Retrospective study · October 2026
A clearer picture of
recorded progress.
What changed in patients’ weight and blood tests—and how much we can learn from the records.
Internal exploratory research
Analysis: 8 October · Frozen data: 7 October, 01:31 IST
Weight change over time
Actual recorded measurements, grouped into comparable follow-up windows.
See the plotted numbers
Progress by recorded medication
Semaglutide and Mounjaro / tirzepatide, compared within the same observation window.
See the plotted numbers
Blood tests: a smaller, mixed picture
First-to-last changes, with the actual time between tests. Each marker has its own patient cohort.
HbA1c decreased on average. Total and LDL cholesterol also had lower means, but their intervals include no change. Triglycerides rose slightly on average; HDL was essentially unchanged.
There are no supported 30-, 60- or 90-day blood-test estimates. The available repeat measurements are too sparse within those windows.
Whiskers show the 95% interval for the mean change. Each row has its own axis and unit; compare the numbers, not lengths across rows. A decrease does not automatically mean better health.
Explore all 34 blood markers with displayed paired results
How recorded weight change differs
Explore patient characteristics and care-record patterns. These are descriptive comparisons over available follow-up.
See the plotted numbers
Other daily measures
Sleep, water, movement and protein have smaller logged cohorts, with short follow-up intervals.
The intervals for these mean changes include zero. These logs do not show a consistent improvement across the four measures. Reported units are preserved; missing logs are never treated as zero.
Know the strength of the evidence
More records do not always mean more usable before-and-after comparisons.
Where time-based results are available
Weight rows below use the first-recorded-weight clock, independent of the clock selected above.
“Not enough paired data” means fewer than five eligible pairs in the window, or results withheld by disclosure checks. It does not mean no change.
What the data supports
Weight changes among people who recorded repeat measurements. Medication and subgroup descriptions with explicit patient counts and actual timing. First-to-last laboratory changes for selected biomarkers.
What remains uncertain
How much change was caused by treatment, which drug is superior, results for people without follow-up, independently confirmed adherence, and time-specific metabolic effects. No usable waist-circumference series was available.
How to read this report
Time windows, not exact dates
Around day 30 uses an observation in days 15–44; day 60 uses 45–74; day 90 uses 75–105. We use the nearest actual observation to the target and show the actual median interval. No missing result is interpolated.
Two different clocks
First recorded weight is not necessarily the weight before treatment. A corroborated recorded start requires a credible dose record within 14 days. The primary treatment baseline may be -14 to +7 days; the stricter option requires a prior or same-day baseline.
What the percentages mean
Weight percentages are calculated per person, then averaged. They differ from the percentage change in the group’s average weight. HbA1c absolute changes are percentage points, not relative percentages.
What the whiskers mean
95% patient-bootstrap intervals describe sampling variability. They do not cover extraction errors, selective follow-up, self-report errors or unmeasured differences between groups. Within-group intervals do not test which medication is better.
Missing data and privacy
Each outcome and window has its own cohort. Patient counts must not be added across markers or groups. Cells below five and unsafe complementary partitions are withheld. No recorded dose is not an untreated control.
Study status
Post hoc retrospective research, not a randomized trial. Clinical review and extraction completeness remain unfinished. Values needing review were excluded. This visual edition summarizes the accepted aggregate study without a new statistical analysis.
Analysis coverage and data sources
Counts here refer to displayed aggregate rows, not unique patients. Some factors only have unknown or single-category results and do not yield a useful comparison.
This visual edition includes released aggregate results. Detailed analysis files remain in the study archive. · STROBE reporting guidance · Understanding HbA1c