Informing the public health response to COVID-19: a systematic review of risk factors for disease, severity, and mortality

BACKGROUND

Severe Acute Respiratory Syndrome coronavirus-2 (SARS-CoV-2) has challenged public health agencies globally. In order to effectively target government responses, it is critical to identify the individuals most at risk of coronavirus disease-19 (COVID-19), developing severe clinical signs, and mortality. We undertook a systematic review of the literature to present the current status of scientific knowledge in these areas and describe the need for unified global approaches, moving forwards, as well as lessons learnt for future pandemics.

METHODS

Medline, Embase and Global Health were searched to the end of April 2020, as well as the Web of Science. Search terms were specific to the SARS-CoV-2 virus and COVID-19. Comparative studies of risk factors from any setting, population group and in any language were included. Titles, abstracts and full texts were screened by two reviewers and extracted in duplicate into a standardised form. Data were extracted on risk factors for COVID-19 disease, severe disease, or death and were narratively and descriptively synthesised.

RESULTS

One thousand two hundred and thirty-eight papers were identified post-deduplication. Thirty-three met our inclusion criteria, of which 26 were from China. Six assessed the risk of contracting the disease, 20 the risk of having severe disease and ten the risk of dying. Age, gender and co-morbidities were commonly assessed as risk factors. The weight of evidence showed increasing age to be associated with severe disease and mortality, and general comorbidities with mortality. Only seven studies presented multivariable analyses and power was generally limited. A wide range of definitions were used for disease severity.

CONCLUSIONS

The volume of literature generated in the short time since the appearance of SARS-CoV-2 has been considerable. Many studies have sought to document the risk factors for COVID-19 disease, disease severity and mortality; age was the only risk factor based on robust studies and with a consistent body of evidence. Mechanistic studies are required to understand why age is such an important risk factor. At the start of pandemics, large, standardised, studies that use multivariable analyses are urgently needed so that the populations most at risk can be rapidly protected.

REGISTRATION

This review was registered on PROSPERO as CRD42020177714 .

Epidemiology of and risk factors for mortality due to carbapenemase-producing organisms (CPO) in healthcare facilities

BACKGROUND

Carbapenemase-producing organisms (CPO) have been largely responsible for the extensive spread of carbapenem resistance, and their prevalence is increasing in many parts of the world.

AIM

To evaluate clinical and molecular epidemiology and mortality associated with CPO among patients.

METHODS

All CPO from clinical and long-term healthcare surveillance cultures across Scotland in 2003-2017 were reviewed retrospectively. Polymerase chain reaction was used to detect genes coding for carbapenemases. A generalized linear mixed model was used to identify risk factors for mortality.

FINDINGS

In total, 290 individuals with CPO were identified. The overall incidence increased over time (P<0.001) from 0.02 to 1.38 per 100,000 population between 2003 and 2017. A total of 243 distinct CPO isolates were obtained from 269 isolations in 214 individuals with available metadata. The majority of the isolates were Enterobacterales (206/243, 84.8%), and Klebsiella pneumoniae (65/206, 31.6%) and Enterobacter cloacae (52/206, 25.2%) were the most common species. VIM (75/243, 30.9%) and NDM (56/243, 23.0%) were the most common carbapenemases. The crude 30-day mortality rate was 11.8% (25/211), while the case fatality rate was 5.7% (12/211). Age >60 years [adjusted odds ratio (aOR) 3.36, 95% confidence interval (CI) 1.06-10.63; P=0.033], presence of non-fermenters (aOR 4.88, 95% CI 1.64-14.47; P=0.005), and systemic infection or organ failure (aOR 4.21, 95% CI 1.38-12.81; P=0.032) were independently associated with 30-day mortality.

CONCLUSION

The incidence of CPO in Scotland is low but increasing. Awareness is required that inpatients aged >60 years, patients with systemic infection or organ failure, and patients presenting with non-fermenters are at higher risk of death from CPO.

Modelling the impact of curtailing antibiotic usage in food animals on antibiotic resistance in humans

Consumption of antibiotics in food animals is increasing worldwide and is approaching, if not already surpassing, the volume consumed by humans. It is often suggested that reducing the volume of antibiotics consumed by food animals could have public health benefits. Although this notion is widely regarded as intuitively obvious there is a lack of robust, quantitative evidence to either support or contradict the suggestion. As a first step towards addressing this knowledge gap, we develop a simple mathematical model for exploring the generic relationship between antibiotic consumption by food animals and levels of resistant bacterial infections in humans. We investigate the impact of restricting antibiotic consumption by animals and identify which model parameters most strongly determine that impact. Our results suggest that, for a wide range of scenarios, curtailing the volume of antibiotics consumed by food animals has, as a stand-alone measure, little impact on the level of resistance in humans. We also find that reducing the rate of transmission of resistance from animals to humans may be more effective than an equivalent reduction in the consumption of antibiotics in food animals. Moreover, the response to any intervention is strongly determined by the rate of transmission from humans to animals, an aspect which is rarely considered.

