RWS Clinical Research Paper Reference Model (Clinical Research Protocol)

1. Basic Principles of Research Protocol Design

Many scholars have misunderstandings about RWS, believing it is similar to clinical retrospective research, both involving retrospectively collecting clinical data and analyzing it into papers. This view is extremely incorrect. Like traditional clinical research, RWS also requires strict, scientific design, planning, and implementation, and must follow a series of basic principles and concepts of classical clinical research. Unlike a specific clinical research method in traditional clinical research, RWS is an integrated clinical research methodology system. It can be either experimental research or purely observational research; it can be prospective research or retrospective research; and in prospective RWS, randomization and other clinical research methods can also be used. Therefore, before conducting RWS, it is necessary to first clarify the practical problems that the research needs to solve. The PICO principle in evidence-based medicine [participants, intervention, comparisons, outcomes] can be used to decompose the problem, understand its essence, and thus clarify the research思路.

2. Prospective or Retrospective Design

RWS does not specify that the research must be retrospective clinical research or must be prospective clinical research. In RWS, whether the research implementation method adopts a prospective design or a retrospective design is determined by the form of data acquisition. If the data needed for the research is newly collected data, a prospective research design should be adopted; if the collected data is existing data, a retrospective research design can be adopted. Moreover, in retrospective RWS, data can come from past medical insurance data or hospital medical record registration data, as well as other previously registered and recorded data.

3. Randomization and Intervention

RWS can be either observational research, observing the relationship between potential variables and endpoint indicators; or interventional research, comparing the effects of two or even more interventions on endpoint indicators. Therefore, whether there is intervention or not is not the key point to distinguish RWS from traditional clinical research. However, it should be noted that RWS emphasizes whether the researcher has strict limits on the intervention measures. RWS requires that all intervention measures of physicians are carried out in a real medical environment. Physicians can communicate with patients, coordinate clinical diagnosis and treatment measures, and researchers only need to strictly observe and record the corresponding data and results.

Randomized or non-randomized design is also not a criterion for judging whether it is RWS. Randomization, as a method in clinical research, is also applicable to RWS. Moreover, in the category of RWS, there is a special and very important experimental research design method, namely the pragmatic randomized controlled trial (pRCT), also known as the practical randomized controlled study. pRCT refers to comparing the treatment outcomes of different intervention measures by using randomized, controlled and other research methods in a real medical environment. Although pRCT is essentially interventional research, its regulations on the specific implementation of intervention measures are lower than those of traditional RCT. pRCT does not emphasize that it must be strictly implemented according to the provisions of the research plan, and it is more flexible in the formulation of intervention strategies. Therefore, pRCT has good application prospects in the field of surgical clinical research, especially in the comparative research of surgical strategies and surgical techniques.

4. Data Collection, Acquisition and Analysis of Real World Research

Unlike classical RCT which needs to strictly define the inclusion/exclusion criteria to specify the入选人群 and intervention measures, the focus of RWS is on the authenticity and accuracy of research data. Data is the top priority of RWS. Where does the data in RWS come from? How is the data collected? How is the data processed? These three questions are the core issues of RWS, involving the three dimensions of data acquisition, data management, and data analysis and application in RWS. These three-dimensional issues are not only the key to ensuring the authenticity and reliability of RWS data, but also the main difference from traditional interventional clinical research.

Where does the data come from: RWS emphasizes that the data comes from a real medical environment. The scientific research field often says “Garbage in, Garbage out”, that is, wrong data will produce wrong evidence, which also applies to clinical research. This theory also applies to RWS. The authenticity of data is an important guarantee for the reliability of clinical evidence generated based on data. Therefore, this puts forward higher requirements for the recording of materials, data and information in the routine clinical diagnosis and treatment work of the research units that intend to carry out RWS. Considering that there are many confounding factors in surgical research, such as individual differences of patients, differences among surgeons, differences in anesthesia, and heterogeneity of tumors, etc., all may affect the judgment of research endpoint indicators.

If the corresponding clinical diagnosis and treatment information cannot be accurately recorded, it will have a great impact on the accuracy and reliability of the research conclusions. For example, in surgical clinical practice, the evaluation of intraoperative blood loss is a clinical work with relatively large bias, but this indicator is a key indicator for comparing the differences between various surgical methods. At present, there are多种方式 that can be used to evaluate intraoperative blood loss, such as: estimation by experienced intraoperative visual inspection, calculation based on vital signs or laboratory test results, or calculation by weighing intraoperative gauze and drainage bottles. However, uniformly and constantly choosing a relatively accurate evaluation method and recording it is one of the key factors to reduce possible bias.

Therefore, carrying out RWS puts forward higher requirements for daily surgical clinical work. We suggest that by standardizing preoperative laboratory examination evaluation standards, standardizing intraoperative situation recording, standardizing postoperative recovery indicator measurement, standardizing postoperative pathological standard judgment and regular follow-up evaluation, and structuring the collected data as much as possible, it will help to ensure the truthfulness, accuracy and reliability of the research data from the bottom.

