Vaccination Against Measles, Mumps, Rubella and Incident Inflammatory Bowel Disease in a National Cohort of Privately Insured Children

Abstract

Background

Infection is believed to be a potential trigger for inflammatory bowel disease (IBD). Whether vaccination against childhood infections including measles, mumps, and rubella may reduce risk of IBD is uncertain.

Methods

We conducted a retrospective cohort study using de-identified claims data from a national private payer (Optum Clinformatics Data Mart). Eligible infants were born between 2001 and 2018 and were continuously enrolled with medical and pharmacy coverage from birth for at least 2 years (n = 1 365 447). Measles, mumps, and rubella vaccination or MMR is administered beginning at 12 months of age. Cox proportional hazard regression models were used to compare time with incident disease in children by category of vaccination, after adjustment for sex, birth year, region of country, history of allergy to vaccines, and seizure history.

Results

The incidence of early pediatric IBD increased between 2001 and 2018. Ten percent (n = 141 230) of infants did not receive MMR, and 90% (n = 1 224 125) received at least 1 dose of MMR. There were 334 cases of IBD, 219 cases of Crohn’s disease, and 164 cases of ulcerative colitis. Children who had received at least 1 dose of MMR had lower risk for IBD than children who did not (hazard ratio, 0.71; 95% confidence interval, 0.59-0.85). These associations did not change after further adjustment for childhood comorbid conditions, preterm birth, or older siblings affected with IBD. Similar associations were observed for MMR with Crohn’s disease and ulcerative colitis, although these did not reach statistical significance.

Conclusion

MMR is associated with decreased risk for childhood IBD.

Introduction

Inflammatory bowel disease (IBD), consisting of Crohn’s disease and ulcerative colitis (UC), currently affects approximately 1.3 million people in the United States.¹ Over the past 2 decades, the incidence of pediatric IBD has increased globally,² particularly in children younger than 7 years of age.³ The explanation for these increases is not entirely clear, although it is likely due to a range of environmental factors including earlier detection; changes in the microbiome resulting from changes in antibiotic prescription patterns and utilization of appendectomy⁴; and viral gastrointestinal infections.^5–7

The measles virus infects a range of immune cells including antigen-presenting cells and B and T cells; it also leads to antibody derangements against a range of infections including and in addition to measles.⁸ Thus, it is possible that childhood vaccination against measles, mumps, rubella and varicella could be protective against IBD due to reduced risk of these specific infections and reduced risk of other infections that might alter the gut microbiome. The first dose of the measles-mumps-rubella vaccine or measles-mumps-rubella-varicella vaccine (both called MMR in this report) is recommended for children at age 12 to 15 months,⁹ and a single dose confers >93% prevention.¹⁰ However, studies conflict regarding the impact of MMR upon IBD risk, noting protective,^11,12 null,^13,14 or even greater risk of IBD¹⁵ after MMR.¹⁶ The inconsistent results may be due to historically high MMR vaccination rates along with the fact that pediatric IBD typically presents in adolescence rather than early childhood, thus limiting the statistical power to detect associations.

MMR has declined in recent years and particularly during the coronavirus 2019 (COVID-19) pandemic: in one network, the average proportion of 16-month-old children with MMR declined from 72% (March 2017 to March 2020) to 62% in August 2020.¹⁷ Such declines have been attributed to a number of factors other than the pandemic, including parental concerns regarding risks for chronic diseases after MMR.¹⁸ Therefore, it is important to examine whether MMR might confer risk for IBD. Using a large national dataset including more than 2 million children—the largest examination of its kind to date—we examined whether receipt of MMR was associated with decreased risk of IBD in a cohort whose vaccine status had been recorded from birth. We hypothesized that after adjustment for confounders between receipt of vaccination and IBD, MMR would be associated with decreased risk.

Materials and Methods

We conducted a retrospective cohort study using de-identified claims data from Optum Clinformatics Data Mart, a product of OptumInsight, Inc. (Eden Prairie, MN). The administrative claims dataset includes more than 30 million commercially insured individuals each year from a geographically diverse US population (16% West, 20% Midwest, 36% South, and 27% Northeast) beginning January 1, 2001, and continuing to December 31, 2020.¹⁹ These claim files include both commercially insured patients and Medicare Advantage enrollees and cover the full set of service utilization such as inpatient hospitalizations, emergency department visits, other outpatient services, pharmacy claims, and laboratory claims. All enrolled beneficiaries must maintain both medical and pharmacy coverage throughout their insurance enrollment. The University of Michigan’s review board declared the study exempt based on its use of existing de-identified data.

