| Objectives: The change in the inhabitant death rate and the corresponding causes is one of the important indexes reflecting the health condition of the people, and also is the scientific basis for making health policy and evaluating the health work quality and effect. In some areas of death registration systems, due to under-reporting death data cannot reflect the actual death rate and mortality rate was low. The conventional method of estimating the number of the death from illness is to sum up the data from several different sources and removes the repeated recordes to get the estimated death number finally. However, due to the incompleteness of the data from different sources, the final result is still incomplete.It is very meaningful to estimate the actued number of death using incomplete data from several different sources. Capture - recapture method provides a quite efficient approach to solving this problem.Capture-recapture method was originally used in ecology to estimate wildlife population in the restrictive region. The simplest capture-recapture model is to capture some wild animals from restrictive region. The caught samples are returned to the population after marking label, and then capture animals to get the second sample using the same procedure. Then using the number of individuals caught in both sets of samples and the number caught only in one set of sample respectingly, the number not caught in both sample is estimated and the total population size is summaried finally. Therefore, capture-recapture is also called capture-mark-recapture in ecology.Capture-recapture method started to be applied in the human health field in 1940's. But it was used in epidemiology in 1969 by Wittes and Sidel for evaluating the frequency of birth defect. Due to its simple principle and inexpensive cost, capture-recapture method is getting more and more used in epidemiology(infectious disease, drug abuse, injuries and tumors) " , for estimating under-reporting death number and evaluating the completeness of the monitoring system. The World Health Organization (WHO) also sets the capture-recapture method as a standard method for monitoring the incidence of insulin-dependence diabetes mellitus.There are some essential assumptions and limitations in using the capture-recapture method. Among them, an important statistically assumption is that sources must be independent using the capture-recapture method. Researchers have found for human populations, this assumption cannot be met. At least three or more samples are needed to adjust for dependencies among the samples. Thus, based on three sets of incomplete data from the third death cause review investigation, this paper applies the capture-recapture method to estimate the actual death number in Linzhou city from 2004 to 2005 and assess the integrality of the death data in all death cause registry.Method: 1. The death data were collected from the all-death cause registry, civil administration and police registries in linzhou city during 2004 to 2005. Data without name, cause of death and date of death was regarded as omissive case which was neglected. 2. Cases in every source were matched on the name, ID card, age, sex, date of death, cause of death and resident address. 3. For the 3 sources, log-linear models were used to estimate the number of under-reporting death cases. So, the estimated total number of death was got by the observed death number plus the estimated under-reporting death number. 4. In order to explain that using log-linear model could avoid underestimate caused by dependencies between 2 samples when the samples were used, the estimates got by using the 2-sample capture-recapture method were provided as a comparison. 5. Considering the possible affect of demographic characteristics on the probability of capture, the death data from cancer, road traffic injure and child under age 5 was stratified by age (group) and gender. The number of under-reporting cases for each stratum was calculated using log-linear model method and the total number of underreporting death was obtained by summing the stratified age-gender estimates. 6. In order to evaluate the practical value of applying for capture-recapture methods in death monitoring, the estimate of capture-recapture was compared with the result of the third-time death cause research in Linzhou.Results: 1. It was reported that death number of all death cause was 13007 from combined three sources during 2004 to 2005. Among them, the number of death from malignant tumor, road traffic injures and child under age 5 were respectively3069,434 and 359; 2. The capture-recapture method estimated that the death numberof all-death causes was 13402 (95%CI 13353-13458) and the mortality was 6.62%o in Linzhou city during 2004 to 2005. Among of these, 3139 (95% CI 3128-3170) and 155.0 per 100,000 population for malignant tumor, 461 (95%CI 448-486) and 22.8 per 100,000 population for road traffic injuries, 398(95%CI 383-424) and 23.9‰for child under age 5. 3. Taking the estimate of capture-recapture method as denominator, the underreporting rate of the registry alone ranged form 6.6% to 26.9%. Among them, the underreporting rate of malignant tumor was 6.6%, while the child under age 5 was 26.9%. After three sources were aggregated, the underreporting rate ranged from 2.3% to 9.8%. And results were also different in age-gender strata. Female deaths were more likely to be underreported than male for child under age 5 but for road traffic injure the result was just the opposite. Low-age group was more likely to be underreported than the other age group. 4. Capture-recapture estimates were very close to those obtained by the third death investigation except the death data of child under age 5.Conclusion: 1. Registry office or aggregated database underestimated the death number in linzhou city during 2004 to 2005. On the base of data which were collected from police, village symposium, civil and registry, using the capture-recapture method can obtain the turn death number. 2. The death data of malignant tumor was real but the data of road traffic injure and child under age 5 was imcoplete in Linzhou city. 3. To compare the result of capture-recapture estimate with the third-time death cause investigation, we found that true estimates of death were obtained more quickly and economically using capture-recapture method. It is suggested that various places should establish the perfect death administration system, enhance professional training and coaching, improve the quality of data, and guarantee validity and integrity of data in order to use the information resource effectively. |
PDF Full Text Request