Efficiency and accuracy of disease monitoring systems: application of capture-recapture methods to injury monitoring

Gaining further insights into the COVID-19 pandemic in Australia: Evidence using capture-recapture methods

Objective

We re-examined the reported number of COVID-19 cases in Australia and across its states during the first wave of the pandemic. We provided estimates of the total number of cases, adjusted for under-reporting.

Methods

Publicly available data sourced from Australian governments at federal, state and territory levels included records on cumulative confirmed COVID-19 cases and cumulative deaths occurring in Australia and across its states on a daily basis. Lower bound and upper bound estimates of the total number of COVID-19 cases in Australia and across its states, that included the undetected cases that have not been recorded, were estimated.

Results

On January 25, 2020, Australia recorded its first 4 cases of COVID-19 and the first death occurred on March 3, 2020. On April 1, 2020, 4864 cases had been reported with 21 deaths. Our estimation showed that on April 1, 2020, the minimum and maximum number of COVID-19 cases in Australia were in fact 10,160 (95 % CI: 9781-10,538) and 21,748 (95 % CI: 21,607-22,014) respectively. We estimated that the total number of cases were at least twice and at most four times the observed cases recorded. These differences were also found at the state level where in New South Wales there was a minimum and maximum of 207 and 447 cases in total for every 100 reported cases, while in Victoria these figures were much lower at 157 and 265 respectively for every 100 reported cases.

Conclusion

Case ascertainment during the pandemic is known to have been underestimated due to difficulties in testing and contact tracing, amongst others. Capture-recapture methods provided a measure of the gap between the official number of cases recorded and the actual number during the first wave of the pandemic.

Estimation of the incidence of invasive meningococcal disease using a capture-recapture model based on two independent surveillance systems in Catalonia, Spain

OBJECTIVES

Invasive meningococcal disease (IMD) is an urgent notifiable disease and its early notification is essential to prevent cases. The objective of the study was to assess the sensitivity of two independent surveillance systems and to estimate the incidence of IMD.

DESIGN

We used capture-recapture model based on two independent surveillance systems, the statutory disease reporting (SDR) system and the microbiological reporting system (MRS) of the Public Health Agency of Catalonia, between 2011 and 2015. The capture-recapture analysis and 95% CIs were calculated using the Chapman formula. Multivariate vector generalised linear model was performed for adjusted estimation.

MEASURES

The variables collected were age, sex, year of report, size of municipality (<10 000 and ≥10 000), clinical form, death, serogroup, country of birth and type of reporting centre (private and public).

RESULTS

The sensitivity of the two combined surveillance systems was 88.5% (85.0-92.0). SDR had greater sensitivity than the MRS (67.9%; 62.7-73.1 vs 64.7%; 59.4-70.0). In 2014-2015, the sensitivity of both systems was higher (80.6%; 73.2-87.9 vs 73.4%; 65.2-81.6) than in 2011-2013 (59.3%; 52.6-66.0 vs 58.3%; 51.6-65.1). In private centres, the sensitivity was higher for SDR than for MRS (100%; 100-100 vs 4.8%; -4.4-13.9). The adjusted estimate of IMD cases was lower than that obtained using the Chapman formula (279; 266-296 vs 313; 295-330). The estimated adjusted incidence of IMD was 0.7/100 000 persons-year.

CONCLUSIONS

The sensitivity of enhanced surveillance through the combination of two complementary sources was higher than for the sources individually. Factors associated with under-reporting in different systems should be analysed to improve IMD surveillance.

Hepatitis A surveillance: sensitivity of two information sources

BACKGROUND

The frequency of mild forms of hepatitis A, especially in children, could lead to underreporting. The objective of the study was to investigate the sensitivity of two surveillance systems, mandatory Statutory Disease Reports and the Microbiological Reporting System of Catalonia, using capture-recapture techniques.

