Improved allocation of costs through analysis of variation in data: planning of laboratory studies.

BACKGROUND: When developing a new laboratory test for study of human diseases, it is important to identify and control internal and external sources of variation that affect test results. It is also imperative that the precision of the test not only meets pre-established requirements and not exceed allowable total error, but also that these objectives are reached without undue expenditure of either time or financial resources. METHODS: This study applies statistical principles in designing a cost-effective experimental approach for determining the analytical precision of a new test. This approach applies the statistical concept of variance components to the problem of balancing a pre-established level of analytical precision against expenses incurred in achieving this precision. RESULTS: We demonstrated (1) estimation of variance components, (2) use of these estimates for improving allocation of costs within the experiment, and (3) use of these estimates for determining the optimal number of replicate measurements. CONCLUSIONS: Although elimination of all sources of variation that can affect laboratory test results is unlikely, the application of analysis of variance (ANOVA) statistical techniques can lead to a cost-effective allocation of resources for estimating the precision of a laboratory test.

Second-order schedules of token reinforcement with pigeons: effects of fixed- and variable-ratio exchange schedules.

Pigeons' key pecks produced food under second-order schedules of token reinforcement, with light-emitting diodes serving as token reinforcers. In Experiment 1, tokens were earned according to a fixed-ratio 50 schedule and were exchanged for food according to either fixed-ratio or variable-ratio exchange schedules, with schedule type varied across conditions. In Experiment 2, schedule type was varied within sessions using a multiple schedule. In one component, tokens were earned according to a fixed-ratio 50 schedule and exchanged according to a variable-ratio schedule. In the other component, tokens were earned according to a variable-ratio 50 schedule and exchanged according to a fixed-ratio schedule. In both experiments, the number of responses per exchange was varied parametrically across conditions, ranging from 50 to 400 responses. Response rates decreased systematically with increases in the fixed-ratio exchange schedules, but were much less affected by changes in the variable-ratio exchange schedules. Response rates were consistently higher under variable-ratio exchange schedules than tinder comparable fixed-ratio exchange schedules, especially at higher exchange ratios. These response-rate differences were due both to greater pre-ratio pausing and to lower local rates tinder the fixed-ratio exchange schedules. Local response rates increased with proximity to food under the higher fixed-ratio exchange schedules, indicative of discriminative control by the tokens.

Differential response to low-fat diet between low and normal HDL-cholesterol subjects.

Heart attacks frequently occur in normolipidemic subjects with low concentration of high density lipoproteins (35 mg/dL). We hypothesized that as subjects with low HDL-C already have low HDL concentrations, the major decrease of HDL-C will occur in subjects with normal HDL-C when a low-fat diet is consumed. Normolipidemic male subjects consumed three diets differing in total fat and saturated fat composition (AAD: 37%, Step-1: 28%, Step-2: 24% total fat) for 6 weeks in a three-period double-blind randomized crossover design. Plasma lipids and apolipoproteins were determined and changes in distribution of HDL subpopulations were evaluated. As a result of a low-fat diet, low HDL-C individuals slightly decreased their HDL-C, but substantially decreased their LDL-C resulting in a significant improvement in the LDL-C/HDL-C ratio. However, subjects with normal HDL-C levels decreased both their LDL-C and HDL-C resulting in an unchanged LDL-C/HDL-C ratio. We also observed significant differences in response to low-fat diets in HDL-C and alpha(1) concentrations between low and normal HDL-C subjects. In the normal HDL-C group, consumption of a low-fat diet also resulted in redistribution of apoA-I-containing HDL subpopulations, indicated by a decrease in the large apoA-I-only alpha(1) subpopulation. These data demonstrate that male subjects with low HDL-C respond to a low-fat diet differently than individuals with normal HDL-C.

Fasciotomy for exertional anterior compartment syndrome: is lateral compartment release necessary?

