Evaluating sampling strategies and logistic regression methods for modelling complex land cover changes

The role of land cover change as a significant component of global change has become increasingly recognized in recent decades. Large databases measuring land cover change, and the data which can potentially be used to explain the observed changes, are also becoming more commonly available. When developing statistical models to investigate observed changes, it is important to be aware that the chosen sampling strategy and modelling techniques can influence results. We present a comparison of three sampling strategies and two forms of grouped logistic regression models (multinomial and ordinal) in the investigation of patterns of successional change after agricultural land abandonment in Switzerland. Results indicated that both ordinal and nominal transitional change occurs in the landscape and that the use of different sampling regimes and modelling techniques as investigative tools yield different results. Synthesis and applications. Our multimodel inference identified successfully a set of consistently selected indicators of land cover change, which can be used to predict further change, including annual average temperature, the number of already overgrown neighbouring areas of land and distance to historically destructive avalanche sites. This allows for more reliable decision making and planning with respect to landscape management. Although both model approaches gave similar results, ordinal regression yielded more parsimonious models that identified the important predictors of land cover change more efficiently. Thus, this approach is favourable where land cover change pattern can be interpreted as an ordinal process. Otherwise, multinomial logistic regression is a viable alternative.

Validation of Logistic Regression Models for Landslide Susceptibility Maps

A wide range of numerical models and tools have been developed over the last decades to support the decision making process in environmental applications, ranging from physical models to a variety of statistically-based methods. In this study, a landslide susceptibility map of a part of Three Gorges Reservoir region of China was produced, employing binary logistic regression analyses. The available information includes the digital elevation model of the region, geological map and different GIS layers including land cover data obtained from satellite imagery. The landslides were observed and documented during the field studies. The validation analysis is exploited to investigate the quality of mapping.

Estimating attributable mortality due to nosocomial infections acquired in intensive care units.

BACKGROUND: The strength of the association between intensive care unit (ICU)-acquired nosocomial infections (NIs) and mortality might differ according to the methodological approach taken. OBJECTIVE: To assess the association between ICU-acquired NIs and mortality using the concept of population-attributable fraction (PAF) for patient deaths caused by ICU-acquired NIs in a large cohort of critically ill patients. SETTING: Eleven ICUs of a French university hospital. DESIGN: We analyzed surveillance data on ICU-acquired NIs collected prospectively during the period from 1995 through 2003. The primary outcome was mortality from ICU-acquired NI stratified by site of infection. A matched-pair, case-control study was performed. Each patient who died before ICU discharge was defined as a case patient, and each patient who survived to ICU discharge was defined as a control patient. The PAF was calculated after adjustment for confounders by use of conditional logistic regression analysis. RESULTS: Among 8,068 ICU patients, a total of 1,725 deceased patients were successfully matched with 1,725 control patients. The adjusted PAF due to ICU-acquired NI for patients who died before ICU discharge was 14.6% (95% confidence interval [CI], 14.4%-14.8%). Stratified by the type of infection, the PAF was 6.1% (95% CI, 5.7%-6.5%) for pulmonary infection, 3.2% (95% CI, 2.8%-3.5%) for central venous catheter infection, 1.7% (95% CI, 0.9%-2.5%) for bloodstream infection, and 0.0% (95% CI, -0.4% to 0.4%) for urinary tract infection. CONCLUSIONS: ICU-acquired NI had an important effect on mortality. However, the statistical association between ICU-acquired NI and mortality tended to be less pronounced in findings based on the PAF than in study findings based on estimates of relative risk. Therefore, the choice of methods does matter when the burden of NI needs to be assessed.

Circulating estrogens and progesterone during primiparous pregnancies and risk of maternal breast cancer.