Seroprevalence of respiratory viral pathogens of indigenous calves in Western Kenya

Most studies of infectious diseases in East African cattle have concentrated on gastro-intestinal parasites and vector-borne diseases. As a result, relatively little is known about viral diseases, except for those that are clinically symptomatic or which affect international trade such as foot and mouth disease, bluetongue and epizootic haemorrhagic disease. Here, we investigate the seroprevalence, distribution and relationship between the viruses involved in respiratory disease, infectious bovine rhinotracheitis virus (IBR), bovine parainfluenza virus Type 3 (PIV3) and bovine viral diarrhoea virus (BVDV) in East African Shorthorn Zebu calves. These viruses contribute to the bovine respiratory disease complex (BRD) which is responsible for major economic losses in cattle from intensive farming systems as a result of pneumonia. We found that calves experience similar risks of infection for IBR, PIV3, and BVDV with a seroprevalence of 20.9%, 20.1% and 19.8% respectively. We confirm that positive associations exist between IBR, PIV3 and BVDV; being seropositive for any one of these three viruses means that an individual is more likely to be seropositive for the other two viruses than expected by chance.

Using genomics data to reconstruct transmission trees during disease outbreaks

Genetic sequence data from pathogens present a novel means to investigate the spread of infectious disease between infected hosts or infected premises, complementing traditional contact-tracing approaches, and much recent work has gone into developing methods for this purpose. The objective is to recover the epidemic transmission tree, which identifies who infected whom. This paper reviews the various approaches that have been taken. The first step is to define a measure of difference between sequences, which must be done while taking into account such factors as recombination and convergent evolution. Three broad categories of method exist, of increasing complexity: those that assume no withinhost genetic diversity or mutation, those that assume no within-host diversity but allow mutation, and those that allow both. Until recently, the assumption was usually made that every host in the epidemic could be identified, but this is now being relaxed, and some methods are intended for sparsely sampled data, concentrating on the identification of pairs of sequences that are likely to be the result of direct transmission rather than inferring the complete transmission tree. Many of the procedures described here are available to researchers as free software.

Efficient national surveillance for health-care-associated infections

Background

Detecting novel healthcare-associated infections (HCAI) as early as possible is an important public health priority. However, there is currently no evidence base to guide the design of efficient and reliable surveillance systems. Here we address this issue in the context of a novel pathogen spreading primarily between hospitals through the movement of patients.

Methods

Using a mathematical modelling approach we compare the current surveillance system for a HCAI that spreads primarily between hospitals due to patient movements as it is implemented in Scotland with a gold standard to determine if the current system is maximally efficient or whether it would be beneficial to alter the number and choice of hospitals in which to concentrate surveillance effort.

Results

We validated our model by demonstrating that it accurately predicts the risk of meticillin-resistant Staphylococcus aureus bacteraemia cases in Scotland.

Using the 29 (out of 182) sentinel hospitals that currently contribute most of the national surveillance effort results in an average detection time of 117 days. A reduction in detection time to 87 days is possible by optimal selection of 29 hospitals. Alternatively, the same detection time (117 days) can be achieved using just 22 optimally selected hospitals. Increasing the number of sentinel hospitals to 38 (teaching and general hospitals) reduces detection time by 43 days; however decreasing the number to seven sentinel hospitals (teaching hospitals) increases detection time substantially to 268 days.

Conclusions

Our results show that the current surveillance system as it is used in Scotland is not optimal in detecting novel pathogens when compared to a gold standard. However, efficiency gains are possible by better choice of sentinel hospitals, or by increasing the number of hospitals involved in surveillance. Similar studies could be used elsewhere to inform the design and implementation of efficient national, hospital-based surveillance systems that achieve rapid detection of novel HCAIs for minimal effort.

Electronic supplementary material

The online version of this article (doi:10.1186/s12889-015-2172-9) contains supplementary material, which is available to authorized users.

Relative associations of cattle movements, local spread, and biosecurity with bovine viral diarrhoea virus (BVDV) seropositivity in beef and dairy herds

The success of bovine viral diarrhoea virus (BVDV) eradication campaigns can be undermined by spread through local transmission pathways and poor farmer compliance with biosecurity recommendations. This work combines recent survey data with cattle movement data to explore the issues likely to impact on the success of BVDV control in Scotland. In this analysis, data from 249 beef suckler herds and 185 dairy herds in Scotland were studied retrospectively to determine the relative influence of cattle movements, local spread, and biosecurity on BVDV seropositivity. Multivariable logistic regression models revealed that cattle movement risk factors had approximately 3 times greater explanatory power than risk factors for local spread amongst beef suckler herds, but approximately the same explanatory power as risk factors for local spread amongst dairy herds. These findings are most likely related to differences in cattle husbandry practices and suggest that where financial prioritization is required, focusing on reducing movement-based risk is likely to be of greatest benefit when applied to beef suckler herds. The reported use of biosecurity measures such as purchasing cattle from BVDV accredited herds only, performing diagnostic screening at the time of sale, implementing isolation periods for purchased cattle, and installing double fencing on shared field boundaries had minimal impact on the risk of beef or dairy herds being seropositive for BVDV. Only 28% of beef farmers and 24% of dairy farmers with seropositive herds recognized that their cattle were affected by BVDV and those that did perceive a problem were no less likely to sell animals as replacement breeding stock and no more likely to implement biosecurity measures against local spread than farmers with no perceived problems. In relation to the current legislative framework for BVDV control in Scotland, these findings emphasize the importance of requiring infected herds take appropriate biosecurity measures to prevent further disease transmission and conducting adequate follow-up to ensure that biosecurity measures are being implemented correctly in the field.

