Quality assessment of the forensic surveillance system: A data-quality monitoring system for substance-abuse detection in unnatural deaths in Thailand Authors: Patrinee Traisathit, Wisarn Worasuwannarak, Apinun Aramrattana, Chaisiri Angkurawaranon, Wichuda Jiraporncharoen, Nida Buawangpong, Kanicnan Intui, Apinut Rankantha, Yuphayong Thanuta, Pimwarat Srikummoon Published: 2026 ABSTRACT AbstractObjectiveThis study aimed to assess the reliability and identify sources of discrepancies to strengthen data governance and system integrity of the national forensic toxicology surveillance database from postmortem examinations called by the Forensic Drug Abuse Monitoring Network (FDAMN), Thailand.MethodsA facility-based cross-sectional study was conducted across 25 forensic institutions participating in the FDAMN. 10% (1441 cases) of postmortem toxicology records during January 1 to December 31, 2021, were randomly selected for independent re-entering records for verification during fiscal year 2023. Subsequently, 638,100 paired data points from the original and re-entered datasets were compared. Data consistency was analyzed using descriptive statistics, and a qualitative review was performed to identify underlying causes of discrepancies.ResultsA total of 1418 matched records (98.4%) were successfully verified. All of the data fields were identical between the datasets in 98.5% of cases, with 1.5% showing discrepancies. Administrative variables exhibited 96.3% consistency, while demographic and toxicological data exceeded 94% consistency. Circumstance variables, such as the date and time of death, showed the highest discrepancy rate (6.0%). Most inconsistencies arose from administrative coding variations, typographical errors, and/or legitimate data updates following confirmatory analyses.ConclusionsThailand's national forensic toxicology surveillance database demonstrates high data consistency, thereby affirming the robustness of the FDAMN system. Remaining discrepancies are primarily procedural and can be mitigated through standardization of coding, automation of data-validation checks, and feedback-driven quality improvement. These findings may provide a practical framework for strengthening forensic drug surveillance systems and data-quality assurance in other countries facing evolving illicit drug threats.Keywords: Forensic toxicology, Data quality, Thailand public-health informatics, Drug monitoring, Evidence-based decision-making Introduction The world drug situation is characterized by dynamic and evolving markets that significantly impact global health, governance, and security. In 2023, 316 million people worldwide are known to have used illicit drugs [1]. This crisis is increasingly driven by the spread of synthetic drugs, including fentanyl that is up to 50 times stronger than heroin and hundreds of new psychoactive Substances (NPSs) identified annually [2]. In Southeast Asia, the Golden Triangle (the border region between Myanmar, Lao PDR, and Thailand) remains one of the world's most significant zones for illicit drug manufacture and distribution. Thailand's geography makes it both a major destination for methamphetamine and a key pathway for sending illicit drugs to other countries [3]. The regional market has been further complicated by the arrival of new synthetic mixtures, such as ‘Happy Water’ or ‘k-powder’ cocktails, which are often surreptitiously distributed in entertainment venues [4]. The presence of these complex substances elevates the importance of postmortem toxicology in accurately identifying causes of death that might otherwise remain hidden. Because many of these synthetic cocktails and NPSs evade standard drug screening, advanced forensic analysis is required to detect their presence during postmortems. To enhance drug control and public health responses in Thailand and across the ASEAN region, the forensic information system should fulfill several critical roles, including regional standardization and data sharing [5]. The detection of drugs and toxic substances in postmortem biological specimens is a cornerstone of forensic and toxicological investigations. It provides crucial evidence for determining the cause and manner of death, particularly in cases involving overdose, illicit drug use, or deaths under suspicious circumstances [6]. The reliability of these toxicological findings is necessary for both the judicial process and public-health surveillance. Inconsistencies in toxicology data can lead to the misclassification of drug-related deaths and, ultimately, to misguided policy responses [7,8]. Accurate and reliable data are fundamental to the development of effective public-health and forensic surveillance systems. In the context of substance-abuse monitoring, high-quality data not only support epidemiological assessments but also play a crucial role in shaping evidence-based policies and interventions. In Thailand, forensic toxicology data derived from biological specimens of deceased individuals are routinely collected as part of the national surveillance system for substance abuse in unnatural deaths. These data are instrumental in identifying emerging drug trends, assessing the burden of substance-related mortality, and guiding the national response to illicit drugs [9]. The Office of the Narcotics Control Board (ONCB) of Thailand coordinates a nationwide network of 49 forensic institutions that collect and report toxicology results through the Postmortem Toxicology Information System, known as the “Forensic Drug Abuse Monitoring Network”. Established in Thailand since 2007, this system serves as the backbone for national