Chronic pancreatitis and pancreatogenic type 3c diabetes risk: a systematic review and meta-analysis.

Chronic pancreatitis (CP) is a progressive inflammatory disease of the pancreas characterized by irreversible fibrotic destruction of the pancreatic parenchyma, leading to a permanent loss of both exocrine and endocrine functions, a pathological cascade that culminates in exocrine insufficiency, metabolic dysregulation, and an increased risk of severe complications. 1 This multifactorial process is driven primarily by toxic factors (chronic alcohol use, smoking) and genetic mutations, which disrupt protease regulation and predispose the pancreas to autodigestion. 2 The clinical presentation of CP is highly heterogeneous. While in some patients, the disease is asymptomatic and is discovered incidentally, in others, it manifests with severe pain syndrome, steatorrhea, and loss of body mass. 3 The epidemiology of CP varies significantly by region, influenced by factors such as alcohol consumption, smoking, gallstone disease, metabolic disorders, and genetic predispositions. According to the Global Bremen Disease Survey 2021, the total number of reported cases of pancreatitis—including both acute and chronic forms—increased from 1.73 million in 1990 to 2.75 million in 2021. These central Asian regions, which include Kazakhstan, have some of the highest incidence rates in the world. 4 Furthermore, metabolic factors, including hyperlipidemia and hypercalcemia, also play a role by exacerbating inflammation and encouraging the formation of calcifications and scar tissue in the pancreas. 5 , 6 The clinical course of CP is fraught with debilitating complications, including persistent abdominal pain, pancreatic exocrine insufficiency leading to malnutrition, and the development of pancreatic pseudocysts. 7 Furthermore, patients face risks of biliary obstruction, metabolic complications such as osteoporosis and osteopenia, 8 and a significantly elevated lifetime risk of pancreatic adenocarcinoma. 9 Rare but serious complications include splenic vein thrombosis, pancreatic ascites, and pancreaticopleural fistulas. 10 One of the most significant and debilitating long-term complications of CP is the development of pancreatogenic type 3c diabetes mellitus (T3cDM). T3cDM is a form of secondary diabetes that arises from the primary disease of the exocrine pancreas, which subsequently damages the insulin-producing islet cells. 11 The primary etiological factor for T3cDM is chronic pancreatitis, with studies reporting a diabetes prevalence ranging from 25% to 80% in CP patients depending on the cohort and duration of follow-up. 12 Other causes include pancreatic carcinoma, cystic fibrosis, hemochromatosis, and previous pancreatic surgery. 13 The diagnostic criteria for T3cDM, as per recent consensus, include the presence of pancreatic exocrine insufficiency, pathological pancreatic imaging, and the absence of an autoimmune etiology typically associated with type 1 diabetes. Minor criteria for diagnosis include the assessment of beta-cell function, the absence of insulin resistance, impaired incretin secretion, and low serum concentrations of fat-soluble vitamins (A, D, E, and K). 14 It is often misdiagnosed as type 2 diabetes mellitus (T2DM) due to overlapping features, yet it is distinct in its pathophysiology, clinical presentation: a higher propensity for hypoglycemia due to concurrent glucagon deficiency, and management requirements. Such diagnostic errors lead to suboptimal therapy: instead of the timely initiation of adequate insulin therapy, patients are often prescribed oral hypoglycemic agents. These medications are not only ineffective in the setting of absolute insulin deficiency but are also potentially dangerous, as they increase the risk of hypoglycemia or diabetic ketoacidosis. 15 Despite growing recognition, significant gaps remain in our understanding of T3cDM. The epidemiological data on the prevalence of T3cDM, persistently recirculated across publications, require critical re-evaluation. The widely cited prevalence of 0.5%–1.15% for North America originates from research in the late 1980s and has been perpetuated in subsequent reviews citing these outdated sources. 11 Similarly, the understanding of its prevalence in regions like Southeast Asia is also built on foundational but older sources, including seminal research and book chapters on fibrocalculous pancreatopathy (Tropical Chronic Pancreatitis) from the 1980s and 1990s. 16 However, more recent analyses demonstrate that the actual prevalence of T3cDM may reach 5%–10% among all diabetes cases in Western populations. 17 Consequently, the prevailing narrative significantly underestimates the true epidemiological significance of this disease. Furthermore, there is an acute lack of recent data on the prevalence of T3cDM in other regions of the world, particularly in Eastern Europe and Asia. If we consider that a common cause of T3cDM is CP or other diseases of the pancreas, the prevalence of this type of diabetes among patients with CP is of particular interest. A recent meta-analysis revealed a pooled prevalence of new-onset DM after a diagnosis of CP of 30% (and 17% for insulin-dependent DM), demonstrating an increase in risk from 15% within the first 36 months to 33% after 60 months of observation. 18 Nevertheless, comprehensive epidemiological data on T3cDM as a whole still require urgent updating. The primary aim of this systematic review and meta-analysis is to determine the pooled frequency and risk of developing diabetes mellitus (DM) in patients with a confirmed diagnosis of chronic pancreatitis. By quantitatively synthesizing the available global literature, this study seeks to provide a definitive estimate that will inform clinicians, researchers, and healthcare policymakers, ultimately improving screening strategies and early intervention for this vulnerable patient population.