We included children who had both medical and pharmacy coverage and who were born between January 1, 2001, and December 31, 2018, who were continuously enrolled in the health plan from birth to 18 months of age. MMR receipt was defined as having a Current Procedural Terminology (CPT) or ICD-9/10 procedure code indicating receipt of each component (measles, mumps, rubella with or without varicella; Supplemental Table 1). Inflammatory bowel disease was determined through ICD-9/10 diagnosis codes (ICD-9-CM codes: 555.x for Crohn’s disease or 556.x for UC) for a single inpatient encounter or 2 outpatient claims on different days.²⁰ Optum data and medical record review have high concordance for childhood rotavirus vaccination and diphtheria-tetanus-pertussis vaccinations, although MMR has not been compared against medical records.^21,22 MMR vaccine administration is lower in Optum than parental self-reported rates in the National Immunization Survey, although it was unclear whether this was due to parental over-reporting of vaccinations or under-recording of MMR vaccinations.²³ The algorithm we used to identify IBD in pediatric and older persons is similar to those used in other studies that find comparable rates between claims data and epidemiologic estimates, as well as medical record review.^24,25 In Optum,²⁶ the use of 2 claims results in slightly lower pediatric prevalence than reported in Canada³ during a similar time frame and thus may under-represent IBD cases or reflect differences in detection and coding between the United States and Canada. Thus, these criteria were expanded to include 2 claims from different days or a single hospitalization; in sensitivity analyses, restricting the definition to 2 claims yielded a similar pattern of results and significance (Supplemental Table 3), and thus the expanded definition was used for the main results.

Cox proportional hazards regression was used to compare time to IBD in children by category of MMR. In our base model, we adjusted for sex, birth year, region of country, and conditions that might confound decisions to vaccinate, namely history of allergy to vaccines, seizure history, and health care utilization as indicated by attendance of a well-child visit (Model 1). We observed that all children who had a diagnosis of IBD, Crohn’s Disease or UC also had a well-child visit within the first 2 years; only 2989 children did not have a well-child visit. Thus, we constructed models with and without well-child visit as an adjuster and found a similar pattern of results, and so regression models including only children with well-child visits are presented.

We conducted several additional analyses. First, to determine if childhood frailty might be a confounder, an additional model adjusted for pre-term birth and childhood comorbidity as fixed covariates (Model 2). We used the modified claims-based version of the childhood chronic conditions score,²⁷ which incorporates neuromuscular, cardiovascular, respiratory, renal, gastrointestinal, hematologic or immunologic, metabolic, other congenital or genetic defects, and malignant neoplasms between birth and 2 years. Second, since parents might be less likely to vaccinate if they already had another child with IBD, we also adjusted for older sibling history of IBD using the approach described by Jain et al²³ (Model 3). Older siblings needed to also have been continuously enrolled for at least 6 months between the beginning and end of the study period (January 1, 2001, through December 31, 2019). Older siblings of index children were identified using a family identifier variable associated with the insurance policy; siblings had to be between 6 months and 17 years older than the index child to be included. For examination of the association between MMR and subtypes of IBD, similar models were constructed. Seizure history and allergy to vaccination were modelled as monthly time-varying covariates.

We also conducted several sensitivity analyses. Because IBD is uncommon for children younger than 5 years of age, we restricted the analyses only to children who enrolled for at least 5 years. We also adjusted for antibiotic exposure; specifically, any oral or intravenous antibiotic use was included as a time-varying covariate. Models were constructed for UC and Crohn’s disease.

Statistical significance testing of unadjusted rate ratios was conducted using the Yates X2 test, and statistical significance testing of hazard ratios (HRs) estimated by maximum likelihood were conducted using Wald X2 statistics. Likelihood test ratios were used to test the statistical significance of Cox proportional hazards models with and without interaction terms. Proportional hazards assumption was assessed using the log rank test. All adjusted hazard ratios were estimated by maximum likelihood. All statistical tests were 2-sided, and the alpha for all tests was 0.05. Analyses were performed using SAS version 9.4 (SAS Institute Inc).

Results

Table 1 shows the characteristics of the study sample by MMR status. Almost all of the children in the study population had undergone at least 1 well-child visit between birth and 2 years of age. Children who were not vaccinated with MMR were enrolled for a shorter amount of time compared with children who received at least 1 dose of MMR. At the end of the enrollment period, children who had received both MMR vaccinations had a median age of 6 years, with an interquartile range of 5 to 9 years and a range of 2 to 19 years compared with children who had not received any MMR vaccination, who had a median age of 3 years with an interquartile range of 3 to 4 years and a range of 2 to 19 years. Other factors associated with longer length of enrollment included greater comorbidity including seizure history, vaccine allergies, and chronic conditions score (Supplemental Table 2). Children who did not receive MMR were less likely to report histories of seizures, vaccine allergies, or preterm birth and had a lower burden of chronic conditions than children who did receive MMR. The proportion of children who had older siblings with IBD, Crohn’s disease, and ulcerative colitis was low and did not differ markedly by MMR status.

