The PREEMPT Protocol: Why Fixed-Site, Fixed-Dose Was the Trial Design, and What It Means in Practice

The PREEMPT protocol is the reason chronic migraine injection looks nothing like aesthetic injection, and the reason injectors who are excellent at one are frequently poor at the other. It is a fixed-site, fixed-dose regimen: predetermined locations, predetermined units, delivered the same way in every patient regardless of where that individual's pain or movement happens to be. To an aesthetic injector trained to read a face and respond to it, that is close to offensive. It is also the entire basis of the approval, and understanding why is the difference between delivering the regimen that was studied and delivering something that merely resembles it.

This is a trial-design literacy piece. The clinical instruction in it is short. The reasoning is the point.

What PREEMPT actually was

PREEMPT — Phase 3 REsearch Evaluating Migraine Prophylaxis Therapy — was two multicentre pivotal trials of identical design (ClinicalTrials.gov NCT00156910 and NCT00168428). Each comprised a 24-week randomised, double-blind, placebo-controlled phase followed by a 32-week open-label phase. Patients were randomised 1:1 to onabotulinumtoxinA 155–195 Units or placebo injections every 12 weeks, with study visits every four weeks. A total of 1,384 adults were randomised: 688 to active treatment, 696 to placebo (Dodick DW, Turkel CC, DeGryse RE, et al. OnabotulinumtoxinA for treatment of chronic migraine: pooled results from the double-blind, randomized, placebo-controlled phases of the PREEMPT clinical program. Headache. 2010;50(6):921–936).

The enrolled population was tightly defined. The FDA label records that during a 28-day baseline period, participants had more than 15 headache days lasting four hours or more, with more than 50% being migraine or probable migraine, and were not using any concurrent headache prophylaxis.

The primary pooled endpoint was mean change from baseline in frequency of headache days. At week 24 that was −8.4 with onabotulinumtoxinA versus −6.6 with placebo (p < .001).

Hold those three facts together, because they are what "the PREEMPT protocol" is actually anchored to: a specific population, a specific regimen, and a specific endpoint measured at a specific timepoint. Change any one of them and the outcome data stop applying to what you are doing.

What "fixed-site, fixed-dose" means

In a fixed-site, fixed-dose paradigm, the injector does not choose the sites and does not choose the units. Both are specified in advance and administered identically in every participant.

Dr. Croley describes it in teaching as a relevant absence of movement: these are the 31 injection sites that these patients were treated with, regardless of any kind of movement. Fixed dose, fixed location, for every patient in the trial.

The contrast is the follow-the-pain approach, in which the clinician determines the number of sites and the dose within a protocol-specified range according to where the individual patient's head pain and tenderness are located. Follow-the-pain feels more like clinical medicine, and it is closer to how an aesthetic injector already thinks.

The history matters here, because it explains why both exist. Early botulinum toxin headache work used sets of fixed pericranial sites. The approach then evolved to add sites corresponding to the location of pain and tenderness in the individual patient, in addition to the fixed ones. PREEMPT ended up using both: a fixed core, plus optional additional follow-the-pain sites considered according to individual symptoms (Blumenfeld A, Silberstein SD, Dodick DW, Aurora SK, Turkel CC, Binder WJ. Method of injection of onabotulinumtoxinA for chronic migraine: a safe, well-tolerated, and effective treatment paradigm based on the PREEMPT clinical program. Headache. 2010;50(9):1406–1418).

That is where the 155–195 Unit range in the trial description comes from. The fixed core was 155 Units across 31 sites; physician discretion permitted an additional 40 Units across three muscle groups, for a maximum of 195 Units across 39 sites (Blumenfeld AM, Silberstein SD, Dodick DW, Aurora SK, Brin MF, Binder WJ. Insights into the Functional Anatomy Behind the PREEMPT Injection Paradigm: Guidance on Achieving Optimal Outcomes. Headache. 2017;57(5):766–777).

