Reina Haque

A most stubborn bias: No adjustment method fully resolves confounding by indication in observational studies

OBJECTIVE To evaluate the effectiveness of methods that control for confounding by indication, we compared breast cancer recurrence rates among women receiving adjuvant chemotherapy with those who did not. STUDY DESIGN AND SETTING In a medical record review-based study of breast cancer treatment in older women (n=1798) diagnosed between 1990 and 1994, our crude analysis suggested that adjuvant chemotherapy was positively associated with recurrence (hazard ratio [HR]=2.6; 95% confidence interval [CI]=1.9, 3.5). We expected a protective effect, so postulated that the crude association was confounded by indications for chemotherapy. We attempted to adjust for this confounding by restriction, multivariable regression, propensity scores (PSs), and instrumental variable (IV) methods. RESULTS After restricting to women at high risk for recurrence (n=946), chemotherapy was not associated with recurrence (HR=1.1; 95% CI=0.7, 1.6) using multivariable regression. PS adjustment yielded similar results (HR=1.3; 95% CI=0.8, 2.0). The IV-like method yielded a protective estimate (HR=0.9; 95% CI=0.2, 4.3); however, imbalances of measured factors across levels of the IV suggested residual confounding. CONCLUSION Conventional methods do not control for unmeasured factors, which often remain important when addressing confounding by indication. PS and IV analysis methods can be useful under specific situations, but neither method adequately controlled confounding by indication in this study.

Nutritional Factors and Susceptibility to Arsenic-Caused Skin Lesions in West Bengal, India

There has been widespread speculation about whether nutritional deficiencies increase the susceptibility to arsenic health effects. This is the first study to investigate whether dietary micronutrient and macronutrient intake modulates the well-established human risk of arsenic-induced skin lesions, including alterations in skin pigmentation and keratoses. The study was conducted in West Bengal, India, which along with Bangladesh constitutes the largest population in the world exposed to arsenic from drinking water. In this case–control study design, cases were patients with arsenic-induced skin lesions and had < 500 μg/L arsenic in their drinking water. For each case, an age- and sex-matched control was selected from participants of a 1995–1996 cross-sectional survey, whose drinking water at that time also contained < 500 μg/L arsenic. Nutritional assessment was based on a 24-hr recall for major dietary constituents and a 1-week recall for less common constituents. Modest increases in risk were related to being in the lowest quintiles of intake of animal protein [odds ratio (OR) = 1.94; 95% confidence interval (CI), 1.05–3.59], calcium (OR = 1.89; 95% CI, 1.04–3.43), fiber (OR = 2.20; 95% CI, 1.15–4.21), and folate (OR = 1.67; 95% CI, 0.87–3.2). Conditional logistic regression suggested that the strongest associations were with low calcium, low animal protein, low folate, and low fiber intake. Nutrient intake was not related to arsenic exposure. We conclude that low intake of calcium, animal protein, folate, and fiber may increase susceptibility to arsenic-caused skin lesions. However, in light of the small magnitude of increased risks related to these dietary deficiencies, prevention should focus on reducing exposure to arsenic.

Arsenic in Drinking Water and Skin Lesions: Dose-Response Data from West Bengal, India

Background. Over 6 million people live in areas of West Bengal, India, where groundwater sources are contaminated with naturally occurring arsenic. The key objective of this nested case-control study was to characterize the dose-response relation between low arsenic concentrations in drinking water and arsenic-induced skin keratoses and hyperpigmentation. Methods. We selected cases (persons with arsenic-induced skin lesions) and age- and sex-matched controls from participants in a 1995–1996 cross-sectional survey in West Bengal. We used a detailed assessment of arsenic exposure that covered at least 20 years. Participants were reexamined between 1998 and 2000. Consensus agreement by four physicians reviewing the skin lesion photographs confirmed the diagnosis in 87% of cases clinically diagnosed in the field. Results. The average peak arsenic concentration in drinking water was 325 &mgr;g/liter for cases and 180 &mgr;g/liter for controls. The average latency for skin lesions was 23 years from first exposure. We found strong dose-response gradients with both peak and average arsenic water concentrations. Conclusions. The lowest peak arsenic ingested by a confirmed case was 115 &mgr;g/liter. Confirmation of case diagnosis and intensive longitudinal exposure assessment provide the basis for a detailed dose-response evaluation of arsenic-caused skin lesions.

Children's intellectual function in relation to arsenic exposure.

