Statnote 24: Multiple regression

In previous statnotes, the application of correlation and regression methods to the analysis of two variables (X,Y) was described. These methods can be used to determine whether there is a linear relationship between the two variables, whether the relationship is positive or negative, to test the degree of significance of the linear relationship, and to obtain an equation relating Y to X. This Statnote extends the methods of linear correlation and regression to situations where there are two or more X variables, i.e., 'multiple linear regression’.

Effects Of Therapeutic Approach On The Neonatal Evolution Of Very Low Birth Weight Infants With Patent Ductus Arteriosus.

To analyze the effects of treatment approach on the outcomes of newborns (birth weight [BW] < 1,000 g) with patent ductus arteriosus (PDA), from the Brazilian Neonatal Research Network (BNRN) on: death, bronchopulmonary dysplasia (BPD), severe intraventricular hemorrhage (IVH III/IV), retinopathy of prematurity requiring surgical (ROPsur), necrotizing enterocolitis requiring surgery (NECsur), and death/BPD. This was a multicentric, cohort study, retrospective data collection, including newborns (BW < 1000 g) with gestational age (GA) < 33 weeks and echocardiographic diagnosis of PDA, from 16 neonatal units of the BNRN from January 1, 2010 to Dec 31, 2011. Newborns who died or were transferred until the third day of life, and those with presence of congenital malformation or infection were excluded. Groups: G1 - conservative approach (without treatment), G2 - pharmacologic (indomethacin or ibuprofen), G3 - surgical ligation (independent of previous treatment). Factors analyzed: antenatal corticosteroid, cesarean section, BW, GA, 5 min. Apgar score < 4, male gender, Score for Neonatal Acute Physiology Perinatal Extension (SNAPPE II), respiratory distress syndrome (RDS), late sepsis (LS), mechanical ventilation (MV), surfactant (< 2 h of life), and time of MV. death, O2 dependence at 36 weeks (BPD36wks), IVH III/IV, ROPsur, NECsur, and death/BPD36wks. Student's t-test, chi-squared test, or Fisher's exact test; Odds ratio (95% CI); logistic binary regression and backward stepwise multiple regression. Software: MedCalc (Medical Calculator) software, version 12.1.4.0. p-values < 0.05 were considered statistically significant. 1,097 newborns were selected and 494 newborns were included: G1 - 187 (37.8%), G2 - 205 (41.5%), and G3 - 102 (20.6%). The highest mortality was observed in G1 (51.3%) and the lowest in G3 (14.7%). The highest frequencies of BPD36wks (70.6%) and ROPsur were observed in G3 (23.5%). The lowest occurrence of death/BPD36wks occurred in G2 (58.0%). Pharmacological (OR 0.29; 95% CI: 0.14-0.62) and conservative (OR 0.34; 95% CI: 0.14-0.79) treatments were protective for the outcome death/BPD36wks. The conservative approach of PDA was associated to high mortality, the surgical approach to the occurrence of BPD36wks and ROPsur, and the pharmacological treatment was protective for the outcome death/BPD36wks.

Regional and total body bone mineral density in elite collegiate male swimmers

Back,ground To examine the role of long-term swimming exercise on regional and total body bone mineral density (BMD) in men. Methods. Experimental design: Cross-sectional. Setting: Musculoskeletal research laboratory at a medical center, Participants:We compared elite collegiate swimmers (n=11) to age-, weight-, and height-matched non-athletic controls (n=11), Measures: BMD (g/cm(2)) of the lumbar spine L2-4, proximal femur (femoral neck, trochanter, Ward's triangle), total body and various subregions of the total body, as well as regional and total body fat and bone mineral-free lean mass (LM) was assessed by dual-energy X-ray absorptiometry (DXA, Hologic QDR 1000/W). Results. Swimmers, who commenced training at 10.7+/-3.7 yrs (mean+/-SD) and trained for 24.7+/-4.2 hrs per week, had a greater amount of LM (p<0.05), lower fat mass (p<0.001) and percent body fat (9.5 vs 16.2 %, p<0.001) than controls. There was no significant difference between groups for regional or total body BRID, In stepwise multiple regression analysis, body weight was a consistent independent predictor of regional and total body BMD, Conclusions. These results suggest that long-term swimming is not an osteogenic mode of training in college-aged males. This supports our previous findings in young female swimmers who displayed no bone mass benefits despite long-standing athletic training.

