Remember me
Asia is a vast and diverse continent that also represents varied health care systems and socioeconomic challenges. Multiple evidence-driven approaches tailored to each nation’s unique health care and research context are required to draw essential data to support the ambitious goals for universal health coverage in each country [,]. The strength and necessity of real-world data (RWD) and their concrete data analytical interference in terms of real-world evidence (RWE) are integral to this evidence generation. RWE has the potential to inform health technology assessments (HTAs), guide evidence-driven policies, and streamline service delivery []. However, as crucial as RWE is, the Asian health care landscape lacks a cohesive framework to harness its full potential despite its promise in pharmacoeconomics, pharmacovigilance, and pharmacoepidemiology [,].
The utility of RWD and RWE becomes even more apparent with large integrated research databases within health systems. The integrated warehouses offer vast connected data sets that sustain the statistical rigor and can assist in providing insights with minimal bias and confounding []. However, these data reservoirs have not been vastly studied across health care systems, which limits their broader utility in health care research; RWE data generation; and, consequently, universal health coverage [].
Recognizing this potential, our previous research explored integrated RWD warehouses within 3 diverse health care pilots for Taiwan, India, and Thailand []. Our systematic research identified strong differences in the types of RWD and their warehouses in the 3 countries. Still, the results only partly reflected their divergent economic, social, and clinical settings. Hence, we continued to conduct similar research in many other diverse Asian health care systems in line with our published protocol [].
ObjectivesThe literature on RWD practices and awareness of corresponding warehouses in certain Asian countries such as China, Japan, and South Korea is significant [,-], partly because these countries also have recommendations on the utility of RWE by external regulators []. This study sought to understand the evolving landscape of RWD use and its implications across Hong Kong, Indonesia, Malaysia, Pakistan, the Philippines, Singapore, and Vietnam, where RWD practices are emerging or undergoing significant development []. Our selection of countries for this scoping review was strategically based on selecting a contrasting spectrum of HTA maturity across countries with evolving HTA systems, ranging from relatively mature systems in Singapore, Thailand, and Malaysia to emerging frameworks in Indonesia, the Philippines, and Vietnam and systems in the nascent stages in Pakistan [,]. Each nation, with its individualistic health care challenges and unique research capabilities, underscores the need for understanding recent patterns in RWD research and use of clinical research warehouses, especially in light of the marked underrepresentation of specific Asian demographics in traditional randomized clinical trials []. By systematically analyzing both single-country studies (SCSs) and cross-country collaboration studies (CCCSs), this research aimed to delineate the current state of RWE generation and collaborative research initiatives for RWE from integrated databases across different nations in Asia. Our objectives also included obtaining a comprehensive understanding of the preference for RWD methodologies by contrasting the emphasis on comparative effectiveness research (CER) with descriptive studies and discerning the preferred and popular real-world research databases.
The cyclical interplay between a nation’s economic strength, health care infrastructure, and research capacity perpetuates disparities in RWD generation. We hypothesized that Asian countries with less extensively documented RWD research trends could be effectively clustered based on systematic patterns in RWD generation. This streamlined our objective to evaluate trends in RWD generation and shed light on targeted capacity-building strategies essential for informed health care policy making. Through this rigorous extended scoping research, we aimed to present insights that resonate with clinical stakeholders, medical researchers, and health policy makers, thereby guiding the formulation of strategies attuned to each nation’s health care challenges and research diversities and complexities.
Our research approach was methodically aligned with the guidelines set forth by the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) []. Our published protocol specified a preliminary focus on 3 countries—India, Thailand, and Taiwan—as a representative pilot to explore the diversity of health care systems and RWD use in Asia []. The outcomes from our initial study covering Taiwan, India, and Thailand have been previously published []. Relevant insights from the latter publication were incorporated into the archetyping of the nations wherever applicable. In this study, we expanded our protocol to 7 other countries. However, we maintained consistency with the original protocol’s methodological framework to ensure comparability across all countries studied. This expansion was aligned with our initial intent to potentially include more countries following the first study across 3 countries.
