Life table methodology applied to NHANES II database for analysis of mortality associated with cholesterol/HDL ratios

This article describes the association between total cholesterol to high-density lipoprotein (HDL)-cholesterol ratios and all-cause mortality in a large cohort of Americans with nearly 17 years of follow-up. Detailed actuarial life table methodology was used. It concludes that the relationship is best described as a J-shaped curve.

Body mass index in mid-life women: relative influence of menopause, hormone use, and ethnicity

OBJECTIVE

To assess the relative influence of menopausal status and hormone use on body mass index (BMI) among a multiethnic sample of mid-life women.

DESIGN

Cross-sectional telephone survey conducted at seven sites where each site targeted an ethnic minority group and Caucasians as part of Study of Women's Health Across the Nation (SWAN).

SUBJECTS

A total of 7181 Caucasians, 3949 African-Americans, 1660 Hispanics, 562 Chinese Americans, and 803 Japanese Americans between ages of 40 and 55 y residing in or near Boston, Chicago, Detroit, Los Angeles, Newark, NJ, Oakland, CA, and Pittsburgh, PA.

MEASUREMENTS

Self-reported BMI based on weight in kg divided by height in m(2) menopausal status, physical inactivity, postmenopausal hormone use, ethnicity, and age in years.

RESULTS

Compared to premenopausal women (covariate adjusted M=27.3), women reporting a surgical menopause (M=28.2) or being in the perimenopausal transition (M=27.7 for early and 27.9 for late perimenopause) had higher BMI. Women reporting a natural menopause (M=27.4) did not have a higher BMI than premenopausal women, after adjusting for chronological age and other covariates. Hormone use was associated with lower BMI (M=26.5 vs 27.3). A comparison of effect sizes showed that menopausal status (F=13.1), followed by chronological age (F=24.0), were the least powerful predictors of BMI, whereas the more powerful predictors were physical activity level (F=1377.1) and ethnicity (F=400.5).

CONCLUSIONS

The menopausal transition affects body mass index in mid-life, but the effect is small relative to other influences. Interventions to increase physical activity are highly recommended to prevent increases in adiposity common in mid-life.

HIV-1 co-infection in children hospitalised with tuberculosis in South Africa

SETTING

Hospitals associated with the Department of Paediatrics at the University of the Witwatersrand, Johannesburg, South Africa.

OBJECTIVES

To define the prevalence of human immunodeficiency virus (HIV) co-infection and differences in clinical presentation between HIV-infected and non-infected hospitalised children with tuberculosis.

DESIGN

Children were prospectively enrolled between August 1996 and January 1997.

RESULTS

Of 161 children enrolled, 42% were HIV-infected, including 67/137 with pulmonary tuberculosis (PTB) and 1/24 with extra-pulmonary disease (EPTB). Positive microscopy or bacteriology did not differ by HIV status for children with either PTB or EPTB. Although age did not differ between HIV-infected and non-infected children with PTB, non-HIV-infected children with EPTB were significantly older than those with PTB only (median age 32 months vs 14.5 months, P = 0.004). Chronic weight loss, malnutrition and the absence of BCG scarring were more common in HIV-infected children with PTB. HIV-infected children were also more likely to show cavitation (P = 0.001) and miliary TB (P = 0.01) on chest X-ray. Reactivity to tuberculin (> or = 5 mm and > or = 10 mm in HIV-infected and non-infected children, respectively) was significantly lower in HIV-infected children, as were CD4+ lymphocyte levels. The mortality rate during the study was 13.4% in HIV-infected children compared with 1.5% in non-HIV-infected children (P = 0.03).

CONCLUSIONS

There is a high prevalence of HIV co-infection in children with TB. Progressive PTB and death are more common in HIV-infected children. Tuberculin skin testing is of limited use in screening for TB in HIV-infected children even when using a cut-point of > or = 5 mm.

Relational database design

Relational databases are the predominant method for storing repetitive data in computers because they allow efficient and flexible storage of that data. While medical directors and underwriters are more likely to use a spreadsheet than a database program to analyze their business, the data they wish to study are often stored in corporate databases. Or the data may be complex enough to require being keyed into or downloaded into a personal computer (PC) database program for storage, even if the data are then output to a spread-sheet for numerical analysis. In many circumstances, one can benefit from an understanding of efficient database design. After a brief overview, the reader is led step-by-step through a practical explanation of database design, from a flat file to a relational model.

Life table analysis

Life table analysis is an effective way to present and evaluate survival data in a number of circumstances. The summary tables and survival curves that are commonly seen in the medical literature are frequently derived through life table analysis. A basic understanding of life table construction is of benefit to the medical director who wishes to abstract comparative mortality data from study reports. This article reviews the basic methodology of life table analysis with a particular focus on handling right-censored study participants as withdrawals.