Bluetongue and epizootic haemorrhagic disease virus in local breeds of cattle in Kenya

The presence of bluetongue virus (BTV) and Epizootic Haemorrhagic Disease virus (EHDV) in indigenous calves in western Kenya was investigated. Serum was analysed for BTV and EHDV antibodies. The population seroprevalences for BTV and EHDV for calves at 51 weeks of age were estimated to be 0.942 (95% CI 0.902–0.970) and 0.637 (95% CI 0.562–0.710), respectively, indicating high levels of circulating BTV and EHDV. The odds ratio of being positive for BTV if EHDV positive was estimated to be 2.57 (95% CI 1.37–4.76). When 99 calves were tested for BTV and EHDV RNA by real-time RT-PCR, 88.9% and 63.6% were positive, respectively. Comparison of the serology and real-time RT-PCR results revealed an unexpectedly large number of calves that were negative by serology but positive by real-time RT-PCR for EHDV. Eight samples positive for BTV RNA were serotyped using 24 serotype-specific real-time RT-PCR assays. Nine BTV serotypes were detected, indicating that the cattle were infected with a heterogeneous population of BTVs. The results show that BTV and EHDV are highly prevalent, with cattle being infected from an early age.

Disease transmission on fragmented contact networks: livestock-associated Methicillin-resistant Staphylococcus aureus in the Danish pig-industry

Animal trade in industrialised livestock-production systems creates a complex, heterogeneous, contact network that shapes between-herd transmission of infectious diseases. We report the results of a simple mathematical model that explores patterns of spread and persistence of livestock-associated Methicillin-resistant Staphylococcus aureus (LA-MRSA) in the Danish pig-industry associated with this trade network. Simulations show that LA-MRSA can become endemic sustained by animal movements alone. Despite the extremely low predicted endemic prevalence, eradication may be difficult, and decreasing within-farm prevalence, or the time it takes a LA-MRSA positive farm to recover a negative status, fails to break long-term persistence. Our results suggest that a low level of non-movement induced transmission strongly affects MRSA dynamics, increasing endemic prevalence and probability of persistence. We also compare the model-predicted risk of 291 individual farms becoming MRSA positive, with results from a recent Europe-wide survey of LA-MRSA in holdings with breeding pigs, and find a significant correlation between contact-network connectivity properties and the model-estimated risk measure.

Evaluation of risks of foot-and-mouth disease in Scotland to assist with decision making during the 2007 outbreak in the UK

An outbreak of foot-and-mouth disease (FMD) occurred in Surrey on August 3, 2007. A Great Britain-wide ban on livestock movements was implemented immediately. This coincided with the start of seasonal sheep movements off the hills in Scotland; the majority of these animals are sold via markets. The ban therefore posed severe economic and animal-welfare hardships if it was to last through September and beyond. The Scottish Government commissioned an analysis to assess the risk of re-opening markets given the uncertainty about whether FMD had entered Scotland. Tracing of livestock moved from within the risk zone in England between July 16 and August 3 identified contact chains to 12 Scottish premises; veterinary field inspections found a further three unrecorded movements. No signs of infection were found on these holdings. Under the conservative assumption that a single unknown Scottish holding was infected with FMD, an estimate of the time-dependent probability of Scotland being FMD free given no detection was made. Analyses indicated that if FMD was not detected by early to mid-September then it was highly probable that Scotland was FMD free. Risk maps were produced to visualise the potential spread of FMD across Scotland if it was to spread either locally or via market sales.

The predicted impact of immunosuppression upon population age-intensity profiles for schistosomiasis

The slow development of acquired immunity is thought to be responsible for the characteristic convex age-intensity curve seen in human schistosome infection, which peaks earlier in more heavily infected populations (this is described as a peak shift). Schistosomes are able to suppress protective host responses, and it is hypothesized that this suppression is responsible for the delayed development of protective responses. A deterministic mathematical model is used to describe levels of infection and immunity in an endemic population, incorporating protective immune responses which either reduce adult worm burden or reduce superinfection. Suppression, related to current worm burden, is also included and acts against one or both protective responses. If suppression acts against the entire protective response, it is able to delay the development of protective immunity, and the peak shift is predicted to be reversed at higher infection intensities, with removal of the peaks altogether at the highest levels of infection and/or suppression. If only the anti-adult worm protective immune response is vulnerable to suppression, while the anti-reinfection response remains intact, then suppression does not remove the peak in the age-intensity curve. These findings are discussed in the light of existing field and experimental data.

Epidemiological implications of the contact network structure for cattle farms and the 20-80 rule

The network of movements of cattle between farm holdings is an important determinant of the potential rates and patterns of spread of infectious diseases. Because cattle movements are uni-directional, the network is unusual in that the risks of acquiring infection (by importing cattle) and of passing infection on (by exporting cattle) can be clearly distinguished, and there turns out to be no statistically significant correlation between the two. This means that the high observed degree of heterogeneity in numbers of contacts does not result in an increase in the basic reproduction number, R0, in contrast to findings from studies of other contact networks. Despite this, it is still the case that just 20% of holdings contribute at least 80% of the value of R0.

Modelling the epidemiology and transmission of Verocytotoxin-producing Escherichia coli serogroups O26 and O103 in two different calf cohorts

Mathematical models are constructed to investigate the population dynamics of Verocytotoxin-producing Escherichia coli (VTEC) serogroups O26 and O103 in two different calf cohorts. We compare the epidemiological characteristics of these two serogroups within the same calf cohort as well as the same serogroups between the two calf cohorts. The sources of infection are quantified for both calf cohort studies. VTEC serogroups O26 and O103 mainly differ in the rate at which calves acquire infection from sources other than infected calves, while infected calves typically remain infectious for less than 1 week regardless of the serogroups. Fewer than 20% of VTEC-positive samples are the result of calf-to-calf transmission. PFGE typing data are available for VTEC-positive samples to further subdivide the serogroup data in one of the two calf cohort studies. For serogroup O26 but not O103, there is evidence for unequal environmental exposure to infection with different PFGE types.

