Who actually decides what Ebola PPE looks like

Write "per WHO guidance" into a PPE specification and you have specified almost nothing. Not because WHO is vague, but because at least four bodies publish a position on filovirus PPE, they do not agree, and the disagreements fall on exactly the items that cost money: the respirator, the coverall and the doffing protocol.

None of the four is wrong. They are answering different questions. WHO is answering what the evidence supports. CDC is answering what a nurse can do in a room at three in the morning. ECDC is answering what a European hospital should hold as standard. MSF is answering what has worked in a treatment centre in an outbreak, having run more of them than anyone else. A buyer who does not know which of those questions their specification is copying will end up with an ensemble nobody actually asked for.

This article is drawn from documents read in full: WHO's Infection prevention and control guideline for Ebola and Marburg diseases, 2026 edition, and the Preferred Product Characteristics for Personal Protective Equipment for the Health Worker on the Frontline Responding to Viral Hemorrhagic Fevers in Tropical Climates, 2018, which the guideline points to for product characteristics. The field and procurement positions are taken from public catalogues and published evaluations, cited in place.

Key takeaways

What do the four positions actually say?

Four columns, four answers, and only one of them is what most tenders think they are citing.

What does WHO recommend, and how strongly?

WHO specifies the ensemble by role, not as one kit list, and the tiers are further apart than most specifications assume. Screening at one metre or more with a strict no-touch approach requires no PPE at all, provided the engineering and administrative controls hold. Screening closer than a metre requires scrubs, covered shoes, one pair of nitrile gloves, a medical mask with eye protection and a fluid-resistant gown. Triage moves up to a fluid-resistant coverall and two pairs of gloves. Direct and indirect patient care adds a head-and-neck covering where a gown is used, plus an apron. Cleaning, hygiene, mortuary and burial work takes the direct-care ensemble and adds heavy-duty outer gloves, a heavy-duty apron and waterproof boots.

The mask recommendation rests on how filoviruses move. WHO's evidence review concludes that transmission in humans "does not naturally occur via the airborne route or via the respiratory droplet route". Blood is the most infectious body fluid. The mask is there to stop splashes reaching the nose and mouth, not to filter air. So the requirement is a fluid-resistant medical mask with a structured design that does not collapse against the mouth, a duckbill or cup shape, because masks get wet in tropical conditions and a flat pleated mask collapses onto the lips when they do.

Then comes the qualification almost everyone misses. Respirators may be substituted for medical masks where an off-the-face design is wanted and a suitable mask is not available, "and not for protection from aerosols". A bid offering FFP2 is therefore not automatically non-compliant with WHO. A specification demanding FFP2 as aerosol protection for routine care is.

The strength of the recommendations matters more than the content, and it is the part nobody reads. The great majority of the ensemble recommendations are conditional on very low certainty evidence: the screening ensemble, the triage ensemble, the direct-care ensemble, double gloves, the apron, eye protection worn under the head covering. WHO's systematic reviews found no primary studies comparing coveralls with gowns for infection outcomes at all. The strong recommendations are a short list: mucous membranes completely covered, a face shield or goggles, the structured fluid-resistant medical mask, a respirator for aerosol-generating procedures, nitrile preferred over latex, hand hygiene, and the recommendation against spraying.

One change since the 2016 edition is instructive. Double gloving used to be a strong recommendation on moderate certainty evidence. It is now conditional, because "a reassessment of the evidence resulted in a lower rating of the certainty". The practice did not change. The confidence behind it did.

Head-and-neck covering is the most honest passage in the document. Two crossover trials found more contamination without it but more errors during removal with it, and simulation studies found heat outcomes significantly worse with it. WHO's own conclusion is that the balance "probably favours covering the mucous membrane only", and that the covering stays in the recommendation because "the status quo in terms of current practice includes the head-and-neck skin coverage and stronger evidence is needed to change these practices".

