Numbers do not interpret themselves. A single measurement describes one worker on one day under one set of conditions, and the distance between that fact and a statement about a whole workforce is where most exposure assessments succeed or fail. This page covers similar exposure groups, how many samples support a conclusion, the two kinds of variability that matter, and what a result close to a limit actually means.
Exposure assessment begins after valid measurement results are available. Its purpose is to determine what those results say about a defined workforce, operating condition and period of work.
A result should first be checked for analytical and contextual validity. The assessment then asks whether it represents one worker, one task, one day, a group of workers or a wider range of operating conditions. This distinction prevents a precise number from being given more evidential weight than the monitoring design supports.
The central questions are: which workers and conditions does the dataset represent; how variable are the results; is the dataset large and balanced enough to support a conclusion; how does uncertainty affect comparison with the selected criterion; is further evidence necessary; and does the evidence support a conclusion that exposure is adequately characterised.
This page does not address how samples are collected. Personal collection procedures and laboratory analysis are covered separately.
A similar exposure group, or SEG, is a group of workers expected to have comparable exposure patterns because they perform similar tasks, use similar materials, operate comparable processes and work under similar conditions.
Job title alone is rarely a sufficient basis for grouping. Two workers with the same title may spend different proportions of the shift near a source, use different equipment or cover different areas. Conversely, workers with different titles may have similar exposures if their tasks and conditions are substantially alike.
A defensible SEG description should consider tasks and their duration; materials and agents; process and equipment; operating rate; location and proximity; shift pattern; environmental conditions; frequency of non-routine work; and any relevant differences between workers.
Results should be reviewed for evidence that the proposed SEG is too broad. A consistent subgroup of higher values may indicate a distinct exposure pattern rather than random variation. Splitting an SEG may improve validity, although very narrow groups can leave too little data for meaningful evaluation.
Recognised strategies described by BOHS, AIHA and EN 689 use grouping to make assessment practicable. Foreign guidance and standards are evidence of recognised practice and are not automatically binding in the UAE.
There is no universal sample number that guarantees a reliable conclusion. Required numbers depend on the assessment objective, expected variability, proximity to the decision criterion, SEG size, data quality and desired confidence.
One measurement describes one measured circumstance. It cannot establish the exposure distribution for an entire SEG. Two or three results may provide an initial indication, especially if all are very low and conditions are stable, but they offer limited ability to identify variability or higher exposures.
EN 689:2018 provides one recognised framework for comparing inhalation exposure with occupational exposure limits. Its preliminary test uses small datasets under specified decision rules, while its statistical test requires at least six valid measurements. That structure is recognised practice rather than a UAE legal requirement. A minimum threshold should not be mistaken for an ideal dataset.
Where results are variable or close to the assessment criterion, more measurements across different workers and days will usually be needed. A larger dataset improves the estimate of the exposure distribution but cannot correct systematic bias, poor SEG definition or unrepresentative monitoring.
Confidence depends as much on coverage as count. Ten measurements taken from one worker on similar quiet days may provide less information about a group than a smaller but deliberately balanced set spanning different workers, shifts and operating conditions.
Occupational exposure measurements often follow a right-skewed distribution: many results occur at lower levels, with fewer high values. Lognormal statistical models are therefore commonly used, although the suitability of the assumed distribution should be checked.
Between-worker variability reflects consistent differences among people. These may arise from task allocation, technique, position, pace or time spent near an emission source.
Between-day variability reflects changes in production, material, maintenance state, weather, work sequence and other daily conditions. A worker may therefore have low exposure on one day and substantially higher exposure on another without either measurement being erroneous.
A robust dataset should spread measurements across workers and days. Repeated measurements on selected individuals can help separate the two sources of variability. Where data are sufficient, useful summaries may include the number of valid results; the arithmetic mean; the geometric mean; the geometric standard deviation; the range; a confidence interval; and an estimated upper percentile of the exposure distribution.
The arithmetic mean estimates average exposure burden, while the geometric mean describes the centre of a lognormal distribution. An upper percentile is often more useful when the objective is to assess whether most of the SEG is likely to remain below a criterion.
Results below the LOQ require a documented statistical treatment. Replacing every non-quantified result with zero creates downward bias, while replacing all such results with the LOQ may create upward bias. The chosen method should reflect the proportion of censored data and the decision being made.
