연구질문의 설정 Defining a review question

Presentation on theme: "연구질문의 설정 Defining a review question"— Presentation transcript:

1 연구질문의 설정 Defining a review question

2 코크란 체계적 문헌고찰의 과정 Steps of a Cochrane systematic review
연구 질문 설정 선정 기준 계획 연구 방법 계획 연구자료 (원자료) 검색 선정 기준 적용 자료 수집 비뚤림 위험 평가 결과 분석 및 제시 분석결과 해석 및 결론 도출 문헌고찰 개선 및 주기적 갱신 Defining your review question is the first step in starting your Cochrane systematic review. You begin this process at the title registration stage, but there’s often more work to be done in really thinking through your question in detail.

3 개요 Outline 잘 정의된 연구 질문의 중요성 연구 질문의 구성 요소 연구계획서 (protocol) 에 포함되어야 할 내용
See Chapter 5 of the Handbook

4 연구 질문 설정 Defining your question
문헌고찰의 필수적인 첫 단계 연구 방법 상 여러 요소들의 설정에 중요한 역할을 함. 선정 기준 연구자료 검색 전략 자료 추출 및 분석 문헌고찰 시작 전 숙고가 필요함 문헌고찰을 위한 업무 계획에 도움이 됨 비뚤림 방지에 도움이 됨 연구 주제 변경 시, CRG (Cochrane Review Group) 과 상의할 것 A well-defined review question is essential to your review. It’s important that you’re clear about what your review will cover, as the question will define many of the subsequent steps of the review: it will determine your eligibility criteria – that is, which studies are included in the review, and which are excluded it will determine how you go about your search strategy, and what kinds of studies you find. If you’re not clear enough about your question at this stage, you may miss papers that are really of interest to you, or find lots of paper that’s not what you wanted. your question will also determine your analysis of the results of the studies – which outcomes do you want to measure, are there specific issues you want to explore, what information do you need to find in the included studies? It’s worth spending time to think this through in some detail at the beginning of the process. Not only can it help you plan your work, but a clear idea of the scope of your review will help you avoid bias later on. By laying down your parameters objectively in advance, you can reduce the risk of making decisions or changing your mind later, on the fly, in a way that might be influenced by the results of particular studies. Your question will be discussed and agreed with your CRG at the title registration stage. It’s not impossible to make changes to the scope of your review later on, but make sure you speak to your CRG if you plan to do so.

5 연구주제와 범위 설정 Planning your topic and scope
연구주제의 중요도와 연관성을 고려하여 연구질문을 설정할 것 보건의료에 미치는 영향력 고려 – 인구집단 및 개별 환자 중재법의 잠재적 영향력 고려 모든 주요 당사자들 (소비자, 보건의료 전문가, 정책결정자) 을 고려 의사 결정 시 실제로 선택에 직면하게 되는 요소들을 다루어야 함 국내외 보건의료 환경을 모두 고려한 관점을 유지할 것 Take some time to think about the importance of your question: who is it relevant to? Does it address an important burden of disease? What is the possible impact of your intervention on that burden? Try to select a question of real importance to consumers and care providers. It’s also important to select a question that addresses the real choices faced in decision-making. What are the intervention options available? What are the outcomes that influence those choices? Cochrane reviews should always consider an international perspective. The burden of specific diseases may vary greatly in different countries or different populations. The intervention options available and the way health systems operate may also be different, as well as the outcomes of importance. Cochrane reviews may be used by people anywhere the world, in contexts very different from yours, so it’s important not to make assumptions when planning or interpreting the results of your review.

