The study followed a scoping review methodology outlined by the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines [20,21]. For the purpose of this review, MLTCs were defined based on a consensus process involving experts in the field and reported in detail previously [22,23]. This definition includes 56 LTCs selected for their chronic, long-term impact on health, requiring ongoing management or treatment, based on established diagnostic criteria, significant population prevalence, and known effects on morbidity and mortality. The list of conditions is provided in Table S1 in Multimedia Appendix 1.

A review protocol for this study does not exist in any public domain.

Systematic searches were conducted in 5 electronic databases: OVID MEDLINE, Embase, CINAHL, the Cochrane Database of Systematic Reviews, and Bielefeld Academic Search Engine on November 27, 2025, for studies published from inception to date. These databases were selected to capture a range of relevant studies in the biomedical, nursing, and behavioral sciences literature.

The search strategy is reported in line with the PRISMA-S guidelines [24]. Terms to include in our search strategy were identified based on previous similar or relevant studies, searches for MeSH (Medical Subject Headings) terms and keywords or index terms in each database, and support from a librarian at the University of Southampton. Initial searches were conducted in Ovid MEDLINE and Embase. Both databases were searched for MeSH terms and free-text keywords related to “remission,” “resolution,” and “electronic health records” or “electronic medical records,” and “chronic disease” or “long-term condition.” Searches were adjusted for both databases according to their specific indexing. We screened the first 100 records captured in each database to identify key terms used in the title and abstracts of these records. We then refined our search strategy based on our findings, including any additional terms identified in our initial screening. We applied our refined search strategy to each of our 5 databases, adapting the search strategy as required. The search strategies for each database are available in Multimedia Appendix 2. Reference lists of included studies were not searched for additional papers, and study authors were not contacted for any additional information. Gray literature searches were conducted mainly through Bielefeld Academic Search Engine database and Google Scholar searches to identify any additional relevant references (eg, conference abstracts and theses). We did not contact authors to identify additional sources.

Quantitative studies were eligible for inclusion if they were conducted using EHRs, focused on an adult (aged ≥18 y) population, and assessed 1 or more of the 56 LTCs agreed by consensus work within the definition of MLTC [22,23]. In line with most existing research on MLTC, an adult population was selected as MLTC is more common in adults [25], and the criteria used to define remission may vary for adults and children. Studies were excluded if the definition of remission was not reported or unclear. Studies on remission of cancer were excluded as this is already a well-researched area [26]. Registry data and studies that did not assess remission using EHRs were excluded. Studies that reported on remission of symptoms (eg, seizures) rather than remission of an LTC of interest were also excluded. Search limits for studies on humans were applied. No language restrictions were applied.

Search results were exported to EndNote (Clarivate) for deduplication and then imported into the Rayyan collaborative systematic review platform for screening and final study selection. Titles and abstracts of records were first screened against eligibility criteria independently by 2 reviewers (BB and HH). Discrepancies in study selection were resolved by discussion between the 2 reviewers. Full texts of potentially relevant records were sought, retrieved, and reviewed in detail by HH.

Data were extracted from each eligible study using a structured table developed a priori by HH (following discussions with the research team). Data extracted included the LTCs assessed, author name and publication date, country, study design, population studied, study aims, remission definition (eg, clinical criteria and biomarkers), and key findings related to remission. Data were extracted by HH, and 50% of the extracted data were validated by BB. Where a single study (from the same first author) generated multiple papers, reports, or abstracts, we included the paper with the most comprehensive definition of remission. We did not contact investigators to obtain or confirm information.

Quality of the identified studies was not assessed as this is not a requirement of scoping reviews.

A narrative approach was taken to summarize our findings from included studies. The total number of included studies, characteristics of studies (eg, study country, study design, population studied, and setting) was described. Studies were categorized into groups of interrelated conditions, and the number of studies that focused on each specific condition and the characteristics of the studies within each group summarized. Similarities and differences in definitions and methods used to identify remission of LTCs across studies were highlighted.

To ensure the clinical applicability of our findings, the results were discussed with a group of clinicians and data experts. These discussions were pivotal in refining the identified operational definition of remission within EHRs, incorporating insights from real-world clinical practice. Key considerations included the variability in clinical decision-making, the complexities of disease progression, and the practical limitations of current EHR systems. This collaborative process was essential for aligning our findings with the realities of patient care and the capabilities of existing health informatics systems.

