QUEBEC EMERGENCY ROOM HOURLY ARCHIVE
Compiled by Han Kim (Urgences Quebec / Boussole Sante)

WHAT THIS IS
Every hourly reading I have archived from the Quebec health ministry's emergency
room feed, from 2026-08-10 14:00 to 2026-09-14 10:00 local time.

  100,069 hourly readings
  120 installations
  837 consecutive hours (35 days), no gaps

This supersedes the Aug 10 - Sep 13 file. Same archive, 21 hours longer. Nothing
in the earlier rows changed.

The ministry publishes the current hour and overwrites it. Nothing on their site
shows yesterday, last week, or last month. This file is that history.

SOURCE AND LICENCE
Hourly source file, published by the Ministere de la Sante et des Services sociaux
under CC-BY 4.0 via Donnees Quebec:
  https://www.donneesquebec.ca/recherche/dataset/fichier-horaire-des-donnees-de-la-situation-a-l-urgence
Direct CSV:
  https://msss.gouv.qc.ca/professionnels/statistiques/documents/urgences/Releve_horaire_urgences_7jours_nbpers.csv
Every number here comes from that public file. Nothing is modelled, estimated or
imputed. My only contribution is that I kept it every hour instead of letting it
be overwritten.

FILES
1. quebec-er-hourly-archive.csv - the full raw archive, one row per hospital per
   hour, 100,069 rows.
2. quebec-er-daily-by-hospital.csv - the same data rolled up to one row per
   hospital per day, 4,320 rows. Opens in Excel, good for a first pass.

COLUMNS (hourly file)
  hospital                      installation name as the ministry writes it
  establishment                 parent establishment (CIUSSS / CISSS / CHU)
  region_code                   ministry region number
  region_name                   region name
  facility_id                   ministry installation number
  timestamp_local               local hour of the reading, e.g. 2026-09-14T10:00
  functional_stretchers         functional stretcher capacity
  stretchers_occupied           stretchers occupied
  stretcher_occupancy_pct       occupancy percentage as the ministry publishes it
  patients_over_24h             people on a stretcher more than 24 hours
  patients_over_48h             people on a stretcher more than 48 hours
  patients_present              total people physically in the ER
  patients_waiting_for_doctor   people waiting to be seen
  avg_wait_ambulatory_min       average length of stay, ambulatory patients, minutes
  avg_stay_stretcher_min        average length of stay on a stretcher, minutes

COLUMNS (daily file)
  hospital, establishment, region, date
  hours_reported                hours that day with a row
  avg_stretcher_occupancy_pct   mean occupancy that day
  max_stretcher_occupancy_pct   peak occupancy that day
  avg_patients_present          mean people in the ER
  max_patients_present          peak people in the ER
  avg_patients_waiting          mean people waiting for a doctor
  max_patients_waiting          peak people waiting for a doctor
  avg_wait_ambulatory_min       mean ambulatory wait, minutes
  avg_stay_stretcher_min        mean stretcher stay, minutes
  avg_functional_stretchers     mean functional capacity
  avg_patients_over_24h         mean people over 24h on a stretcher
  avg_patients_over_48h         mean people over 48h on a stretcher

FIVE THINGS TO KNOW BEFORE YOU QUOTE IT

1. Six real hospitals publish nothing at all. Sacre-Coeur-de-Montreal, Jean-Talon,
   Fleury, Sainte-Croix, Hotel-Dieu d'Arthabaska and Centre-de-la-Mauricie appear in
   the ministry feed with every occupancy field null, for all 837 hours. They are in
   this file as blank rows on purpose. The absence is itself a fair question for the
   ministry. Seven other blank sites are CHSLDs and multiservice centres with no ER,
   including Albert-Prevost.

2. "Longest wait" has two meanings and they rank differently. Sort by
   patients_waiting_for_doctor and you get the biggest waiting rooms. Sort by
   avg_wait_ambulatory_min or avg_stay_stretcher_min and you get the longest time a
   patient actually spends there. Those are different hospitals. Say which one you
   mean.

3. Small sites make percentages lie. The single highest occupancy in the archive is
   300 percent at Centre de Sante Rene-Ricard in Nord-du-Quebec. That is four people
   in a tiny community clinic. Occupancy percentage is unstable anywhere
   functional_stretchers is in single digits. Filter on it before ranking.

4. A handful of rows carry a malformed timestamp from the ministry feed. They are
   excluded here; every timestamp in this file matches 2026-MM-DDTHH:00.

5. stretcher_occupancy_pct is the ministry's own published value and it does not
   always equal stretchers_occupied divided by functional_stretchers. Where they
   disagree, both numbers are reproduced exactly as published. I have not corrected
   either one.

Questions on any column, ask me.
