Estimating correlations from epidemiological data in the presence of measurement error.

Analysis of a major multi-site epidemiologic study of heart disease has required estimation of the pairwise correlation of several measurements across subpopulations. Because the measurements from each subpopulation were subject to sampling variability, the Pearson product moment estimator of these correlations produces biased estimates. This paper proposes a model that takes into account within and between sub-population variation, provides algorithms for obtaining maximum likelihood estimates of these correlations and discusses several approaches for obtaining interval estimates.

Reliability audit of a regional cardiac surgery registry.

OBJECTIVE: The British Columbia Provincial Cardiac Registry collects demographic and clinical data on all patients who undergo cardiac surgery procedures in the province. The purpose of this study was to compare the reliability of data contained in Registry with data contained in hospital charts. METHODS: Registry and hospital charts were compared for 480 cases. Thirty cases were randomly selected for the province's 16 cardiac surgeons. For each case, 10 distinct fields were selected for analysis and classified as consistent, inconsistent, or rejected (data unavailable in one or other source). RESULTS: The overall rate of consistency between charts and the Registry was 86.4%, with an inconsistency rate of 9.9% and a rejection rate of 3.7%. Consistency rates varied significantly across the 10 fields and among the 16 surgeons. Pairwise comparisons of rates between fields indicated that specific field types were problematic and should be targeted for improvement. In addition, pairwise comparisons of rates between surgeons indicated that further education on Registry use is required. CONCLUSIONS: Recommendations for database design and management include provision of standard definitions for all fields; education of users; extension of the number of mandatory fields; revision of check-off box fields to yes/no/unsure fields; and collection of data close to the time that it is generated.

Mixed Poisson regression models with covariate dependent rates.

This paper studies a class of Poisson mixture models that includes covariates in rates. This model contains Poisson regression and independent Poisson mixtures as special cases. Estimation methods based on the EM and quasi-Newton algorithms, properties of these estimates, a model selection procedure, residual analysis, and goodness-of-fit test are discussed. A Monte Carlo study investigates implementation and model choice issues. This methodology is used to analyze seizure frequency and Ames salmonella assay data.

Development of the pulmonary airways in the fetal rat and its relation to the prenatal environment.

We studied the left lung using multi-focus microphotography in 378 rat fetuses, assessing airway branching from day 13 to day 19 of gestation, and lung growth variables from day 13 to day 21. Longitudinal growth, and monopodial and dichotomous branching brought about a consistent airway pattern with variations within each day of gestation and a small overlap between adjacent days. Amniotic fluid weight and pole to pole (PTP) distance of the lung increased quadratically with age, while fetal weight and the peripheral airway count (PAC) increased exponentially. The location of the fetus within the uterus had no effect on fetal variables, but correlations were found between maternal weight gain and both fetal weight and PTP. Fetal weight was the best predictor of PAC from gestational ages 15 to 19 days (P < 0.008). The method described allows for observations that are reproducible within the environmental variations present in normal gestation and can be used to study the effect of external factors on lung development.

Prophylaxis of urinary tract infection in persons with recent spinal cord injury: a prospective, randomized, double-blind, placebo-controlled study of trimethoprim-sulfamethoxazole.

