent confidence interval, 0.86-1.21). The CHD risk among people with and without chronic dental infections remained constant over time with respect to each other (test for increasing or decreasing trend over time: not significant, chi2(1) = 0.48; P = .93).

CONCLUSIONS

People who had a complete, definitive and long-term elimination of all potential dental infections through extraction of all teeth did not have lower CHD risk when compared with people with diagnosed periodontitis.

CLINICAL IMPLICATIONS

Until evidence is found to the contrary, the authors suggest that prevention of CHD should not be used as the basis for recommending treatment to eliminate chronic dental infections.

Microalbuminuria is common, also in a nondiabetic, nonhypertensive population, and an independent indicator of cardiovascular risk factors and cardiovascular morbidity

OBJECTIVES

To assess the prevalence of microalbuminuria in the general population, especially in nondiabetic and nonhypertensive subjects, and its association with known cardiovascular risk factors and cardiovascular morbidity.

DESIGN

Cross-sectional cohort study.

SETTING

Inhabitants of the city of Groningen, the Netherlands.

SUBJECTS

All inhabitants, aged between 28 and 75 years, were send a postal questionnaire and a vial to collect an early morning urine sample (n = 85 421). Of these 40 856 subjects (47.8%) responded. Cardiovascular risk factors and morbidity were validated in a well defined nondiabetic and nonhypertensive group of 5241 subjects.

MAIN OUTCOME MEASURES

Microalbuminuria, self-reported cardiovascular risk and cardiovascular morbidity in the total study cohort, and additionally more detailed measurements in a subset of the total population.

RESULTS

Microalbuminuria (20-200 mg L-1) was present in 7.2% of the subjects and independently associated with age, gender, hypertension, diabetes, smoking, previous myocardial infarction and stroke. Some of these associations were already observed at albuminuria levels of 10-20 mg L-1. After exclusion of the diabetic and hypertensive subjects, microalbuminuria was still prevalent in 6.6% of the subjects.

CONCLUSIONS

Microalbuminuria appears to be common not only in the general population but also in a nondiabetic, nonhypertensive population and is independently associated with increased cardiovascular risk factors and cardio-vascular morbidity. Importantly, some of these associations are present at urinary albumin levels currently considered to be normal. These findings suggest that urinary albumin measurements may be useful in early risk profiling and prevention of cardiovascular disease in the population at large.

Multiple risk factors and population attributable risk for ischemic heart disease mortality in the United States, 1971–1992

The objective of this study was to assess the associations and population attributable risks (PAR) of risk factor combinations and ischemic heart disease (IHD) mortality in the United States. We used logistic regression models to assess the association of risk factors with IHD in the First National Health and Nutrition Examination Survey (1971-1974) and Epidemiologic Follow-up Study (1982-1992) among white and black men and women. We examined eight modifiable risk factors: hypertension, elevated serum cholesterol, diabetes, overweight, current smoking, physical inactivity, depression, and nonuse of replacement hormones. Risk factors associated with IHD mortality were the same among white and black men (i.e., age, education, smoking, diabetes, hypertension, and serum cholesterol). Age, education, smoking, diabetes, and hypertension were the risk factors among white and black women. Physical inactivity, nonuse of replacement hormones, serum cholesterol, and overweight were the additional risk factors among white women. Adjusted for demographic risk factors, overall PARs for study risk factors were 41.2% for white men, 60.5% for white women (with five risk factors only), 49.2% for black men, and 71.2% for black women. Much IHD mortality attributable to individual risk factors is caused by those factors in combination with other risk factors; relatively little mortality is attributable to each risk factor in isolation. Analysis that does not examine risk factor combinations may greatly overestimate PARs associated with individual risk factors.

Is food insufficiency associated with health status and health care utilization among adults with diabetes?

OBJECTIVES

Preliminary studies have shown that among adults with diabetes, food insufficiency has adverse health consequences, including hypoglycemic episodes and increased need for health care services. The purpose of this study was to determine the prevalence of food insufficiency and to describe the association of food insufficiency with health status and health care utilization in a national sample of adults with diabetes.

METHODS

We analyzed data from adults with diabetes ( n =1,503) interviewed in the Third National Health and Nutrition Examination Survey. Bivariate and multivariate analyses were used to examine the relationship of food insufficiency to self-reported health status and health care utilization.

RESULTS

Six percent of adults with diabetes reported food insufficiency, representing more than 568,600 persons nationally (95% confidence interval, 368,400 to 768,800). Food insufficiency was more common among those with incomes below the federal poverty level (17% vs 4%, P < or = .001). Adults with diabetes who were food insufficient were more likely to report fair or poor health status than those who were not (63% vs 43%; odds ratio, 2.2; P=.05). In a multivariate analysis, fair or poor health status was independently associated with poverty, nonwhite race, low educational achievement, and number of chronic diseases, but not with food insufficiency. Diabetic adults who were food insufficient reported more physician encounters, either in clinic or by phone, than those who were food secure (12 vs 7, P<.05). In a multivariate linear regression, food insufficiency remained independently associated with increased physician utilization among adults with diabetes. There was no association between food insufficiency and hospitalization in bivariate analysis.

CONCLUSIONS

Food insufficiency is relatively common among low-income adults with diabetes and was associated with higher physician utilization.

Adolescent marijuana use and adult occupational attainment: a longitudinal study from age 18 to 28

This report uses data from a nationally representative longitudinal sample to examine the relationship of marijuana use by high school seniors to occupational attainment 10 years later. Analyses were conducted separately by gender, with and without controlling for other variables. Control variables, all measured when respondents were seniors, were: academic performance, educational aspirations, and occupational aspirations. Results indicate that the influence of marijuana use on occupational attainment is considerably different for males and females.

