Evaluation and application of automated quality control in clinical coagulation testing

Abstract

Background

Internal quality control (IQC) is indispensable for ensuring the accuracy of clinical laboratory testing. While fully automated track systems have enabled standardized IQC for clinical chemistry and hematology, coagulation assays—despite their diagnostic importance—still depend on manual handling. Variability during critical preparatory steps such as rewarming, thawing, mixing, and aliquoting of control materials often leads to QC instability and misinterpretation.

Objectives

The aim of this study was to evaluate the clinical applicability of automated internal quality control (auto-QC) detection using a fully automated sample processing system.

Methods

To analyze the performance of auto-QC in coagulation testing, the stability, detection efficiency, and QC completion time of the auto-QC and concurrently acquired manual QC (immediate reconstitution) methods were compared at the Department of Laboratory Medicine at Suining Central Hospital (October 2024 to February 2025) and at Ordos Central Hospital (November 2024 to February 2025).

Results

At Suining Central Hospital, auto-QC yielded lower coefficients of variation (CVs) than manual QC for activated partial thromboplastin time (aPTT), prothrombin time (PT), and multiple levels of D-dimer (DD), as well as low and high levels of fibrin degradation product (FDP), but had a higher CV for FDPs at the middle level (2.63% vs 2.46%). For thrombin time (TT), auto-QC had a lower CV at the normal level (1.00% vs 1.56%) but a higher CV at the abnormal level (1.96% vs 1.80%). For fibrinogen (FIB), auto-QC had a higher CV at the normal level (3.43% vs 3.14%) but a lower CV at the abnormal level (2.31% vs 2.56%). At Ordos Central Hospital, auto-QC yielded lower CVs than manual QC for APTT, TT, FIB, DD, and FDP at both levels, and for PT at the normal level; for PT at the abnormal level, auto-QC had a marginally higher CV (3.37% vs 3.26%). Wilcoxon signed-rank test showed significantly higher σ values for auto-QC (P < .05 for both centers). Regarding daily workflow, auto-QC was completed approximately 1.5 to 2.5 hours faster than manual QC. In routine coagulation testing, auto-QC reduced materials consumption by 16.5% to 38.1% compared with manual QC.

Conclusion

The fully automated sample processing system’s QC detection meets laboratory quality requirements, significantly shortens processing time, reduces workload, and improves work efficiency.

Essentials

• •Manual quality control of coagulation tests is time-consuming and labor-intensive, calling for optimized automatic detection modes for clinical laboratories.
• •This study compared stability, efficiency, and cost of automated and manual internal quality control for 6 coagulation parameters in 2 tertiary hospitals.
• •Automated quality control meets national laboratory standards with consistent detection trends and no significant difference in σ values vs manual detection.
• •The automated system significantly shortens quality control time and reduces reagent consumption, lowering workload and laboratory operation costs.

1. Introduction

Internal quality control (IQC) is an essential component of the quality control (QC) system in a clinical laboratory. Reliable evidence helps doctors provide clinical diagnosis and treatment, and ensuring the accuracy and reproducibility of laboratory specimen test results requires QC testing [1,2]. Recent advancements in laboratory medicine have catalyzed the adoption of fully automated track systems, which can significantly reduce human-induced errors, enhance test stability, alleviate laboratory personnel’s workload, and improve operational efficiency [[3], [4], [5]]. Currently, several clinical laboratories have integrated fully automated IQC tracks for clinical chemistry and hematology, achieving automation and standardization of the IQC process after manual loading of QC materials, thereby minimizing the impact of operator variability on QC results during IQC execution. Multiple studies have indicated that coagulation assays are a critical routine component of clinical laboratories [[6], [7], [8]], providing essential guidance for diagnosing coagulation disorders and monitoring anticoagulant therapy. However, the IQC for measurement of these analytes still faces challenges. Research has indicated that human-induced QC failures account for a considerable proportion of IQC process issues in coagulation-related tests, which reflects how manual variability during critical preparatory steps such as rewarming, thawing, mixing, aliquoting, recovery, and freezing of QC materials may significantly compromise the stability of QC testing results and interfere with their accurate interpretation [[9], [10], [11]]. Therefore, this study integrated a fully automated QC (auto-QC) sample processing system with advanced automated coagulation analyzers to establish a standardized, end-to-end IQC workflow. In addition, a comprehensive performance analysis of the auto-QC system was conducted to evaluate its advantages in terms of stability and operational efficiency, based on a comparison of concurrently acquired auto-QC and manual QC data at 2 hospital laboratories.

