1. Number of observers and completion time. The number of people who need to be observed in the case-control study is small, and the field work is completed after the investigation is completed without follow-up. So it is much more economical than cohort study. If we want to use the cohort method to study rare diseases, almost all the main energy will be spent on the follow-up of those who are not sick. Suppose it is planned to study whether the use of estrogen by mothers before and after conception will increase the risk of congenital heart disease in their children. Assuming that eight unexposed women suffer from congenital heart disease every 1 0,000 deliveries, in order to find out the risk that may be higher than 1 delivery (RR=2), the cohort study needs to observe the pregnancy results of 3889 exposed women and 3889 unexposed women [let α=0.05 by formula (Annex 5- 18). However, if the case-control method is adopted and the proportion of exposed women is 0.30, the required sample size can be estimated by formula (Annex 5- 14) or its simplified formula. Under the same conditions (α=0.05, β=. 10, RR=2), only 188 cases and controls are needed, thus,
Mainly because of this problem, although cohort study has two advantages, in the epidemiological study of cancer (a rare disease with an incidence of several ten thousandths to several ten thousandths), case-control method is still mainly used.
2. Quantitative case-control study of research factors In a study, the relationship between multiple factors and diseases can be studied at the same time. When the cause of a disease is unknown, it is more appropriate to explore the role of various possible factors, that is, it can be used to verify a specific hypothesis and generate a hypothesis. Cohort study can examine the relationship between one factor and multiple diseases at the same time. For a factor with extensive risk effects, only through prospective cohort study can its effects, especially long-term health effects, be fully revealed.
3. The possibility of prejudice
⑴ Selection bias: It is more likely that there will be selection bias in the controlled study of diseases, especially when the inpatients are the objects. However, there may also be selection bias in cohort studies, such as voluntary participation, latent diseases not found at the beginning of the study, and missing or incomplete records of some subjects in retrospective cohort studies, all of which are the sources of bias.
⑵ Information bias: Case-control method requires respondents to recall the exposure history of several factors in the past, and there is a greater possibility of memory bias; The information about exposure and outcome of the queue method does not depend on memory, so it is objective. Because of the long follow-up, the cohort study may be biased due to the loss of follow-up: the loss of follow-up rate may be different between the exposed group and the unexposed group, and the outcome rate of the lost and unvisited people may be different, which may be the source of bias.
4. Case-control study on rate and related indicators In general, the number of cases and patients in the total population is unknown, so it is impossible to calculate morbidity, mortality, RR and excess risk. However, OR can be calculated, and PAR can also be calculated. At this time, the exposure rate of the control group can be used to estimate the population exposure rate. Although case-control study can judge whether there is a statistical connection between exposure and disease, it is not easy to draw a causal conclusion. From the results of prospective cohort study, it is easy to draw the conclusion whether there is causal connection, which is second only to the experimental results.
In short, according to the advantages and disadvantages of the two methods, case-control study is generally used as the first stage of etiology research, screening suspicious etiology, establishing hypothesis, and cohort study is used as the second stage to test hypothesis. At this time, the etiology and natural history of the disease have been fully understood through other channels, otherwise few people will rashly conduct cohort studies because of the high cost of failure. The recent trend is to advocate embedding case-control studies in cohort studies to take advantage of their advantages. But in fact, it still needs to be large-scale, which is generally difficult to do.
Finally, Table 4- 15 summarizes the characteristics of the three main epidemiological methods.
Table 4- 15 Comparison of three epidemiological research methods
Methodological characteristics Prospective cohort study Case-control study survey
The sample consists of disease-free individual cases and control exposure, current patients or survivors.
Grouping criteria for exposure or non-exposure before illness or non-illness
The present situation of time series is forward-looking (from cause to result) and retrospective (from result to cause)
Compare the morbidity or mortality of exposed and unexposed cases with the past exposure, and compare the exposure of the exposed or sick.
Percentage of exposure to morbidity or mortality
Risk degree, relative risk degree, rate difference, PAR or PAR relative risk degree and rate difference; face value
Advantages: More accurate exposure data; Can calculate the incidence and risk; The relationship between one exposure and multiple diseases can be studied simultaneously; Used to test the hypothesis that the sample is small and the result is fast; Low cost; There are no lost accesses; The relationship between a disease and multiple exposures can be studied at the same time to screen the causes; It can be used in the study of rare diseases with rapid results.
Disadvantages require large samples and long-term follow-up; High cost; There are many problems of lost visit; Not suitable for rare diseases, the sample is not representative, and the control selection is not easy to be appropriate; Memory exposure history is biased; It can only be counted as or, and the causal relationship is not easy to determine; Only investigating survivors is not suitable for diseases with short course of disease and fast death; Rare diseases need to be investigated in large samples and are not applicable.