Analysis and visualization of energy use for university campus

The reduction of greenhouse gas emissions is one of the big global challenges for the next decades due to its severe impact on the atmosphere that leads to a change in the climate and other environmental factors. One of the main sources of greenhouse gas is energy consumption, therefore a number of initiatives and calls for awareness and sustainability in energy use are issued among different types of institutional and organizations. The European Council adopted in 2007 energy and climate change objectives for 20% improvement until 2020. All European countries are required to use energy with more efficiency. Several steps could be conducted for energy reduction: understanding the buildings behavior through time, revealing the factors that influence the consumption, applying the right measurement for reduction and sustainability, visualizing the hidden connection between our daily habits impacts on the natural world and promoting to more sustainable life. Researchers have suggested that feedback visualization can effectively encourage conservation with energy reduction rate of 18%. Furthermore, researchers have contributed to the identification process of a set of factors which are very likely to influence consumption. Such as occupancy level, occupants behavior, environmental conditions, building thermal envelope, climate zones, etc. Nowadays, the amount of energy consumption at the university campuses are huge and it needs great effort to meet the reduction requested by European Council as well as the cost reduction. Thus, the present study was performed on the university buildings as a use case to: a. Investigate the most dynamic influence factors on energy consumption in campus; b. Implement prediction model for electricity consumption using different techniques, such as the traditional regression way and the alternative machine learning techniques; and c. Assist energy management by providing a real time energy feedback and visualization in campus for more awareness and better decision making. This methodology is implemented to the use case of University Jaume I (UJI), located in Castellon, Spain.

Effect of biological surlie ageing on the sparkling wines’ quality and elaoration

Report for the scientific sojourn at the Université de Bourgogne, France, from July until October 2007..Surlie ageing after second fermentation is a fundamental operation in the production of quality sparkling wine like Cava and Champagne. Recently, the importance of the interaction between wine and lees cell surface has been reported. Cell surface properties depending on wall biochemical composition are major determinants in microbial interactions, having important repercussions in several technological aspects. Sorption and flocculation are especially important in sparkling wine production, and are governed by distinct cell surface properties. The aim of the present research carried out during the four months of the stage was to know the implication of lees surface modifications occurring during surlie ageing in sparkling wine quality and elaboration. The relationship between physico-chemical properties such as hydrophobicity, charge and electron-donor characteristics, and the yeast surface sorption capacities, we determined these factors in a model system. Then, real industrial lees samples were investigated. The surface properties of sparkling wine lees from the same strain of Saccharomyces cerevisiae were characterized according to the time of surlie ageing, and their possible influence on lees sorption and flocculation capacity was evaluated. Surlie ageing after second fermentation is a fundamental operation in the production of quality sparkling wine like Cava and Champagne. Recently, the importance of the interaction between wine and lees cell surface has been reported. Cell surface properties depending on wall biochemical composition are major determinants in microbial interactions, having important repercussions in several technological aspects. Sorption and flocculation are especially important in sparkling wine production, and are governed by distinct cell surface properties. The aim of the present research carried out during the four months of the stage was to know the implication of lees surface modifications occurring during surlie ageing in sparkling wine quality and elaboration. The relationship between physico-chemical properties such as hydrophobicity, charge and electron-donor characteristics, and the yeast surface sorption capacities, we determined these factors in a model system. Then, real industrial lees samples were investigated. The surface properties of sparkling wine lees from the same strain of Saccharomyces cerevisiae were characterized according to the time of surlie ageing, and their possible influence on lees sorption and flocculation capacity was evaluated.