Nutritional and health consequences are associated with food insecurity among U.S. elderly persons

The purpose of this study was to examine the consequences associated with food insecurity for the nutritional and health status of the elderly in the United STATES: The data analyzed were from the Third National Health and Nutrition Examination Survey (1988-1994) and the Nutrition Survey of the Elderly in New York State (1994). Multiple logistic and linear regression analyses were used to assess the extent to which food-insecure elderly were likely to have lower nutrient intake, skinfold thickness, self-reported health status and higher nutritional risk. Regardless of food insecurity status, older people consumed less than the recommended dietary allowance for eight nutrients. Food-insecure elderly persons had significantly lower intakes of energy, protein, carbohydrate, saturated fat, niacin, riboflavin, vitamins B-6 and B-12, magnesium, iron and zinc, as well as lower skinfold thickness. In addition, food-insecure elderly persons were 2.33 (95% confidence interval: 1.73-3.14) times more likely to report fair/poor health status and had higher nutritional risk. These results indicate that food-insecure elderly persons have poorer dietary intake, nutritional status and health status than do food-secure elderly persons. It is necessary to ensure the nutritional well-being of all elderly persons who are at nutritional and health risk, including those who are food insecure and have even poorer nutritional and health status than those who are food secure.

Understanding needs is important for assessing the impact of food assistance program participation on nutritional and health status in U.S. elderly persons

This study aimed to assess the impact of food assistance programs on nutritional and health status of nutritionally needy elderly persons. Two cross-sectional and one longitudinal data sets were used: Third National Health and Nutrition Examination Survey (1988-94), Nutrition Survey of the Elderly in New York State (1994) and Longitudinal Study of Aging (1984-1990). Multiple logistic and linear regression analyses were used to examine whether food assistance participants among food insecure elderly (i.e., those whose needs for food assistance programs are met) have better nutrient intake, skinfold thickness and self-reported health status and less nutritional risk, hospitalization and mortality than nonparticipants (i.e., those whose needs are unmet) and whether the benefit is larger than that among food secure elderly persons. Across three data sets, food insecure elderly persons had poorer nutritional and health status than food secure elderly persons. Contrary to the hypotheses, among food insecure elderly persons, food assistance participants had similar or poorer nutrient intakes, skinfold thickness, nutritional risk, self-reported health status, hospitalization and mortality than nonparticipants. Food secure participants had similar nutritional and health status as food secure nonparticipants. Lack of information on the dynamic nature and changes in needs with program participation in the three data sets likely did not allow accurate estimation of the impact of food assistance participation. Different study designs, as well as theory and knowledge of needs that clarifies need status and its change within each older individual across an appropriate time interval, are necessary to accurately assess impacts of food assistance programs.

Building bridges between populations and samples in epidemiological studies

The increased use of rigorous population-sampling methods and the analysis of data from those samples in cross-sectional surveys, case-control studies, longitudinal-cohort investigations, and other epidemiological research efforts have raised important statistical issues for health analysts. We describe the origin, implications, and some plausible resolutions for several of these issues. Some of the main issues we consider include (a) establishing whom the sample represents; (b) using sample weights; (c) understanding the role of other important features, such as the use of sampling stratification and the selection of clustered groups of population members; and (d) finding ways to analyze study data with key sampling features in mind. Ultimately, resolution of all of these issues requires that analysts clearly define a reference population and then understand the role of design features in relating sample results to that population.

The potential impact of incentives on future blood donation behavior

BACKGROUND

It is important to assess the potential efficacy and safety of offering donation incentives as part of recruitment and retention programs.

STUDY DESIGN AND METHODS

In 1995, 7489 allogeneic donors responded to an anonymous mail survey that inquired about demographics, donation history, infectious disease risks, and the potential appeal of incentives.

RESULTS

The projected net effect of offering blood credits and medical testing would be to motivate, respectively, 58 percent and 46 percent of donors to return, whereas offering an item of limited value would motivate 20 percent to do so. First-time and younger donors reported more frequently than repeat or older donors that incentives would appeal to them. Donors attracted by cash were 60 percent more likely to have a risk for transfusion-transmissible infections (p = 0.03). Although not statistically significant, the odds of being an at-risk donor were higher among individuals attracted by tickets to events (OR 1.5) and extra time off work (OR 1.2).

CONCLUSION

These findings suggest that offering blood credits and (though to a lesser extent) items of limited value could be safe and effective strategies for retaining donors. Although medical tests were found to have broad appeal, studies are needed to identify tests in which donors would be most interested.

The association of tobacco and alcohol consumption with the use of health care services in Spain

BACKGROUND

Information on the impact of tobacco and alcohol consumption on the use of health services is scant and partially inconsistent. This paper examines the relationship between tobacco and alcohol consumption and the use of health care services in Spain.

METHODS

Data were drawn from the 1993 Spanish National Health Survey, covering a random 21,120-person representative sample of Spain's noninstitutionalized population ages 16 years and older. Information was obtained through home-based interviews.

RESULTS

Compared with never smokers, male smokers of more than 20 cigarettes/day tend to be hospitalized more frequently (odds ratio (OR) 1.31; 95% confidence limits (CL) 0.89-1.93) and make greater use of hospital emergencies (OR 1.51; 95%CL 1.13-2.01; P < 0.01). Among female smokers of more than 20 cigarettes/day, hospitalizations (OR 1.62; 95%CL 0.80-3.26) and medical visits (OR 1. 35; 95%CL 0.79-2.30) are also higher than among never smokers, although the associations do not reach statistical significance. Compared with never smokers, ex-smokers of both sexes make greater use of health care services (P < 0.01 for most services). There is a negative dose-response relationship (P < 0.001) between alcohol consumption and utilization of hospital and ambulatory services, for both sexes. Results are reasonably consistent across all age groups and are observed after adjustment for the principal confounding factors. We have found no evidence of a tobacco-alcohol interaction with the use of health care services.