PURPOSE: To determine the efficacy of trimethoprim-sulfamethoxazole (TMP-SMX) for prophylaxis of urinary tract infection in persons with recent spinal cord injury, during the first 4 months of intermittent catheterization. PATIENTS AND METHODS: One hundred twenty-nine adults (112 men, 17 women) with recent acute spinal cord injury participated in a randomized, double-blind, placebo-controlled trial, and were studied for up to 16 weeks. Low-dose TMP-SMX (TMP 40 mg, SMX 200 mg) or placebo was given once daily. Clinical observations, urine cultures, and cultures of rectal and urethral swabs were made weekly. Subjects who developed breakthrough bacteriuria received conventional antimicrobial therapy and prophylaxis was continued. RESULTS: Sixty-six TMP-SMX recipients (57 men, 9 women) and 60 placebo recipients (52 men, 8 women) are evaluable for efficacy. Among male subjects, bacteriuria was present during 50% or more of study weeks in 30% of TMP-SMX recipients and in 56% of placebo recipients (p = 0.003). The interval to the onset of bacteriuria was prolonged in TMP-SMX recipients (p < 0.0001). TMP-SMX recipients without bacteriuria in any given week had a lower probability of having bacteriuria the subsequent week (0.26) than did placebo recipients (0.49) (p < 0.0001). At least 1 episode of definite symptomatic bacteriuria (bacteriuria and fever and at least 1 classical manifestation of urinary infection) occurred in 4 of 57 TMP-SMX-treated men and in 18 of 52 placebo-treated men (p < 0.0003). We observed similar trends in women, but differences did not reach statistical significance, perhaps because the number of females was small. Adverse events suspected to be due to medications were frequent in this population of patients with recent severe injuries and led to discontinuation of the study in 10% of the TMP-SMX group and 8% of the placebo group. Adverse events included neutropenia (TMP-SMX: two; placebo: three), pseudomembranous colitis (TMP-SMX: one), severe skin rash (TMP-SMX: two; placebo: one), and drug fever (TMP-SMX: one). The proportion of all episodes of bacteriuria that were due to TMP-SMX-resistant organisms was unexpectedly high in the placebo group (51%), and increased progressively according to year of enrollment in the study. By Year 3, all subjects in the placebo group had at least one episode of TMP-SMX-resistant bacteriuria. Gram-negative enteric bacilli resistant to TMP-SMX were recovered from rectal swabs (TMP-SMX 49%, placebo 42%) and urethral swabs (TMP-SMX 33%, placebo 26%) in similar proportions of subjects in the two study groups. CONCLUSIONS: Prophylaxis with TMP-SMX significantly reduces bacteriuria and symptomatic urinary tract infection in persons with recent acute spinal cord injury during bladder retraining using intermittent catheterization. However, adverse reactions attributable to TMP-SMX are common in this population. Colonization and breakthrough bacteriuria with TMP-SMX-resistant organisms are frequent and may seriously limit the usefulness of this strategy, particularly in an institutional setting.

In vitro branching morphogenesis of the fetal rat lung.

We studied the pattern of airways branching in the fetal rat lung in vitro. Lung primordia of gestational ages 13, 14, and 15 days were allowed to grow in culture to a gestational age equivalent to 21 days. The first generation airways appear by a single new bud (monopodial budding) from the left main airway (lateral appearing before the medial). They elongate to form branches and then bud dichotomously (2 buds occurring simultaneously and adjacent to each other) at their tips. Then monopodial branching takes place along their sides. The same cycle of budding and branching seems to be repeated for the following generation of the airways. The total number of the peripheral (subpleural) buds was greatest in the day 15 explants and least in day 13 explants throughout the whole culture period, but the statistical model used indicated faster budding in the 13 day explants. Morphometric assessment showed no difference in the ratios between the lung components in the 3 age groups and that the peripheral epithelial measurements were the same in the 3 groups at an equivalent gestational age of 21 days. We have also shown that lobes do not form in the right lung, although appropriate airways do. This may indicate the importance of mesothelial covering of the lung in the process of lobe formation. The method is useful for studying the control of lung morphogenesis.

Maximum-penalized-likelihood estimation for independent and Markov-dependent mixture models.

This paper concerns the use and implementation of maximum-penalized-likelihood procedures for choosing the number of mixing components and estimating the parameters in independent and Markov-dependent mixture models. Computation of the estimates is achieved via algorithms for the automatic generation of starting values for the EM algorithm. Computation of the information matrix is also discussed. Poisson mixture models are applied to a sequence of counts of movements by a fetal lamb in utero obtained by ultrasound. The resulting estimates are seen to provide plausible mechanisms for the physiological process.

Exact likelihood evaluation in a Markov mixture model for time series of seizure counts.

This paper provides an alternative to Albert's (1991), Biometrics 47, 1371-1381) approximation to the E-step when using the EM algorithm for parameter estimation in Markov mixture models. Use of a recursive algorithm of Baum et al. (1970, Annals of Mathematical Statistics 41, 164-171) results in exact evaluation of the likelihood, optimal parameter estimates, and very efficient computation. Applications to time series of seizure counts and fetal movements clearly show the advantages of this exact approach.

Optimal positioning of endotracheal tubes for ventilation of preterm infants.

Accurate knowledge of upper-airway dimensions is required to prevent malpositioning of endotracheal tubes in preterm infants. We measured vocal cord-carina, oral-carina, and nasal-carina distances in situ at autopsy of two groups of infants (less than 1000 and greater than or equal to 1000 g). In all 24 infants, crown-heel length, crown-rump length, and occipitofrontal circumference were better than weight in predicting upper-airway dimensions. Flexion of the neck decreased and extension increased both nasal-carina and oral-carina distances. Lateral rotation produced no significant changes. The postmortem data were not different from nasal-carina distances measured radiologically in 40 living, nasally intubated and ventilated infants, confirming the clinical validity of our findings. Regression equations were derived to predict optimal endotracheal tube lengths based on the external measurements of crown-rump length and crown-heel length.