Enhancement of bacterial competitive fitness by apramycin resistance plasmids from non-pathogenic Escherichia coli

The study of antibiotic resistance has in the past focused on organisms that are pathogenic to humans or animals. However, the development of resistance in commensal organisms is of concern because of possible transfer of resistance genes to zoonotic pathogens. Conjugative plasmids are genetic elements capable of such transfer and are traditionally thought to engender a fitness burden on host bacteria. In this study, conjugative apramycin resistance plasmids isolated from newborn calves were characterized. Calves were raised on a farm that had not used apramycin or related aminoglycoside antibiotics for at least 20 months prior to sampling. Of three apramycin resistance plasmids, one was capable of transfer at very high rates and two were found to confer fitness advantages on new Escherichia coli hosts. This is the first identification of natural plasmids isolated from commensal organisms that are able to confer a fitness advantage on a new host. This work indicates that reservoirs of antibiotic resistance genes in commensal organisms might not decrease if antibiotic usage is halted.

Low-coverage vaccination strategies for the conservation of endangered species

The conventional objective of vaccination programmes is to eliminate infection by reducing the reproduction number of an infectious agent to less than one, which generally requires vaccination of the majority of individuals. In populations of endangered wildlife, the intervention required to deliver such coverage can be undesirable and impractical; however, endangered populations are increasingly threatened by outbreaks of infectious disease for which effective vaccines exist. As an alternative, wildlife epidemiologists could adopt a vaccination strategy that protects a population from the consequences of only the largest outbreaks of disease. Here we provide a successful example of this strategy in the Ethiopian wolf, the world's rarest canid, which persists in small subpopulations threatened by repeated outbreaks of rabies introduced by domestic dogs. On the basis of data from past outbreaks, we propose an approach that controls the spread of disease through habitat corridors between subpopulations and that requires only low vaccination coverage. This approach reduces the extent of rabies outbreaks and should significantly enhance the long-term persistence of the population. Our study shows that vaccination used to enhance metapopulation persistence through elimination of the largest outbreaks of disease requires lower coverage than the conventional objective of reducing the reproduction number of an infectious agent to less than one.

Molecular characterisation of bovine faecal Escherichia coli shows persistence of defined ampicillin resistant strains and the presence of class 1 integrons on an organic beef farm

Antimicrobial use is heavily restricted on organic farms; however, few studies have been conducted to investigate the impact this has on the epidemiology of resistance in pathogenic and commensal bacteria. We investigated the persistence of antimicrobial resistant Escherichia coli within an organic beef herd over a period of 28 months. Faecal samples collected monthly from three calf cohorts and annually from adult cattle and environmental samples, were screened for the presence of ampicillin, apramycin and nalidixic acid resistant E. coli. The prevalence of ampicillin resistance ranged from 27.3 to 40.7% in the annual herd and environmental samplings (n=22-55) and was greater in the calf cohorts, with a peak cohort prevalence of >47% in all 3 years (n=16-18). Apramycin and nalidixic acid resistant E. coli were rare. Pulsed-field gel electrophoresis (PFGE) identified 10 main genotype groups within the herd, with evidence of strain transmission between different livestock groups, animal species and years. Multiple resistance was found in >44% of isolates tested, with ampicillin, neomycin, sulphamethoxazole and tetracycline carriage the commonest phenotype identified. PCR detected the presence of class 1 integrons in <5% of resistant isolates, 6/7 of which were of cattle origin. These data demonstrate that ampicillin resistant E. coli was common on the farm despite restricted antimicrobial use, although strain diversity was low. Persistence of defined genotype groups was observed between years, together with the transmission of resistant strains between different animal species on the farm.

Super-shedding cattle and the transmission dynamics of Escherichia coli O157

The prevalence of Escherichia coli O157 displays striking variability across the Scottish cattle population. On 78% of farms, in a cross-sectional survey of 952, no shedding of E. coli O157 was detected, but on a small proportion, approximately 2%, very high prevalences of infection were found (with 90-100% of pats sampled being positive). We ask whether this variation arises from the inherent stochasticity in transmission dynamics or whether it is a signature of underlying heterogeneities in the cattle population. A novel approach is taken whereby the cross-sectional data are viewed as providing independent snapshots of a dynamic process. Using maximum-likelihood methods to fit time-dependent epidemiological models to the data we obtain estimates for the rates of immigration and transmission of E. coli O157 infection - parameters which have not been previously quantified in the literature. A comparison of alternative model fits reveals that the variation in the prevalence data is best explained when a proportion of the cattle are assumed to transmit infection at much higher levels than the rest - the so-called super-shedders. Analysis of a second dataset, comprising samples taken from 32 farms at monthly intervals over a period of 1 year, additionally yields an estimate for the rate of recovery from infection. The pattern of prevalence displayed in the second dataset also strongly supports the super-shedder hypothesis.