2. Methods

2.1. Materials

This study selected IQC data for coagulation testing from the Clinical Laboratory Department of Suining Central Hospital (Eastern Campus) from October 2024 to February 2025, and from Ordos Central Hospital (Kangbashi District) from November 2024 to February 2025.

2.1.1. Equipment

The Mindray AQC-100 fully automated sample processing system included the Mindray CX-9000 fully automated coagulation analyzer along with its matching QC materials and reagents. Calibrated pipettes with volumes ranging from 100 to 1000 μL were used.

2.1.2. Reagents

Activated partial thromboplastin time (aPTT), prothrombin time (PT), thrombin time (TT), and fibrinogen (FIB) coagulation assay kits, DD assay kits, and fibrin degradation product (FDP) assay kits, along with their corresponding calibrators and QC materials, were purchased from Shenzhen Mindray Bio-Medical Electronics Co, Ltd.

2.1.3. QC materials

Coagulation QC materials at levels 1 and 2 (LOT: 8624070118, 8724070118), D-dimer QC materials at low, moderate, and high levels (LOT: 6224060116, 6324060116, 6424060116), and FDP QC materials at low, moderate, and high levels (LOT: 6624060116, 6724060116, 6824060116) were used. Lot numbers were verified daily, and no lot changes occurred during the study period.

2.2. Methods

This study used identical dry powder QC reagents for both automated and manual processing. All reagents were reconstituted and processed following identical protocols in strict accordance with the manufacturer’s instructions for at least 30 consecutive days.

2.2.1. Auto-QC procedures

QC materials were loaded into the AQC-100 fully automated sample processing module on the afternoon prior to use. Each day, the testing system automatically initiated a start-up sequence, entering standby mode after completing a power-on self-test. Subsequently, the sample processing system retrieved the corresponding QC materials and dissolving solutions from the instrument’s integrated refrigerator and transported them via a rail system to the designated instrument modules. After automated rewarming, the system added the dissolving solution to fully dissolve and mix the QC materials, enabling automated testing. After aliquoting, the QC materials were transferred via the rail system to the capping unit for capping, sealed, and subsequently returned to the sample storage refrigerator.

2.2.2. Manual QC procedures

The traditional manual QC preparation protocol was performed daily upon arrival at work, with the specific steps shown as follows: QC materials were removed from storage at 2 °C to 8 °C and allowed to equilibrate to room temperature for 15 minutes. Subsequently, 0.5 mL purified water was aspirated and dispensed into the QC vial. The vial was then sealed vertically and incubated undisturbed at room temperature for 20 minutes. Following homogenization, the vial was placed on the dedicated QC rack and introduced into the analyzer for QC material testing.

2.2.3. Within-run precision assessment

To assess within-run precision, 2 levels (normal and abnormal) of QC materials and fresh patient plasma samples were tested. Each sample was continuously repeated 10 times within a single run. The mean, SD, and CV were calculated for each analyte. Acceptable CV criteria were based on industry standards (WS/T 406-2024 for coagulation assays, WS/T 477-2015 for D-dimer assay, and one-fourth or one-third of the total error from the 2025 National Center for Clinical Laboratories external quality assessment for FDP assay).

2.2.4. Between-day stability assessment

According to the standard operating protocol at Suining Central Hospital, each QC level was run once daily at 7:30 am after startup and once at 4:00 pm before shutdown on weekdays. In accordance with the operating procedures of Ordos Central Hospital, level-specific QC is conducted at 7:30 am on each working day. (1) To compare variability between the automated and manual methods, the mean and CV were calculated for each coagulation parameter under both operational modes to determine whether the variability of both control methods met the laboratory’s established acceptance criteria. (2) Bias data were derived from the proficiency testing results reported by the National Center for Clinical Laboratories for the 2025 evaluation period. For this study, 2 to 3 corresponding bias values were obtained based on 2 to 3 concentrations of each test item in this study. For each test item, the bias result from the proficiency testing concentration closest to the study’s median level was selected as the final evaluation datum. The overall analytical performance sigma (σ) value was then calculated based on this data.

2.2.5. Statistical analysis of QC operational failure rates

The failure of QC during instrument operation can be attributed to various factors, including human error, issues with reagents, equipment, or software. We conducted a statistical analysis of the rates and root causes of these failures in the auto-QC system.

