SEARCH · PubMed Health

Results for “multiple linear regression”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 127 records · Page 7Linked to original sources

PubMed1997

Determinants of parental authorization for involvement of newborn infants in clinical trials.

OBJECTIVE: Parents have the right to decide on behalf of their infants whether to enroll them in controlled clinical trials. We determined the degree to which such parental decisions are influenced by risk and benefit considerations compared with other factors. DESIGN: Cross-sectional survey. PARTICIPANTS: Parents who had recently given or declined consent to one of three controlled trials in the neonatal intensive care unit. INTERVENTION: Parents were asked to complete a questionnaire that consisted of 15 sociodemographic items and 13 scaled responses to statements assessing the probability and magnitude of risk and benefit as well as perceived illness severity, attitudes toward research, and the consent process. ANALYSIS: Responses were subjected to factor analysis to identify underlying constructs. The sample was then randomly split, and multiple regression was performed on each half. RESULTS: The response rate was 83% (103 of 124) for those who had consented and 86% (37 of 43) for those who had declined. Factor analysis yielded three factors: (1) illness severity, (2) perceptions of risk or benefit and attitudes to research, and (3) sociodemographic characteristics. Multiple linear regression showed a significant multiple correlation of consent decision (r = .502), but only the second factor contributed. The analyses on split halves of the sample were comparable. Thirty-two percent of all parents agreed with the statement, "I would prefer to have the doctors advise me whether my baby should be in the study, rather than asking me to decide." CONCLUSIONS: In making consent decisions on behalf of their newborn infants, parents are influenced by risk and benefit assessments, attitudes toward research, and the integrity of the consent process. Illness severity or sociodemographic characteristics do not seem to be of similar importance. Rather than making the decision alone, a significant minority of parents would prefer to have the physicians advise them whether to volunteer their infants for a clinical trial.

Attitude to Health↗

PubMed1993

Black clouds. Work load, sleep, and resident reputation.

OBJECTIVE: Although it is assumed that residents in a specific training program will have comparable experiences, residents commonly perceive that some have consistently more difficult times on call. Such residents in our program are said to have "black clouds." We sought to determine if these perceptions were related to differences in real work load. METHODS: We collected data about the on-call experiences of our first-year pediatric residents (PL-1s) for 358 days (1355 on-call experiences) during the 1984-1985 academic year. Every PL-1 (n = 19) reported the following data the morning after each night on call: hours of sleep, number of admissions, total number of patients, number of deaths, number of transfers to the pediatric intensive care unit, number of delivery room trips, and a subjective assessment of work load, using a three-point scale. The reputation of each house officer was determined by asking all residents in the program (PL-1s, PL-2s, and PL-3s) to rate each other three times during the year regarding how hard they worked on call. RESULTS: There were significant differences among PL-1s in how difficult they perceived their work load to be and in how much they slept (P < .001 using analysis of variance). However, actual work load (as measured by the number of either admissions or patients) did not vary significantly among the residents. There was a strong negative association between self-perception of work load and hours of sleep (r = -.75; 95% confidence interval, -0.73 to -0.76). Sleep was the major predictor of perceived work load (multiple R2 = .563 using multiple linear regression analysis). The absence of an association between perceived and actual work load is attributed to large differences in the residents' working styles. This is evidenced by a wide range of correlations among PL-1s between the number of admissions and hours of sleep (range of r values, -.66 to -.16). A reputation for difficult on-call experiences was strongly associated with few hours of sleep (r = -.77; 95% confidence interval, -0.49 to -0.91), but not with actual work load measured by the number of admissions, patients, deaths, or other variables. Sleep was the major predictor of reputation (multiple R2 = .567 using multiple linear regression analysis). CONCLUSIONS: Some residents did have a black cloud; they slept less, perceived that they worked harder than average, and had a reputation for having difficult on-call experiences. Residents with a black cloud function differently from their colleagues; for example, some may be inefficient, while others may create extra work for themselves. Residency program directors must recognize these functional differences to effectively evaluate and counsel house officers.

Analysis of Variance↗

PubMed1996

Insulin-like growth factor I as a predictor of cortical bone mass in a long-term study of ovariectomized and estrogen-treated rats.