Spatial and temporal epidemiology of sporadic human cases of Escherichia coli O157 in Scotland, 1996-1999

In Scotland, between 1995 and 2000 there were between 4 and 10 cases of illness per 100000 population per year identified as being caused by Escherichia coli O157, whereas in England and Wales there were between 1 and 2 cases per 100000 population per year. Within Scotland there is significant regional variation. A cluster of high rate areas was identified in the Northeast of Scotland and a cluster of low rate areas in central-west Scotland. Temporal trends follow a seasonal pattern whilst spatial effects appeared to be distant rather than local. The best-fit model identified a significant spatial trend with case rate increasing from West to East, and from South to North. No statistically significant spatial interaction term was found. In the models fitted, the cattle population density, the human population density, and the number of cattle per person were variously significant. The findings suggest that rural/urban exposures are important in sporadic infections.

Heterogeneous shedding of Escherichia coli O157 in cattle and its implications for control

Identification of the relative importance of within- and between-host variability in infectiousness and the impact of these heterogeneities on the transmission dynamics of infectious agents can enable efficient targeting of control measures. Cattle, a major reservoir host for the zoonotic pathogen Escherichia coli O157, are known to exhibit a high degree of heterogeneity in bacterial shedding densities. By relating bacterial count to infectiousness and fitting dynamic epidemiological models to prevalence data from a cross-sectional survey of cattle farms in Scotland, we identify a robust pattern: approximately 80% of the transmission arises from the 20% most infectious individuals. We examine potential control options under a range of assumptions about within- and between-host variability in infection dynamics. Our results show that the within-herd basic reproduction ratio, R(0), could be reduced to <1 with targeted measures aimed at preventing infection in the 5% of individuals with the highest overall infectiousness. Alternatively, interventions such as vaccination or the use of probiotics that aim to reduce bacterial carriage could produce dramatic reductions in R(0) by preventing carriage at concentrations corresponding to the top few percent of the observed range of counts. We conclude that a greater understanding of the cause of the heterogeneity in bacterial carriage could lead to highly efficient control measures to reduce the prevalence of E. coli O157.

Prevalence and virulence factors of Escherichia coli serogroups O26, O103, O111, and O145 shed by cattle in Scotland

A national survey was conducted to determine the prevalence of Escherichia coli O26, O103, O111, and O145 in feces of Scottish cattle. In total, 6,086 fecal pats from 338 farms were tested. The weighted mean percentages of farms on which shedding was detected were 23% for E. coli O26, 22% for E. coli O103, and 10% for E. coli O145. The weighted mean prevalences in fecal pats were 4.6% for E. coli O26, 2.7% for E. coli O103, and 0.7% for E. coli O145. No E. coli O111 was detected. Farms with cattle shedding E. coli serogroup O26, O103, or O145 were widely dispersed across Scotland and were identified most often in summer and autumn. However, on individual farms, fecal shedding of E. coli O26, O103, or O145 was frequently undetectable or the numbers of pats testing positive were small. For serogroup O26 or O103 there was clustering of positive pats within management groups, and the presence of an animal shedding one of these serogroups was a positive predictor for shedding by others, suggesting local transmission of infection. Carriage of vtx was rare in E. coli O103 and O145 isolates, but 49.0% of E. coli O26 isolates possessed vtx, invariably vtx1 alone or vtx1 and vtx2 together. The carriage of eae and ehxA genes was highly associated in all three serogroups. Among E. coli serogroup O26 isolates, 28.9% carried vtx, eae, and ehxA-a profile consistent with E. coli O26 strains known to cause human disease.

Modelling the spread of scrapie in a sheep flock: evidence for increased transmission during lambing seasons

Presence of scrapie infectivity in the placenta suggests the possibility of increased transmission of scrapie during the lambing season. This hypothesis was explored here using a mathematical model of scrapie transmission dynamics which has previously been successfully used to study several scrapie outbreaks in Scottish sheep flocks. It was applied here to the Langlade experimental sheep flock (INRA Toulouse, France), in which a natural scrapie epidemic started in 1993. Extensive data were available, including pedigree, scrapie histopathological diagnoses and PrP genotypes. Detailed simulations of the scrapie outbreak reveal that the observed patterns of seasonality in incidence can not be accounted for by seasonality in demography alone and provide strong support for the hypothesis of increased transmission during lambing. Observations from several other scrapie outbreaks also showing seasonal incidence patterns support these conclusions.

Quantifying the level of under-detection of Trypanosoma brucei rhodesiense sleeping sickness cases

To formally quantify the level of under-detection of Trypanosoma brucei rhodesiense sleeping sickness (SS) during an epidemic in Uganda, a decision tree (under-detection) model was developed; concurrently, to quantify the subset of undetected cases that sought health care but were not diagnosed, a deterministic (subset) model was developed. The values of the under-detection model parameters were estimated from previously published records of the duration of symptoms prior to presentation and the ratio of early to late stage cases in 760 SS patients presenting at LIRI hospital, Tororo, Uganda during the 1988--1990 epidemic of SS. For the observed early to late stage ratio of 0.47, we estimate that the proportion of under-detection in the catchment area of LIRI hospital was 0.39 (95% CI 0.37--0.41) i.e. 39% of cases are not reported. Based on this value, it is calculated that for every one reported death of SS, 12.0 (95% CI 11.0--13.0) deaths went undetected in the LIRI hospital catchment area - i.e. 92% of deaths are not reported. The deterministic (subset) model structured on the possible routes of a SS infection to either diagnosis or death through the health system or out of it, showed that of a total of 73 undetected deaths, 62 (CI 60-64) (85%) entered the healthcare system but were not diagnosed, and 11 (CI 11--12) died without seeking health care from a recognized health unit. The measure of early to late stage presentation provides a tractable measure to determine the level of rhodesiense SS under-detection and to gauge the effects of interventions aimed at increasing treatment coverage.

