Dataset Information

Capture-Recapture Estimators in Epidemiology with Applications to Pertussis and Pneumococcal Invasive Disease Surveillance.

ABSTRACT:

Introduction

Surveillance networks are often not exhaustive nor completely complementary. In such situations, capture-recapture methods can be used for incidence estimation. The choice of estimator and their robustness with respect to the homogeneity and independence assumptions are however not well documented.

Methods

We investigated the performance of five different capture-recapture estimators in a simulation study. Eight different scenarios were used to detect and combine case-information. The scenarios increasingly violated assumptions of independence of samples and homogeneity of detection probabilities. Belgian datasets on invasive pneumococcal disease (IPD) and pertussis provided motivating examples.

Results

No estimator was unbiased in all scenarios. Performance of the parametric estimators depended on how much of the dependency and heterogeneity were correctly modelled. Model building was limited by parameter estimability, availability of additional information (e.g. covariates) and the possibilities inherent to the method. In the most complex scenario, methods that allowed for detection probabilities conditional on previous detections estimated the total population size within a 20-30% error-range. Parametric estimators remained stable if individual data sources lost up to 50% of their data. The investigated non-parametric methods were more susceptible to data loss and their performance was linked to the dependence between samples; overestimating in scenarios with little dependence, underestimating in others. Issues with parameter estimability made it impossible to model all suggested relations between samples for the IPD and pertussis datasets. For IPD, the estimates for the Belgian incidence for cases aged 50 years and older ranged from 44 to58/100,000 in 2010. The estimates for pertussis (all ages, Belgium, 2014) ranged from 24.2 to30.8/100,000.

Conclusion

We encourage the use of capture-recapture methods, but epidemiologists should preferably include datasets for which the underlying dependency structure is not too complex, a priori investigate this structure, compensate for it within the model and interpret the results with the remaining unmodelled heterogeneity in mind.

SUBMITTER: Braeye T

PROVIDER: S-EPMC4987016 | biostudies-literature | 2016

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Capture-Recapture Estimators in Epidemiology with Applications to Pertussis and Pneumococcal Invasive Disease Surveillance.

Braeye Toon T Verheagen Jan J Mignon Annick A Flipse Wim W Pierard Denis D Huygen Kris K Schirvel Carole C Hens Niel N

PloS one 20160816 8

<h4>Introduction</h4>Surveillance networks are often not exhaustive nor completely complementary. In such situations, capture-recapture methods can be used for incidence estimation. The choice of estimator and their robustness with respect to the homogeneity and independence assumptions are however not well documented.<h4>Methods</h4>We investigated the performance of five different capture-recapture estimators in a simulation study. Eight different scenarios were used to detect and combine case ...[more]

PMID: 27529167

Similar Datasets

Project description:BackgroundEpidemiology of Neisseria meningitidis has been changing since the introduction of universal vaccination programmes against meningococcal serogroup C (MenC) and meningococcal serogroup B (MenB) has now become dominant. This study aimed to analyse the cases reported in institutional data recording systems to estimate the burden of invasive meningococcal diseases (IMDs) and assess the effectiveness of surveillance in Veneto region (Italy).MethodsAnalysis was performed from 2007 to 2014 on data recorded in different systems: Mandatory Notification System, National Surveillance of Invasive Bacterial Diseases System and Laboratories Surveillance System (LSS), which were pooled into a combined surveillance system (CSS) and hospital discharge records (HDRs). A capture-recapture method was used and completeness of each source estimated. Number of cases with IMD by source of information and year, incidence of IMD by age group, case fatality rate (CFR) and distribution of meningococcal serogroups by year were also analysed.ResultsCombining the four data systems enabled the identification of 179 confirmed cases with IMD, achieving an overall sensitivity of 94.7% (95% CI: 90.8% to 98.8%), while it was 76.7% (95% CI: 73.6% to 80.1%) for CSS and 77.2% (95% CI: 74.1% to 80.6%) for HDRs. Typing of isolates was done in 80% of cases, and 95.2% of the typed cases were provided by LSS. Serogroup B was confirmed in 50.3% of cases. The estimated IMD notification rate (cases with IMD diagnosed and reported to the surveillance systems) was 0.48/100 000 population, and incidence peaked at 6.2/100 000 in children aged <1?year old (60.9% due to MenB), and increased slightly in the age group between 15 and 19 years (1.1/100 000). A CFR of 14% was recorded (8.7% in paediatric age).ConclusionsQuality of surveillance systems relies on case ascertainment based on serological characterisation of the circulating strains by microbiology laboratories. All available sources should be routinely combined to improve the epidemiology of IMD and the information used by public health departments to conduct timely preventive measures.

