Patients with Inflammatory Bowel Diseases are at Higher Risk for Meningitis

Structured Abstract

Background

The epidemiology of meningitis is unknown in inflammatory bowel disease (IBD) patients.

Goals

We aimed to determine the incidence of and risk factors for meningitis in IBD patients.

Study

We conducted a retrospective cohort and nested case-control study in the Quintiles IMSLegacy PharMetrics Adjudicated Claims Database from January 2001 to June 2016. We matched IBD patients to those without IBD on age, sex, enrollment and region. Meningitis was defined as one code for meningitis associated with an emergency department visit or hospitalization. Meningitis risk was calculated with incidence rate ratios (IRR). In a nested case-control study of IBD patients, predictors for meningitis were determined with multivariable conditional logistic regression models.

Results

We identified 50,029 patients with Crohn’s disease (CD) and 59,830 patients with ulcerative colitis (UC) matched to 296,801 non-IBD comparators. There were 85 CD patients, 77 UC patients and 235 comparators with meningitis. CD patients had 2.17 times the rate of meningitis and UC patients had 1.63 times the rate of meningitis as non-IBD comparators. After adjusting for relevant covariates among those with IBD, treatment with mesalamine was associated with a significantly lower odds of a meningitis claim (OR: 0.40, 95%CI:0.26–0.62). Having at least one co-morbidity was associated with a significantly higher odds of a meningitis claim (OR: 2.21, 95%CI:1.76–2.77).

Conclusions

While the overall rate of meningitis is low, IBD patients are at an increased risk compared with non-IBD comparators. Comorbidities are a risk factor for meningitis in IBD patients. Pneumococcal and meningococcal vaccinations should be discussed.

Introduction

Patients with inflammatory bowel diseases (IBD), including both Crohn’s disease (CD) and ulcerative colitis (UC), are at increased risk for infections.¹ Treatment of IBD frequently involves systemic immunosuppression, which also confer an increased risk of infections.² Soft tissue infections, urological infections, gynecological infections, herpes infections and pneumonia are some infections that have been shown to be increased in IBD patients.^1–5

Bacterial meningitis is a rare, but serious infection. Streptococcus pneumoniae is the leading cause of bacterial meningitis in the United States; it is associated with a 30% mortality rate. Neisseria meningitidis is the second leading cause of bacterial meningitis. Both of these serious infections have available inactivated vaccines. These vaccines contributed to a significant decrease in the incidence and mortality from bacterial meningitides over time.⁶ There is limited evidence on the incidence of meningitis in IBD patients to inform vaccination guideline recommendations.⁷ To date, incidence rates for meningitis in adult patients with IBD have not been estimated.

Understanding the epidemiology of meningitides in those with IBD can be important in counseling patients regarding this serious infection. Furthermore, understanding the epidemiology of bacterial meningitis can help guide vaccination strategies in adults with IBD, especially those about to start systemic immunosuppression. Therefore, in this study, we aimed to describe the population of patients with IBD who had evidence of meningitis based on health care claim data. We also report risk factors for development of meningitis in patients with IBD.

Methods

We analyzed insurance claims in the Quintiles IMS Legacy PharMetrics Adjudicated Claims Database from January 2001 to June 2016. Prior research has reported that this database is representative of the national commercially insured population reflecting a variety of demographic factors.⁸ This database has also been used previously in the epidemiological studies of IBD.^{1, 9} At the time of this analysis, the database contained enrollment information on over 27 million people across the United States (US), including nearly 130 million person-years of follow up data.

STUDY DESIGN

We performed a retrospective cohort study to determine the overall risk of meningitis in IBD patients compared with a matched non-IBD comparison cohort. We then performed a nested case-control study to determine the effects of demographics and IBD-related medication use on the risk of meningitis in IBD patients. A similar design has been used previously by our group and others to evaluate the incidence of disorders in IBD patients as well as the effects of medications commonly used by IBD patients.^{1, 10}

COHORT STUDY

All patients aged 18–64 years at the time of index IBD claim with at least 12 months of continuous enrollment in a health plan were eligible for this analysis. We chose 64 as the upper age limit because individuals who may be dual eligible for Medicare, which begins at 65 years, may have the possibility of missing data. Individuals were also required to have ongoing pharmacy coverage to fully capture medication exposures. We identified cases of CD and UC using a previously validated definition in administrative claims data.¹¹ In brief, IBD was defined as ≥3 healthcare contacts with an ICD (International Classification of Disease)-9 or ICD-10 diagnosis code for CD (555.x or K50.x) or UC (556.x or K51.x) on different days or 1 healthcare contact with a diagnosis code for CD or UC with 1 IBD related medication dispensation. IBD related medications were defined as the following: budesonide, systemic corticosteroids, available brands of 5-aminosalicylates (5-ASA), methotrexate, 6-mercaptopurine (MP), azathioprine, infliximab, adalimumab, certolizumab and golimumab. Vedolizumab and ustekinumab were not included based on dates of approval and limited follow up available. For patients who had claims for both CD and UC, disease assignment was made according to the majority of the last nine claims or a majority of the total claims if there were fewer than nine. We matched each IBD patient to at least two non-IBD comparators, matching exactly on sex, year of birth, date of enrollment and region of the country.