An occupational exposure limit is not a precise boundary between harmless and harmful conditions. It is an assessment criterion with a defined substance, metric and averaging period. A measured result close to that criterion requires cautious interpretation.
The reported value has measurement uncertainty. The worker's exposure also varies from day to day. A single result just below a limit therefore does not prove that exposure will remain below it on other days. A single result just above a limit may likewise require confirmation of validity and circumstances, although it should not be dismissed as mere variation.
Interpretation should consider analytical uncertainty; sampling and contextual uncertainty; how representative the day was; the distribution of other results; whether higher-exposure conditions were included; the consequence of an incorrect conclusion; and whether the dataset supports an upper-percentile assessment.
Within Abu Dhabi's Occupational Standards and Guideline Values document (2016), Schedule A generally adopts ACGIH Threshold Limit Values for airborne chemical agents. Section 3.2 states that those values "shall be adopted as maximum allowable limits", but that directive wording sits inside a document whose own introductory note describes its values as currently non-mandatory requirements, and which sits in the Standards and Guideline Values layer of the framework rather than among the mandatory Codes of Practice. Abu Dhabi Public Health Centre (ADPHC) now records the document as suspended, directing entities to comply with relevant local or federal standards in force. The schedules are therefore a published reference point, not an enforceable UAE limit. Schedule B adopts NIOSH occupational noise limits, although noise assessment is outside this page.
A conclusion that exposure is adequately characterised should be based on the dataset as a whole rather than on whether the highest result is below a limit.
The evidence is stronger where the SEG is coherently defined; relevant workers and conditions are represented; measurements span more than one day; the analytical method is sufficiently sensitive; qualified or invalid data are treated transparently; variability is evaluated; the result distribution is comfortably separated from the criterion; and uncertainty would not reasonably reverse the conclusion.
Evidence is weaker where all results come from one person, quiet days dominate, non-routine activities are absent, the LOQ is close to the criterion or a small dataset is highly variable.
The outcome may be expressed as adequate evidence of low exposure, evidence of potential exceedance, or an indeterminate position requiring more information. "Indeterminate" is a legitimate technical conclusion where the available numbers do not support a defensible decision.
The resulting judgement and its assumptions should be recorded in the exposure monitoring report rather than left as an undocumented calculation.
Two workers with the same title may spend different proportions of the shift near a source. A consistent subgroup of higher values may indicate a distinct exposure pattern rather than random variation.
Ten measurements taken from one worker on similar quiet days may provide less information about a group than a smaller but deliberately balanced set spanning different workers, shifts and operating conditions.
Between-worker variability reflects consistent differences among people; between-day variability reflects changes in production, material, maintenance state and weather. Repeated measurements on selected individuals can help separate them.
A single result just below a criterion does not prove that exposure will remain below it on other days. Measurement uncertainty and day-to-day variability both apply.
EN 689:2018 provides one recognised framework for comparing inhalation exposure with occupational exposure limits, using a preliminary test on small datasets and a statistical test requiring at least six valid measurements. That structure is recognised practice rather than a UAE legal requirement. Within Abu Dhabi's Occupational Standards and Guideline Values document (2016), Schedule A generally adopts ACGIH Threshold Limit Values for airborne chemical agents and Schedule B adopts NIOSH occupational noise limits. Section 3.2 states that those values shall be adopted as maximum allowable limits, but that directive wording sits inside a document whose own introductory note describes its values as currently non-mandatory requirements, and which sits in the Standards and Guideline Values layer of the framework rather than among the mandatory Codes of Practice. Abu Dhabi Public Health Centre (ADPHC) now records the document as suspended, directing entities to comply with relevant local or federal standards in force. The schedules are therefore a published reference point, not an enforceable UAE limit.
No general statement that UAE law requires exposure monitoring or a particular statistical test should be made without a named primary legal source.
No. It describes one measured circumstance. Variability, representativeness and uncertainty determine how far the conclusion can be extended.
No. Six is the minimum for certain statistical procedures in EN 689, not a universal guarantee of confidence. More may be needed where variability is high or results are close to the criterion.
Grouping allows measurements from selected workers to inform an assessment of others with genuinely comparable exposure patterns.
Neither should be used in isolation. The appropriate statistic depends on the assessment question, averaging period, distribution and decision framework.
It means the available evidence cannot reliably establish the exposure position. Additional or better-targeted data may be necessary.