A medical statistician is a specialist who applies statistical methods to health research, from the first draft of the protocol to the final reply to peer reviewers. Many studies, however, receive no statistical input until the data are collected. At that point, a statistician can analyze the data but cannot fix a study that was too small or poorly designed. A common mistake is to contact a statistician only after a journal rejects the manuscript. This guide explains what a medical statistician does, when to hire one, what to prepare, what affects the cost, and how a statistician supports one hypothetical clinical trial.
Table of Contents
Free tool: calculate the sample size or power for your study with the StatisMed sample size calculator. It shows the exact R code and ready-to-use wording for your Methods section. Open the calculator →
What Is a Medical Statistician?
A medical statistician designs studies, calculates sample sizes, analyzes data, and helps researchers report results correctly. The field is part of biostatistics, and the job titles below are often used interchangeably.
Medical Statistician vs Biostatistician
Both terms describe a statistician who works with health data. “Biostatistician” is common in the United States, while “medical statistician” is common in the United Kingdom and Europe. In practice, both have training in statistics and experience with clinical, epidemiological, or public health data.
Clinical Research Statistician
A clinical research statistician works mainly on clinical trials and observational clinical studies. This role includes writing the statistical analysis plan, preparing the randomization list, supporting the data monitoring committee, and producing the tables required by regulators and journals.
Statistical Consultant
A statistical consultant offers biostatistics consulting on a project basis. Consultants are often hired by hospitals, universities, postgraduate students, and companies that do not employ a full-time statistician. Statistical consulting may cover one task, such as a sample size calculation, or the whole study.
Importance of a Medical Statistician in Medical Research
Valid Study Design
The design determines what the data can show. A medical statistician checks that the study question, the outcome, the comparison groups, and the sample size match each other. Notably, a design error cannot be corrected by any analysis after the study ends.
Fewer Statistical Errors
Statistical errors are common in published clinical studies, and the inappropriate use of statistical analysis can lead to incorrect conclusions (Slutsky, 2013). Typical errors include the wrong test for the data type, multiple testing without correction, and p-values reported without effect sizes or confidence intervals.
Faster Peer Review
Many journals send manuscripts to a statistical reviewer. Therefore, a manuscript with a clear statistical methods section, correct tests, and complete reporting is less likely to be rejected or returned for major revision.
When to Hire a Medical Statistician: 5 Key Times
1. Before Writing the Protocol
The best time to involve a medical statistician is at the planning stage. The statistician helps define the primary outcome, the hypothesis, the type of study, and the methods to control bias, such as randomization and blinding.
2. When Calculating the Sample Size
Ethics committees and funders require a justified sample size. The calculation must be performed before the study starts, and it depends on the expected effect, the significance level, and the power (Das et al., 2016).
3. During Data Collection and Management
A statistician can design the data collection form and the coding sheet. Moreover, early checks of the data detect missing values, impossible values, and coding errors while they can still be corrected.
4. At the Analysis Stage
At this stage, the medical statistician cleans the data, chooses the appropriate tests, runs the analysis, and prepares the tables and figures. Our guide to choosing the right statistical test shows how many factors this choice depends on.
5. During Peer Review and Revision
Reviewers often ask for additional analyses, such as adjustment for confounders or sensitivity analyses. A statistician can perform these analyses and help write a clear, point-by-point response to the reviewers’ statistical comments.
What to Prepare Before Contacting a Medical Statistician
A well-prepared request saves time and reduces the cost. Before the first meeting, prepare the following:
- The research question, the study objectives, and the primary outcome.
- The study design (for example, randomized trial, cohort, case-control, or cross-sectional survey).
- The protocol or proposal, if available, and the target journal or university requirements.
- The data file in Excel or SPSS format, with one row per participant and one column per variable.
- A coding sheet that explains each variable, its units, and the meaning of each code.
- The deadline and the deliverables you need (sample size, analysis, tables, or a written Results section).