The search strategy is described in Table S1 in . We filtered our search to include only English-language publications from the last 5 years, aiming to highlight current and internationally relevant RWE or RWD. As the conversion of RWD to RWE emphasizes the stringent analytical processes necessary to yield valuable and credible findings, we intentionally chose to rely on PubMed as an exclusive source of relevant citations for screening. Our goal was to assess studies yielding robust RWD featured in esteemed, indexed, and peer-reviewed journals while reducing potential duplicates. By focusing solely on PubMed, we tried to identify research representing this standard and offering evidence of the utmost scientific integrity. This strategy aligns with the specifications outlined in our protocol [].
Screening Eligible Studies for Data AnalysisAll retrieved study abstracts were directly imported into the Covidence software (Veritas Health Innovation) for subsequent screening and data extraction. Studies were initially screened against predefined eligibility criteria to capture research from integrated RWD. The criteria encompassed 4 domains described in the original protocol: database type and requirement for research across >1 hospital or clinic, publication nature, RWD study type, and publication scope []. The scope of publication was adapted in this study to include citations on databases involving 1 of the target nations (Hong Kong, Indonesia, Malaysia, Pakistan, Philippines, Singapore, or Vietnam). The inclusion criteria also considered studies featuring nontarget countries as long as 1 of the 7 target nations was involved. Table S2 in provides a snapshot of the eligibility criteria used in this study.
Duplicate removal and a 2-step eligibility screening process were conducted in Covidence. The initial step (phase 1) assessed titles and abstracts, with relevant studies advancing to full-text evaluation in the second phase. Given the study volume, the screening for both phases was divided between 2 reviewers. An independent reviewer examined a random sample of 20% of the studies to maintain accuracy. Any ambiguities or discrepancies were collaboratively resolved, and another reviewer was consulted if needed. The final step involved data extraction and data analysis for eligible studies.
Data Extraction and AnalysisWe used Covidence for data extraction through a custom template that covered the following:
Basic study details: Covidence identifier (ID) based on the first author’s last name and publication year and title.Presence of cross-country collaboration in the research (CCCS or SCS).Nature of publication (clinical study or protocol).Study categorization: CER versus descriptive study (non-CER), with CER definitions adapted from Medical Subject Headings. We expanded the criteria for CER to standardize its meaning in the context of this research as the “studies comparing interventions and strategies (including the comparison between active and nonactive interventions and strategies) to prevent, diagnose, treat, and monitor health conditions using validated methods for confounders elimination, e.g., matching, and statistical adjustments like stratification, weighting, regression, instrumental variable analysis etc” [].Research source database classification involving medical records, health insurance claims, clinical registries, pharmacy claims, or composite databases.Disease specifics: name and area of the target disease under study (defined by primary diagnosis and pathophysiology or by prime medical specialty in charge if they intersected). The disease categories encompassed cardiology and metabolic disorders (CVM), oncology, inflammatory and autoimmune disorders, infectious diseases and vaccines (IDV), and others. These categories represent major research fields in clinical medicine with significant disease burdens, selected to provide pertinent insights into RWD and RWE applications within these critical domains.Outcome types: clinical (treatment effect or safety), cost, or patient-reported outcomes (PROs), with PROs capturing direct patient responses.Demographics (adults, children, or both), number of centers, study participants, and length of study and duration between last data collected and year of publication. The length or duration represented the span from the study’s commencement to completion as specified by the authors.The unique names for the databases used. When provided, the name of the specific database used in each study was collected and organized by target nation, database type, and disease area.A total of 2 reviewers collaboratively managed data extraction, and all extractions underwent quality checks by another reviewer to ensure the accuracy and reliability of the extracted data. However, this process was not conducted independently or blinded to the other reviewer’s decisions. Disagreements between the 2 reviewers were settled through discussions, and an additional reviewer was involved whenever there was a need for a consensus.
The final search was conducted on May 9, 2023, covering the preceding 5 years; to account for partial yearly data in 2018 and 2023, we calculated the equivalent of annual publication count using 365 multiplied by the average daily number of publications. We used linear regression, using the year as a continuous predictor variable, to understand the annual trend in nation study counts. This provided insights into the average annual trend in study numbers throughout the search period. To further even out year-to-year variations, a 2-year simple moving average (SMA) was applied to enhance the clarity of the data trends. This SMA approach was consistent with our previous research methodologies []. Given the study’s descriptive nature, there was no a priori statistical hypothesis. Statistical analyses were conducted to calculate point estimates and their associated errors. Categorical data were presented as frequencies and percentages, whereas continuous data were presented as means and SDs. We used Microsoft Excel (Microsoft Corp) for all data analyses. Adobe Illustrator was used for crafting high-definition figures for the main manuscript.
