Search

The problem of survival analysis with covariates when observations are subject to irregular interval censoring is considered. A proportional hazards model is assumed. A pseudo likelihood based on the comparison of all pairs of individuals whose survival times can be ordered, is proposed. Statistical properties of the maximum pseudo likelihood estimator and its score test are explored. Several examples are presented, one of which is an examination of the effect of covariates on HIV seroconversion.

Aims: To assess the prevalence of general health status, use of sleep medication, and use of medication for cardiovascular diseases, and to study their relation to aircraft noise exposure. Methods: These health indicators were measured by a cross-sectional survey among 11 812 respondents living within a radius of 25 km around Schiphol airport (Amsterdam). Results: Adjusted odds ratios ranged from 1.02 to 2.34 per 10 dB(A) increase in Lden. The associations were statistically significant for all indicators, except for use of prescribed sleep medication or sedatives and frequent use of this medication. None of the health indicators were associated with aircraft noise exposure during the night, but use of non-prescribed sleep medication or sedatives was associated with aircraft noise exposure during the late evening (OR = 1.72). Vitality related health complaints such as tiredness and headache were associated with aircraft noise, whereas most other physical complaints were not. Odds ratios for the vitality related complaints ranged from 1.16 to 1.47 per 10 dB(A) increase in Lden. A small fraction of the prevalence of poor self rated health (0.13), medication for cardiovascular diseases or increased blood pressure (0.08), and sleep medication or sedatives (0.22) could be attributed to aircraft noise. Although the attributable fraction was highest in the governmentally noise regulated area, aircraft noise had more impact in the non-regulated area, due to the larger population. Conclusions: Results suggest associations between community exposure to aircraft noise and the health indicators poor general health status, use of sleep medication, and use of medication for cardiovascular diseases.

We used disease mapping for health impact assessment of the national airport of the Netherlands. Spatio-temporal models were used to relate hospital discharge data for acute myocardial infarction and bronchitis in 1991, 1992 and 1993 to noise and distance from the airport. To compare models a discrepancy measure (expected predictive deviance) proposed by Carlin and Louis was used. The best fitting model was the most general one with inclusion of spatial and temporal components. Results on the effects of the covariates noise and distance from the airport were somewhat inconsistent between men and women and between the two diseases: for women no association between bronchitis and distance from the airport was found, whereas for men no association between acute myocardial infarction and noise was found.

to investigate knowledge, attitude and behaviour toward infection control in two teaching hospitals on the island of Java by means of a questionnaire and to evaluate the use of the questionnaire as a tool.we investigated knowledge, attitude and behaviour toward infection control in two teaching hospitals on the island of Java by means of a questionnaire to identify problem areas, barriers and facilitators. The target was to include at least 50% of all health care workers (physicians, nurses, assistant nurses and infection control nurses) in each hospital, department and profession. Differences between demographic variables and scores for individual questions and groups of questions were compared using the chi-square statistic and analysis of variance and Spearman's rho was used to test for correlations between knowledge, attitude, self-reported behaviour and perceived obstacles.more than half of the health care workers of the participating departments completed the questionnaire. Of the 1036 respondents (44% nurses, 37% physicians and 19% assistant nurses), 34% were vaccinated against hepatitis B, 77% had experienced needle stick accidents and 93% had been instructed about infection control. The mean of the correct answers to the knowledge questions was 44%; of the answers to the attitude questions 67% were in agreement with the correct attitude; obstacles to compliance with infection control guidelines were perceived in 30% of the questions and the mean self-reported compliance was 63%. Safe handling of sharps, hand hygiene and the use of personal protective equipment were identified as the most important aspects for interventions. Significant positive correlations were found between knowledge, attitude, self-reported behaviour and perceived obstacles.the questionnaire in conjunction with site visits and interviews was a valuable strategy to identify trouble spots in the hospitals and to determine barriers to facilitators of change that should be taken into account when planning interventions. Successful interventions should cover hospital management, the infection control team, as well as the health care workers on the wards.