drug-surveillance reporting, trend analysis, and the identification of emerging psychoactive substances [10]. As such, any discrepancy or delay in this dataset can distort assessments of drug problems and lead to suboptimal national policy decisions. Recognizing this challenge, the present study examined the consistency of data within this national database to assess its reliability, with the aim of fostering data transparency across institutions and strengthening Thailand's capacity for evidence-based drug-policy formulation. Similar surveillance frameworks employed internationally include the European Union Drugs Agency (EUDA) early warning system in Europe [11], the Centers for Disease Control (CDC) overdose surveillance programs in the US [12], and Australia's Illicit Drug Reporting System [13]. These initiatives emphasize the importance of high-quality, standardized data for detecting emerging drug threats and informing policy responses. The consistency of forensic toxicology data depends on multiple factors, including analytical techniques, specimen handling, data transcription, and inter-laboratory consistency. When discrepancies are not systematically monitored, even minor data-handling errors such as a date of death exceeding the current date, transposed digits in age, miscoded cause of death, an incorrect incident location code, miscoding of substances, duplicate records, and inconsistencies in free-text entries could propagate across institutional databases, thereby distorting national surveillance outcomes [14,15]. Robust discrepancy analysis provides a structured approach to evaluating data quality, identifying weaknesses in laboratory and data-management processes, and harmonizing standards among diverse data-producing institutions [15]. In forensic toxicology, such evaluations are particularly vital when multiple specimen types and analytical methods are used [16]. However, the reliability and validity of this database depend largely on the precision and uniformity of data entry and management across the various reporting sites. Errors such as missing values, inconsistent terminology, or transcription mistakes can undermine data integrity and lead to inaccurate interpretations of national trends [17]. Hence, systematic data-quality assessment is essential for ensuring that the database accurately reflects forensic findings and remains a trustworthy resource for both policy development and research. This study introduces a surveillance-integrity framework for the discrepancy analysis of postmortem toxicology data, not merely as a quality-control measure but also as an indicator of system reliability. By identifying patterns of data discrepancies, updates, and harmonizations, the aim of the framework is to enhance the credibility of forensic evidence for national surveillance purposes by establishing standardized data-quality practices across institutions. The results of this analysis are intended to strengthen confidence in forensic and analytic outcomes; support informed policy decisions in drug prevention and control; and contribute to the development of a reliable, modern, and sustainable forensic drug-surveillance system over the long term. Methods A facility-based cross-sectional study was conducted to assess data consistency within the Forensic Surveillance System for substance-abuse detection in unnatural deaths in Thailand. The study forms part of a national initiative by the Forensic Drug Abuse Monitoring Network (FDAMN) under the ONCB. Operational since 2007, the surveillance system has been used to continually collect toxicological results from forensic and medico-legal laboratories across the country, including 25 forensic laboratories from medical schools, regional and provincial hospitals, and other forensic agencies distributed across Thailand. Data verification sample and process Data verification was undertaken during the 2023 fiscal year using information originally entered in 2021. The study population consisted of data on deceased individuals who had died of unnatural causes and undergone biological specimen testing for illicit drugs during the year 2021. The inclusion criteria required network institutions to have recorded at least five cases during that year, resulting in the enrollment of 25 eligible institutions (representing 51% of the total number of institutions). From 14,387 cases (data as of February 16, 2022), a 10% proportionate stratified sample was drawn from each institution to ensure balanced representation. The sampling within each institution was executed via simple random sampling without replacement using Stata/MP version 17.0. Due to the rounding of sample sizes to the nearest whole number within each stratum, the final analytical sample size was 1441 cases. This process was conducted by a team of statisticians who were not involved in the routine data entry of forensic surveillance system. Two datasets were used for quality assessment. The second dataset (re-entering records for verification) consisted of the same data re-entered independently from primary forensic source documents, by staff of participating institutions for quality verification purposes. To ensure consistency, staff of participating institutions were trained in data entry according to the established protocol before re-entering the second dataset. A subset of records was double-reviewed, discrepancies were recorded and resolved by staff of the participating institutions. Both datasets were obtained from the centralized web-based platform (https://forensic.nccd.go.th) managed by the ONCB and the FDAMN. Record linkage and verification Data verification involved