Optimal choice of prognostic variables with an application to cardiac monitoring using M-mode echocardiography.

This paper provides a methodology for the optimal choice of a subset from a large number of interrelated diagnostic variables. We use predetermined abnormal ranges for each measurement and code subjects as abnormal or normal on this basis. We present a procedure to determine the smallest subset of measurements that identifies any subject abnormal on at least one measurement. We formulate and solve the problem using integer programming. We then apply this methodology to study the use of M-mode echocardiography to determine potentially cardiotoxic side effects of chemotherapy and compare its performance to several multivariate methods. Extensions and modifications are discussed.

Neonatal outcome after prolonged preterm rupture of the membranes.

Pulmonary hypoplasia was diagnosed in 14 of 88 infants (16%) delivered in 1983 to 1986 after rupture of the membranes of greater than or equal to 7 days with onset before 29 weeks' gestation. Logistic regression analysis applied to examine the relative importance of perinatal risk factors in the prediction of pulmonary hypoplasia showed that gestational age at onset of rupture of the membranes had a significant effect (p = 0.002) on the odds that pulmonary hypoplasia developed in the neonate, whereas the duration of rupture of the membranes (p = 0.11) and the degree of oligohydramnios (p = 0.65) did not. Postnatally, the presence of pulmonary hypoplasia was associated with the severity of skeletal compression deformities (p less than 0.0001). The development of skeletal compression deformities was associated with severe oligohydramnios (p = 0.05) and duration of rupture of the membranes (p = 0.02) but not gestational age at rupture of the membranes (p = 0.77). Gestational age at onset of rupture of the membranes being the best single predictor of pulmonary hypoplasia suggests that the stage of lung development at rupture of the membranes may be important.

Kinetics of isotype-specific humoral immunity in rubella vaccine-associated arthropathy.

The present study documents the relationship between the development of rubella vaccine-associated arthropathy and isotype-specific rubella antibody responses in a prospectively studied population of 44 adult rubella hemagglutination inhibition (HAI) seronegative females undergoing rubella immunization. Rubella-specific IgM, IgG, and IgA antibody responses were evaluated prior to and at 1, 2, 3, 4, 5, 6, 12, and 24 weeks postvaccine. Detectable preimmunization rubella antibody of the IgG or IgA class was present using ELISA techniques in 6 of 6 individuals developing acute arthritis, 13 of 17 developing acute arthralgia, and in 15 of 21 with no joint manifestations postvaccine. Significantly elevated HAI IgM responses were noted 3 and 4 weeks postvaccine in the acute arthritis group but no significant differences were found in IgG and IgA rubella antibody levels postvaccine in relation to the presence or absence of joint manifestations at any time period postvaccine. The data support rubella reinfection as an important feature of rubella vaccine-associated arthropathy but do not support a role for quantitative differences in rubella IgG and IgA antibody in the pathogenesis of this syndrome.

Effect of linoleic acid-rich infant formula feeding on brain synaptosomal lipid accretion and enzyme thermotropic behavior in the piglet.

The effects of a vegetable oil-based infant formula, virtually devoid of n-6 and n-3 long chain polyenoid fatty acids (LCP) and high in 18:2(n-6) and 18:2(n-6)/18:3(n-3) ratio, on brain synaptosome lipid composition and enzyme thermotropic behavior were studied in neonatal piglets. Term gestation piglets were fed either sow milk (SMF) or formula (FF) from birth for 5, 10, 15, or 25 days. Synaptosomal cholesterol, total lipid phosphorus, and phospholipid class composition did not differ between SMF and FF piglets. Synaptosomal fatty acid composition, however, was influenced by diet. The proportion of n-3 LCP, especially 22:6(n-3), was decreased, while the n-6 LCP, especially 22:4(n-6) and 22:5(n-6), were increased in FF compared to SMF piglets. These diet-related changes were most pronounced in the ethanolamine glycerophospholipid fraction and increased with the duration of feeding. FF thus reversed an apparent developmental increase in the synaptosomal n-3/n-6 LCP ratio. The monoene content, especially 18:1, was also reduced in the synaptosomes of FF compared to SMF pigs. FF had no effect on the activity of synaptosomal acetylcholinesterase. However, higher transition temperatures for this enzyme, indicating decreased membrane fluidity, were found in the FF compared to SMF piglets. The data suggest that exclusive feeding of proprietary formulae, devoid of LCP and high in 18:2(n-6) and/or the 18:2 (n-6)/18:3(n-3) ratio, may compromise normal fatty acid accretion and physical properties of brain synaptosomal membranes.