What the label approves, which is narrower

Here is the fact that gets lost in translation between the trial and the clinic: the fixed-dose, fixed-site paradigm — not the follow-the-pain extension — was the basis for US regulatory approval (Blumenfeld et al., 2017).

The US prescribing information specifies a recommended total dose of 155 Units, administered as 0.1 mL (5 Unit) injections per site across 31 sites, divided among seven head and neck muscle areas: corrugator, procerus, frontalis, temporalis, occipitalis, cervical paraspinal muscle group and trapezius. With the exception of the procerus, which is injected at a single midline site, all muscles are injected bilaterally with half the sites on each side. The recommended re-treatment schedule is every 12 weeks. The site-by-site allocation is in Table 3 of the BOTOX prescribing information, which is where you should read it rather than from any article, including this one.

In some other countries a total of up to 195 Units is approved, allowing additional injections into the temporalis, occipitalis and/or trapezius by the follow-the-pain method (Blumenfeld et al., 2017). That is a jurisdictional difference, not a clinical preference, and it is worth knowing about because a great deal of the online material an injector will encounter is written to a different regulatory regime than the one they practise under.

Why a regulator wants a rigid regimen

A fixed regimen buys three things that a discretionary one cannot.

It is blindable. In a placebo-controlled trial where the comparator is an injection, the number and location of injections has to be identical between arms, or the blind leaks. A protocol in which the clinician decides how many injections to give on the basis of examining the patient is very difficult to conceal from either party.

It is reproducible. Every participant received the same intervention, so the effect estimate belongs to a defined thing. In a follow-the-pain design, the intervention differs between participants and you are pooling outcomes across a family of related treatments rather than measuring one.

It is writable as a label. An approved indication has to specify what is approved. "155 Units across 31 defined sites every 12 weeks" can be printed. "Between 155 and 195 Units at the physician's discretion, guided by tenderness" is a description of clinical practice, not a dosing instruction — which is precisely why the US label took the fixed core and left the discretionary extension out.

The mistake: "I follow the movement"

This is the one that costs efficacy, and it comes from a habit that is correct in the other half of your practice.

In aesthetics we assess dynamic movement, symmetry and the patient's aesthetic concerns, and we tailor to them. That is the right method: the target is a visible motor outcome, so the motor examination is the guide. Empire's material on Botox injection sites, the Botox face chart and frontalis dosing all operate in that logic.

Import it into a migraine patient and the reasoning breaks, because the sites in the PREEMPT paradigm were not chosen for motor function. They were chosen because sensory nerve endings, belonging to neurons whose cell bodies lie in the trigeminal and cervical ganglia, are distributed throughout those muscles — and because many of the injection sites sit adjacent to cranial sutures, emissary canals and fissures through which those nerves pass between the extracranial and intracranial compartments (Burstein R, Blumenfeld AM, Silberstein SD, Manack Adams A, Brin MF. Mechanism of Action of OnabotulinumtoxinA in Chronic Migraine: A Narrative Review. Headache. 2020;60(7):1259–1272).

So "there is no movement there" is not an argument for skipping a site. Nor is "she has no lines in her occipital region." You are not treating movement. You are placing drug where sensory terminals are, in a pattern that was validated as a whole.

The second mistake: shaving the protocol

Injectors shave protocols for understandable reasons — a patient complained of neck pain last time, the trapezius injections stung, the dose felt large, the patient wanted to come in at ten weeks.

The published anatomy guidance names the behaviour and its consequence directly. Unwanted outcomes may cause injectors to lower the dose, or to omit treatment of certain muscles or areas altogether, or to delay repeat treatments while events resolve. The authors caution that "lowering doses, avoiding muscles, or delaying repeat treatments may lead to suboptimal efficacy" (Blumenfeld et al., 2017).

There is a cleaner way to think about it. The moment you drop a muscle group or cut the dose, you are no longer administering the regimen that produced the outcome data. You are administering an untested variant that happens to share a name with a studied one. That may still help the patient — but you have forfeited the right to tell them what to expect, because the expectation you were quoting came from the intact protocol.

