Kybella safety: what 1,968 FDA FAERS reports reveal

Kybella (deoxycholic acid) is the only injectable drug the FDA has ever approved for fat reduction — specifically for the convexity or fullness of submental fat, the "double chin," in adults (NDA 206333, approved April 2015). The label carries two named warnings by design: marginal mandibular nerve injury, which can produce an uneven smile or facial weakness, and dysphagia, or trouble swallowing. The clinical trials that supported approval — 513 Kybella subjects against 506 placebo — recorded nerve injury in 4% of treated subjects and dysphagia in 2%, with every case resolving on its own.

Those trial numbers are the figures most often quoted. The FDA's postmarket drug database, FAERS (FDA Adverse Event Reporting System), is where the real-world signal lives after a drug reaches the market. A full-extract analysis of FAERS records naming deoxycholic acid as the primary suspect drug — 1,968 reports received between 2015 and 2025 — confirms the label's two flagged effects, quantifies them at a scale the trials could not, and surfaces a pattern the label does not emphasize: a cluster of reports in which the treated fat grew larger rather than smaller, the same paradoxical response that dominates the safety record of cryolipolysis.

This article is a database-signal profile, not a side-effects explainer. For the labeled common effects (swelling, bruising, pain, numbness, hardness) and the trial context, the companion piece on Kybella side effects covers the label and the trials in detail. Here the focus is what the full FAERS extract shows — and, importantly, what it cannot tell you.

The dataset at a glance

FAERS is a passive, spontaneous-reporting system. It cannot be used to calculate how often something happens in the real world, it undercounts heavily, and a report records an event that followed a drug, not proof the drug caused it. A report coded "serious" means it met an FDA seriousness criterion (death, hospitalization, life-threatening, disability, required intervention, or other medically important condition), not that causation was established. With those caveats, FAERS is the closest thing to a public postmarket signal for an injectable drug — and deoxycholic acid is a useful case because its labeled warnings are narrow and specific, so the database can be read against the label directly.

The 1,968 figure is a full primary-suspect extract — every report in which deoxycholic acid (the generic name for Kybella) is listed as the primary suspect drug, across the full post-approval period. Smaller counts you may see elsewhere reflect narrower filters: queries on the brand name "Kybella" alone, a tighter seriousness criterion, or a shorter window all return far fewer reports (the companion Kybella side-effects article draws on one such narrower cut). The figures here are the larger denominator behind those signals; the reaction-term ranking below is read against that denominator.

Reporting is consumer-driven as much as it is clinician-driven: 1,118 reports (57%) originated from consumers, 845 (43%) from a healthcare professional (525 other HCP, 316 physicians, 4 pharmacists). Women outnumber men roughly eight to one among reports where sex is recorded (1,309 female vs. 167 male), and the median age is 40 — both consistent with the submental-cosmetic population the drug is approved for.

What the reports actually say

Counting reaction terms across 1,968 reports produces a ranking that the trial table does not. A single report can list several reaction terms, so the counts below overlap.

Read thematically, four signals dominate the profile.

1. The nerve-injury signal the label warns about — at scale

The label's first warning is marginal mandibular nerve injury. In FAERS, the unambiguous nerve-injury terms — facial paresis (109), nerve injury (100), and facial paralysis (26) — combine to 235 reports, about 12% of the total, with another 47 reports of facial asymmetry and 27 of jaw pain that are frequently nerve-related in the narratives. This is the label's flagged effect showing up in the wild at a volume the 513-subject trial could not have produced. The narratives describe the expected presentation — an uneven or crooked smile, drooping of one side of the mouth, facial weakness — appearing within days of injection and, in most reports, resolving over weeks to months. A minority describe persistence, and these are the reports that drive the "disabling" and "not recovered" outcomes below.

2. Effectiveness failure and off-label use rank near the very top

The single most frequent reaction term is drug ineffective (263, 13.4%), and off label use (243, 12.3%) ranks third. Together with "condition aggravated" (62), this is the second-largest signal in the database after the local injection-site effects. It reflects two things at once: patients reporting that the treatment did not produce the expected reduction in submental fullness, and patients who received deoxycholic acid somewhere other than under the chin — the jawline, the jowls, the abdomen, the arms — areas where the drug was never studied and where the safe-injection boundaries mapped for the submentum simply do not apply.

This matters because the nerve-injury risk is anatomical: the marginal mandibular nerve runs along the jaw, and the label's injection-technique instructions exist to keep the drug away from it. When the drug is injected off-label into the lower face or neck without those boundaries, the nerve-injury signal is harder to control. The label states plainly that "the safe and effective use of KYBELLA for the treatment of subcutaneous fat outside the submental region has not been established and is not recommended."

3. The local injection-site effects — exactly as the label predicts

Swelling (244 injection-site + 207 general), pain (146 + 59), nodules and masses (109 + 86), induration (67), bruising and contusion (60 + 75), and hypoaesthesia (78) together account for the bulk of the remaining reports. These are the effects the trials recorded at high rates — 87% swelling, 72% bruising, 70% pain, 66% numbness in the pooled trial population — and their prominence in FAERS is expected rather than surprising. The injection-site nodule and mass terms (195 combined) are the ones worth flagging to patients, because firmness and small nodules under the chin can persist for weeks and occasionally longer; the trials noted injection-site numbness, swelling, pain, and induration lasting beyond 30 days in 42%, 20%, 16%, and 13% of subjects respectively.

