Estimating the Relative Probability of Direct Transmission between Infectious Disease Patients

Van Ness, S. E.; Lee, R. S.; Sebastiani, P.; Horsburgh, C. R.; Jenkins, H. E.; White, L. F.

doi:10.1101/612945

Home/Archive/bioRxiv

bioRxiv · 10.1101/612945

Estimating the Relative Probability of Direct Transmission between Infectious Disease Patients

Van Ness, S. E.·Lee, R. S.·Sebastiani, P.·Horsburgh, C. R.·Jenkins, H. E.·White, L. F.

View original source ↗Export metadata ↓

Abstract

BackgroundEstimating infectious disease parameters such as the serial interval (time between symptom onset in primary and secondary cases) and reproductive number (average number of secondary cases produced by a primary case) are important to understand infectious disease dynamics. Many estimation methods require linking cases by direct transmission, a difficult task for most diseases. MethodsUsing a subset of cases with detailed genetic or contact investigation data to develop a training set of probable transmission events, we build a model to estimate the relative transmission probability for all case-pairs from demographic, spatial and clinical data. Our method is based on naive Bayes, a machine learning classification algorithm which uses the observed frequencies in the training dataset to estimate the probability that a pair is linked given a set of covariates. ResultsIn simulations we find that the probabilities estimated using genetic distance between cases to define training transmission events are able to distinguish between truly linked and unlinked pairs with high accuracy (area under the receiver operating curve value of 95%). Additionally only a subset of the cases, 10-50% depending on sample size, need to have detailed genetic data for our method to perform well. We show how these probabilities can be used to estimate the average effective reproductive number and apply our method to a tuberculosis outbreak in Hamburg, Germany. ConclusionsOur method is a novel way to infer transmission dynamics in any dataset when only a subset of cases has rich contact investigation and/or genetic data. KEY MESSAGESO_LIThis method provides a way to calculate the relative probability that two infectious disease patients are connected by direct transmission using clinical, demographic, geographic, and genetic characteristics. C_LIO_LIWe use a naive Bayes, a machine learning technique to estimate these probabilities using a training set of probable links defined by contact investigation or pathogen WGS data on a subset of cases. C_LIO_LIThese probabilities can be used to explore possible transmission chains, rule out transmission events, and estimate the reproductive number. C_LI

Source connections

Published article ↗

Explore related subjects

epidemiology Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Van Ness, S. E., Lee, R. S., Sebastiani, P., Horsburgh, C. R., Jenkins, H. E., White, L. F.. 2019-04-26. Estimating the Relative Probability of Direct Transmission between Infectious Disease Patients. https://doi.org/10.1101/612945

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2019

Rainfall and other meteorological factors as drivers of urban transmission of leptospirosis

BackgroundLeptospirosis is an important public health problem affecting vulnerable urban slum populations in developing country settings. However, the complex interaction of meteorological factors driving the temporal trends of leptospirosis remain incompletely understood.\n\nMethods and findingsFrom 1996 to 2010, we investigated the association between the weekly incidence of leptospirosis and climatic variables in the city of Salvador, Brazil by using a dynamic generalized linear model that accounted for time lags, overall trend and seasonal variation. Our model showed an increase of leptospirosis cases associated with rainfall, lower temperature and higher humidity. There was a lag of one-to-two weeks between weekly values for significant meteorological variables and leptospirosis incidence. Independent of the season, a weekly cumulative rainfall of 20 mm increased the risk of the leptospirosis by 10% compared to a week without rain. Finally, over the 14 year study period the incidence of leptospirosis decreased significantly by four fold (12.8 versus 3.6 per 100,000 people), independently of variations in climate.\n\nConclusionsStrategies to control leptospirosis should focus on avoiding contact with contaminated sources of Leptospira as well as on increasing awareness in the population and health professionals within the short time window after both low-level and extreme rainfall events. Increased leptospirosis incidence was restricted to one-to-two weeks after those events suggesting that infectious Leptospira survival may be limited to short time intervals.\n\nAuthor SummaryTo determine the role of meteorological variables, seasonal variation and temporal trends in the incidence of leptospirosis, we investigated the time series of leptospirosis incidence amongst residents of Salvador, Brazil, from 1996 to 2010. Exploratory and confirmatory statistical methods detected associations between meteorological factors and disease incidence. Results showed the importance of extreme meteorological conditions, particularly rainfall, as short-term predictors of leptospirosis incidence. In addition, we found a long-term decreasing trend of in disease incidence over the observation period.

