Lab TLF Examples: Tables, Listings and Figures in Trials

A lab TLF is a table, listing or figure that presents laboratory results from a clinical trial. The most common examples are a change-from-baseline summary table, a shift table (normal to abnormal), a listing of subjects with clinically significant abnormalities, and graphics such as box plots and eDISH plots. Below you will find each one described, with a layout you can adapt and a plain explanation of what the reviewer is looking for.

This guide is written for clinical operations leads, project managers, study directors and sponsor teams who need to see what “good” looks like before they write a shell or choose a laboratory partner. It also helps site and practice teams who want to understand where their draws end up.

TL;DR

• TLF means Tables, Listings and Figures. TFL is the same thing with the letters reordered, and both are used interchangeably (FDA).
• Lab tables summarize groups (change from baseline, shifts, values beyond limits). Listings show patient-level rows. Figures show trends and outliers, such as box plots and eDISH plots.
• Every output must trace to a pre-specified Statistical Analysis Plan, and the numbers must reconcile with the analysis datasets and the raw data.
• Clean TLFs start at the laboratory: consistent units, reference ranges, standard test codes and clear sample labeling.

What TLF means, and whether it is TLF or TFL

TLF stands for Tables, Listings and Figures. These are the standardized outputs that present study results in regulatory submissions and clinical study reports. You will also see TFL. Both refer to the same collection of summary, patient-level and visual displays, so use whichever your sponsor or statistical team already uses and keep it consistent across documents (FDA, Standard Safety Tables and Figures).

The three formats do different jobs:

• Tables summarize data across treatment groups, typically with counts, percentages and statistical comparisons.
• Listings show raw, patient-level detail, one row per subject and time point.
• Figures give a visual narrative, such as a Kaplan-Meier curve or a dose-response plot (Weltrix).

Laboratory data feed all three. In most trials, lab results are the largest single block of safety data, which is why lab TLFs deserve their own planning.

What counts as laboratory data in a trial

Before looking at outputs, it helps to be clear about the inputs. Laboratory data in a trial typically include:

• Hematology: red and white blood cell counts, hemoglobin, platelets.
• Blood chemistry: liver enzymes such as ALT and AST, kidney markers such as creatinine, and electrolytes such as sodium and potassium.
• Urinalysis.
• Specialty data: genomic, pharmacokinetic and biomarker panels, depending on the protocol (FDA).

Each of these needs a unit, a reference range, a collection time point and a link to a subject and visit. If any of those is missing or inconsistent between sites, the TLF programmer ends up fixing the problem downstream. Our guide to analysis-ready lab data covers how to prevent that at the source.

Lab table examples

Lab tables answer group-level questions: did a value move, did it move in one arm more than another, and did it cross a threshold anyone should worry about?

Summary of lab parameters over time (change from baseline)

This is the workhorse. For each parameter, such as ALT, creatinine or hemoglobin, the table shows the observed value and the mean change from baseline at each scheduled visit, by treatment group (FDA).

An illustrative layout (placeholder values, not real data):

What the reviewer looks for: a drift in one arm that does not appear in the other, and whether the number of subjects at each visit (the n) is shrinking in a way that could bias the means.

Incidence of values meeting criteria of clinical concern

Instead of averages, this table counts subjects whose results met a pre-specified threshold, for example Level 2 criteria for a given test (FDA). Averages can hide a handful of subjects with large excursions. Counts do not.

The last row reflects the combination that matters for liver safety. We return to it in the eDISH section below.

Values above or below the limits of normal

These summaries count subjects whose post-baseline values fall above or below the laboratory’s limits of normal, usually by parameter and treatment group (FDA). They depend entirely on the reference ranges supplied with the data, so a laboratory that reports ranges consistently, with age and sex adjustments where relevant, makes this table easier to build and easier to defend.

Shift tables (normal to abnormal)

A shift table cross-tabulates each subject’s baseline category against their worst post-baseline category. Typical categories are low, normal and high.

Read the off-diagonal cells. Subjects who moved from normal to high are the ones a medical monitor will want to see in the listings. Subjects who started abnormal and stayed there are usually not a treatment signal.

