The fighting is winding down. The WHO’s record of it is permanent

The ceasefire has held, more or less, since October 2025, and at the end of July the next phase was announced. Whatever comes of it, the war is moving out of the field and into the archives. What gets written down now — in UN submissions, in prosecution dossiers, in the academic literature — is largely fixed.

Much of it rests on a single database, and Gregory Rose, a professor of law at the University of Wollongong, and I have spent months analyzing it for a report published by the Henry Jackson Society and The Hague Initiative for International Cooperation. We found that the database cannot bear the weight.

The World Health Organization’s Surveillance System on Attacks on Health Care, launched in 2017, holds 3,229 records for what it classifies as “Occupied Palestinian Territory”. That figure, or numbers derived from it, appears in special rapporteurs’ statements, in NGO submissions to UN bodies, and in the journals now assembling the case against Israel. It is presented as verified. It is treated as a casualty count. It is neither.

Start with the container. The WHO folds East Jerusalem, the West Bank and Gaza into one reporting unit. A checkpoint delay in Jerusalem, an ambulance search near Jenin, and an airstrike in Khan Younis all land in the same bucket and surface as one number. Anyone citing “over three thousand attacks on health care in the Palestinian territories” is describing a war in Gaza using a figure substantially composed of events that happened somewhere else.

How much of it? By our analysis, 38.6 percent of the events in that bucket occurred during periods when there was no armed conflict at all. Under international law, an “attack” is an act of violence connected to an armed conflict. For nearly four in ten of these records, the precondition for the word “attack” is simply absent.

The word is doing other work it cannot support. The WHO defines an attack as any act of verbal or physical violence, threat of violence, psychological violence, or obstruction interfering with health services. This includes insults, arrests, detentions and searches. By the WHO’s own written methodology, a doctor who arrives late for a shift after a checkpoint delay is an “attack on health care.” So is an ambulance searched for weapons, a practice the Red Cross accepts as lawful and one Israel has obvious cause to maintain.

The consequence shows up in the severity data. We classified as “severe” those events involving heavy-weapons violence with no recorded militarization of the facility. In the WHO’s Palestinian bucket, 26.1 percent clear that bar; in Ukraine, 88.6 percent do. Only 6.3 percent of the Palestinian records involve a death, against 42.1 percent in Somalia and 45.5 percent in South Sudan.

Then there is the question the methodology does not ask. Under the Geneva Conventions, a hospital forfeits its protected status when used to commit acts harmful to the enemy — as a command post, a weapons store, a tunnel access. The WHO records “militarization” as an event category, yet publishes nothing indicating that such a facility is treated differently from a protected one. The database appears to hold 112 events for Gaza between November 11 and 24, 2023, when the IDF surrounded Al-Shifa and searched it after three days of failed evacuation talks. An operation conducted because a hospital had lost its protection sits in the same ledger as an attack on one that had not.

Nor does the WHO verify what it publishes. Staff score their confidence that an event occurred; they do not check reported deaths and injuries against death certificates or medical records, and they do not determine responsibility. One record that appears to correspond to the killing of a Médecins Sans Frontières physiotherapist — whom Israel identified as a member of Palestinian Islamic Jihad — reports six killed and six injured as an attack on health care, with no public evidence the others were medical personnel.

And consider the entry at the top of the entire global dataset. The deadliest event the WHO has recorded anywhere, in any conflict, is the explosion at Al-Ahli Arab Hospital in October 2023 — blamed on Israel within hours, then traced by forensic analysis to a rocket that misfired on launch inside Gaza. The record stands as first entered. Israel cannot contest it. No state can.

Which brings us to who is filing. The WHO relies on country offices and pre-identified local partners. In Gaza, the health ministry answers to Hamas. A health worker who reported the terror group’s use of a hospital would lose his job at best; that same worker has every reason to report Israeli conduct in detail. When we tried to match a random sample of records against public sources, we could not verify more than 90 percent.

None of this means violence against health care in Gaza did not occur, or that it does not deserve documentation. It did, and it does. Our point is narrower: a dataset assembled this way cannot support the comparisons, the attributions, and the legal conclusions now being built on it.

There is a moment to press this. Nominations for the next WHO director-general close on September 24, with the appointment due in May 2027. The organization should be asked for neutral terminology in place of “attack”; a prominent statement that its data supports neither cross-country comparison nor findings of legal responsibility; verification of reported deaths and injuries; and a public right of reply for the states these records describe.

The war is passing into history. The record of it should be one that can survive examination.