METHODS

The study was conducted in Catalonia between 2011 and 2015. Hepatitis A cases reported to two independent surveillance systems were included: Statutory Disease Reports (SDR) and Microbiological Reporting System of Catalonia (MRS). The variables collected were: age, sex, year of declaration, size of municipality (< 10,000 and ≥ 10,000), country of birth (Spain or abroad), reporting centre (primary care/hospital) and notification method (electronic or paper). The capture-recapture analysis and the estimate of 95% confidence intervals were made using the Chapman formula for comparison of two sources, both for the estimate of the total number of cases and the stratification according to variables. Multinomial logistic regression was performed to obtain an adjusted estimate.

RESULTS

The SDR had a greater overall sensitivity than the MRS (48.8%; 43.5-55.6 vs. 19.3%; 17.2-21.9). In cases aged < 15 years the sensitivity of both systems was higher (76.6%; 72.7-81 vs. 25.2%; 20.9-29.5) than in cases aged > 15 years (25.5%; 22.8-28.3 vs. 12.1%; 10-14.2). For those born in Spain, the sensitivity was 57.2% (49.6-67.4) in the SDR and 27.1% (23.5-31.9) in the MRS, lower than that for foreign-born patients (58%; 51.2-66.8 vs. 49.1%; 43.4-56.6). In electronically-reported cases, the sensitivity was much higher in the SDR than in the MRS (47.2%; 42.3-52.1 vs. 9.4%; 6.5-12.3). No differences were observed according to sex, size of municipality, and year of declaration or reporting centre. The estimated total number of cases using the Chapman formula was very similar to the adjusted estimate (1121; 985-1258 vs. 1120; 876-1525), indicating the robustness of the results.

CONCLUSIONS

The sensitivity of the SDR was greater than that of MRS, especially in patients aged < 15 years, although for patients born abroad the difference in sensitivity was lower. Reinforced surveillance combining the SDR and MRS improves the efficiency in the detection of cases.

Hazardous waste site frequency: use of the capture-recapture method

This investigation used a two-source capture-recapture method (CRM) for determining ascertainment and undercounts of non-national priority listed hazardous waste sites in the states of Arizona, Maine and Pennsylvania. These findings suggest that ascertainment of hazardous waste sites vary greatly, with some more accurate (i.e., Maine) than others (i.e., Pennsylvania). These data suggest that nontraditional manufacturing states (e.g., Maine) have a higher ascertainment rate than traditional manufacturing states (e.g., Pennsylvania). These results indicate that resources for locating hazardous waste sites should be more heavily allotted to industrialized areas. We suggest that the CRM is a convenient, low cost and effective method for determining (1) the accuracy of previous estimates, and (2) the number of sites in a locale with 95% confidence intervals along with an estimate of the undercount. Findings suggest that estimates of hazardous waste sites should use the CRM to determine and improve accuracy.

Identification of hidden key hepatitis C populations: an evaluation of screening practices using mixed epidemiological methods

BACKGROUND

Hepatitis C virus (HCV) is a major cause of liver diseases worldwide. Due to its asymptomatic nature, screening is necessary for identification. Because screening of the total population is not cost effective, it is important to identify which risk factors for positivity characterize the key populations in which targeting of screening yields the highest numbers of HCV positives, and assess which of these key populations have remained hidden to current care.

METHODS

Laboratory registry data (2002-2008) were retrieved for all HCV tests (23,800) in the south of the Netherlands (adult population 500,000). Screening trends were tested using Poisson regression and chi-square tests. Risk factors for HCV positivity were assessed using a logistic regression. The hidden HCV-positive population was estimated by a capture-recapture approach.

RESULTS

The number of tests increased over time (2,388 to 4,149, p<.01). Nevertheless, the positivity rate among those screened decreased between 2002 and 2008 (6.3% to 2.1%, p<.01). The population prevalence was estimated to be 0.49% (95%CI 0.41-0.59). Of all HCV-positive patients, 66% were hidden to current screening practices. Risk factors associated with positivity were low socio-economic status, male sex, and age between 36-55. In future screening 48% (95%CI 37-63) of total patients and 47% (95%CI 32-70) of hidden patients can be identified by targeting 9% (men with low socio-economic status, between 36-55 years old) of the total population.