This systematic review and meta-analysis were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines 19 and was prospectively registered in the PROSPERO database (registration number: CRD420251136931). A comprehensive systematic literature search was conducted to identify all relevant published studies. The search was performed across three major electronic databases: Scopus, Web of Science Core Collection, and PubMed, covering the period from each database’s inception to August 2025. The search strategy was designed to capture all studies pertaining to the outcomes of CP and DM. A combination of controlled vocabulary (MeSH terms) and free-text keywords was used to maximize the sensitivity of the search strategy ( Table 1 ). The full database-specific search strategies are provided in Table S1. Search strategy across databases for chronic pancreatitis and pancreatogenic type 3c diabetes risk. Bibliographic records, including complete metadata and cited references, were exported from each database and imported into EndNote citation management software for preliminary processing. A rigorous deduplication process was undertaken, where records were sorted by title and Digital Object Identifiers (DOIs), and duplicates were removed. To ensure a focus on primary human research, the screening process involved the exclusion of specific publication types. Consequently, review articles, case reports, study protocols, conference abstracts, and animal studies were deemed ineligible. It is important to note that gray literature sources, such as clinical trial registries or dissertations, were not searched; this review focused exclusively on published, peer-reviewed journal articles. The resulting final dataset was exported as a CSV file, which served as the foundation for the subsequent screening and selection process. The screening was conducted in a multi-stage, independent manner to ensure reliability. The first reviewer performed an initial screening of titles and abstracts against the predefined eligibility criteria. The entire library and screening decisions were then verified by a second independent reviewer. Any discrepancies were resolved through discussion or by a third adjudicator. Studies selected by both reviewers advanced to full-text assessment and were subsequently included in the quantitative synthesis ( Figure 1 ). Flow diagram of the study selection process for the systematic review and meta-analysis on the risk of pancreatogenic type 3c diabetes in chronic pancreatitis. The inclusion criteria for this review comprised prospective or retrospective cohort studies conducted in adult patients (⩾18 years) with clinically and/or instrumentally confirmed CP and without pre-existing diabetes at baseline. Eligible studies were required to have a minimum follow-up duration of 6 months and to report outcomes related to the incidence of T3cDM and/or the proportion of patients requiring insulin therapy. Studies were excluded if they did not explicitly confirm the absence of pre-existing diabetes or prediabetes at baseline, or if they failed to report the occurrence of T3cDM. Additional exclusion criteria included studies in which diabetes was diagnosed more than 2 years prior to the clinical onset of chronic pancreatitis, the presence of a family history of insulin-dependent DM, or the diagnosis of pancreatic cancer within the first 2 years after the diagnosis of chronic pancreatitis. Studies with a follow-up period shorter than 6 months, non-original research articles (such as reviews, case reports, and conference abstracts), as well as animal studies, were also excluded. Studies focusing exclusively on postoperative outcomes following pancreatic surgery were excluded unless the incidence of diabetes attributable to chronic pancreatitis could be clearly distinguished from surgery-related endocrine insufficiency. A standardized data extraction form was utilized to collect relevant information from each