We further cross-referenced any remaining conditions from our previously agreed list with expert clinical input to derive a refined list of long-term conditions amenable to remission assessment. This was an iterative process underpinned by clinical judgment to ensure both practical feasibility and clinical relevance. We then searched for relevant medical codes used to define remission (based on our agreed definitions) using Clinical Practice Research Datalink (CPRD) Aurum medical and product data dictionaries. CPRD Aurum contains routinely collected primary care data from practices across England [27]. We also searched coding systems (eg, GitHub repositories, London School of Hygiene and Tropical Medicine Data Compass) and published code lists for any relevant codes to ensure a complete code list. Where older code lists were identified, we used description terms from these code lists to identify relevant codes in our data.

Figure 1 illustrates the outcomes of the systematic search and selection process. A total of 2064 records were identified through our searches. Following deduplication (n=87/2064, 4.2%) and title or abstract screening, 252 (12.2%) records met the criteria for full-text review. Of those records, 53.2% (134/252) were excluded following full-text review. A total of 91 (36.1%) final studies were included in our review, which included 4 additional studies identified by the research team.

Included studies were published between 2008 and 2025. The majority of the included studies were conducted in the United States (49/91, 53.8%) [28-76] followed by the United Kingdom (n=13, 14.3%) [13,77-88] and China (n=4, 4.4%) [89-92]. Three (3.3%) studies were conducted in Australia [93-95] and Saudi Arabia [96-98]. Two (2.2%) studies each were conducted in Italy [99,100], Japan [101,102], New Zealand [103,104], Spain [105,106], and Turkey [107,108]. One (1.1%) study each was conducted in Austria [109], Belgium [110], Denmark [111], Finland [112], Kuwait [113], Pakistan [114], South Korea [115], and Sweden [116]. For 1 (1.1%) study, it was unclear which country the study was based [117]. The included studies comprised cohort studies (n=70, 76.9%) [13,28-31,33,35-38,40-42,45,47-51,53-69,71,72,74-79,82-84,86,89-92,95,97,99-106,109-112,114-117], retrospective reviews (n=14, 15.4%) [32,39,43,44,46,52,70,80,81,85,93,98,108,113], cross-sectional studies (n=6, 6.6%) [34,73,88,94,96,107], and an exploratory study (n=1, 1.1%) [87]. Most studies focused on a general adult population (aged ≥18 y) or did not specify an age range, with a small number specifically investigating older populations (eg, veterans or those aged ≥60 y; n=6, 6.6%) [34,35,38,41,81,89] or specific age ranges (n=6, 6.6%) [36,63,75,96,97,112]. Sample sizes ranged from 12 patients to 72.9 million adults. Summary characteristics of the included studies are presented in Textbox 1. Full details of all included studies are available in Supplementary Table 2 in Multimedia Appendix 2.

The risk of bias in studies was not assessed, in line with scoping review methodology.

The majority of the included studies focused on remission in inflammatory bowel disease (IBD; 45.1%, 41/91) [43-61,80-85,90-95,99-101,106,109-111,113,115,117], type 2 diabetes (n=15, 16.5%) [13,62-69,86,87,102-104,114] (one additionally assessing remission of hypertension [103]), and depression (n=15, 16.5%) [32-42,76,78,105,108]. Eight (8.8%) studies assessed remission of drug or alcohol misuse [70-75,107,112], and 3 (3.3%) studies each assessed remission of asthma [29-31] and multiple sclerosis [96-98]. One (1.1%) study each assessed remission in each of anemia [28], chronic kidney disease (CKD) [77], autoimmune pancreatitis [116], epilepsy [79], and heart failure [89]. Additionally, 1 (1.1%) study investigated remission in MLTCs [88].

Sheikh et al [103] assessed hypertension remission based on the absence of antihypertensive medication. We did not identify other studies that assessed remission of hypertension (rather than control). However, 1 (1/91, 1.1%) of our included studies (that assessed remission in depression) assessed hypertension control using blood pressure measurements at 6-month follow-up [39]. Different cutoffs were used for different age groups (18-59 and 60-85 y) and presence of diabetes.

Figure 4 presents an evidence map of remission definition across studies included in the review and highlights potential gaps in the literature. Our review highlighted a lack of studies that examined the extent to which varying follow-up periods or the use of different remission indicators (laboratory tests vs diagnostic remission codes) influenced remission rates, except for studies on depression and IBD that reported remission rates across different follow-up durations. Some studies (eg, on IBD) did not specify the absence of medication in their definition of remission, making it less clear whether patients were truly in remission and making cross-study comparisons difficult.

