Protocol census · a finding

The same procedure. A different person deciding.

A paramedic decompresses a chest on their own authority in one county and telephones a physician for permission in the next. Both crews trained to the same national standard. Comparing 190 agencies’ protocol books across 44 states shows how far that reaches, and it reaches further than the procedure list does.

Procedures where at least 5 agencies make it a standing order and at least 5 require contacting a physician first. Each bar counts agencies, not pages: an agency whose book names a procedure on nine pages is one practice, not nine votes.

Across 2,985 agency-and-procedure pairs

18%

42%

13%

27%

Standing order — act, then report Contact required — call first Depends on the presentation Not stated in the book

“Depends” is a real answer, not a muddle. Those books give BOTH answers for the same procedure — a standing order in one presentation, a call to a physician in another, most often cardiac arrest against everything else. Counting such an agency on both sides would make the columns add up to more agencies than exist, and picking a side for it would invent a policy its book does not have. So it is counted once, here. The book names the procedure and does not say on whose authority. That is a reading of a real page, not a gap — so it is counted and shown, never folded into either answer.

Where agencies disagree about who decides

27 of 33 procedures in the vocabulary are contested: enough agencies commit to each single answer that neither is one document’s habit. Ordered by how many agencies took a side, so the best-evidenced disagreements sit at the top. Every agency appears in exactly one column of a row, and the four columns add up to the agencies naming that procedure.

Procedure	Agencies naming it	Standing order	Contact required	Depends	Not stated
12-lead ECG acquisition and transmission	170	22	86	34	28
Supraglottic airway	158	34	73	19	32
CPAP / BiPAP (non-invasive ventilation)	162	30	70	40	22
Intraosseous access	176	27	71	57	21
Transcutaneous pacing	146	33	65	11	37
Needle chest decompression	148	32	60	31	25
Synchronized cardioversion	136	26	65	14	31
Endotracheal intubation	164	21	69	54	20
Tourniquet application	158	21	69	24	44
Spinal motion restriction / selective clearance	153	22	66	18	47
Surgical cricothyrotomy	126	20	61	5	40
Field termination of resuscitation	115	15	57	11	32
Hemostatic dressing / wound packing	122	34	38	15	35
Mechanical CPR device	106	25	32	9	40
Mechanical ventilation (transport ventilator)	101	14	39	7	41
Nasogastric / orogastric tube	89	19	33	2	35
Manual defibrillation	77	14	32	3	28
Physical restraint	71	9	34	2	26
Pelvic binder	73	15	27	4	27
Rapid / delayed sequence intubation	96	6	36	15	39
Central venous access / existing central line use	75	15	27	10	23
Blood product administration	58	9	27	9	13
External jugular IV	57	14	21	1	21
Needle cricothyrotomy / transtracheal jet insufflation	57	13	19	0	25
Obstetric delivery maneuvers	48	10	19	0	19
Chest tube / finger thoracostomy	42	9	18	0	15
Meconium / deep tracheal suctioning	41	12	9	2	18

These columns do not add up to a national picture, and should not be read as one. They count the agencies whose published book we have read. An agency absent from a row did not decline the procedure — it is simply not in the corpus, or its book does not name the procedure in words this extractor reads.

Where states disagree at the same certification level

This is the one comparison in the census where a no is publishable. 6 states publish a real competency matrix — every skill against every certification level, each cell ticked or not — and an unticked cell in an enumeration like that is the state saying no, not the document being quiet. Below are the 15 skills where two of those states, at the same level, give opposite answers.

Skill, as the state names it	Level	What each state’s matrix says
Acetaminophen	EMT	NC yes NV yes UT no VA no
Acetaminophen	AEMT	NC no NV yes UT yes
Albuterol	EMT	NV no UT yes VA yes
Amiodarone	AEMT	NC no NV no UT yes
Acetaminophen	EMR	NC yes UT no
Atropine	AEMT	NC no UT yes
Diazepam	AEMT	NV no UT yes
Glucose, oral	EMR	NC yes UT no
Hydroxocobalamin	AEMT	NV no UT yes
Intranasal	EMT	MA yes MS no
Lidocaine	AEMT	NV no UT yes
Midazolam	AEMT	NV no UT yes
Pulse Oximetry	EMR	MS no NC yes
Rectal	Paramedic	MA yes MS no
Subcutaneous	AEMT	MA yes MS no

Skills are matched on the state’s own wording, so this table undercounts: one state’s “Cardiac Pacing” and another’s “Transcutaneous Pacing” are the same skill and will not pair here. An instrument that can only miss disagreements is the right way round for this question. 2 further states (OR, SD) publish scope as prose rather than a matrix. Nothing they leave out appears here as a no, because a list of what a level may do is not a list of what it may not.

What a certification level is, and when we cannot tell you

2,533 of 5,419 readings name the level that may perform the procedure. The rest split into two groups that this page refuses to pool, because they are not the same fact:

Not stated — 2,886 readings: The page names the procedure and never names a level. We could not read the answer, so we do not print one. This is an absence of evidence in our reading, not a statement by the agency.
Set elsewhere by design — none in this build: The document deliberately declines to set the tier. Michigan prints one statewide protocol and each local Medical Control Authority ticks a box for which level may perform the skill, so the answer is not missing from the page — the page is telling you the answer is set somewhere else. No reading in the current extract carries that flag, so the page shows none. It is kept here, and counted separately the moment one appears, because the distinction is the point: a document that declines to set a tier has said something, and folding it into “not stated” would throw it away.

What this page will never tell you

We publish what a protocol book SAYS. A procedure missing from an agency's book is missing from that book — it is not evidence the agency does not do it. A protocol is not a scope of practice, and nothing on this page should be read as one.

Nowhere on this page does an agency appear as not doing something. The census reads what books say, and a protocol book is written to guide care, not to enumerate a scope of practice: it can omit a procedure the crew performs daily under a separate policy, a state rule or a medical director’s standing memo. The only negatives we publish are the state matrix cells above, and they are negatives because a matrix is an enumeration — it lists every skill and every level and ticks the boxes, so an unticked box is the document speaking.

How far this reaches

190 agencies across 44 states, of roughly 26,883 licensed EMS agencies we count nationally. That is a sample, and a sample chosen by which agencies publish their protocols in a readable form — which is not a random slice of American EMS. Read every count here as “of the books we have read”, never as “of the country”.

Every reading carries the document and the page it came from, and is rebuilt from the source documents on each run rather than edited in place.

Open the protocol census · How these books change over time · Census methodology