How a Medical Statistician Supports a Study: A Clinical Example
Consider a hypothetical surgical team that plans a randomized trial. The team wants to test whether a new antimicrobial wound dressing reduces surgical site infection within 30 days compared with the standard dressing. The steps below show the work of a medical statistician from planning to reporting. All the numbers were calculated with Python.
Step 1: Define the Question and the Primary Outcome
The statistician and the surgeons agree on one primary outcome: surgical site infection within 30 days (yes or no). The null hypothesis (H0) states that the infection rate is the same with both dressings. The alternative hypothesis (H1) states that the rates differ. The significance level is 0.05 (two-sided).
Step 2: Calculate the Sample Size
Based on hospital records, the expected infection rate with the standard dressing is 15%. The team considers a reduction to 6% clinically important. With a power of 80%, the formula for comparing two proportions gives 181 patients per group. After allowing for 10% loss to follow-up, the target is 202 patients per group.
| Parameter | Value |
|---|---|
| Expected infection rate, standard dressing | 15% |
| Expected infection rate, new dressing | 6% |
| Significance level (two-sided) | 0.05 |
| Power | 80% |
| Required sample size per group | 181 |
| Sample size per group after 10% dropout | 202 |
| Total patients to recruit | 404 |
Step 3: Show What Happens Without a Statistician
Suppose the team had recruited only 50 patients per group, as initially suggested. With the same infection rates, the power of this study is only 31%. If the results were 8 of 50 (16.0%) vs 3 of 50 (6.0%), the difference would not be significant (χ² = 2.55, df = 1, p = 0.110). The relative risk would be 0.38, but the 95% confidence interval (0.11 to 1.33) would include 1. Therefore, a real benefit could be missed simply because the study was too small.
Step 4: Prepare the Statistical Analysis Plan
Before unblinding, the statistician writes a statistical analysis plan. It states that the primary analysis follows the intention-to-treat principle, that the infection rates will be compared with the chi-square test, and that the effect will be reported as a relative risk with a 95% confidence interval. Pre-specifying the plan prevents selective reporting.
Step 5: Analyze the Data
At the end of the hypothetical trial, 200 patients per group completed follow-up. Infection occurred in 30 patients (15.0%) with the standard dressing and in 12 patients (6.0%) with the new dressing. The smallest expected count is 21, so the chi-square test is appropriate.
| Group | Infection, n (%) | No infection, n (%) | Total |
|---|---|---|---|
| Standard dressing | 30 (15.0%) | 170 (85.0%) | 200 |
| New dressing | 12 (6.0%) | 188 (94.0%) | 200 |
| Total | 42 (10.5%) | 358 (89.5%) | 400 |
The chi-square test showed a significant difference (χ² = 8.62, df = 1, p = 0.003). However, a p-value alone does not show the size of the benefit. For this reason, the statistician also calculates the effect measures in Table 3.
Step 6: Calculate the Effect Size and the Confidence Intervals
| Measure | Estimate | 95% CI |
|---|---|---|
| Relative risk (RR) | 0.40 | 0.21 to 0.76 |
| Absolute risk reduction (ARR) | 9.0 percentage points | 3.1 to 14.9 |
| Number needed to treat (NNT) | 12 | – |
The new dressing reduced the risk of infection by 60% (RR = 0.40). In absolute terms, 9 fewer infections occurred per 100 patients, so about 12 patients need the new dressing to prevent one infection (1 ÷ 0.09 = 11.1, rounded up). Our posts on odds ratio vs relative risk and confidence intervals explain these measures in more detail.
Step 7: Report the Results
Finally, the statistician prepares the tables, the statistical methods section, and the Results paragraph, as shown below.
How to Report the Work of a Medical Statistician in a Manuscript
In the Methods section, describe the sample size calculation and the analysis. For example: “A sample size of 181 patients per group was required to detect a reduction in infection from 15% to 6% with 80% power at a two-sided significance level of 0.05. The chi-square test was used to compare infection rates, and the relative risk was calculated with a 95% confidence interval.” In the Results section, report the significant findings only, as in the model paragraph below.