That is a guideline telling you, in print, that it is holding a position for want of evidence to move it. It deserves to be read as such.

Why do CDC and ECDC land somewhere else?

Because they are optimising for something different, and both say so.

CDC's position is a human-factors judgement rather than a virological one: "Because there might not be time for staff to leave the room to don proper PPE for an aerosol-generating procedure, CDC recommends that all healthcare workers entering the room of a patient with VHF wear respiratory protection that would protect them during an aerosol-generating procedure." WHO's rule is per procedure. CDC's is per entry. Neither disputes the transmission route. They disagree about whether a worker can reasonably be expected to leave, re-don and return when a patient deteriorates.

ECDC, updating on 2 June 2026, makes FFP2 or higher the hospital standard and states that "coveralls are preferred to gowns, which incompletely cover the head and neck". In some jurisdictions the question is not guidance at all: Cal/OSHA legally mandates powered air-purifying respirators where an aerosol-generating procedure might occur, with no skin exposed.

Two further pressures push the same way. WHO's own responder survey, published by Den Boon and colleagues in F1000Research in 2018, found that 33 of 44 respondents, 77 per cent, preferred the more robust ensemble components. WHO recommended the lighter ensemble anyway, which is a defensible reading of the evidence and an uncomfortable one to hand to a worker. And optimised supportive care has put CPAP and high-flow oxygen into the standard of care for Ebola without any change to the PPE guidance. Both are aerosol-generating. If a treatment centre offers them, the "except during aerosol-generating procedures" clause stops being an exception and starts being the working day.

What does MSF specify, and why does that matter?

Because MSF is the field implementer whose practice much of the sector copies, and because MSF says so about itself. Its Stockholm Evaluation Unit review of the West Africa response records that "because of its recognised experience in dealing with Ebola, MSF rapidly became one of the major references for setting the technical standards regarding PPE", and that MSF Supply provided over 1.5 million euros of PPE to third parties including the French Red Cross, International Medical Corps and Save the Children. Specification changes it made, the review says, "led to the creation of MSF standards (e.g. overalls) which became unique on the commercial market".

MSF's international supply catalogue is public and gives item-level specifications. The viral haemorrhagic fever isolation module contains a hoodless chemical protective coverall certified to EN 14126 for protection against infective agents from risk groups 1 to 4, to ISO 13688, and to chemical protection types 3, 4, 5 and 6. Over that goes a separate nonwoven hood with an integrated Type IIR surgical mask, certified to EN 14126, with fabric rated at least 4 of 6 on both ISO 16603 and ISO 16604, seams sealed or ultrasonically welded, PPE Category III. And with that, an unvalved FFP2/N95 duckbill respirator certified to EN 149+A1:2009, to NIOSH 42 CFR 84 and to EN 14683 Type IIR, at 1,680 units per module, with the catalogue noting that "first choice is the large duckbill design", that other models "are not appropriate for VHF / Ebola contexts", and directing users to order a fit-test kit. The rest of the module is panoramic indirect-vent goggles, a reusable nitrile apron, white rubber boots, extended-cuff nitrile examination gloves with a minimum 290 mm cuff, sterile surgical latex gloves for the second pair, and a four-poster donning and doffing protocol set with mirrors.

Two things there are worth pausing on. First, MSF carries both mask types simultaneously, a Type IIR built into the hood and a respirator worn with it, which is not how the divergence is usually described. Second, the divergence from WHO is real but narrow: it is the respirator, specified as a standard component in volumes consistent with routine use and backed by fit-testing, where WHO specifies a medical mask.

The second documented divergence is doffing, and it is WHO that documents it. In the justification for its strong recommendation against spraying, WHO writes that other organisations including CDC and ECDC do not recommend spraying, and then: "MSF guidelines do still include spraying of health and care workers during the PPE-removal process. The chlorine spraying is done once at the beginning and again immediately prior to PPE removal; hand hygiene is performed during the remaining PPE-removal process."

