Quality and consistency of genetic newborn screening reports in China: insights from a multi-laboratory analysis using simulated samples Authors: Jiaming Wu, Penghui Feng, Jinming Zhang, Honghui Zhou, Weijie Tian, Xingtong Chen, Chongyi Hao, Falin He Published: 2026 ABSTRACT AbstractBackgroundChina currently lacks unified reporting standards for genetic newborn screening (gNBS), resulting in substantial variation in report content, structure, and variant interpretation across laboratories. This study aims to evaluate the quality of gNBS reports nationwide and provide practical evidence to support the development of standardized reporting guidelines.MethodsThrough an external quality assessment (EQA) program organized by the National Center for Clinical Laboratories (NCCL), we collected reports generated by 34 laboratories for three bioinformatically simulated samples. A structured questionnaire, panel comparison, and content evaluation were employed to systematically assess differences in gene panel composition, consistency in variant detection and classification, approaches to distinguishing primary and secondary findings, completeness of key report elements, and the scope of genetic counseling recommendations.ResultsThe number of genes included in screening panels varied substantially across laboratories, with only about one-third demonstrating high inter-laboratory similarity. Most laboratories achieved good consistency in detecting primary variants and classifying their pathogenicity, whereas considerable discrepancies were observed in the reporting of secondary findings, evidence grading, and specific descriptive details. Some reports lacked essential information, including testing purpose, laboratory signature, and submitting institution. Although most laboratories provided genetic counseling recommendations, these suggestions were generally not comprehensive.ConclusionThe gNBS reports in China exhibit a certain degree of consistency in presenting core information, yet further standardization is needed to improve report structure and completeness. This study provides practical support for developing unified reporting standards and highlights areas where observed variability may guide mutual learning and quality improvement across laboratories.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12887-026-07023-x.Keywords: Genetic testing, Genetic counseling, Newborn screening, External quality assessment. Introduction Both the Global Health Strategy for Women, Children and Adolescents (2016-30) and Healthy China 2030 highlight the goal of promoting the health of children through equitable access, early prevention, and sustainable development [1, 2]. Over the past seven decades, China has made remarkable progress in reproductive, maternal, newborn, child, and adolescent health (RMNCAH), leading to a substantial reduction in neonatal mortality [3]. As the RMNCAH agenda continues to advance, improving birth quality and preventing rare genetic diseases have become new public health priorities. Newborn screening (NBS) is a key component of the “three-stage prevention strategy”, enabling early detection and intervention to reduce mortality and disease burden [4, 5]. NBS originated in the 1960s with the introduction of dried blood spot (DBS) testing for phenylketonuria, which established the technical foundation for expanding programs to multiple disorders [6]. With continuous advances in technology, modern NBS laboratories employ diverse analytical platforms, including immunoassays, tandem mass spectrometry, PCR-based assays, DNA sequencing for variant detection [7]. Currently, next-generation sequencing (NGS) has rapidly emerged as a powerful complement to traditional biochemical approaches. Owing to its high throughput, accuracy, and decreasing cost, NGS has been widely adopted for diagnosing rare genetic disorders and reveals great promise for broader integration into genomic newborn screening (gNBS) programs [8, 9]. Internationally, organizations such as the American College of Medical Genetics and Genomics (ACMG) [10], the European Society of Human Genetics (ESHG) [11], and the Association for Clinical Genomic Science (ACGS) [12] have issued guidelines that standardize genetic testing reports, specifying requirements for testing procedures, result interpretation, and report content. However, these guidelines vary across regions due to differences in healthcare systems and professional frameworks. In China, laboratories primarily follow domestic expert consensus statements [13] and ACMG technical standards [14], but a universally accepted national guideline for NBS report format and content is still lacking. Therefore, laboratories often design their own report templates based on individual experience and interpretation, leading to substantial variation