Reliability of a sleep quality questionnaire for use in epidemiologic studies

Background

The longer-term health impacts of poor sleep quality are of increasing interest, as evidence suggests that there are rising levels of sleep disturbance in the community. Studies have reported links between sleep quality and increased morbidity and mortality. However, the results of these studies are constrained by limitations in the measurement of sleep quality in epidemiologic studies. The Breast Cancer Environment and Employment Study (BCEES) has developed a sleep questionnaire that attempts to address some of the limitations of previous sleep questionnaires. The present study assessed the test-retest reliability of the sleep questionnaire used in the Breast Cancer Environment and Employment Study (BCEES).

Methods

Subjects for this reliability study were women who were participating as controls in the BCEES study. Test-retest reliability was evaluated for individual items, using weighted kappa for categorical variables and intraclass correlation coefficients (ICC) and limits of agreement for continuous variables.

Results

Most sleep questions showed good agreement, ranging from 0.78 to 0.45. The ICC was 0.45 (95% CI 0.32–0.59) for lifetime sleep loss per year and 0.60 (95% CI 0.49–0.71) for symptom severity.

Conclusions

The test-retest reliability of the general sleep questions was good, and future epidemiologic studies of sleep could reliably expand the number of assessed domains of sleep quality. However, reliability decreased as increasing detail was required from participants about specific periods of sleep disturbance, and changes to the questionnaire are warranted.Key words: sleep quality, sleep duration, questionnaire, reliability, test-retest, epidemiology     

Reliability and validity of self-reported male balding patterns for use in epidemiologic studies

PURPOSE: To evaluate the validity and reliability of self-reported male balding patterns in an epidemiologic study. METHODS: Participants were 100 men 50-76 years old who were randomly selected from a cohort study of over 77,000 men and women. To assess hair patterning, men selected from among 3 sets of pictures that best described their hair patterning at age 30, at age 45, and at their current age. The categories were little or no hair loss, frontal baldness, and vertex baldness. The self-reporting was done a second time 3 months later. An interviewer independently assessed each man's current hair patterning using the same classification scheme. RESULTS: Test-retest reliability (kappa) of self-reported hair patterning was 0.74, 0.71, and 0.81 for age 30, age 45, and current age, respectively. The kappa for the comparison of the subject's report of current hair patterning to the interviewer's assessment was 0.47. CONCLUSION: Hair patterning represents a noninvasive biomarker that may be an indicator of increased disease risk. While trained observers may represent the gold standard, our study indicates that men are fairly accurate in self-reporting their balding patterns and are quite reliable in reporting their hair patterning at earlier ages.     

A simple, valid step test for estimating maximal oxygen uptake in epidemiologic studies

The authors' modification of the Astrand-Rhyming Cycle Ergometer Test is of short duration, has low initial and peak work rates and was in an earlier study applied for population fitness testing (N = 587) at a survey center after other cardiovascular risk factor measures were obtained in the home. To add fitness testing in the home, the authors have designed a safe, brief 10 inch (25.4 cm) high step test for estimating maximal oxygen uptake (VO2max). Measured maximal oxygen uptake for step tests has been shown to be approximately 10% higher than that reported for cycle tests. All test instructions and stepping rates were included on a cassette tape; heart rates were monitored by a digital tachograph during the last 30 seconds of stepping. Maximal oxygen uptake was measured directly on a bicycle, estimated by the step test, and measured by the authors' bike test in 48 men and women aged 19-70 years who took part in a community fitness program in Pawtucket, Rhode Island in January-February 1983. No significant differences in maximal oxygen uptake were found between the bicycle protocols. The step test estimate of maximal oxygen uptake (VO2max) was significantly higher (12%) than directly measured VO2max, reflecting the expected difference between stepping and cycling. The correlation between direct and both estimates was 0.92. The cross-validation correlation between the estimates was 0.98. The authors' protocol provides accurate estimates of maximal oxygen uptake and is safe and suitable for in-the-home assessment of fitness of people aged 19-70 years for epidemiologic studies.     

The use of dual or multiple reports in epidemiologic studies

Weak measurement of epidemiologic exposures is an impediment to appreciation of the effects of those exposures. This paper discusses two strategies to assess the true effects of weakly measured exposure. The first is to use external information about the extent of mismeasurement to adjust estimates of the effects of exposure. The second strategy is to use multiple measurement--to repeat the measurement in such a way that measurement errors are not repeated. The major disadvantage of the adjustment strategy is its sensitivity to incorrect specification of mismeasurement structure. The primary disadvantage of the multiple measurement strategy is its inefficiency. Unless epidemiologists are quite confident, about the extent and structure of measurement error in their data, they should rely primarily on multiple measurement, and secondarily on adjustment procedures.     