Modelling the epidemiology of Verocytotoxin-producing Escherichia coli serogroups in young calves

We investigate the epidemiology of 12 Verocytotoxin-producing Escherichia coli (VTEC) serogroups observed in a calf cohort on a Scottish beef farm. Fitting mathematical models to the observed time-course of infections reveals that there is significant calf-to-calf transmission of VTEC. Our models suggest that 40% of all detected infections are from calf-to-calf transmission and 60% from other sources. Variation in the rates at which infected animals recover from infection by different VTEC serogroups appears to be important. Two thirds of the observed VTEC serogroups are lost from infected calves within 1 day of infection, while the rest persist for more than 3 days. Our study has demonstrated that VTEC are transmissible between calves and are typically lost from infected animals in less than 1 week. We suggest that future field studies may wish to adopt a tighter sampling frame in order to detect all circulating VTEC serogroups in similar animal populations.

Shedding patterns of verocytotoxin-producing Escherichia coli strains in a cohort of calves and their dams on a Scottish beef farm

Rectal fecal samples were taken once a week from 49 calves on the same farm. In addition, the dams of the calves were sampled at the time of calf birth and at the end of the study. Strains of verocytotoxin-producing Escherichia coli (VTEC) were isolated from these samples by using PCR and DNA probe hybridization tests and were characterized with respect to serotype, verocytotoxin gene (vtx) type, and the presence of the intimin (eae) and hemolysin (ehxA) genes. A total of 170 VTEC strains were isolated during 21 weeks from 130 (20%) of 664 samples from calves and from 40 (47%) of 86 samples from their dams. The characteristics of the calf strains differed from those strains isolated from the dams with respect to verocytotoxin 2 and the presence of the eae gene. In addition, no calf shed the same VTEC serogroup (excluding O?) as its dam at birth or at the end of the study. The most frequently detected serogroups in calves were serogroup O26 and provisional serogroup E40874 (VTEC O26 was found in 25 calves), whereas in dams serogroup O91 and provisional serogroup E54071 were the most common serogroups. VTEC O26 shedding appeared to be associated with very young calves and declined as the calves aged, whereas VTEC O2 shedding was associated with housing of the animals. VTEC O26 strains from calves were characterized by the presence of the vtx1, eae, and ehxA genes, whereas vtx2 was associated with VTEC O2 and provisional serogroup E40874. The high prevalence of VTEC O26 and of VTEC strains harboring the eae gene in this calf cohort is notable because of the association of the O26 serogroup and the presence of the eae gene with human disease. No association between calf diarrhea and any of the VTEC serogroups was identified.

Spatial and temporal risk factors for the early detection of Trypanosoma brucei rhodesiense sleeping sickness patients in Tororo and Busia districts, Uganda

We have carried out a study of risk factors for early detection of Trypanosoma brucei rhodesiense sleeping sickness. Records of sleeping sickness patients from 1987 to 2001 from Tororo and Busia districts in Uganda were reviewed for their village of origin and clinical stage (early or late). All villages that reported sleeping sickness and fixed post-diagnostic sleeping sickness health units in Tororo and Busia districts were geo-referenced. The spatial distribution of early and late stage patient detection by health units was analysed using Geographical Information Systems (GIS). Of 1316 sleeping sickness patients admitted at the Livestock Health Research Institute and Busolwe hospitals and Lumino health centre from Tororo and Busia districts, 471 (35.8%) were early stage, 825 (62.7%) were late stage, while 20 (1.5%) were not staged. Five hundred and eighty-five (44.5%) came from within a 10 km radius of the reporting health units. After multivariate analysis, the proportion of early stage patients detected was found to be significantly associated with patients originating from within a 10 km radius of the health unit (P < 0.01), with adults (>19 years) (P < 0.01), and with annual parish incidence (P < 0.01). Application of GIS and the early to late stages ratio are an informative and powerful means of determining efficiency of surveillance of sleeping sickness.

High frequency transfer and horizontal spread of apramycin resistance in calf faecal Escherichia coli

OBJECTIVES

The aminoglycoside apramycin has been used extensively in animal husbandry in the UK since 1978. This study aimed to determine both whether calves that had never been treated with aminoglycoside antibiotics harboured apramycin-resistant (apr(R)) commensal Escherichia coli, and the mode of spread of the resistance gene.

METHODS

Apr(R) E. coli from weekly calf faecal samples were typed by pulsed-field gel electrophoresis, antibiotic resistance phenotype, plasmid restriction profiles and plasmid transfer frequencies.

RESULTS

During 4 months of weekly sampling, six of 11 calves were found to harbour apr(R) E. coli. All apr(R) E. coli (45) were cross-resistant to gentamicin and tobramycin, which are both used in human medicine. Resistance was conferred by the aac(3)IV gene, present on three different conjugative plasmids. Two of these plasmids also mediated tetracycline and streptomycin resistance. One plasmid demonstrated very high transfer frequencies and was found in three different genotypes.

CONCLUSIONS

We report the presence of apr(R) commensal E. coli in cattle that have never been treated with aminoglycosides. The presence of one conjugative plasmid in three different genotypes is evidence of horizontal spread of this plasmid. This is the first report of a very high transfer frequency of apr(R) plasmid, demonstrating horizontal spread in the commensal flora of food animals.