in structure, data presentation, and interpretation. Inconsistencies in analysis pipelines, variant annotation databases, and interpretive criteria further exacerbate this variability, posing challenges to clinical application and genetic counseling [15]. With the advancement of precision medicine and increasing public awareness of genomics, the concept of patient-centered reporting has gained growing recognition. There is a rising expectation that genetic testing results should be understandable not only to patients but also to non-specialist healthcare providers [16]. Although many laboratories in China have launched genetic screening programs, systematic evaluations of report quality, content completeness, and consistency are still lacking, leaving the current reporting landscape insufficiently characterized. To narrow this gap, we conducted a nationwide external quality assessment (EQA) project that collected NBS reports from 34 laboratories, each analyzing the same set of simulated samples. Through comparative and content analyses, this study systematically evaluated inter-laboratory variations and reports features, aiming to establish a practical framework for report evaluation and to provide empirical evidence supporting the development of unified national reporting standards. Materials and methods Project workflow In July 2025, the National Center for Clinical Laboratories (NCCL) launched a nationwide EQA program focusing on bioinformatics analysis of gNBS reports. The participating laboratories were required to meet the following inclusion criteria: (1) possessing the appropriate legal qualifications and actively performing genetic testing related to newborn screening; (2) maintaining a stable testing workflow capable of providing complete technical and reporting information; and (3) having previously participated in other genetic testing EQA programs organized by the NCCL, which allowed us to have a baseline understanding of their technical capacity and quality management systems. Ultimately, a total of 34 laboratories participated (coded L01–L34), including 13 public hospitals and 21 third-party testing institutions. The overall project workflow was illustrated in Fig. 1. The EQA test materials consisted of human genomic DNA sequencing data (reference genome GRCh37/hg19) provided in FastQ format. Raw sequencing data were distributed via Baidu Cloud and included three simulated samples (Samples 1–3). Each sample was accompanied by a brief description of relevant clinical information (see Supplementary Table S1). Participating laboratories were first required to complete a survey questionnaire, followed by generating reports based on their respective bioinformatics pipelines, and finally submitting their current gNBS panels. Subsequently, we conducted a comprehensive comparative analysis of these results and released the target-variant details and survey summary two months after receiving all submissions. Overview of the study workflow Selection of target variants The simulated variants (see Supplementary Table S2) were derived from clinically significant genes recommended by national expert consensus documents [13]. Selection was guided by the following criteria: (1) documented carrier frequency and pathogenicity in the Chinese population; (2) gene relevance to commonly screened disorders such as inborn errors of metabolism and hereditary deafness; (3) inclusion of diverse variant types (e.g., missense, nonsense, deletions, and duplications) to evaluate laboratories’ detection capabilities across mutation categories; and (4) preference for regions reliably covered by standard high-throughput sequencing platforms to ensure technical feasibility and comparability across laboratories. Variant simulation This study did not involve actual human or animal subjects; therefore, although it relates to newborn screening, no parental or guardian consent was required. The study was performed in accordance with relevant guidelines and regulations and was approved by the Ethics Committee of the Chinese PLA General Hospital (approval number: S2022-693-01). Three simulated samples were generated from two reference genomes: Samples 1 and 2 from the female NA12892 genome, and Sample 3 from the male NA12891 genome. Selected variants were then bioinformatically introduced into each sample on the NCCL Linux server. File transfer and remote operations were conducted using Xshell 8 and Xftp 8. Variants were inserted in batch mode via shell scripts executed in the Bash environment to ensure reproducibility and process consistency. Following variant introduction, the modified sequences were visually inspected in Integrative Genomics Viewer (IGV, version 2.19.1) to confirm the correct insertion and representation of all target variants. Statistical analysis of laboratory