2.2.6. Evaluation of daily QC completion times

Automated and manual QC completion times were continuously recorded for 30 days. Based on completion times, the QC duration under automated and manual modes was statistically compared and plotted as corresponding line graphs.

2.2.7. Evaluation of QC material consumption

Accurate calculation of QC material consumption is a key component of laboratory quality management that can directly affect cost control, inventory management, and testing stability. In this study, single-use consumption was calculated using the formula: Single-use consumption = 1 mL of control reagent/number of tests. Daily consumption was calculated using the formula: Daily consumption = number of daily tests × single-use consumption.

2.3. Statistical analysis

Excel 2016 was used to calculate the mean, SD, CV, and total analytical performance σ value for each item. The σ value was calculated using the formula: σ = (TEa − |Bias|) / CV, where TEa is acceptable total error. The R ggplot package was employed to plot the z-score graph. Performance at the 6σ level was analyzed. Statistical differences in σ values between manual QC and auto-QC processing for each parameter were evaluated using the Wilcoxon signed-rank test for paired samples. Two-tailed significance tests were performed with a significance level (α) set at 0.05; differences between the 2 QC modes were considered statistically significant at P < .05.

3. Results

3.1. Stability assessment of auto-QC and manual QC

3.1.1. Comparison of CVs in QC results

The within-run precision results are shown in Table 1. The CVs for both QC materials and patient samples met the respective industry standards. The CVs for QC materials ranged from 0.33% to 1.55% at normal and abnormal levels, while those for patient samples ranged from 0.76% to 9.24%. All CVs were within the allowable limits (WS/T 406-2024 for APTT, PT, TT, and FIB; WS/T 477-2015 for DD; and one-fourth or one-third of the total error from the 2025 National Center for Clinical Laboratories external quality assessment for FDP), indicating good within-run precision of the testing system.

Table 1.

Within-run precision of coagulation assays.

Item	QC level	n	Mean ± SD	CV (%)	Patient sample	n	Mean ± SD	CV (%)	Allowable CV
aPTT (s)	N	10	29.11 ± 0.10	0.33	Normal	10	29.99 ± 0.25	0.82	≤4.0%
	P	10	54.15 ± 0.62	1.14	Pathologic	10	53.86 ± 1.79	3.32	≤8.0%
PT (s)	N	10	11.70 ± 0.08	0.67	Normal	10	10.47 ± 0.12	1.09	≤3.0%
	P	10	35.34 ± 0.32	0.91	Pathologic	10	22.37 ± 0.17	0.76	≤8.0%
TT (s)	N	10	15.97 ± 0.10	0.62	Normal	10	14.95 ± 0.17	1.14	≤6.0%
	P	10	36.46 ± 0.47	1.29	Pathologic	10	17.61 ± 0.27	1.55	≤8.0%
FIB (g/L)	N	10	2.51 ± 0.03	1.26	Normal	10	3.01 ± 0.05	1.52	≤6.0%
	P	10	1.22 ± 0.02	1.55	Pathologic	10	3.14 ± 0.11	3.52	≤12.0%
DD (μg/mL)	L	10	0.72 ± 0.01	1.10	Normal	10	0.42 ± 0.02	4.15	≤15.0%
	M	10	2.15 ± 0.01	0.65	Pathologic	10	6.59 ± 0.05	0.82	≤10.0%
FDP (μg/mL)	L	10	7.93 ± 0.14	1.70	Normal	10	1.32 ± 0.14	9.24	≤11.67%
	M	10	20.46 ± 0.34	1.65	Pathologic	10	27.21 ± 0.49	1.79	≤8.75%

aPTT, activated partial thromboplastin time; CV, coefficient of variation; DD, D-dimer; FDP, fibrinogen degradation product; FIB, fibrinogen; L, low level; M, medium level; N, normal level; P, pathologic level; PT, prothrombin time; QC, quality control; TT, thrombin time.

