Sure outcomes of random events: a model for clinical trials.

We consider the outcomes of a clinical trial as determined by one, or several, possibly hidden causes. This paper proposes a statistical model that allows such a distinction of causes not only for the main, or therapeutic, effects but also for the side, or toxic, effects. More specifically, we focus on trials where the effects are naturally dichotomized, that is, where the health of a patient has improved or not, and where a specific adverse effect has occurred or not. A case study provides an example of the way this model can help to solve some problems of suspected drug toxicity. Finally, the model is shown to be a part of a hierarchy of models and the way to select a best model is investigated.

Multiple screening for fetal Down's syndrome with the classic triple test, dimeric inhibin A and ultrasound.

The aim of the study was to determine whether inhibin combined with ultrasound improves the classic second trimester triple test. Eighteen Down's syndrome and 200 euploid pregnancies were included. Maternal serum alpha-fetoprotein (AFP), human chorionic gonadotrophin (hCG), and estriol (E3), measured between 15 and 20 weeks' gestation by radioimmunoassay, were retrieved. Inhibin A was tested with a specific two-monoclonal-antibody ELISA. Ultrasound was performed between 15 and 25 weeks' gestation: femur length, nuchal thickness and morphological abnormalities were assessed. Statistical analysis used a multivariate discriminant analysis. Except for the classic triple test parameters, only inhibin and the standard deviation score (SDS) for femur length allowed us to make a statistically significant distinction between euploid and Down's syndrome pregnancies. For a 5% false-positive rate, the classic triple test sensitivity was 61%. When E3 was replaced with inhibin A, the sensitivity increased to 77.8% and to 66.7% with SDS femur length. Combining hCG, AFP, SDS femur length and inhibin A gave a sensitivity of 66.7% for a false-positive rate of 2.5%. In conclusion, the classic triple test can be more significantly improved by replacing E3 with inhibin than by including ultrasound. However, a combination of ultrasound and inhibin enhances both the sensitivity and specificity of the test. In Down's syndrome screening, the biological parameters improve sensitivity while ultrasound is a better indication of specificity.

Smoking during pregnancy: foetal growth retardation and other risks for the newborn.

In spite of the well-known effect of tobacco on embryo growth retardation, of the higher perinatal mortality of the offspring of smoking mothers, and of the dependence of perinatal mortality risk on small birth weight, it has consistently been found that small infants of smoking mothers have lower mortality rates than small infants of non-smoking mothers. This problem was studied on the perinatal database of a hospital, using adverse outcomes (death or Apgar score <7 at the 10th minute of life) as endpoints rather than perinatal or foetal mortality. A stochastic model constructed to account for cause-effect relations demonstrated that tobacco influences weight and mortality by independent pathways. Furthermore, this model shed some light on the non-tobacco determinants of small birth weight and neonatal morbidity. The method undertaken, based on the use of latent variables, had the advantage of analysing the prevalence, consequences and interactions of some risk factors without identifying them.

Prenatal diagnosis of human cytomegalovirus by culture and polymerase chain reaction: 98 pregnancies leading to congenital infection.

Human cytomegalovirus (HCMV) is the most common cause of viral intra-uterine infection. The experience with prenatal diagnosis remains limited and is based on few reports of small numbers of cases. It is thus difficult to compare the accuracy of the different tests because the groups studied were small and heterogeneous. We describe here our experience on a series of 98 pregnancies leading to HCMV congenital infection, among which 71 have been tested by amniotic fluid (AF) sampling followed by culture and/or polymerase chain reaction (PCR). Independently of the delay between AF sampling and the first HCMV IgM positive result, the mean sensitivity of both culture and PCR was around 70 per cent. The best sensitivity (95.5 per cent) was obtained after a delay > or = 6 weeks in late pregnancy (> or = 23 weeks). The present study demonstrated clearly that the delay between AF puncture and the presumed date of seroconversion is more important for sensitivity than the technique used for the diagnosis (PCR or culture). However, even in the best diagnostic conditions, negative results of HCMV culture or PCR in AF cannot formally exclude intra-uterine infection.