Pregnancy reduces maternal risk of breast cancer in the long term, but the biological determinants of the protection are unknown. Animal experiments suggest that estrogens and progesterone could be involved, but direct human evidence is scant. A case-control study (536 cases and 1,049 controls) was nested within the Finnish Maternity Cohort. Eligible were primiparous women who delivered at term a singleton offspring before age 40. For each case, two individually matched controls by age (±6 months) and date of sampling (±3 months) were selected. Estradiol, estrone and progesterone in first-trimester serum were measured by high-performance liquid chromatography tandem mass spectrometry and sex-hormone binding globulin (SHBG) by immunoassay. Odds ratios (OR) and 95% confidence intervals (CI) were estimated through conditional logistic regression. In the whole study population there was no association of breast cancer with any of the studied hormones. In analyses stratified by age at diagnosis, however, estradiol concentrations were positively associated with risk of breast cancer before age 40 (upper quartile OR, 1.81; CI, 1.08-3.06), but inversely associated with risk in women who were diagnosed ≥age 40 (upper quartile OR, 0.64; CI, 0.40-1.04), p(interaction) 0.004. Risk estimates for estrone mirrored those for estradiol but were less pronounced. Progesterone was not associated with risk of subsequent breast cancer. Our results provide initial evidence that concentrations of estrogens during the early parts of a primiparous pregnancy are associated with maternal risk of breast cancer and suggest that the effect may differ for tumors diagnosed before and after age 40.

Circulating insulin-like growth factor-I in pregnancy and maternal risk of breast cancer.

BACKGROUND: Elevated serum concentrations of insulin-like growth factor (IGF)-1 have been associated with increased risk of breast cancer. Previously, we reported a similar association in samples obtained during pregnancy. The current study was conducted to further characterize the association of IGF-1 during pregnancy with maternal breast cancer risk. METHODS: A case-control study was nested within the Finnish Maternity Cohort. The study was limited to primiparous women less than 40 years of age, who donated blood samples during early (median, 12 weeks) pregnancy and delivered a single child at term. Seven hundred and nineteen women with invasive breast cancer were eligible. Two controls (n = 1,434) were matched to each case on age and date at blood donation. Serum IGF-1 concentration was measured using an Immulite 2000 analyzer. Conditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CI).RESULTS: No significant associations were observed between serum IGF-1 concentrations and breast cancer risk in both the overall analysis (OR 1.08 (95% CI 0.80-1.47) and in analyses stratified by histological subtype, lag-time to cancer diagnosis, age at pregnancy or age at diagnosis.CONCLUSIONS: There was no association between IGF-1 and maternal breast cancer risk during early pregnancy in this large nested case-control study.Impact:Serum IGF-1 concentrations during early pregnancy may not be related to maternal risk of breast cancer.

Association of different lipid measures for atherogenic lipid fractions and risk of coronary events in patients receiving combination antiretroviral therapy: the Swiss HIV Cohort Study

Background: Evidence for a better performance of different highly atherogenic versus traditional lipid parameters for coronary heart disease (CHD) risk prediction is conflicting. We investigated the association of the ratios of sma11 dense low density lipoprotein(LDL)/apoplipoprotein A, aolipoprotein B/apolipoprotein A-I and total cholesterol! HDL-cholesterol and CHD events in patients on combination antiretroviral therapy (cART).Methods: Case control study nested into the Swiss HIV Cohort Study: for each cART-treated patient with a first coronary event between April 1, 2000 and July 31, 2008 (case) we selected four control patients (1) that were without coronary events until the date of the event of the index case, (2) had a plasma sample within ±30 days of the sample date of the respective case, (3) received cART and (4) were then matched for age, gender and smoking status. Lipoproteins were measured by ultracentrifugation. Conditional logistic regression models were used to estimate the independent effects of different lipid ratios and the occurrence of coronary events.Results: In total, 98 cases (19 fatal myocardial infarctions [MI] and 79 non-fatal coronary events [53 definite MIs, 15 possible MIs and 11 coronary angioplasties or bypassesJ) were matched with 392 controls. Cases were more often injecting drug users, less likely to be virologically suppressed and more often on abacavir-containing regimens. In separa te multivariable models of total cholesterol, triglycerides, HDL-cholesterol, systolic blood pressure, abdominal obesity, diabetes and family history of CHD, small dense-LDL and apolipoprotein B were each statistically significantly associated with CHD events (for 1 mg/dl increase: odds ratio [OR] 1.05, 95% CI 1.00-1.11 and 1.15, 95% CI 1.01-1.31, respectively), but the ratiosof small dense-LDLlapolipoprotein A-I (OR 1.26, 95% CI 0.95-1.67), apolipoprotein B/apolipoprotein A-I (OR 1.02, 95% CI 0.97-1.07) and HDL-cholesterol! total cholesterol (OR 0.99 95% CI 0.98-1.00) were not. Following adjustment for HIV related and cART variables these associations were weakened in each model: apolipoprotein B (OR 1.27, 95% CI 1.00-1.30), sd-LDL (OR 1.04, 95% CI 0.99-1.20), small dense-LDLlapolipoprotein A-I (OR 1.17, 95% CI 0.87-1.58), apolipoprotein B/apolipoprotein A-I (OR 1.02, 95% CI 0.97-1.07) and total cholesterolJHDL- cholesterol (OR 0.99, 95% CI 0.99-1.00).Conclusions: In patients receiving cART, small dense-LDL and apolipoprotein B showed the strongest associations with CHD events in models controlling for traditional CHD risk factors including total cholesterol and triglycerides. Adding small dense LDLlapoplipoprotein A-l, apolipoprotein B/apolipoprotein A-I and total cholesterol! HDL-cholesterol ratios did not further improve models of lipid parameters and associations of increased risk for CHD events.

