Endpoints and Outcomes

Endpoints and outcomes

Trials should have a clear primary outcome. Typically, this is the difference between the intervention and control groups in a specific variable – and will determine if your intervention is effective or not. The primary outcome should be an observation of significant clinical relevance – either to patients, doctors or to the health service (ideally, to all three groups). Examples of primary outcome include [difference in… or difference in change in…] blood pressure, recurrence of kidney stones, quality of life, pain score, death or albuminuria.

The ideal primary outcome is important to all stakeholders – especially to patients. Typical examples of patient-centred (i.e. important to patients) outcomes are adverse clinical events (such as infection, hospitalisation or death) or patient-reported outcome measures (PROMs) – typically captured by structured survey-type instruments like the Epworth Sleepiness Scale or the EuroQOL Five Dimensions. These tools provide a quantitative measure of the disease burden to patients.

Primary outcomes that are not directly important to the patient are called surrogate outcomes. These outcomes measure variables known to be associated with adverse events or other events important to patients. Good examples are blood pressure and albuminuria, which typically cause patients no distress but which are treated as we believe that lowering their values leads to better patient outcomes such as fewer cardiovascular events or to the preservation of renal function. Surrogate outcomes are common primary outcomes in nephrology as patient-centred outcomes such as end-stage kidney disease may take many years to develop. As such, it is more feasible to measure change in surrogate variables with the expectation that the surrogate outcome will be linked to the hard, longterm outcome. However, changes in surrogate outcomes due to new treatments do not always lead to the desired changes in clinical outcomes. This can only be proved by a long-term trial that actually measures the clinical outcome.

  • How is it measured? Is this measurement reliable?
  • Is this outcome the preferred measure as specified by guidance documents? For example, it is suggested that blood pressure be measured in a quiet room, in the seating position with the cuff at the level of the heart and that three measurements are taken 2 minutes apart.
  • Does your outcome have a standard definition (eg. an International Classification of Diseases code)? How have other researchers defined it?
  • Is your endpoint recommended by consensus statements such as the Standardized Outcomes in Nephrology (SONG Initiative)?
  • If it is a PROM, is the PROM validated in your population?
  • Is it important that the people assessing the outcome are blinded to the patient’s treatment allocation?
  • How much variability is there between different outcome assessors? For example, if you have two people measuring blood pressure or two people conducting a survey on patients, are these two people doing it in the same way? Do you need to train them together to align their techniques?
  • Is it a surrogate outcome and if so, what is it a surrogate measure of?
  • Is it a patient-centred outcome?
  • How will the outcome be measured?
  • How many missing values might you expect? Will there be participants who cannot come to the follow up visits to have their outcomes measured?
  • Do your machines need to be calibrated?
  • How have other researchers measured this outcome? It is advisable that you measure the outcome in the same way. This allows your study to be properly compared and to form part of a larger meta-analysis.