The median is a statistical measure of central tendency and refers to the value that divides an ordered data set into two equal halves.
In clinical trials, registry analyses, and health economics, the median is frequently used because it is more robust against outliers compared to the arithmetic mean.
Definition and calculation
To calculate it, observations are first sorted. For an odd number of values, the median is the middle value. For an even number, the average of the two middle values is typically used.
It is important to note that the median describes the rank position and not the “typical” deviation from the center; for this purpose, additional measures such as interquartile range or range are reported.
Why the median is so common in clinical research
Many clinical variables are skewed (e.g., length of hospital stay, costs per patient, biomarker values). In such cases, the median provides a stable metric that is not dominated by a few extreme values.
In reports, the median and interquartile range are often presented together to provide a clear picture of both the center and the dispersion.
Median in time-to-event analyses (survival analysis)
In survival analysis, the median survival time is frequently reported. It is the point in time at which 50% of the study participants have experienced an event (e.g., death, progression).
If fewer than 50% of events are observed, the median survival time may be “not reached.” In such cases, alternative metrics such as hazard ratio, event rates, or survival probabilities at defined time points are useful.
Interpretation, limitations, and typical misunderstandings
A common misunderstanding is to read the median as the “average.” While the arithmetic mean takes all values into account, the median is based exclusively on the ranking and the middle position.
Furthermore, the median alone can mask different distributions. Two groups can have the same median but significantly different dispersion or different proportions of extreme values. Therefore, the supplementary presentation of the distribution (e.g., box plots) is important.
Practical relevance: Reporting in study reports and HTA contexts
In clinical study reports and health technology assessment dossiers, medians are frequently used in tables and Kaplan-Meier plots. A consistent definition is crucial, including the handling of censoring and clear indication of whether confidence intervals were calculated.
For HTA and reimbursement, medians can be helpful in cost and benefit data, but should be supplemented by means or modeling depending on the question, as budget impact analyses often require totals and expected values.
FAQ
When is the median more suitable than the arithmetic mean?
When data are highly skewed or contain outliers (e.g., costs, length of stay, laboratory values), the median is often more meaningful for the typical location.
Can clinical relevance be derived from medians alone?
Usually not. Medians are descriptive. Clinical conclusions require additional effect measures, uncertainty data (confidence intervals), and a predefined comparison in the analysis plan.
What does “median not reached” mean in Kaplan-Meier curves?
This means that by the time of analysis, fewer than 50% of participants had experienced an event. The median time-to-event cannot then be estimated; alternative metrics should be reported.
Regulatory references: EU Regulation (EU) No. 536/2014 (Clinical Trials Regulation); ICH E6(R3) Good Clinical Practice.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.
From a sponsor and CRO perspective, it is important that definitions and evaluation rules are established in advance in the protocol and analysis plan. Consistent data collection, plausibility checks, and transparent reporting reduce the scope for interpretation by ethics committees, authorities, and HTA bodies.