This review facilitated the identification of conditions for which remission is measurable within EHRs. Although the scoping review highlighted additional conditions with theoretical potential for remission, discussions with clinicians and CPRD data experts indicated that, in many instances, the lack of suitable EHR-derived metrics or insufficient sample sizes would preclude their meaningful inclusion. Accordingly, we restricted our final list to conditions for which remission was both plausibly measurable and supported by validated indicators within the CPRD dataset, recognizing that this pragmatic approach may omit some conditions capable of achieving remission. Table 1 presents our refined list of long-term conditions based on clinical input and availability of codes in EHRs.

This scoping review synthesized evidence on how remission of LTCs has been defined and operationalized within EHRs. We identified 91 studies [13,27-117] covering 12 LTCs, including type 2 diabetes [13,62-69,86,87,102-104,114], depression [32-42,76,78,105,108], IBD [43-61,80-85,90-95,99-101,106,109-111,113,115,117], anemia [28], epilepsy [79], CKD [77], autoimmune pancreatitis [116], alcohol and/or drug misuse [70-75,107,112], heart disease [89], asthma [29-31], hypertension [103], and multiple sclerosis [96-98]. The review improved our understanding of how remission can be identified and operationalized in EHRs as criteria used to define and identify remission in studies included diagnostic (remission) codes, validated symptom scales, biochemical markers, absence of condition-specific events, and discontinuation of treatment [13,27-117]. However, criteria and follow-up periods differed widely across studies, even within the same condition (Figures 2-4), highlighting a lack of standardized definitions for remission. Using our scoping review findings and discussions with clinicians and data experts, we propose standardized definitions for remission of LTCs that will improve identification in EHRs (Table 1).

Overall, our review indicates that remission of LTCs remains relatively underresearched as many relevant studies identified in our searches did not clearly define remission (Figure 1). Most studies were conducted in the United States [28-76] or the United Kingdom [13,77-88] and focused on IBD [43-61,80-85,90-95,99-101,106,109-111,113,115,117] and diabetes [13,62-69,86,87,102-104,114]. The lack of studies in some countries likely reflects differences in the extent of adoption of EHRs for clinical care, particularly for low-income countries where EHR adoption may be at its earlier stages [118]. These differences may also capture differences in clinician coding, with some studies indicating clinicians in the United States, for example, spend more time actively using the EHR [119]. Furthermore, we did not identify any studies on remission of some conditions known (and deemed by clinicians we consulted) to be amenable to remission, such as peptic ulcer disease, chronic urinary tract infection, chronic liver disease, and alcoholic liver disease, peripheral vascular disease, transient ischemic attack, stroke, tuberculosis, and thyroid disease (standardized definitions provided in Table 1). Our finding that the majority of studies described remission of certain conditions may reflect differences in the clinical focus of health conditions, research priorities, and policies across countries [118,120]. For example, in the United Kingdom, the Quality and Outcomes Framework—a pay-for-performance scheme—is associated with better recording and monitoring of certain long-term conditions (such as diabetes and asthma) in clinical practice [120].

Across studies, variation in remission criteria for the same condition may partly be explained by differences in health systems, as well as differences in the availability of remission information [118,121]. For example, settings with fragmented health systems may not reliably include criteria such as absence of hospitalization in their definition of asthma remission, whereas this would be possible for integrated systems with linked primary and secondary care data [121]. Diagnostic remission codes were used for only a few of the conditions, as may be expected due to a lack of remission codes for many conditions [88]. Remission codes, where they exist, may not be commonly used due to clinician coding burden and uncertainty [122,123]. Thus, it is important to include a range of criteria in the definition of remission to allow more accurate estimates of remission. Follow-up duration for remission also varied for studies, which may impact estimates of remission rates [124]. A better understanding of how remission rates vary over time (and any factors affecting remission over time) could improve understanding of observation periods required for remission definitions.

Previous reviews on remission have examined single conditions (eg, asthma [125,126], psoriasis [127], diabetes [128,129], rheumatoid arthritis [130], psychosis [131], and systemic lupus erythematosus [132]) and did not focus on EHRs. Our findings are consistent with previous studies [125-127] in that we observed varying definitions of remission of each LTC. Moreover, these previous studies identified that individuals with MLTCs were less likely to achieve remission of specific conditions [125,127]; however, the studies did not assess remission of comorbidities [125-127]. Our review offers an opportunity to advance understanding of remission of long-term conditions by providing consensus-driven definitions of remission of multiple LTCs [Table 1]. This will allow studies (including those focused on single conditions) to explore and consider the impact of remission in other comorbidities in a consistent, rigorous way that can allow cross-study comparisons. Unlike these previous reviews on remission, our study focuses on EHRs and presents practical approaches to define remission in real-world settings where barriers such as variation in coding, fragmented health care, and health system factors make it challenging to study remission [118-123]. Our review extends previous work on remission in MLTCs [88] by presenting a comprehensive approach to defining remission in EHRs beyond the use of remission and resolved codes to better identify remission and in turn improves population-based research and clinical outcomes of LTCs.