CONCLUSIONS

Smokers and ex-smokers make greater use of health care services. Control of smoking might reduce the use of such services and the ensuing human and economic costs. However, as alcohol consumption increases, the use of health care services decreases. This finding should not be used to promote even the moderate consumption of alcoholic drinks.

Dietary vitamin C and the risk for periodontal disease

BACKGROUND

Vitamin C has long been a candidate for modulating periodontal disease. Studies of scorbutic gingivitis and the effects of vitamin C on extracellular matrix and immunologic and inflammatory responses provide a rationale for hypothesizing that vitamin C is a risk factor for periodontal disease.

METHODS

We evaluated the role of dietary vitamin C as a contributing risk factor for periodontal disease utilizing the Third National Health and Nutrition Examination Survey (NHANES III) which is representative of the U.S. civilian, non-institutionalized population.

RESULTS

A sample of 12,419 adults (20 to 90+ years of age), with dental measurements and assessment of dietary information as well as demographic and medical histories were included in the studies. Dietary vitamin C was estimated by a 24-hour dietary record. Individuals with periodontal disease were arbitrarily defined as those who had mean clinical attachment levels of > or =1.5 mm. Using multiple logistic regression analysis, we found a relationship between reduced dietary vitamin C and increased risk for periodontal disease for the overall population (odds ratio [OR] = 1.19; 95% CI: 1.05 to 1.33). Current and former tobacco users who were taking less dietary vitamin C showed an increased risk of periodontal disease with OR of 1.28, 95% CI: 1.04 to 1.59 for former smokers, and an OR of 1.21, 95% CI: 1.02 to 1.43 for current tobacco users. There was a dose-response relationship between the levels of dietary vitamin C and periodontal disease with an OR of 1.30 for those taking 0 to 29 mg of vitamin C per day, to 1.16 for those taking 100 to 179 mg of vitamin C per day as compared to those taking 180 mg or more of vitamin C per day.

CONCLUSION

Dietary intake of vitamin C showed a weak, but statistically significant, relationship to periodontal disease in current and former smokers as measured by clinical attachment. Those taking the lowest levels of vitamin C, and who also smoke, are likely to show the greatest clinical effect on the periodontal tissues.

Calcium and the risk for periodontal disease

BACKGROUND

Dietary calcium has long been a candidate to modulate periodontal disease. Animal as well as human studies of calcium intake, bone mineral density, and tooth loss provide a rationale for hypothesizing that low dietary intake of calcium is a risk factor for periodontal disease.

METHODS

We evaluated the role of dietary calcium intake as a contributing risk factor for periodontal disease utilizing the Third National Health and Nutrition Examination Survey (NHANES III), which is representative of the U.S. civilian non-institutionalized population. Dietary calcium intake was determined from a 24-hour dietary recall. The U.S. Department of Agriculture Nutrient Database was used as a source of nutrient composition data. Periodontal disease was measured by attachment loss. In addition, serum calcium was assessed using venous blood samples. Logistic regression analysis was used to examine the association between periodontal disease and dietary calcium intake or serum calcium levels after adjusting for covariants including age, gender, tobacco consumption, and gingival bleeding.

RESULTS

The association of lower dietary calcium intake with periodontal disease was found for young males and females (20 to 39 years of age), and for older males (40 to 59 years of age). The relationship between low dietary calcium intake and increased levels of periodontal disease showed an estimated odds ratio (OR) of 1.84 (95% CI: 1.36 to 2.48) for young males, 1.99 (95% CI: 1.34 to 2.97) for young females, and 1.90 (95% CI: 1.41 to 2.55) for the older group of males. These odds ratios were adjusted for gingival bleeding and tobacco consumption. The dose response was also seen in females, where there was 54% greater risk of periodontal disease for the lowest level of dietary calcium intake (2 to 499 mg) and 27% greater risk in females who took moderate levels of dietary calcium (500 to 799 mg) as compared to those who took 800 mg or more dietary calcium per day. A statistically significant association between low total serum calcium and periodontal disease was found in younger females aged 20 to 39 with OR = 6.11 (95% CI: 2.36 to 15.84) but not for males or older females, after adjusting for tobacco use, gingival bleeding, and dietary calcium intake.

CONCLUSIONS

These results suggest that low dietary intake of calcium results in more severe periodontal disease. Further studies will be needed to better define the role of calcium in periodontal disease and to determine the extent to which calcium supplementation will modulate periodontal disease and tooth loss.

Community-based studies of Alzheimer's disease: statistical challenges in design and analysis

Alzheimer's disease is a chronic disease, primarily of the elderly, characterized by progressive dementia and eventual death. Community-based studies will likely provide a better representation of the spectrum of disease than will studies drawn solely from clinical sources, because an unknown and possibly substantial fraction of the cases do not come to the attention of the medical care system, or are diagnosed only very late in the disease. Community-based studies will provide not only more accurate estimates of prevalence and incidence, but also more directly comparable unaffected people for studies of risk factors for onset and progression. Such studies are likely to consist of a census component where relatively inexpensive but useful auxiliary information is collected and a probability sample from the census, with the detailed and costly clinical diagnosis of Alzheimer's disease restricted to the sample. The statistician faces challenges both in designing a sample that meets multiple objectives efficiently and in analysing data from the resulting complex survey designs.