Design and evaluation of an in-house HIV-1 (group M and O), SIVmnd and SIVcpz antigen capture assay.

An enzyme-linked immuno-sorbent assay (ELISA) for the detection of human immunodeficiency virus type 1 (HIV-1) and simian immunodeficiency virus (SIVcpz/SIVmnd) antigens was designed using immunoreagents from naturally infected individuals, and compared to the commercially available Vironostika HIV-1 Antigen Microelisa System (Organon Teknika). The in-house assay proved to be specific for HIV-1 isolates belonging to group M (A-H) and group O and for SIVcpz and SIVmnd isolates, but was less sensitive than the Vironostika HIV-1 Antigen Microelisa System, except for SIVmnd. For the strains belonging to HIV-2, SIVmac and SIVagm, the in-house assay could not detect antigen to an appreciable degree. This study shows that a considerably less expensive but sufficiently accurate HIV-1 antigen capture assay can be developed to monitor HIV-1 (group M and O), SIVcpv and SIVmnd antigen in the supernatants of virus cultures.

The predictive value of serum thyroglobulin in the follow-up of differentiated thyroid cancer.

A strict and careful strategy has to be adopted to cure thyroid cancer. Diagnostic iodine-131 whole-body scan (WBS) and serum thyroglobulin (Tg) are important tools to detect thyroid remnants after thyroidectomy and radioiodine therapy. The aim of this retrospective study was to compare the relative sensitivity of WBS and Tg in the detection of thyroid remnants or metastases and to evaluate the predictive value of Tg in the clinical and scintigraphic course of the disease. Ninety-three patients were followed up after total thyroidectomy and the administration 4-6 weeks later of an ablative dose of 100 or 150 mCi 131I. Eighty-five percent of the patients were free of regional or distant metastases. The follow-up scheme included clinical examination of the patient followed by WBS, Tg, thyroid-stimulating hormone and free thyroxine measurements performed 4 weeks after thyroxine withdrawal and the observance of a low-iodine diet for at least 1 week. WBS (+) patients received a 100- or 150-mCi therapeutic dose of 131I. All patients were further followed up in the same way every 6 months until both WBS and Tg became negative, and thereafter at 1-, 2- and 4-year intervals. Six months after the postoperative radioiodine treatment (first visit), the sensitivity of WBS and Tg was 87% and 26% respectively. Among patients who were WBS(+) at the first visit, 95% of those who were Tg(-) and 47% of those who were Tg(+) had become disease-free at a median of 4 years after surgery (chi2=13.6; P<0.05). Patients whose tests were both positive required more radioiodine to be cured (335+/-90 vs 250+/-95 mCi; P<0.05). Our data indicate that in early diagnosed thyroid cancer, serum Tg measured 6 months after the postoperative 131I ablative dose is less sensitive than WBS for the demonstration of persistence of residual thyroid tissue but provides predictive information on the disease course. WBS(+) and Tg(-) patients are cured earlier and with less radioiodine than those who remain Tg(+).

Marriage: distance between birthplaces and age difference between mates.

The methodology of potential mate analysis can be used to obtain information about the determinants of marriage. This study investigates the influence of the distance between birthplaces and of age difference between mates. We use a new method of potential mate analysis that involves a sample of potential mates simulated from a sequence of actual mates. The differences between actual and potential mates are investigated using a logistic model and by direct comparison of distributions. The logistic model gives a poor fit. The comparison of distributions reveals no significant difference in distance between birthplaces for actual and potential mates or, at any rate, no difference large enough to be demonstrated by our small sample. A marked difference is observed for marriages in which the male is slightly older. The distance between birthplaces in our sample is shorter than in Great Britain.