The correct response to neck pain from posterior injections is better technique and better site selection within the protocol, not omission of the posterior fields. That is a training problem with a training solution.

Where tailoring is legitimate

None of this means you inject 31 sites by rote at fixed coordinates on every skull. There are three places where judgement belongs, and they are worth naming precisely so that "tailoring" does not become a euphemism for freelancing.

Individual anatomy locates the sites. The standard fingerbreadth measurements used to describe the paradigm are general guidance; the actual injection sites should be located on the basis of the patient's unique anatomy, because muscle mass, brow width and height, and preexisting weakness all vary considerably between patients (Blumenfeld et al., 2017). Fixed-site means fixed relative to the anatomy, not fixed relative to a ruler.

Baseline assessment changes what you watch for. Each patient should be examined before treatment for preexisting eyelid or eyebrow ptosis, pseudoptosis, neck pain and neck weakness. Patients with small frames may be predisposed to weakness after trapezius injection. This does not change the regimen; it changes what you warn them about and what you monitor.

Follow-the-pain, where the regulatory context permits it. Where additional units are allowed, the guidance is specific: additional temporalis drug goes into up to two additional sites per side, not into the standard sites; additional occipitalis drug goes as two injections into one side or one into each, depending on where maximal tenderness is. And there is one explicit exception — the authors state that in their experience neck weakness and neck pain are associated with high doses into the trapezius, and therefore additional trapezius injections should generally be avoided where possible, even under a follow-the-pain strategy (Blumenfeld et al., 2017).

Outside those three, the cash-pay off-label setting is its own conversation. Tailoring there is legitimate precisely because you are no longer claiming to deliver the on-label regimen — but that only works if you have said so to the patient, in writing.

Three questions to ask of any protocol

The habit worth taking from this article generalises well beyond migraine.

What was fixed and what was discretionary? If a regimen came out of a trial, find out which parts the investigators were allowed to vary. The discretionary parts usually did not make it into the label.

Which population generated the number? The PREEMPT effect estimate belongs to adults with more than 15 headache days a month lasting four hours or more, over half of them migrainous, not on other prophylaxis. It does not belong to your episodic patient, and the label's Limitations of Use says so after seven negative studies.

Which endpoint, at which timepoint? Week 24, mean change in headache days per 28 days. Not "patient satisfaction," not "at two weeks." If you assess at a different point against a different measure, you cannot expect the published number.

At the chairside

• Read the site allocation from the prescribing information, not from a diagram on a slide deck.
• Locate sites on the patient's anatomy, not on fingerbreadths alone.
• Inject the whole regimen or say plainly that you are not.
• Hold the twelve-week interval.
• Treat neck pain as a technique problem to solve, not a field to abandon.
• If you are tailoring, be honest with yourself that you have left the on-label regimen — and be honest with the patient in the consent.

Rigorous head and neck anatomy is taught hands-on in Empire's Anatomical Based Aesthetics Training, and neurotoxin fundamentals in Cosmetic Neurotoxins Training.

This guidance reflects Dr. Chris Croley's clinical practice as taught in Empire Medical Training's hands-on curriculum. Technique is learned under supervision; this article is educational and is not a substitute for training.

Related guides in this cluster

Part of Therapeutic Neurotoxin: Chronic Migraine.

Clinical GuideChronic Migraine On-Label Criteria: Deciding Which Patients Actually QualifyChronic migraine criteria decide who you can treat on-label. The label wording, the ICHD-3 definition, the diary, and the documentation Clinical GuideConfirm the Diagnosis First: Secondary-Headache Red Flags Before You InjectSecondary headache red flags decide whether a migraine patient gets injected or referred. The SNNOOP10 framework, the numbers behind it Clinical GuideMigraine Treatment Outcome Measures vs Aesthetic Endpoints: Two Assessment FrameworksMigraine treatment outcome measures are disease burden, not movement. The four domains to record, the responder thresholds to pick, and