Estimation of seroconversion rates for infectious diseases: Effects of age and noise.

The presence of serum antibodies is a biomarker of past infection. Instead of seroclassification aimed at measuring seroprevalence a population sample of serum antibody levels may be used to estimate the incidence of seroconversion. This article expands an earlier study into seroincidence estimation, employing models of the seroresponse that include probability of escaping infection, as well as nonexponential decay kinetics and different sources of noise. As previously, a constant force of infection is assumed. When the seroconversion rate is low, a substantial fraction of the population may not be old enough to have experienced any seroconversions, causing underestimation of seroconversion rates that may be substantial at young ages. A correction is given that can be shown to remove such age dependent bias. Simulation studies show that the updated models provide accurate estimates of seroconversion rates, but also that the presence of noise, when unaccounted for, may introduce considerable bias, especially at low (< 0.1/yr) seroconversion rates and young ages. The revised serocalculator scripts can be used to update the R package "seroincidence."

Kinetics of serum antibodies in response to infection with Yersinia enterocolitica.

Information on the kinetics of the serum antibody response to infection with Yersinia enterocolitica is essential to allow the estimation and comparison of seroconversion rates in a diversity of pools of cross-sectional serum antibody measurements. Data from 94 patients with acute enteritis caused by Yersinia infection were used. The follow-up period for the longitudinal study was 36 months, addressed by questionnaire. An indirect enzyme-linked immunosorbent assay method was adapted to determine the concentration of antibodies against Y. enterocolitica in human sera. A mathematical within-host model was used to describe the interaction between pathogen and immune system and the waning of immunity after clearing of the pathogen. All observed antibodies (IgG, IgM, IgA) reached peak levels shortly after infection and then decayed slowly indicating that the median levels decreased only little during the observation period. Estimated maximum peak antibody levels were highest in IgG. Seroresponse curves of all antibodies showed large individual variation between patients. There was no apparent pattern of variation with age, nor any notable difference between genders. Estimated half-times were very long for all antibodies, and their posterior distributions were highly skewed. IgA appeared to have the most persistent antibody response, compared with IgG and IgM. Median peak levels of all three antibodies were similar. There was no significance found between peak antibody levels and severity of symptoms of gastrointestinal infection and severity of joint pain. Our findings allow the use of cross-sectional serum antibody measurements as biomarkers, to estimate seroconversion rates. Such seroincidence estimates include asymptomatic seroconversions, thereby avoiding under-reporting, and allows the comparison of infection pressures among countries, independent of their healthcare and surveillance systems.

Waning and boosting: on the dynamics of immune status.

The aim is to describe the distribution of immune status (as captured by antibody level) on the basis of a within-host submodel for continuous waning and occasional boosting. Inspired by Feller's fundamental work and the more recent delay equation formulation of models for the dynamics of physiologically structured populations, we derive, for given force of infection, a linear renewal equation. The solution is obtained by generation expansion, with the generation number corresponding to the number of times the individual became infected. Our main result provides a precise characterization of the stable distribution of immune status.

Shedding of norovirus in symptomatic and asymptomatic infections.

Norovirus is the most frequent cause of acute infectious gastroenteritis and it is difficult to control in crowded environments like hospitals and nursing homes. Transmission depends on oral intake of virus deposited in the environment by infectious subjects. Data from volunteer studies indicate that virus concentrations in stool are highly variable, but systematic studies of the time-course of shedding and its individual variation are lacking. This paper quantifies norovirus shedding in a large population of 102 subjects, including asymptomatic shedders, and uses a longitudinal model to generalize shedding patterns. Enhanced surveillance for studies of transmission of norovirus in hospital outbreaks has yielded a considerable number of faecal samples from symptomatic and asymptomatic shedders, both from patients and staff. Norovirus concentrations were determined by real-time PCR. A quantitative dynamic model was fitted to the shedding data, in a multilevel Bayesian framework, to study the time-course of shedding and its variation. The results indicate that shedding in asymptomatic subjects is similar to that in symptomatic infections, both showing considerable variation in peak levels (average 105-109 /g faeces) as well as duration of virus shedding (average 8-60 days). Patients appear to shed higher numbers of virus than staff, for slightly longer durations, but the differences are too small to be significant. Given equal shedding, the greater contribution of symptomatic cases to transmission must be caused by their higher efficiency in spreading these viruses. The results of this study will be helpful for risk studies that need to quantify the deposition of virus in the environment.

