Skip to content

ACC data via eolas

The Accident Compensation Corporation (ACC) administers New Zealand's no-fault accident insurance scheme. ACC publishes detailed claim statistics — broken down by injury type, sport / activity, age cohort, region, and outcome. eolas serves 73 datasets from ACC — the largest single source after Stats NZ.

If you're doing injury-prevention research, healthcare-cost analysis, sport-safety studies, occupational health research, or actuarial work — ACC is the source.

What's in the catalogue

ACC publishes thematically rather than as one giant table. Each dataset typically covers a specific injury type, activity, or cohort. Themes include:

Sport + recreation injuries

Examples
acc_acl_injuries (anterior cruciate ligament)
acc_achilles_tendon_ruptures
acc_climbing_related_claims_lodged_between_1_january_2018_and_31_december_2022
acc_cycling_injury_data
acc_cricket_related_upper_limb_injuries
acc_skiing_and_snowboarding_injuries
Examples
acc_e_scooter_injuries
acc_motorcycle_injuries
acc_pedestrian_injury_data

Workplace + sector injuries

Examples
acc_education_sector_claims
acc_health_and_social_assistance_sector_claims
acc_construction_sector_claims
acc_chainsaw_related_injuries

Body-region injuries

Examples
acc_concussion_tbi_data_update (traumatic brain injury)
acc_eye_injuries
acc_carpal_tunnel_injuries
acc_cubital_tunnel_syndrome_claims

Cohort + demographic breakdowns

Examples
acc_claims_for_children_and_teenagers
acc_falls_dataset (elderly falls — high cost cohort)
acc_falls_dataset_update

Operations + service data

Examples
acc_elective_surgery_data (ACC-funded surgeries)
acc_diagnostic_imaging_acc_claims_by_modality

Browse the full list: eolas.fyi/datasets?source=ACC.

Refresh schedule

Monthly. ACC publishes most datasets quarterly or annually; our refresh runs weekly to catch new releases promptly.

meta = client.info("acc_acl_injuries")
meta["last_refreshed_at"]
meta["source_last_modified_at"]

License

All ACC data is published under CC-BY 4.0 via the ACC OIA + Open Data Programme. Commercial use is fine; attribution required.

Recommended attribution: "Source: Accident Compensation Corporation, served via eolas (eolas.fyi). CC-BY 4.0."

A privacy note: ACC publishes aggregate counts, not individual claims. Small-cell suppression applies to cells below threshold (typically ≤4 claims) — these appear as null with a suppression flag where present.

Common patterns

E-scooter injury trend

import matplotlib.pyplot as plt

es = client.acc("acc_e_scooter_injuries")
# Most ACC datasets have columns: year, region, injury_type, count, cost
by_year = es.groupby("year")["count"].sum()
by_year.plot(title="E-scooter injuries (ACC claims)")
plt.show()

library(dplyr)
library(ggplot2)

es <- eolas_get_acc("acc_e_scooter_injuries")
by_year <- es |> group_by(year) |> summarise(count = sum(count))
ggplot(by_year, aes(year, count)) + geom_line() +
  labs(title = "E-scooter injuries (ACC claims)")

Falls in the elderly — the big-cost cohort

falls = client.acc("acc_falls_dataset")
# Cost per claim by age band
by_age = falls.groupby("age_band").agg({"count": "sum", "total_cost": "sum"}).reset_index()
by_age["cost_per_claim"] = by_age["total_cost"] / by_age["count"]
print(by_age.sort_values("cost_per_claim", ascending=False))

Sector-specific injury rate

education = client.acc("acc_education_sector_claims")
# Filter to most-recent year
latest = education[education["year"] == education["year"].max()]
# By injury type
print(latest.groupby("injury_type")["count"].sum().sort_values(ascending=False).head(10))

Source-specific notes

  • Lots of small thematic datasets: ACC publishes per-topic rather than one giant table. To compare across themes (e.g. ACL injuries vs eye injuries) you'll need to load each dataset separately. Use the search filter on eolas.fyi/datasets?source=ACC to find what's there.
  • Schemas vary: ACC datasets are published as OIA responses — each tends to follow the columns the original requestor asked for. Use client.info(name) to inspect schema before joining.
  • Claims ≠ injuries: an ACC claim is one interaction with the scheme; a single injury can generate multiple claims (initial assessment, follow-up, surgery, rehab). Count the right unit for your analysis.
  • Costs are scheme costs, not full healthcare costs: ACC funds part of the care (specifically the accident component). Hospital admissions, primary care, and prescriptions outside the ACC scheme aren't reflected.
  • Provisional vs final: recent quarters are sometimes provisional and revised upward as late claims are accepted. The most-recent 2-3 quarters are usually still moving.
  • Coverage history: ACC has been operating since 1974. Most datasets start in 2000-2010; very few have data older than that.

Where to find more