The standard surgery for exertional anterior compartment syndrome is fasciotomy of the anterior and lateral compartments of the leg. We prospectively studied the necessity of lateral compartment release, which can add morbidity and extend recovery. We performed 30 anterior compartment releases in 20 patients (10 bilateral operations) with exertional anterior compartment syndrome but not lateral compartment involvement. We alternately performed only an anterior compartment release or both anterior and lateral compartment releases on 10 patients. The 10 patients who underwent bilateral surgery each had, during the same surgery, an anterior and lateral compartment release on one leg and only an anterior compartment release on the other. All were athletes involved in running sports. Overall, satisfactory outcomes were seen in 90% of the limbs--93% (14 of 15) that had anterior compartment release and 87% (13 of 15) that had release of both compartments (no statistically significant difference). For the patients who had unilateral surgery, the average time for full return to sports was 8.1 weeks after anterior release only and 11.4 weeks after release of both compartments, a statistically significant difference. The average time for full return to sports after bilateral surgery was 12.1 weeks. Among these patients, seven said that the leg with only anterior release seemed to recover faster. We concluded that when doing a fasciotomy for exertional anterior compartment syndrome alone, a lateral compartment release is not necessary.

Normolipidemic subjects with low HDL cholesterol levels have altered HDL subpopulations.

Epidemiological studies have established that plasma concentration of HDL is inversely correlated with the risk of coronary heart disease, even in the absence of increased LDL cholesterol levels. We postulate that specific HDL subpopulations may be responsible for antiatherogenic properties of HDL. HDL subpopulations were quantitated by two-dimensional gel electrophoresis in 79 normolipidemic healthy male subjects. To eliminate the influence of diet, volunteers consumed an average American diet for 6 weeks. After the diet period, subjects were stratified according to their HDL cholesterol (HDL-C) levels to low HDL-C < 0.91 mmol/L (< 35 mg/dL), medium > 0.91 < 1.30 mmol/L (> 35 < 50 mg/dL), and high > or = 1.30 mmol/L (> or = 50 mg/dL) groups. Plasma triglycerides and insulin levels were in the normal range, but subjects with low HDL-C levels had higher concentrations of plasma triglycerides and insulin than subjects with medium or high HDL-C concentrations. The absolute concentration (mg/dL) of apoA-I in the largest alpha-migrating HDL subpopulation (alpha 1) was (P < .01) lower in the low HDL-C subjects compared with the medium and high HDL-C groups. The relative concentration (percent distribution) of apoA-I was decreased (P < .01) in alpha 1 and increased (P < .01) in alpha 3 subpopulations. A positive correlation between HDL-C and alpha 1 (P < .001) and a negative correlation between HDL-C and alpha 3 were observed. The inverse correlation of apoA-I distribution (relative concentration) between alpha 1 and alpha 3 suggests an interconversion of alpha 1 and alpha 3 subpopulations, possibly by cholesteryl ester transfer protein. Pre-beta subpopulations showed an inverse trend with HDL-C, while the pre-alpha subpopulation behaved similarly to the alpha-migrating subpopulation. Colocalization of apoA-I and apoA-II particles in the different HDL subpopulations demonstrated that alpha 1, pre-beta 1, and pre-beta 2 subpopulations are apoA-I-only particles rather than apoA-I:A-II particles.

Abnormal lung function in polyurethane foam producers. Weak relationship to toluene diisocyanate exposures.

