Claim at v1.0.0
graph-triad-hot-asthma-absence
e8613ac3d8d2727fb91f15c8f5570bd0c3ad47c1d5933c570e5aad137040a963
Data
{
"has_outcome": true,
"kind": "bridge",
"lineage": {
"audit": "docs/lineage_audit.md; data/derived/lineage_audit.json",
"class": "a",
"class_meaning": "exposure or treatment only, not the outcome",
"classes": [
"a"
],
"early_sources": [
{
"class": "a",
"dataset": "GHCN-Daily v3 (NCEI), doi:10.7289/V5D21VHZ",
"evidence": "Daily temperature (GHCN-Daily) is the exposure, not the outcome. It is weather, fetched for earlier heat claims and reused.",
"fetched": "2026-11-12T20:23:55-05:00"
}
]
},
"outcome": "graph-triad-hot-asthma-absence",
"prereg_branch": {
"name": "prereg/graph-triad-hot-asthma-absence",
"sha": "0c97c85342a559c605b39ae7cdb48327d3089530"
},
"prior": 0.5,
"registered_at": "2026-11-12T22:51:37-0500",
"registrant": "claude (graph triads study)",
"registration": {
"alternatives": [
"Temperature affects attendance directly (discomfort, travel, classrooms without air conditioning, families' choices), not through asthma; a ratio far above the composed interval indicates this.",
"Calendar: late June is hot and attendance falls at the end of the year; winter cold coincides with influenza season and snow. Week-of-year and school-year effects and the snow indicator address this in part; influenza is not controlled.",
"Citywide aggregation hides schools that close or report late on extreme days; the exclusions address this in part.",
"The asthma links come from hospital admissions in Hong Kong, Hefei, Lanzhou and Florida and from 92 children in Minnesota; none is a New York school population."
],
"chain": [
"temperature -> asthma exacerbation (graph)",
"asthma -> days absent (graph)"
],
"composition": {
"composed_ratio_lo_median_hi": [
1.0006761898256078,
1.0025738370025843,
1.0054867760972164
],
"discrimination": {
"D_expected": 0.022492551500257994,
"advice": "the test cannot separate the chain from no effect; find a larger contrast or a more precise endpoint",
"combined_se": 0.018042326046133237,
"discriminates": false,
"endpoint_se": 0.018,
"p_null_passes": 0.9479771784103469,
"z_null_if_chain_true": 0.14997516961236615
},
"exponent": {
"mean": 0.0156,
"meaning": "share of pupil absences attributable to asthma = prevalence U(0.08, 0.15) x excess days 2.21 (1.41-3.01) / mean days absent U(13, 20); prevalence and mean days absent are the registrant's assumptions, not graph findings",
"sd": 0.0045
},
"link": {
"estimate": 1.19,
"hi": 1.34,
"lo": 1.06,
"pubmed_id": "27343213"
},
"method": "kit.bridge.compose, log RR(absence) = share x log RR(asthma exacerbation), 20000 draws, seed 1"
},
"disclosure": "No attendance data had been queried by this project before registration (no file for these datasets in data/cache/socrata; only the catalog's column names were read). The Central Park temperature series had been used by earlier claims, and the counts of exposed days (116 of 843 hot-season weekdays, 145 of 838 cold-season weekdays) were computed from it before registration. The registrant knows in general that New York attendance falls in late June and on snow days. A web search summary of Keivabu 2024 was read before registration.",
"graph_evidence": [
{
"concepts": "temperature family (hot, temperature, air temperature) - asthma",
"direction": "+ for heat in the hot season and + for cold in the cold season; the 32 positive and 8 negative rows mix both, and include heaters and exhaled temperature",
"n_findings": 40,
"n_studies": 24,
"pubmed_ids": [
"27343213",
"34555607",
"39251927",
"36240934"
],
"relationship_id": [
"4FaXcuNJRt|EAshH_Ko3h",
"EAshH_Ko3h|4FaXcuNJRt",
"dINFs04idR|EAshH_Ko3h"
],
"summary": "hot-season temperature 30 C against 27 C and asthma hospitalisations, cumulative RR lags 0-3 = 1.19 (1.06-1.34), Hong Kong 2004-2011, all ages (children least sensitive); also heat index quartile 4 and asthma exacerbation in April, May and October, OR 1.27 (1.19-1.36) to 2.04 (1.92-2.17) by ethnicity, Florida (PMID 34555607)"