record linkage between the original and second-entry datasets. Matching was performed using five key identifiers: year of death, laboratory record ID, institution code, institution name, and institutional sample number. When discrepancies were observed, additional demographic and contextual variables, such as sex, age, and residential information, were used to confirm that the matching was correct. Data processing and comparison were conducted using Stata/MP version 17. Four categories of variables covering the demographic, administrative, circumstances, and toxicological domains were included in the verification process. Key variable categories included administrative variables: codes and names of police stations, hospitals, other referring agencies, provinces, and districts (10 variables); demographic characteristics: specimen status, identification status, sex, age, nationality, ethnicity, education, and occupation (23 variables); event-related information: location, date, time, and cause of death (22 variables); and toxicological findings: detection results for methamphetamine, 3,4-methylenedioxymethamphetamine (MDMA), morphine, heroin, 6-monoacetylmorphine (6-MAM), codeine, cannabis, cocaine, ketamine, benzodiazepines, kratom, alcohol, and other psychoactive substances (29 variables per substance except alcohol 16 variables). Overall, 638,100 data points (entries) were assessed for consistency across the two datasets. Variable comparisons Each variable comparison was categorized into one of four mutually exclusive groups: Exact match: the data fields were identical across both datasets; Updated information: the re-entered data provided additional or more complete information consistent with the original record; Minor textual variation: discrepancies due only to typographical or spacing differences or rounding of decimal values; and Discrepant entry: values that were clearly inconsistent, missing, or contradictory between the two datasets. Only group discrepancies were classified as data errors for the consistency assessment. In cases where inconsistencies were identified between the two datasets, the records were referred back to the participating institutions for verification, confirmation, or correction of each variable. Statistical analysis Data cleaning and analysis were performed using Stata/MP version 17. Descriptive statistics were applied to summarize the level of agreement between the datasets. Consistency was expressed as the proportion of data entries without discrepancies relative to the total number of variables examined. Analyses were conducted for all variables combined and by variable category (administrative, demographic, circumstances, and toxicological). The overall consistency rate was calculated as Qualitative analysis A qualitative review was also performed to identify any patterns in recurrent discrepancies and to evaluate potential sources of error. Based on an established framework for data-quality error classification in medical registries [15,17], discrepancies were classified into three pre-defined error sources: human, technical, and procedural. Human errors included data entry mistakes, such as typos and misclassification, especially in free-text fields. For instance, discrepancies arose in the additional cause-of-death field where one data entry clerk entered a broad category (e.g., ‘accident’), while the second recorded highly specific clinical findings (e.g., ‘skull fracture, brain laceration’). Similar discrepancies occurred in numeric data entered as free text, such as age transpositions (e.g., ‘54.0′ recorded by one data entry clerk versus ‘45.0′ by the other). Technical errors involved inconsistencies in institutional coding and variable definitions between systems. Procedural errors were identified as discrepancies arising from unclear protocols or inconsistent implementation of instructions. This review was conducted by three research team members (A.R., Y.P., and P.S.) using Stata/MP version 17.0; while clear cases were assigned directly to a category, ambiguous cases were resolved through consensus meetings with senior investigators (A.A., W.W., and P.T.) to ensure consistency. Ethics approval and consent to participate The study was conducted in accordance with the Declaration of Helsinki and approved by the Research Ethics Committee of the Faculty of Medicine, Chiang Mai University (No: 0830/2568). Due to the data having been anonymized, the requirement for written informed consent was waived by the Research Ethics Committee of the Faculty of Medicine, Chiang Mai University. Results Data inclusion and completeness A total of 1441 postmortem cases recorded in 2021 were selected for re-entering records for verification across 25 forensic laboratories nationwide. Of these, 1418 records (98.4%) were successfully re-entered into the national database and matched with their corresponding original entries. The final dataset for comparison purposes comprised 638,100 paired data points covering the administrative, demographic, circumstances, and toxicological domains. Fig. 1 provides a summary of the sampling process and data-verification flow, illustrating the linkage process and completeness rate across institutions. The workflow for the sampling and data-verification processes. Overall data consistency Across all data domains, 98.5% of entries were identical between the original and re-entered datasets, while 1.5% (9708 fields) exhibited discrepancies (Table 1): these ranged from missing or inconsistent entries to clearly contradictory information. The high agreement rate demonstrates strong data-entry consistency within