3.1.2. Between-day stability assessment

As shown in Table 2 (Suining Central Hospital Data), the CVs for all coagulation parameters complied with relevant industry standards. For APTT, PT, and DD, auto-QC yielded lower CVs than manual QC at multiple levels. For TT, auto-QC had a lower CV at the normal (N) level (1.00% vs 1.56%) but a higher CV at the pathologic (P) level (1.96% vs 1.80%). For FIB, auto-QC had a higher CV at the N level (3.43% vs 3.14%) but a lower CV at the P level (2.31% vs 2.56%). For FDP, auto-QC had lower CVs at low and high levels (2.59% and 3.73% vs 2.83% and 3.77%, respectively) but a higher CV at the medium level (2.63% vs 2.46%).

Table 2.

Between-day CV and σ values at Suining Central Hospital.

Item	QC level	n	TEa%	Bias%	Automated QC			Manual QC			Allowable CV
					Mean ± SD	CV (%)	σ	Mean ± SD	CV (%)	σ
aPTT (s)	N	66	15	0.63	31.11 ± 0.82	2.63	5.46	30.73 ± 0.84	2.74	5.24	≤6.5%
	P	66	15	0.80	55.88 ± 1.00	1.79	7.93	55.34 ± 1.22	2.21	6.43	≤10.0%
PT (s)	N	66	15	0.92	11.36 ± 0.21	1.88	7.49	11.23 ± 0.22	1.93	7.30	≤6.5%
	P	66	15	4.50	33.84 ± 0.90	2.66	3.95	32.21 ± 0.96	2.98	3.52	≤10.0%
TT (s)	N	66	20	0.00	15.61 ± 0.16	1.00	20.00	15.66 ± 0.24	1.56	12.82	≤10.0%
	P	66	20	0.38	36.66 ± 0.72	1.96	10.01	37.35 ± 0.67	1.80	10.90	≤12.0%
FIB (g/L)	N	66	20	2.28	2.87 ± 0.10	3.43	5.17	2.90 ± 0.09	3.14	5.64	≤9.0%
	P	66	20	0.12	1.41 ± 0.03	2.31	8.61	1.44 ± 0.04	2.56	7.77	≤12.0%
DD (μg/mL)	L	66	50	5.81	0.83 ± 0.02	2.40	18.41	0.87 ± 0.04	4.11	10.75	≤15.0%
	M	66	50	0.72	2.25 ± 0.03	1.19	41.41	2.35 ± 0.04	1.73	28.49	≤15.0%
	H	66	50	0.64	3.75 ± 0.04	1.09	45.28	4.04 ± 0.11	2.65	18.63	≤15.0%
FDP (μg/mL)	L	66	35	0.47	8.15 ± 0.21	2.59	13.33	8.50 ± 0.24	2.83	12.20	≤8.75%
	M	66	35	6.65	20.95 ± 0.55	2.63	10.78	21.05 ± 0.52	2.46	11.52	≤8.75%
	H	66	35	2.88	40.28 ± 1.50	3.73	8.61	40.26 ± 1.52	3.77	8.52	≤8.75%

Total error (TE) was derived from the total error external quality assessment of the 2025 National Center for Clinical Laboratories.

aPTT, activated partial thromboplastin time; CV, coefficient of variation; DD, D-dimer; FDP, fibrinogen degradation product; FIB, fibrinogen; H, high level; L, low level; M, medium level; N, normal level; P, pathologic level; PT, prothrombin time; QC, quality control; TEa, allowable total error; TT, thrombin time.

As shown in Table 3 (Ordos Central Hospital Data), all CVs complied with industry standards. Auto-QC showed lower CVs for APTT, TT, FIB, DD, and FDP at both levels, and PT at the N level. For PT at the P level, auto-QC had a marginally higher CV (3.37% vs 3.26%).

Table 3.

Between-day CV and σ values at Ordos Central Hospital.