},
{
"concepts": "asthma - school attendance",
"direction": "+ (more days absent)",
"n_findings": 1,
"n_studies": 31,
"pubmed_ids": [
"11487757"
],
"relationship_id": [
"EAshH_Ko3h|_l4tNqo_PI"
],
"summary": "children with asthma had 2.21 (1.41-3.01) more days absent per year than matched children without asthma, Rochester, Minnesota (abstract read; the relationship record lists 31 studies but the API returns one finding)"
},
{
"concepts": "temperature family - school attendance",
"direction": null,
"n_findings": 0,
"n_studies": 0,
"pubmed_ids": [],
"relationship_id": null,
"summary": "no direct finding in the graph for hot, temperature, air temperature, surface temperature, heat wave or urban heat island with school attendance (two findings from two studies with absenteeism)"
}
],
"id": "graph-triad-hot-asthma-absence",
"kind": "bridge",
"literature": [
{
"citation": "Keivabu RC. Temperature and school absences: evidence from England. Population and Environment 2024",
"doi": "10.1007/s11111-024-00448-5",
"finding": "hot days increase absences in English schools; cold days show no substantive effect. Not in the graph. Read as a search summary only, not the paper."
}
],
"plan": "Unit: school day, citywide, school years 2006-07 to 2018-19. Outcome: log(sum of absent / sum of enrolled) over all schools by date, from NYC Open Data daily attendance by school (xwxx-rnki, wpqj-3buw, pffu-gbfi, 3y5p-x48f, x3bb-kg5j), aggregated in SoQL. Exclusions fixed now: weekends; days whose citywide enrolment is below 80% of that school year's median; days with an absence rate above 30%. Arm: school days in May, June and September. Exposure: Central Park TMAX >= 30.0 C on the same day (NOAA GHCN-Daily USW00094728). Model: OLS of the log absence rate on the exposure indicator with fixed effects for school year, day of week and ISO week of the year, and indicators for precipitation >= 10 mm, for a school day next to a weekday without school, and for the last five school days of the school year. Standard errors Newey-West with 5 lags. Estimate: exp(coefficient) with its 95% interval. Secondary, not deciding: exposure on any of lags 0 to 2; a linear term per 5 C of TMAX within the arm; the hot arm without the last two weeks of June; the same model on the school-level medians is not planned.",
"prediction": {
"direction": "+",
"magnitude": "Composed through asthma alone: absence ratio 1.0026 (95% interval 1.0007 to 1.0055). With an expected standard error near 0.018 on the log scale the test cannot separate this size from no effect (D = 0.02, below 4). The registrant expects the direct association to be larger than the chain, between 1.00 and 1.08, because temperature can affect attendance by routes other than asthma.",
"refute_if": "the absence ratio is below 1 and its 95% interval excludes 1.",
"support_if": "the absence ratio for exposed days is above 1 and its 95% interval (Newey-West) excludes 1. The direction composed from the triad is then confirmed. Whether the interval contains the composed interval is reported as a second, registered quantity (the share of the association that the asthma path can account for = composed log ratio / city log ratio)."
},
"prior": 0.5,
"registered_at": "2026-11-12T22:51:37-0500",
"registrant": "claude (graph triads study)",
"sha256": "d47c0c36d36d614de5c5bdb0d943f9a98a09d59a3f467093d410e6276ceb3c0e",
"statement": "On hot school days New York City public-school absence is higher than on other school days of the same week of the year, as the open triad temperature - asthma - school attendance in the System Public Graph implies; the graph holds no finding that links temperature to school attendance directly."