included study. The extracted data encompassed the following: first author’s name, publication year, country of origin, study design, duration of follow-up, total cohort size, mean or median age, percentage of male participants, body mass index (BMI), incidence of T3cDM, and the number of patients requiring insulin therapy. The methodological quality and risk of bias of the included cohort studies were assessed using an adapted Newcastle–Ottawa Scale (NOS). Because the primary analysis was a single-arm incidence meta-analysis among patients with chronic pancreatitis rather than a direct exposed-versus-non-exposed comparison, the “selection of the non-exposed cohort” and “comparability” items were considered not applicable and were not included in the total score. Six applicable items were scored: representativeness of the chronic pancreatitis cohort, ascertainment of chronic pancreatitis diagnosis, confirmation that diabetes was absent at baseline, assessment of incident diabetes, adequacy of follow-up duration, and completeness of follow-up. Each applicable item was assigned 1 point when adequately reported, giving a maximum score of 6. Studies scoring 5–6 were considered high quality, 3–4 moderate quality, and 0–2 low quality. The full item-level NOS assessment, including not-applicable items, is presented in Figure 2 for transparency. Two reviewers independently performed the quality assessment, and disagreements were resolved by consensus or consultation with a third reviewer. (a) Item-level heatmap of Newcastle–Ottawa Scale (NOS) quality assessment for cohort studies on chronic pancreatitis and pancreatogenic type 3c diabetes risk. (b) Summary distribution of “Yes,” “No,” and “NA” responses across the eight NOS items in the same set of studies. NOS, Newcastle–Ottawa Scale. A random-effects model was used for all quantitative syntheses because clinical and methodological heterogeneity was expected across studies. For each study, the incidence proportion of new-onset diabetes was calculated as the number of incident diabetes cases divided by the total number of chronic pancreatitis patients at risk. Pooled proportions and 95% confidence intervals were calculated using inverse-variance random-effects models. Between-study heterogeneity was assessed using Cochran’s Q test, τ², and the I ² statistic, with I ² values above 75% interpreted as substantial heterogeneity. Prediction intervals were reported where appropriate to indicate the expected range of effects in future comparable settings. Prespecified subgroup analysis was conducted according to follow-up duration: 120 months. Differences between subgroups were evaluated using a random-effects test for subgroup differences. Because some studies reported outcomes at more than one follow-up timepoint, the follow-up-stratified analysis was treated as exploratory and interpreted separately from the primary pooled incidence estimate. Meta-regression analyses were conducted in RStudio using the meta and metafor packages to explore whether study-level covariates explained between-study heterogeneity. Candidate moderators included mean age, proportion of male participants, smoking prevalence, alcohol-related etiology, BMI, follow-up duration, abdominal pain prevalence, and pancreatic calcification. Regression coefficients, p values, residual heterogeneity, and the proportion of heterogeneity explained by the model were reported. Because these analyses used aggregated study-level data and involved a limited number of studies, they were interpreted as exploratory. Small-study effects and potential publication bias were assessed for the primary outcome by visual inspection of a funnel plot and Egger’s regression test, because more than 10 studies were available. Formal small-study-effect testing was not performed for the insulin-requiring diabetes outcome because fewer than 10 studies contributed data, making funnel plot and regression-based tests unreliable.