Determinants of basal fat oxidation in healthy Caucasians.

In a retrospective study, we examined several determinants of basal fat oxidation in 720 healthy Caucasian volunteers. Adult men (n = 427) and women (n = 293) were characterized for resting energy expenditure and substrate oxidation by indirect calorimetry (after a 12-h overnight fast), peak O2 consumption by a treadmill test to exhaustion, body composition by hydrodensitometry, food intake from a 3-day food diary, and hormonal status by fasting hormone concentrations. Fat oxidation was negatively correlated with fat mass in men (r = -0.11; P < 0.05), but no statistical relationship was found in women. In a stepwise multiple regression analysis, fat oxidation was best predicted by peak O2 consumption and fat-free mass in men (model R2 = 0.142) and by free thyroxine, fat-free mass, and fasting insulin in women (model R2 = 0.153). Relative rates of fat oxidation (fat oxidation adjusted for differences in resting energy expenditure) were not correlated with fat mass in either gender. Women showed a lower rate of basal fat oxidation (both absolute and adjusted) than did men. Our results show that fat oxidation is not greater in individuals with a greater fat mass. Furthermore, our results support a sexual dimorphism in basal rates of fat oxidation.

Quetiapine affects neuropeptide Y and corticotropin-releasing hormone in cerebrospinal fluid from schizophrenia patients: relationship to depression and anxiety symptoms and to treatment response.

Cumulative evidence indicates that neuropeptides play a role in the pathophysiology of schizophrenia. Early data showed increased neuropeptide Y (NPY) in cerebrospinal fluid (CSF) from schizophrenia patients and data from rodents show that antipsychotic drugs modulate NPY levels in and release from selected rat brain regions. In view of these findings we investigated whether the atypical antipsychotic quetiapine, originally used as an antipsychotic but subsequently shown to be efficient also in major depressive disorder and in both poles of bipolar disorder, would affect NPY-like immunoreactivity (-LI), and corticotropin-releasing hormone (CRH)-LI levels in CSF of schizophrenia patients. NPY-LI and CRH-LI in CSF were determined in 22 patients with schizophrenia. Lumbar puncture was performed at baseline and again after 4 wk of quetiapine treatment (600 mg/d). Patients were assessed with the Positive and Negative Syndrome Scale (PANSS) at baseline and at weekly intervals. Quetiapine treatment was associated with a significant increase in NPY-LI (p<0.001) and decrease in CRH-LI (p<0.01). Stepwise multiple regression analysis revealed that ΔNPY-LI and ΔCRH-LI levels predicted 63% (p<0.001) of the variability of the ΔPANSS total score, ΔNPY-LI 42% (p<0.05) of the ΔPANSS anxiety items (G2) and ΔCRH-LI 40% (p=0.05) of the ΔPANSS depression items (G6). These results suggest that while quetiapine's effects on monoamines are probably related to its antipsychotic properties, the modulation of NPY and CRH accounts for its antidepressant and anxiolytic effects and can be markers of response.

The predictive value of trabecular bone score (TBS) on whole lumbar vertebrae mechanics: an ex vivo study.

