In today’s increasingly refined medical informatization, laboratory quality control is no longer the mission of single-point calibration and manual review. How to standardize, digitize, and intelligentize the quality control process with the help of the LIS system is a challenge that every product manager must face. This article will provide an in-depth analysis of the core value and design essentials of quality control modules in the LIS system, combined with practical cases, to reveal the product logic and optimization path behind the functions.
Under the red light of the emergency room, an error of 0.5mmol/L on a blood glucose test report may cause doctors to misjudge the rehydration regimen of diabetic ketoacidosis patients – excessive insulin may cause hypoglycemic shock, while insufficient dose can delay acidosis correction.
Similarly, in the oncology ward, subtle deviations in tumor marker test results may cause early patients to miss out on surgery opportunities, and may also cause advanced patients to suffer unnecessary chemotherapy side effects. The accuracy of clinical test results is never an internal matter in the laboratory, but a lifeline hanging above the patient’s head.
The quality control module of LIS (Laboratory Information System) is the core line of defense to protect this lifeline. It is not like a biochemical analyzer that can intuitively display the detection values, nor does it have a roaring running sound like a centrifuge, but it is like an invisible net, from the cold chain temperature monitoring during specimen collection, to the reagent matching verification during the test, to the rule review before the release of the report, through every key node of the whole inspection process.
The module was designed to be firmly rooted in the stringent requirements of international standards such as ISO15189 – such as mandatory requirements for the traceability of quality control products, but also to fit the actual scenario where laboratory personnel were racing against time: the quality control results of emergency specimens must be fed back within 30 minutes, and the time of repeated operations should be avoided during batch testing. After all, no matter how perfect the theoretical design is, if it cannot solve the pain points that are prone to errors when the front line is in a hurry, it is only on paper after all.
Next, we will dismantle how to make the LIS quality control module truly become the patron saint of quality inspection from three dimensions: functional module design, rule system construction, and data analysis application.
1. Quality control specimen processing
Quality control specimens are the reference material for testing quality, and every link in their circulation process may affect the final result. Therefore, the whole process management of quality control specimens needs to be as rigorous as tracking the calibration records of precision instruments, and control the quality trajectory from the source.
After 10 years of interaction design, why did I transfer to product manager?
After the real job transfer, I found that many jobs were still beyond my imagination. The work of a product manager is indeed more complicated. Theoretically, the work of a product manager includes all aspects of the product, from market research, user research, data analysis…
View details >
1.1 Specimen information management
The laboratory processes dozens or even hundreds of quality control specimens every day, from whole blood, serum to cerebrospinal fluid, from conventional biochemistry to gene amplification (PCR). Designing a smart ID card system for quality control specimens is the basis for avoiding such errors.
1.1.1 Uniqueness encoding
Coding rules need to take into account both readability and information, not too long to remember, but also contain key operational information, not only identification, but also traceability. For example, in QC-BLD-202501-001, BLD stands for blood, 202501 is the batch number, and 001 is the serial number—but this is only the basic. For special controls, the code needs to be further customized:
- Cold chain sensitive type: If the coagulation factor control product needs to be stored at -20°C, the code can be set to QC-COAG-202501-001-C2 (C2 represents -20°C);
- Molecular diagnostics: For example, the new coronavirus nucleic acid quality control product needs to contain gene fragment information, and the code can be set to QC-PCR-SARS2-N-202501-001 (SARS2-N represents the new coronavirus N gene);
- Time-sensitive type: If blood ammonia quality control products need to be tested within 1 hour after collection, the code can add 1H marking, such as QC-AMM-202501-001-1H.
The core value of this coding system lies in the traceability of the whole process. When the quality control results of a batch are abnormal, the test records of all specimens in the same batch can be quickly located through the code: whether only one instrument detects abnormalities (indicating instrument problems) or the whole laboratory is abnormal (indicating problems with the quality control products themselves), thereby shortening the troubleshooting time.