Analyzing health surveys for cancer-related objectives

Large-scale health surveys conducted by government agencies record information on a large number of health-related variables. We review the use of these data for performing analyses that address cancer-related objectives. After describing the conduct of a large-scale health survey (the third National Health and Nutrition Examination Survey [NHANES III]), we discuss some of the issues involved in analyzing data collected in such a survey. In particular, the use of sample weights in the analysis and the importance of accounting for the complex survey design when estimating standard errors are discussed. Six applications are then presented that involve the following: 1) estimating demographic factors associated with snuff use, 2) estimating the association of type of health insurance with the probability of receiving a digital rectal examination, 3) estimating the association of body iron stores with the probability of later developing cancer, 4) estimating the changing rates of mammography screening in the United States between 1987 and 1992, 5) evaluating smoking and alcohol consumption as risk factors for digestive cancer by use of a population-based, case-control study, and 6) evaluating a randomized community-intervention experiment to encourage smoking cessation. These applications use data from the National Health Interview Survey, the NHANES I Epidemiologic Followup Study, the 1986 National Mortality Followback Survey, and the Community Intervention Trial for Smoking Cessation. The availability of public-use data files is discussed for surveys sponsored by the U.S. government that collect health-related information. We demonstrate that statistical methods and computer software are available for analyzing public-use data files of surveys to address different types of cancer-related objectives.

Alcohol and prostate cancer in the NHANES I epidemiologic follow-up study. First National Health and Nutrition Examination Survey of the United States

PURPOSE

We prospectively investigated the association between alcohol consumption and prostate cancer in the Epidemiologic Follow-up Study (NHEFS) of the first National Health and Nutrition Examination Survey (NHANES I).

METHODS

There were two cohorts: 1) Cohort I, followed from baseline (1971-75) through 1992, included 5766 men ages 25-74 years (median follow-up = 17 years); and 2) Cohort II, followed from the first follow-up round for Cohort I (1982-84) through 1992, included the 3868 men in Cohort I free of prostate cancer in 1982-84 (median follow-up = 9 years). Alcohol consumption was assessed at baseline as usual consumption, and at follow-up as usual consumption and as distant past consumption at the ages of 25, 35, 45, and 55.

RESULTS

There were 252 incident cases of prostate cancer. Consistent with most previous studies, we found no significant associations between usual total alcohol consumption and prostate cancer in Cohorts I or II [p = non significant (NS)], except for a significant inverse association at the heaviest level of drinking in Cohort II [relative risk (RR) = 0.23, 95% confidence interval (CI) = 0.06-0.95]. Further study of heavy drinkers in Cohort II revealed significant inverse associations between distant past heavy drinking (defined as > 25 drinks/week) and prostate cancer at age 25 (RR = 0.20, 95% CI = 0.06-0.63), age 35 (RR = 0.30, 95% CI = 0.12-0.77), and age 45 (RR = 0.39, 95% CI = 0.17-0.93), but not at age 55 (RR = 0.43, 95% CI = 0.17-1.10).

CONCLUSIONS

These results suggest that it may be important to consider distant past alcohol consumption in etiologic studies of prostate cancer. However, our results were based on small numbers of cases who were heavy drinkers and require replication.

Factors associated with seat belt use: an evaluation from the Ontario Health Survey

This study examines the factors associated with seat belt use for drivers and passengers in Ontario. Using the 1990 Ontario Health Survey, a population-based survey of non-institutionalized Ontario residents, factors associated with seat belt use among drivers and passengers were identified and are reported as unadjusted and adjusted odds ratios (OR; 95% CI). Seat belt non-use in Ontario drivers was most strongly associated with younger age (p < 0.0001), high risk health behaviours (drinking and driving (OR: 2.43), speeding (OR: 2.04), smoking (OR: 1.66)), being male (OR: 1.87), living in northern (OR: 1.80) or rural (OR: 1.69) regions, and achieving lower education (OR: 1.46). Seat belt non-use in passengers was associated with younger age (p < 0.0001), smoking (OR: 1.62), being male (OR: 1.68), living in northern (OR: 1.63) or rural (OR: 1.46) regions, and low education (OR: 1.31). Vehicular trauma is a major public health problem and seat belt use has been shown to reduce injuries in the event of a crash. Any strategy to increase seat belt use in Ontario should be targeted to involve both drivers and passengers. Attention should be paid to increasing seat belt usage by younger adults, males, and especially those living in northern and rural regions.

Who is driving when unrestrained children and teenagers are hurt?

This paper examines driver characteristics in crashes where child and teenage motor vehicle crash victims were injured, in particular factors that determine whether or not the victim was restrained. Analyzing the data on children and teenagers who are injured revealed that the presence of a second adult in a vehicle increased the likelihood that these passengers were unrestrained. Other findings are more predictable: victim restraint use generally mirrored driver restraint use; a male driver, a young driver, a drinking driver, a speeding or reckless driver, an unlicenced or suspended driver, and a night-time trip each independently raised the odds that child and teenage passengers were not restrained when they were injured.

Are sleep complaints an independent risk factor for myocardial infarction?

PURPOSE

To investigate whether subjective sleep complaints are an independent predictor of myocardial infarction (MI) in a community of older adults and to gain clues as to why the association between sleep complaints and incident MI exists.

METHODS

Using longitudinal data from the Piedmont study on 2960 adults aged 65 or older who were free of symptomatic heart disease at baseline, we screened 19 potential confounders to determine if any, alone or in combination, could explain the observed relationship between incident MI and sleep complaints.

RESULTS

During the three-year follow-up period, there were 152 incident MIs. Restless sleep (incidence density ratio (IDR) = 1.58, 95% confidence interval (CI) = 1.11, 2.24) and trouble falling asleep (IDR = 1.68, 95% CI = 1.09, 2.60) predicted incident MI after adjusting for age, gender, and race. IDRs were not substantially impacted by controlling for smoking, blood pressure, diabetes or obesity. After adjustment for education, number of prescription medicines, self-rated health, and depression score, all IDRs were nullified. In particular, self-rated health and depression were strong independent risk factors for MI.

