Laboratory QA/QC & Data Quality

Laboratory data are used to make treatment decisions, demonstrate compliance, troubleshoot processes, and identify trends. For those data to be useful, operators must have confidence that the sample, instrument, method, calculations, and records are reliable.

Quality assurance and quality control provide the structure used to produce and evaluate dependable laboratory results. Operators do not need to be laboratory managers, but they should understand the basic QA/QC concepts that help distinguish real process changes from analytical error.

What Is Quality Assurance?

Quality assurance, or QA, is the overall system used to support reliable laboratory work.

QA can include:

• approved procedures;
• staff training;
• instrument maintenance;
• calibration programs;
• sample-handling requirements;
• documentation;
• data review;
• corrective-action procedures.

What Is Quality Control?

Quality control, or QC, consists of specific checks used to evaluate whether an analytical process is working properly.

QC samples and checks may include:

• blanks;
• duplicates;
• standards;
• spikes;
• calibration verification;
• control samples.

QA Versus QC

QA describes the broader quality system.

QC describes individual tests and checks used within that system.

Both are needed for dependable laboratory data.

Data Quality

Good laboratory data should be suitable for the decision being made.

Important data-quality characteristics include:

• accuracy;
• precision;
• representativeness;
• completeness;
• comparability;
• traceability.

Accuracy

Accuracy describes how close a result is to the true or accepted value.

For example, if a standard has a known value of 10.0 mg/L and the laboratory measures 10.1 mg/L, the result is close to the accepted value.

Precision

Precision describes how closely repeated measurements agree with each other.

For example, results of:

• 10.1 mg/L;
• 10.0 mg/L;
• 10.1 mg/L

show good precision.

Accuracy and Precision Are Different

A set of results can be:

• accurate and precise;
• accurate on average but not precise;
• precise but inaccurate;
• neither accurate nor precise.

Precise but Inaccurate

Suppose the true value is 10.0 mg/L but repeated measurements are:

• 12.0 mg/L;
• 12.1 mg/L;
• 12.0 mg/L.

The results agree closely with each other, so precision is good.

However, they are far from the true value, so accuracy is poor.

Representativeness

A result can be analytically accurate but still misleading if the sample does not represent the process.

Representativeness depends on:

• sampling location;
• sampling time;
• sample type;
• sample handling.

Completeness

Completeness describes whether the required amount of valid data was successfully collected and analyzed.

Missing samples, invalid results, or failed analyses reduce completeness.

Comparability

Comparability means results can reasonably be compared because they were produced using compatible:

• methods;
• units;
• sample locations;
• collection procedures;
• reporting conventions.

Traceability

Traceability means a reported result can be connected back to:

• sample;
• collection time;
• analytical method;
• instrument;
• calibration;
• analyst;
• original records.

Laboratory Blanks

A blank is used to detect contamination or background interference.

A blank should contain little or none of the analyte being measured, depending on the specific QC procedure.

Why Blanks Matter

If a blank shows an unexpected concentration, possible causes include:

• contaminated reagent;
• dirty glassware;
• contaminated sampling equipment;
• carryover;
• laboratory contamination.

Method Blank

A method blank is processed through the analytical procedure without a normal field sample.

It helps identify contamination introduced by:

• reagents;
• glassware;
• analytical preparation;
• laboratory environment.

Field Blank

A field blank is handled in the field to evaluate contamination associated with:

• sampling;
• field handling;
• transport.

Equipment Blank

An equipment blank can be used to determine whether reusable sampling equipment was cleaned adequately.

Trip Blank

A trip blank travels with specified sample containers and can be used for certain volatile analyses to detect contamination during transport and handling.

Duplicate Samples

Duplicates are repeated analyses or samples used to evaluate precision.

Duplicates can help reveal:

• analytical variability;
• sample heterogeneity;
• sampling variability.

Laboratory Duplicate

A laboratory duplicate is created from the same sample and analyzed separately.

This primarily evaluates analytical precision.