Project description:BackgroundVaccine-serotype (VT) invasive pneumococcal disease (IPD) rates declined substantially following introduction of 7-valent pneumococcal conjugate vaccine (PCV7) into national immunization programs. Increases in non-vaccine-serotype (NVT) IPD rates occurred in some sites, presumably representing serotype replacement. We used a standardized approach to describe serotype-specific IPD changes among multiple sites after PCV7 introduction.Methods and findingsOf 32 IPD surveillance datasets received, we identified 21 eligible databases with rate data ? 2 years before and ? 1 year after PCV7 introduction. Expected annual rates of IPD absent PCV7 introduction were estimated by extrapolation using either Poisson regression modeling of pre-PCV7 rates or averaging pre-PCV7 rates. To estimate whether changes in rates had occurred following PCV7 introduction, we calculated site specific rate ratios by dividing observed by expected IPD rates for each post-PCV7 year. We calculated summary rate ratios (RRs) using random effects meta-analysis. For children <5 years old, overall IPD decreased by year 1 post-PCV7 (RR 0.55, 95% CI 0.46-0.65) and remained relatively stable through year 7 (RR 0.49, 95% CI 0.35-0.68). Point estimates for VT IPD decreased annually through year 7 (RR 0.03, 95% CI 0.01-0.10), while NVT IPD increased (year 7 RR 2.81, 95% CI 2.12-3.71). Among adults, decreases in overall IPD also occurred but were smaller and more variable by site than among children. At year 7 after introduction, significant reductions were observed (18-49 year-olds [RR 0.52, 95% CI 0.29-0.91], 50-64 year-olds [RR 0.84, 95% CI 0.77-0.93], and ? 65 year-olds [RR 0.74, 95% CI 0.58-0.95]).ConclusionsConsistent and significant decreases in both overall and VT IPD in children occurred quickly and were sustained for 7 years after PCV7 introduction, supporting use of PCVs. Increases in NVT IPD occurred in most sites, with variable magnitude. These findings may not represent the experience in low-income countries or the effects after introduction of higher valency PCVs. High-quality, population-based surveillance of serotype-specific IPD rates is needed to monitor vaccine impact as more countries, including low-income countries, introduce PCVs and as higher valency PCVs are used. Please see later in the article for the Editors' Summary.