Acute salpingitis complicating cervical gonococcal infection is a significant cause of infertility. Relatively little data are available concerning the pathophysiologic mechanisms of this disease. A cohort of 243 prostitutes residing in Nairobi were followed between March 1985 and April 1988. Gonococcal cultures were performed at each visit, and acute salpingitis was diagnosed clinically. Serum at enrollment was tested by immunoblot for antibody to gonococcal outer membrane proteins. 8.6% (146/1689) of gonococcal infections were complicated by salpingitis. Increased risk of salpingitis was associated with younger age, shorter duration of prostitution, HIV infection, number of gonococcal infections, and episodes of nongonococcal salpingitis. Rmp antibody increased the risk of salpingitis. Antibody to Opa decreased the risk of salpingitis. By logistic regression analysis, antibody to Opa was independently associated with decreased risk of gonococcal salpingitis (adjusted odds ratio [OR], 0.35; 95% confidence interval [95%CI], 0.17-0.76); HIV infection (adjusted OR, 3.5; 95% CI, 0.96-12.8) and episodes of nongonococcal salpingitis (adjusted OR, 3.4; 95% CI, 1.8-6.4) were independently associated with an increased risk of salpingitis. Antibody to Opa appears to protect against ascending gonococcal infection, perhaps by interfering with Opa mediated adherence and endocytosis. The demonstration of natural immunity that protects against upper genital tract infection in women suggests that a vaccine to prevent gonococcal salpingitis is possible.

We investigated the impact of a short-lived policy of charging fees to patients attending public-sector outpatient health facilities in Kenya by collecting data on attendance at Nairobi's Special Treatment Clinic for sexually transmitted diseases (STDs) before (23 months), during (9 months), and after (15 months) the user-charge period. During the user-charge period, the seasonally adjusted total mean monthly attendance of men decreased significantly to 40% (95% Cl 36-45) of that before fees were levied. Attendance rose in the post-user-charge period, but reached only 64% (59-68) of the pre-user-charge level. For women, the adjusted total mean monthly attendance during the user-charge period was reduced significantly to 65% (55-77) of the pre-user-charge level. Mean monthly attendance by women rose in the post-user-charge period to 22% (9-37) above the pre-user-charge level. There was no evidence of an increase in attendance over the course of the user-charge period among either men or women. The introduction of user fees probably increased the number of untreated STDs in the population, with potentially serious long-term health implications. The user-fee experience in Kenya should be carefully evaluated before similar measures are introduced elsewhere.

SUMMARYWe report the first population-based case-control study on acute hepatitis B in a very low-incidence country. A case was a Netherlands resident, notified between May 1999 and July 2000 with symptoms and serology compatible with acute hepatitis B. Population controls were randomly selected, with oversampling from men and persons aged 20–39 years. Risk factors were studied using logistical regression, distinguishing confounders and mediators through hierarchical analysis. Participants were 120 cases and 3948 controls. The risk of acute hepatitis B was increased in men who have sex with men, with reporting to have had more than two partners in the past 6 months the only significant risk. In children, adult females and heterosexual males, having parents born in a hepatitis B endemic country was a significant risk. For adult females and heterosexual males, this was largely explained by having a foreign partner. For children this was partly explained by parenteral exposures abroad.

1) To assess the regression to normal cytology in women with cervical smears diagnosed as atypical squamous or glandular cells of undetermined significance (ASCUS/AGUS) and absence or clearance of human papillomavirus (HPV) infection; 2) To evaluate the association between viral load, semi-quantitatively evaluated, and cytological or histological outcome.In this cohort study HPV test and biopsy was taken in 148 women with ASCUS/AGUS cytology. After 12-18 months a HPV test and cervical smear were repeated in 121 women.Absence or clearance of HPV showed significantly more regression to normal cytology than persistent or newly acquired infected women, odds ratio 27 (95% confidence interval; 7-103). The viral load of the HPV test at enrollment was not correlated with the follow-up cytological outcome (Spearman correlation coefficient 0.2, p = 0.2). A marked association between viral load and histological outcome at enrollment was shown (Spearman correlation coefficient 0.43, p0.0001).Absence or clearance of HPV can predict regression to normal cytology. Viral load at enrollment cannot predict cytological regression. There was a marked association between viral load and the underlying CIN at enrollment. However, there was large overlapping of viral loads among the grades of CIN. Therefore, viral load is not a useful parameter to predict high-grade lesions in women with ASCUS/AGUS cytology.