Abnormality summaries by organ system or parameter

Some packages group abnormalities by organ system (liver, kidney, hematologic) rather than by single test, so a reader can see the pattern across related parameters at once (FDA).

Lab listing examples

Listings are where summaries meet individual people. A lab listing shows the raw, patient-level record: who, when, what test, what result, and how it compares to the reference range. A common one is the listing of subjects with clinically significant laboratory abnormalities (FDA).

An illustrative layout (placeholder values):

Good listings share a few habits:

1. Sort predictably. Subject, then test, then date, so a reader can follow one person’s trajectory.
2. Show the flag and the reference range beside the result. Nobody should have to look up whether a value is high.
3. Include baseline. Without it, a reviewer cannot tell a new problem from an old one.
4. Keep units visible. Mixed units, such as mg/dL and umol/L for creatinine, are a classic source of errors.
5. Add the retest. If a value was repeated, show both, with a clear indicator of which was used in analysis.

Public statistical analysis plans show how sponsors lay these out. One example is a table, figure and listing shell document posted on ClinicalTrials.gov for a Phase 1 study, which lists its planned lab outputs, including urinalysis frequency tables, before any data exist.

Lab figure examples

Tables can hide a pattern that a picture makes obvious. Three figures show up most often for lab data (FDA).

Box plots over time

Box plots show the distribution of a lab value at each visit, by arm. You see the median, the spread and the outliers together. They pair naturally with the change-from-baseline table.

eDISH plots

The eDISH plot is a scatter plot used to find possible Hy’s Law cases in liver safety. Each point is a subject, with peak ALT or AST on one axis and peak total bilirubin on the other, both as multiples of the upper limit of normal. Points in the upper right quadrant deserve immediate review. FDA’s definition of a Hy’s Law case includes ALT or AST above 3x ULN and total bilirubin above 2x ULN, without an initial finding of cholestasis (Gastroenterology).

Statisticians favor this kind of graphic because it tells the story of drug-induced liver injury at a glance, where row-based tables often fail to (ASA proceedings, 2015). Each point can also be tied back to that subject’s full clinical and laboratory record, which is why the eDISH plot and the patient listing belong together.

Kaplan-Meier plots for time to a lab event

A Kaplan-Meier plot can show the time to first occurrence of a specific laboratory abnormality, for example the first ALT above 3x ULN, by treatment arm (FDA).

How trial teams use lab TLFs

Teams read these outputs against pre-specified safety endpoints. Mean change-from-baseline plots point to trends. Listings let the team drill into individual cases with abnormal values and confirm whether a clinically important change really occurred (FDA).

A practical review sequence looks like this:

1. Start with the incidence table for criteria of clinical concern.
2. Check the shift table for movement from normal to abnormal.
3. Open the listing for every subject flagged.
4. Confirm with the figure (box plot or eDISH) that the picture is consistent.
5. Return to the source record for any case that still looks unclear.

The same logic applies during the study, not only at the end. Our post on safety lab monitoring windows explains how the timing of draws and review shapes what the clinician sees.

Traceability: from the SAP to the raw sample

FDA expects every TLF in a clinical study report to trace back to a pre-specified Statistical Analysis Plan, which supports transparency and independent audit (FDA). In US filings, results in the report text must also align with the derived datasets (ADaM), which in turn must map back to the raw tabulated data (SDTM) (Pharma Regulatory). Mismatches between report text and TLFs, or between TLFs and datasets, erode credibility.

Volume makes this harder. A single pivotal trial can require hundreds of individual outputs, each needing independent verification (Weltrix). Teams that borrow verification habits from other lab disciplines, such as lot-specific certificate-of-analysis checks described in COA verification practices, tend to catch unit and labeling mismatches earlier.

Three upstream habits prevent most downstream TLF rework:

• Map lab tests to standard codes early. Our walkthrough of CDASH to SDTM lab mapping with LOINC shows where mappings usually go wrong.
• Label samples to a standard. Illegible or inconsistent labels become orphan results. See our sample labeling standards guide.
• Know what electronic lab records must prove. Our overview of Part 11 lab data covers what counts and how to show it.