Children who had received both MMR vaccines had a median age of 8 years at IBD diagnosis, with an interquartile range of 5 to 11 years compared with children who had received no MMR, who had a median age of 4.5 years at IBD diagnosis, with an interquartile range of 3.5 to 8 years. Table 2 shows the associations between MMR status and risk of IBD after adjustment for additional variables including sex, year of birth, history of seizures, vaccine allergy, geographic region, and well-child visit (Model 1). Compared with children who had not received MMR, children who received MMR had a reduced risk of IBD. The protective association for MMR persisted after further adjustment for childhood comorbidity, as represented by preterm birth and childhood chronic conditions score (Model 2), as well as having an older sibling with IBD (Model 3). Similar patterns were observed for the association between MMR and Crohn’s disease as well as between MMR and ulcerative colitis—except the associations only were statistically significant for IBD overall.

Table 3 shows the associations between other risk factors besides MMR status and IBD. The risk of IBD, Crohn’s disease, and UC increased between 2001 to 2018. Children born in earlier years had decreased risk of IBD compared with children born more recently. Compared with children born in Pacific states, children born in New York, Pennsylvania, or the southeast had a greater risk of IBD. History of seizures was associated with increased risk of IBD, Crohn’s disease, and ulcerative colitis. However, sex and allergy to vaccines were not associated with risk of any IBD. In Model 2, which adjusted for childhood comorbidities, children born preterm and with a greater comorbidity burden as represented by childhood chronic conditions had an increased risk of IBD, Crohn’s disease, and UC. In Model 3, having an older sibling affected by IBD was associated with significantly increased risk of childhood IBD and childhood Crohn’s disease, although there were too few cases with affected older siblings to draw conclusions about risk of childhood UC based upon an older sibling with childhood UC.

In sensitivity analysis, we restricted our examination to the population of children who were enrolled for at least 5 years, and we also examined whether adjustment for antibiotic use altered the pattern of results. In these analyses, vaccination with MMR was still associated with lower risk of IBD after adjustment for covariates noted in Model 1 (HR, 0.78; 95% CI, 0.64-0.96), Model 2 (HR, 0.80; 95% CI, 0.65-0.98), and Model 3 (HR, 0.78; 95% CI, 0.64-0.96). Vaccination with MMR was still associated with lower risk after adjustment for antibiotics as a time-varying covariate (HR, 0.77; 95% CI, 0.63-0.95), although antibiotic use itself was not associated with significantly altered risk of IBD (HR, 1.24; 95% CI, 0.76-2.02). Along similar lines, vaccination with MMR had similar associations with UC and Crohn’s disease when restricted to children who were enrolled for at least 5 years and with and without adjustment for antibiotics (results not shown).

Discussion

Using a large national insurance database from the United States, we found that the risk of early pediatric IBD increased between 2001 and 2018 in the United States in young children and also that MMR was associated with decreased risk of childhood IBD. This association persisted after adjustment for factors potentially confounding the association between MMR and IBD including sex, year of birth, geographic factors, childhood comorbidities, and affected older siblings. This analysis, including more than 2 million children, is the largest examination of this association to date. Our findings suggest that MMR is safe to administer and may even be protective against future IBD in young children.

Although a potential link between measles infection and pediatric IBD was reported 20 years ago,²⁸ the relationship between MMR has focused primarily upon potential risks of the vaccine rather than benefits. Previous studies of the association between MMR and IBD date back to a since-discredited report in the 1990s, which also linked MMR with autism.²⁹ Since that retracted study, other reports note inconsistent results regarding the impact of MMR upon IBD risk. In a 2001 case-control study from the Vaccine Safety Datalink project, including 142 persons with IBD, persons who had received MMR were at significantly lower risk for IBD compared with persons who had not.¹¹ A similar protective effect was reported in another registry-based study with approximately 222 cases of Crohn’s disease.¹² However, other studies with similar sample sizes found no association^13,14 and increased risk of IBD¹⁵ after MMR. Thus, the overall protective effect of MMR upon IBD has been inconclusive.¹⁶ We may have found a protective association due to the relatively larger number of children observed. Of note, among children, the age of onset is typically between 15 and 30 years of age, and the current report examined children who were younger. Thus, due to the low incidence of IBD in childhood and the difficulty of distinguishing between Crohn’s disease and UC in young children, we could not determine whether this protective effect was specific to one phenotype of IBD vs another. Compared with the study of risk factors for IBD in adults, studies of demographic and behavioral risk factors among children with IBD is relatively sparse. Among adults, there may be a higher risk of UC in northern latitudes,³⁰ which we did not find in our study. This association may reflect dietary factors such as vitamin D intake,³¹ although the directional of causality and mechanism remain speculative. Among children, UC may have a slightly male predominance,³² which we did not find in our study.