Project description:BackgroundTen-valent pneumococcal conjugate vaccine (PCV10), delivered at 6, 10, and 14 weeks of age was introduced in Kenya in January, 2011, accompanied by a catch-up campaign in Kilifi County for children aged younger than 5 years. Coverage with at least two PCV10 doses in children aged 2-11 months was 80% in 2011 and 84% in 2016; coverage with at least one dose in children aged 12-59 months was 66% in 2011 and 87% in 2016. We aimed to assess PCV10 effect against nasopharyngeal carriage and invasive pneumococcal disease (IPD) in children and adults in Kilifi County.MethodsThis study was done at the KEMRI-Wellcome Trust Research Programme among residents of the Kilifi Health and Demographic Surveillance System, a rural community on the Kenyan coast covering an area of 891 km2. We linked clinical and microbiological surveillance for IPD among admissions of all ages at Kilifi County Hospital, Kenya, which serves the community, to the Kilifi Health and Demographic Surveillance System from 1999 to 2016. We calculated the incidence rate ratio (IRR) comparing the prevaccine (Jan 1, 1999-Dec 31, 2010) and postvaccine (Jan 1, 2012-Dec 31, 2016) eras, adjusted for confounding, and reported percentage reduction in IPD as 1 minus IRR. Annual cross-sectional surveys of nasopharyngeal carriage were done from 2009 to 2016.FindingsSurveillance identified 667 cases of IPD in 3 211 403 person-years of observation. Yearly IPD incidence in children younger than 5 years reduced sharply in 2011 following vaccine introduction and remained low (PCV10-type IPD: 60·8 cases per 100 000 in the prevaccine era vs 3·2 per 100 000 in the postvaccine era [adjusted IRR 0·08, 95% CI 0·03-0·22]; IPD caused by any serotype: 81·6 per 100 000 vs 15·3 per 100 000 [0·32, 0·17-0·60]). PCV10-type IPD also declined in the post-vaccination era in unvaccinated age groups (<2 months [no cases in the postvaccine era], 5-14 years [adjusted IRR 0·26, 95% CI 0·11-0·59], and ≥15 years [0·19, 0·07-0·51]). Incidence of non-PCV10-type IPD did not differ between eras. In children younger than 5 years, PCV10-type carriage declined between eras (age-standardised adjusted prevalence ratio 0·26, 95% CI 0·19-0·35) and non-PCV10-type carriage increased (1·71, 1·47-1·99).InterpretationIntroduction of PCV10 in Kenya, accompanied by a catch-up campaign, resulted in a substantial reduction in PCV10-type IPD in children and adults without significant replacement disease. Although the catch-up campaign is likely to have brought forward the benefits by several years, the study suggests that routine infant PCV10 immunisation programmes will provide substantial direct and indirect protection in low-income settings in tropical Africa.FundingGavi, The Vaccine Alliance and The Wellcome Trust of Great Britain.

Project description:BackgroundLittle information is available about the effect of pneumococcal conjugate vaccines (PCVs) in low-income countries. We measured the effect of these vaccines on invasive pneumococcal disease in The Gambia where the 7-valent vaccine (PCV7) was introduced in August, 2009, followed by the 13-valent vaccine (PCV13) in May, 2011.MethodsWe conducted population-based surveillance for invasive pneumococcal disease in individuals aged 2 months and older who were residents of the Basse Health and Demographic Surveillance System (BHDSS) in the Upper River Region, The Gambia, using standardised criteria to identify and investigate patients. Surveillance was done between May, 2008, and December, 2014. We compared the incidence of invasive pneumococcal disease between baseline (May 12, 2008-May 11, 2010) and after the introduction of PCV13 (Jan 1, 2013-Dec 31, 2014), adjusting for changes in case ascertainment over time.FindingsWe investigated 14 650 patients, in whom we identified 320 cases of invasive pneumococcal disease. Compared with baseline, after the introduction of the PCV programme, the incidence of invasive pneumococcal disease decreased by 55% (95% CI 30-71) in the 2-23 months age group, from 253 to 113 per 100 000 population. This decrease was due to an 82% (95% CI 64-91) reduction in serotypes covered by the PCV13 vaccine. In the 2-4 years age group, the incidence of invasive pneumococcal disease decreased by 56% (95% CI 25-75), from 113 to 49 cases per 100 000, with a 68% (95% CI 39-83) reduction in PCV13 serotypes. The incidence of non-PCV13 serotypes in children aged 2-59 months increased by 47% (-21 to 275) from 28 to 41 per 100 000, with a broad range of serotypes. The incidence of non-pneumococcal bacteraemia varied little over time.InterpretationThe Gambian PCV programme reduced the incidence of invasive pneumococcal disease in children aged 2-59 months by around 55%. Further surveillance is needed to ascertain the maximum effect of the vaccine in the 2-4 years and older age groups, and to monitor serotype replacement. Low-income and middle-income countries that introduce PCV13 can expect substantial reductions in invasive pneumococcal disease.FundingGAVI's Pneumococcal vaccines Accelerated Development and Introduction Plan (PneumoADIP), Bill & Melinda Gates Foundation, and the UK Medical Research Council.

Dataset Information

Capture-Recapture Estimators in Epidemiology with Applications to Pertussis and Pneumococcal Invasive Disease Surveillance.

Introduction

Methods

Results

Conclusion

Publications

Capture-Recapture Estimators in Epidemiology with Applications to Pertussis and Pneumococcal Invasive Disease Surveillance.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure