? Consider the audience of the review
? Think about what you already know
? Use a framework
You’ve decided to go ahead. You have identified a gap in the evidence and you know that conducting a systematic review, with its explicit methods and replicable search, is the best way to fill it – great choice ?.
The review will produce useful information to enable informed decision-making and to improve patient care. Your review team’s first job is to capture exactly what you need to know in a well-formulated review question.
At this stage there is a lot to plan. You might be recruiting people to your review team, thinking about the time-frame for completion and considering what software to use. It’s tempting to get straight on to the search for studies ?.
Take it slowly: it’s vital to get the review question right. A clear and precise question will ensure that you gather the appropriate data to answer your question. Time invested up-front to consider every aspect of the question will pay off once the review is underway. The review question will shape all the subsequent stages in the review, particularly setting the criteria for including and excluding studies, the search strategy, and the way you choose to present the results. So it’s worth taking the time to get this right!
Let’s take a look at five key steps in formulating the question for a standard systematic review of interventions. It’s a process that requires careful thought from a range of stakeholders and meticulous planning. But what if, once you have started the review, you find that you need to tweak the question anyway? Don’t worry, we’ll cover that too ✅.
Who will use this review? What do they want to know? How do they measure effectiveness? Good review teams partner with the people who will use the evidence and make sure that their research plan (or protocol) asks a question that is relevant and important for patients.
How much do you need to know about the topic area at this stage? Ideally, enough to come up with a relevant, useful question but not so much that your knowledge influences the way in which you phrase it. Why? Because setting a review question when you are already familiar with the data can introduce bias by allowing you to direct the question in favour of achieving a particular result. In practice, the review team is very likely to have some knowledge of relevant studies and some preconceived ideas about how the treatments work. That’s fine – and it’s useful – but it’s also good practice to recognise the influence this knowledge and these ideas might have on the choice of question. Issues of bias will come up again as we work through the rest of these steps.
If not enough is known about the subject area to ask a useful question, you might undertake a scoping review. This is a separate exercise from a systematic review and is sometimes used by researchers to map the literature and highlight gaps in the evidence before they start work on a systematic review.
Faced with a heady mixture of concepts, ideas, aims and outcomes, researchers in every field have come up with question frameworks (and some great backronyms) to help them. Question frameworks impose order on a complex thought process by breaking down a question into its component parts. A commonly used framework in clinical medicine is PICO:
? Population (or patients) refers to the characteristics of the people that you want to study. For example, the review might look at children with nocturnal enuresis.
? Intervention is the treatment you are investigating. For example, the review might look at the effectiveness of enuresis alarms.
? Comparison, if you decide to use one, is the treatment you want to compare the intervention with. For example, the review might look at the effectiveness of enuresis alarms versus the effectiveness of drug therapy.
?Outcomes are the measures used to assess the effectiveness of the treatment. It’s particularly important to select outcomes that matter to the end users of the review. In this example, a useful outcome might be bedwetting. (Helpfully, some clinical areas use standardised sets of outcomes in their clinical trials to facilitate the comparison of data between studies ?.)