Exposures to toluene diisocyanate (TDI) were studied for effects on respiratory health of workers in two plants manufacturing polyurethane foams. Intensive personal monitoring was used to characterize job exposures. Of 4,845 12-min personal samples, 9% exceeded 5 ppb and 1% exceeded 20 ppb. Initial questionnaire and spirometry were obtained in 386 workers (88.7% of target population). Current smoking was associated with lower mean FEV1 and FEF25-75, but percent predicted (% pred) means were normal in all smoking categories. Multiple regression showed significant adverse effects of cumulative TDI exposure on initial level of FVC and FEV1 of current smokers, and an effect at borderline significance (p less than 0.063) on FEF25-75 over all smoking categories. Logistic regression showed that chronic bronchitis was more prevalent among those with higher cumulative exposures, after controlling for smoking, age, and sex. Methacholine (MCh) reactivity was associated with reduced airway function, -8.5% pred for FEV1 and -20.0% pred for FEF25-75. In 227 with adequate follow-up, the slopes of annual change were abnormal, for example, FEV1 of -67 ml/yr in current and -53 ml/yr in never smokers. Men had worse FEV1 declines than did women, -71 ml/yr versus -43 ml/yr. TDI exposure, lifetime or concurrent, had no significant effect on slopes, despite its demonstrated effects on initial level of lung function and on prevalence of chronic bronchitis.

Fetal anthropometry at term: effect of menstrual age and relative fetal size.

In a study of 324 term pregnancies, fetal measurements (biparietal diameter, femur length, and abdominal diameter) obtained within one week of delivery correlated more strongly with relative birth weight than with menstrual age.

Biocompatibility testing of an experimental fluoride releasing resin using human gingival epithelial cells in vitro.

Cell culture is a valuable method of evaluating the biocompatibility of new dental materials. The purpose of this study was to compare the in vitro biocompatibility of an experimental fluoride composite resin with fluoride and non-fluoride-releasing materials currently available. The dental materials tested were: MQ Silicate (silicate cement), KETAC-CEM and FUJI (type II glass ionomer cements), VISIO DISPERS (a light-cured, nonfluoridated, microfilled composite resin), and FR-17 (an experimental fluoride-releasing composite resin). The Smulow-Glickman (S-G) human gingival epithelial cell line, which exhibits semidifferentiated characteristics, was used in the study as a test system. Biocompatibility was quantified by counting the viable cells per unit area remaining after 24 and 48 h at two radial distances from cured specimens immersed in the cell culture medium. The test materials were observed to be most toxic to cells nearest the materials. A Time-Distance Cytotoxicity Index (TDCI) was calculated to relate the percentage of dead cells to viable cells at each diffusion distance for each exposure time compared to a nontoxic control. The relative toxicity ranking of the materials tested based on the TDCI was VISIO DISPERS (91%), FUJI (82%), FR-17 (30%), MQ Silicate (23%), and KETAC-CEM (10%), which exhibited the least toxicity. The cytotoxicity of the experimental resin FR-17 was within the range of cytotoxicity of currently accepted restorative materials.

Levels of selected cardiovascular risk factors in a sample of geriatric participants--the Dunedin Program.

A sample of 310 white, ambulatory elderly adults (n = 116 men, n = 194 women, ages 65 to 99 years) from the Dunedin Program in Florida were examined for serum total and lipoprotein cholesterols, weight, and blood pressure. Women had higher age-adjusted mean levels of total cholesterol (p less than .0001), low density lipoprotein cholesterol (p less than .0001), high density lipoprotein cholesterol (p less than .0005), and weight/height (p less than .0001). Systolic blood pressure increased with age in both sexes, but the trend rose more sharply in women (p less than .0001). Both systolic (p less than .05) and diastolic (p less than .025) blood pressure increased with obesity in women but were unrelated to obesity in men. Approximately 70% (218) of the sample had elevated levels of at least one risk factor. This included 12% (38) having elevated levels of two risk factors and 1% (3) with three. The distribution of risk factors for coronary heart disease in this sample indicates a homogeneous population derived through selective survival remains by age 70 years. These results emphasize the need for additional investigations of genetic and environmental variables promoting longevity.

Measurement error and reliability in four pediatric cross-sectional surveys of cardiovascular disease risk factor variables--the Bogalusa Heart Study.