6 연구 질문의 범위 설정 A broad or narrow question?
협소한 질문 광범위한 질문 장점 작성이 쉬워짐 읽기 쉬워짐 포괄적임 외적 타당도 증가 단점 여러 개의 세부적 review가 필요함 저자가 원하는 결과를 얻기 위한 목적으로 질문이 설정될 수 있음 복잡해짐 하위집단(subgroup)별 효과차이를 간과하고 지나칠 수 있음 문헌고찰에 대한 종설 (Overview of reviews) 이 더 적절할 수 있음 Consider whether you wish to answer a narrow question, e.g. looking at one specific intervention in a specific patient group, or a broader question, e.g. looking at all interventions to address a particular health condition, or something more complex like health system or community interventions. A well-focussed systematic review question will be broad enough to answer a useful question, allowing for some of the variations that will always be observed in a field, but not so broad that it doesn’t make sense to consider all those studies together in one review. Narrower reviews are easier to write and easier to read, but you may need to have multiple reviews to answer all the questions relating to a condition or type of intervention. Broad reviews are more comprehensive, but they can be complex, more difficult for authors to write, and can miss some of the detail possible in a narrower review. In some cases what’s most appropriate can change over time. There was once a Cochrane review, ‘Interventions for shoulder pain’, that looked at all possible interventions – steroid injections, physiotherapy, exercise, pain killers. Over time, however, as more and more studies were published, it became unmanageable, and has now been split into a set of reviews looking at one specific intervention. One way that the Cochrane Collaboration manages sets of narrow reviews is to publish Overviews of Reviews – these are brief papers that summarise the findings of multiple related reviews. More information about these Overviews is available in Chapter 22 of the Handbook, if you’re interested in reading more. There’s no right or wrong answer to this - it depends on what you’re interested in. Your Review Group may have a preference based on their experience and how your review fits with other similar reviews already published. See Chapter 22 of the Handbook

7 개요 Outline 잘 정의된 연구 질문의 중요성 연구 질문의 구성 요소
연구 계획서 (protocol) 에 포함되어야 할 내용

8 연구 질문의 구성 요소 Components of a question
아래 요소를 상세히 설명할 것 연구자가 문헌고찰을 통해 탐색하고자 하는 변수를 고려할 것 P population (연구 대상 집단) I intervention (중재법) C comparison (대조군) O outcomes (결과 평가 변수) You will need to define each aspect of your question very clearly. ASK: Who has seen this PICO format before? It can be used to design a study, or a literature search, and it’s also useful for thinking through the question for your systematic review. You will need to think about each of these components and define in detail: who are you interested in studying in your review? What is the intervention or group of interventions you want to test? What will the interventions be compared to? What outcomes will tell you which intervention is the most effective? Think about what you’re interested in, and what you’re not interested in. Within each component, it’s also important to think about the variations that you would like to explore further, such as different populations for whom the intervention might work differently, or variations on the way the intervention can be applied.

9 선정 기준 Eligibility criteria
문헌고찰에 포함시킬 연구자료의 결정 기준 아래 항목을 기반으로 기준 설정: PICO 요소 전체 혹은 일부 및 선정 가능한 연구자료의 디자인 정의 연구계획서 출판 후 선정기준 변경 시, 문헌고찰 내에 합당한 설명이 제시되어야 함 When you have clearly defined your question, and identified in detail the components you would like to investigate, you can then define the eligibility criteria for your review. Eligibility criteria are the essential components, the boundaries, that determine whether a study is included and discussed in the review, or whether it is excluded. Remember that the eligibility criteria for a review may be different from the ideal criteria you might set if you were designing a primary study yourself. For example, you may wish to look specifically at the impact on people with a certain stage of disease, or you might have a preferred model for implementing a complex intervention. However, the body of studies you find in the literature might show considerable variation from your ideal model, or might incorporate a mixture of participants with different stages of disease. If these studies are included in your eligibility criteria, you can explore the effect of these differences within the review. If you rule them out, they will be excluded altogether. It’s important to articulate each component carefully, so that when you come to making decisions about including studies in the review, you are doing so as objectively as possible. Having clear criteria to refer to will help you avoid ad hoc decisions that may introduce bias. We’ll go through each of these components in more detail.