the FDAMN, thereby reflecting robust data management across institutions. Table 1Overall data-entry consistency in the Forensic Toxicology Surveillance database.Table 1Data StatusNumber of EntriesPercentage (%)Exact match628,39298.5Discrepant entries97081.5Total638,100100.0 Overall data-entry consistency in the Forensic Toxicology Surveillance database. Consistency by variable category Administrative variables Among 14,180 administrative entries (institution, police station, province codes, etc.), 96.3% were consistent and 2.6% showed discrepancies (Table 2), primarily due to inconsistent or incomplete recording of referring-agency information. Cases with “updated” data (1.1%) indicate improved data completeness after the second entry (e.g., identifying sub-units within referring hospitals). Table 2Administrative variable consistency in the Forensic Toxicology Surveillance database.Table 2Data statusNumber of entriesPercentage (%)Exact match13,65096.3Updated information1601.1Discrepant entries3702.6Total14,180100.0 Administrative variable consistency in the Forensic Toxicology Surveillance database. Demographic variables These showed generally high agreement across the two datasets, with exact matches exceeding 93% for most fields (Table 3). Both specimen status and identification status demonstrated perfect consistency (100%), thus indicating complete consistency in these administrative demographic indicators. Table 3Demographic variable consistency in the Forensic Toxicology Surveillance database.Table 3VariableExact match (%)Updated (%)Minor Textual Variation (%)Discrepancy (%)Specimen status100.00.00.00.0Identification status100.00.00.00.0Sex98.20.10.01.7Age95.30.12.12.5Nationality96.20.40.03.4Ethnicity100.00.00.00.0Education99.60.10.00.3Occupation84.711.10.04.2Residence/current address93.14.50.02.4 Demographic variable consistency in the Forensic Toxicology Surveillance database. Among the personal demographic variables, sex and age showed high agreement rates (98.2% and 95.3%, respectively), although minor discrepancies were observed. Discrepancies in age (2.5%) mainly arose from rounding or estimation during data entry. Nationality data exhibited a 3.4% discrepancy rate, largely due to misclassification between Thai and non-Thai nationals. Occupation had the highest proportion of updated entries (11.1%) and a discrepancy rate of 4.2%, thus indicating post-entry completion or revision during data verification. In contrast, ethnicity and education were highly consistent, with over 99% exact matches and negligible discrepancies (0.0% and 0.3%, respectively). The residence or current address variables demonstrated slightly lower agreement (93.1% exact match, 2.4% discrepancy), likely due to incomplete geographic coding or differences in administrative area naming conventions between datasets. Circumstances variables These showed slightly lower agreement than the demographic data (Table 4). Incident location was highly consistent (96.8%), with minor discrepancies due to non-matching subdistrict codes. Detailed event descriptions had moderate agreement (87.6%), with small variations or omissions of location details in the later entry. The date and time of the event showed the highest discrepancy rate (6.0%), mainly from confusion between “date found” and “date of death.” Cause of death and additional cause-of-death notes had undergone substantial updates (16.3% and 22.1%, respectively), thereby indicating revisions made after diagnostic reviews. The manner of death showed 88.9% agreement, with some omissions during re-entering records. Table 4Circumstances variable consistency in the Forensic Toxicology Surveillance database.Table 4VariableExact match (%)Updated (%)Minor Textual Variation (%)Discrepancy (%)Source of DiscrepancyIncident location96.81.30.01.9Non-matching subdistrict identifiersDetailed event description87.63.67.11.7Previously specified event-location details were missing in the later entryDate and time of event93.70.30.06.0Inconsistent date of event; e.g., “date found” recorded instead of “date of death”Cause of death83.716.30.00.0Revised or refined entries during re-entryAdditional cause-of-death notes72.122.15.20.6Mismatched cause of deathManner of death88.97.00.04.1Previously specified manner of death details were missing in the later entry Circumstances variable consistency in the Forensic Toxicology Surveillance database. Toxicological data consistency The toxicological data in Fig. 2 show high overall consistency across all substance types, with most entries demonstrating over 94% exact matches. Heroin, 6-MAM, and codeine show the highest consistency (≥99%), while minor updates were present in less than 1% of records. Moderate discrepancies were found for methamphetamine (2.3%), ketamine (2.3%), cannabis (2.1%), benzodiazepines (2.1%), and alcohol (1.8%), largely reflecting differences in analytical method reporting or confirmation updates. Methamphetamine also accounted for the largest proportion of updated entries (5.1%), followed by benzodiazepines (5.7%) and alcohol (5.3%). Overall, the toxicological results demonstrated strong inter-dataset consistency, with most variations attributable to legitimate laboratory result revisions or refinements rather than data-entry errors. Summary of toxicological data consistency by substance type. Patterns and sources of discrepancies Qualitative review of the verification process revealed that data discrepancies primarily originated from three interrelated factors: administrative inconsistencies, human-entry errors, and data-update dynamics. Administrative inconsistencies commonly stemmed from variations in