Decline of IgG pertussis toxin measured in umbilical cord blood, and neonatal and early infant serum.

Maternal pertussis-specific antibodies are passively acquired by infants during pregnancy. An IgG pertussis toxin (IgG-PT) concentration of >20 U/ml is considered to protect neonates against pertussis. To evaluate the IgG concentration at birth and during the first two months of life, we examined the IgG-PT concentration in the umbilical cord blood and three times during the neonatal and early infant period. IgG-PT was measured by validated IgG-specific enzyme-linked immunosorbent assays (ELISA) in umbilical cord blood and in Guthrie card blood samples of umbilical cord blood in 2,790 children, born between 1 August 2006 and 1 December 2008. These measurements were comparable. All children with concentrations of IgG-PT >30 U/ml were included. IgG-PT was also measured in Guthrie card blood samples, when the neonates or early infants were 5 days, 1 month and 2 months old. The mean concentrations of IgG-PT were calculated. The mean concentration of IgG-PT in umbilical cord blood was 60.1 U/ml (LN 4.1; 0.6 SD; n = 103). At the age of 5 days, 1 month and 2 months, the mean concentration of IgG-PT was 40.6 U/ml (LN 3.7; 0.5 SD; n = 103), 20.7 U/ml (LN 3.0; 0.7 SD; n = 62) and 16.7 U/ml (LN 2.8; 0.9 SD; n = 61), respectively. Four percent of the neonates had a concentration of IgG-PT >30 U/ml in umbilical cord blood, which declined to levels around the concentration needed for protection against pertussis (>20 U/ml) in the first two months of life. Hence, it is of great importance to further investigate the safety of maternal immunisation during pregnancy to prevent life-threatening pertussis in newborns.

Time-course of antibody responses against Coxiella burnetii following acute Q fever.

Large outbreaks of Q fever in The Netherlands have provided a unique opportunity for studying longitudinal serum antibody responses in patients. Results are presented of a cohort of 344 patients with acute symptoms of Q fever with three or more serum samples per patient. In all these serum samples IgM and IgG against phase 1 and 2 Coxiella burnetii were measured by an immunofluorescence assay. A mathematical model of the dynamic interaction of serum antibodies and pathogens was used in a mixed model framework to quantitatively analyse responses to C. burnetii infection. Responses show strong heterogeneity, with individual serum antibody responses widely different in magnitude and shape. Features of the response, peak titre and decay rate, are used to characterize the diversity of the observed responses. Binary mixture analysis of IgG peak levels (phases 1 and 2) reveals a class of patients with high IgG peak titres that decay slowly and may represent potential chronic cases. When combining the results of mixture analysis into an odds score, it is concluded that not only high IgG phase 1 may be predictive for chronic Q fever, but also that high IgG phase 2 may aid in detecting such putative chronic cases.

Campylobacter seroconversion rates in selected countries in the European Union.

As a major foodborne pathogen, Campylobacter is frequently isolated from food sources of animal origin. In contrast, human Campylobacter illness is relatively rare, but has a considerable health burden due to acute enteric illness as well as severe sequelae. To study silent transmission, serum antibodies can be used as biomarkers to estimate seroconversion rates, as a proxy for infection pressure. This novel approach to serology shows that infections are much more common than disease, possibly because most infections remain asymptomatic. This study used antibody titres measured in serum samples collected from healthy subjects selected randomly in the general population from several countries in the European Union (EU). Estimates of seroconversion rates to Campylobacter were calculated for seven countries: Romania, Poland, Italy, France, Finland, Denmark and The Netherlands. Results indicate high infection pressures in all these countries, slightly increasing in Eastern EU countries. Of these countries, the differences in rates of notified illnesses are much greater, with low numbers in France and Poland, possibly indicating lower probability of detection due to differences in the notification systems, but in the latter case it cannot be excluded that more frequent exposure confers better protection due to acquired immunity.

Biomarker dynamics: estimating infection rates from serological data.

The marginal distribution of serum antibody titres in a cross-sectional population sample can be expressed as a function of the infection rate, taking into account heterogeneity in peak levels and decay rates. This marginal model allows estimation of incidences, as well as simple tests for homogeneity across age, gender or geographic strata, using likelihood ratio tests. An example is given using Campylobacter serum antibody data. Using a hierarchical dynamic model to analyse data from a follow-up study in patients with symptomatic Campylobacter infection, we show that the serum antibody response consists of a rapid increase to peak levels followed by a slow decline with a geometric mean halftime of 1.4, 0.6 and 0.3 years for IgG, IgM and IgA, respectively. Antibody peak levels and decay rates were highly variable among subjects. Incidence estimates are consistent among different antibody classes (IgG, IgM and IgA). High seroconversion rates indicate that Campylobacter infection is a frequent event, occurring approximately once every year in any adult person, in the Netherlands, supporting the conclusion that a small fraction of infections leads to symptoms severe enough for notification.