Field Duplicate

A field duplicate is a second sample collected from the same location and time as closely as practical.

It includes both sampling and analytical variability.

Duplicate Results Should Be Similar

If duplicate results differ greatly, possible causes include:

• poor mixing;
• heterogeneous sample;
• sample contamination;
• analytical error;
• instrument instability.

Relative Percent Difference

Duplicate agreement is sometimes evaluated using Relative Percent Difference, or RPD.

A common formula is:

RPD = |Result 1 - Result 2| ÷ Average of Results × 100

RPD Example

Suppose duplicate results are:

• Result 1 = 20 mg/L;
• Result 2 = 22 mg/L.

Average:

(20 + 22) ÷ 2 = 21 mg/L

Difference:

|20 - 22| = 2 mg/L

RPD:

2 ÷ 21 × 100 = 9.5%

Whether that is acceptable depends on the method and laboratory QC criteria.

Standards

A standard contains a known concentration or property and is used to check analytical performance.

Standards may be used for:

• instrument calibration;
• calibration verification;
• accuracy checks;
• method-performance checks.

Calibration

Calibration establishes the relationship between instrument response and known reference values.

Examples include:

• pH buffer calibration;
• spectrophotometer calibration curve;
• conductivity standard;
• DO instrument calibration.

Calibration Verification

A calibration verification check determines whether an existing calibration remains acceptable.

If verification fails, the instrument or method may require:

• recalibration;
• maintenance;
• investigation;
• reanalysis of affected samples.

Calibration Curve

Some laboratory methods use several standards to establish a calibration curve.

The instrument response from an unknown sample is then compared with the calibrated relationship.

Do Not Extrapolate Carelessly

If a sample result is above the established calibration range, the result may require:

• dilution;
• reanalysis;
• another approved procedure.

Do not assume the calibration remains reliable beyond its validated range.

Spikes

A spike is a known amount of analyte added to a sample or prepared matrix.

Spike recovery can help evaluate:

• method accuracy;
• matrix interference;
• analytical recovery.

Percent Recovery

A simplified recovery relationship is:

Percent Recovery = Amount Recovered ÷ Amount Added × 100

Detailed calculations may account for the original sample concentration depending on the QC method.

Matrix Effects

A matrix is everything in the sample other than the specific analyte being measured.

Wastewater can contain:

• solids;
• color;
• salts;
• organic compounds;
• other chemicals

that interfere with an analytical method.

Matrix Spike

A matrix spike adds a known amount of analyte to an actual sample.

Poor recovery may indicate that the sample matrix affects the analytical method.

Control Samples

A laboratory control sample contains a known analyte concentration in a controlled matrix.

It helps evaluate analytical performance independent of the actual sample matrix.

Control Limits

QC results are often compared with established acceptance limits.

If a QC result falls outside the allowed range, the laboratory should follow its corrective-action procedure.

Do Not Ignore Failed QC

A sample result should not automatically be accepted when associated quality-control checks fail.

The cause of the QC failure should be evaluated.

Control Charts

A control chart plots QC results over time.

Control charts can show:

• stable performance;
• gradual drift;
• sudden shifts;
• increasing variability.

Trend Before Failure

An instrument or method may begin drifting before an individual QC result exceeds a formal acceptance limit.

Trend review can help identify developing problems earlier.

Outlier

An outlier is a result that differs substantially from the surrounding data.

An outlier can represent:

• a real process event;
• sampling error;
• analytical error;
• instrument failure;
• data-entry error.

Do Not Delete an Outlier Just Because It Looks Wrong

Unexpected data should be investigated.

Review:

• sample collection;
• QC results;
• instrument condition;
• calculations;
• process conditions.

Real Process Change Versus Bad Data

A sudden laboratory change is more believable when it agrees with independent evidence.

Examples include:

• high effluent ammonia plus low DO;
• higher TSS plus poor clarifier settling;
• lower chlorine residual plus increased chlorine demand;
• higher conductivity plus known industrial discharge.