CONCLUSIONS

Although the current HCV screening policy increasingly addresses high-risk populations, it only reaches one third of positive patients. This study shows that combining easily identifiable demographic risk factors can be used to identify key populations as a likely target for effective HCV screening. We recommend strengthening screening among middle-aged man, living in low socio-economic neighborhoods.

Under-reporting of road traffic mortality in developing countries: application of a capture-recapture statistical model to refine mortality estimates

Road traffic injuries are a major cause of preventable death in sub-Saharan Africa. Accurate epidemiologic data are scarce and under-reporting from primary data sources is common. Our objectives were to estimate the incidence of road traffic deaths in Malawi using capture-recapture statistical analysis and determine what future efforts will best improve upon this estimate. Our capture-recapture model combined primary data from both police and hospital-based registries over a one year period (July 2008 to June 2009). The mortality incidences from the primary data sources were 0.075 and 0.051 deaths/1000 person-years, respectively. Using capture-recapture analysis, the combined incidence of road traffic deaths ranged 0.192-0.209 deaths/1000 person-years. Additionally, police data were more likely to include victims who were male, drivers or pedestrians, and victims from incidents with greater than one vehicle involved. We concluded that capture-recapture analysis is a good tool to estimate the incidence of road traffic deaths, and that capture-recapture analysis overcomes limitations of incomplete data sources. The World Health Organization estimated incidence of road traffic deaths for Malawi utilizing a binomial regression model and survey data and found a similar estimate despite strikingly different methods, suggesting both approaches are valid. Further research should seek to improve capture-recapture data through utilization of more than two data sources and improving accuracy of matches by minimizing missing data, application of geographic information systems, and use of names and civil registration numbers if available.

Identifying work-related motor vehicle crashes in multiple databases

OBJECTIVE

To compare and estimate the magnitude of work-related motor vehicle crashes in Utah using 2 probabilistically linked statewide databases.

METHODS

Data from 2006 and 2007 motor vehicle crash and hospital databases were joined through probabilistic linkage. Summary statistics and capture-recapture were used to describe occupants injured in work-related motor vehicle crashes and estimate the size of this population.

RESULTS

There were 1597 occupants in the motor vehicle crash database and 1673 patients in the hospital database identified as being in a work-related motor vehicle crash. We identified 1443 occupants with at least one record from either the motor vehicle crash or hospital database indicating work-relatedness that linked to any record in the opposing database. We found that 38.7 percent of occupants injured in work-related motor vehicle crashes identified in the motor vehicle crash database did not have a primary payer code of workers' compensation in the hospital database and 40.0 percent of patients injured in work-related motor vehicle crashes identified in the hospital database did not meet our definition of a work-related motor vehicle crash in the motor vehicle crash database. Depending on how occupants injured in work-related motor crashes are identified, we estimate the population to be between 1852 and 8492 in Utah for the years 2006 and 2007.

CONCLUSIONS

Research on single databases may lead to biased interpretations of work-related motor vehicle crashes. Combining 2 population based databases may still result in an underestimate of the magnitude of work-related motor vehicle crashes. Improved coding of work-related incidents is needed in current databases.

Under-recognition and reporting of dengue in Cambodia: a capture–recapture analysis of the National Dengue Surveillance System

Robust disease burden estimates are important for decision-making concerning introduction of new vaccines. Dengue is a major public health problem in the tropics but robust disease burden estimates are lacking. We conducted a two-sample, capture–recapture study in the largest province in Cambodia to determine disease under-recognition to the National Dengue Surveillance System (NDSS). During 2006–2008, community-based active surveillance for acute febrile illness was conducted in 0- to 19-year-olds in rural and urban areas combined with testing for dengue virus infection. Of 14 354 individuals under active surveillance (22 498 person-seasons), the annual incidence ranged from 13·4 to 57·8/1000 person-seasons. During the same period, NDSS incidence rates ranged from 1·1/1000 to 5·7/1000, which was 3·9- to 29·0-fold lower than found in the capture–recapture study. In hospitalized cases, the rate of under-recognition was 1·1- to 2·4-fold. This study shows the substantial degree of under-recognition/reporting of dengue and that reported hospitalized cases are not a good surrogate for estimating dengue disease burden.