That is not a quarrel. MSF staff sat on the Guideline Development Group and among the external reviewers for the guideline itself. It is a guideline recording, transparently, that its strongest field implementer has not adopted one of its strong recommendations. If a customer's protocol still sprays, that is now a documented divergence from WHO rather than a local habit, and it should be written down as such.

Are the procurement specifications the outlier they appear to be?

No, and this is where the received wisdom is most wrong.

The intuition runs that humanitarian responses consume coveralls, N95 respirators and goggles in volume, therefore the procurement specifications must have departed from WHO. Read the specifications and the opposite is true. UNICEF's public supply catalogue lists a Category III Type 3-B coverall requiring "EN 14126:2003 certified passing infectious agent test according to ISO 16604:2004" at a "minimum exposure pressure of 1.75kPa (class 2)", under Regulation (EU) 2016/425, with Type 3 liquid-tightness to EN 14605:2005+A1:2009. ISO 16604 class 2 is WHO's own performance class. The specification has taken WHO's performance requirement and expressed it through the European conformity route. A Type 6-B item sits alongside it for lower-exposure use, which maps onto WHO's role-based tiering rather than departing from it.

The respiratory catalogue is a ladder, not a substitution. Type IIR surgical masks to EN 14683, unvalved FFP2/N95 to EN 149:2001 or NIOSH 42 CFR 84, a fluid-resistant FFP2 surgical N95 citing "EN 149, minimum FFP2 and EN 14683 Type IIR", and FFP3/N100 above that. The third of those is, line for line, WHO's own specification for aerosol-generating procedures. The catalogue stocks the medical mask, the hybrid WHO asks for, and a higher tier. It has not replaced one with another.

So the gap is not between guidance and specification. It is between specification and consumption. A humanitarian framework is a catalogue covering every role, risk tier and scenario a response might meet, and it is written permissively. What differs is which lines get pulled in volume, and that is decided by field protocol and risk perception, not by the standards in the specification. Which is precisely why the four positions above matter more to a buyer than the catalogue does.

What happens when one conformity framework cannot supply the demand?

The specification stops naming one framework, and that is not an accident.

Read the published isolation gown specification in the same catalogue and it does not sit inside the European system at all. It cites AAMI PB70 Levels 1 to 3 with ASTM F3352, EN 13034 Type PB [6] with a minimum hydrostatic head of 50 cmH2O, and, at the top of the range, AAMI PB70 Level 4 with ASTM F3352, or ISO 16604 Class 5. It then closes the list with "or alternative equivalent set of standards". The coverall specification does the same thing more quietly: EN 14605:2005+A1:2009 "or equivalent", and the ISO 16604 class 2 requirement qualified as "or equivalent international standard". The respirator page offers NIOSH 42 CFR 84 or the EN 149 and EN 14683 pair as alternatives, with the same equivalence clause underneath.

That clause is doing real work. A specification written to a single conformity route can only be filled from the pool of product certified to that route, and that pool is smaller than the market. In ordinary times it is large enough. In a surge it is not, and the buyer has two options: widen the specification or wait. Waiting is not an option in an outbreak, so the specification widens. This is the ordinary mechanics of the supply side and it is the reason a European buyer meets American barrier levels in humanitarian work far more often than in domestic healthcare procurement.

Here is the problem that creates, and it is the same document that flagged it. WHO's own product characteristics work says that because the test methods and performance requirements cannot be compared directly, "it is difficult to assign equivalency between surgical gowns classified according to EN 13795 and ANSI/AAMI PB70". So the buyer is instructed by one WHO document that equivalency is hard to assign, and writes an equivalence clause into the specification because supply gives no alternative. Both are correct. They do not sit comfortably together.