Four cross-sectional epidemiologic studies of children 5-17 years of age were conducted between 1973 and 1982 in the community of Bogalusa, La. Anthropometric, blood pressure and serum lipid variables were measured. Measurement error (precision) and reliability (reproducibility) were estimated for these risk factor variables of coronary artery disease. Precision was less than 0.8 cm for height, 0.8 kg for weight, and 1 mm for triceps and subscapular skinfold over the four surveys. Measurement error for the mercury sphygmomanometer and automatic blood pressure instruments was between 4 and 5 mmHg during this same period. Except for triglycerides and total and alpha-lipoprotein cholesterols that stabilized at lower levels after the first survey, precision of laboratory variables remained nearly constant over the 9 years of survey. Anthropometric measurements were found to be the most reliable followed next by the laboratory and then blood pressure measurements. These estimates offer guidelines applicable to data collection and processing in future investigations of risk factor variables for coronary artery disease.

Transitions of cardiovascular risk from adolescence to young adulthood--the Bogalusa Heart Study: I. Effects of alterations in lifestyle.

Adolescence and young adulthood represents a transition period for biologic and lifestyle characteristics. In a preliminary investigation of young adults (ages 18-20 years), the Bogalusa Heart Study documented patterns of alcohol, tobacco, and oral contraceptive use, as well as changes in education, occupational, marital and parenting status. Such behaviors accelerate the cardiovascular disease process and may differentially influence risk factor patterns of race and sex groups. Adverse levels of systolic blood pressure and alpha-lipoprotein cholesterol were more frequent in married vs single men; elevated triglyceride levels were more frequent in married vs single whites. However adverse levels of beta- and alpha-lipoprotein cholesterol were more frequent in nonparents than in parents. Cigarette smoking and oral contraceptive use were independently related to elevated beta-lipoprotein cholesterol and decreased alpha-lipoprotein cholesterol levels of young white women. Alcohol consumption was highest among white males, with 32% reporting daily consumption of the equivalent of two or more beers or one mixed drink. Alcohol consumption was negatively correlated with blood pressure in white males and positively correlated with alpha-lipoprotein cholesterol in black males. Since such lifestyle factors are related to physiologic risk factors that result in heart disease and adult cardiovascular morbidity and mortality in the older ages, early targeting during adolescence and young adulthood is important.

Subtle electrocardiographic changes in children with high levels of blood pressure.

One hundred thirty-four children, aged 8 to 17 years, were studied to evaluate differences in left ventricular (LV) voltages with mid-range and high blood pressure (BP) levels. The children were selected from 1,604 children after a series of replicate measurements of BP over a 4-month period. Ninety children were selected from those whose BP persisted at the upper 10% and 44 whose BP remained at the middle range (between the fiftieth and sixtieth percentiles). Factors influencing the electrocardiographic voltage (i.e., triceps skinfold, weight, age, race and sex) were also analyzed. Children in the upper 10% of BP levels showed larger LV voltages R1 + SV1 + RV4 than did those in the middle range. This difference was strongly influenced by larger voltages shown by white children. Males showed larger voltages than females using the same index. Triceps skinfold, age and BP account for 6 to 14% of the electrocardiographic variability in the high BP group. Thus, subtle voltage differences in children with early elevations of BP levels do occur. The larger ventricular voltages on the electrocardiogram may represent an early sign of increased LV mass, as has been noted by echocardiography.

Serum lipoprotein lipid and apoprotein (B and A-I) responses to saturated fat-rich diet in cynomolgus monkeys: effects of carbohydrate type and cholesterol.

The effects of carbohydrate type and cholesterol in a high saturated fat diet (40% cal) on serum lipoprotein lipid and apoprotein responses were studied in 12 male cynomolgus monkeys (Macaca fascicularis). For 6 wk each monkey was fed one of four semipurified diets that provided by calorie 49% as starch or sucrose without and with 1 mg/kcal cholesterol. Starch and sucrose diets without added cholesterol resulted in similar levels of serum total cholesterol and apoB; whereas starch produced significantly higher values for these variables than sucrose when cholesterol was added to these diets. Starch diet, irrespective of dietary cholesterol, yielded significantly lower apoA-I levels than sucrose diet. Carbohydrate type did not affect serum triglyceride levels. Cholesterol-enriched starch diet resulted in relatively higher intermediate- and low-density lipoproteins and lower high-density lipoproteins in comparison to similar sucrose diet. Thus, the magnitude of responses of serum lipoproteins can be affected by the type of carbohydrate.