institutional or police station codes, thereby indicating the absence of standardized naming conventions across reporting sites. Human-entry errors were also evident, including typographical mistakes, omissions, and inconsistent spelling in free-text variables such as occupation and incident descriptions. These types of discrepancies are largely attributable to manual data input and insufficient use of data-validation checks in the online system. In addition, a proportion of the identified differences reflected data-update dynamics, especially where laboratory results were revised following confirmatory analysis or updated after delayed communication between the regional laboratories and the central database, rather than true errors. Interestingly, institutions showing a higher proportion of updated information also demonstrated stronger internal quality-assurance practices, thus suggesting that systematic data verification was actively implemented rather than an indication of data inconsistency. Overall, these findings show that most discrepancies were procedural or administrative in nature and highlight the need for enhanced standardization and automated validation mechanisms within the Forensic Surveillance System. Discussion This study provides a comprehensive evaluation of internal consistency within Thailand's national forensic toxicology surveillance database via large-scale paired verification of more than 638,000 data fields from 25 institutions. The overall consistency rate of 98.5% demonstrates that the FDAMN operates with a high degree of reliability and data integrity. These findings affirm that the system provides a robust foundation for drug-surveillance reporting and policy formulation consistent with global recommendations emphasizing data validity as being essential for forensic and public-health decision-making [18]. Moreover, the data derived from this system are utilized in research related to alcohol-associated unnatural deaths in Thailand [19]. Although overall agreement was excellent, discrepancies were not uniformly distributed across the variable domains. The administrative and circumstances data were slightly less consistent than the demographic and toxicological data. The highest discrepancy rate (6.0%) occurred in the date and time variables, often due to confusion concerning the date of death. Similar temporal-entry errors have been reported in international forensic databases, thus highlighting the vulnerability of free-text or manually entered fields [20]. In contrast, toxicological results exhibited very high consistency (exceeding 94%) for all major analytes, and discrepancies were largely attributable to legitimate updates rather than transcription errors. These patterns suggest that structured, coded data formats and drop-down selection menus can play a key role in minimizing manual-entry variation. Importantly, a notable proportion of discrepancies were classified as “updated information” rather than true errors due to post-entry corrections, laboratory result confirmations, and/or expanded details obtained after the initial report. Rather than indicating weakness, these updates demonstrate the presence of dynamic quality control, a process whereby ongoing verification contributes to system learning and improvement [21,22]. Comparable findings have been documented in other health-surveillance systems, where iterative re-validation cycles strengthen both data accuracy and completeness [23]. Moreover, institutions showing higher rates of updated data often demonstrate stronger internal quality-assurance mechanisms, highlighting that discrepancy monitoring is a proxy indicator of institutional quality maturity. The study also identified specific areas for system enhancement. Administrative inconsistencies, particularly in institutional and police station codes, reflect the need for standardized nomenclature and uniform coding systems [24]. Human-entry errors, including typographical and narrative inconsistencies, highlight the importance of automated validation checks, staff retraining, and feedback loops between local laboratories and the central ONCB database. Implementing real-time discrepancy dashboards or automated alerts could facilitate proactive monitoring and correction [25]. Furthermore, adopting double-entry verification for key identifiers would reduce transcription errors [26], while integrating expert reviews by forensic pathologists for interpretive fields such as cause and manner of death could ensure medical validity of complex variables [27]. Together, these measures would create a multi-layered quality-assurance framework combining automation, human oversight, and institutional accountability. Since several approaches have been employed for illicit drug use surveillance (e.g., toxicology testing of patients in emergency departments, surveying sentinel populations, wastewater-based epidemiology, and forensic pathology [[28], [29], [30], [31]], the data therefrom could be shared and integrated to improve the quality and efficiency of illicit drug use monitoring and surveillance. The Overdose Data to Action (OD2A) cooperative agreement launched by the Centers for CDC in the US has suggested some strategies to enhance overdose surveillance infrastructure to track nonfatal and fatal overdoses, including rapidly identifying drug overdose hotspots and suspected drug overdoses, both in survivors and during postmortems [32]. In addition to timely toxicology testing, ensuring data quality along with strategies for data modernization could improve nonfatal and fatal drug overdose surveillance. A previous study in Pennsylvania