Background: Very little evidence exists concerning the possible impairment of childrens intellectual function in relation to arsenic exposure in utero and during childhood. Methods: We conducted a cross-sectional study among 351 children age 5 to 15 years who were selected from a source population of 7683 people in West Bengal, India, in 2001–2003. Intellectual function was assessed with 6 subtests from the Wechsler Intelligence Scale for Children as well as with the Total Sentence Recall test, the Colored Progressive Matrices test, and a pegboard test. Arsenic in urine and lifetime water sources (including during the pregnancy period) were assessed using measurements of samples from 409 wells. The test scores were analyzed with linear regression analyses based on the method of generalized estimating equations incorporating relevant covariates. Results: Stratifying urinary arsenic concentrations into tertiles, we found associations between arsenic and reductions in the adjusted scores of the vocabulary test (0, −0.14, −0.28; P for trend = 0.02), the object assembly test (0, −0.16, −0.24; P for trend = 0.03), and the picture completion test (0, −0.15, −0.26; P for trend = 0.02). These findings correspond to relative declines of 12% (95% confidence interval =0.4% to 24%) in the vocabulary test, 21% (−0.8% to 42%) in the object assembly test, and of 13% (0.3% to 24%) in the picture completion test in the upper urinary arsenic tertile. However, we did not find evidence of an association between test results and arsenic water concentrations during pregnancy or childhood. Conclusions: Current arsenic concentrations in urine, which reflect all sources of recent exposure, including water and food, were associated with small decrements in intellectual testing in school-aged children in West Bengal. We did not see associations between long-term water arsenic concentrations and intellectual function.

Statin use and risk of prostate cancer in the California Men's Health Study cohort.

Statins have known anticarcinogenic effects, however, evidence for long-term statin use as effective chemoprevention for prostate cancer is inconsistent. We examined the association between statin use and risk of prostate cancer among 69,047 eligible participants in the California Mens Health Study, a prospective cohort of Northern and Southern California Kaiser Permanente (KP) members, ages 45 to 69 years, initiated in 2002. Prostate cancer cases were identified by linkage to the KP California Cancer Registries. Statin exposure, estimated from automated KP outpatient pharmacy records (available since 1991 in Southern California and since 1994 in Northern California), was treated as time-varying and defined as the cumulative days dispensed of any statin from the first dispensing until a prostate cancer diagnosis, radical prostatectomy, termination of membership, or end of study (December 31, 2004). Cox proportional hazards models with age as the time scale were used to estimate rate ratios, while controlling for confounding variables. During follow-up, 888 prostate cancer cases, including 131 advanced cases, were identified. There was no association between ever statin use or <5 years use and prostate cancer. Conversely, ≥5 years use was associated with a 28% lower risk for prostate cancer compared with nonuse (adjusted rate ratio, 0.72; 95% confidence interval, 0.53-0.99). This association did not differ markedly for advanced disease. However, the association did seem to be restricted to those who regularly take nonsteroidal anti-inflammatory drugs. Our findings suggest that long-term statin use might be associated with a reduced risk of prostate cancer but perhaps only among regular nonsteroidal anti-inflammatory drug users. (Cancer Epidemiol Biomarkers Prev 2007;16(11):2218–25)

Bronchiectasis in persons with skin lesions resulting from arsenic in drinking water

Background: Arsenic is a unique human carcinogen in that it causes lung cancer by exposure through ingestion (in drinking water) as well as through inhalation. Less is known about nonmalignant pulmonary disease after exposure to arsenic in drinking water. Methods: We recruited 108 subjects with arsenic-caused skin lesions and 150 subjects without lesions from a population survey of over 7000 people in an arsenic-exposed region in West Bengal, India. Thirty-eight study participants who reported at least 2 years of chronic cough underwent high-resolution computed tomography (CT); these scans were read by investigators in India and the United States without knowledge of the presence or absence of skin lesions. Results: The mean (± standard deviation) bronchiectasis severity score was 3.4 (± 3.6) in the 27 participants with skin lesions and 0.9 (± 1.6) in the 11 participants without these lesions. In subjects who reported chronic cough, CT evidence of bronchiectasis was found in 18 (67%) participants with skin lesions and 3 (27%) subjects without skin lesions. Overall, subjects with arsenic-caused skin lesions had a 10-fold increased prevalence of bronchiectasis compared with subjects who did not have skin lesions (adjusted odds ratio = 10; 95% confidence interval = 2.7–37). Conclusions: These results suggest that, in addition to being a cause of lung cancer, ingestion of high concentrations of arsenic in drinking water may be a cause of bronchiectasis.