Studies on LTCs using EHRs, including those assessing prevalence and trajectories, often do not account for remission of conditions [88], which may have implications for our understanding of MLTC [133-135]. This may be partly due to a lack of consensus for identifying remission, as identified by this review. We combined findings from our review with discussions with a clinical team, data experts, and a review of existing codes for remission in EHRs. This comprehensive approach can be used to advance research in this area and improve understanding of remission, particularly in the context of MLTC [136]. Future work may validate whether remission, as identified in EHRs, reflects true remission rates [137]. As with all EHR-based studies, there is bias due to incomplete data or missing information from subgroups [138]. Future work will need to consider missing data handling techniques to address this in order to ensure remission rates are not overestimated (eg, due to patients not consulting or being tested). Further research on remission may improve understanding of which conditions are more likely to achieve remission in patients living with MLTC, as well as factors and outcomes associated with remission. This understanding can improve the evidence base on remission and allow clinicians and patients to work toward remission, as well as improve care and outcomes of MLTC [139]. Furthermore, the range of criteria used to define remission of LTCs (Figure 2) highlights a need for a comprehensive approach to clinical coding to allow accurate identification and monitoring of remission. Further research is needed to assess the reliability of EHRs in capturing remission and to explore the wider factors associated with the documentation of remission in health records.

This study methodology [20] offers a comprehensive and systematic examination of how the remission of LTCs has been defined and operationalized within EHRs. A key strength lies in our integrative approach, which combined evidence from the literature with expert consultation and real-world coding frameworks. This methodological triangulation enhances the relevance and applicability of our findings [136], particularly in the context of MLTCs, where the concept of remission may inform future intervention design and policy development. However, several limitations should be acknowledged. First, although we conducted thorough electronic and gray literature searches, it is possible that some relevant studies, particularly those not using MeSH terms related to EHR, may not have been captured. Additionally, studies involving pediatric populations, cancers, or LTCs outside our predefined list for MLTCs were excluded. This decision was informed by input from our public advisory group, who emphasized the need to focus on noncancer conditions due to the relative lack of understanding around their remission. Nonetheless, this narrowed scope limits the generalizability of our findings to other potentially resolvable conditions [140], and further research will be necessary to develop comparable frameworks for these excluded areas.

Second, for some LTCs, only a small number of eligible studies were identified (n≤1), resulting in limited evidence on which to base definitions of remission. As a result, our synthesis may reflect early or preliminary interpretations for these conditions rather than consensus-based definitions. Similarly, different study characteristics, as well as varying remission indicators and duration of follow-up, made it difficult to directly compare study findings [140]. There was a lack of studies comparing whether remission rates varied based on the choice of remission indicators. Finally, our operational framework for identifying remission was derived largely from coding structures available in the UK primary care and linked datasets. Conditions for which complete or long-term follow-up is less feasible in primary care were not included. As such, the applicability of our findings may vary in international contexts or within health care systems using different data architectures [140]. Future efforts should aim to validate and refine this framework across diverse settings and data environments to enhance its robustness and global utility.

This review demonstrates that the remission of multiple LTCs can be identified and operationalized using EHRs, although approaches to defining remission and the duration of follow-up varied considerably across studies. While several LTCs have been studied, many others with the potential for remission remain underrepresented in the literature. Our review is innovative, as it uses both evidence synthesis and consultation with clinical and data experts to propose practical approaches to define and implement remission in EHR-based research. The review offers standardized definitions for multiple LTCs. These findings lay the groundwork for improving the identification of remission, particularly in the context of MLTCs, where a better understanding of disease trajectories is critical. Due to variation in clinical coding and missing data (eg, due to lack of test results) [137,138], it is necessary to adopt a comprehensive approach and include all these indicators (as appropriate for the specified condition) to reliably capture remission and evaluate outcomes for patients. Future research may also report on how the choice of remission indicators (eg, diagnostic codes vs biochemical biomarkers) influences findings. Further research may apply, compare, and evaluate standardized definitions across different data sources to assess generalizability and further improve our understanding of the remission of LTCs.