Questionable Data

A result may deserve investigation when:

• it is physically impossible;
• it conflicts with multiple independent measurements;
• QC fails;
• sample identification is uncertain;
• holding time was exceeded;
• instrument calibration failed.

Data Validation

Data validation is the review of results and associated information to determine whether the data are acceptable for their intended use.

Validation may include review of:

• sample identification;
• holding time;
• preservation;
• calibration;
• blanks;
• duplicates;
• standards;
• calculations;
• method requirements.

Data Qualification

Sometimes a result can still be reported but must be qualified because a condition did not fully meet normal criteria.

Qualification should explain the limitation rather than hide it.

Invalid Data

A result may be unusable when the analytical or sampling problem is serious enough that the original condition cannot be reliably determined.

Examples can include:

• wrong sample;
• severe contamination;
• critical holding-time failure;
• major calibration failure.

Documentation

Laboratory records should allow another qualified person to understand what was done.

Records may include:

• sample ID;
• date and time;
• analyst;
• method;
• instrument;
• calibration;
• raw data;
• calculations;
• QC results;
• final result.

Record Raw Data

Raw observations should be retained according to applicable procedures.

Examples include:

• instrument readings;
• weights;
• volumes;
• titration endpoints;
• absorbance values.

Do Not Backfill Data

Laboratory and operating records should be completed honestly and at the appropriate time.

Do not invent or reconstruct values that were never actually measured.

Correcting Records

When a paper record requires correction, good practice is to preserve the original entry so the change remains traceable.

Corrections should follow laboratory or facility procedures.

Transcription Errors

A correct analytical result can become wrong when transferred incorrectly into:

• spreadsheet;
• log sheet;
• report;
• regulatory system.

Check Units During Data Entry

Common unit errors include confusion between:

• mg/L and µg/L;
• mg/L as N and mg/L as NO3;
• mg/L as P and mg/L as PO4;
• gallons and liters.

Decimal-Point Errors

A misplaced decimal can create a result ten, one hundred, or one thousand times different from the real value.

Unusual results should be checked against expected ranges.

Detection Limit

Analytical methods have limits on how small a concentration can be reliably detected or reported.

A result below the method's reporting capability should not automatically be treated as an exact zero.

Non-Detect Results

A reported non-detect generally means the analyte was not detected at or above the applicable method reporting threshold.

It does not prove that absolutely none of the substance exists.

Significant Figures

Laboratory results should not imply more precision than the method supports.

Reporting many unnecessary decimal places can create a false impression of analytical certainty.

Replicate Measurements

Repeated measurements can improve confidence when the method calls for them.

However, repeatedly testing until a preferred number appears is not valid quality control.

Instrument Maintenance

Reliable laboratory data depend on instruments being:

• clean;
• calibrated;
• maintained;
• operated within specified conditions.

Reagent Quality

Reagents can affect analytical accuracy.

Check:

• correct reagent;
• expiration date;
• storage condition;
• preparation date where applicable;
• contamination.

Standard Preparation

Errors in preparing a standard can affect every sample analyzed from that calibration.

Accurate preparation requires correct:

• mass;
• volume;
• dilution;
• labeling.

Glassware and Volumetric Equipment

Laboratory measurements may depend on equipment such as:

• pipettes;
• burettes;
• volumetric flasks;
• graduated cylinders.

The appropriate device should be selected for the accuracy required.

Contaminated Glassware

Dirty glassware can introduce:

• analyte contamination;
• chemical interference;
• residue from previous samples.

Laboratory Temperature and Environment

Some analytical procedures can be affected by:

• temperature;
• dust;
• humidity;
• vibration;
• chemical vapors.

Analyst Technique

Human technique can influence:

• pipetting;
• titration;
• sample mixing;
• timing;
• endpoint recognition;
• instrument operation.

Training and Competency

Laboratory personnel should be trained for the procedures they perform.

Consistent technique improves data comparability and reliability.

Corrective Action

When QC fails, corrective action may include:

• checking calculations;
• recalibrating;
• preparing fresh reagent;
• cleaning equipment;
• repeating QC;
• reanalyzing samples;
• documenting the problem.

