Abstract
Latin America has experienced a rise in the prevalence and incidence of inflammatory bowel disease (IBD). Differences in IBD phenotype between Hispanics in Latin America and those in the USA have not been described. We conducted a systematic review with meta-analysis of population-based and cohort studies comparing the phenotype of ulcerative colitis (UC) and Crohn’s disease (CD) in Latin Americans and US Hispanics. A systematic search was conducted up to March 2019 using MEDLINE, EMBASE and Google Scholar. Inclusion criterion includes studies describing IBD phenotype in Latin Americans or in US Hispanics. Exclusion criterion includes prevalence or incidence studies not describing phenotype. A random effects model was chosen “a priori” for analysis of pooled proportions. A total of 46 studies were included from Latin America and 7 studies from the USA. The predominant IBD subtype in Latin America was UC with a more balanced UC:CD ratio noted in Puerto Rico (0.53) and Brazil (0.56). UC-related extensive colitis was more common in US Hispanics (0.64) than in Latin Americans (0.38), p<0.001. CD phenotype was similar between US Hispanics and Latin Americans. UC is the predominant IBD subtype in Latin America, with the exception of Puerto Rico and Brazil which demonstrate a more balanced UC:CD ratio. In UC, extensive colitis was more frequently seen in US Hispanics than in Latin Americans. CD phenotype was similar in both US Hispanics and Latin Americans.
Significance of this study
What is already known about this subject?
There is a growing incidence and prevalence of inflammatory bowel disease (IBD) in Latin America.
Hispanics in the USA are more likely to have ulcerative colitis (UC).
Based on a previous meta-analysis, Hispanics and non-Hispanic whites in the USA share a common IBD phenotype.
What are the new findings?
Hispanics in Latin America have a predominance of UC except in Brazil and Puerto Rico where the UC to Crohn’s disease (CD) ratio is nearly 1:1.
US Hispanics are more likely to have extensive UC than Latin Americans.
US Hispanics and Latin Americans share a common CD phenotype.
US Hispanics and Latin Americans have a similar need for surgery in both UC and CD.
How might these results change the focus of research or clinical practice?
Further research should be focused on identifying the different triggers for IBD within Latin American countries.
Due to the low use of biologics in Latin America, efforts to provide these medications to this underserved community should be sought.
INTRODUCTION
The incidence and prevalence of inflammatory bowel disease (IBD) is increasing in Latin America.1–3 Concurrently, a growing number of academic centers have been reporting the phenotype of ulcerative colitis (UC) and Crohn’s disease (CD) in this region. Environmental exposures4 such as industrialization and a Western diet5 have been considered to be influential in the development of this disease6 and Latin America has experienced an increase in both industrialization7 and westernization of the food industry.8 There is a need to further understand the phenotype of IBD in this region as it is plausible that evolving environmental and dietary factors are shaping the phenotypic characteristics of IBD in Latin Americans. The Latin American population is a heterogeneous group comprised of a blend of indigenous, African and European descent. It is conceivable that IBD phenotype differs within Latin America and this was explored in a recent systematic review which included all types of cohort and cross-sectional studies.2 Latin Americans also comprise the largest minority group9 and largest migrant group to the USA.10 On migration to the USA, Latin Americans are exposed to the Western diet and environmental factors which have been long considered to be influential in the development of IBD.4 5
Based on a recent meta-analysis, US Hispanics were more likely to have UC but shared a similar IBD phenotype to non-Hispanic whites.11 However, it remains unclear whether the phenotype seen in US Hispanics is similar to the one seen in Latin America. Thus, we also aimed to compare the pooled IBD Latin American phenotype with IBD phenotypes in US Hispanics. As secondary endpoints, we aimed to assess differences in age at diagnosis, presence of a family history of IBD, smoking status, medication use and surgical outcomes.