Human chorionic gonadotropin in pregnancy and maternal risk of breast cancer.

Full-term pregnancies are associated with long-term reductions in maternal risk of breast cancer, but the biological determinants of the protection are unknown. Experimental observations suggest that human chorionic gonadotropin (hCG), a major hormone of pregnancy, could play a role in this association. A case-control study (242 cases and 450 controls) nested within the Northern Sweden Maternity Cohort included women who had donated a blood sample during the first trimester of a first full-term pregnancy. Total hCG was determined on Immulite 2000 analyzer. Odds ratios (OR) and 95% confidence intervals (CI) were estimated through conditional logistic regression. Maternal breast cancer risk decreased with increasing hCG (upper tertile OR, 0.67; CI, 0.46-0.99), especially for pregnancies before age 25 (upper tertile OR, 0.41; CI, 0.21-0.80). The association diverged according to age at diagnosis: risk was reduced after age 40 (upper tertile OR, 0.60; CI, 0.39-0.91) and seemed to increase before age 40 (upper tertile OR, 1.78; CI, 0.72-4.38). Risk was reduced among those diagnosed 10 years or longer after blood draw (upper tertile OR, 0.60; CI, 0.40-0.90), but not so among those diagnosed within 10 years (upper tertile OR, 4.33; CI, 0.86-21.7). These observations suggest that the association between pregnancy hCG and subsequent maternal risk of breast cancer is modified by age at diagnosis. Although the hormone seems to be a determinant of the reduced risk around or after age 50, it might not confer protection against, or it could even increase the risk of, cancers diagnosed in the years immediately following pregnancy.

Lung cancer in the Swiss HIV Cohort Study: role of smoking, immunodeficiency and pulmonary infection.

BACKGROUND: Immunodeficiency and AIDS-related pulmonary infections have been suggested as independent causes of lung cancer among HIV-infected persons, in addition to smoking. METHODS: A total of 68 lung cancers were identified in the Swiss HIV Cohort Study (SHCS) or through linkage with Swiss Cancer Registries (1985-2010), and were individually matched to 337 controls by centre, gender, HIV-transmission category, age and calendar period. Odds ratios (ORs) were estimated by conditional logistic regression. RESULTS: Overall, 96.2% of lung cancers and 72.9% of controls were ever smokers, confirming the high prevalence of smoking and its strong association with lung cancer (OR for current vs never=14.4, 95% confidence interval (95% CI): 3.36-62.1). No significant associations were observed between CD4+ cell count and lung cancer, neither when measured within 1 year (OR for <200 vs ≥500=1.21, 95% CI: 0.49-2.96) nor further back in time, before lung cancer diagnosis. Combined antiretroviral therapy was not significantly associated with lung cancer (OR for ever vs never=0.67, 95% CI: 0.29-1.52), and nor was a history of AIDS with (OR=0.49, 95% CI: 0.19-1.28) or without (OR=0.53, 95% CI: 0.24-1.18) pulmonary involvement. CONCLUSION: Lung cancer in the SHCS does not seem to be clearly associated with immunodeficiency or AIDS-related pulmonary disease, but seems to be attributable to heavy smoking.