},
"registration_commit": {
"commit": "86c42fc401e6682a570afeb801f9df9f74a10d41",
"committed_at": "2026-11-12T22:51:43-05:00"
},
"registration_file_sha256": "88b3ae7fd467af9a76250d94c5386f38311a6a905a2849a7b910b75abbdfe9cc",
"registration_path": "claims/registered/graph-triad-hot-asthma-absence.json",
"registration_sha256": "d47c0c36d36d614de5c5bdb0d943f9a98a09d59a3f467093d410e6276ceb3c0e",
"registration_text": "{\n \"id\": \"graph-triad-hot-asthma-absence\",\n \"kind\": \"bridge\",\n \"statement\": \"On hot school days New York City public-school absence is higher than on other school days of the same week of the year, as the open triad temperature - asthma - school attendance in the System Public Graph implies; the graph holds no finding that links temperature to school attendance directly.\",\n \"graph_evidence\": [\n {\n \"relationship_id\": [\n \"4FaXcuNJRt|EAshH_Ko3h\",\n \"EAshH_Ko3h|4FaXcuNJRt\",\n \"dINFs04idR|EAshH_Ko3h\"\n ],\n \"concepts\": \"temperature family (hot, temperature, air temperature) - asthma\",\n \"n_findings\": 40,\n \"n_studies\": 24,\n \"direction\": \"+ for heat in the hot season and + for cold in the cold season; the 32 positive and 8 negative rows mix both, and include heaters and exhaled temperature\",\n \"summary\": \"hot-season temperature 30 C against 27 C and asthma hospitalisations, cumulative RR lags 0-3 = 1.19 (1.06-1.34), Hong Kong 2004-2011, all ages (children least sensitive); also heat index quartile 4 and asthma exacerbation in April, May and October, OR 1.27 (1.19-1.36) to 2.04 (1.92-2.17) by ethnicity, Florida (PMID 34555607)\",\n \"pubmed_ids\": [\n \"27343213\",\n \"34555607\",\n \"39251927\",\n \"36240934\"\n ]\n },\n {\n \"relationship_id\": [\n \"EAshH_Ko3h|_l4tNqo_PI\"\n ],\n \"concepts\": \"asthma - school attendance\",\n \"n_findings\": 1,\n \"n_studies\": 31,\n \"direction\": \"+ (more days absent)\",\n \"summary\": \"children with asthma had 2.21 (1.41-3.01) more days absent per year than matched children without asthma, Rochester, Minnesota (abstract read; the relationship record lists 31 studies but the API returns one finding)\",\n \"pubmed_ids\": [\n \"11487757\"\n ]\n },\n {\n \"relationship_id\": null,\n \"concepts\": \"temperature family - school attendance\",\n \"n_findings\": 0,\n \"n_studies\": 0,\n \"direction\": null,\n \"summary\": \"no direct finding in the graph for hot, temperature, air temperature, surface temperature, heat wave or urban heat island with school attendance (two findings from two studies with absenteeism)\",\n \"pubmed_ids\": []\n }\n ],\n \"literature\": [\n {\n \"citation\": \"Keivabu RC. Temperature and school absences: evidence from England. Population and Environment 2024\",\n \"doi\": \"10.1007/s11111-024-00448-5\",\n \"finding\": \"hot days increase absences in English schools; cold days show no substantive effect. Not in the graph. Read as a search summary only, not the paper.\"\n }\n ],\n \"composition\": {\n \"method\": \"kit.bridge.compose, log RR(absence) = share x log RR(asthma exacerbation), 20000 draws, seed 1\",\n \"link\": {\n \"estimate\": 1.19,\n \"lo\": 1.06,\n \"hi\": 1.34,\n \"pubmed_id\": \"27343213\"\n },\n \"exponent\": {\n \"mean\": 0.0156,\n \"sd\": 0.0045,\n \"meaning\": \"share of pupil absences attributable to asthma = prevalence U(0.08, 0.15) x excess days 2.21 (1.41-3.01) / mean days absent U(13, 20); prevalence and mean days absent are the registrant's assumptions, not graph findings\"\n },\n \"composed_ratio_lo_median_hi\": [\n 1.0006761898256078,\n 1.0025738370025843,\n 1.0054867760972164\n ],\n \"discrimination\": {\n \"endpoint_se\": 0.018,\n \"combined_se\": 0.018042326046133237,\n \"z_null_if_chain_true\": 0.14997516961236615,\n \"D_expected\": 0.022492551500257994,\n \"discriminates\": false,\n \"p_null_passes\": 0.9479771784103469,\n \"advice\": \"the test cannot separate the chain from no effect; find a larger contrast or a more precise endpoint\"\n }\n },\n \"prediction\": {\n \"direction\": \"+\",\n \"magnitude\": \"Composed through asthma alone: absence ratio 1.0026 (95% interval 1.0007 to 1.0055). With an expected standard error near 0.018 on the log scale the test cannot separate this size from no effect (D = 0.02, below 4). The registrant expects the direct association to be larger than the chain, between 1.00 and 1.08, because temperature can affect attendance by routes other than asthma.