A perspective from our lab

We see TLFs from the other end of the pipeline. A listing flag that looks like a clinical problem is, surprisingly often, a data problem: a unit that changed between sites, a reference range that was not updated, a repeat test with no indicator. By the time that row reaches a statistician, it is expensive to untangle.

So we think about TLFs from the first requisition. Which reference ranges will the sponsor use? How will repeat and unscheduled draws be identified? Will imaging-derived measurements need to line up with lab values in the same subject-level view? The questions are answered once, in writing, before enrollment, and they save weeks at database lock.

We also believe a central laboratory partner should behave like part of the study team, not a vendor that sends a file and disappears. That means direct communication, fast turnaround and flexibility when a protocol amendment changes a panel. Those are the things that keep lab outputs clean enough for the TLF team to trust.

— Kohealth Labs

Planning your lab TLF shells

Shells are mock tables, listings and figures with placeholder values, written into or alongside the SAP. A useful lab shell set includes:

• A list of all lab parameters, units and scheduled visits.
• The reference range source and how out-of-range flags are derived.
• Definitions of clinical concern criteria, with the exact thresholds.
• Rules for repeats, unscheduled visits and early termination.
• The population used for each output (safety set, per-protocol and so on).
• Titles and footnotes that state the data source and the program that created each output.

A quick test: could a programmer who has never met your team produce the table from the shell and the datasets alone? If not, the shell needs more detail.

Where the lab fits into your TLF workflow

A laboratory partner supports TLFs in practical ways that never appear in the output itself: consistent units, stable reference ranges, clean specimen identifiers, and data delivered in a structure your programmers can use. Our pieces on DICOM SR traceability for SDTM and on getting lab data analysis-ready show how this extends to imaging-derived values.

If you are running a study and want lab and imaging data delivered in a form your biostatistics team can use, see our clinical trials laboratory services. To understand the flow from specimen pickup to results delivery, our how it works page lays it out step by step.

Common mistakes in lab TLFs

• Mixed units across sites without a conversion rule documented in the SAP.
• Reference ranges that change mid-study and are not versioned.
• Averages only. Means that hide a few extreme cases.
• Listings with no baseline or no repeat indicator.
• Shift tables with unclear categories because the low, normal and high definitions are not footnoted.
• Figures with no link to the listing. A reviewer who spots an outlier should be able to find that subject in one step.

FAQ

Is it TLF or TFL?

Both are used for the same thing: Tables, Listings and Figures. Teams choose one by habit or sponsor convention. What matters is that you use the same term and numbering across the SAP, shells and report.

What is a lab listing in a clinical trial?

A lab listing is a patient-level display of raw laboratory results, with one row per subject, test and time point. It usually includes the result, unit, reference range, an abnormal flag and the baseline value. Reviewers use it to check individual cases flagged by summary tables.

What are the most common lab tables in a clinical study report?

The most common are change-from-baseline summaries by visit, incidence tables for values meeting criteria of clinical concern, summaries of values above or below limits of normal, and shift tables. Together they show group trends and the subjects who crossed important thresholds.

What figure is used to look for liver safety signals?

The eDISH plot. It graphs peak ALT or AST against peak total bilirubin, each as a multiple of the upper limit of normal, so possible Hy’s Law cases stand out in one quadrant. Each point links back to that subject’s data for review.

Sources

• FDA: Standard Safety Tables and Figures, Integrated Guide
• Weltrix: TLF in Clinical Trials, Programming Best Practices
• ASA proceedings: Graphical Approaches to Evaluate Liver Safety Data
• Pharma Regulatory: CTD Module 5 Clinical Study Reports, US Data Presentation
• Gastroenterology: Drug-Induced Liver Injury, Morbidity, Mortality, and Hy’s Law
• ClinicalTrials.gov: Statistical Analysis Plan, Table/Figure/Listing Shells

For more information about our diagnostic testing, clinical research laboratory services, or partnership opportunities, visit https://kohealthlabs.com or contact KoHealth Labs at (469) 792-8439 and ask for Amy Patterson.