The difference in mean costs as a pharmacoeconomic outcome variable: power considerations.

The difference in mean costs between a placebo and a drug treated group is used for the pharmacoeconomic evaluation of a new drug. The distribution of costs without treatment (null hypothesis) is determined through hospital records. The sampling properties are established by simulation. The results of the simulation allow determination of a break point of differences corresponding to a given level of significance. For power determination, using this break point, clinical hypotheses are used to generate the distribution of costs under treatment (alternative hypothesis).

Assessing asthma management from interviews of patients and family physicians.

Directed self-care is recommended in asthma. Adequate patient education and follow-up are nevertheless necessary to optimize outcomes. We compared the agreement between detailed information on asthma history and management, collected from the patient and the family physician, to validate the files of physicians and to assess patients' knowledge, attitude, and behavior concerning asthma. A sample of 54 asthma patients were interviewed in detail about use of medications and self-care practice; 36 family physicians (FPs) were interviewed concerning asthma therapy, history, and attitudes of the same patients. Forty-eight percent of the patients expressed negative attitudes toward inhaled corticosteroids, for reasons of safety or lack of efficacy. Less than 20% of the patients made regular use of a peak flow meter. Eighty-three percent of the patients usually obtained prescriptions for asthma therapy from their FP, but on average, only 40% of these prescriptions were provided during visits specific to asthma. FPs were not optimally informed of actual treatments and outcomes and had poor perception of patients' attitudes toward treatment. Nonetheless, in about 30% of the patients, FPs identified risk factors for adverse outcome, such as depression and family conflicts. A majority of interviewed patients had a negative perception of anti-inflammatory therapy, specifically relating to issues of safety and efficacy. Peak flow meters were seldom used and therapy was commonly prescribed outside visits specific to asthma. Despite being centrally involved in the care of asthma patients, FP did not optimally assess therapy and outcomes. The findings suggest suboptimal education and health status in this asthma population.

Phenytoin hydroxylation in a healthy Caucasian population: bimodal distribution of hydroxyphenytoin urinary excretion.

The phenytoin hydroxylation index was studied in 122 unrelated Caucasian volunteers. Following a 100 mg oral dose of phenytoin, phenytoin and hydroxyphenytoin were measured in urine from 0-32 hr after administration. As phenytoin was not found in all urine collections, a phenytoin hydroxylation index was expressed as follows: Phenytoin hydroxylation index = amount of phenytoin administered/0-32 hr urinary output of hydroxyphenytoin. Phenytoin hydroxylation index values appear to be bimodally distributed, 92% of the population showing a mean (+/- S.E.M.) value of 0.639 +/- 0.099 and 8% a mean (+/- S.E.M.) value of 1.001 +/- 0.180 (log10 values). These results are in favour of the existence of a phenytoin genetic polymorphism. Since misuse of urinary metabolite excretion data in drug metabolism studies is a well-known phenomenon, our data emphasize the need for future population studies on phenytoin pharmacokinetics as well as on CYP2C9 genotyping before concluding about existence of a phenytoin genetic polymorphism.

Practical considerations on the use of the Charlson comorbidity index with administrative data bases.

To develop a measure of the burden of comorbid disease from the MED-ECHO data base (Québec), the so-called Charlson index was adapted to International Classification of Disease (ICD-9) codes. The resulting comorbidity index was applied to the study of inpatient death in 33,940 patients with ischemic heart disease. Multiple logistic regression was used to relate inpatient death to its predictors, including gender, principal diagnosis, age, and the comorbidity index. Various transformations of the comorbidity score were performed, and their effect on the predictive accuracy was assessed. The comorbidity index was constantly and strongly associated with death. From a statistical viewpoint, the best results were obtained when the index was transformed into four dummy independent variables (the area under the receiver-operating curve is then 0.87). In a validation analysis performed on 1990-1991 MED-ECHO data (36,012 admissions with ischemic heart disease), the comorbidity index has the same statistical properties. We conclude that the Charlson index may be an efficient approach to risk adjustment from administrative data bases, although it should be tested on other conditions.