10 연구 대상 집단 Population 연구 대상 집단을 명확히 정의할 것 두 가지 고려 요소
건강 상태 및 질환 진단이 어떻게, 누구에 의해 내려졌는가? 연구 대상 및 연구가 수행되었던 환경 연구 대상 집단을 제한할 시, 명확하고 합리적인 기준을 제시해야 함 혹은, 연구 대상 집단의 제한 없이 문헌고찰 대상으로 포함시킨 뒤 하위그룹 분석 (Subgroup analysis) 을 수행하는 방법도 있음 First: the population (sometimes also called the participants, or people). What are the most important characteristics that describe the people you are interested in? The level of precision you need will depend on the topic of your review, but should always be detailed enough to identify which studies you wish to exclude from the review. For example, if you are interested in a specific population with defined diagnostic criteria, e.g. a particular stage of breast cancer, or a particular genetic marker, you will need to describe the criteria in some detail. Keep in mind that your ideal diagnostic criteria may not be universally used, or even available, and older studies may not have access to the newest available tests. Alternatively, if you are looking at a broader group of people with a health condition, e.g. any stage of breast cancer, or even a healthy population with no health condition, less detail will be needed. Whatever limits you place around the group of people should have a clear rationale: if you are limiting to a specific stage of disease, or to a specific age group, is this because you expect this group to react differently to an intervention than other people? Remember: once you have defined your eligibility criteria, you can also identify other differences of interest that you would like to explore in the review.

11 혼합된 연구 대상 집단 Mixed populations
일부 피험자만 연구 대상 선정 기준을 충족하는 경우 예) 선정기준: 16세 이하의 어린이, 그러나 검색된 연구에는 18세 이하의 청소년이 포함됨. 문헌고찰의 목적에 가장 잘 부합하는 방법 선택 검색된 연구의 자료 전체를 포함시킴 선정 가능한 기준 설정 가능 (예, 피험자의 80% 혹은 대다수가 연구대상 선정 기준에 부합하는 경우) 연구 대상 선정 기준에 부합하는 일부의 피험자 자료만 포함 원하는 자료만 별도로 보고되지는 않을 수 있음. 자료 전체를 제외시킴 연구 계획서 단계에서, 이러한 연구들을 어떻게 처리할지 사전에 계획하고 설명함. Sometimes a study includes some ‘eligible’ participants and some ‘ineligible’ participants, for example when an age cut-off is used in the review’s eligibility criteria. In this situation, how will you apply your eligibility criteria? Should the study be included or excluded? You have several options for how to treat populations where only a subset meets your criteria: 1. You could include the whole study – although this will mean some participants who don’t meet your criteria will be excluded, it means you won’t lose all the information about those participants who do meet your criteria. You may wish to select a threshold or cut-off point for this – for example, you could decide to include any study in which at least half, or at least 80% of the population meets your criteria. 2. You could decide to only include information on those participants who meet your criteria. This ensures that you don’t include any information from ineligible participants. However, detailed information on which participants met your inclusion criteria and separate results for those participants may not be available. You can try asking the authors of the study for separate information, but you may not be able to obtain it. 3. Finally, you can decide to exclude the whole study. That is, you can decide that the risk of including some inaccurate information from ineligible participants outweighs the benefit of including the study at all. If possible, it is good practice to specify in your protocol how you plan to manage studies where only some of the participants meet your eligibility criteria, particularly where you know there are many mixed studies in your field (such as reviews of joint replacements that often combine kip and knee replacements), or where you have given a cut-off point (such as an age group) that you know will not always match the cut-off points of your included studies. A rationale should be given for why you have chosen a particular approach. Remember that it’s not always possible to predict when this situation will arise, and that in some cases, although post hoc decisions should be avoided, it may be better to consider studies on a case by case basis, rather than stick to an arbitrary rule. When making post hoc decisions, you should consider the overall aims and rationale for defining the scope of your review, and your decisions should be documented in your review. You may wish to conduct a sensitivity analysis on your results to assess the impact of your decisions on the review’s findings.