Temporal shedding patterns and virulence factors of Escherichia coli serogroups O26, O103, O111, O145, and O157 in a cohort of beef calves and their dams

This study investigated the shedding of Escherichia coli O26, O103, O111, O145, and O157 in a cohort of beef calves from birth over a 5-month period and assessed the relationship between shedding in calves and shedding in their dams, the relationship between shedding and scouring in calves, and the effect of housing on shedding in calves. Fecal samples were tested by immunomagnetic separation and by PCR and DNA hybridization assays. E. coli O26 was shed by 94% of calves. Over 90% of E. coli O26 isolates carried the vtx(1), eae, and ehl genes, 6.5% carried vtx(1) and vtx(2), and one isolate carried vtx(2) only. Serogroup O26 isolates comprised seven pulsed-field gel electrophoresis (PFGE) patterns but were dominated by one pattern which represented 85.7% of isolates. E. coli O103 was shed by 51% of calves. Forty-eight percent of E. coli O103 isolates carried eae and ehl, 2% carried vtx(2), and none carried vtx(1). Serogroup O103 isolates comprised 10 PFGE patterns and were dominated by two patterns representing 62.5% of isolates. Shedding of E. coli O145 and O157 was rare. All serogroup O145 isolates carried eae, but none carried vtx(1) or vtx(2). All but one serogroup O157 isolate carried vtx(2), eae, and ehl. E. coli O111 was not detected. In most calves, the temporal pattern of E. coli O26 and O103 shedding was random. E. coli O26 was detected in three times as many samples as E. coli O103, and the rate at which calves began shedding E. coli O26 for the first time was five times greater than that for E. coli O103. For E. coli O26, O103, and O157, there was no association between shedding by calves and shedding by dams within 1 week of birth. For E. coli O26 and O103, there was no association between shedding and scouring, and there was no significant change in shedding following housing.

Foot-and-mouth disease in the UK: what should we do next time?

Mathematical models were used to guide the UK foot-and-mouth disease (FMD) control policy during the 2001 epidemic. Based on data collected during the epidemic, prospective modelling using a variety of approaches gave the same conclusions: (i) that the epidemic had not been brought under control by 'traditional' methods, and (ii) that neighbourhood control measures (the contiguous cull) could bring the epidemic under control and result in a net saving of livestock. Retrospective analyses suggest that the subsequent course of the epidemic was consistent with a beneficial impact of the contiguous cull and that it would have been difficult to achieve a better outcome using reactive vaccination, which would have required very large-scale vaccination programmes to have been implemented quickly. Perhaps the most important lesson to be learned is the vital importance of rapid and decisive intervention in response to an outbreak, including the earliest possible implementation of a national ban on the movement of livestock once the presence of disease is confirmed.

Contrasting cellular responses in Schistosoma haematobium infected and exposed individuals from areas of high and low transmission in Zimbabwe

The study compared cytokine profiles of individuals from two areas with different transmission patterns for Schistosoma haematobium. One area was a high transmission (HT) while the other was a low transmission (LT) area for S. haematobium. Observations on cellular immune responses were made on stimulated peripheral blood mononuclear cells (PBMC), which were collected pre-treatment, then at 12 and 18 months post treatment. Stimulation was with schistosome worm and egg antigens and a mitogen, phaetohaemaglutinin (PHA). Observations were made on PBMC proliferation and the profiles of cytokine produced over a 5-day incubation period. The two distinct areas showed significant differences on both levels of proliferation and cytokine production for all the measured classes (IL-4, IL-5, IL-10 and IFN-gamma). PBMC from individuals from the LT area had high levels of proliferation but low cytokine production to both antigen stimulants while PBMC from individuals from the HT area showed low levels of proliferation but high cytokine production levels. Prior to treatment, individuals not excreting schistosome ova in the HT area had higher levels of proliferation to the stimulants, than the infected individuals. However, after treatment re-infected individuals showed high levels of proliferation. Before treatment, both infected and uninfected groups showed low and similar ratios, respectively, of IL-4:IFN-gamma, IL-5:IFN-gamma and IL-10:IFN-gamma, while IFN-gamma was high in the infected individuals. After treatment the non re-infected had higher levels of IL-4, IL-5 and IL-10, with the infected having high levels of IFN-gamma. Th1-like response dominated during infection with the Th2-like responses dominating post treatment and in uninfected individuals. The results indicated that the cytokine balance determines, in part, susceptibility or resistance to S. haematobium infection.

Neighbourhood control policies and the spread of infectious diseases

We present a model of a control programme for a disease outbreak in a population of livestock holdings. Control is achieved by culling infectious holdings when they are discovered and by the pre-emptive culling of livestock on holdings deemed to be at enhanced risk of infection. Because the pre-emptive control programme cannot directly identify exposed holdings, its implementation will result in the removal of both infected and uninfected holdings. This leads to a fundamental trade-off: increased levels of control produce a greater reduction in transmission by removing more exposed holdings, but increase the number of uninfected holdings culled. We derive an expression for the total number of holdings culled during the course of an outbreak and demonstrate that there is an optimal control policy, which minimizes this loss. Using a metapopulation model to incorporate local clustering of infection, we examine a neighbourhood control programme in a locally spreading outbreak. We find that there is an optimal level of control, which increases with increasing basic reproduction ratio, R(0); moreover, implementation of control may be optimal even when R(0) < 1. The total loss to the population is relatively insensitive to the level of control as it increases beyond the optimal level, suggesting that over-control is a safer policy than under-control.