The purpose of this research was to model the familial clustering of breast cancer and to provide an accurate risk estimate for individuals from the general population, based on their family history of breast and ovarian cancer. We constructed a genetic model as an extension of a model by Claus et al. (E. B. Claus et al., Am. J. Hum. Genet., 48: 232-242, 1991), with three breast cancer genes, BRCA1, BRCA2, and a hypothetical BRCAu, in two variants, one in which BRCAu was dominant and one in which BRCAu was recessive. The model parameters were estimated using published estimates of population incidence and relative risks. Risk estimation was performed for a set of 196 counselees and for a set of simulated counselees with both the dominant BRCAu and the recessive BRCAu model, and compared relating to medical management. Estimates of the model parameters were found. Relative risks among family members were comparable between the model of Claus et al. (E. B. Claus et al., Am. J. Hum. Genet., 48: 232-242, 1991) and our model. The dominant and the recessive model provided approximately similar lifetime risks for breast cancer. Our model is suitable for breast cancer risk estimation in a health care setting.

The standard assay for onchocerciasis diagnosis is microscopical detection of microfilariae in skin snips. Skin snipping is painful, requires appropriate sterilization of equipment, and may fail to diagnose light infections. Two alternatives are a polymerase chain reaction (PCR) test which detects parasite DNA in pieces or scrapings of skin and a test based on allergic reactions to topical application of diethylcarbamazine (DEC). We compared these 2 diagnostics with standard skin snip microscopy in 313 individuals from 2 villages in Guinea, with low prevalence after over 10 years of control by the Onchocerciasis Control Programme. Lower and upper bounds on sensitivities and specificities of these 3 tests were estimated. In addition, these parameters were estimated using 5 different statistical models. Where prevalence was low, PCR and the DEC patch test appeared to be more sensitive than skin snipping which has low sensitivity. As the DEC test is non-invasive, simple and cheap, it may provide a good alternative to skin snipping alone for surveillance in low prevalence areas.

The risk of TB infection among schoolchildren in Kenya from the increased incidence of tuberculosis (TB) due to HIV infection was evaluated. The study was based on the 1986-96 tuberculin survey carried out in randomly selected primary schools in 12 districts via the probability-proportional-to-size methods. Districts were grouped according to the year in which TB notification rates started to increase defined as the first year in which notification rate was at least 30% higher than the average during 1986-89. Odd ratio of infection was adjusted via logistic regression analyses involving age sex districts and the presence of a BCG vaccination. HIV prevalence in TB patients and changes in TB infection prevalence were compared between districts. Results showed that tuberculous infection prevalence rates increased strongly in districts where TB notification rates had increased before 1994 (odds ratio 3.1; 95% confidence interval 2.3-4.1) but did not increase in districts where notification rates had increased more recently or not at all. HIV prevalence rates in TB patients were 50% in districts with an early increase in notification rates and 28% in the other study districts. In conclusion countries with increasing HIV prevalence rates may face an increasing TB problem in individuals both with and without HIV infection. Thus additional resources for TB control are needed as a consequence of increased risk of TB infection.

Background and Objectives: The objective was to assess reliability of self-reported sexual histories among sexually transmitted disease clinic attendees who enrolled in a study in 1994. Goal of this Study: Knowledge about the reliability of sexual data is important to decide whether these measures of sexual behavior can be used in epidemiologic studies of sexually transmitted diseases. Study Design: In 288 attendees, degree of agreement was assessed in responses to an identical set of sexual questions asked independently by a medical doctor and a public health nurse and in responses made by members of the same couple (n = 50) to a public health nurse. Results: In the test-retest comparison, high agreement was found for most questions: kappa-values and exact agreement ranged from 0.73 to 0.96 and 54% to 99%, respectively. Participants interviewed by the medical doctor reported significantly lower numbers of partners and a higher age at first intercourse. Stratified analyses showed variability in agreement across subgroups. Most consistent, women provided more reliable reports than men. In the comparison of couples, substantial agreement was found for the municipality where they met (88% agreement; kappa = 0.72) and contraceptive method (87% agreement; kappa = 0.60), but only moderate agreement was found for frequency of sexual intercourse (26% agreement; kappa = 0.50). Conclusion: The authors conclude that data on sexual behavior can be collected reliably among sexually transmitted disease clinic attendees, although reporting bias does occur. The frequency of sexual intercourse was not sufficiently reliable and should be interpreted as an estimate only.