12 특정 인구집단과 형평성 Equity and special populations
특정 인구 집단에 대한 형평성 및 연관성이 문헌고찰에서 중요한 요소인지 고려할 것 성별, 연령, 인종, 지역적 / 경제적 상태, 교육 정도 등 왜 이러한 요소들을 고려해야 하는가? 유병률, 질병의 진행 및 건강에 미치는 영향력 등이 각 집단별로 다를 수 있음. 해당 중재법의 효과 및 안전성이 각 집단별로 다를 수 있음. 결과 평가 변수의 중요도가 각 집단별로 다를 수 있음. Cochrane reviews seek to be relevant internationally and across contexts, and it’s important to consider your question in relation to all the different types of people who may be affected. Prevalence and progress – e.g. antibiotics for sore throat – may not be useful in high-income settings, where impact of the disease is not great. In some low-income countries there is a greater likelihood of progression to rheumatic fever, with very serious consequences, such that you might come to a different conclusion about the importance of intervention. Different effects or safety – e.g. some interventions may work differently in different cultural settings (e.g. communication and education interventions), or there may be biological differences in the way interventions work among different ethnic groups, in women compared to men, or in children compared to adults. Different outcomes of importance – e.g. for interventions in children, developmental milestones or attendance at school may be important outcomes that don’t apply to adults. You may need to measure outcomes differently in some populations, e.g. measuring pain in newborn babies. The cost of an intervention may be very important for people without health insurance, or for policy makers planning implementation in resource-poor settings. The duration of treatment required may be important for participants in rural settings who travel long distances to access treatment. Although these may not form part of your eligibility criteria, they will assist you in considering other aspects of your question as you plan your review.

13 중재법 Intervention 최대한 상세히 서술할 것 중재법 제한 시 명확하고 합리적인 기준을 제시해야 함
중재법의 효과를 발현시키는 활성 구성 요소 용량, 강도 수행 방법 / 투여 경로 수행/투여 시간, 빈도, 기간 중재 수행에 사용된 의료기구/장비 중재 수행 인원 (자격, 교육 기간) 중재가 수행된 지역 및 맥락 단독 투여/수행 혹은 기타 중재법과 병행 여부 중재법 제한 시 명확하고 합리적인 기준을 제시해야 함 혹은, 중재법의 제한 없이 문헌고찰 대상으로 포함시킨 뒤 하위그룹 분석 (Subgroup analysis) 를 수행하는 방법도 있음 Next, you will need to describe the intervention in enough detail to distinguish your intervention of interest from those interventions that fall outside the scope of the review. So, if your review is looking at a very specific topic, such as a specific dose of a specific drug, you can be very narrow in your definition. If you are looking at a more general topic, such as a class of drugs, or a more complex, multi-component intervention, you can be less detailed in your criteria, allowing for expected variations in the literature. However, you must be able to clearly identify the core or common factors that determine whether each study, each variation, falls within the review, or is so different that it should be excluded – whether this is a minimum clinically effective dose or duration of drug treatment, or essential components of a complex intervention. Based on the objective of your question, think carefully about the advantages and disadvantages of including studies that test your intervention in combination with other interventions. If you are interested in the direct effects, say, of phsyiotherapy for shoulder pain, then a study that looks at physiotherapy in conjunction with painkillers will not answer that question, because the effects of physiotherapy are confounded with those of painkillers . Alternatively, you might be interested in the effects of multi-component interventions, in which case a study of physiotherapy plus painkillers might be appropriate. It is particularly important to be as clear and informative as you can when describing complex interventions, such as multi-component clinical interventions like rehabilitation programs, educational or behavioural interventions, community-based interventions, etc. These are often poorly described in the literature, which makes it difficult to make decisions about inclusion, and difficult for users of the review to implement the intervention if it is found to be effective.

14 지역 및 맥락 Location and context
중재법은 특정 환경에서는 효과적일 수 있지만, 다른 환경에서는 그렇지 않을 수 있음 적용 가능성 및 접근성 의료기구 및 장비 활용가능한 의료진의 경험 및 숙련도 우선 해결 필요 과제의 지역별 차이 비용 지불 체계 문화적, 언어적 다양성 사회경제적 지위 전원/도시 환경 Again, it’s important to consider your intervention and how it might be applied internationally, and in a variety of contexts. Interventions may work differently in some contexts, or need to be applied differently. Factors you may wish to consider can include: whether the intervention is available in every context, or does it depends on specialised equipment itself that may not be available, where that’s an expensive scanning machine or something simple like running water? are there specific skills or training that the staff implementing the intervention will need? Will different staff be implementing the intervention in different settings? is the intervention acceptable in all cultural contexts, and will it work the same way? is the intervention affordable to people with different incomes, or under different insurance systems? See Section of the Handbook