The database search identified 24,101 records. After the removal of duplicates, the remaining records were screened by title and abstract. Studies that clearly did not meet the eligibility criteria were excluded at this stage. A total of 310 articles were eligible for full-text review. Of these, 291 studies were excluded for predefined reasons, including ineligible study design, absence of chronic pancreatitis-specific outcome data, presence of pre-existing diabetes at baseline, insufficient follow-up duration, or insufficient data for quantitative synthesis. The final systematic review included 19 studies, of which 18 were eligible for quantitative meta-analysis of diabetes incidence after chronic pancreatitis ( Figure 1 ). The mean age of patients ranged from 25 to 62.8 years, and the proportion of men was 75.2%. In addition to the outcome, we extracted the following data from each article included in the review: country of study, study design, patient age, cohort size, total number of DM cases, number of insulin-requiring new-onset diabetes mellitus (NODM) cases, follow-up duration, and diagnostic criteria used to verify DM and CP. Detailed characteristics of the extracted data are presented in Table 2 . Characteristics of studies related to chronic pancreatitis and pancreatogenic type 3c diabetes risk included in the meta-analysis. Data expressed as median, mean (SD), or range. CP, chronic pancreatitis; CT, computed tomography; DM, diabetes mellitus; ERCP, endoscopic retrograde cholangiopancreatography; GAD, Glutamic Acid Decarboxylase; ICA, Islet Cell Antibodies; MRPT, magnetic resonance pancreatography; NODM, new-onset diabetes mellitus; OGTT, oral glucose tolerance test. The methodological quality of the included studies was assessed using the NOS. The appraisal demonstrated the high methodological rigor of the included articles. Mean scores were 5.5/6 for the NOS. Majority of studies (58%) were rated as high-quality based on the NOS ( Figure 2 ). The primary random-effects meta-analysis of incident diabetes after chronic pancreatitis included 19 studies and 11,585 participants. The pooled incidence was 30% (95% CI: 27–34), with substantial heterogeneity ( I ² = 94.8%, p  < 0.0001) and a prediction interval of 15%–49% ( Figure 3 ). The wide prediction interval indicates that the incidence of diabetes may vary considerably across clinical settings and study populations. Forest plot of diabetes mellitus incidence in chronic pancreatitis. In the exploratory analysis stratified by follow-up duration, diabetes incidence increased with longer follow-up. The pooled incidence was 18% (95% CI: 14–22) for follow-up 120 months. The test for subgroup differences was statistically significant ( p  = 0.0184), supporting a time-dependent pattern; however, this analysis should be interpreted cautiously because some cohorts contributed more than one follow-up timepoint, and residual heterogeneity remained high ( Figure 4 ). Forest plot of diabetes mellitus incidence in chronic pancreatitis by follow-up duration. Among patients who developed diabetes, a separate analysis of seven studies including 968 participants showed that 59% (95% CI: 46–71) required insulin therapy. Heterogeneity was also high for this outcome ( I ² = 91.7%, p  < 0.0001), and the prediction interval was wide (19%–93%), indicating substantial variation in insulin requirement across study populations and clinical practice settings ( Figure 5 ). Forest plot for the pooled prevalence of insulin-dependent diabetes mellitus following chronic pancreatitis. Leave-one-out sensitivity analysis demonstrated that sequential exclusion of individual studies did not materially alter the pooled incidence estimate ( Table S2 ). Across all iterations, the recalculated pooled estimates remained consistent with the primary analysis, indicating that no single study exerted a dominant influence on the overall results. These findings support the robustness and stability of the pooled incidence estimate. A meta-regression analysis was performed to identify factors potentially influencing the variability (heterogeneity) of the pooled estimates of diabetes incidence. Covariates included demographic, behavioral, and clinical characteristics aggregated at the study level. The analysis revealed no significant influence of demographic or behavioral factors on diabetes incidence: proportion of males (β = −0.614, p  = 0.275), prevalence of alcohol use (β = −0.213, p = 0.689), or proportion of smokers (β = −0.476, p = 0.509). The coefficient of determination ( R ²) was 0.0% for all these covariates, indicating their inability to explain the observed between-study variability. Meta-regression analysis additionally evaluated the influence of age, follow-up duration, and the prevalence of abdominal pain reported in the included cohorts. No statistically significant association was observed between mean age and the incidence of diabetes after chronic pancreatitis ( Figure 6(a) ). Similarly, study-level follow-up duration and pain