Human beings are highly susceptible to low doses of Trichinella spp.

Trichinella is an important foodborne pathogen causing considerable morbidity and mortality. To prevent human trichinellosis, meat inspection for Trichinella spp. at slaughter is a key instrument. Current testing is based on minimal infectious dose in humans, but a scientific basis for this approach is lacking. To this end, a dose-response model must be developed, allowing translation of exposure into disease burden at the population level. We developed novel methods for dose-response assessment using outbreak data incorporating sexual reproduction of the parasite. A selection of suitable outbreak studies, reporting numbers exposed and infected, as well as estimated doses, was collated from a literature study. Humans appear to be highly susceptible: exposure to low doses (few larvae) is associated with a considerable risk of infection. As a consequence, levels of Trichinella in meat must be low to maintain acceptable health risks.

Combining risk assessment and epidemiological risk factors to elucidate the sources of human E. coli O157 infection.

E. coli O157 can be transmitted to humans by three primary (foodborne, environmental, waterborne) and one secondary (person-to-person transmission) pathways. A regression model and quantitative microbiological risk assessments (QMRAs) were applied to determine the relative importance of the primary transmission pathways in NE Scotland. Both approaches indicated that waterborne infection was the least important but it was unclear whether food or the environment was the main source of infection. The QMRAs over-predicted the number of cases by a factor of 30 and this could be because all E. coli O157 strains may not be equally infective and/or the level of infectivity in the dose-response model was too high. The efficacy of potential risk mitigation strategies to reduce human exposure to E. coli O157 using QMRAs was simulated. Risk mitigation strategies focusing on food and environment are likely to have the biggest impact on infection figures.

Seroepidemiological studies indicate frequent and repeated exposure to Campylobacter spp. during childhood.

The annual number of episodes of clinical gastroenteritis caused by Campylobacter spp. in The Netherlands is estimated to be 75 000, i.e. once per 200 person life-years. This number is based on extrapolation of culture results from population-based studies. The number of culture-confirmed cases of Campylobacter infection peaks in the first 3 years of life and again between the ages of 20 and 25 years. The seroepidemiology of Campylobacter describes the relationship between age and exposure to Campylobacter and reflects both symptomatic and asymptomatic infections. Using a validated ELISA system, antibodies to Campylobacter were measured in an age-stratified sample (n=456) of the PIENTER serum collection of the Dutch general population. The seroprevalence of Campylobacter IgG antibodies increased with age, reaching almost 100% at age 20 years. Antibody levels steadily increased with age until young adulthood, suggesting repeated exposure to Campylobacter. In conclusion, seroepidemiological data demonstrated repeated exposures to Campylobacter throughout life, most of which do not lead to clinical symptoms. From young adulthood, >95% of the population in The Netherlands had serological evidence for exposure to Campylobacter.

Enteric virus infection risk from intrusion of sewage into a drinking water distribution network.

Contaminants from the soil surrounding drinking water distribution systems are thought to not enter the drinking water when sufficient internal pressure is maintained. Pressure transients may cause short intervals of negative pressure, and the soil near drinking water pipes often contains fecal material due to the proximity of sewage lines, so that a pressure event may cause intrusion of pathogens. This paper presents a risk model for predicting intrusion and dilution of viruses and their transport to consumers. Random entry and dilution of virus was simulated by embedding the hydraulic model into a Monte Carlo simulation. Special attention was given to adjusting for the coincidence of virus presence and use of tap water, as independently occurring short-term events within the longer interval that the virus is predicted to travel in any branch of the distribution system. The probability that a consumer drinks water contaminated with virus is small, but when this happens the virus concentration tends to be high and the risk of infection may be considerable. The spatial distribution of infection risk is highly heterogeneous. The presence of a chlorine residual reduces the infection risk.

Long-term inactivation study of three enteroviruses in artificial surface and groundwaters, using PCR and cell culture.