The relationship between serum insulin-like growth factor I (IGF-I) and cortical bone mass, formation and resorption and length of bone in a long-term experiment on intact, ovariectomized and estrogen-treated/substituted rats was studied by using multiple linear regression analysis. The study comprised intact rats killed at 2, 6, 9, 12, 15, and 24 months of age, rats ovariectomized 6 months old and killed at 9, 12, 15, and 24 months of age, and intact and ovariectomized rats treated with a low dose of estrogen for 8 months before they were killed at 24 months of age. Serum IGF-I, bone length and total, subperiosteal and subendocortical bone mass in mid-diaphyseal cross sections of the femur were determined. Changes in the latter two variables, respectively, represent the net result of subperiosteal bone formation and subendocortical bone resorption. Multiple linear regression analysis showed that IGF-I was a positive determinant of cortical bone mass and subperiosteal bone formation. In aged rats, IGF-I was also a positive determinant of bone length, whereas IGF-I was not found to be a determinant of subendocortical bone resorption. The analyses showed that long-term treatment of aged rats with a low dose of estrogen had a dual effect on cortical bone by inhibiting subperiosteal formation and subendocortical resorption. The results revealed a relationship between endogenous circulating IGF-I and local anabolic actions of the growth factor in bones. Thus, IGF-I may be a valuable serum marker of cortical bone formation and length of long bones when considering estrogen-depleted and estrogen-treated rats.

Animals↗

PubMed1976

Specificity of release of biogenic amines from isolated rat brain tissue as a function of the meta substituent of N-ethylamphetamine derivatives.

The actions of several meta-substituted N-ethylamphetamines on the release, uptake and catabolism of biogenic amines were investigated in vitro in two regions of rat brain. Each of the compounds released 3H-norepinephrine from chopped cerebral cortex, and 3H-dopamine and 3H-5-hydroxytryptamine from chopped corpus striatum. Stepwise multiple linear regression analyses indicated that the potencies (EC50 values) with which the compounds released 3H-norepinephrine were inversely related (r = 0.94) to the size of the meta substituent. Addition of substituent lipophilic or electronic effects to the regression equation did not greatly improve the correlation. However, the maximal release of 3H-norepinephrine produced by these compounds was directly related only to substituent lipophilic effects (r = 0.95). The potencies and maxima with which the compounds released 3H-dopamine or 3H-5-hydroxytryptamine and the potencies with which the compounds inhibited the uptake of the 3H-amines were not correlated with substituent size, lipophilicity or electronegativity. Variations in meta substituent also altered the degree to which the release 3H-amines were metabolized by monoamine oxidase. Multiple linear regression analyses indicated that for each of the 3H-amines, the maximal percent increase in 3H-deaminated metabolites produced by the drugs was directly correlated only with substituent size (r = 0.85, 0.89 and 0.95 for 3H-norepinephrine, 3H-dopamine and 3H-5-hydroxytryptamine, respectively). Comparisons among the potencies with which compounds released the various 3H-amines indicated that, in general, as substituent size increased, the compounds became relatively less potent in releasing 3H-catecholamines and relatively more potent in releasing 3H-5-hydroxytryptamine. Thus, these studies indicate that variations in the meta substituent group of N-ethylamphetamine derivatives determine the affinity of the derivatives for norepinephrine nerve endings and the relative affinities of the derivatives for norepinephrine, dopamine and 5-hydroxytryptamine nerve endings.

Amphetamines↗

PubMed1991

All cause mortality and its determinants in middle aged men in Finland, The Netherlands, and Italy in a 25 year follow up.