Impact of Breast Cancer Subtypes and Treatment on Survival: An Analysis Spanning Two Decades

Background: We investigated the impact of breast cancer molecular subtypes and treatment on survival in a cohort of medically insured women followed for more than 20 years. Methods: We examined 934 female members of an integrated health care delivery system newly diagnosed with invasive breast cancer between 1988 and 1995 and followed them through 2008. Tumors were classified into four molecular subtypes on the basis of their expression profile: luminal A; luminal B; basal-like; and HER2-enriched. We followed women from the surgery date to death, health plan disenrollment, or studys end. HR and 95% confidence intervals (CI) were fit using Cox proportional hazards models adjusting for cancer treatments and tumor characteristics. Results: A total of 223 (23.9%) women died because of breast cancer during the 21-year study period. Compared with women with luminal A tumors, women with HER2-enriched (HR 2.56, 95% CI 1.53–4.29) and luminal B tumors (HR 1.96, 95% CI: 1.08–3.54) had roughly a two-fold increased adjusted risk of breast cancer mortality. In addition, the survival curves suggest that risk of late mortality persists in women with luminal A tumors. Conclusion: Among women with health care coverage, molecular subtypes were important predictors of breast cancer mortality. Women with HER2-enriched tumors and luminal B subtypes had the poorest survival despite adjusting for important covariates. Impact: In a cohort followed for more than 20 years, women with HER2-enriched tumors had worse survival, but interestingly, the survival curve for women with luminal A tumors continued to steadily decline after 10 years of follow-up. Cancer Epidemiol Biomarkers Prev; 21(10); 1848–55. ©2012 AACR.

Prostatitis, sexually transmitted diseases, and prostate cancer: the California Men's Health Study.

Background Prostatitis and sexually transmitted diseases (STDs) have been positively associated with prostate cancer in previous case-control studies. However, results from recent prospective studies have been inconclusive. Methodogy/Principal Findings We investigated the association between prostatitis, STDs, and prostate cancer among African American, Asian American, Latino, and White participants of the California Mens Health Study. Our analysis included 68,675 men, who completed a detailed baseline questionnaire in 2002–2003. We identified 1,658 incident prostate cancer cases during the follow-up period to June 30, 2006. Cox proportional hazards models were used to estimate relative risks and 95% confidence intervals. Overall, men having a history of prostatitis had an increased risk of prostate cancer than men with no history (RR = 1.30; 95% CI: 1.10–1.54). Longer duration of prostatitis symptoms was also associated with an increased risk of prostate cancer (P trend = 0.003). In addition, among men screened for prostate cancer (1 or 2 PSA tests), a non-significant positive association was observed between prostatitis and prostate cancer (RR = 1.10; 95% CI: 0.75–1.63). STDs were not associated with overall prostate cancer risk. In racial/ethnic stratified analysis, Latinos reporting any STDs had an increased risk of disease than those with no STDs (RR = 1.43; 95% CI: 1.07–1.91). Interestingly, foreign-born Latinos displayed a larger risk associated with STDs (RR = 1.87; 95% CI: 1.16–3.02) than U.S. born Latinos (RR = 1.15; 95% CI: 0.76–3.02). Conclusion In summary, results from this prospective study suggest that prostatitis and STDs may be involved in prostate cancer susceptibility. While we cannot rule out the possible influence of incidental detection, future studies are warranted to further investigate the role of infectious agents related to prostatitis and STDs in prostate cancer development.

Sleep disturbance, inflammation and depression risk in cancer survivors

Over two-thirds of the 11.4 million cancer survivors in the United States can expect long-term survival, with many others living with cancer as a chronic disease controlled by ongoing therapy. However, behavioral co-morbidities often arise during treatment and persist long-term to complicate survival and reduce quality of life. In this review, the inter-relationships between cancer, depression, and sleep disturbance are described, with a focus on the role of sleep disturbance as a risk factor for depression. Increasing evidence also links alterations in inflammatory biology dynamics to these long-term effects of cancer diagnosis and treatment, and the hypothesis that sleep disturbance drives inflammation, which together contribute to depression, is discussed. Better understanding of the associations between inflammation and behavioral co-morbidities has the potential to refine prediction of risk and development of strategies for the prevention and treatment of sleep disturbance and depression in cancer survivors.

Surgical margins and survival after head and neck cancer surgery

BackgroundMixed results exist as to whether positive surgical margins impact survival. The aim of this study was to determine whether positive surgical margins are indeed associated with decreased survival in patients with primary head and neck cancer.MethodsWe conducted a retrospective cohort study of 261 cases diagnosed with cancer of the larynx or tongue between 1995 and 1999. Cases were followed through December 31, 2002. Survival curves by margin status were generated by Kaplan-Meier methods. Categorical data were evaluated with odds ratios (OR).ResultsAll-cause mortality was markedly higher in cases with positive margins as compared with those with negative margins (54% versus 29%, P = 0.005). This pattern also appeared after adjusting for age and sex (OR = 2.97, 95% CI: 1.29 – 6.84).ConclusionOur findings suggest that positive surgical margin status is associated with increased mortality. This association also generally persists after adjustment for tumor size, stage, and adjuvant therapy.