Retesting and tracking over time
Retesting means collecting a second stool sample and comparing it to an earlier result from the same person. Valid comparison requires the same laboratory pipeline (extraction, sequencing region or shotgun depth, bioinformatics version), a stable enough context (diet, medications, recent antibiotics), and a specific question, not a vague hope that the report will confirm a supplement worked. Longitudinal studies in healthy adults show person-specific signatures that persist over months (Franzosa et al., 2019), but technical and biological variation between any two samples can still swamp small real shifts if methods or timing are mismatched.
Consumer panels rarely tell you whether a change is signal or noise. A 15% relative rise in Bifidobacterium after four weeks of a probiotic may sit inside normal week-to-week fluctuation for that individual, or reflect a different batch of reference data.
For report routing: Reading your microbiome report. For post-antibiotic timing: Post-antibiotic gut recovery.
What not to conclude
| Retest pattern | Weak conclusion | More accurate framing |
|---|---|---|
| ↑ “Beneficial” taxon after probiotic | Probiotic colonised and fixed gut health | Possible transient presence; colonisation is strain- and person-specific, one trial showed some probiotic regimens delayed post-antibiotic reconstitution (Suez et al., 2018) |
| ↓ Dysbiosis score | Clinical improvement proven | Proprietary index vs updated reference cohort; not a symptom endpoint |
| Different lab, “better” profile | Objective progress | Not comparable, extraction, primers, and databases differ |
| Retest at 2 weeks on new fiber | Fiber failed if diversity unchanged | Fermentation and taxa shifts may need 4–8+ weeks of sustained intake; gas symptoms can precede compositional change |
| Monthly kits for a year | Trend line = medical monitoring | No validated clinical protocol for monthly consumer sequencing; cost and artefact risk high |
| Stable taxa, worse symptoms | Report must be wrong | Composition ≠ motility, visceral sensitivity, or diet trigger, see Multi-marker synthesis |
Minimum requirements for a valid comparison
| Requirement | Why it matters |
|---|---|
| Same company and assay version | 16S V4 vs V3–V4 vs shotgun are different instruments |
| Documented interval | Days since antibiotics, probiotics, colonoscopy prep, acute gastroenteritis |
| Medication log | PPIs, metformin, GLP-1 agonists, opioids reshape reports, Medications and microbiome |
| Diet note | Short-term diet shifts can move composition within days in controlled feeding studies (David et al., 2014) |
| Same clinical question | ”Recovering from amoxicillin” ≠ “Did this strain persist?” |
| Pre-intervention baseline (ideal) | Strongest design available on consumer testing, still not an RCT |
The Human Microbiome Project reported relative stability within individuals over months when health was stable (HMP Consortium, 2012), but stability is not immobility. Beta-diversity between your own samples is usually smaller than between you and a stranger, that is the premise of personalised baselines, yet vendor “optimal ranges” are built from other people’s pools, not your prior sample.
Typical timelines before a retest is interpretable
These are research-informed heuristics, not validated retest schedules on consumer kits.
| Context | Suggested minimum wait | What you might see |
|---|---|---|
| After antibiotic course | 4–12 weeks for trend; 6+ months if asking “fully back?” | Diversity often recovers before composition matches pre-course (Palleja et al., 2018) |
| Probiotic trial | Finish course + 2–4 week washout if asking about native community | Probiotic reads may disappear after stop, that is not always failure |
| High-fiber or prebiotic ramp | 4–8 weeks sustained intake | Gas may rise before taxa associated with fermentation change |
| Low-FODMAP reintroduction | Compare habitual diet windows, not active elimination vs liberal diet | Deliberately different fermentation load, not a fair before/after |
| Acute gastroenteritis | Wait until bowel habit baseline ≥4 weeks | PI-IBS workup is clinical, Post-infectious IBS |
| FMT or serious clinical event | Specialist-guided only | Consumer retest timing is not standardised |
Stopping a probiotic the day before retest and calling the sample “off probiotic” is usually too short a washout if the question is whether your baseline community changed.
What changes may matter vs noise
Plausible signal (especially if same pipeline, documented intervention, aligned symptoms):
- Sustained shift in dominant genera in the direction of dietary change (e.g. more Prevotella-rich pattern after sustained high-fiber diet in some people)
- Persistent presence of a probiotic strain if the panel resolves strain-level IDs (many do not)
- Large post-antibiotic diversity drop that partially recovers on repeat, timing must match antibiotic story
Often noise or artefact:
- Single-taxon wiggles within vendor “normal” bands
- Pathway score changes without taxa movement (database update)
- Diversity score shifts when reference cohort version updates
- Seasonal or travel-related bumps without clinical correlate
Meta-reanalyses caution that alpha diversity is not uniformly lower across diseases (Duvallet et al., 2017), a diversity “improvement” on retest does not prove you moved from diseased to healthy.
When retesting is low value
- No baseline from the same lab, you are comparing to a stranger pool
- Multiple simultaneous changes (new probiotic + fiber + PPI stop + holiday diet)
- Active symptom flare without red-flag workup, sequencing does not replace calprotectin or clinical assessment when IBD is possible
- Monthly “optimisation” without a hypothesis, high cost, low validated yield
- Expecting stool to track breath-test outcomes for SIBO/IMO, geography differs (Methane and colonic gas)
If symptoms improved but the report worsened (or the reverse), trust the symptom trajectory and clinical follow-up over the panel for management decisions.
What to do next
- Log date, lab, kit version, antibiotics, probiotics, and major diet change on every sample.
- Wait until one variable has been stable long enough for your question (table above).
- Prefer same-lab before/after over cross-vendor comparison.
- Pair retest with clinical markers when inflammation is a concern, not sequencing inflammation scores alone.