Pyuria: its relationship to bacteriuria in spinal cord injured patients on intermittent catheterization.

Observations made during a study of intermittently catheterized spinal cord injured patients suggested that leukocyte counts yield higher results in aliquots of terminal-catheter urine (TCU) than in midstream-catheter urine (MCU) or suprapubic aspirate (SPA). The purpose of this study was to confirm that observation, to examine the relationship of leukocyte counts in TCU, MCU, and SPA to the leukocyte excretion rate (LER), and of pyuria to bacteriuria in this population. We collected sets of urine specimens obtained by SPA and intermittent catheterization (for leukocyte counts and quantitative culture) and timed urine collections (for LER determination). Fifty-two patients were studied for an average of five days. Leukocyte counts were performed in 241 SPA, 250 MCU, and 236 TCU specimens, and LER in 131 timed collections. The mean of the logarithm of leukocyte counts differed significantly between TCU and both MCU and SPA (p less than .0001). The difference between TCU and MCU was greater than 150 leukocytes/mm3 for 25% of paired specimens (mean 624 leukocytes/mm3, median 15 leukocytes/mm3). The statistical correlation between LER and leukocyte counts in all catheter specimens was significant; however, SPA and MCU frequently underestimated LER and TCU overestimated LER. Estimates of pyuria do not clearly separate bacteriuric from abacteriuric specimens. In spinal cord injured patients on intermittent catheterization, aliquots of catheter urine are not suitable for estimation of pyuria, and estimation of pyuria is not a feasible screening test for bacteriuria.

Prognostic factors in diffuse proliferative lupus glomerulonephritis.

A number of clinical laboratory and biopsy-derived parameters were assessed for their prognostic significance in the short (24 months), intermediate (60 months) and long terms in 45 patients (43 female, 2 male) with diffuse proliferative lupus glomerulonephritis (DPGN). The factors evaluated were serum creatinine (SCr) and urinary protein at time of biopsy, initial dose of prednisone and immunosuppressive after biopsy, activity index (AI), chronicity index (CI), their individual components, extent of extraglomerular (tubulo-interstitial) immune deposits (EGD) and mean number of intraglomerular monocytes per glomerulus (NSE index). Using proportional hazards analysis to evaluate the parameters, SCr (P = 0.003), AI (P = 0.005) and NSE index (P = 0.038) were shown to be significant predictors of outcome when all variables except the components of AI and CI were considered. When AI and CI were omitted but their components included, SCr (P = 0.0005), NSE index (P = 0.024), extent of karyorrhexis (P = 0.035) and glomerulosclerosis (P = 0.033) were then demonstrated to be significant prognostic factors of DPGN. The results suggest that intraglomerular monocyte infiltration has a protective effect and confirm that AI index is a relatively powerful predictor of outcome. Histologic and nonhistologic biopsy factors contribute significant additional prognostic information to that provided by SCr.

Evaluation of seven immunoassays for detection of rotavirus in pediatric stool samples.

The performance of seven commercially manufactured rotavirus assays was evaluated with 144 pediatric stool specimens and compared with electron microscopy (EM) findings. The four enzyme-linked immunosorbent assays used were Rotazyme II, Pathfinder, IDL rotavirus immunoassay, and Enzygnost (Behring) rotavirus assay. The three latex tests were Meritec rotavirus detection test, Virogen Rotatest, and Bartels rotavirus latex test. Test outcomes were compared with EM on the basis of sensitivity, specificity, positive-negative predictive value, and the kappa statistic. Relative to EM, Meritec had the highest specificity (97%), followed by Virogen (95%), IDL (91%), Pathfinder (85%), Behring (81%), Bartels (72%), and Rotazyme (71%). The sensitivities were as follows: Rotazyme (92%), Pathfinder (89%), Bartels (86%), Virogen (86%), Behring (82%), Meritec (71%), and IDL (75%). Patient age and sex did not influence test results. Owing to the absence of a true standard, the tests were also compared with each other on the basis of the kappa statistic, the frequency of positive test results, and the frequency of samples in which a test differed from all other tests. Using these measures, the assays could be classified into three groups with progressively decreasing utility: group 1 (Virogen, Meritec, IDL, and EM), group 2 (Pathfinder and Behring), and group 3 (Rotazyme and Bartels). Laboratory criteria were also compared. Latex tests were faster and required less equipment than enzyme-linked immunosorbent assays. The Virogen latex assay showed the best overall performance, which made it our choice for rapid and accurate rotavirus diagnosis. However, in children who have gastrointestinal symptoms with negative rotavirus test results, EM will be useful until such time as immunological tests for other enteric viruses are available.