Detection of antimicrobial activity in urine for epidemiologic studies of antibiotic use.

Antibiotic resistance is the inevitable consequence of the selective pressure of antimicrobial drug use and the adaptive plasticity of the microorganisms. Excessive and irrational use of antimicrobial drugs is a problem in all countries. It is particularly troublesome in developing countries where there is a heavy burden of infectious diseases. This study was designed to determine whether detection of antimicrobial activity in the urine might be a useful tool for epidemiologic studies of the interaction between antibiotic use and resistance in developing countries. A laboratory marker is necessary because the history of antimicrobial drug use may be unreliable. Serial specimens or spontaneously voided urine were obtained from healthy volunteers given a single oral dose of commonly used antimicrobial drugs. Urine was also obtained from hospitalized patients the morning after the last dose of an antimicrobial drug and from untreated controls. Assays were performed with standard American Type Culture Collection (Rockville, MD) stains of Bacillus stearothermophilus, Escherichia coli, and Streptococcus pyogenes. Antimicrobial activity could not be detected in pretreatment urine. After a single oral dose, the beta lactam antibiotics and erythromycin could be detected for about 12 to 24 hours, whereas clindamycin, tetracycline, trimethoprim/sulfamethoxazole, and ciprofloxacin could be detected for 48 or more hours. In hospitalized patients, receiving multiple drugs, the following were the sensitivity and specificity for detection of antimicrobial activity: for B. stearothermophilus, 100.0% and 85.9%, respectively; for S. pyogenes, 94.9% and 94.9%, respectively; and for E. coli, 71.8% and 98.7%, respectively. The combination of E. coli and Streptococcus pyogenes exhibited a sensitivity of 97.4% and specificity of 94.9%. Detection of antimicrobial activity in urine is a promising method to determine antimicrobial drug use in epidemiologic studies, particularly in populations in which drug use history is unreliable.     

High-fiber rye bread and insulin secretion and sensitivity in healthy postmenopausal women

BACKGROUND: Fiber and whole-cereal intakes may protect against hyperinsulinemia and the risk of type 2 diabetes. OBJECTIVE: The aim was to study whether the long-term use of high-fiber rye bread and white-wheat bread modifies glucose and insulin metabolism in healthy postmenopausal women. DESIGN: The study was a randomized crossover trial consisting of 8-wk test and 8-wk washout periods. The subjects were 20 postmenopausal women [macro x +/- SD age: 59 +/- 6.0 y; body mass index (in kg/m(2)): 27.5 +/- 2.9; baseline fasting serum cholesterol: 6.5 +/- 0.8 mmol/L], of whom 3 had impaired glucose tolerance as determined by a 2-h oral-glucose-tolerance test. The test breads were high-fiber rye and white-wheat breads, planned to make up > or =20% of energy. Fasting blood samples were collected for the measurement of plasma glucose and insulin at the beginning and at the end of both bread periods. The frequently sampled intravenous-glucose-tolerance test was performed at the run-in and at the end of both bread periods. The acute insulin response, insulin sensitivity, and glucose effectiveness were calculated. RESULTS: The rye bread made up 23.4 +/- 4.3% and wheat bread 26.7 +/- 8.2% of total energy intake. Compared with that during the run-in period, the acute insulin response increased significantly more during the rye bread period (9.9 +/- 24.2%) than during the wheat bread period (2.8 +/- 36.3%; P = 0.047). Other measured variables did not change significantly during the study. CONCLUSIONS: Modification of carbohydrate intake by high-fiber rye bread did not alter insulin sensitivity in postmenopausal, hypercholesterolemic women. High-fiber rye bread appears to enhance insulin secretion, possibly indicating improvement of b cell function.     

