Twelve Samples, One 'Typical' Week: Is This Solvent Exposure Acceptable?
At a 40-employee furniture refinishing shop, three technicians spray a solvent-based lacquer containing toluene (internal 8-hour TWA OEL of 25 ppm) in a 9,000 ft3 booth. You collected 12 full-shift personal samples on consecutive days during what the plant manager called 'a typical week.' Results (ppm, 8-hour TWA): 5, 7, 9, 11, 12, 14, 16, 18, 20, 23, 27, 31. Sampling and analytical methods were validated, field blanks were clean, and no result was below the limit of detection. Two of the three highest results came from a technician who was wearing a half-mask organic vapor respirator during spraying because, she said, 'it smells bad otherwise.' The manager averages the results to 16 ppm and concludes the shop is 'well under the limit.' Production is scheduled to increase 30% next quarter.
1. Is this exposure assessment adequate to support the manager's conclusion? Show your reasoning using the data.
Read the worked answer
The mean is the wrong decision statistic. For a chronic-effect agent, the 8-hour OEL is a level that no single shift should exceed, so what matters is the distribution of day-to-day exposures, not the average of 12 days.
- A lognormal fit to the 12 results gives a geometric mean of about 14 ppm and a geometric standard deviation of about 1.7.
- The estimated fraction of shifts above 25 ppm is about 15% (observed: 2 of 12 = 17%) - roughly one shift in seven.
- The point estimate of the 95th percentile is about 35 ppm, already above the OEL, and with only n = 12 the upper confidence bound on that percentile is far higher.
- The data therefore cannot demonstrate compliance; they are more consistent with a routinely overexposed group.
Two further weaknesses: all samples came from one week on consecutive days, so between-week and seasonal variability are unmeasured, and a 30% production increase will shift the entire distribution upward. The respirator wearer's samples remain valid - they measure the airborne concentration generated by the task, which drives everyone else's exposure; her inhaled dose is lower but the source strength is unchanged.
2. Design the next sampling round: strategy, number of samples, duration, and quality assurance. State what decision the results will support.
Read the worked answer
Strategy. Move from convenience sampling to a task-based, randomly selected strategy with defined similar exposure groups (SEGs): spray gun operator, gun cleaning/setup, booth entry/cleanout, mixing. Randomize which technician and which day is sampled so high-exposure tasks are not systematically under- or over-represented.
Number and duration. Collect 6–10 full-shift (at least 7 hours) personal samples per SEG, plus 15-minute short-term samples during spraying and gun cleaning to capture peaks and any ceiling limit. Sample outside the respirator (ambient concentration at the breathing zone) and record respirator use as a separate variable so you can model both ambient exposure and inhaled dose.
QA. Pre- and post-calibration within 5%, one field blank per 10 samples minimum, media from the same lot, temperature/humidity recorded, chain of custody, and documentation of production rate and booth conditions so a 30% increase can be modeled.
Decision rule. Compute the 95% upper confidence limit on the exceedance fraction. If it is below 5%, exposure is reasonably characterized as controlled and you move to periodic confirmation. If not - as the current data suggest - the sampling justifies engineering controls, not more sampling.
3. The plant proposes to install a new spray booth in nine months. What do you recommend in the interim, and how do you handle the respirator issue?
Read the worked answer
Respirators are the last line of defense, not a control strategy, and an informal, self-issued respirator is a compliance gap in itself.
Interim engineering and work-practice controls (in order of preference):
- Substitution - evaluate a waterborne or lower-solvent coating; this is the only step that reduces source strength for everyone.
- Enclosure and local exhaust at the gun-cleaning station (a major unmeasured contributor) and improvement of booth airflow within existing capacity.
- Work practice - lid on solvent trays, no open pans, gun washer instead of hand cleaning, staggered spray schedules to reduce peak concentrations.
Respirator program. Because a respirator is already in use, it must be program-managed now: written program, medical evaluation, fit testing, training, cartridge change schedule, and storage. Do not let informal use continue, because it creates a false sense of protection and hides the real exposure data.
Documentation. State in writing that exposures are not controlled under the current configuration, that interim measures are time-bound to booth installation, and that effectiveness will be verified by repeat sampling after start-up.