15 대조군 Comparison 문헌고찰의 목적에 기반할 것 중재법과 마찬가지로, 대조 중재법의 구성요소를 최대한 자세히 정의할 것
‘무처치 대조군’ 일 경우, 그 의미를 명확히 할 것 예) 어떠한 치료도 제공하지 않음, 플라시보 대조군, 일상관리군 등 확인된 대조군을 모두 포함할 수 있지만, 그 내용을 명확히 할 것 The comparisons you’re testing against your intervention of interest should be clearly based on the objective of your review question. If you review is comparing the intervention against a specific active alternative (or more than one), e.g. one drug versus another, or health professional versus peer-led interventions for mental illness, the comparisons should be described in as much detail as the intervention. You may be interested in the effects of caffeine compared to any alternative intervention, or no intervention at all. In which case, you don’t need to specify a specific comparison for your eligibility criteria. Even if your objective is to test your intervention against no intervention, remember that may be important differences between inactive comparisons – such as truly doing nothing, or using a placebo, or leaving in place whatever ‘usual care’ is for this population. Remember, too, that ‘usual care’ may vary a lot from study to study, so plan to examine these differences carefully if this applies to your review. Again, remember to consider the real choices available to people making decisions in this area – what are the options currently available in different contexts? Have some intervention options become obsolete (remembering that interventions that are obsolete in high-income countries and large urban hospitals may still be in use in other places)? If your review is broad, you can choose to include studies comparing your intervention to any possible comparison – if that’s the case, make sure you state this clearly.

16 결과 Outcomes 대체로 선정 기준 항목이 아님 연구계획서 단계에서 명시할 것
결과 항목의 보고 여부로 연구를 배제시키는 것은 비뚤림 유발 가능 연구결과는 임상시험 저자에 의해 선택적으로 보고되었을 수도 있음 (저자에게 연락하여) 추가 자료를 얻을 가능성도 있음. 연구질문의 정의 상 결과 항목이 중요할 경우, 선정기준에 포함시키는 것이 적절할 수도 있음 예방 vs 치료, 최소 한 가지 이상의 질환에 사용된 중재법 등 연구계획서 단계에서 명시할 것 Moving on to outcomes, note that it is not always expected that outcomes will form part of the eligibility criteria for a review. Commonly, studies that do not measure your outcomes of interest are still included and noted in the review, although they cannot contribute to your analysis of those outcomes. It is important to include these studies so that you are presenting a complete picture of the literature. It’s particularly important to consider the possibility of outcome reporting bias. ASK: Can anyone explain the meaning of outcome reporting bias? You are probably familiar with the concept of publication bias – that is, studies with positive or significant findings are more likely to be published than non-significant studies. That also applies within studies to the reporting of results for individual outcomes - outcomes with positive or significant findings are more likely to be fully reported in a paper, which means they’re more available to be included in your review. Given that, it’s important to consider the possibility that your outcomes of interest may have been measured by a study, even if they were not reported. By excluding these studies without exploring further, you run the risk of introducing bias to your review. Don’t forget, also, that it may be possible to contact the authors of a trial and obtain additional information for these studies. In some cases it is appropriate to include the outcomes in your eligibility criteria – for example, if you are interested in prevention, then your included studies must measure prevention, and not only treatment outcomes. For interventions with many applications, you may wish to define your review to cover one particular field – the Cochrane Library currently has separate reviews looking at hormone replacement therapy for menopausal symptoms, for cardiovascular outcomes, and cognitive outcomes. Whether your outcomes are part of the eligibility criteria or not should be clearly stated in your protocol. Source: Dwan K, Gamble C, Kolamunnage-Dona R, Mohammed S, Powell C, Williamson PR. (2010) Assessing the potential for outcome reporting bias in a review: a tutorial. Trials. 11:52