Cross-sectional household surveys are extensively used for data collection, priority setting and programme evaluation in developing countries. They are now being promoted to assess a number of health care interventions such as Control of Diarrhoeal Diseases (CDD), Expanded Programme on Immunization (EPI), AIDS control and child survival programmes. Few field studies of the validity and precision of data generated from these surveys have been carried out, in part because such work is rather demanding of resources. The purpose of this study was to draw conclusions on validity and reliability of data from household surveys through a comparison of results from large-scale surveys on diarrhoea conducted by CDD and Demographic and Health Surveys (DHS) programmes in developing countries. Diarrhoea prevalence and treatment were compared for nine surveys for which little time had passed in between the CDD and DHS survey. The variation in results between the surveys was in many instances too large to be explained only by a true variation in the variable studied. A literature review suggested that validity problems could be due to response and recall errors. The authors caution the use of frequent household surveys for programme impact evaluation. Their cost-effectiveness should be carefully assessed, especially when services' evaluations already have provided evidence that a programme has had a positive effect on the behaviour of health workers and target groups in the community. It is recommended that more research be carried out on how selection and training of surveyors can be improved to make the quality of household surveys in developing countries better.

The aim of this study was to assess the rate of defaulting from treatment among tuberculosis patients diagnosed in the Netherlands in the period 1993-1997, whether risk groups for defaulting can be identified at the start of treatment and the trend of defaulting over time. The Netherlands Tuberculosis Register provided data on all patients diagnosed in the Netherlands during the period 1993-1997. Defaulting probabilities were determined using Kaplan-Meier survival analysis and risk factors were identified with Cox's proportional hazard analysis. Of 7,529 patients with reported treatment outcome, 718 (10%) defaulted or left the country within 1 yr after starting treatment. Defaulting probabilities were 9% (95% confidence interval (CI) 8-10%) among 5,256 patients in low-risk groups, 17% (95% CI 14-19%) among 1,437 asylum seekers and 29% (95% CI 24-34%) among 836 patients in other high-risk groups (other recent immigrants, illegal immigrants, the homeless, prisoners and nationals from Eastern Europe). Defaulting probabilities decreased over time from 12% in 1993 to 7% in 1997. Risk groups for defaulting can be recognized at the start of treatment. The decreasing defaulting probabilities were probably due in part to shortening treatment from 9 to 6 months and improved follow-up of asylum seekers. However, additional measures are needed to reduce defaulting among the homeless, recent immigrants, illegal immigrants and prisoners.

SUMMARY A generalization of the coefficient of determination R2 to general regression models is discussed. A modification of an earlier definition to allow for discrete models is proposed.

In survival clinical trials the response, i.e. (time to) death is delayed. As a result there are at any time patients that have not responded, yet. Given a stopping rule based on the number of deaths, the distribution of the total number of patients that will enter the trial is examined. Besides, a simple approximation to the expected number of patients is presented.

We explore the estimation of sensitivity and specificity of diagnostic tests when the true disease state is unknown. Instrumental variables which subdivide the patient population are used. A logistic model, relating these instrumental variables to the (unknown) true disease state is proposed. It is shown that this procedure allows the goodness-of-fit to the resulting model to be tested.

The concept of clustering of events within elements of a population is explored. Test statistics for testing the null hypothesis of a Poisson process versus the alternative of clustering are developed and their conditional null expectation and variance are derived. An application to an EEG registration of an epilepsy patient is presented.