[Dr. John Arbuthnot, inventor of statistical testing]. / Le docteur John Arbuthnot, inventeur des tests statistiques.

John Arbuthnot, Queen Anne's medic was very fond of probabilities and statistics expectations. Showing a great interest to new born's sex forecast, he devised a new way: the statistics. Then, he ruled out the case according to uncertain fluctuations connected with an average of 50% male births the excess of male births during 82 years in London. In that mind, he forged ahead to demonstrate Providence deed.

Modelling mortality and morbidity of newborns.

Data of 10,514 singleton births collected over a ten year period at a single hospital are analysed and models linking birthweight and gestational age with mortality and morbidity defined by low Apgar scores are constructed and compared. Based on these models, charts of mortality and morbidity are drawn and compared with common charts of birthweight centiles. Classification rules for newborns at risk are defined by iso-mortality contours, marginal birthweight centiles, and birthweight centiles adjusted by gestational age, respectively, and compared using receiver operating characteristic (ROC) curves. The results suggest that, as far as neonatal mortality is concerned, classification rules based on simple marginal birthweight centiles perform almost as well as iso-mortality contours and considerably better than birthweight centiles adjusted for gestational age.

[Effect of local medications on intraocular pressure control following trabeculectomy]. / Influence des médications locales sur le contrôle de la pression intra-oculaire après trabéculectomie.

Longterm administration of topical antiglaucoma therapies had been indicated to be a serious risk for failure of trabeculectomy through inducing a subclinical inflammation of conjunctival tissues and stimulating secondary fibroblastic proliferation. Our study had 3 main aims: (1) to check if this pejorative effect is present in our population of glaucomatous patients, (2) in order to improve the rate of surgical success in theoretically high risk patients (duration of topical treatment longer than 36 months), to determine the usefulness of the following actions: suppression of beta blocking agents 8 days before surgery, suppression of miotics and eventual topical administration of fluorometholone one month before surgery. (3) to correlate the observed clinical results to the cellular profile of preoperative conjunctival biopsies. We presented in a first step the results of a retrospective analysis including 88 trabeculectomies performed between January '93 and April '95 in 62 GCAO patients older than 40 years and without any previous ocular surgery. The mean age of patients was 66,8 years (41 to 87 years). Patients were classified into "complete success", "qualified success" and "failures" according to the "target" IOP was or was not reached in the last examination. Mean follow up was 5,8 months (1 to 26 months). The second prospective part of our study included 20 trabeculectomies performed in 16 patients we chose according to the nature, the duration of the preoperative topical therapy and the preoperative eventual preparation of the conjunctiva. This study also included a semiquantitative count of the different conjunctival cellular colonies in peroperative biopsies. Mean follow up of this study was 2,3 months (1 to 5 months). 31 "complete success", 52 "qualified success" and 5 "failure" were observed in the retrospective study. The reduction of IOP following trabeculectomy was statistically significant at all examinations (mean IOP in the last examination: 14,9 +/- 4,3 mmHg) but the level of success was not significantly different according to the nature, (mono against polytherapy), the duration of preoperative medical topical treatment, as well as the preoperative actions in order to improve the operative success in high risk patients. Preliminary results of our second prospective study did not yet show any failure and did not exhibit any clear correlation between the counts of the different cellular colonies and the level of operative success on one hand and the different topical therapies in the other hand. The results and limitations of this study were compared to those of the literature.

Comparison of thermostable macromolecular antigens from leprosy-associated bacteria.

