Alex Potapov

Wildlife disease elimination and density dependence.

Disease control by managers is a crucial response to emerging wildlife epidemics, yet the means of control may be limited by the method of disease transmission. In particular, it is widely held that population reduction, while effective for controlling diseases that are subject to density-dependent (DD) transmission, is ineffective for controlling diseases that are subject to frequency-dependent (FD) transmission. We investigate control for horizontally transmitted diseases with FD transmission where the control is via culling or harvest that is non-selective with respect to infection and the population can compensate through DD recruitment or survival. Using a mathematical model, we show that culling or harvesting can eradicate the disease, even when transmission dynamics are FD. Eradication can be achieved under FD transmission when DD birth or recruitment induces compensatory growth of new, healthy individuals, which has the net effect of reducing disease prevalence by dilution. We also show that if harvest is used simultaneously with vaccination, and there is high enough transmission coefficient, application of both controls may be less efficient than vaccination alone. We illustrate the effects of these control approaches on disease prevalence for chronic wasting disease in deer where the disease is transmitted directly among deer and through the environment.

Quality Indicators of End-of-Life Care in Patients With Cancer: What Rate Is Right?

PURPOSE To develop data-driven and achievable benchmark rates for end-of-life quality indicators using administrative data from four provinces in Canada. METHODS Indicators of end-of-life care were defined and measured using linked administrative data for 33 health regions across British Columbia, Alberta, Ontario, and Nova Scotia. These were emergency department use, intensive care unit admission, physician house calls and home care visits before death, and death in hospital. An empiric benchmark was defined using indicator rates from the top-ranked regions to include the top decile of patients overall. Funnel plots were used to graph each regions age- and sex-adjusted indicator rates along with the overall rate and 95% confidence limits. RESULTS Rates varied approximately two- to four-fold across the regions, with physician house calls showing the greatest variation. Benchmark rates based on the top decile performers were emergency department use, 34%; intensive care unit admission, 2%; physician house calls, 34%; home care visits, 63%; and death in hospital, 38%. With the exception of intensive care unit admission, funnel plots demonstrated that overall indicator rates and their confidence limits were uniformly worse than benchmarks even after adjusting for age and sex. Few regions met the benchmark rates. CONCLUSION There is significant variation in end-of-life quality indicators across regions in four provinces in Canada. Using this studys methods-deriving empiric benchmarks and funnel plots-regions can determine their relative performance with greater context that facilitates priority setting and resource deployment. Applying this studys methods can support quality improvement by decreasing variation and striving for a target.

Allee threshold and stochasticity in biological invasions: Colonization time at low propagule pressure

We consider the problem of estimating the time needed for species colonization. The analysis is based upon the known population dynamic models by Dennis with minor modification to the Allee effect description, which allows us to obtain an analytical expression for the colonization time. For the stochastic counterpart of the models in diffusion approximation, we (1) propose the description of immigration stochasticity, (2) provide the estimates of time required for the population to overcome strong demographic Allee effect, and (3) consider the numerical results for mean colonization time and its uncertainty. Strong Allee effect strictly disallows populations at small immigration rates to colonize new habitats, unless the stochasticity in immigration, environment, or demography is present, or incorporated into the model. Immigration stochasticity, complementing with environmental and demographic stochasticity, enables the populations to overcome the Allee threshold even at low values of propagule pressure.

Models of lake invasibility by Bythotrephes longimanus, a non-indigenous zooplankton

We built a family of hierarchical risk models for the spread of invasions by the spiny waterflea (Bythotrephes longimanus) in lakes in Ontario, Canada. Knowledge of covariates determining lake invasibility and ability to predict risk of future invasions may help to develop management policy and slow the invasions in the future. The models are based on two component submodels. The first component was a stochastic gravity submodel for the propagule pressure between lakes via recreational boaters. The second component was a submodel for establishment risk, given that the invader has already been introduced to a lake. This component was a logistic regression model, incorporating up to 17 measured covariates that describe the physical and chemical condition of the lake. Variants of the risk model, each incorporating different subsets of the covariates, were calibrated using presence/absence data from a 300-lake survey conducted in 2005–2006 by the Canadian Aquatic Invasive Species Network (CAISN). The predictive capacity of the best model was high, giving AUC values close to 0.94. Of the model covariates considered, the most important predictors of existing invasions were propagule pressure and lake pH, and, to lesser extents, phosphorus (P) and lake elevation. Our fitting of the propagule pressure submodel demonstrated a significant Allee effect for Bythotrephes. Our development of the establishment risk predictor showed that it is essential to account for temporal variability in lake physico-chemistry. We demonstrated that invasions of lake networks by the spiny waterflea follow highly predictable patterns which can be understood with a properly calibrated, hierarchical risk model.