1.1.2 Status Tracking
From warehousing to testing completion, quality control specimens need to go through multiple circulation nodes, and delays or improper operations of any node may affect the results. The system needs to update the state in real time in a visual language and clarify the trigger conditions for state transitions:
- Green dot (collected): with collection time, operator ID, and environmental data at the time of collection (such as room temperature and humidity, especially important for PCR quality control products);
- Yellow moving arrow (in transit): record the transporter, transfer tool (such as special cold chain box number), real-time temperature (synchronized through cold chain box sensor), if the transportation time exceeds the maximum tolerance time of the quality control product (such as blood ammonia specimen exceeds 1 hour), automatically trigger the timeout warning;
- Blue pending inspection mark (signed): displays the list of instruments to be assigned, marking the current load of each instrument (e.g., instrument A: 5 copies have been queued, expected to wait for 20 minutes);
- Red exclamation mark (abnormal): subdivision of contamination (e.g., bacterial culture quality control products contaminated by bacteria), expiration (past expiration date), damage (container leakage) and other sub-states, and automatically associate the corresponding SOP (Standard Operating Procedures) treatment guidelines (e.g., contaminated specimens need to be destroyed according to biosafety level).
1.2 Specimen testing process management
The daily work of the laboratory is like a symphony of precision: people, instruments, and specimens need to work together seamlessly. The management of the testing process of quality control specimens is the conductor of this symphony, which must not only ensure the accuracy of testing, but also avoid waste of resources.
1.2.1 Task assignment
The task allocation logic of the system should simulate the experience and judgment of senior laboratory directors, and dynamically allocate them based on instrument status, personnel qualifications, and project characteristics:
- Instrument matching: For high-precision items such as troponin, chemiluminescence instruments that have just completed maintenance but are not calibrated are automatically excluded, and are preferentially assigned to instruments with an operating time of < 24 hours after calibration; For routine biochemical projects (such as liver function), the load of the instrument is evenly distributed to avoid excessive business of an instrument;
- Personnel matching: Coagulation function testing needs to be assigned to a technician with a special coagulation operation certificate, and the technician’s pass rate for coagulation quality control operations in the past 3 months needs to be ≥95%; Novices (6 months of employment <) can only assign regular projects and automatically associate the ID of senior employees with one-on-one guidance.
- Urgency: Quality control specimens marked with STAT (emergency) (such as emergency biochemistry), skip the regular queue, directly insert the first place in the testing queue, and automatically push the priority review request after the test is completed.
1.2.2 Optimization of detection sequence
In the laboratory, time is life is never just a slogan, emergency priority, efficiency is king. The system should prioritize the quality control specimen row according to the characteristics of the project, instrument compatibility, and urgency to reduce unnecessary waiting time:
- Priority of urgency: specimens marked with STAT directly jump the queue and alert technicians with flashing red light on the instrument operation interface;
- Instrument compatibility optimization: On the same hematology analyzer, complete blood count (no residual effect of reagents), then blood sedimentation (anticoagulants required, no effect on the former), and finally reticulocytes (special staining is required to avoid staining agent contamination of the first two).
- Batch combination: The same batch and type of quality control specimens (such as 10 biochemical quality controls) are centrally arranged for testing in the same time period to reduce the number of instrument cleaning and calibration (e.g., biochemical instruments are calibrated once before each batch of testing, instead of once per copy).
2. Generation of quality control report
A high-quality quality control report should not be a dense pile of numbers, but should be like an accurate narrator: use clear logic to tell clinicians whether the results are reliable, use intuitive charts to remind the laboratory director where the potential risks are, and use a standardized format to let regulators quickly judge whether they are compliant.
2.1 Report template
Laboratory needs are never one-size-fits-all: clinicians need to quickly see if they are under control during ward rounds, laboratory directors analyze trends in the past three months when summarizing monthly, and regulatory authorities pay attention to whether they meet ISO15189 requirements when inspecting. Therefore, the report template must be flexible. Thousands of people, customized on demand.
