PubMed Health⌕ Search
Home/Archive/PubMed

PubMed · 11338312

Data, design, and background knowledge in etiologic inference.

J M Robins
View original source ↗Export metadata ↓

Abstract

I use two examples to demonstrate that an appropriate etiologic analysis of an epidemiologic study depends as much on study design and background subject-matter knowledge as on the data. The demonstration is facilitated by the use of causal graphs.

Explore related subjects

Case-Control StudiesCohort StudiesData CollectionEndometrial NeoplasmsEpidemiologic StudiesFemaleHormone Replacement TherapyHumansPostmenopauseRandomized Controlled Trials as TopicResearch DesignStatistics as TopicJournal Article

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

J M Robins. 2001. Data, design, and background knowledge in etiologic inference.. https://doi.org/10.1097/00001648-200105000-00011

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related citations

PubMed2002

Interleukin-10 promoter polymorphisms and cervical cancer risk in Korean women.

The aims of this study were to determine whether polymorphisms of the interleukin (IL)-10 promoter might be associated with an increased risk of cervical cancer, and further whether systemic IL-10 concentration might be influenced by the genotypes in Korean women. Peripheral blood samples from patients with invasive cervical cancer (ICC, n=144) and non-cancer controls (NCC, n=179) were used to detect three biallelic IL-10 promoter polymorphisms at -1082, -819, and -592 sites by polymerase chain reaction-restriction fragment length polymorphism assay using MnlI, MaeIII and RsaI, respectively. The IL-10 serum concentration was measured with enzyme-linked immunosorbent assay. We compared the distribution of genotypes in both groups, evaluated the serum IL-10 level according to the genotypes, and analyzed the association of these polymorphisms with the risk of cervical cancer. The genotype at the -1082 site exhibited only the *A homozygote, and only two haplotypes were found in Korean women, one being -1082*A/-819*T/-592*A (ATA) and the other -1082*A/-819*C/-592*C (ACC). No association was found for IL-10 promoter polymorphisms among the ICC patients in comparison with the NCC subjects. The risk of cervical cancer was not increased in either genotype and the IL-10 serum concentration was not influenced by the genotypes in either group. Polymorphisms of the IL-10 promoter do not appear to be associated with cervical cancer risk or systemic IL-10 production in Korean women.

Case-Control Studies↗
PubMed2002

The association between schizophrenia and cancer: a population-based mortality study.

BACKGROUND: For most of this century there has been speculation that persons diagnosed with schizophrenia have a reduced incidence of cancer. OBJECTIVE: To determine if a history of cancer was more common in persons diagnosed with schizophrenia when compared with the general population, controlling for known risk and demographic factors. DESIGN: We used the 1986 National Mortality Followback Survey (NMFS) which sampled 1% of all deaths in the US from that year. Data were obtained from death certificates and records of hospitalizations in the last year of life. Additional health and demographic data were obtained through interviews with decedents' families and other informants. We compared persons diagnosed with schizophrenia (n=130) to individuals without schizophrenia (n=18,603) and used logistic regression to determine the odds ratio for the occurrence of cancer in persons diagnosed with schizophrenia. Adjustment for age at death was done to correct for the fact that persons diagnosed with schizophrenia die on average 10 years younger than the general population. MAIN OUTCOME MEASURE: A diagnosis of cancer on a hospital record or the death certificate. RESULTS: The unadjusted odds ratio for cancer among individuals with schizophrenia was 0.62 (95% confidence interval (CI) 0.40-0.96). After controlling for age, race, gender, marital status, education, net worth, smoking, and hospitalization in the year before death, we determined that the odds ratio for the diagnosis of cancer in persons with schizophrenia was 0.59 (95% CI 0.38-0.93). CONCLUSION: In this population-based study, we demonstrated a reduced risk of cancer among persons diagnosed with schizophrenia.

Case-Control Studies↗
PubMed2002

Effects of glutathione-S-transferase M1, T1, and P1 on childhood lung function growth.

The effects of glutathione-S-transferase (GST) M1, GSTT1, and GSTP1 genotypes on lung function growth were investigated in 1,940 children enrolled in the Children's Health Study as fourth graders (aged 8-11 years) in two cohorts during 1993 and 1996 and were followed annually over a 4-year period. Genotypes for GSTM1 and GSTT1 and GSTP1 codon 105 variants (ile105 and val105) were determined using DNA from buccal cell specimens. We used two-level regression models to estimate the effects of GSTM1, GSTT1, and GSTP1 genotypes on the adjusted annual average lung function growth. GSTM1 null was associated with deficits in annual growth rates for FVC (-0.21%; 95% confidence interval [CI], -0.40, -0.03) and FEV(1) (-0.27%; 95% CI, -0.50, -0.04). Children who were homozygous for the GSTP1 val105 allele had slower lung function growth (FVC -0.35%; 95% CI, -0.62, -0.07; and FEV(1) -0.34%; 95% CI, -0.68, 0.00) than children with one or more ile105 alleles. Children with asthma who were homozygous for the GSTP1 val105 allele had substantially larger deficits in FVC, FEV(1), and maximal mid-expiratory flow than children without asthma. The deficits in FVC and FEV(1) growth associated with both GSTM1 null and the GSTP1 val105 allele were largest and were statistically significant in non-Hispanic white children. We conclude that GSTM1 and GSTP1 genotypes are associated with lung function growth in school children.

Case-Control Studies↗