Chronic Wasting Disease: Transmission Mechanisms and the Possibility of Harvest Management

We develop a model of CWD management by nonselective deer harvest, currently the most feasible approach available for managing CWD in wild populations. We use the model to explore the effects of 6 common harvest strategies on disease prevalence and to identify potential optimal harvest policies for reducing disease prevalence without population collapse. The model includes 4 deer categories (juveniles, adult females, younger adult males, older adult males) that may be harvested at different rates, a food-based carrying capacity, which influences juvenile survival but not adult reproduction or survival, and seasonal force of infection terms for each deer category under differing frequency-dependent transmission dynamics resulting from environmental and direct contact mechanisms. Numerical experiments show that the interval of transmission coefficients β where the disease can be controlled is generally narrow and efficiency of a harvest policy to reduce disease prevalence depends crucially on the details of the disease transmission mechanism, in particular on the intensity of disease transmission to juveniles and the potential differences in the behavior of older and younger males that influence contact rates. Optimal harvest policy to minimize disease prevalence for each of the assumed transmission mechanisms is shown to depend on harvest intensity. Across mechanisms, a harvest that focuses on antlered deer, without distinguishing between age classes reduces disease prevalence most consistently, whereas distinguishing between young and older antlered deer produces higher uncertainty in the harvest effects on disease prevalence. Our results show that, despite uncertainties, a modelling approach can determine classes of harvest strategy that are most likely to be effective in combatting CWD.

Quality of end-of-life cancer care in Canada: a retrospective four-province study using administrative health care data

BACKGROUND The quality of data comparing care at the end of life (eol) in cancer patients across Canada is poor. This project used identical cohorts and definitions to evaluate quality indicators for eol care in British Columbia, Alberta, Ontario, and Nova Scotia. METHODS This retrospective cohort study of cancer decedents during fiscal years 2004-2009 used administrative health care data to examine health service quality indicators commonly used and previously identified as important to quality eol care: emergency department use, hospitalizations, intensive care unit admissions, chemotherapy, physician house calls, and home care visits near the eol, as well as death in hospital. Crude and standardized rates were calculated. In each province, two separate multivariable logistic regression models examined factors associated with receiving aggressive or supportive care. RESULTS Overall, among the identified 200,285 cancer patients who died of their disease, 54% died in a hospital, with British Columbia having the lowest standardized rate of such deaths (50.2%). Emergency department use at eol ranged from 30.7% in Nova Scotia to 47.9% in Ontario. Of all patients, 8.7% received aggressive care (similar across all provinces), and 46.3% received supportive care (range: 41.2% in Nova Scotia to 61.8% in British Columbia). Lower neighbourhood income was consistently associated with a decreased likelihood of supportive care receipt. INTERPRETATION We successfully used administrative health care data from four Canadian provinces to create identical cohorts with commonly defined indicators. This work is an important step toward maturing the field of eol care in Canada. Future work in this arena would be facilitated by national-level data-sharing arrangements.

Empirical Estimation of R0 for Unknown Transmission Functions: The Case of Chronic Wasting Disease in Alberta

We consider the problem of estimating the basic reproduction number R 0 from data on prevalence dynamics at the beginning of a disease outbreak. We derive discrete and continuous time models, some coefficients of which are to be fitted from data. We show that prevalence of the disease is sufficient to determine R 0. We apply this method to chronic wasting disease spread in Alberta determining a range of possible R 0 and their sensitivity to the probability of deer annual survival.