MATERIAL AND METHODS
Search strategy
Author (DJA) and a medical librarian conducted a systematic search up to March of 2019 using MEDLINE. The MESH and non-MESH terms used were as follows: (inflammatory bowel disease* OR ulcerative colitis OR crohn*) AND (Latin America OR Peru OR Ecuador OR Colombia OR Venezuela OR Brasil OR Brazil OR Argentina OR Uruguay OR Paraguay OR Bolivia OR El Salvador OR Panama OR Guatemala OR Mexico OR Honduras OR Nicaragua OR Chile OR Costa Rica OR Dominican Republic OR Puerto Rico OR Cuba). A similar search was conducted using EMBASE (online supplemental file 1). Google Scholar was used to gather unpublished data discovered during cross-referencing. We contacted authors if data were deemed pertinent but not accessible. Two authors (DJA, AC) performed a surveillance search from our group’s recent meta-analysis to gather studies including US Hispanics with IBD.11 The systematic review was reported according to the MOOSE guidelines (online supplemental file 2).12
Supplementary data
Supplementary data
Inclusion/Exclusion criteria
Two authors (DJA and AC) reviewed the studies for inclusion with an interobserver agreement of 94% and any disagreements in inclusion were resolved after a formal discussion. Authors JS and AC extracted the data from each study and compared it for accuracy. Disagreements were resolved with a third author (ST). Studies published in full, abstract form or unpublished work until March of 2019 were eligible for inclusion. Inclusion criterion includes population-based Latin American studies from countries where Spanish or Portuguese was the predominant language. The studies needed to describe IBD phenotype and there were no language restrictions. Each study was reviewed, and the following data were extracted when available: geographical location, study design, number of participants, location/behavior of disease, age at IBD diagnosis, family history of IBD, smoking history (former/current), medication use (eg, steroids, aminosalicylates, immunomodulators, biologics) and need for surgical intervention. Studies were excluded if IBD phenotype could not be integrated into the Montreal Classification. Incidence or prevalence studies which did not provide phenotype were also excluded. Authors JS and ST conducted the study appraisal using the AXIOS tool (online supplemental file 3).13
Supplementary data
Outcomes assessed
The main outcome of this study was to assess the phenotypic characteristics of UC and CD in Latin America according to the Montreal Classification.14 If the Montreal Classification was not reported among the authors, then the location of disease was grouped into the “best-fit” category that would approximate to the Montreal Classification. Particularly, proctosigmoiditis was commonly reported among Latin American studies and this was grouped with E1 disease (proctitis).
We also aimed to compare IBD phenotype between Latin Americans and US Hispanics. The term “US Hispanic” was used to define Hispanics residing in the USA, whereas “Latin Americans” defined Hispanics living in Latin America. Secondary outcomes were assessed when available and included age of IBD diagnosis as defined by the Montreal classification (A1: <17, A2: 17–40, A3: >40 years), smoking status, family history of IBD, aminosalicylate use, immunomodulator and biologic use.
Statistical analysis
STATA V.15 was used for statistical analysis. A random effects pooled proportion model was chosen as a priori for all analyses. The I2 and p value tested for heterogeneity and an I2 >70% or p value <0.05 was considered significant for high heterogeneity. Further sensitivity analysis was conducted when heterogeneity was present.
Categorical outcomes were analyzed using a prevalence rate and effect size (ES) between countries of origin. The pooled proportion of ES with 95% CIs were calculated for each outcome with respect to the Latin America country of origin and Hispanic American/Latin America. A meta-regression analysis was also conducted to determine the relationship between countries of origin. A pooled proportion (ES) was considered significant when the 95% CI did not include 1.00 and the meta-regression p value was less than 0.05.
RESULTS
The search yielded 1365 articles in MEDLINE and 4435 articles in EMBASE. A total of 4627 articles were left for screening after duplicates were removed. Description of further exclusion of studies is depicted in figure 1 (see the Search strategy section). A total of 46 of Latin American unique studies were used in the quantitative analysis for Latin America (tables 1 and 2). The following countries reported at least one study: Costa Rica,15 Argentina,16 17 Colombia,18 19 Chile,20–23 Cuba,24–26 Mexico,27–29 Puerto Rico,30–32 Uruguay,33–35 Peru,36–40 and Brazil.41–60 Seven US Hispanic studies61–67 were included. The US-based studies were gathered from a previously published systematic review11 with an updated surveillance search up to March 2019 which did not yield any new relevant studies.
Comparison of disease phenotype within Latin American countries
Predominant IBD subtype: Latin America
The predominant IBD subtype was UC and it was seen in nearly two-thirds of the IBD Latin American cohort (0.68) (figure 2) (see the Dominant IBD subtype in Latin American countries section). There were more balanced UC:CD ratios (1:1) noted in Puerto Rico (0.53) and Brazil (0.56); however, these findings did not reach statistical significance, p=0.16 (table 3).
UC disease extension: Latin America
Extensive colitis was more frequent in Argentina (0.67) and Mexico (0.54), p=0.006. UC location of disease was otherwise similar across Latin American countries (table 3).