2.1.1 Template library
The preset template library of the system should be like a toolbox, which can be used at any time, and each template has a clear description of the applicable scenario:
- “Conventional Project Template”: The homepage highlights core information such as test results, target values, SD (standard deviation), and whether it is under control, and is intuitively identified with green checkmarks (under control), yellow question marks (warnings), and red crosses (out of control), and a brief conclusion is attached to the bottom (if the test is under control, the results are trustworthy), which is suitable for clinicians to quickly consult;
- “Microbiology Special Template”: In addition to basic information, increase the strain identification compliance rate (such as the degree of agreement with standard strains), drug susceptibility test quality control results (MIC value and target deviation of each antibiotic), and embed the morphological map of standard strains (such as Gram staining pictures) to facilitate comparison and analysis by microbiology technicians;
- “Regulatory Inspection Template”: Formatted in strict accordance with the appendix requirements of the ISO15189, including quality control frequency (such as once a day), out-of-control treatment records (including corrective measures and effect verification), personnel qualification certificates (with training record numbers), instrument calibration certificate numbers and other elements, and even reserve the signature column of inspectors, so that supervisors do not need to ask for additional information.
2.1.2 Customization
Each lab has its own specific needs, and customization needs to be flexible enough to allow the lab to speak according to its habits:
- Layout adjustment: Support dragging and dropping chart positions (such as moving the Levey-Jennings chart from page 3 to the home page), hiding redundant information (such as hiding professional data such as CV value in batches from clinicians).
- New fields: A children’s hospital laboratory needs to add the quality control results of disinfection of the heel blood collection site to the report (such as whether the alcohol disinfection time is ≥ 30 seconds); an infectious disease hospital requires the addition of indoor quality control and inter-laboratory quality evaluation results for HIV antibody testing;
- Rule embedding: The laboratory’s unique quality control judgment rules (such as the target range of neonatal bilirubin detection must be stricter than 20% for adults) can be written into the template, and the system will automatically judge whether it is under control according to this rule.
- SOP association: Insert SOP summaries of key operations in the report (e.g., 15 minutes of reconstitution of quality control products) to facilitate novice control operations and reduce errors.
2.2 Report generation and review
Automatic reporting ≠ all goes well, and it must pass the audit to ensure that the data is correct. The design of the audit process should be like a multi-layer filter, layer by layer to eliminate possible errors.
2.2.1 Autofill
After the system receives the quality control results from the instrument, it needs to automatically settle the accounts like a careful inspector and complete the preliminary sorting:
- Data calculation: Automatically calculate the deviation between the results and the target value (e.g., glucose detection result 8mmol/L, target value 5.0mmol/L, deviation +16%), CV value (coefficient of variation), Z-score (comparison value with interlaboratory quality assessment);
- Chart drawing: Automatically generate Levey-Jennings chart (the horizontal axis is the detection time, the vertical axis is the result, and the range of ±2s and ±3s is marked with red lines), trend chart (fluctuation curve of the results in the past 30 days), and histogram (comparison of the detection results of different instruments);
- Rule judgment: Automatically judge whether it is under control according to the preset Westgard multi-rules (such as 13s, 22s), and mark the basis for judgment (if the result exceeds +3s, trigger the 13s rule, it is judged to be out of control).
This step frees inspectors from tedious calculations and plotting, allowing them to focus more on the issues behind the data (such as why the results of the same project are so biased across two instruments).
2.2.2 Review process
A county-level hospital laboratory once made a serious mistake due to the direct release of automatic reports: the system mistakenly displayed mmol/L as mol/L (decimal point misalignment), and the auditors did not find it, which eventually led to an insulin dose calculation error in a diabetic patient, causing hypoglycemic coma. Therefore, the audit process must be rigidly implemented, and each level of audit has a clear review focus, checking at all levels, and the responsibility is assigned to the person:
- First-level audit (self-inspection by the tester): focus on checking whether the original data is consistent with the reported data (such as whether the instrument display value is the same as the reported filling value), whether the unit is correct (such as whether creatinine is μmol/L or mmol/L), and whether the chart is accurate (such as whether the points in the Levey-Jennings chart correspond to the correct detection time);
- Secondary review (team leader review): focus on the quality control rules to determine whether they are accurate (such as whether the 22s rule is misjudged – 2 consecutive results exceed +2s but not 3s), whether the labeling of abnormal results is reasonable (if a result is within ±2s, but fluctuates by more than 50% compared to the previous result, whether it is marked needs attention);
- Three-level review (key items): The director conducts spot checks on high-risk items (such as troponin and blood group identification) of out-of-control items, focusing on whether the out-of-control treatment is complete (whether there is cause analysis, corrective measures, and effect verification records) and whether it affects patient reports (such as whether patient specimens need to be re-tested during the out-of-control period).