Nonlinear estimation of parameters in biphasic Arrhenius plots.

This paper presents a formal procedure for the statistical analysis of data on the thermotropic behavior of membrane-bound enzymes generated using the Arrhenius equation and compares the analysis to several alternatives. Data is modeled by a bent hyperbola. Nonlinear regression is used to obtain estimates and standard errors of the intersection of line segments, defined as the transition temperature, and slopes, defined as energies of activation of the enzyme reaction. The methodology allows formal tests of the adequacy of a biphasic model rather than either a single straight line or a curvilinear model. Examples on data concerning the thermotropic behavior of pig brain synaptosomal acetylcholinesterase are given. The data support the biphasic temperature dependence of this enzyme. The methodology represents a formal procedure for statistical validation of any biphasic data and allows for calculation of all line parameters with estimates of precision.

Age-related changes in humoral and cell-mediated immunity in Down syndrome children living at home.

Abnormalities of humoral and cell-mediated immunity have been described in Down syndrome but reported findings have been inconsistent. Confounding factors have included age, institutional versus home life, hepatitis B antigenemia, and zinc deficiency. To clarify this problem, we studied 64 children with Down syndrome (DS) compared with an age-matched control group. All children had always lived at home. All the DS children were negative for hepatitis B surface antigen. Serum zinc concentration in the DS group was on average 12 micrograms/dl lower than age-matched control children. They also had significantly lower levels of immunoglobulin M, total lymphocyte count, T and B lymphocytes, and T helper and suppressor cells. In vitro lymphocyte response to phytohemagglutinin and concanavalin A was significantly reduced at all ages in the DS group. Lymphocyte response to pokeweed mitogen increased with age in control children but decreased in the DS children. By 18 yr, the mean response for DS was 60000 cpm lower than controls. The DS group had significantly higher concentrations of immunoglobulins A and G than controls and the difference increased with age. Complement fractions C3 and C4 were also higher in the DS group at all ages. The number of HNK-1 positive cells was higher in the DS group than controls at all ages. When hepatitis and institutionalization are excluded as confounding factors, DS children still differ in both humoral and cell-mediated immunity from an age-matched control group.

Immunity to soluble retinal antigen in patients with uveitis accompanying juvenile rheumatoid arthritis.

Studies of immunity to bovine soluble retinal antigen (antigen S) were carried out using serum and peripheral blood lymphocytes from children with juvenile rheumatoid arthritis and chronic anterior uveitis (JRA-uveitis), children with JRA alone, children with nonrheumatic diseases, and controls who had no ocular or rheumatic disease. Enzyme-linked immunosorbent assay and the lymphocyte transformation assay were used to determine immunity. Antibody to antigen S was present significantly more frequently in children with JRA-uveitis than in children with JRA alone, children with nonrheumatic disorders, or controls. These latter groups did not differ in positivity for this antibody. Lymphocyte transformation occurred more frequently in children with JRA-uveitis than in children with JRA alone or controls. Children with JRA alone and controls had similar frequencies of lymphocyte transformation positivity. Enzyme-linked immunosorbent assay positivity and lymphocyte transformation positivity tended to occur in different children. Children with JRA-uveitis who had HLA-B35 had the highest frequency of antibody to antigen S. Immunity to antigen S may be the result of ocular damage by mechanisms other than a pathogenic mechanism per se.

Normal serial variability for M-mode and digitized echocardiography in a pediatric population.

The normal serial variability of M-mode and digitized echocardiography was assessed in 18 healthy children aged 3-16 years. M-mode echocardiograms were done one month apart in each child and two sets of three beats were analyzed twice for each month's tracing. Ejection fraction and shortening fraction were obtained in the standard manner and the tracing digitized and analyzed for peak and normalized left ventricular velocities, and for posterior wall and septal wall velocities in systole and diastole. The component variabilities were determined by a univariate four-factor hierarchical variance components analysis and the sum of these variability limits established. Standard M-mode echocardiographic measurements of shortening fraction and ejection fraction (15% and 10% of mean) were the least variable. Peak and normalized velocities were more variable with left ventricular velocities (18%-29% of mean), posterior wall velocities (26%-45% of mean) and septal velocities (38%-68% of mean) showing increasing variabilities. Thus there is less variability in M-mode echocardiography than digitized echocardiography. The normal limits of variability are established and should be used for follow-up studies.