Item	QC level	n	TEa%	Bias%	Automated QC			Manual QC			Allowable CV
					Mean ± SD	CV (%)	σ	Mean ± SD	CV (%)	σ
aPTT (s)	N	30	15	0.63	29.46 ± 0.49	1.64	8.76	28.89 ± 1.42	4.91	2.93	≤6.5%
	P	30	15	0.80	53.09 ± 0.88	1.66	8.55	54.55 ± 1.52	2.78	5.11	≤10.0%
PT (s)	N	30	15	0.92	13.08 ± 0.57	4.36	3.23	11.96 ± 0.61	5.08	2.77	≤6.5%
	P	30	15	4.50	38.77 ± 1.31	3.37	3.12	41.22 ± 1.35	3.26	3.22	≤10.0%
TT (s)	N	30	20	0.00	15.49 ± 0.18	1.15	17.39	15.49 ± 0.22	1.43	13.99	≤10.0%
	P	30	20	0.38	39.36 ± 0.43	1.10	17.84	38.05 ± 1.03	2.72	7.21	≤12.0%
FIB (g/L)	N	30	20	2.28	2.92 ± 0.06	2.09	8.48	2.83 ± 0.12	4.22	4.20	≤9.0%
	P	30	20	0.12	1.43 ± 0.04	2.77	7.18	1.36 ± 0.06	4.58	4.34	≤12.0%
DD (μg/mL)	L	30	50	5.81	0.73 ± 0.01	1.92	23.02	0.70 ± 0.02	2.88	15.34	≤15.0%
	M	30	50	0.72	2.28 ± 0.02	0.98	50.29	2.09 ± 0.07	3.20	15.40	≤15.0%
FDP (μg/mL)	L	30	35	0.47	7.14 ± 0.16	2.22	15.55	6.90 ± 0.28	4.09	8.44	≤8.75%
	M	30	35	6.65	18.77 ± 0.42	2.25	12.60	17.83 ± 0.78	4.37	6.49	≤8.75%

Total error (TE) was derived from the total error external quality assessment of the 2025 National Center for Clinical Laboratories.

aPTT, activated partial thromboplastin time; CV, coefficient of variation; DD, D-dimer; FDP, fibrinogen degradation product; FIB, fibrinogen; L, low level; M, medium level; N, normal level; P, pathologic level; PT, prothrombin time; QC, quality control; TEa, allowable total error; TT, thrombin time.

A Wilcoxon signed-rank test (z = 2.062, P = .04 for Suining; z = 2.981, P = .003 for Ordos) demonstrated significant differences in σ values between auto-QC and manual QC for coagulation assay performance.

3.1.3. Comparison of z-score trend plots between auto-QC and manual QC

A z-score graph was plotted using coagulation-related testing data from Suining Central Hospital. Data points representing QC failures or out-of-control events attributable to equipment or human factors were excluded from the analysis. Finally, the results showed consistent overall trends between auto-QC and manual QC (as shown in Figure 1), without any instances of QC drift.

Figure 1.

Z-score graph of quality control results in FIB and FDP automated quality control and manual quality control modes. (A) Z-score results for FIB parameters; (B) Z-score results for FDP parameters. FIB, fibrinogen; FDP, fibrin degradation product; H, high; L, low; M, medium; N, normal; P, pathologic.

3.2. Operational failure rate of auto-QC and cause analysis

Based on QC data from Suining Central Hospital, 25 of the initial 66 auto-QC runs failed during the initial phase, resulting in a failure rate of 37.88%. The primary causes of failure were as follows: (1) Equipment factors (n = 8): Abnormal x/y-axis movement of the cart mechanism or lifting mechanism during scheduling; abnormal sample rack status on the cart; failure to detect sample racks on the cart after removal; abnormal syringe operation. (2) Software issues (n = 2): Abnormal startup or failure of the sample rack number card reader. (3) Reagent factors (n = 13): Insufficient QC buffer solution. (4) Human factors (n = 2): Unavailability of sample racks in the buffer zone and delayed replenishment of reagent sample racks.

Based on QC data from Ordos Central Hospital, 21 of the initial 66 auto-QC runs failed during the initial phase, resulting in a failure rate of 31.82%. The primary causes of failure were as follows: (1) Equipment factors (n = 8): Abnormal x/y-axis movement of the cart mechanism or lifting mechanism during scheduling; reagent loading mechanism malfunction. (2) Software issues (n = 2): Network disconnection between the instrument and computer and failure of initiation of reagent scheduling agent. (3) Reagent factors (n = 8): Insufficient sewing needle cleaning solution, cleaning solution, and complex solution. (4) Human factors (n = 3): Measurement startup failure because of insufficient sewing needle cleaning solution and reactive cups.

3.3. Comparison of completion times for auto-QC and manual QC

The daily completion times for auto-QC and manual QC modes over 30 consecutive days were analyzed based on QC data from Ordos Central Hospital. For auto-QC, the instrument was programmed to initiate a daily start-up sequence at 7:00 am, with each QC batch allocated a fixed duration of 30 minutes. For manual QC, operations were performed after staff arrived on workdays and lasted ∼1 hour. As shown in Figure 2, the daily auto-QC completion times were 1.5 to 2.5 hours faster than those for manual automated control after scheduling the QC function in the AQC module. Sample testing could proceed immediately upon QC acceptance, ensuring both the timeliness of QC and result reporting.