The system needs to record in detail the time, personnel (associated work number), and review opinions at each level of review (if the data is correct, agree to publish; Levey-Jennings diagram drawing errors need to be corrected), forming a traceable chain of responsibility – even if problems arise six months later, the person responsible for each level of review can be quickly located.
3. Automatic alarm
Abnormalities in the inspection process are like invisible bombs: random errors can lead to anomalies in a single result, and systematic errors can cause continuous loss of control, which may affect the reports of one or even several batches of patients if not detected in time. The automatic alarm function is to defuse these bombs before they detonate.
3.1 Alarm rules
Alarm rules cannot be one-size-fits-all, not only to follow international standards, but also to fit the actual situation of the laboratory – after all, the accuracy of instruments in tertiary hospitals is different from that of community hospitals, and the fault tolerance rate of conventional and emergency projects is also different.
3.1.1 Westgard Multi-Rule
Westgard Multi-Rule is like a set of combos that can accurately identify different types of errors, and the system needs to fully integrate and clarify the applicable scenarios of each rule:
- 13s rule (a single result exceeds ±3s): Like sudden abnormalities, it is mostly caused by random errors (such as inaccurate reagent drop dosage during operation, insufficient mixing of specimens). Such alarms need to be prompted immediately (e.g., instrument interface pop-ups) as they may affect the patient specimen currently being tested;
- 22s rule (2 consecutive results beyond ±2s): more likely to be systematic errors (e.g., instrument calibration offset, reagent batch change). These alarms flag potential system issues and recommend checking instrument status (e.g., if recalibration is required);
- R4s rule (more than 4 seconds difference between the two levels of quality control results in the same batch): indicates that the testing process is unstable (such as sudden instrument failure, reagent deterioration). For example, the results of a batch of high-value and low-value blood glucose control products are 0mmol/L (+2.5s) and 3.0mmol/L (-2.0s), respectively, with a difference of 4.5s, triggering the R4s rule, and the system will automatically recommend suspending the instrument’s detection and troubleshooting.
3.1.2 Custom rules
General rules may not be suitable for all scenarios, and the system needs to support laboratories to formulate house rules according to their own situation:
- Seasonal adjustment: A blood station laboratory found that the quality control results of blood specimens fluctuated greatly in summer (high room temperature), so the alarm threshold of all blood quality control products was tightened to ±1.5s from June to August, which was stricter than the conventional standard.
- Equipment adjustment: A community hospital set the first 3 quality control results after the daily start-up and automatically triggered the 12s rule (± 2s is a warning) to prevent errors caused by insufficient equipment preheating in advance.
- Project adjustment: For the gold standard for myocardial infarction diagnosis such as myocardial troponin (cTnI), an alarm is set when a single result exceeds ± 1.8 seconds (stricter than the conventional ±2s), because a deviation of 01ng/mL may affect the diagnosis; For the specific gravity of urine in the urine routine, it is relaxed to ±3s to avoid excessive alarm interference with work.
3.2 Alarm form
If the police are ignored, it is equivalent to a white report. The system must choose the most appropriate notification method according to the personnel role and work scenario to ensure that those who should know know it as soon as possible.
3.2.1 Multi-channel alarm
The work scenarios of laboratory personnel are diverse (operating instruments, writing reports, and holding meetings), and the alarm style should also be adapted to local conditions and reminded in an all-round way:
- Technicians operating the instrument: a red pop-up window (including an immediate processing button) + beep sound (lasting 3 seconds, repeated at 5 second intervals), visual + auditory double reminders to ensure that the operating personnel will not be missed;
- Team leader who is writing a report: Push message in the lower right corner of the computer (with click to view the details link), and the taskbar icon continues to flash, which does not affect the current work but can be reminded in time;
- Director who goes out for a meeting: SMS + email is sent simultaneously, and the content must be concise and clear (such as the quality control of the coagulation project is out of control, instrument A, the 13s rule has been triggered, please pay attention), avoid lengthy information;
- Night shift personnel: If the alarm is triggered at night (such as 2 a.m.), the system will automatically call the mobile phone of the technician on duty (play the preset voice: emergency biochemical quality control is out of control, please deal with it immediately) to ensure that someone responds to the emergency.
