Logistic discrimination of mixtures of M. tuberculosis and non-specific tuberculin reactions

Antibody acquisition models: A new tool for serological surveillance of malaria transmission intensity

Serology has become an increasingly important tool for the surveillance of a wide range of infectious diseases. It has been particularly useful to monitor malaria transmission in elimination settings where existing metrics such as parasite prevalence and incidence of clinical cases are less sensitive. Seroconversion rates, based on antibody prevalence to Plasmodium falciparum asexual blood-stage antigens, provide estimates of transmission intensity that correlate with entomological inoculation rates but lack precision in settings where seroprevalence is still high. Here we present a new and widely applicable method, based on cross-sectional data on individual antibody levels. We evaluate its use as a sero-surveillance tool in a Tanzanian setting with declining malaria prevalence. We find that the newly developed mathematical models produce more precise estimates of transmission patterns, are robust in high transmission settings and when sample sizes are small, and provide a powerful tool for serological evaluation of malaria transmission intensity.

Estimating a Logistic Discrimination Functions When One of the Training Samples Is Subject to Misclassification: A Maximum Likelihood Approach

The problem of discrimination and classification is central to much of epidemiology. Here we consider the estimation of a logistic regression/discrimination function from training samples, when one of the training samples is subject to misclassification or mislabeling, e.g. diseased individuals are incorrectly classified/labeled as healthy controls. We show that this leads to zero-inflated binomial model with a defective logistic regression or discrimination function, whose parameters can be estimated using standard statistical methods such as maximum likelihood. These parameters can be used to estimate the probability of true group membership among those, possibly erroneously, classified as controls. Two examples are analyzed and discussed. A simulation study explores properties of the maximum likelihood parameter estimates and the estimates of the number of mislabeled observations.

Using serological measures to monitor changes in malaria transmission in Vanuatu

Background

With renewed interest in malaria elimination, island environments present unique opportunities to achieve this goal. However, as transmission decreases, monitoring and evaluation programmes need increasingly sensitive tools to assess Plasmodium falciparum and Plasmodium vivax exposure. In 2009, to assess the role of serological markers in evaluating malaria transmission, a cross-sectional seroprevalence study was carried out in Tanna and Aneityum, two of the southernmost islands of the Vanuatu archipelago, areas where malaria transmission has been variably reduced over the past few decades.

Methods

Malaria transmission was assessed using serological markers for exposure to P. falciparum and P. vivax. Filter blood spot papers were collected from 1,249 people from Tanna, and 517 people from Aneityum to assess the prevalence of antibodies to two P. falciparum antigens (MSP-1₁₉ and AMA-1) and two P. vivax antigens (MSP-1₁₉ and AMA-1). Age-specific prevalence was modelled using a simple catalytic conversion model based on maximum likelihood to generate a community seroconversion rate (SCR).

Results

Overall seropositivity in Tanna was 9.4%, 12.4% and 16.6% to P. falciparum MSP-1₁₉, AMA-1 and Schizont Extract respectively and 12.6% and 15.0% to P. vivax MSP-1₁₉ and AMA-1 respectively. Serological results distinguished between areas of differential dominance of either P. vivax or P. falciparum and analysis of age-stratified results showed a step in seroprevalence occurring approximately 30 years ago on both islands, indicative of a change in transmission intensity at this time. Results from Aneityum suggest that several children may have been exposed to malaria since the 2002 P. vivax epidemic.

Conclusion

Seroepidemiology can provide key information on malaria transmission for control programmes, when parasite rates are low. As Vanuatu moves closer to malaria elimination, monitoring changes in transmission intensity and identification of residual malaria foci is paramount in order to concentrate intervention efforts.

A tuberculin skin test survey among Ghanaian school children

Background

Ghana has not conducted a national tuberculin survey or tuberculosis prevalence survey since the establishment of the National Tuberculosis Control Programme. The primary objective of this study was therefore to determine the prevalence of tuberculin skin sensitivity in Ghanaian school children aged 6-10 years in 8 out of 10 regions of Ghana between 2004 and 2006.

Methods

Tuberculin survey was conducted in 179 primary schools from 21 districts in 8 regions. Schools were purposively selected so as to reflect the proportion of affluent private and free tuition public schools as well as the proportion of small and large schools.