The way out is to map at the level of the test method rather than the level name, and the catalogue's own top tier shows how. AAMI PB70 Level 4 is defined by ASTM F1671, no penetration at 2 psi, 13.8 kPa, using a phi-X174 bacteriophage challenge. ISO 16604 uses the same phi-X174 challenge, and its procedure applies 14 kPa. Two frameworks, one test, effectively one pressure. That pairing is defensible, which is why the specification makes it. What is not defensible is treating a level name in one system as a level name in the other, or assuming that a garment class carries across when the underlying methods, the tested areas and the pass criteria differ.

The practical consequences run past the paperwork. Two nominally equivalent gowns from two frameworks are two stocks with two sets of documentation, and the risk assessment, the doffing protocol and the training were written for one of them. If you are going to accept a second framework, and in a surge you will, decide in advance which test result you are actually requiring, ask for that test report on both, and record which framework each delivery was accepted under. That is a five minute decision at specification stage and an unanswerable question afterwards.

This part is our own observation from the supply side rather than something WHO has published. The specifications quoted above are public.

Where do the standards themselves run out?

Further up than anyone would like, and WHO is candid about it. The guideline defers product detail to the 2018 Preferred Product Characteristics document, which is explicitly research and development guidance rather than a specification, and which states that "because there is scant technical data to support PPE protection effectiveness, little or no precise measurements or specifications on PPE can be presented".

Its findings are the honest picture of the field. No standard covers the ensemble: "none exists solely for the preferred characteristics or for evaluating of a full PPE ensemble specifically for health worker at the frontline against Ebola virus." Items are certified individually and the combination is not. No harmonised minimum exists for biological-agent PPE. European and American gown classes are not interchangeable, because the test methods and performance requirements cannot be compared directly, so "it is difficult to assign equivalency between surgical gowns classified according to EN 13795 and ANSI/AAMI PB70". Anyone offering an equivalence table between them is going further than WHO will.

The test methods do not model the conditions either. Generally only the primary material is tested, not seams and junctions. Penetration tests run 60 minutes. Only hydrostatic pressure is applied, not the mechanical pressure of leaning or kneeling. The synthetic blood used has a surface tension that "may not be applicable for the other body fluids which may be more common during Ebola". And simulated Ebola particles were recovered from saturated N95 respirators and surgical masks, leading to the conclusion that "existing standards are therefore not protective enough under conditions of heat and saturation, and must be examined and redefined".

One finding should change how you read a certificate. In a 2015 study of 22 commercial single-use isolation gowns, nine of them, 41 per cent, failed to meet the AAMI PB70 liquid barrier performance at the level their own manufacturer specified. WHO cites that as part of the reason an apron is worn over the gown.

A last detail for a European buyer. WHO's standards references have aged. Directive 86/686/EEC and 89/686/EEC were repealed by Regulation (EU) 2016/425, and the medical device directive by Regulation (EU) 2017/745. The 2018 document's glove table cites EN 455 Part 1: 2002, EN 455 Part 2: 2011, EN 420:2004 and a bare EN 374, where the current editions are EN 455-1:2020+A2:2024, EN 455-2:2024, EN ISO 21420:2020 and the EN ISO 374 series. That is not a criticism of WHO so much as a demonstration of why editions belong in your own specification rather than borrowed from someone else's. Our guide to the EN 455 series sets out what each part now covers, and our examination gloves are specified against the current editions for that reason.

What should a specification actually say?

Name the authority, then do the product specification yourself.

Underlying all of it is a line from the guideline that deserves quoting more often: "Paradoxically, wearing excessive PPE can increase a person's risk of infection." PPE sits at the bottom of the hierarchy of controls, below engineering and administrative measures, and WHO's own background notes that "excessive PPE may lead to increased self-contamination, yet excessive PPE is commonly used". The four positions in this article are not a ladder from cautious to reckless. They are four readings of the same evidence, and the buyer has to pick one deliberately.