Using capture-recapture methods to better ascertain the incidence of fatal child maltreatment

OBJECTIVES

To (1) test the use of capture-recapture methods to estimate the total number of child maltreatment deaths in a single state using information from death certificates, child welfare reports, child death review teams, and uniform crime reports; and to (2) compare these estimates to the number of maltreatment deaths identified through an in-depth "gold standard" review.

METHODS

Child maltreatment deaths were identified in four existing administrative data sources: (1) death reports in our state vital statistics (DC); (2) child death review team reports (CDR); (3) homicide reports filed by our state police agency as uniform crime report (UCR) supplements for the FBI; and (4) abstracted reports of a minor's death from our state child protective services (CPS) agency. Capture-recapture pair-wise and pooled comparisons were then applied to estimate the numbers of abuse and total maltreatment deaths and were compared to the number of cases identified by independent case review.

RESULTS

There were a total of 194 child maltreatment deaths in Michigan during 2000-2001 with 66 due to physical abuse. Capture-recapture analysis estimated the mean number of total child maltreatment deaths as 101.02 (95%CI=92.52, 109.53), with abuse deaths of 64.55 (60.85, 68.25). Most pair-wise and pooled comparisons worked equally well for abuse deaths, but estimates for total child maltreatment deaths were low.

CONCLUSIONS

Capture-recapture methods applied to existing administrative datasets produced accurate estimates of child abuse deaths but were not useful in producing reliable estimates of total child maltreatment deaths due to undercounting neglect-related deaths in all existing administrative data sets. The underlying assumptions for capture-recapture methods were not met for neglect deaths. Local and/or state teams conducting ongoing intensive case review may yet remain the best way to identify the total number of child maltreatment deaths.

PRACTICE IMPLICATIONS

Capture-recapture methods allow for more accurate estimation of the true number of child physical abuse deaths than does using single existing sources of child fatality information, but deaths from causes other than abuse are undercounted. Child maltreatment fatality surveillance requires a systematic process and standard criteria for identifying cases of maltreatment, particularly neglect-related child deaths.

A review of injury epidemiology in the UK and Europe: some methodological considerations in constructing rates

BACKGROUND

Serious injuries have been stated as a public health priority in the UK. However, there appears to be a lack of information on population-based rates of serious injury (as defined by a recognised taxonomy of injury severity) at national level from either official statistics or research papers. We aim to address this through a search and review of literature primarily focused within the UK and Europe.

METHODS

The review summarizes research papers on the subject of population based injury epidemiology published from 1970 to 2008. We examined critically methodological approaches in measuring injury incident rates including data sources, description of the injury pyramid, matching numerator and denominator populations as well as the relationship between injury and socioeconomic status.

RESULTS

National representative rates come from research papers using official statistics sources, often focusing on mortality data alone. Few studies present data from the perspective of an injury pyramid or using a standardized measure of injury severity, i.e. Injury Severity Score (ISS). The population movement that may result in a possible numerator - denominator mismatch has been acknowledged in five research studies and in official statistics. The epidemiological profile shows over the past decades in UK and Europe a decrease in injury death rates. No major trauma population based rates are available within well defined populations across UK over recent time periods. Both fatal and non-fatal injury rates occurred more frequently in males than females with higher rates in males up to 65 years, then in females over 65 years. Road traffic crashes and falls are predominant injury mechanisms. Whereas a straightforward inverse association between injury death rates and socio-economic status has been observed, the evidence of socioeconomic inequalities in non-fatal injuries rates has not been wholly consistent.

CONCLUSION

New methodological approaches should be developed to deal with the study design inconsistencies and the knowledge gaps identified across this review. Trauma registries contain injury data from hospitals within larger regions and code injury by Abbreviated Injury Scale enabling information on severity; these may be reliable data sources to improve understanding of injury epidemiology.