Location of luminal disease and disease behavior for CD: Latin America
The highest proportions for colonic disease were seen in nearly one-half of patients from Uruguay (0.43), Chile (0.46), and Puerto Rico (0.50), whereas the lowest proportions were seen in Cuba (0.15) and Colombia (0.17), p=0.002. The location of disease in CD was otherwise similar across Latin American countries. Similarly, there were no statistically significant differences in disease behavior (table 3).
Comparisons of disease phenotype in Latin Americans versus US Hispanics
Predominant IBD subtype: Latin Americans versus US Hispanics
In Latin America, two-thirds of the IBD population had UC (0.68) which was similar to US Hispanics (0.62), p=0.38 (table 4).
UC disease extension: Latin Americans versus US Hispanics
Proctitis was present in nearly one-third of patients in Latin America (0.37) but this was less common in US Hispanics (0.08), p=0.003 (figure 3) (see the IBD subtypes: Latin America vs United States section). The fewer incidence of proctitis in the US setting is likely due to the use of topical medications as most studies reported in the USA were from tertiary care centers and these patients likely had exposure to topical medications prior to presenting to the tertiary care center. Another explanation is that dietary and environmental exposures in the USA may contribute to differences in phenotype between Latin Americans and US Hispanics. Left-sided colitis was equally seen in one-fourth of patients from both groups, p=0.82. Extensive colitis was more common in US Hispanics (0.64) than Latin Americans (0.38), p<0.001 (table 4).
Location of luminal disease for CD: Latin Americans versus US Hispanics
The predominant phenotype in Latin America was ileocolonic disease (0.37), which was also the predominant disease location in US Hispanics (0.46), p=0.88. Ileal disease was more commonly seen among Latin Americans (0.31) than US Hispanics (0.19), but this difference did not reach statistical significance, p=0.13. Colonic disease was similarly seen in one-third of Latin and US Hispanics, p=0.42. Upper gastrointestinal involvement was infrequently seen in both groups. Overall, there were no statistically significant differences in location of disease between Latin Americans and US Hispanics (table 4).
Disease behavior for CD in Latin Americans versus US Hispanics
Inflammatory behavior (B1) was predominant for Latin America and US Hispanics (0.51) vs (0.64), p=0.24, respectively. Stricturing disease was nearly two times greater in Latin Americans (0.22) than US Hispanics (0.12), but this did not reach statistical significance, p=0.12. Stricturing CD was likely more common in Latin Americans due to less access to biologics. Penetrating disease was similarly seen in one-fourth of patients for both groups, p=0.94. Perianal involvement was also similar with nearly one-third of patients showing this phenotype in both groups, p=0.82 (table 4).
Secondary outcomes
Age at IBD diagnosis
Latin Americans
In Latin America, Brazil, Colombia and Peru provided age of IBD diagnosis according to the Montreal Classification. In UC (0.52) and CD (0.58), the predominant age group was between 17 and 40 years. Among US-based studies, age of IBD diagnosis was not stratified according to IBD subtype and could not be determined.
Family history of IBD
Latin Americans
In Latin America, a family history of IBD was infrequent in both UC and CD (0.08) and (0.09), respectively, p=0.85 (table 4). Family history is likely infrequent given that this is a relatively new disease in Latin America. Among studies from the USA, family history data were not provided according to IBD subtype.
Smoking history
Latin Americans
In Latin America, a history of being a former or active smoker was greater in CD than UC, but this did not reach statistical significance (0.35) vs (0.25), respectively, p=0.35. Smoking status for US Hispanics was not stratified by IBD subtype.
Medication use
Latin Americans
In UC, steroids were used in nearly two-thirds of patients in Peru (0.65), Argentina (0.67) and Colombia (0.71). 5-Aminosalicylate (5-ASA) products were almost universally used, except in Peru where it was used in three-fourths of patients (0.76). Immunomodulators were used in one-fifth of patients from Brazil (0.20) and Chile (0.22); one-quarter of patients from Colombia (0.26) and Mexico (0.25), and nearly one-third of patients from Argentina (0.31). Colombia and Argentina had the highest use of biologics (0.10) and (0.16), respectively. Biologic use was the lowest in Cuba (0.01), Mexico (0.01), Peru (0.02), and Brazil (0.04) (table 5).
