Claim at v1.0.0
ems-hospital-turnaround-wait
349cdc3916b0d0cfe5e672b4995b1c7c46bf7d7e702579a4411622505019cfa9
Data
{
"has_outcome": true,
"kind": "confirmation",
"lineage": {
"audit": "docs/lineage_audit.md; data/derived/lineage_audit.json",
"class": "none",
"class_meaning": "the audit found no city-data source fetched before registration",
"classes": [],
"early_sources": []
},
"outcome": "ems-hospital-turnaround-wait",
"prereg_branch": {
"name": "prereg/ems-hospital-turnaround-wait",
"sha": "be7f692245a7be72156b1d7907bd1bc064346f0f"
},
"prior": 0.6,
"registered_at": "2026-11-13T11:54:58-0500",
"registrant": "health-gaps agent (Claude), 13 Nov 2026",
"registration": {
"alternatives": [
"Common daily shock: heat, flu or other surges crowd emergency departments and raise ambulance demand together, beyond what the call count captures; the previous-day model and the severity split partly separate this.",
"Reverse direction: when the system is busy crews may shorten time at hospital, which would bias the elasticity down.",
"Staffing: fewer ambulance tours lengthen waits; year-month effects absorb slow change but not day-to-day staffing gaps.",
"Data artefact: incident close may be recorded late on busy days, inflating measured turnaround when waits are long."
],
"graph_evidence": [
{
"note": "The graph has findings on what slower response costs (docs/congestion_tradeoffs.md: five odds ratios for cardiac-arrest survival per minute of shorter response, 1.02, 1.08, 1.12, 1.22, 1.24, read on PubMed). A variable search for 'offload' and 'ambulance' (13 Nov 2026) returned no finding on ambulance offload or hospital turnaround and ambulance availability; the cause side of the chain is absent from the graph."
}
],
"id": "ems-hospital-turnaround-wait",
"kind": "confirmation",
"literature": [
{
"citation": "Lane DJ et al. (2023). The association between paramedic service system hospital offload time and response time. CJEM.",
"doi": "PMID 37208561",
"finding": "Hourly time series (Ontario): longer offload times go with longer response times, adjusting for call volume; larger effect at high volume in winter."
},
{
"citation": "Patterns in California Ambulance Patient Offload Times by Local EMS Agency (2024). JAMA Netw Open (PMC11650417).",
"doi": "",
"finding": "About half of California agencies consistently exceed the 30-minute offload standard."
},
{
"citation": "NYC Mayor's Management Report, FDNY chapter, FY2026 (read 13 Nov 2026, before this registration).",
"doi": "",
"finding": "Average ambulance turnaround at hospitals 36:17 (FY22), 38:10, 40:50, 40:40, 40:25 (FY26); end-to-end ambulance response to life-threatening emergencies 10:17 (FY22) to 11:21 (FY25) and 13:09 (FY26, with a definition change for upgraded calls); ambulance in-service hours per day 8,891 to 8,109. The annual figures were seen before registration; the daily incident data were not."
}
],
"plan": "Script analyses/health_gaps/ems_model.py and fetch analyses/health_gaps/ems_fetch.py, committed in 6278c85 before this registration and before the daily data are fetched. Data: NYC Open Data 76xm-jjuj, daily citywide aggregates 2017-01-01 to the last complete month, excluding 2020-03-01 to 2020-06-30. Outcome: log daily mean dispatch_response_seconds_qy (incident creation to first unit assigned) among incidents with valid dispatch time. Exposure: log daily mean of incident_close_datetime minus first_hosp_arrival_datetime among transported incidents (disposition 82), trimmed to 0-4 h and to close and arrival in the same calendar month. Controls: log daily incident count, day-of-week dummies, year-month fixed effects (which absorb staffing, fleet and definition changes that move slowly). OLS, Newey-West standard errors with 7 lags. No resampling and no seed. Secondary, not deciding: severity 1-3 and 4-8 separately; previous-day turnaround; annual means; attribution of the 2019-2025 change. A technical test of the SoQL date arithmetic was run on one hour of 3 January 2017 (counts of 98 rows and three histogram bins) before registration.",
"prediction": {
"direction": "+",
"magnitude": "Elasticity of daily mean dispatch wait to daily mean hospital turnaround between 0.5 and 3 (hospital time is roughly 40% of an ambulance's time per job, and wait rises steeply with utilisation); attenuated by measurement noise in daily means, so 0.2 to 0.5 is also plausible.",
"refute_if": "Upper 95% bound of the turnaround elasticity in the primary model below 0.2 (a negligible or negative association).",
"support_if": "Lower 95% bound of the turnaround elasticity in the primary model above 0.2."
},
"prior": 0.6,
"registered_at": "2026-11-13T11:54:58-0500",
"registrant": "health-gaps agent (Claude), 13 Nov 2026",
"sha256": "8441e4128a55fa566516f03ec6b2992030e717cba3c8197f4682524cfc46e173",
"statement": "In New York City, on days when ambulances spend longer at hospitals after transporting a patient, callers wait longer for an ambulance to be assigned, holding the day's call volume, the day of week and the month fixed."
