Published on November 12, 2018

Starting a clinical study is no simple task. From patient recruitment to study completion, Principal Investigators (PI) must undertake a wide array of tasks. In this article, we will look at what should be considered before beginning  your clinical data collection. This guide will in no way be comprehensive, but will provide you with valuable info to get started.

1. Define your hypothesis and endpoints in a proper way

Make sure you define your endpoints using the correct scientific language. For example, your research question could be: “Is treatment A better than treatment B?” You should then define your primary endpoint based on a reasonable estimate of what effect you expect to measure. For example, you could define your endpoint as `10% of participants will have a lower viral load 8 weeks after treatment.’ Secondary endpoints could be secondary factors that you deem important for your research question, such as quality of life. It is important to remember that a scientific study will never prove a hypothesis is true. It can only reject the null hypothesis. A good researcher will also show he/she understands the limitations of the scientific method. Read about Ronald Fisher’s famous Lady Drinking Tea experiment to better understand this.

2. Include adequate bias control measures

The reliability of clinical study results depends on the extent to which potential sources of bias has been avoided. Sources of bias include:

  • Reporting bias also known as outcome reporting bias or selective reporting bias – refers to the bias toward significant outcomes being reported more often than insignificant ones. This can be mitigated by using an EDC and making required fields.
  • Selection bias refers to when your reference groups significantly different in terms of characteristics. This can be mitigated through correct and randomisation.
  • Performance bias refers to when there is significant variability in the care that is provided, leading to exposure to other variables than those of interest. This can be mitigated by proper blinding of your study, to make sure that all subjects receive equal attention.
  • Detection bias refers to the differences between groups in how outcomes are determined and can occur when measuring more qualitative variables such as level of pain. This can again be mitigated by proper blinding.
  • Attrition bias refers to when research data becomes incomplete because of disproportionate withdrawals or exclusions from one study arm. Pray this does not happen.

Phd Humor


3. Define adequate inclusion criteria and stick to your inclusion criteria

Before starting your recruitment process it is important to clearly specify your criteria for inclusion and exclusion. This will later tell other researchers how your study may differ from other studies. For example, your study might target a different age group or disease state than other studies. Once you have defined your inclusion or exclusion criteria, make sure that you actually follow these during patient recruitment. Failing to do so may lead to exclusions later on, or invalidate your study.

4. Make sure your study is ‘powerful’

Many clinical studies do not reach their endpoints because they are underpowered. In other words, they have too few subjects to measure the necessary effect size. Read up on the relationship between power, Type II error, and effect size before deciding how many patients you will need. There are also many software programs available that will allow you to calculate the size of your sample. Don’t forget to save your power size calculation, as you will need to report it later in your methods section.

5. Decide on how to collect the data

It is important that you think through how you will collect your data. Will you use paper or an EDC? Is your study dependent on a clinician filling out Case Report Forms (CRFs) or will you depend on direct patient feedback through Patient Reported Outcome (PRO) surveys? In either case, it is important that you have the forms and systems in place to start collecting data as soon as your first patient is recruited.

Phd Humor

Source :

Stop losing study opportunities to poor data capture! Sign up for our webinar to learn how standardizing data for reuse opens up new research possibilities.

6. Discuss your timeline and budget with the Project Bureau

Many academic hospitals have a project bureau that can assist researchers with timelines and budget planning. Researchers often see this step as a barrier to starting as soon as possible. While this does take time, the project bureau can provide feedback on whether your timeline and budget are realistic. Getting these validated can save you from enormous trouble later on. Once you have created your timeline and budget, make sure to stick to them!

7. Site selection

Does your study include multiple sites? Think hard about what sites fit well with your research and protocol. The sites you select and how well they are able to perform the intervention has a direct impact on the quality of your data. Define your site selection criteria. These could include staff qualifications, facilities and equipment, access to patients, and past trial performance. If different sites perform the intervention in a different way-for example because they work with different equipment-make sure this is covered in the protocol. Failing to do so will lead to protocol deviations.

Phd Humor

Source :

8. Make sure you have the buy-in from the different departments that you are working with

Before you start your study, make sure that the different departments you work with are aware of your study and agree to spend time on your patients. For example, if your study requires an extra scan from radiology or additional tests from hematology, make sure that the staff that needs to perform the tests are informed. Forgetting to do so could lead to unexpected delays.

9. Work with properly trained personnel

Make sure it is clearly documented who will perform your study and that they have received the proper training before the study starts. Training is especially important when it comes to the reporting of adverse events (AEs). Performing a clinical study is highly regulated and protocol deviations (or worst violations) should be clearly documented. The better trained the staff is, the fewer additional paperwork and stress they will encounter.

10. Do not underestimate how hard it is to recruit patients

You as PI should expect to be involved in the recruitment for your study. Once a patient agrees to be contacted for the study, you will most likely perform the informed consent procedure.

Note that the less common it is to perform your intervention, the harder it will be to recruit patients. Plus, not all patients will consent to participate in a clinical study. If you need to halt your study because you cannot enrol enough patients, this looks really bad. In addition to your inclusion criteria, it is very advisable to create a plan for effective patient recruitment.

Phd Humor


Finally, remind yourself that, ultimately, you are running a clinical study to help patients. Patients have privacy and information rights before, during, and after the study. Make sure you have systems and procedures in place that adequately safeguard patient privacy, request consent and provide adequate information.


Find out how Castor EDC can help you capture high data quality

Learn More