reported-results All statistical analyses and visualizations were performed using R (R version 4.5.1) and Microsoft Excel. A structured questionnaire was designed to systematically collect key information on each laboratory’s core procedures (Tables 1 and 2). Additionally, the number of genes included in each laboratory’s screening panel was compiled and verified (Supplementary Figure S1). To assess the similarity among panels, the Jaccard similarity index was calculated for each pair of gene lists (Fig. 2). This index is defined as the number of shared genes divided by the total number of unique genes in the two sets, ranging from 0 to 1, with higher values indicating greater overlap. Table 1Survey results on testing methods and other basic information from 34 laboratoriesInvestigated ItemNumber of Laboratories (%)Investigated ItemNumber of Laboratories (%)Laboratory typeLibrary preparation strategy* Public hospitals13(38.2)Ultrasonic fragmentation → adapter ligation → PCR enrichment → hybrid capture;3(8.8) Third-party institutions21(61.8)Enzymatic fragmentation → adapter ligation → PCR enrichment → hybrid capture;31(91.2)Human genome referenceMultiplex PCR library preparation4(11.8) GRCh37/hg1930(88.2)Capture kit manufacturer GRCh38/hg384(11.8)Nanodigmbio13(38.2)Testing method*Integrated DNA Technologies4(11.8) Panel30(88.2)Twist Bioscience3(8.8) WES10(29.4)iGeneTech3(8.8) WGS1(2.9)Beijing Fujun3(8.8)Specimen type*Magigene2(5.9) Peripheral blood33(97.1)Agilent Technologies1(2.9) Dried blood spot29(85.2)Hunan GeneMatrix1(2.9) Buccal epithelial cells9(26.5)BGI Genomics1(2.9) Saliva3(8.8)Berry Genomics (Fujian)1(2.9)Turnaround time (business days)Guangzhou Yongnuo1(2.9) 72(5.9)In-house (Guangzhou Sheng’an)1(2.9) 1017(50.0) 1510(29.4)Sequencing platform 205(14.7)MGISEQ-200010(29.4)Age restrictionMGISEQ-2002(5.9) None16(47.1)Novaseq 60006(17.6) < 18 years7(20.6)DNBSEQ-T76(17.6) < 14 years1(2.9)MGI T73(8.8) < 12 years3(8.8)Novaseq X Plus1(2.9) < 12 months4(11.8)DNBSEQ-T7RS1(2.9) < 3 months2(5.9)MGISEQ-2000 + DNBSEQ-T71(2.9) < 1 month1(2.9)Illumina nextseq550DX1(2.9)Sequencing modeIllumina nextseq500CN1(2.9) PE15024(70.6)Sailu-EVOR0011(2.9) PE10010(29.4)Surfseq50001(2.9)Items marked with “*” were multiple-choice questions Survey results on testing methods and other basic information from 34 laboratories Items marked with “*” were multiple-choice questions Table 2Survey results on the testing and reporting scope of 34 laboratoriesInvestigated ItemNumber of Laboratories (%)Investigated ItemNumber of Laboratories (%)Reported variant types*Reporting of variants of uncertain significance (VUS) Single nucleotide variant (SNV)34(100.0)No15(44.1) Insertion/deletion variant ≤ 10 bp (Indel)34(100.0)Yes19(55.9) Exon-level deletion/duplication variant25(73.5)⚫ If a pathogenic/likely pathogenic variant is found in a gene for a recessive disorder, an additional suspicious VUS in the same gene is also reported, even if compound heterozygosity is unconfirmed, to support future re-evaluation.6(17.6) Sex chromosome aneuploidy5(14.7)⚫ VUS highly consistent with phenotype and inheritance pattern are reported despite limited evidence, to prompt clinical follow-up.3(8.8) Dynamic mutation1(2.9)⚫ When newborn MS screening shows abnormalities, all related VUS are listed in a supplementary report to aid result interpretation.8(23.5) Inversion2(5.9)⚫ VUS with ACMG scores of 4–5, near the “likely pathogenic” threshold, are included in the report.1(2.9) Other(CNV > 1 Mb)1(2.9)⚫ VUS highly matching the phenotype and inheritance pattern are noted in an additional report to alert clinicians.1(2.9)Report ACMG secondary findings(81 genes)Verification on another technical platform before reporting Yes6(17.6)Yes27(79.4) No28(82.4)No7(20.6)Detect mosaic variantsValidation platform* No25(73.5)Sanger34(100.0) Lowest detectable mosaic ratioqPCR25(73.5) 10%2(5.9)MLPA21(61.8) 15%3(8.8)gap-PCR17(50.0) 20%2(5.9)HR karyotype5(14.7) 30%2(5.9)Report pathogenic or likely pathogenic carrier variants Yes31(91.2) No3(8.8)Items marked with ‘*’ were multiple-choice questions Survey results on the testing and reporting scope of 34 laboratories Report ACMG secondary findings (81 genes) Items marked with ‘*’ were multiple-choice questions Jaccard similarity index quantifying the relatedness of gene panels across 34 laboratories. The Jaccard similarity index ranged from 0 (no overlap) to 1 (complete overlap) and was used to measure the similarity between gene panels; Darker blue colors represented higher similarity values, indicating greater overlap of genes between laboratories; The Jaccard index was calculated using the formula: J(A,B) = ∣A∩B∣/∣A∪B∣; where A and B denoted two gene panels, ∣A∩B∣ indicated the number of shared genes, and ∣A∪B∣ indicated the total number of unique genes across the two panels Furthermore, all