4. A "fat tissue increased" signal that echoes cryolipolysis

The most interesting term in the ranking is one the label does not mention: fat tissue increased (37 reports, 1.9%). In a drug whose entire purpose is to destroy submental fat cells, a cluster of reports describing the treated fat as enlarged rather than reduced is the pharmacovigilance analogue of paradoxical adipose hyperplasia after cryolipolysis — the same unexpected direction of effect, in a different modality. Thirty-seven reports is a small absolute number, and the term is not a validated diagnosis; it may capture post-inflammatory swelling or weight gain misattributed to the injection. But it is the kind of low-frequency, unexpected-direction signal that passive surveillance exists to catch, and it is worth asking whether it reflects a real biological paradoxical response in a small subset of patients. The published literature on deoxycholic acid has not characterized such an effect, which is precisely why a signal of this shape belongs in a transparent safety profile rather than being buried in the reaction list.

Outcome severity and the temporal pattern

Outcome was recorded for roughly a third of reports; the rest are listed as unknown, which is normal for spontaneous reporting. Among reports with a known outcome:

• Not recovered: 661 (33.6% of all reports). This is the most cautionary number in the dataset, but it needs care: "not recovered" includes cases where the outcome was simply not yet established at the time of reporting, and it skews toward the lingering local effects (nodules, numbness, firmness) that the trials already showed can persist beyond a month. It is not the same as "permanent."
• Recovering: 200 (10.2%) and Recovered: 317 (16.1%).
• Recovered with sequelae: 11 and Fatal: 3. The fatal reports are few enough that they cannot be interpreted as a rate; FAERS fatality reports for a cosmetic injectable often involve patients with significant comorbidity or events the database cannot causally link to the drug. They are recorded, not established.

Reporting volume over time peaked in 2018 (340 reports), held between roughly 170 and 290 per year from 2016 through 2023, and fell in 2024–2025 — a pattern that tracks the drug's commercial trajectory (it launched in 2015, peaked in early adoption, and has since faced competition from newer body-contouring options) more than any change in intrinsic risk. Year-to-year counts in FAERS reflect reporting behavior and product volume, not incidence.

How to read this, and what to ask

The honest summary of the full FAERS profile is that it confirms the label rather than contradicting it. The two effects the FDA called out — nerve injury and dysphagia — are exactly what surface at scale in the postmarket data, the local injection-site effects appear at rates consistent with the trials, and the database adds two signals the label understates: the volume of effectiveness-failure and off-label reporting, and a small but notable paradoxical "fat tissue increased" cluster.

For a patient considering Kybella, the questions the database suggests asking are:

• "Are you treating only the submental area, and are you staying within the labeled injection zone?" The off-label reporting is concentrated where the drug was moved outside the submentum. The label exists in part to protect the marginal mandibular nerve; an injector who treats the jawline or lower face is working outside the studied anatomy.
• "What is your plan if I develop an uneven smile or trouble swallowing?" Both are labeled effects. Most resolve, but the FAERS outcomes show a tail of cases that persist. Ask how the practice evaluates and follows up on nerve symptoms, and how long after treatment you should report a crooked smile or swallowing difficulty.
• "How many of your patients see the result they expected?" Drug-ineffective is the most frequent single reaction term. Deoxycholic acid produces a modest reduction across multiple sessions, and candid expectation-setting is part of informed consent.
• "What happens to firmness and nodules if they don't resolve?" Persistent submental firmness is the other tail in the data.

Deoxycholic acid is an approved drug with a narrow, well-characterized indication and a safety profile that the postmarket data largely reaffirms. The reason to look at the full FAERS extract rather than stop at the trial table is that the trials were small and short, and the real world includes off-label use, effectiveness failure, and the occasional unexpected signal — all of which a patient is entitled to understand before the first injection.

Sources

• U.S. FDA, FAERS (FDA Adverse Event Reporting System) public data — full-extract analysis of primary-suspect deoxycholic-acid (Kybella) reports, 2015 – 2025 (run June 2026): https://www.fda.gov/drugs/surveillance/questions-and-answers-fdas-adverse-event-reporting-system-faers
• KYBELLA (deoxycholic acid) injection Prescribing Information — NDA 206333, warnings on marginal mandibular nerve injury (4%) and dysphagia (2%); pooled trial adverse-reaction table (N=513): https://www.accessdata.fda.gov/drugsatfda_docs/label/2022/206333s005lbl.pdf
• KYBELLA Prescribing Information (2015 original approval, NDA 206333): https://www.accessdata.fda.gov/drugsatfda_docs/label/2015/206333Orig1s000lbl.pdf
• Drugs@FDA, KYBELLA (deoxycholic acid) approval history and labeling: https://www.accessdata.fda.gov/scripts/cder/daf/index.cfm?event=overview.process&ApplNo=206333