A guide to literature search and review for medical device regulatory submissions

When preparing for a regulatory submission, a thorough literature search and review is critical to ensure your medical device meets safety and performance standards.

For successful conformity assessment, the literature search and review process must follow a systematic and well documented process. In this post, we’ll walk through the essential steps to do so for medical devices.

What is a literature search and review?

A literature search and review is the process of systematically finding, selecting, appraising, and synthesising existing research to answer key questions about the safety and performance of your medical device. For regulatory purposes, this review provides evidence to support claims about your device and ensures that it is based on current knowledge, and addresses any identified risks.

Key steps in conducting a literature search and review

A structured literature search involves several key steps, ensuring that you identify the relevant studies, assess their quality, and apply their findings to your regulatory submission.

1. Define your objectives and research questions

The first step in any literature search is to define clear objectives and research questions. These questions will guide the search and ensure that you gather the necessary information to support your device’s regulatory submission.

The literature search might cover:

• State of the Art
• Similar Devices
• Alternative Approaches
• Your Device or an Equivalent Device
• Safety and Risk Management

It’s important to remember that “State of the Art” here does not mean the most novel or technologically advanced solutions, but rather what is the current and generally accepted good practice in the field.

Research questions should be carefully constructed using a process such as PICO - Population (or disease/condition), Intervention, Comparator/Control, Outcome.

For example, imagine an AI-based screening software for diabetic retinopathy which automatically analyses retinal images captured by ophthalmologists, detecting early signs of the disease. A PICO statement for a State of the Art literature search might look like:

• P - Adults with diabetes at risk for diabetic retinopathy
• I - Retinal exam or image analysis
• C - (not applicable)
• O - Diagnostic accuracy, sensitivity, specificity, time efficiency, clinical integration

2. Identify data sources

Choosing the right databases and sources is essential for a comprehensive literature search - and multiple databases should be searched. Sources should include scientific literature databases as well as internet and non-published data sources.Sources should include scientific literature databases as well as internet and non-published data sources.

Scientific literature databases

Some common scientific literature databases include:

• MEDLINE or PubMed: A broad source of biomedical literature, including research articles and reviews. It’s a good starting point but it may miss some European studies
• EMBASE: Contains a large number of records from international sources (95 countries), including Europe, making it useful for comprehensive searches
• Cochrane CENTRAL: High sensitivity for identifying randomised controlled trials
• PROSPERO: Includes prospectively registered systematic reviews with health related outcomes including interventions and diagnostic accuracy

Additional considerations:

• Domain-Specific Databases: Such as PsycINFO for psychological studies or CINAHL for nursing and allied health literature.
• Regional Databases: Databases specific to certain regions or countries to capture local studies.

Internet and non-published data sources

The internet search should complement database searches and might include sources such as:

• Manufacturer websites: Look for information on the manufacturers of similar devices. They may have data, technical reports, or updates relevant to your search.
• Professional bodies: Websites for medical associations or regulatory bodies often provide useful guidelines and updates.
• Regulatory authorities: Check for field safety corrective actions or other data published by European Competent Authorities, the U.S. FDA, or similar organisations.

Reference searching

References in the identified literature may also be relevant and should be screened. As information coverage in databases and other sources can change over time, the data sources for your literature search will need to be justified and re-evaluated over time.

3. Develop Search Terms

Once your objectives and data sources are clear, the next step is to develop tailored search terms for each data source. The addition of boolean logic and filters can help narrow the search to identify the most relevant studies while minimising retrieval of irrelevant ones.

Note that constructing complex search queries varies between databases. Always check the documentation of each database on how to correctly build these queries.

4. Screen and select studies

Not all identified studies will be relevant, so it’s important to establish clear inclusion and exclusion criteria as part of your literature search protocol. These criteria should be applied consistently across all literature. Screening is a multi-step process that ensures you focus on high-quality studies that address your research questions.

Steps in Study Selection:

1. Title and Abstract Screening: Review titles and abstracts to quickly assess whether the studies are relevant.
2. Full-Text Screening: For studies that pass the initial review, examine the full text to determine their suitability based on the predefined criteria.

5. Critically Appraise Studies

Once the relevant studies are selected, a critical appraisal assesses the quality, validity, and reliability of the findings. This step is crucial for ensuring that your submission is based on high-quality evidence.

There are two key domains to appraise:

1. Methodological quality and scientific validity
2. Relevance to the clinical evaluation

Methodological quality and scientific validity considerations include aspects such as confounding influences, bias, adequacy of the sample size and endpoints, blinding of users or assessors, and adequacy of the follow up period.

Relevance to the clinical evaluation considerations include if the study is conducted with the device under evaluation and representativeness of the target population or intended users.

There are many established tools and resources for critical appraisal including, but not limited to:

• Cochrane Methods Network
• GRADE (Grading of Recommendations, Assessment, Development, and Evaluations)
• International Medical Device Regulators Forum Clinical evaluation resources
• CASP (Critical Appraisal Skills Programme) checklists

6. Extract data

Relevant data need to be extracted from each publication. Ensure there are strategies to ensure data integrity during extraction such as the use of quality control or a second review of extracted data by an additional reviewer as mentioned in MEDDEV 2.7/1 rev 4 Annex A5.3.

7. Synthesise findings and report results

Finally, you need to synthesise the findings and report the results of the literature review. This should be done in relation to the original research question, but also include safety and performance benchmarks. Even studies excluded on the basis of quality may contribute to identification of risks. It is important to pay attention to and address duplication of data across multiple studies.

Everything from the search protocol, critical appraisal and findings should be documented thoroughly.

Summing up

Conducting a comprehensive literature search and review is an essential part of preparing for medical device regulatory submissions. By following a structured approach—defining research objectives, developing a search strategy, identifying the right sources, screening and appraising studies, extracting relevant data, and synthesising the findings—you ensure that your submission is built on a solid foundation of high-quality, relevant evidence.

To learn about common mistakes when it comes to literature search and review, and how to avoid them, see our post on the topic here.