Evaluation of measures of urinary albumin excretion in epidemiologic studies

Twenty-four-hour urinary albumin excretion (UAE) is considered the gold standard for determining albumin level in epidemiologic studies, but this measure is inconvenient and often unavailable. Simpler alternatives include the albumin:creatinine ratio (ACR) and urinary albumin concentration (UAC) obtained from a single sample. The authors assessed the strengths and weaknesses of ACR and UAC as alternatives to UAE using albumin measurements from two 24-hour urine samples collected in 1996-1999 from 4,678 participants aged 40-59 years in the International Study of Macronutrients and Blood Pressure (17 population samples from four countries). The authors compared ACR and UAC with regard to correlations with UAE, daily within-person variability, and associations with known predictors of UAE. Rank-order correlations of ACR with UAE were 0.949 and 0.942 for men and women, respectively, versus 0.881 and 0.816 for UAC. Mean within-person coefficients of variation were 34.0-40.0% for the three measures, with the smallest values being observed for UAC. Average correlations with blood pressure were similar for UAE, ACR, and UAC, but the correlation with body mass index was lower for ACR (0.118 for ACR and 0.188 for UAC vs. 0.211 for UAE) because of high correlation between body mass index and creatinine level. Thus, UAC and ACR are acceptable alternatives to the more complex UAE, and the simpler UAC may be preferable to ACR in some respects.     

Effects of weight loss on insulin secretion and in vivo insulin sensitivity in obese diabetic and non-diabetic subjects

To investigate the effects of caloric restriction and weight loss program on insulin sensitivity, acute insulin secretion and glucose effectiveness in 2 obese groups with different beta-cell function, we performed a longitudinal clinical intervention study with a 60 week weight loss program (20-25 kcal/kg ideal weight/day) in 44 obese subjects, 20 with Type 2 diabetes (OD) and 24 non-diabetic (OND). Body mass index (BMI) and metabolic parameters were determined at baseline and every 15 weeks during the intervention study. Insulin sensitivity index (Si) [10(-4) min(-1) x (mU/ml)(-1)], acute insulin response to glucose (AIRg) [pm] and glucose effectiveness (Sg) [x10(-2)/min(-1)] were assessed using the Bergman's minimal model (MINMOD). Thirty-eight subjects finished the study. BMI fell significantly in both groups: 36.2 +/- 3.7 kg/m2 to 33.5 +/- 3.4 kg/m2 (OD) and 37.6 +/- 4.2 kg/m2 to 33.3 +/- 2.9 kg/m2 (OND) (p < 0.05). Insulin sensitivity improved from 1.20 +/- 0.83 to 1.82 +/- 0.75 in the OD group (p < 0.05) and from 1.49 +/- 0.79 to 2.47 +/- 1.13 in the OND group (p < 0.001). The pancreatic beta-cell response to glucose after diet intervention was different, in OD there was a slight increase in the AIRg: 261.3 +/- 70.2 to 288.4 +/- 89.1 (p > 0.05), while a marked reduction was observed in OND subjects 1062.1 +/- 367.3 to 673.5 +/- 236.1 (p < 0.01). Glucose effectiveness did not change significantly in both groups. Weight loss significantly improved insulin sensitivity in both groups, however, the AIRg did not change in the OD group, which could represent an exhausted pancreatic beta-cell, while in the OND subjects, with initially exaggerated AIRg, weight reduction resulted in a parallel reduction in the insulin response. It is possible that this insulin release reduction by pancreatic beta-cells could theoretically avoid or delay pancreatic beta-cell exhaustion. Further studies for longer periods of time and with a larger number of subjects are needed to confirm it.     

Association of serum phosphate levels with glucose tolerance, insulin sensitivity and insulin secretion in non-diabetic subjects

BACKGROUND: Hypophosphatemia is associated with impaired glucose tolerance and insulin resistance in primary hyperparathyroidism. However, little is known about the association between serum phosphate and glucose metabolism in healthy subjects. METHODS: We measured fasting serum phosphate levels (SP, normal range 2.6-4.5 mg/dl) and serum calcium (S-Ca, normal range 2.1-2.6 mmol/l) in 881 non-diabetic subjects (341 male/540 female, age: 38+/-1 years, body mass index 25.9+/-0.2 kg/m(2) (mean+/-standard error of the mean). An oral glucose tolerance test (OGTT) with determination of glucose and insulin every 30 min was performed in all subjects. Insulin secretion and insulin sensitivity (IS) were estimated from the OGTT using validated indices. Furthermore, we tested whether serum phosphate predicts glucose tolerance in 115 subjects during a lifestyle intervention program (LIP). RESULTS: Serum phosphate was negatively correlated with 2-h blood glucose levels independent of age, gender and percent body fat (r=-0.13, P<0.0001). This association remained significant after additional adjustment for S-Ca, creatinine and parathyroid hormone. Serum phosphate was positively correlated with IS (r=0.10, P=0.0006), but not with insulin secretion. This was independent of age, gender, percent body fat, S-Ca and serum creatinine. In the subjects taking part in the LIP low serum phosphate levels at baseline were associated with higher postprandial glucose levels. CONCLUSIONS: In non-diabetic subjects, low serum phosphate levels are associated with high 2-h blood glucose levels and reduced IS. Whether low serum phosphate levels are a cause or a consequence of low IS and impairment of glucose tolerance needs to be tested in further studies.     