The construction and analysis of epidemic trees with reference to the 2001 UK foot-and-mouth outbreak

The case-reproduction ratio for the spread of an infectious disease is a critically important concept for understanding dynamics of epidemics and for evaluating impact of control measures on spread of infection. Reliable estimation of this ratio is a problem central to epidemiology and is most often accomplished by fitting dynamic models to data and estimating combinations of parameters that equate to the case-reproduction ratio. Here, we develop a novel parameter-free method that permits direct estimation of the history of transmission events recoverable from detailed observation of a particular epidemic. From these reconstructed 'epidemic trees', case-reproduction ratios can be estimated directly. We develop a bootstrap algorithm that generates percentile intervals for these estimates that shows the procedure to be both precise and robust to possible uncertainties in the historical reconstruction. Identifying and 'pruning' branches from these trees whose occurrence might have been prevented by implementation of more stringent control measures permits estimation of the possible efficacy of these alternative measures. Examination of the cladistic structure of these trees as a function of the distance of each case from its infection source reveals useful insights about the relationship between long-distance transmission events and epidemic size. We demonstrate the utility of these methods by applying them to data from the 2001 foot-and-mouth disease outbreak in the UK.

Modelling vaccination strategies against foot-and-mouth disease

Vaccination has proved a powerful defence against a range of infectious diseases of humans and animals. However, its potential to control major epidemics of foot-and-mouth disease (FMD) in livestock is contentious. Using an individual farm-based model, we consider either national prophylactic vaccination campaigns in advance of an outbreak, or combinations of reactive vaccination and culling strategies during an epidemic. Consistent with standard epidemiological theory, mass prophylactic vaccination could reduce greatly the potential for a major epidemic, while the targeting of high-risk farms increases efficiency. Given sufficient resources and preparation, a combination of reactive vaccination and culling might control ongoing epidemics. We also explore a reactive strategy, 'predictive' vaccination, which targets key spatial transmission loci and can reduce markedly the long tail that characterizes many FMD epidemics. These analyses have broader implications for the control of human and livestock infectious diseases in heterogeneous spatial landscapes.

Dose-dependent responses of sheep inoculated intranasally with a type O foot-and-mouth disease virus

Unlike foot-and-mouth disease (FMD) in cattle and pigs, which spreads rapidly, resulting in easily detectable foci of clinical infection, the disease in sheep is characterized by restricted transmission, low morbidity and sporadic clinical cases. The study described was designed to investigate whether the ability of sheep to transmit and maintain FMD virus was dose-related. The viral isolate used was known to be associated epidemiologically with rapid fade-out of transmission within sheep flocks. Five separate transmission experiments were performed, with different doses of FMD virus, each experiment containing five intranasally inoculated donor sheep and 10 in-contact recipient sheep. The lowest dose required to cause clinical infection by inoculation (10(4) 50% tissue culture infectious doses; 10(4) TCID50) was also the optimum dose for producing in-contact transmission. Inoculation of donor sheep with higher doses (10(5) and 10(6) TCID50) resulted in reduced transmission, characterized by reduced duration and degree of viraemia and an early humoral and cell-mediated immune response. Principal component analysis was used to interpret the complex interactions of the dose-related responses to infection.

Comparative epidemiology of scrapie outbreaks in individual sheep flocks

Data recording the course of scrapie outbreaks in 4 sheep flocks (2 in Cheviot sheep and 2 in Suffolks) are compared. For each outbreak the data on scrapie incidence and sheep demography and pedigrees cover periods of years or decades. A key finding is that the incidence of clinical cases peaks in sheep 2-3 years old, despite very different forces-of-infection. This is consistent with age-specific susceptibility of sheep to scrapie, as has been reported for cattle to bovine spongiform encephalopathy and for humans to variant Creutzfeldt-Jakob disease. Scrapie incidence was higher in ewes than rams and at certain times of years, though these effects were not consistent between flocks. There was no evidence for high levels of vertical transmission.

Transmission patterns of African horse sickness and equine encephalosis viruses in South African donkeys

African horse sickness (AHS) and equine encephalosis (EE) viruses are endemic to southern Africa. AHS virus causes severe epidemics when introduced to naive equine populations, resulting in severe restrictions on the movement of equines between AHS-positive and negative countries. Recent zoning of South Africa has created an AHS-free zone to facilitate equine movement, but the transmission dynamics of these viruses are not fully understood. Here, we present further analyses of serosurveys of donkeys in South Africa conducted in 1983-5 and in 1993-5. Age-prevalence data are used to derive estimates of the force of infection, A. For both viruses, A was highest in the northeastern part of the country and declined towards the southwest. In most of the country, EE virus had a higher transmission rate than AHS. The force of infection increased for EE virus between 1985 and 1993, but decreased for AHS virus. Both viruses showed high levels of variation in transmission between districts within the same province, particularly in areas of intermediate transmission. These data emphasize the focal nature of these viruses, and indicate areas where further data will assist in understanding the geographical variation in transmission.

Just how unstable are agroecosystems?

Although it is widely believed that agroecosystems are less "stable" than natural ecosystems, the idea is only loosely based on empirical evidence. We compare the annual variability in population abundance for published data on temperate terrestrial arthropod communities from both natural and agricultural sites. Annual variability values are, as expected, significantly higher for agricultural communities. We briefly review proposed mechanisms for such an effect but conclude that none is sufficiently well tested to form the basis for any general ecological rule.