epidemiology↗

bioRxiv2019

Cost utility analysis of end stage renal disease treatment in Ministry of Health dialysis centres, Malaysia: hemodialysis versus continuous ambulatory peritoneal dialysis

OBJECTIVESIn Malaysia, there is exponential growth of patients on dialysis. Dialysis treatment consumes a considerable portion of healthcare expenditure. Comparative assessment of their cost effectiveness can assist in providing a rational basis for preference of dialysis modalities.\n\nMETHODSA cost utility study of hemodialysis (HD) and continuous ambulatory peritoneal dialysis (CAPD) was conducted from a Ministry of Health (MOH) perspective. A Markov model was also developed to investigate the cost effectiveness of increasing uptake of CAPD to 55% and 60 % versus current practice of 40% CAPD in a five-year temporal horizon. A scenario with 30% CAPD was also measured. The costs and utilities were sourced from published data which were collected as part of this study. The transitional probabilities and survival estimates were obtained from the Malaysia Dialysis and Transplant Registry (MDTR). The outcome measures were cost per life year (LY), cost per quality adjusted LY (QALY) and incremental cost effectiveness ratio (ICER) for the Markov model. Sensitivity analyses were performed.\n\nRESULTSLYs saved for HD was 4.15 years and 3.70 years for CAPD. QALYs saved for HD was 3.544 years and 3.348 for CAPD. Cost per LY saved was RM39,791 for HD and RM37,576 for CAPD. The cost per QALY gained was RM46,595 for HD and RM41,527 for CAPD. The Markov model showed commencement of CAPD in 50% of ESRD patients as initial dialysis modality was very cost-effective versus current practice of 40% within MOH. Reduction in CAPD use was associated with higher costs and a small devaluation in QALYs.\n\nCONCLUSIONSThese findings suggest provision of both modalities is fiscally feasible; increasing CAPD as initial dialysis modality would be more cost-effective.

epidemiology↗

bioRxiv2019

Social, demographic, health care and co-morbidity predictors of tuberculosis mortality in Amazonas, Brazil: a multiple cause of death approach

OBJECTIVESTo estimate TB mortality rates, describe multiple causes in death certificates in which TB was reported and identify predictors of TB reporting in death certificates in the State of Amazonas, Brazil, based on a multiple cause of death approach.\n\nMETHODSDeath records of residents in AM within 2006-2014 were classified based on tuberculosis reporting in the death certificate as tuberculosis not reported (TBNoR), reported as the underlying cause of death (TBUC) and as an associate cause of death (TBAC). Age standardized annual mortality rates for TBUC, TBAC and with TB reported (TBUC plus TBAC) were estimated for the State of Amazonas, using the direct standardization method and WHO 2000-2025 standard population. Mortality odds ratios (OR) of reporting TBUC and TBAC were estimated using multinomial logistic regression.\n\nRESULTSAge standardized annual TBUC and TBAC mortality rates ranged, between 5.9-7.8/105 and 2.7-4.0/105, respectively. TBUC was associated with residence in the State capital (OR=0.66), female sex (OR=0.87), education level (OR=0.67 and 0.50 for 8 to 11 and 12 or more school years), non-white race/skin colour (OR=1.38) and occurrence of death in the State capital (OR=1.69). TBAC was related to time (OR=1.21 and 1.22 for years 2009-11 and 2012-14), age (OR=36.1 and 16.5 for ages 15-39 and 40-64 years) and when death occurred in the State capital (OR=5.8).\n\nCONCLUSIONTBUC was predominantly associated with indicators of unfavorable socioeconomic conditions and health care access constraints, whereas TBAC was mainly related to ages typical of high HIV disease incidence.\n\nConflicts of interestNone.\n\nFundingFundacao de Amparo a Pesquisa do Estado do Amazonas - FAPEAM

epidemiology↗