The ingestion of saturated fatty acid triacylglycerols acutely affects insulin secretion and insulin sensitivity in human subjects

To assess the effects of acute dietary saturated fat intake on glucose-induced insulin secretion rate (ISR), measured by the C-peptide deconvolution method, and on insulin clearance and sensitivity, five obese and five normal-weight women (controls) were studied after either a 100 g oral butter load or a 100 ml water load. At 120 min after the oral load a hyperglycaemic clamp was performed over 180 min. A dramatic increase of ISR occurred after butter compared with the water challenge in the controls (1305.6 (SE 124.1) v. 616.1 (SE 52.5) pmol/min; P<0.01) and to a lesser degree in the obese subjects (1975.0 (SE 44.1) v. 1417.5 (se 56.0) pmol/min; P<0.05). Insulin sensitivity was impaired after butter (0.60 x 10(-2) (SE 0.11 x 10(-2)) v. 2.26 x 10(-2) (SE 0.32 x 10(-2)) ml/min per kg FFM per (pmol/l); P<0.01) in the controls but not in the obese group. Insulin clearance during the clamp was reduced after butter compared with after the water load only in the controls (0.89 (SE 0.22) v. 1.70 (SE 0.15) litres/min; P<0.01). The data are consistent with the hypothesis that acute excess lipid availability may lead to a compensatory elevation in glucose-induced insulin secretion as a result of the decline in insulin sensitivity and a reduced insulin clearance.     

The use of attributable fraction in the design and interpretation of epidemiologic studies

Because of the etiologic heterogeneity present in many diseases and the interaction among causal factors in the development of disease, relative risks relating any one exposure to a disease may be low, especially in the presence of common exposures. Nevertheless, in the design of epidemiologic studies, arbitrary values of relative risks are often used to determine the sample size required to detect an association between a particular exposure and a disease outcome. Such an approach may not yield adequate statistical power to detect an association. In this commentary, the authors point out the value of using the attributable fraction to determine an appropriate value of relative risk to use for sample size calculations. The approach is particularly useful in cluster investigations where the magnitude of the expected attributable fraction can be readily estimated from the observed and expected rates of disease. Specification of an attributable fraction is also useful in the design of case-control studies of etiologically heterogeneous diseases, especially when common exposures are suspected. Finally, the relationship among attributable fraction, relative risk and exposure frequency is valuable in interpreting results of an epidemiologic study and gaining insight into the differences in relative risk estimates found in various studies.     

Evaluation of insulin sensitivity in clinical practice and in research settings

Insulin resistance is the core metabolic abnormality in type 2 diabetes. Its high prevalence and its association with dyslipidemia, hypertension, hyperinsulinemia, and high coronary and cerebrovascular mortality put it in the forefront as the plausible target for aggressive intervention. Measurements of insulin sensitivity provide clinicians and clinical researchers with invaluable instruments to objectively evaluate the efficiency of both current and potentially useful interventional tools. Although several methods had been developed and validated to evaluate insulin sensitivity, none of these methods can be universally used in all patients. Nonetheless, a method suitable for use in clinical or basic research may not necessarily be a practical method for use in clinical practice or for epidemiologic research. We reviewed the currently used methods for assessment of insulin sensitivity. For each method, we summarized its procedure, normal value, cut-off value for defining insulin resistance, advantages and limitations, validity, accuracy for each patient population, and suitability for use in clinical practice and in research settings. The methods reviewed include fasting plasma insulin, homeostatic model assessment, quantitative insulin sensitivity check index, glucose-to-insulin ratio, continuous infusion of glucose with model assessment, indices based on oral glucose tolerance test, insulin tolerance test, and the so called "gold standard" methods, the hyperinsulinemic euglycemic clamp and the frequently sampled-intravenous glucose tolerance test.     