CONCLUSIONS

A subjective sleep complaint increases the likelihood of a first MI in older adults without overt coronary heart disease (CHD) independently of classic coronary risk factors and appears to be a marker for a syndrome of depression and malaise that may have a causal relationship to MI.

Household participation in the Food Stamp and WIC programs increases the nutrient intakes of preschool children

We examined the effects of household participation in the Food Stamp and WIC Nutrition Programs on the nutrient intakes of preschoolers using data from the 1989-1991 Continuing Survey of Food Intake by Individuals. Nonbreastfeeding children, 1-4 y of age, with 3 d of dietary data and whose households had incomes < 130% of the poverty level were included in the study sample (n = 499). Nutrient adequacy ratios for each of 15 nutrients were the dependent variables in multiple regression models that controlled for the following: age, sex and ethnicity of the individual; income, size and location of the household; schooling of the household head; home ownership; school lunch and breakfast participation; and season in which the interview was conducted. WIC benefits positively influenced (P < 0. 05) the intakes of 10 nutrients. For iron and zinc, the average increase due to WIC represented 16.6 and 10.6%, respectively, of the preschooler recommended dietary allowance (RDA) for these nutrients. The same analyses of the Food Stamp Program revealed increases in five nutrients. For iron and zinc, the average increase due to Food Stamps represented 12.3 and 9.2%, respectively, of the preschooler RDA. The effects of the WIC Program on the intakes of iron and zinc were greater than that of cash income, and neither program affected the intakes of fat, saturated fat or cholesterol.

Biomedical risk factors for hospital admission in older adults

OBJECTIVES

This study examines the influence of risk factors such as cigarette smoking, blood pressure, serum cholesterol, or chronic illness on frequency of hospital admission in a population-based sample.

METHODS

Data from the National Health and Nutrition Examination Survey I Epidemiologic Followup Study for 6,461 adults aged 45 years and older were used to assess the influence of risk factors measured by interview, physical examination, and laboratory tests on frequency of hospital admission over a 12- to 16-year follow-up period. Cox proportional hazard regressions were estimated separately for men and women and for ages 45 to 64 years and 65 years and older. SUDAAN software was used to correct for clustering, stratification, unequal weighting, and multiple observations per respondent.

RESULTS

Risk of hospitalization was higher for current but not former smokers (relative risk [RR] = 1.17-1.34 for different age-sex groups; P < 0.01), higher blood pressure (RR = 1.25-1.28 for ages 45-64; RR = 1.07-1.15 for ages 65 and older; P < 0.01), and lower serum albumin (RR = 1.08-1.14; P < 0.01). Diabetes, lung conditions, heart attack, and ulcer each were associated with higher risk in at least three of the four age-sex groups, as was arthritis among the middle-aged (45-64 years). Serum cholesterol was not associated with hospitalization.

CONCLUSIONS

Chronic conditions with high morbidity as well as many factors associated with mortality are associated with a higher frequency of hospitalization.

Alcohol use and wages: new results from the National Household Survey on Drug Abuse

A recent study published in the Journal of Health Economics by French and Zarkin [French, M.T., Zarkin, G.A., 1995. Is moderate alcohol use related to wages? Evidence from four worksites, Journal of Health Economics 14, 319-344] found evidence of a positive, inverse-U-shaped relationship between wages and alcohol consumption for individuals at four worksites. In this paper, we attempted to replicate French and Zarkin's findings using a combined sample of prime-age workers from the 1991 and 1992 National Household Surveys on Drug Abuse (NHSDA). Whereas French and Zarkin found that individuals who consume approximately 1.5 to 2.5 drinks per day have higher wages than non-drinkers and heavy drinkers, we found no evidence of a turning point at this consumption level for either men or women. Our results do suggest that men who use alcohol have approximately 7% higher wages than men who do not drink, and this apparent wage premium is approximately the same over a wide range of alcohol consumption. For women, the estimated alcohol use premium is approximately half as large as for men and is statistically insignificant.

Moderate wine consumption is associated with decreased odds of developing age-related macular degeneration in NHANES-1

OBJECTIVE

To determine the association between alcohol intake and the risk of developing age-related macular degeneration (AMD).

DESIGN

Case control study.

PARTICIPANTS

The sample consisted of 3072 adults 45 to 74 years of age with macular changes indicative of AMD who participated in a nationally representative sample of the first National Health Nutrition and Examination Survey (NHANES-1) between 1971 and 1975: (a) the ophthalmology data set and (b) the medical history questionnaire.

MAIN OUTCOME MEASURES

Alcohol intake and the risk of developing AMD were measured. AMD was determined by staff at the National Eye Institute by fundoscopy examination using standardized protocol.

RESULTS

Overall, 184 individuals (6%) had AMD. We observed a statistically significant but negative association between AMD and the type of alcohol consumed in a bivariate model (OR 0.86; 95% CI 0.73, 0.99). In the same model, age maintained a consistently strong association with AMD (OR 1.08; 95% CI 1.06-1.11; P < .001). Among the different types of alcohol consumed in NHANES-1 (beer, wine, and liquor), the effect of wine, either alone (OR 0.66; 95% CI 0.55-0.79) or in combination with beer (OR 0.66; 95% CI 0.55-0.79) or liquor (OR 0.74; 95% CI 0.63-0.86), dominated the negative association observed between AMD and alcohol type. Additionally, a statistically significant and negative association between wine and AMD was noted after adjusting for the effect of age, gender, income, history of congestive heart failure, and hypertension (OR 0.81; 95% CI 0.67-0.99).