Methodological issues in motorcycle injury epidemiology

Motorcycle riders are over 30 times more likely than car occupants to die in a traffic crash. While this fact is well known, specific issues of methodology in epidemiological motorcycle-injury research have been rarely researched. To facilitate more-valid research on motorcycle injuries, this article evaluates the current state of our knowledge on how we measure the population at risk of injury, completeness of case finding and identification, validity of crash/injury data sources, and completeness of information on important exposures such as alcohol consumption, helmet status, crash severity, and crash speeds, as well as problems of existing injury severity scales and statistical analyses for correlated injury data.

Estimating road traffic mortality more accurately: use of the capture-recapture method in the West Azarbaijan province of Iran

The study estimates the rate of fatal road traffic injuries (RTIs) by population and road-users group in one Iranian province. The capture - recapture method was employed, using both the death register and the forensic medicine register over one year. They recorded totals of 669 and 665 RTIs respectively, giving a non-overlapping number of 897 cases. An estimate of 1018 fatalities occurred, at rates of 34 per 100,000 of the population for all road users aggregated, 10 per 100,000 for pedestrians and 25 per 100,000 for other road users. Coverage was somewhat better for victims less than 15 years of age, and also for males. The method showed 121 under-reported cases in both sources; however, it can help Iranian policy-makers to produce a good estimation of fatal RTIs number each year, when following up current RTIs-prevention programmes. Yet, given that each registry operates separately, optimum coverage will only be obtained when both sources are integrated and work together.

Estimating infectious diseases incidence: validity of capture-recapture analysis and truncated models for incomplete count data

Capture-recapture analysis has been used to evaluate infectious disease surveillance. Violation of the underlying assumptions can jeopardize the validity of the capture-recapture estimates and a tool is needed for cross-validation. We re-examined 19 datasets of log-linear model capture-recapture studies on infectious disease incidence using three truncated models for incomplete count data as alternative population estimators. The truncated models yield comparable estimates to independent log-linear capture-recapture models and to parsimonious log-linear models when the number of patients is limited, or the ratio between patients registered once and twice is between 0.5 and 1.5. Compared to saturated log-linear models the truncated models produce considerably lower and often more plausible estimates. We conclude that for estimating infectious disease incidence independent and parsimonious three-source log-linear capture-recapture models are preferable but truncated models can be used as a heuristic tool to identify possible failure in log-linear models, especially when saturated log-linear models are selected.

Bayesian inference on the patient population size given list mismatches

In applying capture-recapture methods for closed populations to epidemiology, one needs to estimate the total number of people with a certain disease in a certain research area by using several lists with information of patients. Problems of lists error often arise due to mistyping or misinformation. Adopting the concept of tag-loss methodology in animal populations, Seber et al. (Biometrics 2000; 56:1227-1232) proposed solutions to a two-list problem. This article reports an interesting simulation study, where Bayesian point estimates based on improper constant and Jeffreys prior for unknown population size N could have smaller frequentist standard errors and MSEs compared to the estimates proposed in Seber et al. (2000). The Bayesian credible intervals based on the same priors also have super frequentist coverage probabilities while some of the frequentist confidence intervals procedures have drastically poor coverage. Seber's real data set on gestational diabetics is analysed with the proposed new methods.

O desafio da aplicação da metodologia de captura-recaptura na vigilância do diabetes mellitus em idosos: lições de uma experiência no Brasil

O presente estudo buscou avaliar a factibilidade da implantação de um sistema de vigilância do diabetes mellitus na população idosa, empregando para tal fontes de dados sobre morbidade (hospitalar e ambulatorial) e mortalidade e a metodologia de captura-recaptura. Modelos loglineares foram ajustados para o conjunto da amostra e nos estratos formados pelo sexo e a faixa etária. Setecentos e quarenta pacientes foram identificados no conjunto das fontes. O modelo selecionado para o conjunto da amostra estimou um total de 22.925 casos (IC90%: 11.354-57.269), que é implausivelmente elevado sugerindo uma forte dependência negativa final entre as fontes. A estratificação segundo sexo e faixa etária levou a valores pontuais mais plausíveis, embora ainda imprecisos. O único estrato que apresentou valores precisos foi o de mulheres com 60 a 69 anos (1.290 casos; IC90%: 779-2.476). Os fatores que possivelmente explicam estes resultados são: (a) o número pequeno dos casos captados na fonte ambulatorial atendidos em unidades especializadas e de maior complexidade; (b) o sub-registro do diabetes.

