Taiwan Today

There are two kinds of public health data in Taiwan: vital statistics and general health statistics. Vital statistics deal with population analysis, neonatal and infant mortality, and analysis of causes of death. General health statistics cover everything else. The Office of Statistics of the Department of Health (DOH), Executive Yuan, is in charge of collecting and analyzing public health statistics across the board. The Taiwan Provincial Health Department, the Taipei City Health Department, and the Kaohsiung City Health Department also have departments of statistics. As an example of how the system works, this is the sequence of events after somebody dies, and the person or agency responsible at each stage: Reporting death---the deceased's family. Death certification---physicians, forensic doctors, prosecutor's office. Copying death data---office of population administration and health stations. Collecting data---health department and population administration department of city and county government. Collecting and checking data, coding cause of death, copying by hand, entering data on computer and sending it to the DOH's Office of Statistics via computer tape---Office of Statistics, Taiwan Provincial Health Department. According to the DOH, the three leading causes of death in Taiwan in 1994 were, in descending order, malignant neoplasms (cancer), cerebrovascular disease (strokes), and “accidents and adverse effects,” with heart diseases, diabetes, and chronic liver diseases and cirrhosis coming next. Nephritis, pneumonia, hypertension and bronchitis, emphysema and asthma round off the list of Taiwan's top killers. The government is confident in the accuracy of its statistics, but there are many outside skeptics. Kao Sen-yeong (高森永), an associate professor at the National Defense Medical Center's School of Public Health, has reservations. “Official statistics are not very reliable,” he says. “For various reasons, certain causes of death are underestimated, and sometimes the cause of death is just incorrectly certified.” For instance, Kao believes that the official death rate for newborn babies is wrong . “It's only five per thousand births, which is unbelievably low,” he says. “The system for reporting infant deaths is at fault.” The reporting system requires that the parents of a newborn baby register the birth with the relevant office within fifteen days of the physician or midwife handing over the birth certificate. But the first twenty-four hours of a baby's life are the most hazardous and, in medical terms, human beings are not out of the woods until they've survived for at least a month. “Many newborn babies die within fifteen days, so many parents don't report the birth to the authorities. It follows that they won't report the death, either,” Kao maintains. Things may improve, however, because since March this year the duty of registering new births has been shifted from parents to physicians. Death certification is another problem area. “Many physicians are inadequately trained when it comes to writing death certification, and therefore a lot of the data is garbled or just plain wrong,” Kao says. Complex disease patterns can make it difficult for physicians to detect the real underlying cause of death. Medical schools do not offer training in this field. Factor in the risks of copying data by hand---a common practice---and the scope for error increases exponentially. Statisticians know that their task is far from easy. “The biggest problem when working with statistics is to get accurate data,” says Lee Li-shu (李麗雪), director of the DOH's Office of Statistics. “Many statistics come from surveys and questionnaires. But people are reluctant to reveal themselves to strangers, so a lot of interviewees refuse to answer our surveyors' questions. Then, sometimes, people just don't want to tell the truth.” To reduce the scope for error, data is crosschecked by computers and statisticians. Some errors that are illogical in nature can be detected easily. For example, it's impossible for a three-year-old child to have a job, and any answer suggesting the contrary can be disregarded. Computers root out other errors with the help of special programs designed by statisticians. Even so, some mistakes slip through the net. For example, Lee Li-shu came across figures which suggested that the number of dentists practicing in Taipei in 1994 was lower than the 1993 figure. All her instincts told her that this had to be wrong and sure enough she was able to track down the error to health stations, where staff had misunderstood the definition of “dentist” to be used in compiling data. “It's not enough for a statistician to amass professional knowledge,” she says. “A painstaking approach is essential.” Some other statisticians hope that physicians will make death certificates clearer and more detailed. Lu Shu-mei (盧淑美) is section chief of the vital statistics department in the Taiwan Provincial Government's Department of Health. “It seems that many physicians don't have much time to write out death certificates,” she says. “For example, some doctors just write 'car accident' as the cause of death.” Although statisticians make a point of regularly attending hospitals and clinics in order to raise these questions with physicians, the accuracy of death certification remains far from perfect. There is talk of launching training programs for physicians in this area, and Sun Shu-mei (孫淑梅), who works in the same office as Lu Shu-mei, would like to see that happen. She believes that statistics will only be accurate when everyone at all levels is working to a proper standard. “It's a matter of conscience for me to do this job as best I can,” she says. And Lee Li-shu thinks that statisticians could use a little more encouragement, as well as criticism. “I've been doing statistical work for twenty-eight years now,” she says. “And even if I sometimes get a little upset, I still enjoy what I do.” —by Jessie Cheng