Mobile phone use and brain tumors in children and adolescents: a multicenter case-control study.

Background It has been hypothesized that children and adolescents might be more vulnerable to possible health effects from mobile phone exposure than adults. We investigated whether mobile phone use is associated with brain tumor risk among children and adolescents. Methods CEFALO is a multicenter case-control study conducted in Denmark, Sweden, Norway, and Switzerland that includes all children and adolescents aged 7-19 years who were diagnosed with a brain tumor between 2004 and 2008. We conducted interviews, in person, with 352 case patients (participation rate: 83%) and 646 control subjects (participation rate: 71%) and their parents. Control subjects were randomly selected from population registries and matched by age, sex, and geographical region. We asked about mobile phone use and included mobile phone operator records when available. Odds ratios (ORs) for brain tumor risk and 95% confidence intervals (CIs) were calculated using conditional logistic regression models. Results Regular users of mobile phones were not statistically significantly more likely to have been diagnosed with brain tumors compared with nonusers (OR = 1.36; 95% CI = 0.92 to 2.02). Children who started to use mobile phones at least 5 years ago were not at increased risk compared with those who had never regularly used mobile phones (OR = 1.26, 95% CI = 0.70 to 2.28). In a subset of study participants for whom operator recorded data were available, brain tumor risk was related to the time elapsed since the mobile phone subscription was started but not to amount of use. No increased risk of brain tumors was observed for brain areas receiving the highest amount of exposure. Conclusion The absence of an exposure-response relationship either in terms of the amount of mobile phone use or by localization of the brain tumor argues against a causal association.

Circulating carotenoids and risk of breast cancer: pooled analysis of eight prospective studies.

Background Carotenoids, micronutrients in fruits and vegetables, may reduce breast cancer risk. Most, but not all, past studies of circulating carotenoids and breast cancer have found an inverse association with at least one carotenoid, although the specific carotenoid has varied across studies. Methods We conducted a pooled analysis of eight cohort studies comprising more than 80% of the world's published prospective data on plasma or serum carotenoids and breast cancer, including 3055 case subjects and 3956 matched control subjects. To account for laboratory differences and examine population differences across studies, we recalibrated participant carotenoid levels to a common standard by reassaying 20 plasma or serum samples from each cohort together at the same laboratory. Using conditional logistic regression, adjusting for several breast cancer risk factors, we calculated relative risks (RRs) and 95% confidence intervals (CIs) using quintiles defined among the control subjects from all studies. All P values are two-sided. Results Statistically significant inverse associations with breast cancer were observed for α-carotene (top vs bottom quintile RR = 0.87, 95% CI = 0.71 to 1.05, Ptrend = .04), β-carotene (RR = 0.83, 95% CI = 0.70 to 0.98, Ptrend = .02), lutein+zeaxanthin (RR = 0.84, 95% CI = 0.70 to 1.01, Ptrend = .05), lycopene (RR = 0.78, 95% CI = 0.62 to 0.99, Ptrend = .02), and total carotenoids (RR = 0.81, 95% CI = 0.68 to 0.96, Ptrend = .01). β-Cryptoxanthin was not statistically significantly associated with risk. Tests for heterogeneity across studies were not statistically significant. For several carotenoids, associations appeared stronger for estrogen receptor negative (ER(-)) than for ER(+) tumors (eg, β-carotene: ER(-): top vs bottom quintile RR = 0.52, 95% CI = 0.36 to 0.77, Ptrend = .001; ER(+): RR = 0.83, 95% CI = 0.66 to 1.04, Ptrend = .06; Pheterogeneity = .01). Conclusions This comprehensive prospective analysis suggests women with higher circulating levels of α-carotene, β-carotene, lutein+zeaxanthin, lycopene, and total carotenoids may be at reduced risk of breast cancer.

Influence of HIV-related immunodeficiency on the risk of hepatocellular carcinoma.

