Common medical research methods

Clinical medical research is also composed of experience level and theory level, which are closely linked to complete the understanding of medicine. The experience level, that is, the perceptual knowledge stage, is the stage of collecting various clinical information of patients in clinical practice. In order to get scientific conclusions, the collected clinical information should be representative, objective, accurate and repeatable (accuracy). According to the hypothesis and intervention, the collected clinical information can be divided into observational research, experimental research and working research. Observational research can be divided into descriptive research and analytical research, and the experimental conditions cannot be artificially controlled. Descriptive research has neither assumptions nor interventions, such as case reports; Analytical studies have assumptions but no intervention, such as case-control studies.

There are assumptions and interventions in both experimental research and work research, which can artificially control the research conditions and directly discuss the relationship between one or some research factors and results, such as clinical drug trial research.

On the theoretical level, that is, the rational cognition stage, researchers go deep into things, think about the perceptual cognition stage, and then raise one or some clinical phenomena to rational cognition through logical analysis and synthesis, and sum up the general principles guiding clinical practice.

Cohort study is mainly used to test the hypothesis of etiology. This study can directly observe people's exposure to the cause and its outcome, so as to determine the relationship between risk factors and diseases.

Cohort study is an observation method and a causal study. The exposure factors are objective, and the control group can come from the same group or different groups. Before observation, the two groups of people must be people who have not suffered from diseases at the end of the study. The exposure conditions of the two groups are different, but other conditions should be basically similar, and all the observed objects in the two groups should follow the same time. During the observation period, the morbidity or mortality of the two groups were recorded, that is, the observation results. Calculate and compare the morbidity or mortality between the two groups during the observation period. If there are differences, there is a correlation between exposure factors and the occurrence or death of the disease, and the relative risk (RR) value can be directly calculated.

The observation queue is divided into fixed queues (referring to all people as a queue when a specific event occurs). Sometimes it refers to a relatively stable crowd or a relatively large crowd) and a dynamic queue (it refers to an observation crowd that can add new members at any time).

Prospective cohort studies should have clear test assumptions, and the incidence (mortality) of the diseases studied should generally not

In addition to the above conditions, the historical queue research should also have enough complete records or files, which can quickly produce results and save time and money, but the reliability of the results is poor.

Bidirectional queues have the advantages of the above two kinds of queues and make up for their shortcomings to some extent.

The factors of cohort study include pathogenic factors (risk factors), that is, exposure factors and protection factors that lead to the increase of disease events, that is, exposure factors that lead to the decrease of disease events. For example, smoking is a risk factor for lung cancer, and carotene is a protective factor for lung cancer.

The results of the cohort study should be determined, and the results refer to the outcome events expected by the researchers, that is, the results of the individual appearance of the research object. The general outcome is illness, death or some laboratory index. It should be noted that if the research object is alive, there should be strict diagnostic criteria to judge whether it is sick (recurrent); If the outcome of the research object is death, there must be evidence of death.

The start time and length of follow-up of each subject in cohort study are directly related to the curative effect of cohort study, so the start date and end date of follow-up should be clear. The length of follow-up depends on the intensity of exposure and disease contact and the length of disease incubation period. Follow-up time should be determined in order to observe the occurrence of lesions and save money.

In order to determine the follow-up period, it is necessary to know the induction period and incubation period of the research disease. The induction period of disease refers to a period of time when the etiology begins to act on the occurrence of disease; The incubation period of disease refers to the time interval from the occurrence of disease to clinical discovery.

Follow-up methods include questionnaire survey, medical examination, obtaining relevant records and investigation and testing environment.

The observation object of cohort study must be people who have no disease of a certain research achievement and may have the disease of the research achievement during the observation period. The exposed population should have exposure factors worthy of study, such as special exposure population (for example, asbestos factory workers can be selected to study lung cancer), general population (for example, coronary heart disease in a community) and organized population (for example, the incidence study of lung cancer by British male doctors). Non-exposed group (control group) can be controlled inside and outside the selected population, or both internal and external controls can be set.

When estimating the sample size in a cohort study, the whole population is observed only when the exposure times are few, and a certain number of samples are usually taken, but the sample size should be enough to ensure that the difference in morbidity (mortality) between the exposed group and the non-exposed group can be detected at the specified significance and guarantee level. The significance level is the false positive error value (α value) when testing the hypothesis. The smaller the value, the larger the sample size. Usually α= 0. 05 or 0. 0 1. The assurance level is the probability of false negative error (β value) in hypothesis testing, and the assurance level is 1-β. The higher the requirement of guarantee, the larger the sample size is needed. Usually β= 0. 10, sometimes β= 0. 20. The sample size is also affected by the incidence (P0) of the diseases studied in the general population. The closer P0 is to 0. 50, the larger the sample size. The sample size also depends on the incidence of the diseases studied in the exposed population (P 1). P 1 is not easy to get. We can try to get the RR value, P 1 = RR-P0. The difference (d) between the incidence of general population and exposed population also affects the sample size, and d =P 1-P0. The greater the value of d, the smaller the sample size is required.

The proportion of exposed group and non-exposed group should be determined before the start of cohort study. Generally speaking, the exposure group

For example, a cohort study on the relationship between radiation and leukemia. It is known that the incidence rate of the general population is 0. 000 1, and the radiation population is 0. 00 1. If researchers set α to 0. 05 (bilateral test) and β= 0. 1, then U α = 1. 96 and U β = 65438. Substitution formula: n = 2 (0.00055× 0.99945 )× (1.96+1.282) 2/0.0005438+0-0.001) 2 = 65438+.

Because the follow-up of cohort study is easy to be lost, in order to be 90% sure, the sample size should be increased, 14266 people /0. 9 (assurance degree) ≈ 1585 1 person. In fact, each group needs 1585 1 person.

Look-up table method is to look up table N according to four basic data of α, β, P0 and RR (please refer to: epidemiological research method edited by Wang 199 1. P3 18)

The commonly used ratios in cohort study data analysis include cumulative incidence and incidence density.

Cumulative incidence (CI) refers to the probability that a disease will occur in a fixed population in a certain period.

The formula is: CI = I/P × 100 %( I = the number of new cases during the observation period, P = the population at the beginning of the observation). Cumulative incidence is often used when there are enough observers, the observation crowd is fixed and the data is neat. This indicator has no time unit (probability), so we should consider and pay attention to the observation time when applying it.