reported variants were systematically cataloged and compared against the pre-defined target variants (Fig. 3), while any non-target variants reported by the laboratories were also recorded (Supplementary Table S3 and S4). We analyzed how laboratories distinguished between primary and secondary findings, including reporting through report structure or textual interpretation (Fig. 4). Finally, the content of each laboratory’s report was categorized and quantified (Supplementary Figure S2), covering basic information, testing results, and genetic counseling recommendations. Clustered circular representation of variant reporting among 34 laboratories. A-C corresponded to the three simulated genomic samples (Sample 1-3) included in the external quality assessment (EQA); Each ring represented one variant, and each sector denoted one participating laboratory (L01–L34); Colors indicated how each laboratory classified and reported the variants: pink for primary findings, blue for secondary findings, and grey for unreported variants; Variants marked with an asterisk (*) were classified as likely pathogenic (LP), whereas those without an asterisk were reported as pathogenic (P); The circular layout was hierarchically clustered to group laboratories with similar reporting pattern Classification approaches for differentiating primary and secondary findings across 34 laboratories. The Venn diagram showed different methods used in laboratory reports: A (blue) — direct explanation within the report text; B (green) — annotation as a risk warning; C (pink) — separation via report structure (e.g., independent tables or appendices); D (grey)— no distinction or a single-level report; Overlapping areas represented laboratories that applied multiple approaches; numbers indicated laboratory counts, and percentages showed their proportions The counseling recommendations were further classified using a thematic analysis approach to detail the specific practices of each laboratory (Fig. 5). We categorized them into five major types: (1) clinical context, which referred to recommendations for integrating the patient’s phenotype, clinical findings, and family history; (2) parental testing, which indicated submission of parental samples for familial verification through site-specific first-generation sequencing; (3) cis-trans analysis, which involved determining whether two heterozygous variants were located in trans (compound heterozygous) or in cis (on the same allele) based on inheritance pattern analysis; (4) biochemical test, which represented additional proband testing such as biochemical assays, tandem mass spectrometry, or other related examinations; and (5) reproductive risk, which referred to the assessment of reproductive risk for the patient and recurrence risk for the parents. UpSet plot of clinical counseling recommendations reported by laboratories. The matrix below the bar graph represented each type of counseling recommendation and their intersections; One laboratory did not provide any clinical counseling recommendations and was not included in the figure above; Clinical context: referred to recommendations for integrating the proband’s phenotype, clinical findings, and family history; Parental testing: indicated submission of parental samples for familial verification through site-specific first-generation sequencing (Sanger sequencing); Cis-trans analysis: involved determining whether two heterozygous variants were located in trans (compound heterozygous) or in cis (same allele) based on inheritance pattern analysis; Biochemical test: represented additional proband testing such as biochemical assays, tandem mass spectrometry, or other related examinations; Reproductive risk: referred to the assessment of reproductive risk for the proband and recurrence risk for the parents Results Survey questionnaires The results of the questionnaires were presented in Tables 1 and 2, while the databases used by each laboratory for variant annotation were shown in Supplementary Figure S3. Sixteen laboratories (47.1%) did not set an age limit for test recipients, mainly because some commercial laboratories serve broad client groups and offer multiple types of genetic tests. The laboratories differed in their library preparation methods, capture kits, and sequencing platforms (Table 1). Among them, 90% of the laboratories used enzymatic fragmentation-based capture for library preparation, while a few still used ultrasonic fragmentation or multiplex PCR. Although multiplex PCR is cost-effective and fast, its limited coverage uniformity and GC-related dropout may compromise detection quality, especially in laboratories relying solely on this method. All laboratories detected and reported single nucleotide variants (SNVs) as well as small insertions and deletions (≤ 10 bp). Moreover, more