CIH: Sampling Strategy, Analytical Methods and Statistics
A CIH sampling strategy begins with a defined exposure question, target population, task, and decision criterion.
- Select a representative design, method, sample duration, and analytical approach, then document calibration and field conditions.
- Interpret measurements with variability, detection limits, data quality, and uncertainty rather than treating one result as a complete exposure profile.
On this page12 sections
- Design sampling around a decision
- Define the population and similar exposure group
- Select personal, area, task, or direct-reading methods
- Plan sample size and timing
- Check calibration and field performance
- Choose an analytical method and understand its limits
- Understand distributions and variability
- Handle nondetects, outliers, and missing data transparently
- Compare results with the right criterion
- Make sampling records usable
- Worked design example
- Check whether results support a comparison
Design sampling around a decision
Sampling is a means to answer a decision question, not an end in itself. Define whether the study concerns a full shift, a task, a peak, a control check, or a comparison among workers. Specify what result would change the recommendation. This determines the population, method, timing, number of measurements, and records needed.
The BGC blueprint expects a CIH to design exposure assessment strategies and apply statistical principles to study design. A plan should state assumptions before data collection. If the question is vague, samples can be technically correct yet unhelpful. Write the purpose in plain language and ask a colleague to confirm that the proposed design actually addresses it.
Define the population and similar exposure group
A similar exposure group is a set of workers expected to have comparable exposure because they share relevant tasks, processes, materials, controls, and work practices. Job title alone may be too broad. Observe actual work and ask workers how conditions vary. If a group includes different operations or exposure patterns, divide it into more meaningful subgroups.
The sampling unit can be a worker, shift, task, or workday, depending on the objective. Multiple measurements help characterize variability. Document how members were selected and whether they represent routine or unusual conditions. If the group definition changes after observation, record the revision and explain its effect on interpretation.
Select personal, area, task, or direct-reading methods
Personal samples can estimate exposure associated with an individual’s work during a defined period. Area samples characterize a location or source condition but may not represent breathing-zone exposure. Task-based samples can focus on a short operation. Direct-reading instruments can show changes over time, but their specificity, response, and interferences must be understood.
The method should fit the agent and the question. Consider collection medium, flow rate, duration, analytical technique, expected concentration, detection limit, and sample stability. A monitor with a rapid response may identify peaks but require confirmation. A laboratory method may be specific but take longer and integrate exposure across a period.
Plan sample size and timing
The number of samples depends on the variability, decision stakes, population, and desired confidence. A single sample rarely describes a whole workforce across different days. Collect measurements across workers, shifts, or tasks when variability matters. Include representative production and operating conditions, and document why any unusual or high-exposure condition was included or analyzed separately.
Timing should capture the relevant exposure period. A full-shift method can support a time-weighted question, while a short task may require a task-specific design. Do not convert a task sample into a full-shift conclusion without a valid basis. Account for sample duration, breaks, task sequence, and potential exposure outside the monitored window.
Check calibration and field performance
Calibration and instrument checks support reliable measurements. Follow method requirements for pre- and post-use checks, flow verification, zeroing, maintenance, and environmental limits. Record instrument identification, calibration equipment, dates, flow rates, and any failure. A sample collected with an out-of-tolerance instrument may need to be qualified or excluded under the applicable method.
Field notes should capture worker, task, materials, process rate, controls, location, duration, unusual conditions, and sample handling. These details help interpret results and allow another professional to understand how the data were produced. A laboratory report cannot recreate a missing field record.
Choose an analytical method and understand its limits
Analytical chemistry determines what the collected sample can reveal. Use an appropriate validated method for the agent and matrix. Review detection and quantitation limits, interferences, recovery, sample volume, storage, and stability. A method that cannot detect concentrations near the decision level may not answer the study question, even if the laboratory follows it correctly.
NIOSH’s Manual of Analytical Methods provides method information for occupational exposure assessment. Read the method itself and any current notes, not only a summary. Use the method’s prescribed sampling media and conditions. If a method is modified, document why and evaluate how the modification affects validity.
Understand distributions and variability
Exposure data often vary across workers, shifts, and tasks. A summary should identify the distribution and relevant sources of variability, not just the arithmetic average. Consider whether data are lognormally distributed, whether outliers represent data errors or real conditions, and whether measurements belong to the same exposure group. Statistical assumptions should be checked rather than applied automatically.
Distinguish within-worker variability from between-worker variability. Repeated samples from the same person answer a different question than samples across a group. A dataset collected on one day may not capture between-day variation. The purpose of the study determines which variability matters and what additional data are useful.