\",\n \"support_if\": \"the absence ratio for exposed days is above 1 and its 95% interval (Newey-West) excludes 1. The direction composed from the triad is then confirmed. Whether the interval contains the composed interval is reported as a second, registered quantity (the share of the association that the asthma path can account for = composed log ratio / city log ratio).\",\n \"refute_if\": \"the absence ratio is below 1 and its 95% interval excludes 1.\"\n },\n \"plan\": \"Unit: school day, citywide, school years 2006-07 to 2018-19. Outcome: log(sum of absent / sum of enrolled) over all schools by date, from NYC Open Data daily attendance by school (xwxx-rnki, wpqj-3buw, pffu-gbfi, 3y5p-x48f, x3bb-kg5j), aggregated in SoQL. Exclusions fixed now: weekends; days whose citywide enrolment is below 80% of that school year's median; days with an absence rate above 30%. Arm: school days in May, June and September. Exposure: Central Park TMAX >= 30.0 C on the same day (NOAA GHCN-Daily USW00094728). Model: OLS of the log absence rate on the exposure indicator with fixed effects for school year, day of week and ISO week of the year, and indicators for precipitation >= 10 mm, for a school day next to a weekday without school, and for the last five school days of the school year. Standard errors Newey-West with 5 lags. Estimate: exp(coefficient) with its 95% interval. Secondary, not deciding: exposure on any of lags 0 to 2; a linear term per 5 C of TMAX within the arm; the hot arm without the last two weeks of June; the same model on the school-level medians is not planned.\",\n \"alternatives\": [\n \"Temperature affects attendance directly (discomfort, travel, classrooms without air conditioning, families' choices), not through asthma; a ratio far above the composed interval indicates this.\",\n \"Calendar: late June is hot and attendance falls at the end of the year; winter cold coincides with influenza season and snow. Week-of-year and school-year effects and the snow indicator address this in part; influenza is not controlled.\",\n \"Citywide aggregation hides schools that close or report late on extreme days; the exclusions address this in part.\",\n \"The asthma links come from hospital admissions in Hong Kong, Hefei, Lanzhou and Florida and from 92 children in Minnesota; none is a New York school population.\"\n ],\n \"disclosure\": \"No attendance data had been queried by this project before registration (no file for these datasets in data/cache/socrata; only the catalog's column names were read). The Central Park temperature series had been used by earlier claims, and the counts of exposed days (116 of 843 hot-season weekdays, 145 of 838 cold-season weekdays) were computed from it before registration. The registrant knows in general that New York attendance falls in late June and on snow days. A web search summary of Keivabu 2024 was read before registration.\",\n \"prior\": 0.5,\n \"registrant\": \"claude (graph triads study)\",\n \"chain\": [\n \"temperature -> asthma exacerbation (graph)\",\n \"asthma -> days absent (graph)\"\n ],\n \"registered_at\": \"2026-11-12T22:51:37-0500\",\n \"sha256\": \"d47c0c36d36d614de5c5bdb0d943f9a98a09d59a3f467093d410e6276ceb3c0e\"\n}",
"statement": "On hot school days New York City public-school absence is higher than on other school days of the same week of the year, as the open triad temperature - asthma - school attendance in the System Public Graph implies; the graph holds no finding that links temperature to school attendance directly."
}History in this collection
- addedv1.0.0Nov 14, 2026this version