Measuring transport injuries in a developing country: an application of the capture-recapture method

The purposes of this study are to provide an estimation of the incidence of transport injuries in a defined local community in Nicaragua by using the capture-recapture method, and to compare results using this method when data at different levels of severity are utilized. Two sources of injury data were used to monitor injuries: hospital data (inpatient and outpatient) and traffic police records. Characteristics available for matching included name, age, sex, and date of occurrence. The methodology of capture-recapture was used to estimate the ascertainment degree of both sources of data and the estimate mortality and incidence rates. Estimates were calculated both when all hospital data were taken into account (inpatient and outpatient combined) and when only inpatient records were matched against police records.First, including police records and all hospital data, the mortality and morbidity estimates were 35.5/100000 and 43.7/1000 per year, respectively. Second, when outpatients were excluded from the analysis, the corresponding estimates were 28.6/100000 and 7.5/1000, respectively. In non-fatal cases, the ascertainment-corrected coverage through police records was 2.6% and through hospital surveillance 19.0% when both inpatients and outpatients were included. In fatal cases, the corresponding rates were 56.1 and 22.8%, respectively. The combined data set pointed out pedestrians and cyclists as the main risk groups. Most fatalities were due to head injuries. Our results show that neither police records nor hospital records nor the aggregate database provided acceptable coverage of transport-related injuries. Combining police and hospital data by means of capture-recapture analysis produces more valid estimates, but caution must be given to the issue of severity heterogeneity between the two sources.

Estimating the lesbian population: a capture-recapture approach

STUDY OBJECTIVE

Little is known about the number of women who identify as lesbian. Estimates from the US range from 1% to nearly 10%. Accurate estimates are critical in order to meet lesbian's healthcare needs and to address health problems that may be more prevalent among them. This study used capture-recapture methods to estimate the lesbian population of Allegheny County, Pennsylvania.

DESIGN

Mailing lists from four sources were used to identify lesbians. The capture-recapture method and log-linear modelling were used to estimate the number of lesbians in the defined geographical area, and the percentage of the female population they comprised there was determined through census data.

SETTING

Allegheny County, Pennsylvania, USA.

RESULTS

A total of 2185 unique names were identified. The capture-recapture method estimated that the total lesbian population of Allegheny County was 7031 (95% CI 5850 to 8576). Therefore, based on the 1990 census figures, the county's adult lesbian population was estimated to be 1.87% (95% CI 1.56% to 2.28%) of the adult female population.

CONCLUSIONS

An estimate of the lesbian population is fundamental for addressing lesbian's health needs and for developing appropriate research programmes. Capture-recapture methods have the potential to provide accurate and reliable estimates of this population in any location.

Capture-recapture: a useful methodological tool for counting traffic related injuries?

INTRODUCTION

Although the capture-recapture technique is increasingly employed in studies of human populations to correct for under-ascertainment in traditional epidemiological surveillance, it has rarely been used in injury research.

OBJECTIVES

To estimate the completeness of official data sources on traffic related injuries (TRIs) by using the capture-recapture technique and to calculate an ascertainment corrected number of fatal and serious TRIs among Scottish young people aged 15-24 years. The appropriateness of the approach in this context is also assessed.

METHOD

A two sample capture-recapture technique was applied to two official sources of TRI data. Data on TRIs were obtained from the Scottish Health Service and the STATS19 dataset at the University of Essex Data Archive for 1995. Four standards (A-D) of matching were applied to fatalities and serious TRIs to allow plausible relaxation of matching standards within the context of the data collection setting. The completeness of each data source was assessed, and an ascertainment corrected number of fatalities and serious TRIs calculated.