STUDY OBJECTIVE: The aims were (1) to compare all cause mortality in population samples of different cultures; and (2) to cross predict fatal event by risk functions involving risk factors usually measured in cardiovascular epidemiology. DESIGN: The study was a 25 year prospective cohort study. The prediction of all cause mortality was made using the multiple logistic equation as a function of 12 risk factors; the prediction of months lived after entry examination was made by the multiple linear regression using the same factors. POPULATION SAMPLES: There were five cohorts of men aged 40-59 years, from Finland (two cohorts, 1677 men), from The Netherlands (one cohort, 878 men), and from Italy (two cohorts, 1712 men). SETTING: The Finnish cohorts came from geographically defined rural areas, the Dutch cohort from a small town in central Holland, and the Italian cohorts from rural villages in northern and central Italy. MEASUREMENTS AND MAIN RESULTS: All cause mortality was highest in Finland (557 per 1000), and lower in The Netherlands (477) and in Italy (475). The solutions of the multiple logistic function showed the significant and almost universal predictive role of certain factors, with rare exceptions. These were age, blood pressure, cigarette smoking, and arm circumference (the latter with a negative relationship). Similar results were obtained when solving a multiple linear regression equation predicting the number of months lived after entry examination as a function of the same factors. The prediction of fatal events in each country, using the risk functions of the others, produced limited errors, the smallest one being -2% and the largest +11%. When solving the logistic model in the pool of all the cohorts with the addition of dummy variables for the identification of nationality, it also appeared that only a small part of the mortality differences between countries is not explained by 12 available risk factors. CONCLUSIONS: A small set of risk factors seems to explain the intercohort differences of 25 year all cause mortality in population samples of three rather different cultures.

Adult↗

PubMed1994

The use of epidemiological concepts and techniques to discern factors associated with the nitrate concentration of well water on swine farms in the USA.

This epidemiological study investigates the relationship between various factors associated with swine farms and the nitrate concentration of well water in the USA. Through a random sampling procedure, 605 swine farms located in 18 states were selected for inclusion in this study. A total of 631 well water samples were collected from these farms and tested for a variety of elements and compounds. The concentrations of nitrate, nitrite, sulfate, chloride, sodium, potassium, ammonia, fluoride, bromide and lithium were determined by an ion chromatograph while an inductively coupled argon plasma emission spectrophotometer was used to determine the concentrations of calcium, magnesium, barium, zinc, iron and phosphate. Data concerning various farm factors were gathered via a personally administered questionnaire. The data were examined using both multiple linear regression and logistic regression. Results indicate that 53.6% (338/631) of the wells contained detectable levels of nitrate, 11.7% (74/631) had nitrate levels exceeding 45 ppm and 4.3% (27/631) exceeded 100 ppm. Logistic models demonstrated an association between nitrate concentrations > 45 ppm, increasing water potassium levels and wells < 100 ft deep. Nitrate levels > 100 ppm were related to increasing water concentrations of potassium, magnesium, barium and zinc, wells 6-10 years old, increasing distance from the study farm to the nearest cattle farm and a greater distance to the nearest waterway located off the study farm. A negative association was seen between nitrate concentrations > 100 ppm, the water level of sulfate, and the use of the same well to supply both the household and livestock. Multiple linear regression models revealed a positive association between increasing nitrate concentration and the water levels of chloride, calcium, zinc and the greater number of miles from the study farm to the nearest farm with cattle or sheep. A negative association was noted between the concentration of well water nitrate and the water levels of sulfate and ammonia, the use of water treatment, the number of miles to the nearest farm with poultry, the employment of water treatment and the use of the same well to supply water to both livestock and the household.

Agriculture↗

PubMed1993

Menadione nicotinamide bisulfite is a bioactive source of vitamin K and niacin activity for chicks.

Young chicks were fed a vitamin K-deficient soybean concentrate basal diet containing graded levels of menadione from menadione nicotinamide bisulfite (MNB) or menadione dimethyl-pyrimidinol bisulfite (MPB) to assess prothrombin time as a function of menadione intake. Prothrombin time decreased linearly as menadione dose increased from 0 to 400 micrograms/kg. Multiple linear regression slope-ratio calculations indicated that both sources of menadione were of equal potency. To assess niacin bioactivity of MNB, graded doses of nicotinamide (0 to 5 mg/kg) from MNB or nicotinamide were added to a niacin-deficient diet based upon corn, corn gluten meal and vitamin-free casein. Weight gain increased linearly as a function of nicotinamide dose, and multiple linear regression analysis of weight gain as a function of supplemental nicotinamide intake revealed no significant differences in slope between the two sources of nicotinamide. Using excess doses, MNB was compared with MPB in acute (single crop intubation) or chronic (fed in the diet for 14 d) toxicity trials. With a single menadione dose of 1600 mg/kg body wt, weight gain in the subsequent 14-d period was reduced by MNB but not by MPB. Mortality rates of 25 and 17% occurred for MPB and MNB, respectively, at this dose level. Doses lower than 1600 mg/kg body wt caused neither morbidity nor mortality. When provided in the diet for a 14-d feeding period, menadione doses of 3000 mg/kg diet from MNB reduced gain, feed intake, gain:feed ratio and blood hemoglobin concentration. Menadione doses of 6000 mg/kg diet were required to produce morbidity of this type when MPB was fed.(ABSTRACT TRUNCATED AT 250 WORDS)

Administration, Oral↗

PubMed1991

Predictors and prevalence of obstructive sleep apnoea and snoring in 1001 middle aged men.

One thousand and one men, aged 35-65 years, were identified from the age-sex register of one group general practice. Over four years 900 men were visited at home and asked questions about symptoms potentially related to sleep apnoea and snoring. Height, weight, neck circumference, resting arterial oxygen saturation (SaO2), and spirometric values were also determined. All night oximetry was then performed at home and the tracing analysed for the number of dips in SaO2 of more than 4%. Subjects with more than five dips of 4% SaO2 or more per hour were invited for sleep laboratory polysomnography. Seventeen per cent of the men admitted to snoring "often." Multiple linear regression techniques identified and ranked neck circumference (r2 = 7.2%), cigarette consumption (r2 = 3.4%), and nasal stuffiness (r2 = 2%) as the only significant independent predictors of snoring. Together these account for at least a sixfold variation in the likelihood of being an "often" snorer. Forty six subjects (5%) had greater than 4% SaO2 dip rates of over five an hour and 31 of these had full sleep studies. Three subjects had clinically obvious and severe symptomatic obstructive sleep apnoea, giving a prevalence of three per 1001 men (0.3%; 95% confidence interval 0.07-0.9%). Eighteen men had obstructive sleep apnoea only when supine and in 10 the cause of the SaO2 dipping on the original home tracing was not elucidated. The greater than 4% SaO2 dip rates correlated with the history of snoring. Multiple linear regression techniques identified and ranked neck circumference (r2 = 7.9%), alcohol consumption (r2 = 3.7%), age (r2 = 1%) and obesity (r2 = 1%) as the only significant independent predictors of the rate of overnight hypoxic dipping. This study shows that snoring in this randomly selected population correlates best with neck size, smoking, and nasal stuffiness. Obstructive sleep apnoea, defined by nocturnal hypoxaemia, correlates best with neck size and alcohol, and less so with age and general obesity.

Adult↗

PubMed1996

Interleukin-2 receptor levels in sera of patients with rheumatoid arthritis treated with sulfasalazine, parenteral gold, or placebo.

OBJECTIVE: To evaluate the associations of soluble serum interleukin-2 receptor (sIL-2R) levels in patients with rheumatoid arthritis (RA) with clinical and laboratory measures of disease activity and the predicted response to therapy. METHODS: sIL-2R levels were determined by ELISA in 137 patients with RA, not previously treated with 2nd line therapy. Patients were enrolled in a prospective, randomized, placebo controlled trial of sulfasalazine (SSZ) versus gold sodium thiomalate (GSTM), sponsored by the Cooperative Systematic Studies of Rheumatic Diseases. Using correlation analysis and regression modeling, the clinical utility of sIL-2R as a measure of disease activity and predictor of outcome was assessed. RESULTS: 91 women and 46 men with a mean age of 51 +/- 13 years and mean duration of disease of 64 +/- 78 months participated in this study. The mean sIL-2R level in all patients with RA was markedly elevated (980 +/- 589 U/ml) compared with that in healthy control subjects (446 +/- 196 U/ml; p = < 0.0001). There was no correlation between the sIL-2R levels and the joint pain/tenderness count, either at study entry or completion. There were significant positive correlations between the baseline sIL-2R and baseline erythrocyte sedimentation rate (ESR) and between the change in sIL-2R and the change in ESR. Both a multiple linear regression model and a multiple logistic regression model showed that baseline sIL-2R levels were not predictive of clinical outcome. CONCLUSION: sIL-2R levels are significantly elevated in patients with active RA and correlate positively with ESR. However, these results indicate that in patients with early RA, sIL-2R levels are neither associated with standard disease activity criteria nor predictive of the response to therapy with SSZ or GSTM, even after controlling for the simultaneous effects of other important clinical variables.