prevalence did not demonstrate statistically significant moderating effects in the meta-regression models ( Figure 7(b) and ( c )). However, follow-up duration remained clinically important and was therefore further explored in the predefined subgroup analysis described above. High and statistically significant residual heterogeneity ( I ² > 97%, p  < 0.0001 for the QE test) confirmed that sources of variation remained unaccounted for in these models. Analysis of clinical factors revealed a more nuanced picture. Although the association between the proportion of patients with pancreatic calcification and diabetes risk did not reach conventional statistical significance (β = 1.245, p  = 0.146), a strong positive trend was observed. Notably, this was the only covariate that explained part of the variability ( R ² = 10.14%), suggesting its potential clinical relevance as an effect moderator. The association between BMI and diabetes incidence was also not statistically significant (β = −0.115, p  = 0.529). However, this result should be interpreted with caution due to the limited number of studies ( k  = 8) included in this model, which may have resulted in insufficient statistical power to detect a true effect ( Figures 6 and 7 ). Meta-regression analyses of demographic and behavioral moderators of diabetes incidence after chronic pancreatitis. (a) Mean age of study participants; (b) proportion of male participants; (c) smoking prevalence; and (d) alcohol-related etiology of chronic pancreatitis. Meta-regression analyses of clinical moderators of new-onset diabetes after chronic pancreatitis. (a) BMI; (b) study follow-up duration; (c) prevalence of abdominal pain; and (d) proportion of pancreatic calcifications. BMI, body mass index. Thus, the performed meta-regression analysis, while not revealing a statistically significant influence of specific demographic or behavioral factors (proportion of males, smoking, alcohol consumption), confirms that the observed high heterogeneity ( I ² > 97%) reflects substantial clinical and methodological variability among the included studies. The inability of these factors to explain the variability ( R ² = 0.0%) is consistent with the notion that the heterogeneity is likely attributable to other causes: significant variation in sample sizes, diversity in the etiology of CP (alcoholic, biliary, hereditary, idiopathic) within the pooled cohorts, and the critical dependence of diabetes incidence on follow-up duration, as demonstrated in the subgroup analysis. The partial explanation of variability ( R ² = 10.14%) by the proportion of patients with pancreatic calcification points to the potential role of structural organ damage as one moderator of risk. However, for definitive conclusions and to identify all significant sources of heterogeneity, further studies with larger datasets, unified follow-up protocols, and patient stratification by CP etiology are required. Visual inspection of the funnel plot for the primary outcome did not suggest marked asymmetry ( Figure S1 ). Egger’s regression test did not provide evidence of small-study effects for incident diabetes after chronic pancreatitis (intercept = −0.77, p  = 0.715). Begg’s rank correlation test was also not statistically significant ( p  = 0.211). These results should be interpreted cautiously because funnel plot methods are less reliable in single-arm proportion meta-analyses and in the presence of substantial between-study heterogeneity. Formal small-study-effect testing was not performed for insulin-requiring diabetes because only seven studies contributed data.

This systematic review and meta-analysis aimed to determine the overall incidence and risk factors for the development of DM in patients with a confirmed diagnosis of CP. Analysis of 19 studies with a total of 11,585 participants yielded reliable and meaningful estimates. From a clinical perspective, the observed increase in diabetes incidence with longer follow-up suggests that patients with chronic pancreatitis may benefit from continued long-term glycemic monitoring. However, because the evidence is derived from observational cohorts with substantial heterogeneity, the present meta-analysis should not be interpreted as defining a universal screening interval. Rather, monitoring frequency should be individualized according to disease duration, pancreatic structural damage, exocrine insufficiency, symptoms, comorbidities, and local clinical guidance. The main finding of this systematic review and meta-analysis is that approximately one-third of patients with chronic pancreatitis developed diabetes during follow-up. This estimate is consistent with previous evidence, including the meta-analysis by Zhu et al., 18 which reported a pooled frequency of newly diagnosed diabetes of 30% after chronic pancreatitis diagnosis. Our updated synthesis supports the concept that endocrine dysfunction is a common long-term complication of chronic pancreatitis. Nevertheless, because the included studies were observational