Handle nondetects, outliers, and missing data transparently
A nondetect is censored by an analytical limit; it is not a measured zero. Record the reporting limit and decide how the result will be handled under the selected statistical method. If the limit is too high relative to the decision criterion, consider another method or a different sample design. Do not replace nondetects with zero without a justified method.
Review outliers against field notes and laboratory quality information. An unusual value may indicate a data issue, but it may also represent a real task, failure, or high-exposure condition. Document any exclusion and its rationale. Missing samples and failed pumps should remain visible in the study record and may affect conclusions.
Compare results with the right criterion
Before comparing, verify the agent identity, units, averaging time, notation, and source of the exposure limit. A ceiling or short-term limit is not interchangeable with a time-weighted average. Some criteria include skin notation or other qualifiers. State whether the benchmark is a regulation, consensus standard, recommended limit, or internal target.
A comparison should consider analytical uncertainty, sampling error, and whether the sample represents the population. A result below a limit does not automatically mean no risk, and a value above a criterion does not by itself explain the source or control needed. Interpret the number in its health and work context.
Make sampling records usable
An exposure study file should allow a reviewer to understand the question, design, collection, analysis, and interpretation. Keep the protocol, field sheets, calibration records, chain of custody, laboratory reports, calculations, and communications together according to the organization’s record practices. Identify any deviations and explain whether they affect use of the data.
Protect confidential worker and medical information. Report results to affected parties through the appropriate process and avoid unnecessary identifiers in broader reports. The BGC blueprint includes field records and communication as professional skills because data quality and trust depend on how the assessment is documented and shared.
Worked design example
A facility wants to evaluate solvent exposure among operators who perform mixing and occasional filter changes. A defensible plan first observes both tasks and determines whether the same workers, materials, and controls apply. It defines whether the decision concerns shift exposure, filter-change peaks, or both. It then selects a personal method and a task-focused method if required, with analytical limits suitable for the comparison criterion.
The plan documents who is sampled, when, under what production rate, and how the instruments are checked. The analyst reviews variability and task notes before summarizing. If filter changes were missed, the report states that limitation and recommends targeted follow-up rather than claiming the monitoring characterized every task. This demonstrates the link between study design and a defensible conclusion.
| Design choice | Questions to resolve | Record |
|---|---|---|
| Population | Who shares exposure determinants? | Tasks, processes, group rationale |
| Measurement | What method fits agent and time? | Sampler, duration, analytical method |
| Quality | Were instruments and samples valid? | Calibration, handling, deviations |
| Statistics | What variability and limits remain? | Distribution, nondetects, uncertainty |
| Decision | What can the data support? | Criterion, conclusion, next action |
Check whether results support a comparison
A comparison is defensible only when the groups and conditions are meaningfully comparable. Check whether samples represent similar tasks, shifts, processes, and control states. A difference in measured values may reflect a true exposure difference, a change in operating conditions, or measurement variability. Describe the sampling frame, missing observations, detection limits, and any reason a result was excluded. Do not treat a convenient sample as a random sample or generalize beyond the group represented. When data are sparse, emphasize the limitation and use additional lines of evidence, such as task observation, process knowledge, or repeat sampling. Choose statistical summaries that match the distribution and decision. Report units consistently and retain the original data and calculation steps so another reviewer can reproduce the interpretation.
These concepts support exam study, not a site-specific professional assessment. Workplace exposure evaluations require qualified judgment, current methods, applicable criteria, and site context.
Common questions
What is a similar exposure group?
It is a group of workers expected to have comparable exposure because they share relevant tasks, processes, materials, controls and work practices. Define it from actual exposure determinants, not job title alone.
When are area samples appropriate?
Area samples can characterize a location, source, or control condition, but may not represent personal breathing-zone exposure. Select them only when they answer the stated question or supplement a broader design.
How should a nondetect be treated statistically?
A nondetect is below an analytical reporting or detection limit, not zero. Record the limit and use a justified method suited to the dataset and decision. If the limit is too high, the method may not resolve the question.
Why collect more than one sample?
Multiple measurements can characterize variability across workers, tasks, and days. The number and timing depend on the population, variability, and decision purpose. One result rarely represents all conditions.
What belongs in exposure sampling field notes?
Record the task, worker or group, materials, process conditions, controls, sampling duration, instrument and calibration details, deviations, and unusual events needed to interpret the measurements.