OBJECTIVE: To investigate HIV-related immunodeficiency as a risk factor for hepatocellular carcinoma (HCC) among persons infected with HIV, while controlling for the effect of frequent coinfection with hepatitis C and B viruses. DESIGN: A case-control study nested in the Swiss HIV Cohort Study. METHODS: Twenty-six HCC patients were identified in the Swiss HIV Cohort Study or through linkage with Swiss Cancer Registries, and were individually matched to 251 controls according to Swiss HIV Cohort Study centre, sex, HIV-transmission category, age and year at enrollment. Odds ratios and corresponding confidence intervals were estimated by conditional logistic regression. RESULTS: All HCC patients were positive for hepatitis B surface antigen or antibodies against hepatitis C virus. HCC patients included 14 injection drug users (three positive for hepatitis B surface antigen and 13 for antibodies against hepatitis C virus) and 12 men having sex with men/heterosexual/other (11 positive for hepatitis B surface antigen, three for antibodies against hepatitis C virus), revealing a strong relationship between HIV transmission route and hepatitis viral type. Latest CD4+ cell count [Odds ratio (OR) per 100 cells/mul decrease = 1.33, 95% confidence interval (CI) 1.06-1.68] and CD4+ cell count percentage (OR per 10% decrease = 1.65, 95% CI 1.01-2.71) were significantly associated with HCC. The effects of CD4+ cell count were concentrated among men having sex with men/heterosexual/other rather than injecting drug users. Highly active antiretroviral therapy use was not significantly associated with HCC risk (OR for ever versus never = 0.59, 95% confidence interval 0.18-1.91). CONCLUSION: Lower CD4+ cell counts increased the risk for HCC among persons infected with HIV, an effect that was particularly evident for hepatitis B virus-related HCC arising in non-injecting drug users.

Physical activity and energy expenditure in rheumatoid arthritis patients and matched controls.

OBJECTIVES: To compare daily energy expenditure between RA patients and matched controls, and to explore the relationship between daily energy expenditure or sedentariness and disease-related scores. METHODS: One hundred and ten patients with RA and 440 age- and sex-matched controls were included in this study. Energy expenditure was assessed using the validated physical activity (PA) frequency questionnaire. Disease-related scores included disease activity (DAS-28), functional status (HAQ), pain visual analogue scale (VAS) and fatigue VAS. Total energy expenditure (TEE) and the amount of energy spent in low- (TEE-low), moderate- (TEE-mod) and high-intensity (TEE-high) PAs were calculated. Sedentariness was defined as expending <10% of TEE in TEE-mod or TEE-high activities. Between-group comparisons were computed using conditional logistic regression. The effect of disease-related scores on TEE was investigated using linear regression. RESULTS: TEE was significantly lower for RA patients compared with controls [2392 kcal/day (95% CI 2295, 2490) and 2494  kcal/day (2446, 2543), respectively, P = 0.003]. A significant difference was found between groups in TEE-mod (P = 0.015), but not TEE-low (P = 0.242) and TEE-high (P = 0.146). All disease-related scores were significantly poorer in sedentary compared with active patients. TEE was inversely associated with age (P < 0.001), DAS-28 (P = 0.032) and fatigue VAS (P = 0.029), but not with HAQ and pain VAS. CONCLUSION: Daily energy expenditure is significantly lower in RA patients compared with matched controls, mainly due to less moderate-intensity PAs performed. Disease activity and fatigue are important contributing factors. These points need to be addressed if promoting PA in RA patients is a health goal. Trial registration. ClinicalTrials.gov, http://clinicaltrials.gov, NCT01228812.

Early pregnancy sex steroids and maternal risk of epithelial ovarian cancer.

Well-established associations between reproductive characteristics and epithelial ovarian cancer (EOC) support an involvement of sex steroid hormones in the etiology of EOC. Limited previous studies have evaluated circulating androgens and the risk of EOC, and estrogens and progesterone have been investigated in only one of the previous studies. Furthermore, there is little data on potential heterogeneity in the association between circulating hormones and EOC by histological subgroup. Therefore, we conducted a nested case-control study within the Finnish Maternity Cohort and the Northern Sweden Maternity Cohort to investigate the associations between circulating pre-diagnostic sex steroid concentrations and the histological subtypes of EOC. We identified 1052 EOC cases among cohort members diagnosed after recruitment (1975-2008) and before March 2011. Up to three controls were individually matched to each case (n=2694). Testosterone, androstenedione, 17-hydroxyprogesterone (17-OHP), progesterone, estradiol (E2), and sex hormone-binding globulin levels were measured in serum samples collected during the last pregnancy before EOC diagnosis. We used conditional logistic regression to estimate odds ratios (ORs) and 95% CIs. Associations between hormones and EOC differed with respect to tumor histology and invasiveness. Sex steroid concentrations were not associated with invasive serous tumors; however, doubling of testosterone and 17-OHP concentration was associated with approximately 40% increased risk of borderline serous tumors. A doubling of androgen concentrations was associated with a 50% increased risk of mucinous tumors. The risk of endometrioid tumors increased with higher E2 concentrations (OR: 1.89 (1.20-2.98)). This large prospective study in pregnant women supports a role of sex steroid hormones in the etiology of EOC arising in the ovaries.