CONCLUSION

Moderate wine consumption is associated with decreased odds of developing AMD. Health promotion and disease prevention activities directed at cardiovascular disease may help reduce the rate of AMD-associated blindness among older people. The nature and pathophysiology of this association warrant further investigation.

Nutrient intakes of individuals from food-insufficient households in the United States

OBJECTIVES

Understanding the nutritional consequences of food insufficiency is important for informed policy-making that addresses the problem of domestic hunger. This study estimated the extent to which individuals from food-insufficient households were likely to have low intakes of energy and 14 other nutrients.

METHODS

The diets of pre-schoolers, adult women, and the elderly were analyzed with 24-hour recall data from the 1989 through 1991 Continuing Survey of Food Intake by Individuals. Logistic regression analysis was used to study the association of self-reported household food insufficiency with nutrient intakes below 50% of the recommended daily allowance.

RESULTS

For adult women, food insufficiency was significantly associated with low intakes of eight nutrients, including energy, magnesium, and vitamins A, E, C, and B6. Elderly individuals in the food-insufficient group were also more likely to have low intakes of eight nutrients, including protein, calcium, and vitamins A and B6. Household food insufficiency was not significantly associated with low intakes among preschoolers.

CONCLUSIONS

The results validate the use of self-reported hunger measures in nutritional surveillance and highlight nutrients of concern for food assistance and nutrition education efforts targeted at individuals from food-insufficient households.

Relative weight and mortality in U.S. blacks and whites: findings from representative national population samples

PURPOSE

To examine the impact of relative weight on mortality in black and white men and women.

METHODS

Two representative national populations samples were used: the NHANES-I Epidemiologic Follow-up Study (NHEFS), and the National Health Interview Survey (NHIS). The principal analysis focused on 13,242 participants in the NHEFS and 114,954 in the NHIS. Minimum mortality was estimated from both categorical analysis and a logistic model.

RESULTS

Minimum mortality ranged from a body mass index (BMI) of 25 to 32 kg/m2. The model-estimated BMI of minimum mortality for NHEFS was 27.1 (24.8-29.4, 95% CI), 26.8 (24.7-28.9, 95% CI), 24.8 (23.8-25.9, 95% CI) and 24.3 (23.2-25.4, 95% CI); for black men, black women, white men and white women, respectively, whereas for NHIS the corresponding values were 30.2 (24.8-35.6, 95% CI) 26.4 (24.2-28.7, 95% CI), 27.1 (25.5-28.7, 95% CI), and 25.6 (24.2-27.0, 95% CI). In all groups the shape of the relative risk curve was virtually identical and a broad range of BMI values in the middle of the distribution was associated with low relative mortality risk. Averaging the results from both surveys, the observed BMI of minimum risk was 3.1 kg/m2 higher in black men and 1.5 kg/m2 higher in black women than in their white counterparts; when adjusted for covariates these differences were only of borderline statistical significance, however.

CONCLUSIONS

Because of the wide range of BMI values associated with low risk, and the consistency of the point of the up-turn in risk, group specific definitions of optimal values do not appear to be warranted.

Reform of primary health care: the case of Spain

Different approaches to health reform are proposed in many countries to overcome inefficiencies in care delivery. This paper assesses an incremental reform initiated in Spain 10 years ago, which sought to improve the efficiency of the entire health system through changes in the organization and delivery of primary care. In this study, aspects of accessibility, comprehensiveness, longitudinality and technical quality of reformed versus unreformed care were assessed for respondents to a household interview survey conducted in the Basque Region of Spain in 1992. According to this study, aspects of care such as longitudinality and technical quality seemed improved with the reform, whereas other aspects such as accessibility and comprehensiveness remained unchanged. The authors conclude that system related characteristics (more associated with access and comprehensiveness) may be impeding the achievement of the goals of the reform and argue that attempts to encourage more autonomy of care delivery may be required.

Some correlates of self-rated health for Australian women

OBJECTIVES

This study aimed to identify some of the correlates of self-rated health for young to middle-aged Australian women.

METHODS

Regression analyses were based on a 4-year longitudinal study using a random sample of Sydney women 20 to 59 years of age at baseline. Participants were interviewed in 1986/87 and 1990.

RESULTS

Cross-sectional relationships between self-assessed health and other health measures varied significantly by age, although physical health was a common correlate. Sixty-three percent of participants reported a similar rating of health over the 4-year period between the surveys. Changes in self-assessed health were sensitive to chronic disease. Also, participants' self-ratings of health were related to their subsequent chronic disease status.

CONCLUSIONS

Self-rated health reflects a complex process of internalized calculations that encompass both lived experience and knowledge of disease causes and consequences. Women seem to take into consideration a broad range of factors, including lifestyle, vitality, mental attitude, and age, and, if they have a health condition, the chronicity of their disease, duration since diagnosis, and treatment.

Innovative strategies using SUDAAN for analysis of health surveys with complex samples

Large-scale health surveys provide a wealth of information for addressing problems in health sciences research. Designed for multiple purposes, these surveys frequently have large sample sizes and extensive measurements of demographic and socioeconomic characteristics, risk factors, disease outcomes and health care service use and costs. Complex features of the sampling design typically employed to select the survey sample, coupled with the vast amount of information available from the survey database, underlie issues that must be addressed during data processing and analysis. Numerous articles in the literature have focused on the debate of whether or not, and how, to control for features of the sample design during data analysis. Traditional statistical methods for simple random samples and the software that accompanies them have historically not had the capacity to account for the survey design. Recent advancements in statistical methodology for survey data analysis have greatly expanded the analytical tools available to the survey analyst. Commercial software packages that incorporate these methods offer the analyst convenient ways for applying such tools to large survey databases in an easy and efficient manner. We present an overview of analysis strategies for survey data and illustrate their application via the SUDAAN software system. Examples for analyses are provided through data from two large US health surveys, the National Health Interview Survey and the Longitudinal Study of Aging. Questions of both a cross-sectional and longitudinal nature are addressed. The examples involve logistic regression, time-to-event analysis, and repeated measures analysis.