We investigated the association of trabecular bone score (TBS) with microarchitecture and mechanical behavior of human lumbar vertebrae. We found that TBS reflects vertebral trabecular microarchitecture and is an independent predictor of vertebral mechanics. However, the addition of TBS to areal BMD (aBMD) did not significantly improve prediction of vertebral strength. INTRODUCTION: The trabecular bone score (TBS) is a gray-level measure of texture using a modified experimental variogram which can be extracted from dual-energy X-ray absorptiometry (DXA) images. The current study aimed to confirm whether TBS is associated with trabecular microarchitecture and mechanics of human lumbar vertebrae, and if its combination with BMD improves prediction of fracture risk. METHODS: Lumbar vertebrae (L3) were harvested fresh from 16 donors. The anteroposterior and lateral bone mineral content (BMC) and areal BMD (aBMD) of the vertebral body were measured using DXA; then, the TBS was extracted using TBS iNsight software (Medimaps SA, France). The trabecular bone volume (Tb.BV/tissue volume, TV), trabecular thickness (Tb.Th), degree of anisotropy, and structure model index (SMI) were measured using microcomputed tomography. Quasi-static uniaxial compressive testing was performed on L3 vertebral bodies to assess failure load and stiffness. RESULTS: The TBS was significantly correlated to Tb.BV/TV and SMI (râeuro0/00=âeuro0/000.58 and -0.62; pâeuro0/00=âeuro0/000.02, 0.01), but not related to BMC and BMD. TBS was significantly correlated with stiffness (râeuro0/00=âeuro0/000.64; pâeuro0/00=âeuro0/000.007), independently of bone mass. Using stepwise multiple regression models, we failed to demonstrate that the combination of BMD and TBS was better at explaining mechanical behavior than either variable alone. However, the combination TBS, Tb.Th, and BMC did perform better than each parameter alone, explaining 79 % of the variability in stiffness. CONCLUSIONS: In our study, TBS was associated with microarchitecture parameters and with vertebral mechanical behavior, but TBS did not improve prediction of vertebral biomechanical properties in addition to aBMD.

Cytochrome P450 and ABCB1 genetics: association with quetiapine and norquetiapine plasma and cerebrospinal fluid concentrations and with clinical response in patients suffering from schizophrenia. A pilot study.

Variability in response to atypical antipsychotic drugs is due to genetic and environmental factors. Cytochrome P450 (CYP) isoforms are implicated in the metabolism of drugs, while the P-glycoprotein transporter (P-gp), encoded by the ABCB1 gene, may influence both the blood and brain drug concentrations. This study aimed to identify the possible associations of CYP and ABCB1 genetic polymorphisms with quetiapine and norquetiapine plasma and cerebrospinal fluid (CSF) concentrations and with response to treatment. Twenty-two patients with schizophrenia receiving 600 mg of quetiapine daily were genotyped for four CYP isoforms and ABCB1 polymorphisms. Quetiapine and norquetiapine peak plasma and CSF concentrations were measured after 4 weeks of treatment. Stepwise multiple regression analysis revealed that ABCB1 3435C > T (rs1045642), 2677G > T (rs2032582) and 1236C > T (rs1128503) polymorphisms predicted plasma quetiapine concentrations, explaining 41% of the variability (p = 0.001). Furthermore, the ABCB1 polymorphisms predicted 48% (p = 0.024) of the variability of the Δ PANSS total score, with the non-carriers of the 3435TT showing higher changes in the score. These results suggest that ABCB1 genetic polymorphisms may be a predictive marker of quetiapine treatment in schizophrenia.

Family burden related to clinical and functional variables of people with intellectual disability with and without a mental disorder

Few studies have been found that to assess the factors that explain higher levels of familyburden in adults with intellectualdisability (ID) and intellectualdisability and mental disorders (ID-MD). The aims of this study were to assess familyburden in people with ID and ID-MD and to determine which sociodemographic, clinical and functionaldisabilityvariables account for familyburden. The sample is composed of pairs of 203 participants with disability and their caregivers, of which 33.5% are caregivers of people with ID and 66.5% of ID-MD. Assessments were performed using scales of clinical and functionaldisability as the following instruments: Weschler Adult Intelligence Scale-III (WAIS-III), Inventory for Client and Agency Planning (ICAP), Psychiatric Assessment Schedule for Adults with Development Disability (PAS-ADD checklist), Disability Assessment Schedule of the World Health Organization (WHO-DAS-II) and familyburden (Subjective and Objective FamilyBurden Inventory - SOFBI/ECFOS-II). People with ID-MD presented higher levels of functionaldisability than those with ID only. Higher levels of familyburden were related to higher functionaldisability in all the areas (p < 0.006-0.001), lower intelligence quotient (p < 0.001), diagnosis of ID-MD (p < 0.001) and presence of organic, affective, psychotic and behavioral disorders (p < 0.001). Stepwise multiple regression showed that behavioral problems, affective and psychotic disorder, disability in participation in society, disability in personal care and presence of ID-MD explained more than 61% of the variance in familyburden. An integrated approach using effective multidimensional interventions is essential for both people with ID and ID-MD and their caregivers in order to reduce familyburden.