Do Not Automatically Reanalyze Until a Passing Result Appears

The cause of a failed QC check should be understood.

Repeated testing without investigating the cause can hide a real analytical problem.

Preventive Laboratory Quality Control

Good QA/QC is easier when problems are prevented through:

• routine calibration;
• clean equipment;
• fresh reagents;
• proper sample preservation;
• consistent methods;
• complete records.

Review Trends in QC Data

QC results can be trended just like process data.

Useful warning signs include:

• steady calibration drift;
• increasing duplicate difference;
• declining spike recovery;
• increasing blank contamination.

Process Data and Laboratory Data Should Support Each Other

Laboratory results are strongest when they are consistent with:

• online instrumentation;
• process observations;
• equipment status;
• historical trends.

Example: Suspicious Ammonia Result

Suppose effluent ammonia is normally 1 mg/L and suddenly reports 18 mg/L.

Before making a major process change, review:

• sample ID;
• sample location;
• QC results;
• DO;
• pH;
• alkalinity;
• nitrate trend;
• other recent ammonia results.

If process indicators also show nitrification failure, the high result becomes more credible.

Example: Suspicious TSS Result

If a reported TSS value increases sharply but:

• effluent remains visibly clear;
• turbidity is unchanged;
• clarifier performance is normal;

review sample handling, filter preparation, weighing, and calculations before concluding that the process failed.

Common QA/QC Mistakes

• Confusing accuracy with precision.
• Ignoring blank contamination.
• Ignoring large duplicate differences.
• Using an instrument after calibration verification fails.
• Discarding unexpected results without investigation.
• Accepting results even though holding time or preservation failed.
• Failing to document corrective action.
• Transcribing correct results incorrectly.
• Confusing units or reporting basis.
• Treating non-detect as an exact zero.
• Reporting more decimal places than the method supports.
• Repeating analyses until a preferred result is obtained.

A Practical Laboratory Result Review

1. Confirm the sample ID and location.
2. Confirm collection date and time.
3. Review preservation and holding time.
4. Review calibration status.
5. Review blanks, duplicates, standards, and other required QC.
6. Check calculations and units.
7. Compare the result with historical data.
8. Compare with independent process measurements.
9. Investigate unusual or inconsistent results.
10. Document any qualification or corrective action.

A Practical QC Failure Review

1. Identify which QC check failed.
2. Stop accepting affected results automatically.
3. Check calculations and transcription.
4. Check calibration and instrument condition.
5. Check reagents and standards.
6. Check sample and glassware contamination.
7. Repeat appropriate QC according to procedure.
8. Reanalyze affected samples when required.
9. Document the cause and corrective action.
10. Verify acceptable performance before normal analysis resumes.

What to Remember for the Exam

• Quality assurance is the overall system used to produce reliable laboratory data.
• Quality control consists of specific checks used to evaluate analytical performance.
• Accuracy is closeness to the true or accepted value.
• Precision is agreement among repeated measurements.
• A result can be precise but inaccurate.
• Representative sampling is part of overall data quality.
• Blanks help identify contamination or background interference.
• Duplicates help evaluate precision and variability.
• Standards contain known values and are used for calibration and accuracy checks.
• Calibration establishes the relationship between instrument response and known reference values.
• Calibration verification checks whether an existing calibration remains acceptable.
• RPD can be used to evaluate agreement between duplicate results.
• Spikes can help evaluate recovery and matrix interference.
• QC results should be compared with established acceptance criteria.
• Failed QC should be investigated rather than ignored.
• Outliers can represent real process changes or analytical problems and should be investigated.
• Data validation reviews sampling, preservation, calibration, QC, calculations, and other information before results are accepted.
• Non-detect does not necessarily mean an exact concentration of zero.
• Traceable records should connect a reported result to the sample, method, instrument, analyst, and raw data.
• Laboratory data should be interpreted together with process observations and independent measurements.