Hospital and nursing home use in the last three months of life

We describe decedents' days of care and changes in residence due to episodes of hospital and institutional care in the last 90 days of life. Data are from the National Institute on Aging's Survey of the Last Days of Life (SLDOL) with informants' responses for a sample of 1,227 decedents age 65 years and older from Fairfield County, Connecticut. Overall, three-fourths of the decedents made only one transition to or from a health care facility in the last 3 months of life and another 10% made two or three transitions. Hospital days decreased slightly with age, but nursing home days increased dramatically. Physical disability, lack of social contacts, incontinence, and a diagnosis of dementia were significantly associated with a long, terminal, institutional stay. Future cost savings are likely to be achieved more through the reduction of nursing home care in the last months of life than in a reduction of hospital care.

Use of care and subsequent mortality: the importance of gender

OBJECTIVE

In light of recent discussions on access that have emphasized the need to relate access measures to outcomes, we examined the relationship between three self-reported utilization and access to care measures and the risk of subsequent mortality. DATA SOURCES AND DESIGN: A nationally representative sample from the first National Health and Nutrition Examination Survey that included adults 25-64 years of age without publicly funded health insurance was followed prospectively from initial interview in 1971 through 1975.

DATA COLLECTION

Complete baseline and follow-up information was obtained on 4,491 persons (90 percent). Baseline access and use was assessed with answers to three questions: having a usual source of care, obtaining a general checkup, and not obtaining needed care (or forgone care). The relationships between the access and use measures and mortality by 1987 in men and women were examined using survival analyses. The analyses adjusted for race, and for baseline age, education, income, residence, insurance status, employment status, the presence of morbidity on examination, self-rated health, smoking status, leisure exercise, alcohol consumption, and obesity.

PRINCIPAL FINDINGS

After adjusting for all other baseline variables, not obtaining a general checkup was associated with higher mortality in women (hazard ratio = 1.64 [95% confidence interval = 1.16, 2.32]), but not in men (hazard ratio = 1.07 [95% confidence interval = 0.80, 1.42]). Reporting a usual source of care and forgone care were not related to subsequent mortality in either women or men.

CONCLUSIONS

Reporting a general checkup is an outcome-related utilization measure in women only. Further development of access and use indicators should address gender differences in health care use.

Health consequences of social-role careers for women: a life-course perspective

This study aimed to demonstrate the importance of early adulthood social-role participation in determining the health of women over the life course. It used Australian Family Project data from a nationally representative proportional stratified sample of 1-in-1000 women aged between 20 and 59 years in 1986-87 (n = 1678); and data from a follow-up survey conducted in 1990 using project participants who were living in Sydney at the time of the baseline (n = 291). Social-role participation was measured over each participant's life using retrospective histories of employment, marital status and parental status. Validity and reliability tests supported the use of these histories. Health was measured by retrospective accounts of serious chronic disease onset and indicators of self-rated health. After considering a broad range of confounders, early adulthood social-role careers were found to vary significantly in their risk of serious chronic disease and levels of self-rated health. These findings suggest that the influence of society's social structures on health for women goes beyond causes related to conventional socioeconomic differentials or arbitrary periods of social-role participation. The nature of the associations are complex and dynamic, involving both the mixture and timing of social events and transitions. This conclusion supports a life-course approach in which social careers are perceived as cumulative, providing women with lessons, liabilities and resources that influence the way they age and meet the realities of life.

Trends in obesity differences by educational level in Spain

Our objective was to study the trend in differences in the frequency of obesity by educational level in the general population 20 to 64 years of age. We used data from two cross-sectional health surveys carried out in 1987 and 1993 in representative samples of the Spanish population. We investigated the relation between obesity and educational level during the periods 1987 and 1993, taking into account the main factors confounding the relation. We used, as setting, the National Health Interview Surveys representative of the whole Spanish population. In both men and women, the highest odds ratios (ORs) for obesity were observed at lower educational levels. These differences increased in women between 1987 and 1993, while they decreased in men during the same period. Evidence of increased educational differences in the frequency of obesity indicates that future studies should focus on the evaluation and monitoring of this trend in the population.

The intergenerational relationship between mother's birthweight, infant birthweight and infant mortality in black and white mothers

The relationship between the birthweight of white and black mothers and the outcomes of their infants were examined using the 1988 National Maternal and Infant Health Survey. White and black women who were low birthweight themselves were at increased risk of delivering very low birthweight (VLBW), moderately low birthweight (MLBW), extremely preterm and small size for gestational age (SGA) infants. Adjustment for the confounding effects of prepregnant weight and height reduced the risks of all these outcomes slightly, and more substantially reduced the maternal birthweight associated risk of moderately low birthweight among white mothers. There was little effect of maternal birthweight on infant birthweight-specific infant mortality in white mothers; however, black mothers who weighed less than 4 lbs at birth were at significantly increased risk of delivering a normal birthweight infant who subsequently died. Although the risks for the various outcomes associated with low maternal birthweight were not consistently higher in black mothers compared with white mothers, adjustment for prepregnant weight and height had a greater effect in white mothers than in black mothers. We suggest that interventions to reduce the risks for adverse pregnancy outcomes associated with low maternal birthweight should attempt to optimise prepregnant weight and foster child health and growth.