In CD, Argentina had the highest use of steroids (0.86). Steroids were used in nearly two-thirds of patients from Costa Rica (0.64) and Colombia (0.68). 5-ASA products were used in nearly three-quarters of patients from Chile (0.74) and Argentina (0.79) and two-thirds of patients from Peru (0.65) and Brazil (0.68). Immunomodulators were used in nearly one-half of patients from Brazil (0.50) and Chile (0.51); and three-fourths of patients from Costa Rica (0.79). Biologics were used in one-half of patients from Argentina (0.50) and infrequently used in Mexico (0.02) and Cuba (0.04) (table 5).
US Hispanics
In the USA, only one study stratified medication use according to Hispanic ethnicity and IBD subtype62; therefore, a pooled analysis for US Hispanics was not conducted.
Surgical outcomes
Latin Americans
In UC, Brazil (0.22), Puerto Rico (0.23) and Cuba (0.35) had the highest rates of need for surgery. In CD, need for surgery was highly prevalent across all Latin American countries but more pronounced in Peru (0.50), Puerto Rico (0.52), Mexico (0.73), and Cuba (0.81).
Latin Americans versus US Hispanics
In UC, surgery was needed in similar proportions in Latin America and US Hispanics (0.15) and (0.19), respectively, p=0.62. In CD, surgery was also needed in similar proportions between Latin America and US Hispanics (0.43) and (0.41), p=0.85, respectively (table 6).
Assessment for heterogeneity
Overall, high levels of heterogeneity occurred across Latin America and the USA. In Latin America, Brazil contained the most population-based and cohort studies (n=20), thus, sensitivity analyses were conducted within this group. A sensitivity analysis was conducted and the common factors that appeared to differentiate these studies were sample size, large effect size, and incomplete phenotype data. Based on these criteria, five studies45 50 52 55 58 from the UC analysis were removed from the meta-analysis. With this exclusion, heterogeneity was controlled (I2=0.00%). The same concept was applied when reviewing Brazilian patients with CD. The initial analysis found a high level of heterogeneity (I2=91.07%). The same five studies were removed from the analysis45 50 52 55 58 which again controlled the heterogeneity at I2=0.00%. While reviewing the US Hispanic UC studies, they again showed a high level of heterogeneity (I2=92.55%). Using the same selection/exclusion criteria, four studies were removed from the analysis.62 65–67 With this exclusion, heterogeneity in patients with UC within the USA was reduced to I2=47.73%. Studies that evaluated CD within the USA were also found to have high levels of heterogeneity (I2=92.5%). Based on the previous selection/exclusion criteria, five studies were removed from the analysis.61 62 64–66 With this exclusion, heterogeneity in US Hispanics with CD was controlled (I2=0.00%). Results remained unchanged after sensitivity analyses.
DISCUSSION
Based on this meta-analysis, we found that the predominant IBD subtype in Latin America is UC; however, we found that Puerto Rico and Brazil had a more balanced UC to CD ratio (1:1). Within Latin America, Argentina and Mexico are more likely to have extensive UC, however, Latin Americans as a whole are less likely to have UC extensive disease as compared with US Hispanics. In UC, extensive colitis is associated with a higher need for biologics, hospitalizations, increased risk of colorectal cancer and higher rates for colectomy.68 A plausible hypothesis for extensive UC being more common in US Hispanics than Latin Americans may stem from the environmental and dietary triggers found in the USA, which are often absent in Latin America.4 5 69–71 The implication of this finding is that certain Latin American groups with UC (eg, Argentineans and Mexicans) and US Hispanics may have a more complicated course. This increased risk may serve as a prognostic factor which can aid in clinical decision-making for these cohorts.