Questions we get asked

Do we need FFP2 or FFP3 respirators for an Ebola response? It depends whose logic you are buying. WHO specifies a fluid-resistant structured medical mask, with a respirator reserved for aerosol-generating procedures. CDC recommends respiratory protection for every room entry, ECDC makes FFP2 the hospital default, and MSF carries a fit-tested FFP2/N95 as a standard module component. All four are defensible. Decide which one your specification is following and say so.

Is a Type 3-B or 4-B coverall required? WHO does not specify an EN 14126 Type for filovirus work. It requires fabric tested for resistance to penetration by blood and body fluids or by bloodborne pathogens, and it uses ISO 16604 class 2 as the performance level. Humanitarian catalogues express that as a Category III Type 3-B garment. Specify the property, the test and the class, and let the supplier show how their certification meets it.

Can we treat EN 13795 and AAMI PB70 levels as equivalent? Not as level names. WHO states that the test methods and performance requirements cannot be compared directly and that equivalency is difficult to assign. Where the two systems share a test, the comparison is sound: AAMI PB70 Level 4 is defined by ASTM F1671 at 2 psi, 13.8 kPa, with a phi-X174 challenge, and ISO 16604 uses the same challenge with its procedure applying 14 kPa. So specify the test and the pressure, not the level name, and you can accept product from either framework without inventing an equivalence.

We cannot get enough product certified to one framework. What now? Widen the specification deliberately rather than by exception. Published humanitarian specifications already do this: the isolation gown page cites AAMI PB70 Levels 1 to 3 with ASTM F3352, EN 13034 Type PB [6] at 50 cmH2O hydrostatic head, and AAMI PB70 Level 4 or ISO 16604 Class 5, then adds "or alternative equivalent set of standards". That is a buyer accepting in advance that one certified pool will not fill a surge. Copy the structure, name the test method for the tier you actually care about, and keep a record of which framework each delivery came in under.

Does chlorine spraying during doffing still have a place? WHO makes a strong recommendation against it, on a survey of 1,550 people including 500 health workers, of whom 92 per cent were sprayed when leaving a treatment unit: 64 per cent reported chest symptoms, 48 per cent eye problems including after a single exposure, and about a third skin irritation. Some had symptoms despite wearing PPE, which WHO reads as suggesting the PPE did not protect them from the chlorine. WHO's own guideline records that MSF protocols still include spraying. If a protocol sprays, treat that as a documented divergence and record the decision.

How long can staff work in full PPE? The 2018 document defines a work period as "between 40 minutes and 4 hours, subject to IPC protocol", and notes that during the West African response most workers could tolerate around 40 minutes. Plan consumption on that basis rather than on a shift length.

Sources

WHO, Infection prevention and control guideline for Ebola and Marburg diseases, 2026, ISBN 978-92-4-011133-2. WHO, Preferred Product Characteristics for Personal Protective Equipment for the Health Worker on the Frontline Responding to Viral Hemorrhagic Fevers in Tropical Climates, 2018, ISBN 978-92-4-151415-6. Den Boon S et al., "Development of a Preferred Product Characteristics document for personal protective equipment", F1000Research, 2018. MSF Stockholm Evaluation Unit, OCB Ebola Review Part 5: Supply functioning, December 2015. MSF international supply catalogue, viral haemorrhagic fever isolation protective equipment module, public item specifications. UNICEF Supply Catalogue, public item specifications for Category III coveralls, disposable isolation gowns and respiratory protection. ANSI/AAMI PB70 barrier levels and the ASTM F1671 test conditions as recorded by NIOSH PPE-Info. ECDC outbreak guidance, 2 June 2026. Standards editions are as recorded in our own standards library, verified against the source documents.

Niels Kristian BitschManaging director, Eastwest Medico ApS

Seventeen years sourcing and supplying medical gloves and PPE to governments, UN agencies, NGOs and healthcare systems. Writes on glove standards, quality management and responsible sourcing for institutional buyers, working from the source standards rather than secondary summaries.

View profile on LinkedIn