Risk factors for injury: similarities and differences for traffic crashes and other causes

Actions to prevent injury may address either individual behaviour or population practice. Factors associated with injuries in a variety of circumstances may be best suited to prevention at the level of population whereas more specific causes would not be. Data from the Ontario Health Survey were used to assess the similarities and differences between risk factors and injury in traffic crashes, at work and in other circumstances. Respondents reporting injury were compared with age- and sex-matched controls reporting no injury for reported alcohol ingestion, binge drinking and drinking and driving. In addition, data on the use of licit and illicit drugs, occupation and other demographic factors were compared. Variables showing consistent associations with injury included drug use, some measures of alcohol ingestion and reports of co-existing chronic disease. Many variables showed a variable pattern, including occupation, education and income. In many instances, these differences reflected differences in exposure to risk. Efforts at prevention related to alcohol or medication could be broadly based, whereas for other factors, more targeted prevention efforts would be appropriate.

The disablement experienced by traumatically brain-injured adults living in the community

The disablement that occurs following traumatic brain injury (TBI) can be extensive and severe and consequently has been difficult to report on in a comprehensive and thorough manner. We were able to address this difficulty by analysing a sub group of data from the Canadian Health and Activity Limitation Survey (HALS) using the theoretical framework of disablement developed by the World Health Organization, the International Classification of Impairment, Disability and Handicap (ICIDH). There were 454 survey respondents (representing 12,290 in the Canadian population) with disability resulting from a TBI and a mean time post-injury of 13 years. Three handicaps identified in the ICIDH were the focus of the study: physical independence, work, social integration. The prevalence of long term handicap was very high with 66% of the sample reporting the need for ongoing assistance with some activities of daily living, 75% not working, and 90% reporting some limitations or dissatisfaction with their social integration. Multivariate regression analysis was used to investigate the determinants of the handicaps. The determinants included: age, gender, level of education, living alone, physical environment, and specific disabilities. The implications of these findings are discussed in relation to rehabilitation issues, the usefulness of the ICIDH as a model to investigate outcomes, and directions for future research.

The association between leisure-time physical activity and dietary fat in American adults

Relations between leisure-time physical activity and dietary fat were examined in a population-based probability sample of 29,672 adults in the 1990 Behavioral Risk Factor Surveillance System. Consumption of 13 high-fat food items and participation in physical activities were measured, and fat and activity scores were calculated. Dietary fat and physical activity were strongly and inversely associated. This association was independent of nine other demographic and behavioral risk factors. Etiologic researchers should consider that diet and physical activity can potentially confound each other, and creators of public health messages that target one behavior should consider including the other.

Depression and unemployment: panel findings from the Epidemiologic Catchment Area study

Studies that have found an association between unemployment and psychological depression often fail to establish the direction of causal influence. Analyses of Epidemiologic Catchment Area panel data revealed that of employed respondents not diagnosed with major depression at first interview, those who became unemployed had over twice the risk of increased depressive symptoms and of becoming clinically depressed as those who continued employed. Although the increase in symptoms was statistically significant, the effect on clinical depression was not, possibly because of the low power of the test. The reverse causal path from clinical depression at Time 1 to becoming unemployed by Time 2 was not supported. The unemployment rate in the respondent's community at time of interview was not related directly to psychological depression but appeared associated indirectly with depression via its impact on the risk of becoming unemployed. Implications for policy and further research were discussed.

Diet, indicators of kidney disease, and later mortality among older persons in the NHANES I Epidemiologic Follow-Up Study

OBJECTIVES

The purpose of this study was to determine whether diet adversely affected survival among 2572 older persons with indicators of kidney disease in a population-based cohort. Average follow-up time for survivors, of whom 1453 (57%) had died at analysis, was 14.5 years.

METHODS

Kidney disease indicators were a "yes" response to "Has a doctor ever told you that you have kidney disease or renal stones?" and/or trace or greater amounts of protein in urine. Dietary protein intakes were calculated from 24-hour recalls.

RESULTS

Cox proportional hazards models were used, stratified by sex, with age, body mass index, blood pressure, education, smoking status, total caloric intake, and diabetes mellitus as covariates. Relative risk of total mortality with an additional 15 g of protein per day was 1.25 (95% confidence interval [CI] = 1.09, 1.42) among White men with kidney disease indicators, vs 1.00 (95% CI = 0.95, 1.06) among those without them; relative risks of renal-related mortality were 1.32 (95% CI = 0.97, 1.79) and 0.95 (95% CI = 0.81, 1.11), respectively. No significant differences were found for White women.

CONCLUSIONS

Once chronic renal disease is present, diet may be associated with earlier mortality in White males.

Estimating Increment-Decrement Life Tables with Multiple Covariates from Panel Data: The Case of Active Life Expectancy

A fundamental limitation of current multistate life table methodology-evident in recent estimates of active life expectancy for the elderly-is the inability to estimate tables from data on small longitudinal panels in the presence of multiple covariates (such as sex, race, and socioeconomic status). This paper presents an approach to such an estimation based on an isomorphism between the structure of the stochastic model underlying a conventional specification of the increment-decrement life table and that of Markov panel regression models for simple state spaces. We argue that Markov panel regression procedures can be used to provide smoothed or graduated group-specific estimates of transition probabilities that are more stable across short age intervals than those computed directly from sample data. We then join these estimates with increment-decrement life table methods to compute group-specific total, active, and dependent life expectancy estimates. To illustrate the methods, we describe an empirical application to the estimation of such life expectancies specific to sex, race, and education (years of school completed) for a longitudinal panel of elderly persons. We find that education extends both total life expectancy and active life expectancy. Education thus may serve as a powerful social protective mechanism delaying the onset of health problems at older ages.