Adult↗

PubMed1995

Use of anthropometric measurements in assessing risk for coronary heart disease: a useful tool in worksite health screening?

The study examined the association between the anthropometric measurements body mass index (BMI), waist/hip ratio (WHR), and waist/thigh ratio (WTR) and cardiovascular risk factors, and assessed whether a combination of BMI and WHR could be used in routine screening of risk for cardiovascular arteriosclerotic disease at worksites. The data were obtained from a cross-sectional survey designed to assess the nutritional situation, with special reference to cardiovascular risk factors. The study population comprised 372 healthy men working on platforms in the North Sea. Serum cholesterol, triglyceride, fibrinogen, and blood pressure were positively related to the anthropometric variables, while high-density lipoprotein (HDL) was inversely related with them. The relations remained after adjusting for possible confounders, such as age, smoking, physical activity, and an indicator of dietary fat intake. In stepwise multiple linear regression models, BMI, WHR, and WTR were positively related to serum cholesterol, triglycerides, fibrinogen, diastolic blood pressure, and systolic blood pressure, and inversely related to HDL. When controlling for the anthropometric variables WHR and WTR, BMI was not independently related to fibrinogen and risk score. WHR and WTR were not independently related to systolic and diastolic blood pressure, and WTR was in addition not related to triglycerides when controlling for BMI. Overall, the anthropometric variables BMI and WHR were considered the best predictors for CAD risk when taking several risk factors into consideration. A joint variable between BMI and WHR, called "body score", constituted the four categories lean, lean android, overweight gynoid, and overweight ovoid. This body score was positively associated with levels of serum lipids, fibrinogen, and blood pressure, and inversely associated with HDL. In stepwise multiple linear regression models, controlling for possible confounding variables, body score was positively related to CAD risk. Dividing the risk score into tertiles, about 51% of the lean were in the first, while 46% of the overweight ovoid were in the third tertile. Those classified as lean android or overweight gynoid had about the same distribution, namely between 31% and 39% in each tertile if the two categories were combined. These data support the hypothesis that BMI, WHR, and WTR are independent predictors for risk factors for CAD among oil workers, and that combinations of BMI and WHR are strong enough predictors to be useful in routine screening for CAD risk at worksites. Based on these findings, supported by data from the literature, a matrix aimed at screening for follow-up at worksites is proposed.

Adult↗

PubMed1997

Serum leptin is increased in growth hormone-deficient adults: relationship to body composition and effects of placebo-controlled growth hormone therapy for 1 year.

The gene product from the ob gene, leptin, has recently been characterized in humans. The circulating level of leptin is related to body mass index (BMI) and more closely to estimates of total body fat, whereas visceral fat has been reported to be of minor importance. However, it is unknown if leptin is directly regulated by hormones that influence substrate metabolism and body composition. We studied leptin in adult growth hormone (GH)-deficient (GHD) patients substituted with GH treatment for 12 months in a parallel double-blind, placebo-controlled study. Twenty-seven GHD adults aged 44.9 +/- 1.9 years underwent anthropometric measurements for determination of regional and total body fat (BMI, waist to hip ratio [WHR], computed tomographic [CT] scan, dual-energy x-ray absorptiometry [DEXA] scan, and bioimpedance analysis [BIA]) before and after 12 months of placebo-controlled GH substitution (2 IU/m2) in a parallel design. The same measurements were performed in 42 healthy adults aged 39.1 +/- 1.7 years. The logarithm of serum leptin levels correlated positively with abdominal subcutaneous fat and total body fat (BIA and DEXA) in untreated GHD patients and healthy subjects. Fasting insulin did not correlate with leptin levels in either of the groups. After 12 months of GH administration, the body composition of GHD patients was significantly changed with respect to a marked decrease in body fat. The relations of leptin to the estimates of body fat were maintained, and leptin was furthermore related to BMI and fasting insulin. In multiple linear regression analyses, additional estimates of visceral adiposity (intraabdominal fat and maximal anterior-posterior diameter determined by CT scan) were significant determinants of leptin in the healthy subjects. The increase in fasting insulin levels during GH substitution correlated negatively with the reduction in leptin levels (r = -.823, P = .003). At baseline, leptin levels were increased in the patients compared with controls in both sexes (women, 21.8 +/- 3.3 v 11.3 +/- 1.4 ng/mL, P = .002; men, 8.1 +/- 1.2 v 4.7 +/- 0.7 ng/mL, P = .008). Leptin levels were similar in GHD patients treated for 12 months compared with healthy controls for both women and men (women, 15.9 +/- 2.3 and 11.3 +/- 1.4 ng/mL, P = .163; men, 7.1 +/- 2.8 and 4.7 +/- 0.7 ng/mL, P = .759). In healthy adults and in GHD patients, leptin levels were significantly higher in women than in men (11.3 +/- 1.4 v 4.7 +/- 0.7 ng/mL, P < .001; 21.8 +/- 3.3 v 8.1 +/- 1.2 ng/mL, P < .001). Gender remained a significant determinant of leptin levels in several models of multiple linear regression analysis also including age, estradiol levels, insulin, and estimates of body fat. We conclude that leptin is increased but not differently regulated in GHD patients compared with normal subjects, and that leptin levels are closely related to estimates of body fat. This relationship is maintained during a decrease in body fat due to GH substitution.