than 70% of laboratories extended their analysis to include exon-level deletions or duplications. Regarding the reporting of secondary findings, around 80% of laboratories did not include the 81 genes recommended by the ACMG. Furthermore, 15 laboratories (44.1%) chose not to report variants of uncertain significance (VUS). Panel comparison The number of genes included in the screening panels used by the 34 laboratories varied widely, ranging from 114 to 1,355, with a median of 246 (Supplementary Figure S1). As displayed in Fig. 2, approximately one-third of the laboratories (L3, L4, L15, L22-L24, L27, L29, L32, and L34) exhibited a relatively high degree of similarity in their gene lists, all of which contained 246 genes. In contrast, laboratories with panels comprising fewer than 200 genes (L01, L08, L11, L12, L14, L18, L20, L21, L30, L33) showed low mutual similarity, as did those with more than 400 genes (L05, L06, L16, L17, L19, L28, L31). Notably, pairs such as L21-L30, L06-L31, and L12-L18 demonstrated high pairwise similarity. Importantly, laboratories L17 and L19 included substantially larger gene panels — 805 and 1,355 genes, respectively — and demonstrated low similarity with each other. This observation suggested that their expanded gene panels might have been derived from distinct sources or designed based on different selection strategies. Variant findings The reports of target variants across the three simulated samples by different laboratories were presented in Fig. 3. Overall, more than half of the laboratories demonstrated a high degree of consistency in which variants were reported for each sample. For Sample 1 and Sample 2 (Fig. 3A and B), laboratories indicated relatively consistent classification of the target variants’ pathogenicity as well as agreement on whether these variants were reported as primary or secondary findings, indicating a high level of concordance in reporting. In contrast, for Sample 3 (Fig. 3C), four target variants exhibited discrepancies in pathogenicity classification (pathogenic or likely pathogenic) among laboratories, with approximately two-thirds of laboratories classifying the GJB2 variant as likely pathogenic. Additionally, in Sample 2 (Fig. 3B), only one laboratory reported a variant in the mitochondrial gene POLG as a secondary finding and classified it as pathogenic, whereas the remaining laboratories did not report it. Overall, laboratories showed high concordance in reporting primary findings for each sample, although some differences persisted. Notably, some laboratories also reported non-target variants, which across the three samples involved five different genes. The reporting details and specific variant information were provided in Supplementary Tables S3 and S4. On average, laboratories reporting these non-target variants accounted for approximately 16% of the total. Furthermore, considerable differences were observed among laboratories in how they distinguished between primary and secondary findings in their reports (Fig. 4). The distinction methods mainly fell into three categories: direct explanation within the report text, annotation as a risk warning, or separation via report structure, such as independent tables or appendices. The number of laboratories using each method was 16 for text explanations, 11 for risk warnings, and 12 for report structure. A relatively large proportion of laboratories (about 20% each) relied solely on text, solely on structural separation, or a combination of text and structural approaches. Additionally, 10 laboratories (29.4%) employed two or more distinction methods, whereas 6 laboratories (17.6%) did not indicate any classification in their reports. Report content To assess differences in the content and structure of gNBS reports, we systematically reviewed and summarized the reports submitted by the 34 laboratories. The inclusion of various report components was illustrated in Supplementary Figure S2. Some laboratories omitted essential items such as the reason for testing request, authorized signatures, or physician and hospital information. In addition, only about half of the laboratories provided detailed evidence for variant classification, while several lacked references or information on population frequencies from databases. Moreover, 7 laboratories also incorporated the protein predictive tool REVEL to support the evaluation of potential variant pathogenicity. Furthermore, we analyzed the genetic counseling recommendations included in the reports (Fig. 5) and categorized them into five major types. Among them, the most frequent recommendations involved integrating clinical context and parental testing, reported by 7 laboratories. Overall, recommendations addressing clinical context and parental testing were the most common, each provided by more than