The search was conducted on May 9, 2023, and yielded 1483 studies with 1 duplicate. Of these 1483 studies, 553 (37.3%) were included in phase-2 screening, and 369 (24.9%) studies were eligible for data extraction ().
The vast majority of the publications (361/369, 97.8%) were original research, whereas the remaining 2.2% (8/369) were study protocols. The country-wise distribution of SCSs and CCCSs is illustrated in .
Figure 1. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flowchart for the selection of eligible studies. *Search date: May 9, 2023. EHR: electronic health record; EMR: electronic medical record; PCT: pragmatic clinical trial; RCT: randomized clinical trial. Geographic Distribution and Collaboration RelationshipAmong the 369 studies that qualified for data extraction, Singapore, Hong Kong, and Malaysia each contributed to ≥100 publications, with respective counts of 157 (42.5%), 130 (35.2%), and 100 (27.1%). The other 4 nations—Indonesia, Pakistan, Vietnam, and the Philippines—were involved in fewer publications, with 8.9% (33/369), 8.4% (31/369), 7.3% (27/369), and 6% (22/369), respectively, and each had >50% of their studies classified as CCCSs (Table S3 in ). Given their lower overall study numbers and the predominance of CCCSs, these 4 nations were categorized as global collaborators in certain subsequent analyses ( []).
Figure 2. Archetype by number of publications and percentage of cross-country collaboration studies (CCCSs) from real-world study databases in the last 5 years from 10 nations in Asia. Scatter plot archetyping for countries with a relatively high percentage of CCCSs but fewer total publications as global collaborators (highlighted within the dashed lines). Solo scholars cluster includes countries with a lower percentage of CCCSs (≤50%) and a higher number of publications (≥100 real-world data publications from integrated databases in the last 5 years). Countries such as Taiwan show a high number of total publications with a relatively low percentage of CCCSs, signifying a tendency to conduct independent research. The data for India, Taiwan, and Thailand were derived from our previous publication that used the same methodology as that used in this research []. Countries beyond the 7 target nations of this study that were involved in collaborations were labeled as nontarget countries. The cross-country collaboration network across the 7 target nations and nontarget countries is described in Table S4 in . The average number of collaborative countries (ANC) indicates the cross-country interconnection for research of a given nation. The ANC varied from 2.2 for Singapore to 4.1 for the Philippines. Despite having the lowest ANC, Singapore was involved in 30.3% (77/254) of the studies, making it the highest contributor to CCCSs. On the other hand, Malaysia, with a higher ANC of 3.1, participated in 50% (50/100) of CCCSs, making it the most engaged collaborator within the solo scholars cluster. Malaysia participated in 56% (10/18) of CCCSs from Pakistan, 75% (15/20) of CCCSs from Vietnam, 81% (21/26) of CCCSs from Indonesia, and 84% (16/19) of CCCSs from the Philippines.
Nontarget countries were common collaborators in CCCSs across all 7 target countries, with their involvement ranging from 84% (16/19) to 98% (43/44).
Time Trendand depict the yearly average counts and growth rates of SCSs and CCCSs across the 7 target nations from the solo scholar and global collaborator clusters, respectively. Between 2018 and 2023, every target nation except for the Philippines, which experienced a decline of –14.5%, exhibited an upward trend in the average number of studies published. Vietnam led with the steepest growth rate at 24.5%, trailed by Pakistan at 21.2%. Among solo scholars, there were growing trends for all 3 nations, with a growth rate of 6.2% for Hong Kong, 8.7% for Singapore, and 16.3% for Malaysia. Due to the small average number of studies by year in some of the individual global collaborator nations, growth trends are also presented collectively for global collaborators as a cluster in . Duplicate studies within clusters were adjusted; thus, the sum of CCCSs from all 4 nations might be larger than the number of CCCSs of the cluster as whole. The growth rate was 21.2% in the global collaborators cluster after adjusting for duplicates.