How to collect data: Under the premise of ensuring the reliability of the data source, the traceability of data collection is another characteristic of RWS. Data traceability requires that all data included in the final statistical analysis decision can be traced back to its fundamental source. Although this point is consistent with the requirements for data collection in other research methods such as prospective randomized controlled studies, due to the wide range of data dimensions involved in RWS, the large number of samples, and the huge workload of data collection, how to transform the繁杂的原始病历信息 into usable data information in the database must be重点考量 at the beginning of RWS design. When designing the research plan, this problem can be solved by establishing a data quality management plan.

This data quality management plan not only stipulates where the data is collected from, but more importantly, it is reflected in structured data batch collection, unstructured data conversion, abnormal data monitoring, data logic error correction, and data Double-Check verification, etc., so as to comprehensively, systematically and accurately collect various subjective and objective medical information data during the diagnosis and treatment of subjects. With the popularization of informatization, by using the electronic medical record system, past unstructured clinical information is converted into a structured clinical data mode, and exported through the port, which greatly reduces the workload of data collection. However, how to set the structured data fields in the electronic medical record system still needs comprehensive consideration based on factors such as disease type and treatment mode. For RWS in the surgical field, the focus and difficulty of data collection lies in the regulations and collection of surgical variables.

The surgical operation habits among surgeons, individual differences among patients, and anesthesia methods are all factors that may affect the short-term and long-term observation endpoint indicators of surgical clinical research. Therefore, when designing the research plan and formulating the data quality management plan for surgical RWS, it is also necessary to重点考虑 the definition of various variables related to various surgeries. For example, in gastric cancer surgical clinical research, the scope of surgical lymph node dissection can be defined based on the postoperative pathological report of lymph node分组 and the provisions for the scope of D1, D2 and D2+ lymph node dissection in the Japanese gastric cancer treatment guidelines. However, considering that the categories of data that need to be collected in RWS are numerous and the amount of data is large, designing a complete data quality management plan is particularly important in RWS research.

How to analyze data: Data processing and analysis is the process of transforming real-world data into real-world evidence. This is both the focus and the difficult problem in RWS. The main purpose of conducting RWS is to explore and obtain key events related to disease diagnosis and treatment outcomes from massive clinical data through statistical methods “fishing in the sea”. Because classical RCT has controlled confounding and bias factors to the greatest extent from the perspective of research design, while RWS has various possible confounding and bias factors. Therefore, in the process of RWS data analysis, it is the key to ensuring the reliability of real-world research evidence to reasonably and effectively discover and control bias and confounding factors through scientific statistical methods. At present, in RWS statistical analysis, multiple analysis models are often used, such as propensity score matching, stratification or subclassification, and regression adjustment, etc., to control bias and confounding factors. Of course, at the beginning of the research design, it is also an essential环节 to formulate a reasonable statistical analysis plan with the assistance of statistical experts.

5. Other Considerations for RWS Design

Because RWS is also a type of clinical research, its research conduct needs to follow the basic norms and general principles of clinical research conduct. RWS also needs to review and consider its ethical issues in accordance with the principles and methods of ethical review of all clinical research. The principles普遍适用 in clinical research, such as “respect for people”, “benefit”, and “justice”, also apply to the ethical considerations of RWS. However, considering the relative particularity of RWS, there are certain particularities compared with classical clinical research in the process of research application and implementation. For example, the intervention measures of RWS are similar to routine clinical diagnosis and treatment, and the risks undertaken by subjects participating in the research are not greater than the risks of receiving routine clinical diagnosis and treatment.

In addition, because the real world pays more attention to the collection, analysis and mining of massive information, the possible risks of patient privacy leakage and information security involved in the process of collecting, storing, analyzing and mining this information in RWS are issues that RWS needs to focus on. Furthermore, all clinical research involving patient data requires the informed consent of the patient themselves. However, in RWS based on existing data, such as real-world research on medical insurance data, informed consent can usually be waived, but it needs to be reviewed and filed by the ethics committee. In prospective real-world research, obtaining patients’ informed consent is an essential环节, and subject informed consent is a不小的挑战 that must be faced in conducting RWS.

On the one hand, unlike traditional clinical research which has fixed preset intervention measures, the clinical intervention measures of RWS changes, and researchers usually need to fully and clearly inform patients of the purpose of the research and various clinical possibilities, so that potential subjects without medical background can clearly understand the purpose and risks of the research and voluntarily participate in the research;

On the other hand, because RWS needs to obtain comprehensive and massive data information as much as possible, patients’ refusal to participate in the research may affect the coverage of the research population, thereby affecting the research efficacy. Therefore, to obtain the consent of each subject to participate in the research as much as possible, and to ensure the external authenticity of the research results, is what needs to be paid attention to in the implementation process of prospective RWS.