half of the laboratories, whereas recommendations related to reproductive risk were the least frequent (5 laboratories). Only one laboratory did not provide any genetic counseling recommendation. It is worth mentioning that a deeper examination of the report content revealed additional notable features. Four laboratories appended glossaries explaining genetic terminology and specialized terms used in the report. Two laboratories included flowcharts outlining their bioinformatics analysis pipelines, and another four listed ACMG criteria for pathogenicity classification. Interestingly, one laboratory also attached a summary of its internal historical data, which illustrated the proportions of different variant types detected in its previous testing experience. Discussion Birth defects are a leading cause of childhood morbidity and mortality [17, 18]. Rare diseases, often linked to genetic birth defects, frequently face diagnostic delays due to phenotypic complexity and limited resources, imposing substantial burdens on families [19, 20]. The gNBS plays a pivotal role in the early detection and precise intervention of rare genetic disorders, highlighting the critical importance of improving gNBS report quality and implementing robust quality assessment frameworks. Through comprehensive analyses, our study identified substantial variability across laboratories in variant interpretation, report structure, and other aspects, revealing areas that warrant further refinement. We observed that some laboratory reports lacked essential pre-analytical information (Supplementary Figure S2). Patient phenotype and a clear reason for request constitute key evidence for variant classification [10]; omission of this information may disconnect results from the clinical context, increasing the risk of misinterpretation or miscommunication. Complete patient identification also minimizes confusion during report delivery and interpretation and embodies a patient-centered approach to quality management. Additionally, 4 laboratories omitted authorized signatures, a significant oversight, as signatures are crucial for result verification and accountability within laboratory quality management systems. The absence of signatures does not comply with ISO 15,189 standards and may compromise traceability [21]. Distinct from conventional DBS-based EQA programs [22, 23], our study employed server-generated simulated samples, eliminating the need for specimen collection, transportation, and storage. It reduces logistical and ethical burdens, enables rapid multi-round testing, and enhances the flexibility and sustainability of EQA. Compared with traditional EQA programs distributing limited physical samples annually [24], simulated samples facilitate the rapid creation of diverse variant datasets, allowing a more comprehensive evaluation of laboratory performance across variant types and providing an efficient solution for quality control. Moreover, the virtual sample system can be expanded to construct virtual pedigrees, generating simulated sequences for both neonates and their parents. This enables validation of allele phasing, analysis of inheritance patterns, and interpretation of compound heterozygosity, thereby reducing false-positive findings caused by cis mutations or missing parental information [25]. In this study, we observed that two laboratories (L17 and L19) included substantially more genes in their screening panels than the others, with limited overlap between them. The two laboratories differ markedly in their testing strategies, including age restrictions, capture kit manufacturers, and sequencing platforms. Although both laboratories refer to the disease lists recommended by ACMG and national NBS consensus documents, some conditions have multiple subtypes involving different numbers of associated genes. Decisions such as targeting only hotspot regions versus providing broader coverage further contribute to discrepancies in gene lists. Differences in geographic location and client profiles may also influence panel design. In practice, despite the large overall gene sets, both laboratories tailor sub-panels according to hospital requirements. Regarding costs, both reported that pricing varies based on hospital procurement processes, making direct comparison difficult. While expanded panels may benefit certain neonates by covering more genes, premature or excessive inclusion of genes could pose challenges in implementation and cost management. Thus, adherence to fundamental NBS principles — focusing on early-onset, treatable, and highly penetrant conditions — is essential [8]. Although consensus exists in China regarding diseases to be screened, there remains no standardized guideline on which genes to include or variants to report. Minten et al. [26] indicated that gene selection