Variability in blood pressure response to dietary sodium intake among African green monkeys (Cercopithecus aethiops).

Since salt-sensitivity to blood pressure (BP) response is a species-related phenomenon, the response of BP to dietary sodium was examined in 14 adult male African green monkeys (Cercopithecus aethiops). The animals were exposed to graded increase of dietary NaCl at 0 (control), 3.0, 4.5, and 6.0%, respectively for 45, 139, 37, and 90 days. Systolic and diastolic BP and body weights were measured twice weekly. Body weight did not differ significantly between successive diets. The mean (+/- SD) systolic/diastolic BP for control, 3.0, 4.5, and 6.0% NaCl diets were 88.8 +/- 15.4/65.9 +/- 13.7, 101.9 +/- 17.6/72.1 +/- 13.0, 90.5 +/- 20.9/62.2 +/- 16.2, and 115.3 +/- 22.6/81.0 +/- 16.4, respectively. BP changes between successive diets were significant (p less than 0.05). Monkeys with high or low initial mean arterial BP levels on control diet tended to maintain the same order with significantly high correlations between BP levels for successive diet periods (r = 0.88 to 0.95, p less than 0.0001). Thus as a group, this nonhuman primate species responds to increased NaCl intake with elevated mean systolic and diastolic BP. Individual variations in salt sensitivity and consistency in persistence of BP levels in this species has bearing on human hypertension.

Influence of persistent obesity in children on cardiovascular risk factors: the Bogalusa Heart Study.

The relationship between body fat indexes, lipid and lipoprotein levels, and blood pressure was studied in 2230 children, each examined during 1973, 1976, and 1978. The children were grouped with the use of seven percentile (P) intervals of triceps skinfold thickness that were specific for race, sex, and age; cardiovascular risk factor variables were assessed over time. Of the 238 children initially in the lowest P group (less than 15P), 44% remained there, and 65% had skinfold thicknesses below 30P on the three occasions. Of the 352 children in the highest P group (greater than or equal to 85P), who were considered to be very obese, 39% remained at this level while 69% remained at 70P or greater. Of the 366 considered to be obese (greater than or equal to 70, less than 85%P), 38% remained at or above 70P. At baseline, children in the seven groups differed in weight, ponderosity (wt/ht3), systolic and diastolic blood pressures, serum triglyceride levels, and pre-beta- and beta-lipoprotein cholesterol levels. Pairwise comparison of data from children in each of the six other groups with those from children in the middle range (greater than or equal to 40, less than 60P) showed that the obese and very obese children had significantly higher systolic blood pressures (p less than .05), while only those in the highest P group had significantly greater diastolic blood pressures (p less than .05). These differences increased and diverged over time. Those in the obese and very obese groups showed a striking drop over time in alpha-lipoprotein cholesterol levels and increases in pre-beta-lipoprotein cholesterol levels and systolic blood pressure. Triglyceride levels decreased over time for the lowest and middle range groups but remained at higher levels in obese and very obese children. There was a strong tendency for tracking (remaining in the same P group over time) in lean, obese, and very obese children. Those who tracked showed definite differences in risk factor variables at the baseline level and over time when compared with the middle range group. Since consistent obesity in early life enhances cardiovascular risk, the measurement over time of skinfold thickness in children is a useful method to detect the potential for adult cardiovascular disease.

Distributions of serum lipoproteins in children by repeated measurements.

