The laboratory as an institution
Run laboratories whose results can be trusted and reconstructed after the investigator has gone: written procedures, calibrated instruments, raw data recorded and archived, quality checked by people independent of the work, hazards controlled by assessment, and competence proven to an outside body.
Everything for this step
- Field guide
- Chemistry → 9 sections
- Builds
- None yet.
- Reference sheets
- None.
- Track
- Science and measurement, level 2 of 3 · next: Knowing whether a result is real
Why this matters
A gifted experimenter working alone produces results nobody can check, in notebooks nobody can read, with instruments nobody else can use. The laboratory as an institution turns that into a public, repeatable service. The OECD's Principles of Good Laboratory Practice set the rules for safety studies that governments rely on: written procedures for every routine operation, every entry "recorded directly, promptly, accurately, and legibly" and signed, no correction that obscures what it replaces, a quality assurance programme "independent of study conduct", and an archive from which any study can be reconstructed (OECD GLP).
Testing and calibration laboratories prove their competence against ISO/IEC 17025, the criteria accreditation bodies use (ISO); results from a laboratory accredited by a signatory of the ILAC arrangement are accepted by all the other signatories (ILAC).
The shape of it
- Fund the laboratory, not the investigator: a standing budget for technicians, maintenance and the archive.
- Write a standard operating procedure for every routine operation, approved by management.
- Inspect, clean, maintain and calibrate every instrument on a schedule, traceable to national standards.
- Record raw data directly, promptly and legibly, signed and dated, and never obscure a correction.
- Give quality assurance to people independent of the work, reporting straight to management.
- Archive every study so it can be reconstructed, and keep the archive if the laboratory closes.
- Assess the hazard of every procedure before it starts, and control it to match.
- Prove competence from outside: accreditation and proficiency tests.
Doing it, step by step
The summary above says what has to happen. This says how, with the numbers that decide whether it works.
1 · Fund the institution, not the investigator
Instruments, technicians and records must outlive any one scientist. Give the laboratory a standing budget, separate from any single project, for:
- technicians who maintain the instruments and train newcomers;
- maintenance and calibration of every instrument;
- the archive and its keeper;
- safety: ventilation, storage and waste.
Name a manager responsible for the facility, and have them appoint qualified staff for both the experimental work and an independent quality assurance function (OECD GLP). Keep records of every person's qualifications, training and experience.
2 · Write the procedures down
A standard operating procedure (SOP) is the laboratory's written method for a routine task, approved by management, and revised only with management's approval (OECD GLP 7.1). Write one at least for:
- receiving, labelling, sampling and storing test materials;
- using, maintaining, cleaning and calibrating each apparatus;
- validating, operating, securing and backing up computer systems;
- preparing and labelling reagents and solutions;
- record keeping, reporting, storage and retrieval.
Keep the historical file of every SOP, so it is known which version was used in any study.
3 · Calibrate and maintain every instrument
Apparatus should be periodically inspected, cleaned, maintained and calibrated according to SOPs, with records kept, and calibration, where appropriate, traceable to national or international standards of measurement (OECD GLP). If the laboratory keeps its own reference standards, have them calibrated by a competent body at prescribed intervals — the chain that ends at the national standards of standard units.
Label each instrument with its calibration date and the date the next is due, and take it out of use when it is overdue.
4 · Record raw data so it can be trusted
The OECD rules for raw data are the core of the laboratory's credibility:
- Record all data directly, promptly, accurately and legibly, by the person who made the observation, signed or initialled and dated.
- Make any change so as not to obscure the previous entry, with the reason, the date and the signature of the person making it.
- In a computer system, identify who entered each item, and keep a full audit trail that shows every change without obscuring the original.
Use bound notebooks with numbered pages, in ink, and never loose sheets. Publish through the journal only results that can be traced back to these records.
5 · Keep quality assurance independent
A quality assurance programme is "a defined system, including personnel, which is independent of study conduct" (OECD GLP). Its staff:
- inspect studies, facilities and processes on a schedule, and some at random;
- report every finding promptly, in writing, to management and the study's director;
- sign a statement in each final report listing their inspections and confirming that the report reflects the raw data.
They take no part in the work they monitor, and the person responsible for quality assurance reports directly to top management.
6 · Archive everything needed to reconstruct a study
OECD GLP section 10 lists what to keep, for the period the authorities set:
- the study plan, raw data, samples, specimens and the final report;
- quality assurance inspection records;
- staff qualification and training records;
- instrument maintenance and calibration records;
- the historical file of all SOPs.
Index the archive, let only authorised staff in, record every item that moves in or out, and protect it from deterioration. If the laboratory closes with no successor, transfer the archive to the study's sponsor.
7 · Assess the hazard, then control it
The WHO Laboratory Biosafety Manual (4th edition, 2020) replaces fixed rules with an evidence-based risk assessment of each procedure, in five steps (WHO):
- gather information on the agent, the procedure and the staff's competence;
- evaluate the risks;
- develop a risk control strategy;
- select and implement control measures;
- review the risks and the controls.
Apply the core requirements — the minimum tools, training and physical and operational controls — to all work, and add heightened or maximum containment measures only where the assessment calls for them. Assess chemical and physical hazards the same way.
8 · Prove competence from outside
A laboratory's own word is not enough. Seek accreditation against ISO/IEC 17025 by a national accreditation body: accreditation bodies use it as their criteria for testing and calibration laboratories (ISO). Laboratories accredited by members of ILAC take part in proficiency testing — analysing the same samples as other laboratories and comparing results — as a condition of keeping their accreditation (ILAC).
Investigate every unsatisfactory proficiency result to its cause before reporting more results of that kind.