differences across programs are primarily influenced by public health recommendations, disease natural history, and treatment feasibility, providing a reference for further panel optimization. Regarding the distinction between primary and secondary findings (Fig. 4), integrating risk warning with a structured report may be the most intuitive and effective approach. Without visual cues such as bold text or color coding, key information can be obscured, reducing readability and increasing cognitive load. Similarly, Sam et al. [27] suggested that structured and tiered presentation of primary and secondary findings, accompanied by explanatory notes or letters, facilitates comprehension by clinicians and patients, supporting our recommendation for combining risk annotation with structured segmentation. Only one laboratory reported the POLG p.R964C variant (Fig. 3), and only seven panels included this gene, indicating low overall coverage and reporting frequency and limiting its utility as a target variant. Follow-up inquiries showed that although most laboratories were technically capable of detecting this sequence variant, its reporting depended largely on internal reporting strategies rather than analytical sensitivity. For example, one laboratory (L17) reported the variant as pathogenic (P) in the context of carrier findings, whereas another laboratory (L19) exhibited regional differences in reporting practices—sites in northern regions classified and reported it as likely pathogenic (LP), while sites in southern regions retained the variant as VUS and did not report it. Although we initially expected the variant to be detectable in more laboratories, the findings demonstrated insufficient coverage and inconsistent reporting. Future studies may therefore benefit from expanding the survey to include more laboratories and systematically comparing panel coverage, thereby providing more reliable evidence for selecting target variants. Besides, a minority of laboratories (< 20%) detected off-target variants (Supplementary Tables S3 and S4). Variations in bioinformatics pipelines, including overly permissive variant calling or filtering thresholds, as well as minor biases introduced during variant design or sequence construction (e.g., coordinate conversion or file integration errors), may contribute to these findings. Overall, off-target variants were infrequent and had limited impact on the evaluation results, yet underscore the need for stricter target variant definition and validation mechanisms to ensure consistency and traceability. While genetic counseling has become an established profession with formal training and certification programs in regions such as America and Europe, its development in China remains limited [28–30]. Although China has launched several pilot training programs [31], a comprehensive framework for cultivating the next generation of genetic counselors remains lacking. Accordingly, non-specialist physicians often have limited capacity to interpret complex molecular reports, and patients generally possess insufficient foundational knowledge. Therefore, clear and actionable next-step counseling or follow-up guidance on report front pages is critical to facilitate result translation. Our analysis of counseling recommendations across laboratory reports (Fig. 5) revealed substantial variability in content and depth. Only 13 laboratories (38.2%) suggested integrating clinical or parental information, five addressed reproductive risk, and only one specified further clinical testing. This inconsistency partly reflects divergent understanding of the genetic counseling role, with counseling depth and format heavily dependent on individual experience, lacking institutional standardization. These observations align with the study (Tutty et al.) which identified insufficient informed consent procedures and counselor availability as major barriers to gNBS implementation [32]. Enhancing systematic training and certification for genetic counselors, clarifying their role in reporting, and optimizing patient-centered report content are key future priorities. Surprisingly, some laboratories have explored innovative report enhancements, including glossaries of genetic terminology, bioinformatics workflow diagrams, ACMG variant classification criteria, and historical data summaries. From a patient-centered perspective, such reports not only convey laboratory findings but also facilitate communication and education [16, 33]. Incorporating explanatory content and visualization elements reduces professional barriers and supports shared decision-making between patients and clinicians, a practice worth promoting in future report standardization. This study has several limitations. Although 34 laboratories participated, they account for less than 5% of all laboratories