Last updated: 2026-11-17
On 2026-10-30 Judge William H. Orrick of the Northern District of California granted class certification in In re Meta Pixel Healthcare Litigation, No. 3:22-cv-03580-WHO, the consolidated case over the Meta Pixel and other Meta Business Tools on healthcare provider websites and apps. Four classes of Facebook users are certified, for a class period from June 17, 2018 to the present. From the conclusion of the 37-page order (Dkt. 1425): "plaintiffs' motion to certify the classes is GRANTED, except for the limited Section 631 and redefined CDAFA class." Certification means the claims proceed as class claims. Meta's liability is undecided, and the facts below about Meta's data collection are plaintiffs' allegations and expert opinions.
What are the four classes in the Meta Pixel healthcare lawsuit?
The class definitions are on page 3 of the order. "Class Healthcare Providers" are the entities plaintiffs identify as covered by HIPAA, on a revised list the court let plaintiffs file (Dkt. 1377; footnote 1, page 3).
| Class | Definition (quoted from the order) |
|---|---|
| Patient Status Button-Click | "All Facebook users who had at least one patient portal, medical appointment, bill payment, or diagnostic assessment button click obtained by Meta from Class Healthcare Providers." |
| Patient Health Information | "All Facebook users who had at least one communication or fact identifying a doctor, condition, or treatment acquired by Meta from Class Healthcare Providers, excluding Button-Click events described above." |
| Prescription Drug Information | "All Facebook users whose communications with a prescription drug website for which they have a prescription, were obtained by Meta." |
| Patient Device Intrusion | "All Facebook users for whom Meta placed an _fbp cookie and/or used computing resources by sending data to Meta through their device and/or by computing Core Setup logic on their device while at a Healthcare Provider property." |
Claims for breach of contract, breach of the implied covenant, intrusion upon seclusion and under California's Comprehensive Computer Data Access and Fraud Act (CDAFA, Penal Code § 502) are certified for all four classes; claims under the federal Electronic Communications Privacy Act (ECPA) and the California Invasion of Privacy Act (CIPA) are certified for the first three. For CIPA § 631, Judge Orrick certified "subclasses of patients who either communicated from California or exchanged a communication with a California-based healthcare provider" (page 16). For the device class, plaintiffs agreed to amend the class to remove Conversions API ("CAPI") and "customer lists" uploads after Meta argued those do not affect any plaintiff's device (footnote 16, page 17). The classes are certified under Rule 23(b)(3) for damages and Rule 23(b)(2) for an injunction (page 20).
HIPAA bears on the case as evidence for the pleaded claims. Judge Orrick held that HIPAA is relevant "to establishing reasonable consumer expectation" and possibly to a reasonable consumer's understanding of Meta's contractual promises "requiring partners to have 'lawful rights' to the information sent to Meta" (footnote 10, page 13).
Does the class cover healthcare apps or only websites?
Both. Plaintiffs sought certification for Facebook users whose information Meta obtained, or whose devices Meta affected, "when they were on their healthcare providers' websites or applications ('apps')" (page 1). Meta argued that because the classes "now includes healthcare apps and pharmacies", Judge Orrick's prior rulings foreclose a finding that a class action is the superior method. He answered that in 2023, when he declined to relate cases against "healthcare-adjacent" providers (apps for birth control, healthcare coupons and counselling), he left open whether a class action against Meta was appropriate: as he had told counsel, "the shape of this case would be determined on class certification. Here we are." (page 19).
Most class evidence discussed in the order comes from websites: button clicks, prescription drug site URLs, the _fbp browser cookie. How much of the record covers events sent by Meta's App Events SDK from healthcare apps is not stated in the order; the expert reports, filed partly under seal, would establish it. For apps, from Meta's Android App Events documentation: "certain App Events are automatically logged and collected for Events Manager, unless you disable Automatic App Event Logging".
How did the court answer Meta's Pixel configurability defense?
Meta argued "that since the Pixel is configurable and websites and apps were able to refuse the Business Tools defaults, there are some websites and apps that did not share protected healthcare data of patients with Meta" (page 5). In answer, Judge Orrick pointed to the rebuttal report of plaintiffs' expert Dr. Zubair Shafiq, "citing academic research showing that Automatic Events configuration was active on 99.6 percent of healthcare websites from 2019 to 2022" (page 6). If summary judgment or trial establishes that from some date Meta's default settings stopped collecting this data, or that Meta had told healthcare entities how to change them, he held, "[t]hat may result in a narrowing of the class periods or available damages or restitution" (page 6).
Judge Orrick called Meta's data retention "[t]he related, and admittedly more difficult issue for plaintiffs": Meta argues that under its standard retention policies its oldest classwide data is from March 2022, and the class period starts in June 2018 (page 6). He held that the gap can be tested "at summary judgment, at trial, or through other case management techniques" (page 7). From page 2:
Meta's opposition confuses complexity – which undoubtedly this action presents– and individualized questions.
What did the order say about consent?