Influence of first morning urine volume, fasting blood glucose and glycosylated hemoglobin on first morning urinary albumin concentration

The aim of the present study was to evaluate the effect of first morning urinary volume (collected on three different non-consecutive days), fasting blood glucose (determined on the first and third days of urine collection), and glycosylated hemoglobin (determined on the first and third days of urine collection) on the albumin concentration in first morning urine samples collected on three different days. We found 3.6% asymptomatic bacteriuria in the urine samples; therefore, every urine sample must be tested to exclude infection. One hundred and fifty urine samples were provided by 50 IDDM patients aged 21.9 ± 7 (12-38) years with a disease duration of 6.8 ± 5.8 (0.4-31) years attending the Diabetes Clinic at the State University Hospital of Rio de Janeiro. There were no differences in albumin concentration (6.1 vs 5.8 vs 6.2 µg/ml; P = NS) or urinary volume (222.5 vs 210 vs 200 ml) between the three samples. In addition, there were no differences in fasting blood glucose (181.9 ± 93.6 vs 194.6 ± 104.7 mg%; P = NS) or glycosylated hemoglobin (HbA1)(8.4 ± 1.3 vs 8.8 ± 1.5%; P = NS) between the first and third blood samples. Six patients (group 1) had a mean urinary albumin concentration of more than 20 µg/ml for the three urine samples. This group was compared with the 44 patients (group 2) with a mean urinary albumin concentration for the three urine samples of less than 20 µg/ml. No difference was found between groups 1 and 2 in relation to fasting blood glucose (207.1 ± 71.7 vs 187.6 ± 84.6 mg/dl), HbA1 (8.1 ± 0.9 vs 8.6 ± 1.1%) or urinary volume [202 (48.3-435) vs 246 (77.3-683.3) ml]. Stepwise multiple regression analysis with albumin concentration of first morning urine samples as the dependent variable, and urinary volume, fasting blood glucose and glycosylated hemoglobin as independent variables, showed that only 12% (P = 0.01) of the albumin concentration could be accounted for by the independent effect of morning urine volume on the first day of urine collection. No urine samples showed a change in the cutoff level of 20 µg/ml of albumin concentration as the result of volume. Fasting blood glucose and glycosylated hemoglobin did not influence the urinary albumin concentration. Considerable variability in urinary albumin concentration was found in the three morning urine samples with a mean intraindividual coefficient variation of 56%. In conclusion, in the present study, urinary volume had a minimal, though not constant, effect on first morning urinary albumin concentration. Day-to-day metabolic and clinical control of IDDM patients, except probably for ketoacidosis, should not contraindicate microalbuminuria screening in first morning urine samples

Bone mineral density of the lumbar spine of Brazilian children and adolescents aged 6 to 14 years

The authors performed a study of bone mass in eutrophic Brazilian children and adolescents using dual-energy X-ray absorptiometry (DXA) in order to obtain curves for bone mineral content (BMC) and bone mineral density (BMD) by chronological age and correlate these values with weight and height. Healthy Caucasian children and adolescents, 120 boys and 135 girls, 6 to 14 years of age, residents of São Paulo, Brazil, were selected from the Pediatric Department outpatient clinic of Hospital São Paulo (Universidade Federal de São Paulo). BMC, BMD and the area of the vertebral body of the L2-L4 segment were obtained by DXA. BMC and BMD for the lumbar spine (L2-L4) presented a progressive increase between 6 and 14 years of age in both sexes, with a distribution that fitted an exponential curve. We identified an increase of mineral content in female patients older than 11 years which was maintained until 13 years of age, when a new decrease in the velocity of bone mineralization occurred. Male patients presented a period of accelerated bone mass gain after 11 years of age that was maintained until 14 years of age. At 14 years of age the mean BMD values for boys and girls were 0.984 and 1.017 g/cm², respectively. A stepwise multiple regression analysis of paired variables showed that the "vertebral area-age" pair was the most significant in the determination of BMD values and the introduction of a third variable (weight or height) did not significantly increase the correlation coefficient.