Using logistic regression, we analyzed the predictive value of a number of entry variables with respect to the outcome variables delayed cerebral ischemia, rebleeding, and poor outcome (death or severe disability) in patients with aneurysmal subarachnoid hemorrhage. The entry variables were clinical condition on admission (grades on the Glasgow Coma Scale, Hunt and Hess system), the amount of subarachnoid and intraventricular blood and the presence of hydrocephalus on the admission computed tomogram, and antifibrinolytic treatment with tranexamic acid. We used data from a prospectively studied population of 176 patients admitted within 72 hours after subarachnoid hemorrhage. The risk of delayed cerebral ischemia was best predicted by the amount of subarachnoid blood, intraventricular blood, and antifibrinolytic treatment irrespective of clinical condition and hydrocephalus. The site of delayed cerebral ischemia was not related to the location of the subarachnoid hemorrhage. Antifibrinolytic treatment was the only entry variable (negatively) predicting the risk of rebleeding. Death or severe disability after 3 months was best predicted by the amount of subarachnoid blood and the initial clinical condition reflected by the grade on the Glasgow Coma Scale.

The consequences of the misspecification of a regression model are considered. For small effects of covariates a proportional consistency theorem is derived. The consistent estimation of the covariance matrix of the estimates is discussed.

The cross-over design for clinical trials when responses are binary is discussed. Three tests which have been proposed for the analysis of this problem are compared by an assessment of their assumptions. A simple test to establish whether it is appropriate to include observations from the second period is presented.

Summary: Two groups of long-stay mentally subnormal epileptic patients, those with a chronically high seizure frequency and those who had become seizure-free, were studied for clinical and EEG factors relating to the prognosis of seizures. The mean period of observation was 20 and 22 years, respectively. All patients had a detailed clinical examination including psychometric testing, and for each, two EEG records were selected for blind semi-quantitative interrater analysis: an admission EEG and an EEG obtained within 6 months of the start of the study. Early onset of seizures, a high initial seizure frequency, multiple seizure types, upper motor neuron signs, and severe mental retardation characterized the seizure group. Significant admission EEG findings in this group included an absence of posterior dominant rhythmic activity, generalized delta activity, and frequent generalized paroxysmal discharges. Results of a linear discriminant analysis confirm that the admission EEG and clinical findings provide a basis for predicting outcome with a reliability on the order of 80%.

The use of cryopreserved semen offers the possibility of home insemination by the instructed partner. A comparative study was designed whereby participants were randomly allocated to use home or clinic insemination for six cycles. If no pregnancy had occurred after six cycles, the site of insemination was switched to the opposite location for a maximum of six further cycles. Fifty-three women with primary infertility fulfilling all entry criteria entered the study. In the first 6 cycles out of 29 home starters, 13 pregnancies were conceived, whereas in 24 clinic starters 11 pregnancies occurred, yielding no statistical difference in pregnancy rate. Of 138 couples who did not meet the criteria in the same period, 45 opted for home insemination, resulting in 20 home-inseminated pregnancies. Again, for comparable subgroups no statistical difference in pregnancy rate between home and clinic insemination was found.

A comparison is made between two approaches to testing goodness of fit of Cox's regression model for survival data. The first approach is based on the inclusion of time dependent covariates, whereas the second one is based on the autocovariance of successive contributions to the derivative of the loglikelihood. It appears that the second test is most appropriate for testing in situations where the structure of the departure from proportional hazards is not known a priori. An approximate expression for the relative efficiency of the two test procedures is presented.

A double-blind single-dose trial was performed on 13 patients with primary open-angle glaucoma. Metoprolol 1%, 2%, and 4% produced a median fall in IOP of 5.6, 5.4, and 6.8 mm Hg, respectively, in the treated eye. The differences in effect between the 1%, 2%, and 4% solutions were not statistically significant. There was no significant fall in IOP in the untreated eyes. There were no significant changes in blood pressure, pulse rate, or pupillary diameter. We could not detect any local objective or subjective side effects during the single-dose study. The clinical usefulness of metoprolol may be limited due to local toxic reactions after treatment with multiple doses. Furthermore, there is the problem of tachyphylaxis that could limit extended treatment with topical beta-adrenergic blocking drugs.