Judge Orrick treated consent as a question for common proof. Under the Ninth Circuit's Calhoun v. Google decision, consent must be "actual", and what a reasonable user would understand from the disclosures is an objective question (pages 11–14). Evidence of what each provider told its patients about sharing data with Meta "covers the hundreds, thousands, or tens of thousands of class members who secured healthcare from each of those entities" (page 12). Judge Orrick cited Frasco v. Flo Health, the Flo app case, for putting on the defendant the burden of showing that implied consent requires a "class-member-by-class-member" determination (page 15). Meta's intent and knowledge were held common too: how Meta responded to what it received "is common evidence" (page 15).
What did the court rule on the expert reports?
Every motion to exclude an expert, from either side, was denied; the one against Dr. L. Glenn Cohen was denied without prejudice to objections at trial (page 28). Shafiq's Opinion 1, quoted in the order: "Meta can and does programmatically identify healthcare provider websites and apps where it collects data via Meta Pixel and other Meta Business Tools" (page 21). For the button-click class, Shafiq started from "7.3 million buttons associated with 1,844 domains", using manual review, hand-written labelling patterns and an AI model. Meta called the AI extension to "the up to 650,000 domains at issue" "unscientific, untestable, and unverifiable"; Judge Orrick held these are "sampling methodologies that can be run, verified, and challenged by Meta's own experts" (pages 22–23). The motions to exclude the damages surveys of Dr. Hal J. Singer and the data valuation and profit models of Todd A. Zigrang were denied with the rest (pages 23–28).
Source code and the table and field names of systems Meta no longer uses may remain sealed for now, "subject to revisiting the sealing issue on summary judgment or trial" (pages 35–36). Judge Orrick ordered that "General descriptions of how Meta acquired data from Class Healthcare Providers, how Meta used that information within its own systems, including how Meta matched that information with its users and repurposed that information for improving its own systems or serving advertisements to its users, should be disclosed" (page 36). Among the passages Meta asked to keep sealed are what plaintiffs describe as "disclosures from Meta demonstrating that the '[f]ilter did essentially nothing'" (page 34).
How does this order fit with other pixel rulings?
Judge Orrick distinguished four 2026 Northern District of California rulings in tracking cases. In Ingraham v. Capital One, individualized issues predominated where at least eight types of technology from eleven third parties were at issue; "the only technology at issue here is Meta's" (footnote 3, page 6). Smith v. Rack Room Shoes (certification denied) and Gershzon v. Meta (case dismissed) turned on Article III standing; he distinguished both by "[t]he nature of the information at issue in this case" and the expert testimony of acquisition and matching (footnote 7, page 10). In Hinton v. Planned Parenthood, plaintiffs relied on "generalized evidence about Meta and Google's business goals and promises"; here, the evidentiary showing "extends far beyond public reports on Meta's business operations" (page 11).
Meta Pixel and Conversions API events from a telehealth company are alleged in the FTC's lawsuit against Hims & Hers. The _fbp cookie is also at issue in the Meta localhost case over the Facebook and Instagram Android apps, where on 2026-11-02 Judge Rita F. Lin let most claims against Meta proceed.
What happens next in the case?
Judge Orrick ordered a revised class definition within 10 days (2026-11-09), and a proposed notice plan and one joint sealing chart within 30 days (2026-11-29); the parties may stipulate to extend the 30-day deadlines. Stipulations, orders on stipulations, a case management statement and notices were filed on the docket from 2026-11-09 to 2026-11-12; their texts are on PACER only, and whether they move these deadlines is open until those documents are read. Under Rule 23(f), a petition for permission to appeal a class certification order is filed with the circuit clerk within 14 days after the order is entered; for this order, that period ended on 2026-11-13. No press report of a petition by Meta was found as of 2026-11-17; the Ninth Circuit docket, where a petition is filed, is unchecked.
What does an app scan show for Meta App Events in a health app?
On the app side, the counterpart of the Pixel is Meta's App Events SDK. When I scan a build, the CanITrustThat static findings include:
- the Meta App Events (Facebook) SDK linked in an Android or iOS build;
- automatic event logging left at its default on Android: no
com.facebook.sdk.AutoLogAppEventsEnabledmanifest entry set tofalseand no call from app code turning it off (the developer can also turn it off in Meta's Events Manager, a setting absent from the build); - a reference in Android code to
AppEventsLogger.setUserData, the Advanced Matching call through which hashed email, phone and other user fields are attached to app events; - an ad or attribution SDK, Meta App Events with automatic logging left on, or one of several analytics SDKs (Firebase Analytics with collection left on among them) in an Android app listed in a health or medical category on Google Play.
A static finding is evidence of presence in the build. App Events requests to Meta's graph.facebook.com host, with their event names and parameters, are recorded in a traffic capture, before and after any consent screen. Both are limited to the build scanned, on the scan date. The federal rule for health apps that share data with ad platforms is covered in the FTC Health Breach Notification Rule post.
When I scan an app, every static finding is linked to its file and line, and every traffic finding to its capture. What a scan checks, what a capture records and their limits are on the methodology page. I run app tracking investigations for lawyers and the researchers working with them, on one app or a list in bulk. Sign up for a CanITrustThat account to run your own research, let us run an investigation for you, or read the app teardowns.