3.2.2 Graded Notice
Alarms are also prioritized, and hierarchical notifications prevent information overload – if all alarms are notified to the director, it can lead to overwhelming of important alarms. The system needs to be graded by severity:
- Level 1 alarm (serious loss of control): such as 13s+22s triggered at the same time (indicating serious system error), R4s rule triggering (prompting the collapse of the detection process), notifying all relevant personnel (tester, team leader, director, and even the head nurse of the department, because it may affect patient reporting);
- Level 2 alarm (warning): If a single result exceeds +2s but does not reach 3s, and the inventory of quality control products is lower than the safety line, only the inspector and team leader will be notified, and they will judge whether it needs to be upgraded;
- Level 3 alarm (slight abnormality): If the status of the specimen is to be confirmed (such as the label is blurry but recognizable), the operator’s qualification is about to expire (such as a technician’s PCR operation certificate expires in 10 days), only notify the tester or the party to avoid disturbing others.
4. Quality control product management
Quality control products are the yardstick for quality inspection, and if the scale is inaccurate, no matter how accurate the results are, it is useless. Therefore, the management of quality control products must be meticulous, covering the entire life cycle from procurement and warehousing to destruction.
4.1 Information registration
Each batch of quality control products should have an ID card + physical examination report, and the system needs to record complete information to ensure that the source can be checked, the characteristics can be known, and the status can be controlled.
4.1.1 Basic information
- Basic attributes: name (such as high-value biochemical quality control products), manufacturer (such as Roche, Abbott), batch number (20250315), production date, expiration date (2026-03-14), specification (such as 100 servings/box);
- Project characteristics: The target value and acceptable range of each test item (e.g., creatinine target value 80 μmol/L, range 72-88 μmol/L; blood glucose target value 0mmol/L, range 4.5-5.5mmol/L);
- Storage requirements: storage temperature (such as 2-8°C refrigeration, -20°C freezing), expiration date after opening (such as 7 days), whether it needs to be protected from light (such as bilirubin control products need to be stored in a dark place), transportation conditions (such as cold chain transportation temperature range 2-8°C).
4.1.2 Traceability information
The traceability of quality control products is a core requirement of ISO15189, and the system needs to record in detail:
- Whether it has passed the CRM (Certified Reference Substance) calibration with a CRM number (e.g., GBW09154);
- Comparison data with international standards (e.g., deviation rate from WHO standards);
- The calibration certificate number and query link provided by the manufacturer (for easy verification by the regulatory authorities).
4.2 Inventory management
Controls shortages can lead to unstandard testing – laboratories are not allowed to publish patient reports without the support of quality control results, which can be life-threatening in emergency settings. Inventory management must be prepared for a rainy day.
4.2.1 Real-time monitoring
The system needs to update the inventory status in real time like a warehouse manager and visualize it to achieve inventory transparency:
- Quantity details: the remaining quantity of each batch of quality control products (such as batch number 20250315, remaining 12 boxes), used quantity, cumulative consumption;
- Status identification: Mark the inventory status with color – green sufficient (≥ 3 months’ supply), yellow warning (1-3 months supply), red shortage (< 1 month’s supply);
- Related information: Displays the last purchase time (e.g., 2025-01-10), arrival cycle (e.g., 7 days from order placement to warehousing), supplier contact information (phone, email, emergency contact).
4.2.2 Intelligent early warning
When the inventory falls below the safety line (the system can automatically calculate the historical usage, usually 1.5 times the average monthly usage), the system automatically triggers a multi-level warning:
- Level 1 Warning (Yellow): Send a restocking reminder email to the buyer, with historical purchase price (convenient price comparison) and usage trend chart in the past 6 months (to assist in judging the purchase quantity);
- Level 2 Warning (Orange): If the purchase order feedback is not received within 48 hours, the laboratory signboard (such as the electronic screen in the hall) will display the urgent need for purchase: blood lipid control products (batch number 20250315), and remind the team leader to follow up;
- Level 3 warning (red): If there are only 3 days left in the inventory and it has not arrived, automatically call the phone number of the buyer and laboratory director and play a voice reminder (if the blood lipid control product is about to be cut off, please deal with it immediately).