Children with IBD are particularly susceptible to infection due to immunosuppressive medication, and vaccination prior to onset is ideal. Awareness of the benefits of vaccination can improve vaccination rates.³³ However, the ongoing perception that MMR may be unsafe particularly for children at high risk for IBD persists and may contribute to imperfect vaccination rates. One hypothesis has noted that exposure to measles antigens via vaccination or infection may induce molecular mimicry and autoimmunity; although reports have not found greater risk of IBD,³⁴ measles may be a particular concern among parents with a family history of IBD. In a 2022 report of children with IBD from Australia, only 84% were up to date on their vaccination schedules, and only 67% had evidence of serologic immunity for measles, 52% carried antibodies for rubella, and only 42% carried antibodies for varicella at the time of IBD diagnosis.³⁵ A similar report from Italy noted that only 89% of children had MMR at the time of diagnosis.³⁶ Our results should reassure parents, particularly those with older children affected with IBD, and those who are hesitant about vaccinating their younger children. Interestingly in our analysis, children with greater comorbidities were more rather than less likely to be vaccinated.

Children who received MMR were enrolled for slightly longer than children who did not receive MMR. Due to their shorter enrollment period, unvaccinated children may have had less opportunity to be vaccinated. This is unlikely to be the main driver of the association between MMR and reduced risk of IBD because we examined only children who were enrolled for at least 2 years from birth, and MMR is recommended beginning at 1 year of age. Thus, all children in this analysis were insured for at least a year after they could receive the vaccine. Due to their shorter enrollment period, unvaccinated children had less of an opportunity to develop IBD that would be captured by claims data. However, we conducted a sensitivity analysis only including children who had been enrolled for at least 5 years, with similar findings. It is possible that residual confounding by covariates such as length of enrollment, seizure history, vaccine allergy, preterm birth, and other chronic conditions affected the association between MMR and IBD. However, these factors associated with MMR are also linked with closer surveillance, which would have been expected to increase the detection of IBD. We found the opposite, namely that MMR administration was actually associated with reduced risk of IBD. Finally, children’s enrollment in private insurance depends upon parental enrollment, which in turn is contingent upon employment. How might these parental factors influence whether children received MMR and IBD risk? It is possible that parents with more stable employment were also more likely to vaccinate their children. In order to explain the association between MMR and IBD, these parents would need to have been less, rather than more, likely to have children with detected IBD.

Strengths of this report include the large size of the database, examination of a broad population of risk, and capture of vaccination status, which does not rely solely upon parental report. Our report has several limitations. First, this report is observational, and thus we cannot account for confounding by other factors related to the decision to vaccinate against MMR. Second, due to the younger age of most of the children, we may not have captured associations between MMR and onset of disease in older children. Third, we relied on administrative claims data, which in turn relies upon physician diagnosis, which may or may not rest upon biopsy results and thus is likely less specific than studies incorporating these measures. We were unable to distinguish between inflammatory vs structuring vs penetrating disease, and it is possible that MMR may dispose to a particular subtype. However, reduced precision in the diagnosis would bias any associations to the null. Fourth, despite the number of participants enrolled, the number of children affected with IBD was low, and the number of children vaccinated was high. Thus, the power to detect associations between MMR and UC vs. MMR and Crohn’s disease was limited. Fifth, we were unable to adjust for socioeconomic status; although we were able to adjust for regional variations in vaccination and Crohn’s disease, which we have previously reported and thus reduced confounding by these variables and associated factors. Finally, it is possible that an interaction between vaccination and genetic risk exists, where higher or lower benefit is observed among children at higher genetic risk. To our knowledge, such data are not available.

We conclude that MMR may be protective against the development of IBD in young children. Importantly, we did not find any increased risk with this vaccine, even accounting for several measures of childhood comorbidity. Our findings should be replicated in datasets that are linked with clinical data that can provide more granular information upon type of IBD. In the meantime, parents hesitant to vaccinate their children due to poor health status should be reassured that the risk of chronic conditions is minimal with this vaccine, and parents hesitant to vaccinate due to another child affected with autoimmune disease should also be reassured that the vaccine does not seem to be associated with increased risk.

Ethical Considerations

This study was approved by the University of Michigan Institutional Review Board.

Funding

This work was supported by NIDDK R21DK128586.

Conflicts of Interest

None declared.

References