},
"registration_commit": {
"commit": "be7f692245a7be72156b1d7907bd1bc064346f0f",
"committed_at": "2026-11-13T11:54:58-05:00"
},
"registration_file_sha256": "d119d143a0c2c2b32c718493432d6335bcb54cfa21bc1c9dfabf98fdea66b310",
"registration_path": "claims/registered/ems-hospital-turnaround-wait.json",
"registration_sha256": "8441e4128a55fa566516f03ec6b2992030e717cba3c8197f4682524cfc46e173",
"registration_text": "{\n \"id\": \"ems-hospital-turnaround-wait\",\n \"kind\": \"confirmation\",\n \"statement\": \"In New York City, on days when ambulances spend longer at hospitals after transporting a patient, callers wait longer for an ambulance to be assigned, holding the day's call volume, the day of week and the month fixed.\",\n \"graph_evidence\": [\n {\n \"note\": \"The graph has findings on what slower response costs (docs/congestion_tradeoffs.md: five odds ratios for cardiac-arrest survival per minute of shorter response, 1.02, 1.08, 1.12, 1.22, 1.24, read on PubMed). A variable search for 'offload' and 'ambulance' (13 Nov 2026) returned no finding on ambulance offload or hospital turnaround and ambulance availability; the cause side of the chain is absent from the graph.\"\n }\n ],\n \"literature\": [\n {\n \"citation\": \"Lane DJ et al. (2023). The association between paramedic service system hospital offload time and response time. CJEM.\",\n \"doi\": \"PMID 37208561\",\n \"finding\": \"Hourly time series (Ontario): longer offload times go with longer response times, adjusting for call volume; larger effect at high volume in winter.\"\n },\n {\n \"citation\": \"Patterns in California Ambulance Patient Offload Times by Local EMS Agency (2024). JAMA Netw Open (PMC11650417).\",\n \"doi\": \"\",\n \"finding\": \"About half of California agencies consistently exceed the 30-minute offload standard.\"\n },\n {\n \"citation\": \"NYC Mayor's Management Report, FDNY chapter, FY2026 (read 13 Nov 2026, before this registration).\",\n \"doi\": \"\",\n \"finding\": \"Average ambulance turnaround at hospitals 36:17 (FY22), 38:10, 40:50, 40:40, 40:25 (FY26); end-to-end ambulance response to life-threatening emergencies 10:17 (FY22) to 11:21 (FY25) and 13:09 (FY26, with a definition change for upgraded calls); ambulance in-service hours per day 8,891 to 8,109. The annual figures were seen before registration; the daily incident data were not.\"\n }\n ],\n \"prediction\": {\n \"direction\": \"+\",\n \"magnitude\": \"Elasticity of daily mean dispatch wait to daily mean hospital turnaround between 0.5 and 3 (hospital time is roughly 40% of an ambulance's time per job, and wait rises steeply with utilisation); attenuated by measurement noise in daily means, so 0.2 to 0.5 is also plausible.\",\n \"support_if\": \"Lower 95% bound of the turnaround elasticity in the primary model above 0.2.\",\n \"refute_if\": \"Upper 95% bound of the turnaround elasticity in the primary model below 0.2 (a negligible or negative association).\"\n },\n \"plan\": \"Script analyses/health_gaps/ems_model.py and fetch analyses/health_gaps/ems_fetch.py, committed in 6278c85 before this registration and before the daily data are fetched. Data: NYC Open Data 76xm-jjuj, daily citywide aggregates 2017-01-01 to the last complete month, excluding 2020-03-01 to 2020-06-30. Outcome: log daily mean dispatch_response_seconds_qy (incident creation to first unit assigned) among incidents with valid dispatch time. Exposure: log daily mean of incident_close_datetime minus first_hosp_arrival_datetime among transported incidents (disposition 82), trimmed to 0-4 h and to close and arrival in the same calendar month. Controls: log daily incident count, day-of-week dummies, year-month fixed effects (which absorb staffing, fleet and definition changes that move slowly). OLS, Newey-West standard errors with 7 lags. No resampling and no seed. Secondary, not deciding: severity 1-3 and 4-8 separately; previous-day turnaround; annual means; attribution of the 2019-2025 change. A technical test of the SoQL date arithmetic was run on one hour of 3 January 2017 (counts of 98 rows and three histogram bins) before registration.\",\n \"alternatives\": [\n \"Common daily shock: heat, flu or other surges crowd emergency departments and raise ambulance demand together, beyond what the call count captures; the previous-day model and the severity split partly separate this.\",\n \"Reverse direction: when the system is busy crews may shorten time at hospital, which would bias the elasticity down.\",\n \"Staffing: fewer ambulance tours lengthen waits; year-month effects absorb slow change but not day-to-day staffing gaps.\",\n \"Data artefact: incident close may be recorded late on busy days, inflating measured turnaround when waits are long.\"\n ],\n \"prior\": 0.6,\n \"registrant\": \"health-gaps agent (Claude), 13 Nov 2026\",\n \"registered_at\": \"2026-11-13T11:54:58-0500\",\n \"sha256\": \"8441e4128a55fa566516f03ec6b2992030e717cba3c8197f4682524cfc46e173\"\n}",
"statement": "In New York City, on days when ambulances spend longer at hospitals after transporting a patient, callers wait longer for an ambulance to be assigned, holding the day's call volume, the day of week and the month fixed."
}History in this collection
- addedv1.0.0Nov 14, 2026this version