5. Design of quality control rules
Quality control rules are the soul of the LIS quality control module – it determines which anomalies the system can identify and how to judge whether the results are reliable. The design of the rules requires two legs to walk: one leg to step on the international standard, and the other leg to be rooted in the laboratory.
5.1 Anchoring international standards
ISO15189 is not a multiple-choice question, but a compulsory question. The system must be designed to align with standard requirements and translate clauses into executable functions.
5.1.1 Closely follow the core clauses of ISO15189
1) Standard 6.2 stipulates that the laboratory should determine the appropriate quality control frequency according to the characteristics of the inspection items. The system needs to support dynamic frequency setting according to items, instruments, and detection amounts:
- High-frequency items (such as blood routine, daily test volume >200 servings): 1 time per day;
- Medium frequency items (such as biochemical routine, 50-200 servings per day): 1 time per batch;
- Low-frequency items (such as trace elements, < 10 copies per week): Quality control is done synchronously during each test;
- Emergency items: do it once after each turn-on, and add 1 time every 4 hours.
2) Standard 6.3 requirements: When quality control is out of control, corrective measures should be taken and recorded. The system needs to forcibly associate the out-of-control processing record module, including three required fields for cause analysis, corrective action, effect verification, and failure to continue testing the item (similar to not allowing class if homework is not completed).
Following these standards is not only to pass certification, but also to ensure that there is a minimum guarantee of inspection quality – regardless of the size of the laboratory, the rules constrain the basic quality to ensure the basic quality.
5.2 Meet clinical needs
The clinical significance of different inspection items is very different, and quality control rules also need to be taught according to their aptitude to avoid excessive alarms or omissions caused by one-size-fits-all.
5.2.1 Adjust the rules according to the risk level of the project
- High-risk items (such as troponin, blood grouping, HIV antibodies): The results of such items directly affect diagnosis and treatment decisions, and need to be strict – the alarm threshold is tightened from ±2s to ±1.5s, and the patient testing needs to be suspended immediately after it is out of control until the problem is solved;
- Medium-risk items (such as liver function, kidney function): Perform according to the usual rules (±2s), and evaluate whether it affects the tested patient specimens after the loss of control (such as re-testing of the specimen within 1 hour before the loss of control);
- Low-risk items (such as urine specific gravity and color of urine routine): Clinically allow certain fluctuations, which can be relaxed – the alarm threshold can be relaxed to ±3s, and only recording and processing is required, and there is no need to suspend the test.
5.3 Expanding Westgard’s multi-rules
Westgard’s multi-rule is classic, but the ever-changing situation in the laboratory (e.g., contamination risk for molecular diagnostics, trend drift in long-term monitoring projects) requires flexible expansion to upgrade the rules from reactive identification of anomalies to proactive risk prevention.
5.3.1 Special expansion rules
- Molecular Diagnosis Special: New Contamination Monitoring Rules – When the amplification curve (Ct value <38) appears in the negative control, an alarm will be triggered regardless of whether other results are under control or not, and it will automatically prompt that there may be nucleic acid contamination, and environmental disinfection and pipette calibration are recommended;
- Long-term monitoring items: Increase trend rule – if the average value rises or falls by more than 5% for 6 consecutive months (even if a single result is under control), trigger a system drift warning, indicating that there may be problems such as reagent batch changes and instrument aging, and it is recommended to calibrate or replace the reagent;
- Inter-laboratory quality assessment related rules: When the inter-laboratory quality evaluation results of a project are unqualified, the indoor quality control rules of the project will be automatically tightened (such as from ± 2s to ±1.8s) for 3 months until the next inter-laboratory quality assessment is qualified.
6. Data statistical analysis
Quality control data is not a dead number, but a living clue to improve quality. The statistical analysis function of the system should be able to dig out gold mines from massive data and provide direction for quality improvement.
6.1 General statistics
The basic statistics are quality physical examination forms, which are generated regularly (such as daily, weekly, and monthly) so that laboratory personnel can quickly judge which items are of good quality and which ones need improvement.