and highly heterogeneous, the pooled estimate should be interpreted as an average across diverse cohorts rather than a precise risk applicable to every clinical setting. In the meta-analysis conducted by Zhu et al., a clear temporal dependence was revealed: the frequency of diabetes development increased over time from 15% with less than 3 years of follow-up to 33% with more than 5 years of follow-up. This trend is statistically significant ( p  = 0.0252) 18 and underscores the progressive nature of the damage to the endocrine apparatus of the pancreas in CP. The data we obtained not only fully align with the previously identified temporal relationship but also provide further granularity: we observed a progressive increase in the frequency of diabetes, from 18% (95% CI: 0.14–0.22) during a follow-up period of ⩽5 years, to 25% (95% CI: 0.15–0.35) over a period of 5–10 years, and up to 41% (95% CI: 0.31–0.60) after more than 10 years of follow-up ( p for subgroup differences = 0.0184). These results are consistent with the pathophysiological model in which continuous inflammatory and fibrotic processes gradually destroy the pancreatic islets of Langerhans. 39 Our follow-up-stratified analysis suggested a time-dependent increase in diabetes incidence, from 18% before 60 months to 44% after more than 120 months. This pattern is biologically plausible because chronic inflammation, fibrosis, ductal obstruction, and progressive parenchymal destruction may gradually impair islet-cell function. However, the subgroup analysis was exploratory and residual heterogeneity remained substantial. Therefore, these findings support the need for awareness of increasing long-term risk but do not establish an exact time threshold for screening or intervention. Among patients who developed diabetes after chronic pancreatitis, 59% required insulin therapy. This finding is compatible with the pathophysiology of pancreatogenic diabetes, in which progressive pancreatic damage can reduce insulin secretion and may also impair counter-regulatory glucagon responses. However, insulin requirement varied widely across studies, as reflected by the broad prediction interval and high heterogeneity. Therefore, this result should be interpreted as evidence of a substantial treatment burden rather than as a recommendation for universal or early insulin initiation. Treatment decisions should remain individualized and based on glycemic status, pancreatic function, nutritional condition, hypoglycemia risk, and clinical guidelines. The extremely high heterogeneity ( I ² > 94%) significantly limits the precision of pooled estimates and suggests that the reported incidence should be interpreted as an average effect rather than a universally applicable risk. The heterogeneity likely arises from differences in study design (prospective vs retrospective), diagnostic criteria for both CP and diabetes, geographic variation, and differences in follow-up duration. Although subgroup analysis partially addressed follow-up duration, residual heterogeneity remained substantial, indicating that additional unmeasured factors contribute to variability. Importantly, the wide prediction interval (15%–49%) reflects uncertainty in estimating future outcomes in individual clinical settings. The meta-regression analysis conducted showed that demographic and behavioral factors such as the proportion of males, alcohol consumption, and smoking did not explain the observed variability ( R ² = 0.0%). This suggests that the risk of developing diabetes is equally inherent to all etiological subtypes of CP (alcoholic, biliary, idiopathic, hereditary) and is not directly dependent on the primary damaging factor once the disease has manifested. 40 The lack of statistically significant associations for sex, smoking, and alcohol-related etiology should be interpreted cautiously. These findings may reflect insufficient statistical power due to the limited number of studies included in the meta-regression models rather than a true absence of effect. Additionally, aggregation at the study level may obscure individual-level associations, a known limitation of meta-regression analyses. Our analysis revealed a non-significant positive trend between pancreatic calcifications and an increased risk of diabetes ( p  = 0.146), accounting for 10.14% of outcome variability. This trend is pathophysiologically plausible: calcification represents a marker of advanced fibrosis and acinar tissue destruction, which correlates with parallel islet apparatus damage and endocrine insufficiency development. 