RESULTS

The ascertainment corrected number of TRI fatalities among 15-24 year olds using standard D was 104. This represents only a small increase in the number of fatalities using capture-recapture than when using each individual dataset. The completeness of the Scottish Health Service database for TRI fatalities was 93%. The STATS19 database was 95% complete. The ascertainment corrected number of TRI hospital admissions was 1969. The STATS19 and the Scottish Health Service databases were approximately two thirds and three quarters complete respectively for non-fatal TRIs requiring hospitalisation.

CONCLUSIONS

Injury researchers have advocated the linkage of major datasets to supplement and improve the quality of injury data. Using capture-recapture we found that routine databases enumerate TRI fatalities accurately, in contrast to injury morbidity databases that do not. Capture-recapture is a potentially useful method of evaluating the completeness of data sources and identifying biases within datasets. However, ascertainment corrected rates should be viewed with caution. A number of requirements of the capture-recapture technique are unachieved in this study of injury in the human population.

[Capture-recapture methodology: an option for surveillance of non-communicable diseases in the elderly]

Capture-recapture methodology is used in Ecology to estimate the total size of wild animal populations. This method can be used in Epidemiology to estimate the incidence and prevalence of communicable and non-communicable diseases in a relatively inexpensive and quick way. Surveillance systems based on the use of this methodology are a potential alternative for monitoring non-communicable diseases in Brazil. This paper presents a brief review of fundamental capture-recapture methodology and its applications to Epidemiology. We also present a theoretical model for implementation of a diabetes mellitus surveillance system in the elderly using publicly available morbidity and mortality data sources and the capture-recapture methodology.

Children are not goldfish--mark/recapture techniques and their application to injury data

OBJECTIVES

Mark/recapture (or capture-recapture) is a simple technique commonly applied to estimate the hypothetical total (including undercount) in a register composed of cases from two or more independent and separately incomplete case lists. This paper seeks to illustrate serious drawbacks in the use of the mark/recapture technique when applied to injuries.

SETTING AND SUBJECTS

Northumbrian children under 15 years of age who were seriously injured in motor vehicle accidents (MVAs) over a five year period ascertained from two data sources: police reports and hospital inpatient records.

METHODS

Individuals (n) appearing in both police (S) and hospital (H) case lists are identified using various matching criteria. The separate and combined influence of age, sex, and casualty class (cyclist, passengers, pedestrians) on the probability of such matching is estimated using multivariate techniques. The hypothetical total incidence of child MVA victims (N) is calculated from N = (S x H)/n.

MAIN OUTCOMES

Estimates of the incidences of "serious" injuries in MVAs under various conditions of stratification and matching. The overall procedure is tested for conformity with accepted criteria for valid use of mark/recapture.

RESULTS

About one third of the 1009 police and 836 hospital records could be exactly matched. There were significant variations in matching proportions by class of accident (pedestrian v passenger v cyclist). This selective recapture or "heterogeneity" was not affected by sex, but was independently influenced by the age of the child. Further uncertainty was introduced when matching criteria were slightly relaxed. Estimates of the total population of children with serious injuries vary accordingly from 1729 to 2743. A number of plausible reasons why these two data sources might not be unbiased or mutually independent samples of the total target population are proposed as explanations for this heterogeneity.

CONCLUSION

This typical example of two sample mark/recapture estimation in an epidemiological setting can be shown to violate virtually all the requirements for valid use of the technique. Very little can be deduced accurately about the scale or characteristics of an unobserved group by the use of mark/recapture applied to two overlapping health event registers.

Capture-recapture method to determine the best design of a surveillance system. Application to a thyroid cancer registry

The capture-recapture method is often confronted, when assessing completeness of surveys, to problems of dependence of data sources. The objective of this paper is to discuss the application of capture-recapture methods to choose the optimal combination of sources for a surveillance system. Our approach is based on: (1) using multiple sources, (2) assessing dependence between sources and between pools of dependent sources and other sources, (3) ruling out combinations that yield biased estimates, and (4) choosing the combinations of sources that have the best ratio between precision and applicability. We studied the independence for each pair of sources by computing the capture-recapture odds ratio. We characterized all combinations of sources by their sensitivity, coefficient of variation of the estimated number of cases, and level of resources needed to ascertain cases. Application of the approach is illustrated by data from a survey of thyroid cancer in New Caledonia, where five sources were used to estimate the incidence. The five sources provided 119 cases; the exhaustivity of sources and combinations of sources varied from 27.1 to 99.2%. Determination of dependence revealed ten dependencies out of 22 combinations. Coefficients of variation of the estimated number of cases varied from 0.83 to 27.79. The preferred combination included four sources and had a sensitivity of 97.5 and a coefficient of variation of 0.94. An assessment of dependence, based on simple criteria, can be used to choose the best combination of sources for a registry or a surveillance system.