Case-control analysis of paternal age and trisomic anomalies.

OBJECTIVES: To determine whether older paternal age increases the risk of fathering a pregnancy with Patau (trisomy 13), Edwards (trisomy 18), Klinefelter (XXY) or XYY syndrome. DESIGN: Case-control: cases with each of these syndromes were matched to four controls with Down syndrome from within the same congenital anomaly register and with maternal age within 6 months. SETTING: Data from 22 EUROCAT congenital anomaly registers in 12 European countries. PARTICIPANTS: Diagnoses with observed or (for terminations) predicted year of birth from 1980 to 2005, comprising live births, fetal deaths with gestational age ≥ 20 weeks and terminations after prenatal diagnosis of the anomaly. Data include 374 cases of Patau syndrome, 929 of Edwards syndrome, 295 of Klinefelter syndrome, 28 of XYY syndrome and 5627 controls with Down syndrome. MAIN OUTCOME MEASURES: Odds ratio (OR) associated with a 10-year increase in paternal age for each anomaly was estimated using conditional logistic regression. Results were adjusted to take account of the estimated association of paternal age with Down syndrome (1.11; 95% CI 1.01 to 1.23). RESULTS: The OR for Patau syndrome was 1.10 (95% CI 0.83 to 1.45); for Edwards syndrome, 1.15 (0.96 to 1.38); for Klinefelter syndrome, 1.35 (1.02 to 1.79); and for XYY syndrome, 1.99 (0.75 to 5.26). CONCLUSIONS: There was a statistically significant increase in the odds of Klinefelter syndrome with increasing paternal age. The larger positive associations of Klinefelter and XYY syndromes with paternal age compared with Patau and Edwards syndromes are consistent with the greater percentage of these sex chromosome anomalies being of paternal origin.

Circulating sex steroids during pregnancy and maternal risk of non-epithelial ovarian cancer.

BACKGROUND: Sex steroid hormones have been proposed to play a role in the development of non-epithelial ovarian cancers (NEOC) but so far no direct epidemiological data are available.METHODS: A case-control study was nested within the Finnish Maternity Cohort, the world's largest bio-repository of serum specimens from pregnant women. Study subjects were selected among women who donated a blood sample during a singleton pregnancy that led to the birth of their last child preceding diagnosis of NEOC. Case subjects were 41 women with sex-cord stromal tumors (SCST) and 21 with germ cell tumors (GCT). Three controls, matching the index case for age, parity at the index pregnancy, and date at blood donation were selected (n=171). Odds ratios (OR) and 95% confidence intervals (CI) associated with concentrations of testosterone, androstenedione, 17-OH-progesterone, progesterone, estradiol and sex hormone binding globulin (SHBG) were estimated through conditional logistic regression.RESULTS: For SCST, doubling of testosterone, androstenedione and 17-OH-progesterone concentrations were associated with about 2-fold higher risk of SCST [ORs and 95% CI of 2.16 (1.25-3.74), 2.16 (1.20-3.87), and 2.62 (1.27-5.38), respectively]. These associations remained largely unchanged after excluding women within 2, 4 or 6 years lag-time between blood donation and cancer diagnosis. Sex steroid hormones concentrations were not related to maternal risk of GCT.CONCLUSIONS: This is the first prospective study providing initial evidence that elevated androgens play a role in the pathogenesis of SCST. Impact: Our study may note a particular need for larger confirmatory investigations on sex steroids and NEOC.