In The Netherlands, retinoblastoma patients have been registered in the Utrecht national retinoblastoma registry since 1862. This register is virtually complete from 1945 onwards. We describe a unique epidemiological survey of the occurrence of non-ocular cancer in all patients registered during the period 1945-1970. The occurrence of non-ocular cancer in relatives of patients with hereditary retinoblastoma is also reported. One hundred and forty-one patients with hereditary retinoblastoma were studied for non-ocular second primary cancer. Nineteen patients died of retinoblastoma. The median follow-up of the surviving 122 patients was 25 years. Seventeen of these patients developed a second primary cancer, most frequently soft-tissue sarcoma. The cumulative incidence of non-ocular cancer was 19% at the age of 35, i.e., a 14-fold increase as compared to the general population. Twelve patients with hereditary retinoblastoma died of non-ocular cancer whereas none of 252 patients with non-hereditary retinoblastoma died of non-ocular cancer. Furthermore, among the parents of our hereditary retinoblastoma patients, 24 (born before 1945) had also been affected by retinoblastoma or had affected sibs. In the parents, 4 tumors occurred, of which 2 were rhabdomyosarcomas and 2 were urinary bladder cancers. Both types of non-ocular cancer were also encountered among the 122 patients with hereditary retinoblastoma. In 103 fathers and 103 mothers of patients with hereditary retinoblastoma who did not have retinoblastoma themselves, there was no previous family history of retinoblastoma. The fathers had a relative risk of 8.3 for pancreatic cancer compared to the general population. There was no significant increase in the number of non-ocular tumors in 332 sibs of patients with hereditary retinoblastoma.

Results of argon laser treatment of pigment epithelial detachments and of subretinal neovascular membranes in Junius-Kuhnt's senile disciform macular degeneration have been compared in a prospective, randomized investigation with the natural course of these affections. Of several hundred cases 26 eyes met our criteria for treatment. In control periods ranging from 1 to 3 years, six eyes with pigment epithelial detachments received laser treatment and seven served as controls; seven eyes with subretinal neovascular membranes received laser treatment and six served as controls. The results indicate that impairment of vision progresses more rapidly after argon laser treatment than when no such treatment is given.

The results of a quantitative spectrophotometric enzyme assay, a fluorescent spot test and a cytochemical assay for glucose-6-phosphate dehydrogenase deficiency were compared systematically. The high sensitivity of the spectrophotometric assay and the fluorescent spot test in the detection of severely deficient individuals was confirmed. For the detection of heterozygote females, however both tests were unreliable; the sensitivities of the fluorescent spot test and the spectrophotometric assay being 32% and 11% respectively. Specificities for both tests were high (99%). Introduction of the ratio of glucose-6-phosphate dehydrogenase and pyruvate kinase (G-6-PD/PK ratio) activities increased the sensitivity of the spectrophotometric assay to nearly 100%. It is concluded that the fluorescent spot test should be used for the diagnosis of G-6-PD deficiency in developing countries; whereas if spectrophotometric enzyme assays are available, the G-6-PD/PK ratio should always be performed. In cases where the ratio is less than 0.70, cytochemical analysis is indicated.

The influence of early diagnosis on sight and survival was studied in 130 patients with bilateral retinoblastoma. Nineteen patients died of this condition. Statistical analysis predicted that 12 of these 19 early deaths could have been prevented if doctors' delay had been less than 1 week. Consequently, a reduction of 65% in mortality is possible. Similarly, statistical analysis also predicted that the number of patients with resulting blindness could be reduced by 40%. Central registration and monitoring of retinoblastoma families would greatly improve early diagnosis.

Cox's proportional hazards model for survival analysis (1972) has gained widespread attention from both applied and theoretical statisticians and is central to most work on survival analysis. To verify the assumption of proportional hazards underlying the model, several methods have recently been put forward (Cox, 1979; Schoenfeld, 1980; Andersen, 1982; Kalbfleisch and Prentice, 1980). These methods consist either of graphical techniques designed so as to visualize departures from the proportional hazards (PH) assumption or of formal tests based on parameterizing the interaction between covariates and time. In this paper a simple test statistic is proposed for detecting departures from the PH assumption. The proposed test is computationally and conceptually straightforward. Expressions for the first two moments of the test statistic are presented.