High isoflavone soy diet increases insulin secretion without decreasing insulin sensitivity in premenopausal nonhuman primates

Consuming soy and soy isoflavones has been shown to cause modest improvements in plasma lipids, lipoproteins, and indices of insulin sensitivity in postmenopausal women. The effect of soy on such end points is attributed often to estrogen receptor agonism by isoflavones. Recent in vitro studies suggest that isoflavones, in combination with high estrogen concentrations (within the range seen circulating in premenopausal women), function as estrogen receptor antagonists that potentially may counteract the beneficial effects seen with soy consumption. We studied insulin sensitivity in 15 premenopausal nonhuman primates consuming either a high isoflavone soy diet or a soy-free casein/lactalbumin diet for 4 months. Insulin sensitivity was measured by intravenous glucose tolerance testing, hyperinsulinemic-euglycemic clamps, and insulin-stimulated insulin receptor and protein kinase B phosphorylation levels in muscle. In addition, plasma lipids, adiponectin, thyroid hormone, and body weights are reported. We show that high isoflavones do not adversely affect insulin sensitivity but do significantly alter insulin secretion to glucose stimulation. Small but significant increases in thyroxine and increased high-density lipoprotein cholesterol were observed as has been reported commonly with soy intake. These study results demonstrate that consumption of soy containing high isoflavone levels is not associated with changes in insulin sensitivity in the high estrogen milieu of the premenopausal female.     

Modeling preclinical cardiovascular risk for use in epidemiologic studies: Miami community health study

To develop a method for assessing preclinical cardiovascular disease risk, models of resting cardiovascular regulation and of insulin metabolic syndrome were derived from information collected from 1991 to 1996 in a culturally heterogeneous sample of 319 healthy men and women (aged 25-44 years) from Miami-Dade County, Florida. The model of resting cardiovascular regulation used 8 noninvasive measures of autonomic and cardiovascular function. Three factors were derived: 1) parasympathetic, 2) inotropy, and 3) systemic vascular resistance. The model of insulin metabolic syndrome used 12 measures assessing body mass, insulin, glucose, and lipid metabolism. Four factors were derived: 1) body mass and fat distribution, 2) glucose level and regulation, 3) insulin level and regulation, and 4) plasma lipid levels. Analyses of the association of the two models revealed that subjects with lower cardiac contractility had greater body mass, higher fasting and postload insulin and glucose levels, and lower insulin sensitivity. Subjects with greater vascular resistance had greater body mass, higher total cholesterol and triglyceride levels, and lower high density lipoprotein cholesterol levels. These findings indicate that preclinical cardiovascular disease risk may involve pathophysiologic processes in which cardiac inotropic and vasodilatory functions are linked to specific aspects of insulin metabolic syndrome.     

The use of an autoregressive model for the analysis of longitudinal data in epidemiologic studies

Korn and Whittemore have presented methods for analyzing longitudinal data where the number of observations per individual is large relative to the number of variables considered for each subject. However, this is often not the case in epidemiologic studies, since one usually collects data at relatively few time points, and the quantity of data collected for each individual at each time point is typically extensive. We present here an autoregressive model for analyzing longitudinal data of this type for the case of a continuous outcome variable. Some of the important features of this model are that one can in the same analysis, consider both independent variables that are time-dependent and those that are fixed over time, partially use data for an individual where some examinations are missing, assess relationships between changes in outcome and exposure over short periods of time, use ordinary multiple regression methods. Anderson has considered this type of model, but, to our knowledge, the model has never been applied to biostatistical problems. We illustrate these methods with data from a longitudinal study that seeks to identify the role of personal cigarette smoking on changes in pulmonary function in children.     

Residual abnormalities of insulin secretion and sensitivity after weight loss by obese women

Changes in plasma insulin and glucose concentrations before and after feeding were measured in six female subjects (post-obese) who had regained a normal body mass after a history of severe obesity (mean weight loss, 37.1 +/- 2.6 kg). The responses of the post-obese group were compared with a group of weight- and age-matched subjects who had not been obese (lean). After an overnight fast subjects were fed a meal at 09.00 h and 13.10 h. Fasting and post-prandial insulin concentrations were lower in post-obese than in lean subjects. Immediately after beginning to eat at 13.10 h all subjects showed a rise in insulin concentration with no change in glucose concentration. In this pre-absorptive period there was no significant difference in insulin concentration between post-obese and lean subjects, although the increment in insulin concentration over baseline values was greater in post-obese subjects (P less than 0.05). It is concluded that abnormalities of insulin secretion and action remain after weight loss by obese subjects. These abnormalities may predispose to hyperphagia and accumulation of excess adiposity.