Figure 3. Annual trends in average number of publications by geographical distribution for solo scholar nations. Malaysia showed the highest annual growth rate of 16.3% among the solo scholars cluster, followed by Singapore (8.7%) and Hong Kong (6.2%). 
Figure 4. Annual trends in average number of publications by geographical distribution for global collaborator nations. Vietnam exhibited the most significant increase in total studies among the 7 target countries, followed by Pakistan and Malaysia. The Philippines was the only nation with a decline, whereas Indonesia and Hong Kong maintained a consistent study count. Vietnam and, to a lesser degree, Indonesia and Pakistan demonstrated a rising participation in cross-country collaborative research. Duplicate studies within the cluster were adjusted, and thus, the sum of cross-country collaboration studies (CCCSs) from all 4 nations might be larger than the number of CCCSs in the cluster as whole. Overall AttributesOverviewand show the primary attributes of the eligible studies from each target nation by SCS and CCCS. The variables presented were study type, disease domain, data source, outcomes, participant demographics, length of study duration, time gap from last data collected to publication, sample size, and number of research centers.
Table 1. Study characteristics in the solo scholar cluster across the 7 target countries by single-country study (SCS) and cross-country collaboration study (CCCS; n=369)a.Study characteristicsSingapore (n=157)Hong Kong (n=130)Malaysia (n=100)aStudy numbers for database types and study outcomes may appear as duplicates; hence, the total percentage may not add up to 100. CCCS numbers may appear as duplicates for studies conducted in multiple target countries. The percentages may add up to less or more than 100 because of rounding.
bCER: comparative effectiveness research.
cCVM: cardiology and metabolic disorders.
dIDV: infectious diseases and vaccines.
eIAD: inflammatory and autoimmune disorders.
fEMR: electronic medical record.
gEHR: electronic health record.
hPRO: patient-reported outcome.
Table 2. Study characteristics in the global collaborator cluster across the 7 target countries by single-country study (SCS) and cross-country collaboration study (CCCS; n=369)a.Study characteristicsIndonesia (n=33)Pakistan (n=31)Vietnam (n=27)The Philippines (n=22)aStudy numbers for database types and study outcomes may appear as duplicates; hence, the total percentage may not add up to 100. CCCS numbers may appear as duplicates for studies conducted in multiple target countries. The percentages may add up to less or more than 100 because of rounding.
bCER: comparative effectiveness research.
cCVM: cardiology and metabolic disorders.
dIDV: infectious diseases and vaccines.
eIAD: inflammatory and autoimmune disorders.
fEMR: electronic medical record.
gEHR: electronic health record.
hPRO: patient-reported outcome.
iNot applicable.
Study Type: CER or Non-CER (Descriptive)Of the 369 studies, 221 (59.9%) were CER studies, with the remaining 148 (40.1%) being non-CER or descriptive. The relative representation of CER versus non-CER for SCSs and CCCSs is illustrated in . Singapore, Hong Kong, Malaysia, and Vietnam had a higher number of CER studies in both their SCSs and CCCSs. Vietnam’s SCSs had the predominant CER representation at 71% (5/7), followed by Singapore at 68% (54/80) and Hong Kong at 66% (57/86). Among CCCSs, Hong Kong led with 70% (31/44) CER studies, followed by Pakistan with 67% (12/18) and Singapore with 57% (44/77). There were more descriptive non-CER studies in SCSs from Pakistan, the Philippines, and Indonesia, resulting in the CER study percentages being 8% (1/13), 33% (1/3), and 43% (3/7), respectively.
shows the yearly trends of CER percentages from 2018 to 2023 broken down by SCS and CCCS. The consistency in trends was more noticeable in SCSs across the 7 target nations compared to CCCSs. An upward trend in CER study percentage was observed in SCSs from Hong Kong and the global collaborators. Conversely, Malaysia’s SCSs experienced a steady decrease in CER contribution over the same period.