Three types of bacteria are associated with leprosy: Mycobacterium leprae, leprosy-derived corynebacteria (LDC), and armadillo-derived mycobacteria (ADM). The immunological relationships between these three types of bacteria and Mycobacterium bovis BCG, used as a reference, were determined by cross-immunoelectrophoresis. When compared with the reference, cross-reactions were observed with a variable number of antigens: 2 in the case of strain LDC 15, 4 with M. leprae, and from 1 to 10 in the case of the ADM, depending on their subgroup. Next, thermostable macromolecular antigens (TMAs), the major cross-reactive antigens of leprosy-associated bacteria, were compared by anti-TMA antibody ELISA tests. The LDC TMAs displayed high cross-reactivity between the subgroups and lower cross-reactivity with the TMAs of M. bovis BCG. Evidence for the presence of a species-specific moiety in TMA of the different LDC was obtained by using depleted anti-TMA antisera. Western blot analysis revealed the presence of many proteins in the TMAs of LDC and M. bovis BCG, some of them being species-specific and other cross-reactive.

The probability of a successful treatment of infertility by in-vitro fertilization.

The objective of this study was to define several estimates of the success rate of in-vitro fertilization and embryo transfer treatment in relation to treatment outcome. The data were collected in a retrospective study of 591 patients treated during 3 years in one Belgian university hospital. Several models used for evaluating the treatment of infertility were fitted to the data (logistic model, binomial model, double binomial model, exponential model, the Speirs model. The number of oocytes retrieved and the proportion of fertilized oocytes at the first attempt can be used as prognostic factors. According to these factors, the success rate after five attempts varies from 28 to 88%.

Molecular phylogeny of part of the env gene of HIV-1 strains isolated in Côte d'Ivoire.

OBJECTIVES: To examine the genetic variation of HIV-1 isolates in Abidjan, Côte d'Ivoire, and to determine the extent to which phylogenetic trees based on sequence information of part of the env gene containing the principal neutralizing domain are representative for documenting genetic variability. DESIGN: Phylogenetic comparison of 13 HIV-1 strains isolated from patients in Abidjan with previously documented HIV-1 strains of different geographic origin. METHODS: To sequence a 900 base-pair fragment of the env gene containing V3, V4, V5 and the beginning of gp41 of three to four clones per isolate. Phylogenetic tree analysis was performed with the software package TREECON. RESULTS: Eleven HIV-1 isolates of Abidjan were classified as genotype A, while two were classed as genotypes B and D. Intra-genotype A distances at the nucleotide level were a maximum of 14.1%. Inter-genotype distances between genotype A and genotypes B, C, and D varied from 16.0 to 22.6%. Phylogenetic trees, based on sequence data of a 300 base-pair fragment containing the V3 loop, showed significant differences in tree topology and statistical confidence with phylogenetic trees based on sequence data of the 900 base-pair env fragment. CONCLUSIONS: Genotype A Côte d'Ivoire HIV-1 strains, which comprise 11 out of 13 isolates, predominate in Abidjan, which may indicate a local burst of particular variants. Phylogenetic trees should be interpreted with caution when based on a more limited number of nucleotides, such as the V3 region.

A regression model for the rate of success of in vitro fertilization.

The treatment of infertility by in vitro fertilization involves several aspects of human reproductive physiology. Since failures are still numerous and since improved techniques are becoming increasingly available, it is important to develop a model where the components of success are identified, allowing estimation of the occurrence of these components of success in the infertile population. The model of Speirs reduces these components to two: uterine receptivity and embryo viability. The probabilities of uterine receptivity and embryo viability were estimated here in a retrospective study of patients in a Belgian hospital sample. For comparison, Australian and U.S. data were used. It was shown that the estimate of uterine receptivity for the Belgian sample is located between the estimates for Australian and U.S. samples. In view of the psychological problem of repeated unsuccessful attempts at in vitro fertilization in some patients, it is important to be able to single out some prognostic factors that would allow more precise estimation of the probability for success in each case. A regression method was developed for that purpose. Only the effect of age, in the Belgian sample, was found to influence uterine receptivity significantly.