Serum lipids and lipoproteins were measured three times over 6 years in a total-community study of children in Bogalusa, Louisiana. Interrelationships and changes in the distributions of serum lipoprotein cholesterol levels were examined in terms of variability and precision of the laboratory measurements. Although distributions of serum lipoprotein cholesterol variables shifted among the three surveys, precision increased markedly as indicated by a decrease in the coefficient of variation for measurement error for total cholesterol, alpha-lipoprotein cholesterol, and triglycerides between Year 1 and Year 4 and then stabilized. Pre-beta-lipoprotein cholesterol was the most difficult variable to measure. The range and magnitude of correlation coefficients between lipid and lipoprotein variables were very similar. Despite quality controls, efforts to standardize laboratory analyses, and adherence to protocols, fluctuations in distributions of lipids and lipoproteins occurred when the same population was restudied. Long-term studies for coronary artery disease prevention should include laboratory safeguards that help to distinguish modest population changes in serum lipids and lipoproteins from laboratory drift.

Studies of arterial wall glycosaminoglycans and collagen during experimental regression of atherosclerotic lesions in cynomolgus monkeys.

The effect of various antiatherogenic regimens on glycosaminoglycan and collagen concentrations in aortas from cynomolgus monkeys with diet-induced atherosclerosis was studied. The drugs and materials that were studied included d-thyroxine, [4-chloro-6-(2,3-xylidino)-2-pyrimidinyl]thioacetic acid, cholestyramine, alfalfa, and glucophage. The treatments resulted in varied degrees of regression of lesions. The mean hydroxyproline concentration in aortas among groups of animals treated with different regimens was significantly different within the groups (p less than 0.001) and correlated with the severity of the lesions (p less than 0.01). The mean total glycosaminoglycan concentration among different groups did not differ significantly but correlated positively (p less than 0.05) with the severity of lesions. Heparan sulfate and hyaluronic acid increased with regression and decreased with increasing severity of lesions, whereas chondroitin sulfates followed an opposite trend. These observations show connective tissue components are intimately involved in remodeling the aorta during regression of diet-induced atherosclerosis.

Longitudinal changes of serum lipid and apoB levels in a newborn-infant cohort.

Longitudinal changes in serum apoB levels in relation to serum lipid levels were determined in a subset (n = 30) of a newborn-infant cohort selected from infants (n = 440) of Bogalusa, Louisiana who were born between January 1, 1974 and June 30, 1975. Serum total cholesterol, triglycerides, beta + pre-beta-lipoprotein cholesterol and apoB levels were measured at birth, 6 mo, 1 yr, 2 yr, and 3 yr of age. The mean +/- SD, mg/dl, levels of these variables increased dramatically (p less than 0.0005) from birth to 6 mo (Total cholesterol: 65.0 +/- 15.7 versus 136.3 +/- 27.4; triglycerides: 32.9 +/- 18.9 versus 82.3 +/- 28.9; beta + pre-beta-lipoprotein cholesterol: 31.8 +/- 9.1 versus 87.3 +/- 25.7; apoB: 20.8 +/- 5.3 versus 49.0 +/- 13.5). On the other hand, the ratio of apoB to beta + pre-beta-lipoprotein cholesterol at birth was significantly higher than at 6 months of age (0.69 +/- 0.24 versus 0.58 +/- 0.12, p less than 0.0005). None of the variables changed significantly between 6 mo and 3 yr of age except the increase in serum total cholesterol level from yr 2 to yr 3 (142.6 +/- 26.6 versus 156.4 +/- 23,0, p less than 0.01). A high intra-child year to year correlation for serum apoB was noted beginning at 6 mo of age (r = 0.67, p less than 0.0001); while the apoB levels were proportional to beta + pre-beta-lipoprotein cholesterol levels in some children, others had varied levels of apoB in relation to cholesterol carried by this apoprotein. The fact that apoB levels persist beginning at 6 mo of age and that we are beginning to observe interindividual differences in levels of cholesterol carried by this apoprotein may have clinical implications for coronary artery disease.