Figure 5. Trends in comparative effectiveness research (CER) by single-country studies (SCSs) and cross-country collaboration studies (CCCSs). Trends in CCCSs across the 7 target countries showed more consistency compared to trends in SCSs. The SCSs of Hong Kong and the global collaborators showed an upward trend in CER percentages, whereas Malaysia’s SCSs consistently decreased in CER contribution. The count of CCCSs represents the individual contributions of each country, leading to a total count across countries that exceeds the actual number of CCCSs due to some studies involving multiple collaborators. The 2-year SMA trends for CER and descriptive studies are illustrated in for the biennial average from 2018 to 2023. Hong Kong consistently increased its CER contributions in both SCSs and CCCSs, increasing from 47% (9/19) between 2018 and 2019 to 73% (24/33) between 2022 and 2023 for SCSs and similarly from 61% (11/18) between 2018 and 2019 to 73% (8/11) between 2022 and 2023 for CCCSs. Other notable rises in CER contributions in CCCSs were observed in Malaysia (from 4/12, 33% between 2018 and 2019 to 12/16, 75% between 2022 to 2023), Indonesia (from 3/9, 33% between 2018 to 2019 to 7/10, 70% between 2022 and 2023), Pakistan (from 1/4, 25% between 2018 and 2019 to 8/9, 89% between 2022 and 2023), and Vietnam (from 4/9, 44% between 2019 and 2020 to 6/7, 86% between 2022 and 2023). Conversely, Malaysia’s SCSs saw a consistent decline in CER contribution over the 5 years, dropping from 67% (8/12) between 2018 and 2019 to 53% (10/19) between 2022 and 2023. Furthermore, all of Pakistan’s SCSs (12/12, 100%) were non-CER between 2018 and 2022.
Database TypeOf the 369 studies, 341 (92.4%) used a single database. Exclusive use of clinical registry databases was most common at 50.9% (188/369), followed by electronic medical records (EMRs) or electronic health records (EHRs) at 39.3% (145/369), health insurance and administrative claims at 1.4% (5/369), and pharmacy claims at 0.8% (3/369). The use of multiple databases was found in 7.6% (28/369) of the studies, primarily combining clinical registries and EMRs or EHRs (). Use of EMR or EHR databases was more common for SCSs (120/246, 48.8%; ). On the other hand, the predominant exclusive database warehouse for CCCSs was clinical registries, used in 73.2% (9/123) of the studies. EMRs’ or EHRs’ contribution to CCCSs was lower, representing only 20.3% (25/123) of CCCSs, which is considerably lower than their share in SCSs ( and and ).
The use of the clinical registry database type consistently dominated across all CCCSs from all target nations, whether used on its own or in combination with other databases. For SCSs, (1) there were more clinical registries over EMRs or EHRs used in Indonesia, Malaysia, and the Philippines—the figures were 57% (4/7) versus 29% (2/7), 78% (39/50) versus 22% (11/50), and 67% (2/3) versus 33% (1/3), respectively; (2) Singapore’s use was almost even, with 58% (46/80) of the studies using clinical registries and 54% (43/80) using EMRs or EHRs; and (3) conversely, Hong Kong, Pakistan, and Vietnam used more EMRs or EHRs than clinical registries—80% (69/86) versus 26% (22/86), 69% (9/13) versus 23% (3/13), and 57% (4/7) versus 14% (1/7), respectively.
reveals the evolution of EMR or EHR contributions, both exclusively and in combination with other databases, in the previous 5 years. In SCSs, Malaysia and the global collaborators experienced a consistent decline in EMR or EHR use, whereas Hong Kong exhibited an increase. Malaysia’s EMR or EHR use in SCSs remained consistently at <50% during this period. In contrast to SCSs, where EMR or EHR use was predominant, EMR or EHR use in CCCSs from the target countries was always at <50% from 2018 to 2023, and no consistent time trend pattern was observed.