Birth defects surveillance: assessing the "gold standard"

OBJECTIVES

This study assessed the sensitivity of the Metropolitan Atlanta Congenital Defects Program (MACDP) by capitalizing on the delayed receipt of a data source.

METHODS

In 1997, we reviewed the medical records of potential cases from the 1995 birth certificates that had not previously been identified by the MACDP. Capture-recapture methods produced an estimate of total cases.

RESULTS

We identified 1149 infants with defects, including 20 exclusively from birth certificates. The estimated sensitivity of the MACDP when data from birth certificates were included was 86.9% (95% confidence interval [CI] = 80.6%, 91.9%) at 1 year after birth, increasing to 94.8% (95% CI = 90.3%, 97.8%) at 2 years after birth.

CONCLUSIONS

The MACDP underestimates defects by 13% at 1 year after birth and by 5% at 2 years after birth.

Surveillance of non-fatal workplace assault injuries, using police and employers' reports

Although the majority of work-related homicides are routinely reported in the United States, information on non-fatal events is less complete. Comprehensive surveillance of non-fatal events depends on understanding reporting trends to different agencies. This study characterizes workplace assaults reported to police and through employers in eight southern California cities. Employers' reports filed from October 1, 1994, through January 31, 1995, and police reports filed from June 1, 1994, through March 31, 1995, that involved a non-fatal workplace assault injury were included. Reports from police and employers were linked, and annualized rates combining non-duplicative reports were calculated and event characteristics compared. The combined annualized rate of workplace assault injury for the eight cities was 184.6 per 100,000 workers, which was almost twice the rate found in either reporting source individually. Police reports differed from employers' reports by industry and occupation of victim but not type of event or weapons used. Examination of multiple reporting sources for non-fatal workplace assault injuries is essential to identifying the magnitude of these events. Understanding trends in reporting is important for the effective design of prevention programs.

Occupational injuries among adolescents in Dunedin, New Zealand, 1990-1993

STUDY OBJECTIVE

In this study we describe occupational injuries among adolescents (ages 15 through 19 years) presenting at a hospital ED in Dunedin, New Zealand, 1990-1993.

METHODS

We used a new database to identify work-related injuries, as well as type of injury, cause of injury, injury site, occupation, industry, age, and sex.

RESULTS

During the 4 years of the study, 1,361 work-related injuries were treated at the ED, for an injury rate of 13.8 per 200,000 hours worked (100 full-time equivalents). Males had a rate of 20.6 injuries, females 5.8 injuries, per 200,000 hours. Compared with injury rates from other studies, rates were lower than, but reasonably comparable to, those rates estimated through more detailed surveys. Main injury sites included upper extremities (mostly hands), head (mostly eyes), and lower extremities. Laceration was the main type of injury, followed by sprain/strain and foreign body. External cause of injury was mainly cutting/piercing objects, followed by foreign body and being struck by or against an object. The rate was highest for the 16- and 17-year-olds and decreased for 18- and 19-year-olds. The rates of injuries in the construction sector were the highest of all occupational groups, at 27 injuries per 200,000 hours, followed by transportation/ communication, manufacturing (including meat and fish processing), and business and repair services. Laborers were the highest occupational risk group, with 36 injuries per 200,000 hours, followed by machine operators, precision production workers, and service workers.

CONCLUSION

These findings may be helpful in focusing prevention efforts in high-incidence areas. This study demonstrates how a well-planned data-collection system can overcome some of the previously described difficulties of getting prevention-oriented information from EDs.