aioh-compliance

AIOH Compliance Decision Framework

This skill provides the professional knowledge base for occupational hygiene exposure compliance assessment based on the AIOH Compliance Decision Guide. It covers data treatment, statistical methods, compliance categories, method selection, and interpretation guidance.

For detailed statistical formulae and calculation procedures, read references/statistical-methods.md. For the compliance category decision framework and flowcharts, read references/decision-framework.md. For method disagreement guidance and edge cases, read references/method-guidance.md.

Core Principles

Data Treatment Requirements

Before performing compliance analysis, exposure data must satisfy several prerequisites:

1. Similar Exposure Group (SEG): All measurements must represent a defined SEG — workers with similar exposure profiles due to comparable tasks, materials, environments, and controls. SEG homogeneity is fundamental; mixed groups produce unreliable statistics.

2. Lognormal Distribution: Occupational exposure data is assumed to follow a lognormal distribution. This assumption must be tested using the Shapiro-Wilk W-test on log-transformed data. If W < critical value at α = 0.05, the lognormal fit is questionable and results should be interpreted with caution.

3. Minimum Sample Size: While calculations can be performed with as few as 3 measurements, results are more reliable with larger samples. The U-parameter test requires a minimum of 6 measurements. The AIOH Guide notes that small samples (n < 10) carry substantial uncertainty.

4. Non-Detects and Censored Data: Values below the limit of detection (LOD) should be substituted at LOD/2 for preliminary analysis, or handled using maximum likelihood estimation for more rigorous treatment.

5. Independence: Measurements should be independent — not repeated measures on the same worker on the same day for the same task, unless the sampling strategy specifically accounts for within-worker variability.

Key Statistical Parameters

Parameter	Description	Use
GM (Geometric Mean)	Central tendency of lognormal distribution	Indicates typical exposure
GSD (Geometric Standard Deviation)	Spread of lognormal distribution	Indicates exposure variability
MVUE (Minimum Variance Unbiased Estimator)	Best unbiased estimate of arithmetic mean from log-transformed data	Point estimate for compliance
UCL₉₅% (Upper Confidence Limit, 95%)	Upper bound of 95% CI on MVUE via Land's exact method	Primary compliance metric
P95 (95th Percentile)	Value below which 95% of exposures fall	Indicates upper tail of distribution
Exceedance Fraction	Estimated percentage of exposures exceeding the OEL	Risk characterisation
U-Parameter	(ln ES − ln GM) / ln GSD	Group compliance test statistic

Compliance Categories (AIOH Framework)

The AIOH framework classifies exposure profiles into five categories based on the primary metric (UCL₉₅% of MVUE is recommended):

• Category 0 (Ideal): Metric < 0.1 × OEL — Exposure well controlled, minimal monitoring needed
• Category 1 (Good): Metric < 0.5 × OEL AND P95 < OEL AND Exceedance < 5% — Routine monitoring appropriate
• Category 2 (Acceptable): Metric < OEL AND does not meet Category 1 — Increased monitoring frequency recommended
• Category 3 (Unacceptable): Metric ≥ OEL OR Exceedance ≥ 5% OR P95 > OEL — Controls required
• Category 4 (Critical): Metric > OEL AND observed exceedances > 10% — Immediate action required

Two Compliance Testing Methods

Two statistical methods are available for group compliance assessment. Each answers a different question at a different confidence level. Read references/method-guidance.md for detailed guidance on when to use each method and how to handle disagreements.

UCL₉₅% of MVUE (Land's Exact Method) — the primary Australian method:

• Calculates the 95% upper confidence limit of the arithmetic mean using Land's exact method
• If UCL₉₅% < OEL, the SEG is considered compliant at 95% confidence
• Recommended by AIOH Section 13.6 as the primary compliance metric
• Used by many large Australian organisations

U-Parameter Group Compliance Test (INRS / NF EN 689) — supplementary European method:

• Calculates U = (ln ES − ln GM) / ln GSD
• Compares against Table 13.1 limiting values at 70% confidence
• Tests whether fewer than 5% of exposures exceed the ES
• Requires minimum 6 measurements
• The 70% confidence level provides a balance against false positives and false negatives

Method Selection Guidance (Summary)

Criterion	UCL₉₅% (Land's)	U-Parameter (INRS)
Confidence level	95%	70%
Conservatism	More conservative	Less conservative
Australian practice	Primary method	Supplementary
Minimum n	3 (recommended ≥ 6)	6 (strict requirement)
What it tests	Is the true mean below ES?	Are < 5% of exposures above ES?
Regulatory basis	AIOH Guide Section 13.6	NF EN 689:2018

When Methods Disagree

When the UCL₉₅% and U-parameter test give different compliance verdicts, this is a borderline situation. The guidance is:

1. UCL₉₅% takes precedence in Australian practice — it is the more conservative primary metric
2. Collect more data — the disagreement typically indicates the exposure profile is near the compliance threshold
3. Examine SEG homogeneity — consider whether the SEG should be subdivided
4. Report both results — the U-test provides useful context about how close the profile is to compliance

For full details on disagreement scenarios, read references/method-guidance.md.

Australian English and Reporting Conventions

• Use Australian English spelling throughout (analyse, minimise, behaviour, labour)
• Use conservative, neutral language — describe findings factually without emotive terms
• Reference the Workplace Exposure Standard (WES) or Occupational Exposure Limit (OEL) as applicable
• Cite the AIOH Compliance Decision Guide and relevant sections when providing methodology rationale
• All legislative references should cite the applicable jurisdiction (e.g., WHS Regulation 2011 (Qld))

References

• AIOH (Australian Institute of Occupational Hygienists). Compliance Decision Guide. (Primary reference)
• NF EN 689:2018. Workplace exposure — Measurement of exposure by inhalation to chemical agents — Strategy for testing compliance with occupational exposure limit values
• Land, C.E. (1971). Confidence intervals for linear functions of the normal mean and variance. Annals of Mathematical Statistics, 42(4), 1187–1205
• Leidel, N.A., Busch, K.A. & Lynch, J.R. (1977). Occupational Exposure Sampling Strategy Manual. NIOSH Publication No. 77-173
• Shapiro, S.S. & Wilk, M.B. (1965). An analysis of variance test for normality. Biometrika, 52(3–4), 591–611