Absorptiometry, Photon↗

PubMed1991

When is a diagnosis worth making? A statistical comparison of two prediction strategies.

For a setting in which the sole goal is to predict a criterion accurately, two strategies are compared, (1) multiple linear regression directly from a set of predictors and (2) using predictors to diagnose individuals and then using only the diagnosis to predict the criterion. These are mathematical models of two common methods for making clinical predictions. This article derives equations for each statistical strategy's validity (and a formula comparing them, for a special case). Prediction accuracies are compared over a simplified but broad parameter space and four conclusions follow. Changing disorder prevalences (in the range 0.1 to 0.5) have small effect on the relative preferability of the two strategies; the effect of varying prevalence differs according to population separation on predictors and criterion. As within-population covariances of predictors with criterion decline below zero (assuming variables are scaled so that the less common population scores higher on predictors and criterion), diagnoses become increasingly preferable as a prediction strategy; as they rise above zero the opposite trend is observed. As populations become better separated on predictors or criterion, diagnoses compare more favorably. Most importantly, over almost all of the parameter space which one would expect to encounter in clinical psychology and psychiatry, multiple linear regression is much superior.

Diagnosis, Differential↗

PubMed1990

Mechanical and metabolic determinants of myocardial stunning in extracorporeally perfused pig hearts.

The purpose of this study was to apply step-wise multiple linear regression analysis retrospectively to an array of mechanical and metabolic measurements chosen because they had the potential to predict the extent of contractile recovery from a prescribed duration of myocardial ischemia. Data were acquired from the extracorporeally perfused, intact, working pig heart which was rendered regionally ischemic (60% reduction in anterior descending coronary flow) for 45 minutes and reperfused to aerobic levels for a final 30-50 minutes. Mechanical recovery was defined by the percentage systolic shortening and the area circumscribed by left ventricular pressure-segment length loops. Data were taken from 39 control hearts and from 16 hearts treated with oxfenicine, an agent which we have previously used to alter mechanical function by its interference with fatty acid metabolism. Despite the fixed nature of the protocol in affecting ischemic hypoperfusion, a wide range of mechanical responses encompassing hypo- and dyskinesis was produced during ischemia, followed by mechanical stunning during reflow. Of the parameters surveyed, regional indices of mechanical performance, together with perfusate pH and PCO2, best predicted recovery. Along with the heart rate, these predictors gave correlations of 0.875 for percentage systolic shortening and 0.766 for the length-pressure loop in control hearts. The analyses were also sensitive to the influence of pharmacological intervention with oxfenicine in that several parameters lost statistical significance for percentage systolic shortening and two were added (heart rate and end-diastolic length) for the length-pressure loop. Separate statistical models for oxfenicine-treated hearts gave correlations of 0.905 for percentage systolic shortening and 0.915 for the length-pressure loop. The data suggest that step-wise multiple linear regression analysis provides new insights toward our understanding of the mechanisms of mechanical stunning in myocardial reperfusion.

Animals↗

PubMed1996

Development of an experimental diet for determining bioavailable choline concentration and its application in studies with soybean lecithin.