41 These findings align with existing literature. Liu et al. 31 demonstrated that morphological changes in the main pancreatic duct (including stenoses, stones, and dilation) constitute an independent diabetes risk factor in idiopathic chronic pancreatitis. Similarly, Hao et al. 26 reported that in alcoholic CP patients, pancreatic calcifications were associated with earlier development of both exocrine and endocrine insufficiency compared to idiopathic forms, suggesting calcification serves as an indicator of more aggressive disease progression and consequently higher diabetes risk. The lack of statistical significance and the persistence of high heterogeneity, despite accounting for calcification, can be attributed to methodological and clinical differences between studies. Variability in diagnostic methods, patient characteristics, and evaluation criteria significantly contributes to the observed heterogeneity. To clarify the role of calcifications, further studies with unified assessment protocols and standardized diagnostic criteria are needed. A clinically important distinction should be made between diabetes arising during the natural history of chronic pancreatitis and diabetes developing after pancreatic surgery in patients with CP. Although both fall within the spectrum of pancreatogenic diabetes, postoperative endocrine insufficiency may reflect the combined effects of underlying chronic inflammatory destruction and loss of pancreatic parenchyma due to resection. Therefore, pooled estimates in CP cohorts should be interpreted in light of whether surgical patients were included and how postoperative outcomes were classified. Several limitations inherent to the present systematic review and meta-analysis should be considered when interpreting the findings. First, significant methodological heterogeneity was observed across the included studies. Variations in diagnostic criteria for both CP and T3cDM such as the use of different imaging modalities [CT, MRI, ERCP, Endoscopic Ultrasound (EUS)], functional tests (fecal elastase, secretin stimulation), and glycemic criteria (fasting glucose, OGTT, HbA1c)—may have influenced the consistency and comparability of incidence estimates. Differences in follow-up duration, frequency of monitoring, and definitions of diabetes onset further contribute to clinical and methodological diversity, as reflected in the high statistical heterogeneity ( I ² > 94%) observed in our pooled estimates. Second, although the overall methodological quality of the included studies was moderate to high, many studies lacked adjustment for potential confounding factors such as BMI, smoking status, alcohol consumption, and genetic background. The absence of multivariable adjustment in several cohorts may limit the ability to isolate the independent effect of CP on diabetes risk and could introduce residual confounding. Third, publication bias may be present, as studies reporting significant or positive associations between CP and diabetes incidence are more likely to be published. Potential publication bias and small-study effects were assessed for the primary outcome using funnel plot inspection and Egger’s regression test. No statistically significant evidence of small-study effects was detected. However, these methods have important limitations in single-arm proportion meta-analyses, particularly when between-study heterogeneity is substantial. In addition, formal small-study-effect testing was not performed for the insulin-requiring diabetes outcome because fewer than 10 studies were available. Fourth, the diagnostic overlap between T3cDM and T2DM poses a challenge. Despite applying strict inclusion criteria to exclude pre-existing diabetes, misclassification bias may persist due to incomplete baseline screening or retrospective study designs. This could lead to an overestimation of T3cDM incidence. Not all included studies clearly separated nonsurgical CP from surgically treated CP, which may have contributed to between-study heterogeneity and may affect interpretation of the pooled incidence as representing the natural history of CP alone. The high heterogeneity and observational nature of included studies limit causal inference.

This systematic review and meta-analysis suggests that chronic pancreatitis is associated with a substantial incidence of pancreatogenic diabetes, with an overall pooled incidence of approximately 30% and a higher incidence with longer follow-up. Among patients who developed diabetes, insulin therapy was frequently required, although estimates varied considerably across studies. Because the included evidence was observational and highly heterogeneous, these findings should be interpreted cautiously and should not be considered proof of causality or a basis for fixed surveillance intervals. The results support long-term glycemic monitoring, careful differentiation of pancreatogenic diabetes from other diabetes types, and individualized management of patients with chronic pancreatitis. Further prospective studies using standardized diagnostic criteria and follow-up protocols are needed.

Supplemental material, sj-docx-1-tae-10.1177_20420188261469942 for Chronic pancreatitis and pancreatogenic type 3c diabetes risk: a systematic review and meta-analysis by Mengtay Makashova, Amin Tamadon, Yerlan Bazargaliyev, Khatimya Kudabayeva, Raikul Kosmuratova, Nadiar M. Mussin and Ramazon Safarzoda Sharoffidin in Therapeutic Advances in Endocrinology and Metabolism