6.1.1 Core statistical indicators
- Mean and SD (standard deviation): Reflects the stability of the assay. For example, the monthly average value of an item is 5.2mmol/L, SD0.3, indicating that the results are concentrated and the fluctuation is small. If the SD suddenly increases to 0.8, it indicates a decrease in stability.
- Coefficient of Variation (CV): Assess the precision (CV=SD/mean×100%). It is generally believed that CV<5% represents good precision, such as blood glucose detection CV=3% (better than 5%), indicating good detection repeatability;
- Pass rate: A project has been tested 100 times this month, 98 times under control, and the pass rate is 98%. Projects with a pass rate of less than 95% need to be listed as key improvement targets.
6.2 Trend analysis
A single quality control result is under control, but it does not mean that the overall quality is stable – long-term small drifts may gradually accumulate into systematic errors. Trend analysis is all about capturing this invisible problem.
6.2.1 Trend Identification Methods
The system needs to analyze long-term data using statistical methods (e.g., moving average, exponential smoothing method) to identify the following trends:
- Linear drift: For example, the monthly average value of ALT (palnine aminotransferase) gradually rises from 35 U/L to 42 U/L (3 consecutive months), even if the individual results are within ±2s, it indicates system drift (may be an instrument calibration offset);
- Periodic fluctuations: If the CV value of a project is 5% higher than other times every Monday, it is found that the instrument is idle on the weekend and restarted, resulting in insufficient warm-up, which can be solved by doing one more quality control after starting on Monday.
- Sudden rise and fall: If the SD value of a project suddenly rises from 2 to 0.7, it may be due to reagent batch replacement, instrument failure, etc., and needs to be investigated immediately.
6.3 Quality improvement
The ultimate goal of statistical analysis is to improve. The system needs to act like a quality consultant to provide the laboratory with specific improvement directions and effect evaluations.
6.3.1 Problem localization and improvement support
- Cause correlation: When a project is out of control many times, the system automatically correlates possible causes – such as the last 3 out-of-control occurrences when using batch number A reagent, it indicates that there may be a problem with the reagent; the loss rate of an instrument is 3 times that of other instruments, indicating that the instrument needs maintenance;
- Suggestions: Recommend improvement measures in SOP according to the type of problem (such as contact the manufacturer to replace the batch with the manufacturer’s after-sales phone number);
- Effect verification: After the implementation of the improvement measures, the system automatically compares the data before and after (e.g., CV is reduced from 7% to 3% after reagent replacement) to determine whether it is effective, and generates an improvement report (if the measures are effective, continuous monitoring is recommended).
This closed loop of analysis-improvement-evaluation allows the inspection quality to enter a virtuous circle of upward spiral.
Let the quality control module become a quality partner
The design of the quality control module of the LIS system should not only understand the strictness of ISO15189 (such as traceability requirements), but also understand the difficulty of laboratory operations (such as time pressure in the emergency department), and the urgency of clinical needs (such as the timeliness and accuracy of results).
From compiling a talking code for quality control specimens (including type, batch, and storage requirements) to using trend analysis to detect system drift in advance; From hierarchical alarm to ensure that key personnel are not missed, to inventory early warning to ensure the continuous supply of quality control products, behind every function is the awe of inspection quality.
In the future, with the integration of AI technology, the QC module will become smarter – for example, using machine learning to predict quality control anomalies (e.g., predicting that lot number B reagent may be out of control on the 15th day of use based on historical data), or automatically recommending optimal improvements (e.g., shortening the calibration period from 1 month to 2 weeks for the instrument can reduce the runaway rate by 60%). However, no matter how the technology develops, the original intention of the design with clinical needs as the core and patient safety as the goal will always be the root of the LIS quality control module.
For product managers, creating such a module requires cross-border thinking: they must understand the implementation logic of IT technology (such as database design and interface development), as well as the operation process of the laboratory (such as the reconstitution step of quality control products, instrument calibration cycle), and understand the clinical expectations of test results (such as the timeliness of emergency reports and the accuracy of tumor markers). Only in this way can the quality control module truly become a quality partner for laboratory personnel and escort the accuracy of each inspection report.