Attempts to determine choline bioavailability have encountered criticism of experimental diets and protocol. Our objectives were to develop a choline-deficient soy isolate diet to quantify bioavailable choline concentration of soybean lecithin and to compare results to those obtained with a purified diet. In Assay 1, weight gain of chicks fed a choline-free crystalline amino acid diet responded linearly (P < .01) to graded doses of choline chloride, fluid lecithin (FL), or deoiled lecithin (DL). Multiple linear regression analysis indicated a bioavailable choline content of 2.3 and 3.7% for FL and DL, respectively. In Assay 2, a choline-deficient soy isolate diet was supplemented with 2-amino-2-methyl-1-propanol (AMP) to inhibit choline biosynthesis. Weight gain, feed intake, and feed efficiency increased (P < .05) markedly with the addition of choline chloride, but supplemental methionine or betaine had no effect (P > .10). Addition of 10% soybean meal to the diet severely deficient in choline per se produced a growth response (P < .05), whereas the same addition to the diet made adequate in choline did not elicit a growth response. In Assay 3, addition of graded levels of choline chloride, FL, or DL to the choline-deficient soy isolate diet containing AMP resulted in a linear (P < .01) increase in weight gain and feed intake. Multiple linear regression analysis indicated a bioavailable choline content of 2.0 and 3.5% for FL and DL, respectively. The AMP-containing soy isolate diet seems well suited for determination of bioavailable choline content in products containing secondary nutrients. Bioavailable choline content of FL and DL was similar to estimates of total choline content, suggesting that the choline in these products was fully available.

Animals↗

PubMed1995

Factors associated with weight for height and skinfold thickness in British children.

STUDY OBJECTIVE: To examine the associations of social and biological factors with measures of obesity in children. DESIGN: The study had a cross sectional design. SETTING: The analyses were based on data from two national study of health and growth cross sectional surveys. The "representative sample" comprised 1990 data from 22 English areas and 1990-91 data from 14 Scottish areas; the "inner city sample" comprised 1991 data from 20 English areas. PARTICIPANTS: The subjects were primary school children aged mainly 5-11 years living in England and Scotland. The "representative" sample included 10,628 children--6463 living in England and 4165 living in Scotland. The "inner city" sample included 7049 children--2183 white, 1124 Afro-Caribbean, 2696 Indian subcontinent, and 1046 from other groups. Due to missing values on continuous variables, 8374 children were included in the analyses. MEASUREMENTS AND MAIN RESULTS: The relation between social environment and childhood overweight was studied using several indicators of obesity. Triceps, subscapular, the sum of triceps and subscapular skinfolds, and weight for height were used as dependent variables. The analyses were carried out in two stages. Firstly, multiple linear regression analyses were used to assess the factors associated with dependent variables treated as continuous. Secondly, multiple linear logistic regression analyses were used to examine the association between independent factors and overweight and fatness defined as binary variables. Birth weight, mother's body mass index (BMI), and father's BMI were consistently associated (p < 0.001) in all models and were the variables that contributed most to the explained variation in the dependent variables. In the multiple regression analyses there was a consistent interaction between the effects of ethnic origin and family size on each outcome variable. In the logistic regression analyses the interaction was not significant, and highly significant associations between both overweight and fatness with the number of children were shown. Ethnic group was not significantly associated with overweight but it was with fatness. The strengths of the remaining significant associations were slight and inconsistent in relation to the dependent variables or the type of analysis. CONCLUSION: Very few variables were associated with measures of overweight and fatness. The only useful factor that was highly associated with all measures of fatness was the parents' BMI. Strategies to prevent childhood obesity should be aimed at the total population and special emphasis should be placed on families in which one or both parents are overweight.

Birth Weight↗

PubMed1997

The internal dose of passive smoking at home depends on the size of the dwelling.