Figure 6. Trends in percentage of use of medical records by single-country studies (SCSs) and cross-country collaboration studies (CCCSs). A decline in exclusive or combined use of electronic medical records (EMRs) or electronic health records (EHRs) was observed for SCSs in Malaysia and the global collaborators, whereas Hong Kong saw an increase; in contrast, use of EMR or EHR databases in CCCSs was consistently below 50%, with no uniform trend emerging during this period. The count of CCCSs represents the individual contributions of each country, leading to a total count across countries that exceeds the actual number of CCCSs due to some studies involving multiple collaborators. The 2-year SMA over the previous 5 years indicated that the exclusive use of EMRs or EHRs in SCSs from the 7 target nations increased from 46% (26/56) to 60% (49/82). In contrast, the reliance on clinical registry databases dipped from 50% (28/56) to 38% (31/82); ). For CCCSs, the distribution between clinical registries and EMRs or EHRs remained relatively steady, with clinical registries being the most common ().
Disease AreaThe leading medical research area was CVM, accounting for 36.9% (136/369) of the studies, trailed by oncology and IDV, each with 14.9% (55/369). Inflammatory and autoimmune disorders was the least prevalent area, representing 6.2% (23/369) of the studies. The remaining 27.1% (100/369) of the studies pertained to various other diseases. The proportion of CVM studies grew from 28% (11/39) in 2018 to 39% (14/36) in 2023, peaking in 2020 with 49% (36/73). Conversely, the share contributed by IDV medical area increased from 8% (6/73) in 2020 to 25% (21/84) in 2022 and 17% (6/36) in 2023, surpassing oncology as the second most common disease and therapeutics research area in recent years ().
Study OutcomesMost of the studies (348/369, 94.3%) presented clinical outcomes whether in terms of treatment effect or safety. There were 5.7% (21/369) of the studies that discussed cost outcomes, PROs, or a combination of these with clinical results. In the SCS category, every study from Pakistan focused on clinical outcomes, whereas cost outcomes were observed in SCSs from all countries except Malaysia and Pakistan. One study each from Hong Kong (1/86, 1%) and Malaysia (1/50, 2%) included PRO outcomes. In the CCCS category, none of the selected nations published studies focusing on cost outcomes ( and ).
Study PopulationOf the 369 studies obtained, 273 (74%) investigated adults, 24 (6.5%) focused on the pediatric age group, and 72 (19.5%) encompassed both adult and pediatric participants. Notably, in the SCS category (), Pakistan (9/13, 69%) and the Philippines (2/3, 67%) reported higher proportions of mixed populations than of solely adult participants (Pakistan: 3/13, 23%; the Philippines: 1/3, 33%).
Pediatric representation in the CCCSs was 8.1% (10/123), slightly higher than in SCSs (14/246, 5.7%). Within CCCSs (), Vietnam led in pediatric-focused research with 15% (3/20) of the studies, followed by Indonesia (3/26, 12%) and Pakistan (2/18, 11%).
Study DurationInformation about study duration was reported for 94.6% (349/369) of the studies. The average duration was 7.4 (SD 6.3) years, ranging from 0.01 to 35.8 years. In total, 3.5% (13/369) of the studies had a duration of >20 years. The mean for SCSs was higher at 7.5 (SD 6.0) years compared to that for CCCSs at 7.1 (SD 6.8) years ().
Among SCSs, the Philippines () topped the list with the longest average study duration of 12.0 (SD 8.2) years. As shown in , Hong Kong followed closely with an average of 9.8 (SD 7.3) years. Conversely, Pakistan and Indonesia registered the shortest mean study durations with 2.1 (SD 1.9) years and 3.1 (SD 3.3) years, respectively. Over a 5-year span, based on a 2-year rolling average, the study duration in Indonesia showed an uptick, increasing from 0.6 years between 2018 and 2019 to 1.9 years between 2022 and 2023. Other target countries did not exhibit any consistent study duration trend patterns ().
For CCCSs, Pakistan () led with the longest mean study duration of 9.1 (SD 11.2) years, closely followed by the Philippines with 8.4 (SD 9.8) years. Malaysia () recorded the shortest average study duration at 5.5 (SD 7.4) years. The study duration in Indonesia’s CCCSs averaged 7.1 (SD 9.6) years, which was notably longer than that of its SCSs. Observing trends, there was a decline in the mean study duration of CCCSs in Malaysia, Pakistan, the Philippines, and Vietnam. Conversely, the average study duration in Singapore’s CCCSs steadily rose, increasing from 5.6 years between 2018 and 2019 to 9.1 years between 2022 and 2023 ().