17 결과 Outcomes 유의미한 결과 항목 설정
일반 사용자, 의료진 및 보건의료 전문가, 정책결정자 등을 위함 이상반응을 결과 항목에 포함시킬 것 각 인구집단 별로 연관성 있는 결과 항목을 포함시킬 것 핵심적 평가 시점 설정 수용 가능한 평가 도구 사용 (정의, 등급 등) 간접적 (trivial) 결과 평가도구를 지양 (생화학 지표, 대리표지자, 과정 평가 등) 핵심적 결과 (core outcome) 및 연관된 기존 문헌고찰에서 사용된 결과 항목의 사용을 고려 여러 유사 결과 항목 존재 시, 사전에 취사선택 계획 수립 중요한 결과 항목이라면, 자료 검색 가능성 여부와 상관없이 연구계획서와 문헌고찰에 포함되어야 함. Plan to report outcomes of importance to people making decisions, starting with those people who are affected by the condition, and those who are making decisions about care. It’s important that you plan in advance which outcomes you will report, to minimise bias in the selection of outcomes when you come to complete the review. Some fields have articulated key outcomes, such as the work done by OMERACT for rheumatology studies – if so, take advantage of these standards. If not, you may wish to look at other reviews in similar areas to see the outcomes they report. Don’t forget to always include possible adverse effects. You might consider how outcomes might be measured – for example, are there published, validated measurement instruments for your outcomes? Consider the variety of ways in which an outcome may be measured and reported in the studies you find, and only be restrictive if you have a good rationale – for example, the latest validated test might be the best, but if it was only released last year, or has not been widely adopted internationally, it may exclude the majority of studies. You may also specify important time points for the measurement of outcomes. Is there a minimum period of time before you could expect the intervention to have an effect? e.g. interventions to improve bone density will take longer than a few weeks. Note that it is not always expected that outcomes will form part of the eligibility criteria for a review. Commonly, studies that do not measure your outcomes of interest are still included and noted in the review, although they cannot contribute to your analysis of those outcomes. It is important to include these studies so that you are presenting a complete picture of the literature. In some cases it is appropriate to include the outcomes in your eligibility criteria – for example, if you are interested in prevention, then your included studies must measure prevention, and not only treatment outcomes. For interventions with many applications, you may wish to define your review to cover one particular field – the Cochrane Library currently has separate reviews looking at hormone replacement therapy for menopausal symptoms, for cardiovascular outcomes, and cognitive outcomes. Select your outcomes based on the information you think is important, not based on what you think you will find in the literature. For example, if you’re interested in mortality, include that as an outcome, even if it’s a long-term outcome that few studies will measure. Short-term surrogate outcomes, such as improvement in the regularity of heartbeat, may be easier to measure, but don’t always correspond to longer-term outcomes in the way you expect. You will be expected to report on all your planned outcomes in the final review, even if you don’t find any studies that measure some outcomes – picking and choosing outcomes based on those with good available data can introduce bias, and reporting that outcomes of importance have not been adequately measured is a useful finding.

18 결과 항목의 우선순위 설정 Prioritising outcomes
일차 평가 변수 (최대 3개) 보통 최소 1개 이상의 발생 가능한 이상반응(harm)을 포함시킴 이차 평가 변수 일차 평가 변수 외의 주요 평가 변수 중요도가 떨어지는 기타 평가 변수 주요 평가 변수 (최대 7개) 의사결정에 필수적임 결과분석 및 요약본 작성의 기초가 됨 You should specify a small number of primary outcomes, that will form the basis of the conclusions of your review, including at least one potential harm. The rest of your outcomes will be considered secondary outcomes, but keep in mind that the total number of outcomes to be addressed should be kept as small as is reasonable for your question, and focus on those outcomes that are important for decision-making. It’s important to prioritise your outcomes to help you in reporting the results of your review, particularly when it comes to summarising your results. Cochrane reviews have a range of short formats to summarise results, such as the abstract, plain language summary, or Summary of Findings table of your review. You may not be able to report against all your primary and secondary outcomes in each summary, so we recommend that you identify a group of outcomes – including your primary outcomes and perhaps a small number of your secondary outcomes – that can be designated as your main outcomes, and prioritised in any summary of your results. These should be the outcomes that are most important for decision-making, whether or not you expect to find enough data in the literature. Also, try not to duplicate within your main outcomes – avoid including multiple variations on the same outcome (e.g. weight, BMI), or measures of the same outcome at multiple time points, without a clear rationale. Additional variations and time points can be included among your other secondary outcomes.