As part of a longitudinal study two urine samples (survey 1 in 1991 and survey 2 in 1992) were collected from 602 elementary school children to investigate the relationship between urinary cotinine excretion (UCE) and the daily consumption of cigarettes at home (exposure). Size of the dwelling, educational level, and maternal smoking were taken into consideration as additional predictors. The history regarding parental smoking habits and confounding variables was ascertained by standardized questionnaires completed by the parents. Cotinine was measured using gas chromatography selected ion monitoring. UCE was expressed as cotinine/creatinine (ng/mg). In children with detectable UCE in survey 1 (35%) and in survey 2 (44%) the excretion ranged between 1.5 and 24.7 ng/mg (5-95%) and between 1.2 and 25.2 ng/mg, respectively. UCE measurements in both surveys were highly correlated (r = 0.65, P = 0.0001), and 59.6% of the UCE in survey 2 can be explained in linear regression by the UCE in survey 1. Using multiple linear regression, the categorized number of cigarettes reported to be consumed at home (20 cigarettes and more: 1991, P = 0.0001; 1992, P = 0.0003) and low educational level of the parents (P = 0.011 in 1991, P = 0.04 in 1992) were positively associated with UCE, whereas the size of the dwelling turned out to be negatively associated with UCE (P = 0.12 in 1991, P = 0.001 in 1992). In small dwellings (< or = 80 m2) the effect of exposure on UCE was much more pronounced. In conclusion, a single UCE measurement provides information which is widely stable within a yearly interval and is related to passive smoke history as well as to socio-economic status and the size of the dwelling. The latter variable should be considered as an effect modifier of exposure on internal dose and should be taken into account in future studies on passive smoke exposure.

Analysis of Variance↗

PubMed1990

A comparison of multivariable mathematical methods for predicting survival--III. Accuracy of predictions in generating and challenge sets.

This paper concludes a study of "performance variability" when four methods of multivariable analysis--multiple linear regression, discriminant function analysis, multiple logistic regression, and two arrangements of Cox's proportional hazards regression--were applied to the same stratified random samples of "generating sets" containing seven different statistical distributions of cogent biologic attributes in a composite staging system for a large cohort of patients with lung cancer. Each model developed from the generating sets was also applied for predictions in a previously sequestered "challenge set". Across the different generating sets, the multivariable methods showed good agreement with one another in the stepwise choice of first two powerful predictor variables, but not in the sequence of subsequent choices or in the standardized coefficients assigned to the same collection of "forced" variables. In concordance of predictions for individual patients in the generating sets, the overall proportions of disagreement for pairs of methods ranged from 0 to 28%, and kappa values ranged from 0.49 to 1.00. The accuracy of individual predictions showed relatively similar results when the different methods were applied to the same generating set. Across the generating sets, the different methods showed similar total results but substantial variations in predictions for alive and dead patients. When the models from the generating sets were applied for predictions in the challenge set, the results showed an analogous pattern: similar accuracy within models for overall and live/dead predictions, but substantial variations in live/dead predictions across models derived from different generating sources. The results showed that the multivariable methods often had good agreement with one another in predictions for groups but not for individual persons; and that no single method was superior to the others or to the composite staging system. We conclude that multivariable analytic methods may be most effective and consistent if used to find the few most powerful predictor variables, omitting the many other variables that may be "statistically significant" but less cogent. The powerful predictors may sometimes be best constructed, before the analysis begins, as composite variables containing appropriate unions or ordinal arrangements of elemental candidate variables.

Cohort Studies↗

PubMed1990

The relationship between neck circumference, radiographic pharyngeal anatomy, and the obstructive sleep apnoea syndrome.

We have studied the predictive importance of neck circumference, obesity, and several radiographic pharyngeal dimensions for obstructive sleep apnoea (OSA), in 66 patients. OSA was quantified as the mean hourly number of greater than 4% dips in arterial oxygen saturation during sleep. Neck circumference (correlation coefficient (r) = 0.63, 95% confidence interval (C.I.) 0.46-0.76), obesity index (r = 0.54, 95% C.I. 0.39-0.69), hyoid position (r = 0.40, 95% C.I. 0.17-0.59), soft palate length (r = 0.31, 95% C.I. 0.08-0.51), and hard palate-to-spine angle (r = 0.29, 95% C.I. 0.04-0.49), correlated significantly with saturation dips in single regression analysis. In stepwise multiple linear regression analysis (saturation dip rate as the dependent variable), only neck size and retroglossal space were significant independent correlates (total r2 = 0.42, 95% C.I. 0.22-0.61, p less than 0.0001). We conclude that the relationships between general obesity, hyoid position, soft palate length, and OSA are probably secondary to variation in neck circumference.

Female↗