Lag Between the Research Period and PublicationOf the eligible studies, most (205/369, 55.6%) were published between 2 and 5 years after the time of latest available data studied, 28.5% (105/369) were published after >6 years, and 9.5% (35/369) were published within 2 years. The remaining 6.5% (24/369) of the studies had unspecified year or years of research completion ().
Eligible studies from all the target nations showed a similar trend, with most (205/345, 59.4%) being published within 2 to 5 years after the research period. However, in both Singapore and Hong Kong, the time taken from research completion to publication was notably longer for both SCSs and CCCSs, averaging 5.8 (SD 3.0) years and 5.1 (SD 2.9) years for SCSs and 4.8 (SD 2.5) years and 5.0 (SD 2.6) years for CCCSs, respectively ().
It is worth noting that 15.9% (17/107) of CCCSs were published within 2 years of the research period but only 7.6% (18/238) of SCSs were published within this time frame. We observed upward trends in the time to publication within 2 years from 2018 to 2023. For SCSs, it was from 3% (2/63) to 15% (12/83), and for CCCSs, it was from 16% (6/38) to 20% (5/25; ).
The publication time lag also varied according to the RWD source (). Among studies that relied on a single database, the highest percentage of those published within 2 years after the research consistently used the EMR or EHR database type. This trend held true for both SCSs and CCCSs, with EMRs or EHRs dominating the quick turnaround for publications. Specifically, among SCSs, 12.8% (15/117) of studies using the EMR or EHR database type were published within this 2-year time frame, whereas only 2% (2/94) of those using clinical registry databases had the same publishing speed. CCCSs had a consistent pattern—26% (6/23) of the studies that used EMRs or EHRs were published within 2 years, in contrast to the 14% (11/78) of the studies that used clinical registry databases. Notably, no studies published within the 2-year window used the health insurance and medical claims database type or the pharmacy claims database type.
Study Size: Sample Size and Number of CentersThe sample size was specified in 98.1% (362/369) of the studies and varied considerably, ranging from as few as 16 to >154,500,000. The average sample size was 672,352 (SD 8,364,280). The average in CCCSs was much higher at 1,824,035 (SD 14,530,091) compared to 106,029 (SD 397,050) in SCSs (). The 2-year SMA of sample size in SCSs indicated an increasing trend from 51,706 to 178,675 from 2018 to 2023. In contrast, a sharp decline was observed in CCCS sample sizes from 5,543,271 to 548,107 during the same period (). Hong Kong had the highest average sample size in SCSs (205,006, SD 607,767) as well as in CCCSs (4,187,122, SD 23,979,119; and ).
The number of participating centers was only specified in 44.4% (164/369) of the studies. The mean number of centers was 44.3 (SD 120.3), ranging from 2 to 1119. As noted in and , Hong Kong reported the highest average number of study centers at 42.7 (SD 60.2) for SCSs, followed by Pakistan at 19.9 (SD 13.4) and Malaysia at 19.5 (SD 15.4). For CCCSs, Vietnam had the highest mean number of study centers at 181.9 (SD 236.9), which was higher than twice the overall CCCS mean of 78.4 (SD 178.6).
The 2-year SMA for the number of study centers in SCSs initially rose from 81.8 between 2018 and 2019, reaching a peak of 131.9 between 2019 and 2020 before declining to 99.1; 37.1; and, finally, 30.8 in the subsequent years. In CCCSs, there was a downtrend starting from 16.1 between 2018 and 2019 to 10.7 centers at its lowest point between 2020 and 2021 before bouncing back to 33.9 centers between 2022 and 2023 ().
Table S5 in provides and overview of the number of centers involved in RWD studies across various database types in the target countries. The Philippines reported the highest average number of centers (197.0) in studies using EMR or EHR databases, whereas Vietnam had the highest average (65.8) in studies associated with clinical registry databases. Information on the number of centers was not reported in many studies, particularly those using health insurance and claims databases.
Database Namesprovides the specific name or names of the databases used in each study organized by target country, database type, and disease area.
This scoping review was based on our earlier research completed for Taiwan, India, and T
Comments (0)