Since the transmission of pathogenic viruses via water is indistinguishable from the transmission via other routes and since the levels in drinking water, although significant for health, may be too low for detection, quantitative viral risk assessment is a useful tool for assessing disease risk due to consumption of drinking water. Quantitative viral risk assessment requires information concerning the ability of viruses detected in drinking water to infect their host. To obtain insight into the infectivity of viruses in relation to the presence of virus genomes, inactivation of three different enteroviruses in artificial ground and surface waters under different controlled pH, temperature, and salt conditions was studied by using both PCR and cell culture over time. In salt-peptone medium, the estimated ratio of RNA genomes to infectious poliovirus 1 in freshly prepared suspensions was about 10(0). At 4 degrees C this ratio was 10(3) after 600 days, and at 22 degrees C it was 10(4) after 200 days. For poliovirus 1 and 2 the RNA/infectious virus ratio was higher in artificial groundwater than in artificial surface water, but this was not the case for coxsackievirus B4. When molecular detection is used for virus enumeration, it is important that the fraction of infectious virus (based on all virus genomes detected) decays with time, especially at temperatures near 22 degrees C.

Estimation of incidences of infectious diseases based on antibody measurements.

Owing to under-ascertainment it is difficult if not impossible to determine the incidence of a given disease based on cases notified to routine public health surveillance. This is especially true for diseases that are often present in mild forms as for example diarrhoea caused by foodborne bacterial infections. This study presents a Bayesian approach for obtaining incidence estimates by use of measurements of serum antibodies against Salmonella from a cross-sectional study. By comparing these measurements with antibody measurements from a follow-up study of infected individuals it was possible to estimate the time since last infection for each individual in the cross-sectional study. These time estimates were then converted into incidence estimates. Information about the incidence of Salmonella infections in Denmark was obtained by using blood samples from 1780 persons. The estimated incidence was about 0.094 infections per person year. This number corresponds to 325 infections per culture-confirmed case captured in the Danish national surveillance system. We present a novel approach, termed as seroincidence, that has potentials to compare the sensitivity of public health surveillance between different populations, countries and over time.

Norovirus outbreaks in nursing homes: the evaluation of infection control measures.

Effective infection control measures during norovirus outbreaks are urgently needed in places where vulnerable individuals gather. In the present study, the effect of a number of measures was investigated in daily practice. Forty-nine Dutch nursing homes were monitored prospectively for norovirus outbreaks during two winter seasons. A total of 37 norovirus outbreaks were registered. Control measures were most effective when implemented within 3 days after onset of disease of the first patient. Measures targeted at reduced transmission between persons, via aerosols, and via contaminated surfaces reduced illness in staff and in residents. Reducing illness in staff results in fewer costs for sick leave and substitution of staff and less disruption in the care of residents. The effect of control measures on outbreak duration was limited. This is the first intervention study examining the effect of control measures. Further research is needed to extend and refine the conclusions.

Characterization of drinking water treatment for virus risk assessment.

Removal or inactivation of viruses in drinking water treatment processes can be quantified by measuring the concentrations of viruses or virus indicators in water before and after treatment. Virus reduction is then calculated from the ratio of these concentrations. Most often only the average reduction is reported. That is not sufficient when treatment efficiency must be characterized in quantitative risk assessment. We present three simple models allowing statistical analysis of series of counts before and after treatment: distribution of the ratio of concentrations, and distribution of the probability of passage for unpaired and paired water samples. Performance of these models is demonstrated for several processes (long and short term storage, coagulation/filtration, coagulation/sedimentation, slow sand filtration, membrane filtration, and ozone disinfection) using microbial indicator data from full-scale treatment processes. All three models allow estimation of the variation in (log) reduction as well as its uncertainty; the results can be easily used in risk assessment. Although they have different characteristics and are present in vastly different concentrations, different viruses and/or bacteriophages appear to show similar reductions in a particular treatment process, allowing generalization of the reduction for each process type across virus groups. The processes characterized in this paper may be used as reference for waterborne virus risk assessment, to check against location specific data, and in case no such data are available, to use as defaults.

Use of a Dose-Response Model to Study Temporal Trends in Spatial Exposure to Coxiella burnetii: Analysis of a Multiyear Outbreak of Q Fever.