19 특수한 평가 변수 유형 Special outcome types
관련 자료 및 도움말 제공 웹사이트 Cochrane Adverse Effects Methods Group Cochrane Patient Reported Outcomes Methods Group Campbell and Cochrane Equity Methods Group Campbell and Cochrane Economics Methods Group Cochrane Qualitative Research Methods Group see At these early stages, it’s appropriate to consider whether the important outcomes for your review incorporate some of these outcome types. Cochrane reviews allow for more in-depth consideration of outcome types that are commonly encountered, such as adverse effects and issues of equity. Alternatively, you may wish to incorporate specialised outcome types such as economic or qualitative evidence. Consideration of these topics may require additional systematic literature searching, and more advanced analysis methods. These topics will not be covered in depth in this course, but further information is available in the Handbook, and on the websites of the Methods Groups who contribute to the Cochrane Collaboration in these areas – website details are available on the cochrane.org website. See Chapters 14,15,17,20 of the Handbook

20 연구자료의 디자인 Study designs
연구질문의 해결에 가장 적합한 연구 디자인 선택 디자인 선택의 논리를 항상 제시할 것 대부분의 코크란 문헌고찰에 포함된 디자인: 무작위 대조군 연구 (randomised controlled trials) 각 군간 계통적 이질성의 발생을 방지 종종 비무작위 대조군 연구도 포함 (유사-무작위 대조 연구; quasi-randomised) ASK: who was under the impression that Cochrane reviews only include randomised trials? In fact, Cochrane reviews mostly include RCTs, but in some cases we do include other study designs. We always aim to include the most appropriate study designs to answer the review question. For straightforward questions about the effects of interventions, that will usually be RCTs. Methodological research tells us that RCTs are the best way to avoid bias introduced by systematic differences between the groups in a study. Usually that’s all you need to specify for your eligibility criteria. In some cases you might want to be specific about an aspect of study design for example if you wish to specifically exclude crossover trials from your review , or in some cases we also include quasi-randomised trials (those trials with methods of allocation that are almost random but not quite, such as alternating allocation).

21 비-무작위 대조 연구 Non-randomised studies
해당 CRG의 동의가 있어야 함 명확한 이유가 있어야 함 RCT의 수행이 부적합하거나 비현실적인 분야 (예, 공공 보건, 복합적 보건의료 체계를 다루는 연구 등) 특정 결과 평가 변수 측정 시 (예, 이상반응, 경제성 평가, 질적 평가 변수 등) 단순히 RCT가 없다는 이유는 불가 비무작위 연구의 포함은 특정 연구디자인에만 한정될 수 있음 예; 전후 대조 연구 (controlled before-and-after), 간여 시계열 분석 (interrupted time series) 비무작위 연구 디자인 포함 시, 연구 디자인에 대한 최소한의 선정기준을 세울 것 비뚤림 위험이 증가함을 염두에 둘 것 Some Cochrane reviews will look at non-randomised studies. You must have a clear rationale for doing this – that is, why would RCTs not be able to provide this evidence? It is not enough to say that you don’t believe there are any RCTs measuring this outcome. If an RCT is the most appropriate way to measure an effect, and there are none, then Cochrane reviews can conclude that there is no appropriate evidence, and that new studies should be conducted. Unlike a guideline or a local policy, we are not obliged to provide a recommendation if we don’t believe the evidence is sufficient. However, in some cases other study designs are appropriate to answer questions. For example, in some cases RCTs may not be appropriate, e.g. where they might be unethical, or seriously impractical, such as in the case of large-scale changes to health system structures, or public health interventions across whole communities or even countries. There are also some outcomes that RCTs are not able to measure well. RCTs are not always good at measuring adverse effects, particularly rare or long-term effects. In those cases, you might need to look for cohort or case-control studies, perhaps government monitoring data, etc. If you’re incorporating economics or qualitative data in your reviews, these have their own study designs that you may wish to consider. These kinds of studies aren’t usually incorporated in your primary analysis of the effectiveness of the intervention, but can be used to inform and interpret the results. ASK: Is anyone planning to include non-randomised studies in their review? Whenever you plan to include non-randomised studies, you must have the agreement of your CRG – not all CRGs will be happy for you to do this. You should also specify which non-randomised designs you will include. Further advice on this is available from your CRG, including recommended minimum design criteria for non-randomised studies. Whenever you’re including non-randomised studies, you must be aware of the increased risk of bias in the results. Make sure you are familiar with the risks, how to assess them, and how best to analyse the results of these studies. See Chapters 13, 14,15,20, 21 of the Handbook