demonstrated the benefits of drug overdose surveillance and information sharing via a publicly accessible database [29]. The authors provided examples of processing drug overdose surveillance data via an online platform, including the data entry procedure by the medical examiners or coroners, the review procedure by the trained staff, and the editing and revision procedure after review. Their processes for data entry and revision are similar to those used in the quality monitoring system described in the present study. Our quality monitoring procedures could be beneficial for drug surveillance using death investigation databases in other countries, especially those in the ASEAN Drug Monitoring Network that cooperated and shared the data for the monitoring of illicit drug trends between Thailand and other ASEAN member states. This study offers several strengths. It highlights nationwide forensic data-verification exercises combining quantitative consistency assessment with qualitative review of discrepancy origins. Moreover, the use of independent re-entry by different personnel provides an objective measure of data reliability. Nonetheless, limitations should be acknowledged. The analysis was confined to data from 2021 and may not reflect improvements implemented in later years [33]. In addition, a limitation might have arisen due to choosing 25 agencies out of a total of 49 (51%) for re-entering records for verification. Because of the strict criterion mandating a minimum of five cases per organization, the remaining 24 agencies could not be audited, thereby potentially limiting the national representativeness of the study. Furthermore, the study was focused on data consistency rather than directly assessing the impact of discrepancies on national drug-trend interpretation or policy responses. Finally, staff re-entering the records might have demonstrated heightened attentiveness during verification, thereby potentially inflating the apparent consistency levels. Overall, the results highlight that the strength of Thailand's forensic toxicology surveillance system lies not only in analytical precision but also in its evolving data governance processes. Regular quality audits, standardization, and feedback-driven improvement are critical for sustaining reliability as the system is scaled up to accommodate emerging psychoactive substances and digital data flows. Conclusion Thailand's national forensic toxicology surveillance database demonstrates high overall data consistency and internal consistency (98.5%), thereby confirming the robustness of the FDAMN system. Minor discrepancies were primarily due to administrative and/or procedural variation and can be effectively mitigated through standardization, automation, and enhanced feedback mechanisms. The study highlights that data reliability in forensic surveillance depends equally on laboratory competence and information-governance quality. By institutionalizing regular discrepancy monitoring, implementing standardized coding systems, and integrating expert validation, Thailand can strengthen the credibility, transparency, and long-term sustainability of its forensic toxicology surveillance infrastructure. Beyond its national value, a robust forensic toxicology surveillance system contributes to international monitoring efforts coordinated by organizations such as the UNODC and WHO. Reliable postmortem toxicology data can support early warning systems for emerging substances, facilitate cross-border trend analysis, and strengthen regional responses to evolving illicit drug markets. Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Funding This research project was funded by the Office of the Narcotics Control Board (ONCB) Thailand. CRediT authorship contribution statement Patrinee Traisathit: Conceptualization, Data curation, Formal analysis, Methodology, Resources, Writing – original draft, Writing – review & editing. Wisarn Worasuwannarak: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Resources, Writing – review & editing. Apinun Aramrattana: Conceptualization, Data curation, Resources, Writing – review & editing. Chaisiri Angkurawaranon: Data curation, Resources, Writing – review & editing. Wichuda Jiraporncharoen: Data curation, Resources, Writing – review & editing. Nida Buawangpong: Data curation, Resources, Writing – review & editing. Kanicnan Intui: Data curation, Resources, Writing – review & editing. Apinut Rankantha: Formal analysis, Methodology, Writing – review & editing. Yuphayong Thanuta: Formal analysis, Methodology, Writing – review & editing. Pimwarat Srikummoon: Conceptualization, Data curation, Formal analysis, Methodology, Resources, Writing – original draft, Writing – review & editing. Declaration of competing interest There are no conflicts of interest to declare concerning the conduct and reporting of the research. The manuscript has not been previously published nor considered for publication elsewhere. Acknowledgements We would like to express our sincere gratitude to all personnel of the Forensic Drug Abuse Monitoring Network (FDAMN) for their continued support and invaluable contributions to this project. We also gratefully acknowledge the Northern Substance Abuse Center for their academic collaboration, coordination, and technical support. In addition, we thank the Office of the Narcotics Control Board (ONCB) for their kind assistance, constructive guidance, and support that helped facilitate the successful implementation of this project. This research was partially supported by the Chiang Mai University. References Associated Data Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.