nationwide, and enrollment was voluntary following an open invitation. As a result, the sample may not fully represent the national landscape, and laboratories with a stronger willingness to participate—often those with more mature quality management systems and more standardized reporting practices—may have introduced a bias toward higher-performing institutions, thereby affecting representativeness. In addition, while simulated sequencing samples provide efficiency and scalability, they inherently bypass critical wet-lab steps such as DNA extraction, library preparation, and variability in sequencing depth [34]. Consequently, they are not suitable for evaluating the accuracy of copy number variation (CNV) detection. All variants assessed in this study were limited to single-nucleotide variants (SNVs) and small insertions/deletions, which constrains the ability to assess laboratory performance in CNV detection. Furthermore, the absence of trio-based samples limited our ability to evaluate laboratory performance in family-based variant interpretation. In addition, we did not collect detailed reasons for reporting each variant, which constrained our analysis of reporting decisions and potential areas for improvement. These issues will be addressed in the next round of assessment. Conclusion In this study, we systematically analyzed gNBS reports from 34 laboratories in China, revealing variability in variant interpretation, report structure, and genetic counseling, while also highlighting innovative efforts in patient education and data visualization. The use of simulated samples demonstrated efficiency and scalability, providing a technical approach for future EQA programs, though limitations remain for complex variants such as CNVs. Our findings advocate for standardized gene panel design and reporting frameworks, while the comparative analysis of current practices provides a benchmark to guide laboratory-specific improvements. Supplementary Information Supplementary Material 1: Table S1. Clinical information of three simulated samples. Table S2. Detailed information of five target variants in each of three samples. Table S3. Summary of reported non-target variants and their classification in three samples across 34 laboratories. Table S4. Detailed information on non-target variants reported by the laboratories. Figure S1. Number of genes included in newborn genetic screening panels across 34 laboratories. Figure S2. Composition of key elements in newborn genomic screening reports from 34 laboratories. Figure S3. Number of laboratories using each annotation database Acknowledgements We are grateful to all the staff members for taking part in this study. We thank the laboratories for their cooperation and report contributions. Authors' contributions Methodology, P.F and F.H.; software, J.W. and J.Z.; formal analysis, J.W.; resources, H.Z. and W.T; writing—original draft preparation, J.W.; writing—review and editing, P.F. , C.H., and F.H.; visualization, J.Z. and X.C. All authors have read and approved the submission. Funding This study was supported by the National Key Research and Development Program of China (grant number 2022YFC2705302), the National Natural Science Foundation of China (grant number 82360297) and the Science and Technology Program of Guizhou Province, China (grant number Qian Ke He Ji Chu-ZK[2023] Yi Ban 204). Data availability Data supporting the findings of this study are available from the corresponding author (hefalin@126.com) upon request. Declarations Ethics approval and consent to participate This study did not involve actual human or animal subjects; therefore, although it relates to newborn screening, no parental or guardian consent was required. The study was performed in accordance with relevant guidelines and regulations and was approved by the Ethics Committee of the Chinese PLA General Hospital (approval number: S2022-693-01). Content for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Jiaming Wu, Penghui Feng and Jinming Zhang contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1: Table S1. Clinical information of three simulated samples. Table S2. Detailed information of five target variants in each of three samples. Table S3. Summary of reported non-target variants and their classification in three samples across 34 laboratories. Table S4. Detailed information on non-target variants reported by the laboratories. Figure S1. Number of genes included in newborn genetic screening panels across 34 laboratories. Figure S2. Composition of key elements in newborn genomic screening reports from 34 laboratories. Figure S3. Number of laboratories using each annotation database Data Availability Statement Data supporting the findings of this study are available from the corresponding author (hefalin@126.com) upon request.