The Netherlands underwent a large Q fever outbreak between 2007 and 2009. In this paper, we study spatial and temporal Coxiella burnetii exposure trends during this large outbreak as well as validate outcomes against other published studies and provide evidence to support hypotheses on the causes of the outbreak. To achieve this, we develop a framework using a dose-response model to translate acute Q fever case incidence into exposure estimates. More specifically, we incorporate a geostatistical model that accounts for spatial and temporal correlation of exposure estimates from a human Q fever dose-response model to quantify exposure trends during the outbreak. The 2051 cases, with the corresponding age, gender and residential addresses, reside in the region with the highest attack rates during the outbreak in the Netherlands between 2006 and 2009. We conclude that the multiyear outbreak in the Netherlands is caused by sustained release of infectious bacteria from the same sources, which suggests that earlier implementation of interventions may have prevented many of the cases. The model predicts the risk of infection and acute symptomatic Q fever from multiple exposure sources during a multiple-year outbreak providing a robust, evidence-based methodology to support decision-making and intervention design.

Determination of protection zones for Dutch groundwater wells against virus contamination--uncertainty and sensitivity analysis.

Protection zones of shallow unconfined aquifers in The Netherlands were calculated that allow protection against virus contamination to the level that the infection risk of 10(-4) per person per year is not exceeded with a 95% certainty. An uncertainty and a sensitivity analysis of the calculated protection zones were included. It was concluded that protection zones of 1 to 2 years travel time (206-418 m) are needed (6 to 12 times the currently applied travel time of 60 days). This will lead to enlargement of protection zones, encompassing 110 unconfined groundwater well systems that produce 3 x 10(8) m3 y(-1) of drinking water (38% of total Dutch production from groundwater). A smaller protection zone is possible if it can be shown that an aquifer has properties that lead to greater reduction of virus contamination, like more attachment. Deeper aquifers beneath aquitards of at least 2 years of vertical travel time are adequately protected because vertical flow in the aquitards is only 0.7 m per year. The most sensitive parameters are virus attachment and inactivation. The next most sensitive parameters are grain size of the sand, abstraction rate of groundwater, virus concentrations in raw sewage and consumption of unboiled drinking water. Research is recommended on additional protection by attachment and under unsaturated conditions.

Linking the seroresponse to infection to within-host heterogeneity in antibody production.

A recently published model for the serum antibody response to infection appeared well suited for use in statistical analyses of longitudinal serological data. The published model assumed exponential decay with fixed rates for pathogen and serum antibody kinetics, ignoring any within-host heterogeneity in the seroresponse. A bi-exponential model shows that there is rapid initial decay followed by a prolonged period of persistent low serum antibody concentrations. We propose a small modification of the decay model that greatly increases its flexibility by allowing for non-exponential antibody decay. The modified model produces power functions that may be interpreted as a mixture of exponential decay curves, with a mixing distribution representing the relative contribution of many centres of antibody production to the serum antibody concentration. Fitting the power function decay model to observed longitudinal data for pertussis shows improved goodness of fit compared to the exponential decay model, with estimates for the shape parameter (r=2.2; 95% CI (1.7-2.8)) that differ from exponential shape (r=1). The power function decay model predicts more persistent antibody concentrations in the long term (symptomatic threshold reached >30 years after infection) which, when used in biomarker studies, will lead to lower estimates of seroconversion rates compared to exponential antibody decay.

Mixed plaques: statistical evidence how plaque assays may underestimate virus concentrations.

Even at very low concentrations human pathogenic viruses may result in infection and possibly subsequent disease. Ideally, viruses are quantified by use of cell culture assays to determine their infectivity. Plaque assays are common tools for enumeration of viruses in inocula and this process is straightforward when a plaque results from the offspring of a single infectious virus particle. In the course of a study on the usefulness of sewage monitoring for surveillance of polio-virus transmission, sewage samples containing a mixture of two live polio vaccine strains (type 1 and type 3) were analyzed. The total poliovirus concentration in plaque forming units (pfu) was estimated by means of a monolayer plaque assay on L20B cells. Subsequent typing of virus directly by neutralisation of virus from excised plaques revealed the occurrence of plaques containing both type 1 and type 3 virus. This means that there must be plaques that originate from more than one initial infectious virus particle. As a consequence, the estimated virus concentration is incorrect. We present statistical methods that utilize these mixed plaque counts to estimate the concentrations of either virus type in our sewage samples. We can also calculate a correction factor for the error in virus concentration, which would result from equating a pfu to a single infectious particle. Since many quantitative methods in microbiology are based on colony counts, we conclude that such counts should be interpreted with caution, especially when data are used in quantitative microbial risk assessment to estimate the public health impact.