Figure 2.

Daily quality control completion time statistics for automated and manual quality control modes. (A) Completion time of quality control testing at Ordos Central Hospital; (B) Completion time of quality control testing at Suining Central Hospital.

3.4. Comparison of QC material consumption between auto-QC and manual QC

As shown in Table 4, auto-QC reduced average daily per-level QC material consumption by 17.7% (Suining) and 16.5% (Ordos) for the 4 coagulation assays (APTT, PT, TT, and FIB); by 34.7% and 38.1% for the DD assay; by 25.0% and 27.6% for the FDP assay. The slight between-center differences were attributable to different daily QC frequencies, which did not affect the overall conclusion of consumption reduction.

Table 4.

Statistics of quality control material consumption.

QC assay	Coag4		DD		FDP
	Suining	Ordos	Suining	Ordos	Suining	Ordos
Aliquot volume per QC test (μL)	6, 40, 27, 27a	same	10	10	5	5
Average daily consumption of automated quality control (μL)	270	167	96	70	75	55
Average daily consumption of manual quality control (μL)	328	213	147	113	100	76
Average daily reduction (μL)	58	33	51	43	25	21
Reduction percentage (%)	17.7	16.5	34.7	38.1	25	27.6

DD, D-dimer; FDP, fibrinogen degradation product; QC, quality control.

^aCoag4 includes 4 separate assays with different aliquot volumes per test (activated partial thromboplastin time: 6 μL, prothrombin time: 40 μL, thrombin time: 27 μL, fibrinogen: 27 μL).

4. Discussion

Routine coagulation testing is a widely used clinical screening test for thrombosis and hemostasis and plays a crucial role in diagnosing and treating diseases. However, coagulation reagents are inherently sensitive to external variables, including transportation conditions, storage temperatures, and preparation protocols. Therefore, rigorous IQC is vital for monitoring the analytical accuracy and clinical reliability of these results [1,9]. This study, jointly conducted by Suining Central Hospital and Ordos Central Hospital, utilized the AQC-100 fully automated QC sample processing system to evaluate the effectiveness of automated and traditional manual QC in coagulation testing. The implementation of such automated systems can provide higher-quality testing services to clinicians and enhance patient trust in physicians, making significant contributions to the advancement of overall healthcare quality [12,13].

The present study results indicated that the auto-QC system offers better σ performance than manual QC, while also demonstrating advantages in working efficiency and cost savings, as detailed below.

First, regarding the CV analysis, APTT, PT, TT, and FIB were evaluated using QC materials at 2 distinct concentration levels for QC. In contrast, DD and FDP assessments used QC materials at 3 concentration levels. Based on data from Suining Central Hospital, the CV for each analytical QC material test result was calculated to assess imprecision. The 7 coagulation assays showed that 3 (APTT, PT, and DD) had higher precision in multilevel auto-QC compared with manual QC, with the CV for auto-QC at N level in TT, at P level in FIB, and at low and high levels in FDP, all lower than those of manual QC. This discrepancy may stem from the use of dry powder in QC, given that reagent reconstitution is susceptible to operator technique variations and laboratory environmental fluctuations during traditional manual QC, leading to significant variability in results. However, auto-QC minimizes human-induced errors, yielding clear advantages in the stability of QC results, especially in the CV. The z-score QC graph revealed consistent overall trends between automated and manual QC results. The σ values of IQC in coagulation-related assays were compared in this study, as σ values account for both bias and imprecision, with CV applied as the evaluation index [14]. In contrast, the σ values (which incorporate both bias and imprecision) showed a statistically significant improvement with auto-QC compared with manual QC (Wilcoxon signed-rank test, P < .05), indicating that auto-QC achieves better analytical stability under routine operating conditions. CV alone is insufficient for a complete QC strategy; bias and lot-to-lot consistency must also be considered. This is particularly important for coagulation assays, where PT, APTT, FIB, and DD each face distinct preanalytical challenges. Our σ metrics, incorporating both bias and CV, provide a more comprehensive assessment.