22 연구 제목 설정 Turning a question into a title
코크란 문헌고찰은 표준화된 제목 형식을 사용함. ‘질환/주제’에 대한 ‘중재’ 급성 기관지염에 대한 항생제 사용 (Antibiotics for acute bronchitis) 신체활동 증진을 위한 지역사회 중재 (Community-wide Interventions for increasing physical activity) 기타 세부항목 포함 가능 자궁경부상피내종양에 대한 즉시치료 대 지연치료 비교 (Immediate versus delayed treatment for cervical intraepithelial neoplasia) 미숙아 호흡 부전에 대한 흡입 산화질소 요법 (Inhaled nitric oxide for respiratory failure in preterm infants) 아동 익사 방지를 위한 수영장 담장 설치 효과 (Pool fencing for preventing drowning in children) The titles of Cochrane reviews are usually based on the standard format: ‘intervention’ for ‘population’. In some cases there are variations, for example if you wish to include a specific comparison, a specific population, or if yours is a review of a prevention rather than treatment intervention. This will be agreed with your CRG at the title registration stage.

23 개요 Outline 잘 정의된 연구 질문의 중요성 연구 질문의 구성 요소
연구 계획서 (protocol) 에 포함되어야 할 내용

24 연구 계획서에 포함되어야 할 내용 What to include in your protocol
배경 (Background) 해당 질환/상태 및 치료/중재에 대한 구체적 기술 연구방법 (Methods) 해당 문헌고찰에 포함될 연구자료의 선정 기준 선정 기준의 명확한 기술 사전 계획된 평가 변수의 명기 자료 수집 및 분석 사전 계획된 하위그룹 분석 (subgroup analysis) 명기 There are several places in your protocol where the details of your question are explained. In the Background, you can give your detailed descriptions of the condition of interest, including the burden of disease, the population, and how the condition affects people. In the Methods section, your eligibility criteria are listed in the section called ‘Criteria for considering studies for this review’. Your list of outcomes to be reported is also included here, even if they are not part of your eligibility criteria. Finally, those additional factors and variations that you want to explore within your review can be planned as subgroup analyses in the section called ‘Data collection and analysis’. We’ll be talking in more detail about analysis methods in separate sessions.

25

26

27

28

29 핵심 전달 사항 Take home message
연구 시작 전, 문헌고찰의 내용 및 범위 (scope)를 숙고할 것 (예; PICO 모델 활용) 이를 통해 얻은 정보를, 명확하고 구체적인 선정기준 설정에 활용할 것 이 정보는 검색 전략, 자료 수집 및 분석 방법을 결정할 때에도 도움이 될 것임.

30 References Acknowledgements
O’Connor D, Green S, Higgins JPT (editors). Chapter 5: Defining the review question and developing criteria for including studies. Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version [updated March 2011]. The Cochrane Collaboration, Available from Acknowledgements 편집: Miranda Cumpston 호주 코크란 센터 (Australasian Cochrane Centre) 교육자료를 원자료로 함 본 교육자료는 Cochrane Method Board가 승인하였음 Translated by Kun Hyung Kim, Myeong Soo Lee and Byung-Cheul Shin