Surgical Site Infection by Dressing Type
Surgical site infection within 30 days occurred in 12 of 200 patients (6.0%) in the new dressing group and in 30 of 200 patients (15.0%) in the standard dressing group (χ² = 8.62, df = 1, p = 0.003). The new dressing was associated with a significantly lower risk of infection (RR = 0.40, 95% CI, 0.21 to 0.76), with an absolute risk reduction of 9.0 percentage points (95% CI, 3.1 to 14.9) (Table 2).
Notice that the paragraph presents the numbers without explaining why the dressing worked. The interpretation and the comparison with previous trials belong in the Discussion section.
Common Mistakes When Working with a Medical Statistician
Contacting the Statistician Too Late
Many researchers ask for help only after the data are collected. As the clinical example shows, a study that is too small cannot be rescued by analysis. Therefore, involve the statistician before recruitment starts.
Sending Unstructured Data
Data with merged cells, colors instead of codes, and several values in one cell take a long time to clean. A simple, consistent spreadsheet with a coding sheet reduces both the time and the cost of the analysis.
Asking for Significant Results
A statistician’s role is to analyze the data correctly, not to produce a significant p-value. Running many tests until one is significant inflates false positive findings and is a form of research misconduct.
Changing the Outcome After the Analysis
The primary outcome must stay as specified in the protocol. Changing it after seeing the results is a common reason for criticism by reviewers and trial registries.
How Much Does a Medical Statistician Cost?
The cost of statistical consulting depends on the scope of the work rather than on a fixed price. The main factors are the number of outcomes and variables, the complexity of the methods (for example, a chi-square test vs a mixed-effects model or a meta-analysis), the quality of the data file, the deliverables (tables only, or a written Results section), and the deadline. Hiring a statistician early often reduces the total cost, because problems are prevented rather than corrected.
Running the Example Analysis in SPSS and R
In SPSS, the sample size can be calculated through Analyze → Power Analysis → Proportions → Independent-Samples Binomial Test (SPSS 27 or later). The trial data are analyzed through Analyze → Descriptive Statistics → Crosstabs; select “Chi-square” and “Risk” under “Statistics”, and “Row” percentages under “Cells”. In R, the sample size is obtained with one line of code: power.prop.test(p1 = 0.15, p2 = 0.06, power = 0.80), which gives about 181 patients per group. The chi-square test is obtained with chisq.test(matrix(c(30, 170, 12, 188), nrow = 2, byrow = TRUE), correct = FALSE).
Conclusion
A medical statistician contributes to every stage of a clinical study: the design, the sample size, data management, the analysis, the manuscript, and the reply to reviewers. The hypothetical wound dressing trial shows that early statistical input decides whether a study can detect a real effect. Moreover, a complete report includes the test statistic, the p-value, an effect size, and confidence intervals. If you need a medical statistician for your protocol, data analysis, or manuscript, the StatisMed team of biostatisticians provides statistical analysis services for medical studies. You can request a free quote today.
Frequently Asked Questions
1. What is the difference between a medical statistician and a data analyst?
A data analyst summarizes and visualizes data, often in business settings. A medical statistician has specific training in study design, sample size, clinical trials, and the statistical methods required by medical journals and regulators.
2. Should a medical statistician be a co-author on my paper?
When the statistician contributes substantially to the design, analysis, or interpretation of the study, co-authorship is appropriate under common authorship criteria. For limited support, such as a single analysis, an acknowledgment may be sufficient. Agree on this at the start of the project.
3. Can a statistician help if my data are already collected?
Yes. A statistician can still clean the data, choose appropriate tests, adjust for confounders, and report the results correctly. However, problems in the design, such as a small sample, can only be described as limitations, not corrected.
References
Das, S., Mitra, K., & Mandal, M. (2016). Sample size calculation: Basic principles. Indian Journal of Anaesthesia, 60(9), 652–656. https://doi.org/10.4103/0019-5049.190621
Slutsky, D. J. (2013). Statistical errors in clinical studies. Journal of Wrist Surgery, 2(4), 285–287. https://doi.org/10.1055/s-0033-1359421