Results

Of the 24,778 children registered for the survey, 23,600 (95.2%) were tested of which 21,861 (92.6%) were available for reading. The age distribution showed an increase in numbers of children towards older age: 11% of the children were 6 years and 25%, 10 years. Females were 52.5% and males 47.5%. The proportion of girls was higher in all age groups (range 51.4% to 54.0%, p < 0.001). BCG scar was visible in 89.3% of the children. The percentage of children with a BCG scar differed by district and by age. The percentage of children with a BCG scar decreased with increasing age in all districts, reflecting increasing BCG vaccination coverage in Ghana in the last ten years. The risk of tuberculosis infection was low in the northern savannah zones compared to the southern coastal zones. Using a cut-off of 15 mm, the prevalence of infection ranged from 0.0% to 5.4% and the Annual Risks of Tuberculosis Infection 0.0% to 0.6%. There was an increase in the proportion of infected children after the age of 7 years. Children attending low and middle-class schools had a higher risk of infection than children attending upper-class schools.

Conclusion

Tuberculosis infection is still a public health problem in Ghana and to monitor the trend, the survey needs to be repeated at 5 years interval.

The impact of the HIV epidemic on tuberculosis transmission in Tanzania

OBJECTIVE

To assess the impact of the HIV epidemic on tuberculosis transmission in Tanzania by estimating the trend in annual risk of tuberculosis infection (ARTI) over the period 1983-2003.

DESIGN

Tuberculin survey among school children aged 6-14 years, randomly selected by cluster sampling.

METHODS

Primary outcome was the ARTI among children without a BCG vaccination scar. To obtain time trends, data were reanalysed from three previous surveys carried out at intervals of 5 years since 1983, using identical methods and definitions.

RESULTS

Of 96,226 children included in the analysis (74% of those enrolled), 10,239 (11%) had no BCG scar. The ARTI was 0.68% (95% confidence interval 0.55-0.81). Despite a doubling of notification rates of smear-positive tuberculosis since 1983, this represents an average annual decline since the first survey of 2.7% (P < 0.001). The declining trend in ARTI was observed in 17 of 20 regions, with no association between this trend and region-specific prevalence of HIV infection among patients with tuberculosis (P = 0.575). A similar decline in ARTI was observed among children with a BCG scar and for various ways of estimating the prevalence of tuberculosis infection from the distribution of skin test reactions.

CONCLUSION

Despite substantial increases in tuberculosis incidence, the overall population-level effect of the HIV epidemic on tuberculosis transmission in Tanzania has been limited. This suggests that in the presence of a strong control programme, the HIV epidemic has limited impact on tuberculosis transmission.

Value of dual testing for identifying tuberculous infection

SETTING

A rural community in Chingleput district in Tamil Nadu state in south India.

OBJECTIVE

To determine the value of dual testing with PPD-S and PPD-B for identifying subjects with a tuberculous infection.

DESIGN

About 240,000 subjects in rural south India, all of whom were tested initially with PPD-S and PPD-B, were followed up for 15 years, mainly by total population survey once in every 212 years. The incidence of culture-positive tuberculosis was estimated using life-table technique.

RESULTS

Among 17,530 subjects with an intermediate reaction (8-11 mm) to PPD-S at intake, 285 with an induration to PPD-S exceeding the induration to PPD-B by at least 2mm, had a significantly higher incidence of culture-positive tuberculosis than the remaining (154 and 93 per 100,000), and similarly 481 who had an induration of <10mm to PPD-B compared to those with >or=10 mm (131 and 93 per 100,000). These subjects may be regarded as having a tuberculous infection. Infection with non-tuberculous mycobacteria conferred protection of about 30% against the development of tuberculosis over a 15-year period.

CONCLUSION

In subjects with an intermediate reaction (8-11 mm) to PPD-S, dual testing with PPD-B enabled identification of those with a tuberculous infection. Most of the reactions were due to non-tuberculous mycobacteria.

Estimating the incidence of subclinical infections withLegionella Pneumonia using data augmentation: analysis of an outbreak in The Netherlands

Infections with Legionella bacteria can cause a potentially lethal form of pneumonia known as legionnaires' disease. In 1999 a major outbreak, causing 31 deaths, occurred among visitors and exhibitors of a consumer fair in The Netherlands. The epidemiology of subclinical infections is largely unknown, as there is no reliable method to diagnose such infections. To explore the incidence of subclinical infections, IgG and IgM antibody levels among exhibitors were compared to those among a representative sample of the Dutch population. As exhibitors were assumed to comprise both infected and uninfected individuals, their antibody levels were modelled as a mixture distribution. As infected individuals are expected to cluster around a point source, the spatial aspect of the spread of infections was taken into account. To estimate the distribution of antibody levels among infected individuals and to impute infection status among exhibitors, data augmentation was used. Subclinical infection appeared to be very common and its frequency declined with the distance from the putative source of the outbreak.