Second, in terms of working efficiency, auto-QC streamlines the workflow by eliminating numerous steps associated with QC material aliquoting, loading, and retrieval compared with manual QC. In addition, its scheduling function allows QC to run before staff arrive, significantly improving turnaround time. Specifically, the average daily sample testing time in this study was advanced by 20 minutes, and the daily QC time was reduced by 1.5 to 2.5 hours, thereby significantly improving testing accuracy by eliminating operational variability. This efficiency optimization shortened the laboratory’s turnaround time [[15], [16], [17], [18]], providing more timely testing support for clinical diagnosis and treatment and enhancing the responsiveness of medical services.

Third, in terms of costs, the study results showed that auto-QC offers superior cost-efficiency than manual QC since it significantly reduced QC material consumption, consistent with previous study results [16]. The consumption of QC materials is related to the postreconstitution stability and the frequency of QC runs, with higher testing frequency leading to lower per test waste within the valid dosage range of a reconstituted QC vial, which is particularly pronounced in manual QC, where manual operations are prone to human errors that can cause result in deviations, necessitating repeat testing that not only wastes reagents but also prolongs the testing cycle. In contrast, auto-QC can effectively mitigate the issue by precisely dispensing reagents and samples at scheduled intervals with intelligent management, reducing QC material waste, conserving labor resources, lowering personnel costs, and significantly reducing laboratory operating costs.

An analysis of system malfunctions during the study period revealed that the primary drivers of auto-QC failures were mechanical and logistical. These included lift mechanism malfunctions during cart scheduling, improper positioning of the sample rack, and reagent depletion for QC reconstitution. Collectively, these technical issues resulted in incomplete datasets or a significant reduction in expected QC results. This analysis of failure modes highlights the need for regular maintenance, hardware adjustments, and staff training to minimize equipment-related failures in future applications of the auto-QC system.

We acknowledge several important limitations. First, the automated workflow used materials loaded the afternoon prior, whereas manual QC used fresh reconstitution on the day of the test. Our comparison evaluated a bundled workflow (automation + overnight storage of dry vials) rather than automation alone. Second, the AQC-100 and CX-9000 systems are not widely used outside China, limiting the international generalizability of our findings; however, the AQC-100 shares core features common to most auto-QC platforms (refrigerated storage, automated reconstitution and mixing, precise aliquoting, and scheduling), so the reported performance metrics (CV, σ, time savings, and materials consumption reduction) reflect general benefits of automating QC workflows rather than platform-specific characteristics and can serve as benchmarks for validation of similar auto-QC systems regardless of manufacturer—laboratories using other robotic or track-based systems may achieve comparable improvements under similar operational parameters. Third, FDP is no longer a common test in many international laboratories; however, it remains clinically relevant in our setting for disseminated intravascular coagulation and fibrinolysis assessment. Because each analyte is evaluated independently using its own QC materials, excluding FDP does not affect conclusions for other assays. Fourth, the initial high failure rate of auto-QC (37.9%) underscores that automation introduces new failure modes (mechanical, software) that require vigilant maintenance and training [19]. Fifth, although within-run precision using patient samples was acceptable, this does not replace a formal method comparison study between the 2 QC workflows. Although we demonstrated analytical stability, this study did not directly compare patient sample results between auto-QC and manual QC workflows.

In conclusion, this dual-center evaluation demonstrates that fully automated QC achieves significantly higher σ metrics for coagulation assays compared with manual QC (P < .05), reflecting improved analytical stability under routine operating conditions. However, the initial failure rate of 37.9% underscores the need for robust maintenance and staff training. Auto-QC also offers meaningful workflow efficiency gains (1.5-2.5 hours saved per day) and modest cost savings (16.5%-38.1% reduction in QC material consumption). The auto-QC system meets laboratory quality requirements for QC materials, but direct patient sample validation across normal and abnormal ranges remains the next necessary step before claiming clinical equivalence. Notably, the application of auto-QC also requires robust software and compatible hardware to ensure the automated, standardized management of the QC process, which is essential for fully automated operations [20]. Systemic failures or technical malfunctions within these automated platforms can compromise the integrity of analytical results [19]. Therefore, it is necessary to conduct a comprehensive assessment of relevant reagents, environmental factors, and hardware systems before implementing auto-QC to ensure stability and consistency across all stages, thus enhancing overall efficiency and accuracy. This application assessment provides deeper insights into auto-QC, offering new perspectives and references for the intelligent management of QC in clinical laboratories.

Data availability

The data used to support the findings of this study are available from the corresponding author upon request.