Retrospective solvent exposure: is this an incident worth investigating?

The manager is using a narrow definition of 'incident' - an event with visible harm. The investigation trigger here is different: a cluster of symptoms in a group sharing a common, identifiable exposure, which is a potential occupational ill-health incident even though nothing was reported and no single event can be dated.

Reasoning to put to the manager

• Two of perhaps five or six people doing the same task reporting the same symptoms within weeks is unlikely to be coincidence; the base rate of headaches and dizziness in a working population does not cluster around one bench.
• The exposure is plausible and uncontrolled: dichloromethane in an open 25-litre drum, 2 m from the breathing zone, no local exhaust ventilation, warm ambient temperature increasing evaporation, shutter closed reducing general ventilation.
• The consequences are serious and irreversible in kind (central nervous system effects, and a carcinogen classification), so the potential severity justifies investigation even at low probability.
• There is a learning value: whether the substance can be eliminated or substituted, and whether the controls ever existed.
• Check your jurisdiction's reporting criteria for occupational ill-health and dangerous occurrences, and check the safety data sheet classification - the answer may create a notification duty, but the decision to investigate should not depend on that; it depends on the risk.

What kind of investigation Adapt rather than copy the accident model: there is no scene or single moment. It becomes a retrospective exposure assessment combined with a causal analysis of why the controls were absent - the 'root cause' question is why an unread data sheet, an open drum and no ventilation were tolerated for eight weeks.

Also consider whether others are at risk before deciding how fast to move: if the answer is yes, control the exposure first and investigate in parallel.

What to collect

• Task and time data: how many hours per shift at the bench, how many shifts per week over the eight weeks, and whether the drum was decanted or left open between uses. Ask for production records rather than relying on recall.
• Quantity and product data: purchase invoices and stock records to establish litres used per week; the current safety data sheet; confirmation that the product has not been substituted.
• Workplace data: room volume (floor area × height gives an air volume for dilution estimates), measured ventilation rate with the shutter closed, temperature logs.
• Health timeline: onset dates, whether symptoms improved on days off or the two-day absence (a strong indicator of work-relatedness), and the GP's record. Seek consent before contacting the GP.
• Current measurement: occupational hygiene sampling during a representative shift - personal breathing-zone samples across the full shift, plus a short-term sample for peak tasks - with the drum in its normal state, not a tidied version.
• Comparison data: records for the other workers on the same task; whether anyone else has symptoms.

Limits to state openly in the report

• Sampling today cannot prove what the air concentration was eight weeks ago; you are measuring a current task, and you should say whether conditions have changed.
• Back-calculation requires assumptions (evaporation rate, air changes) that carry wide uncertainty, so present a range, not a figure.
• Symptoms are non-specific and a 'viral illness' diagnosis does not exclude exposure; you cannot diagnose.
• Recall of hours and practices is imperfect, and workers may under- or over-report once they suspect the solvent.

Conclude with an exposure judgement expressed with confidence and assumptions, and recommend health surveillance through an occupational health provider regardless of the sampling result.

Sequencing Interview the two affected workers first, individually and separately, before the supervisor - so their accounts are not shaped by what a manager has said or by each other. Then the supervisor, then whoever buys chemicals, then a sample of other bench workers. Interview the affected pair on different days if possible, so they do not compare notes overnight.

Framing Open with the purpose: to understand the work and the exposure, not to attribute blame, and that the finding goes into a report that will recommend controls. Be honest about confidentiality limits - if you cannot promise anonymity, say so. Make clear that reporting symptoms will not put their jobs at risk; agency or probationary status makes workers especially reluctant.

Question style Use open, non-leading questions: 'Walk me through your shift from starting the frames to finishing', 'When in the shift do you notice the headache?', 'What did you do about it?', 'Did anything change eight weeks ago?' Avoid 'Do you think the solvent is making you ill?' - that invites the answer you expect and contaminates the evidence. Probe for timing: symptoms better at weekends or during the two days off is exactly the detail that supports work-relatedness. Ask what they know about the product and whether anyone explained the data sheet.

Pitfalls

• Suggestion and expectation bias once a hazard is suspected.
• Group interviews, which produce a single shared story.
• Turning the interview into a medical assessment; note what is said and refer.
• Language or literacy barriers - use an interpreter or a colleague the worker trusts, not a manager.
• Recording only conclusions. Take contemporaneous notes, read them back, date and sign, and keep the worker's own words for the report.