Swimming training down-regulates plasma leptin levels, but not adipose tissue ob mRNA expression

The aim of the present study was to assess the effects of endurance training on leptin levels and adipose tissue gene expression and their association with insulin, body composition and energy intake. Male Wistar rats were randomly divided into two groups: trained (N = 18) and sedentary controls (N = 20). The trained group underwent swimming training for 9 weeks. Leptin and insulin levels, adiposity and leptin gene expression in epididymal and inguinal adipose tissue were determined after training. There were no differences in energy intake between groups. Trained rats had a decreased final body weight (-10%), relative and total body fat (-36 and -55%, respectively) and insulin levels (-55%) compared with controls (P < 0.05). Although trained animals showed 56% lower leptin levels (2.58 ± 1.05 vs 5.89 ± 2.89 ng/mL in control; P < 0.05), no difference in leptin gene expression in either fat depot was demonstrable between groups. Stepwise multiple regression analysis showed that lower leptin levels in trained rats were due primarily to their lower body fat mass. After adjustment for total body fat, leptin levels were still 20% (P < 0.05) lower in exercised rats. In conclusion, nine weeks of swimming training did not affect leptin gene expression, but did lead to a decrease in leptin levels that was independent of changes in body fat.

Fish consumption preferences and factors influencing it

Fish consumption preferences are affected by individuals’ socioeconomic characteristics. The aims of the present paper were (i) to obtain information on fish consumption level and frequency; (ii) to investigate the associations between the socioeconomic characteristics of consumers and their preferences; and (iii) to examine the influence of determinants on fish consumption. Data were gathered through a questionnaire completed by a total of 127 randomly selected individuals from different socioeconomic backgrounds from the Antakya, Turkey. The average consumption was found to be 2.98 kg/person/year for fish. Anchovies, gilt-head sea bream, and sea bass were reported as the most consumed three species, respectively. Significant differences in fish consumption were found among age groups, gender groups, and education groups, as well as between marital statuses. A majority of the consumers eat fish once a month throughout the year or only during the winter months. Fish consumption level and frequency were significantly positively correlated with education (p<0.01), income (p<0.05) and total meat consumption (p<0.01). The stepwise multiple regression model explained 41.7% (p<0.01) of the total variance for fish consumption. The amount and frequency of the consumption in the region, which is very far below the world and Turkey average especially for lower socioeconomic groups and for less-consumed fish species, can be increased by certain policies, such as training, advertising and different marketing strategies. Moreover, consumption should be distributed equally throughout the year instead of consuming only in certain seasons.

Personality type and self-directed learning

This correlational study was designed to investigate the relationship between self-directed learning and personality type. A sample of 133 graduate and undergraduate education students completed the MBTI and the SDLRS. Two hypotheses were examined: (a) scores on the intuitive scale will account for a significant amount of the variance in the prediction of selfdirected learning readiness and, (b) scores on the introverted scale will account for a significant amount of the variance in self-directed learning readiness. Stepwise multiple regression analyses indicated that psychological type accounts for 28% of the variance in self-directed learning. Support for the first hypothesis was found with 15% of the variance in selfdirected learning accounted for by intuition. The second hypothesis was not supported. Introversion accounted for 13% of the variance but in a negative manner. Results of this study indicate that personality type does influence the ability of the learner to be self-directed in studies. These findings add another dimension for the adult educator to consider when attempting to develop self-directedness in learners.