We also found notable differences in CD phenotype within Latin American countries. Colonic CD was seen in nearly one-half of patients from Uruguay, Chile, and Puerto Rico, and these cohorts may be at a higher risk for the development of CRC.72 Aside from colonic CD, there were no statistically significant differences in the location of disease nor disease behavior in Latin America. Latin America as a whole had a predominance of ileocolonic CD which was found to be similar to US Hispanics. Ileocolonic CD has also been found to be the predominant CD disease location in Caucasians, Asians and African Americans.73 While CD was overall less common in Latin America, it initially appears to carry the same disease behavior as US Hispanics. However, a confounding factor is that disease modifying drugs (eg, biologics) are not readily accessible in Latin America as compared with the USA.74
In Latin America, there was wide use of 5-ASA products in CD despite their questionable efficacy in inducing and maintaining remission.75 Biologics were used in 5% of patients with UC and 25% of patients with CD. In contrast, between the years of 2007 and 2015, the USA has increased biologic use from 5.1% to 16.2% in UC and from 21.8% to 43.8% in CD.76 A likely reason for the disparity in the use of biologics between the USA and Latin America is the cost, which has been estimated to be $41,109 per year.76 Private insurance companies in Brazil, Mexico and Argentina do not cover biologics or biosimilars.74 In our meta-analysis, we found that within Latin America, biologic use was inversely proportional to need for surgery in both UC (eg, Cuba, Brazil) and CD (eg, Mexico, Cuba). We hypothesize that a subset of IBD Latin American patients with moderate to severe disease are living with suboptimal controlled disease due to lack of access to biologics. However, Latin America as a whole had similar surgical outcomes in both UC and CD when compared with US Hispanics. It is conceivable that with greater biologic use in Latin America, the surgical outcomes would have been better in the Latin American group.77 More importantly, the low access to biologics seen in Latin America should raise a public health concern. While biologics have been shown to improve outcomes in IBD,77 the costs associated with them make them highly inaccessible to a poorer Latin America healthcare system.78
In Latin America, the predominant age of diagnosis was between 17 and 40 years for both UC and CD. This is similar to the age of IBD diagnosis seen in the Caucasian population.79 However, the time lag between symptom onset and IBD diagnosis could not be answered with this meta-analysis and it could be a confounder if Latin Americans are having delays in diagnoses. Among Latin Americans, the presence of a family history of IBD was infrequent in both UC 8% and CD 9%, whereas a recent systematic review estimated the proportion to be 12% in Caucasians.73 It is plausible that IBD in Latin Americans and US Hispanics is driven primarily by environmental factors rather than genetics, as a positive family history of IBD was infrequent in this meta-analysis. In our previous meta-analysis, we also found that a family history of IBD was less frequently seen in US Hispanics as compared with non-Hispanic whites.11 Smoking habits (current/former smoker) were greater in Latin American patients with CD as compared with patients with UC, but this did not reach significance.
A strength of this meta-analysis is we analyzed the Latin American population per country which allowed us to bring out differences within Latin America. We also compared Latin Americans as a whole as compared with US Hispanics. This has not been previously done. Limitations of this meta-analysis include the high degree of heterogeneity seen across the Latin American countries. We were able to determine certain factors that increased heterogeneity which were sample size, large effect size, and a few studies had incomplete phenotype descriptors. After the sensitivity analysis, the heterogeneity was reduced. Other potential contributors to heterogeneity include variations in environmental exposures, and the admixed background of Latin Americans. There are focal concentrations of European ancestry in certain parts of Brazil80 81 and more widespread predominance of European ancestry in countries like Uruguay82 and Argentina.83 In contrast, countries such as Peru and Mexico have a predominant indigenous population.84 In addition, the number of published studies were variable per country, and this may have created a selection bias; however, we accounted for this by searching for unpublished articles. The US studies were also not stratified according to first-generation or second-generation Hispanic but this was not feasible as only one study performed this stratification in the USA.61 The US studies also did not differentiate between Latin American Hispanics as compared Hispanics from Spain. This is unlikely to have affected the results as Spaniards only make up 1.3% of the total Spanish-speaking population in the USA.85 While there are dietary and environmental variations within Latin America, these components are often balanced via meta-analysis.
CONCLUSION
Hispanics in Latin America have a predominance of UC over CD, except in Brazil and Puerto Rico, where the UC:CD ratio is more evenly balanced. Countries such as Argentina and Mexico are more likely to have extensive UC, but US Hispanics as a whole are more likely to have extensive UC than Latin Americans. CD colonic disease is more likely to affect Uruguay, Chile, and Puerto Rico but Latin Americans as a whole have a similar CD phenotype when compared with US Hispanics. There is great variability in the use of biologics in Latin America and many countries with low access to biologics have a higher need for surgery. Future studies should be aimed at studying the genetic and environmental factors accounting for the phenotypic differences in Latin Americans and US Hispanics.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information. This is a meta-analysis. Data were gathered from previously published and unpublished studies.
Ethics statements
Patient consent for publication
Ethics approval
This study does not involve human participants.
Footnotes
Twitter @DannyJAvalos, @JinendraSatMD
Contributors Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work: DJA, JS, AKD, CD, LA, ST, AC. Drafting the work or revising it critically for important intellectual content: DJA, JS, MZ. Final approval of the version to be published: DJA, JS, MZ, AKD, CD, LA. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: DJA, JS, MZ, AKD, CD, LA, AC. Responsible for the overall content as the guarantor: DJA.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.