Each patient was required to have a minimum of 6 months of health plan enrollment before entry into the cohort. Entry into the cohort began at the date of the first IBD claim and entry into the non-IBD cohort occurred in a matched fashion. Each member of the cohort was followed until the first diagnosis of meningitis. If there was no meningitis event, they were censored at the discontinuation of primary or pharmacy insurance coverage.

Assessment of Outcome

Meningitis was defined as one ICD code for a meningitis associated with an emergency department (ED) visit or hospitalization. We used the definition developed by the Armed Forces Health Surveillance Branch to identify viral and bacterial meningitis and included codes for fungal meningitides; we also performed a manual search of codes to ensure that we were not missing any codes for meningitis.^{12, 13} Codes are listed in Appendix A. To ensure that the outcomes observed in the cohort were incident rather than prevalent, we looked at the 30-day claim history prior to entry into the cohort and did not find any claims for meningitis during this period.

Assessment of Exposures

All baseline exposure assessment occurred during the 6 month screening period prior to entry into the cohort. Comorbidities were assessed using the Deyo modification of the Charlson co-morbidity index (CCI).¹⁴ Medications, including systemic corticosteroids, defined as a ≥2 week course of any oral formulation, 5-ASAs, immunomodulators and biological agents were assessed in the exposure period.

Statistical Analysis

We used descriptive statistics to summarize the characteristics of patients with and without IBD. Continuous variables are reported as medians with an interquartile range (IQR), categorical variables are reported as percentages. We calculated incident rates of meningitis per 100,000 person-years and used incidence rate ratios (IRRs) and 95% confidence intervals (CIs) to compare the incidence of meningitis in IBD and non-IBD patients. Analyses were stratified by type of IBD (CD and UC). All analyses were performed using SAS (version 9.3) statistical software (SAS Institute, Cary, NC, USA).

NESTED CASE-CONTROL STUDY

We then conducted a case-control study among IBD patients to evaluate the associations between demographics, medications used to treat IBD: corticosteroids, 5-ASAs, immunomodulators and anti-TNF agents, and meningitis. This analysis was nested in the previously defined cohort of IBD patients.

IBD patients who were diagnosed with meningitis were cases. Controls were sex, age, index IBD year, enrollment year and disease (CD or UC) matched patients without a diagnosis of meningitis.

Assessment of Exposures

Medication use was the primary exposure assessed in this analysis. The medications assessed were systemic corticosteroids, 5-ASAs, thiopurines (6-MP and azathioprine), methotrexate and anti-TNF biologic agents. Medications, co-morbidities and demographics were assessed in the 6 month period prior to development of meningitis or a matched period.

Statistical Analysis

We used descriptive statistics to summarize the characteristics of the meningitis cases and non-meningitis controls. Continuous variables are reported as medians with an IQR, categorical variables are reported as percentages. We constructed multivariable conditional logistic regression models to calculate odds ratios (ORs) with 95% CIs adjusted for age, IBD-related medications, region and co-morbidities to predict the outcome of meningitis. We did sensitivity analyses replicating our models in sub-sets of the population that had a claim for an IBD-related medication and those who had a claim for immunosuppressive agent. All p-values were two-sided with values <0.05 considered statistically significant. Statistical analyses were performed in SAS (version 9.4) statistical software (SAS Institute, Cary, NC USA) and R (version 3.4.2, R Foundation for Statistical Computing, Vienna, Austria).

Results

COHORT STUDY

We identified 50,029 patients with CD and 59,830 patients with UC matched to 296,801 comparators without IBD. CD patients had 307,757 person-years of follow up, UC patients had 371,516 person-years of follow up and matched non-IBD comparators had 1,846,784 person-years of follow up. On average, there were 2.7 non-IBD comparators matched to each IBD patient. Demographics were similar between IBD patients and non-IBD comparators. CD patients had a median age of 57, UC patients had a median age of 55 and matched non-IBD comparators had a median age of 56. In this cohort, 59% of CD patients, 56% of UC patients and 58% of matched non-IBD comparators were female. Over 80% of all the cohorts had a Charlson co-morbidity index of 0. Demographics of the study population are presented in Table 1.

Table 1:

Characteristics of inflammatory bowel disease (IBD) subjects and matched non-IBD comparators in the Quintiles IMS Legacy PharMetrics Adjudicated Claims Database between January 2001 and June 2016

	Non-IBD Comparators	Crohn’s Disease	Ulcerative Colitis
N	296,801	50,029	59,830
% Female	58	59	56
Median current age in years (IQR)	56 (46–64)	57 (47–64)	55 (44–63)
Region
% East	24	24	24
% Midwest	30	31	30
% South	31	30	30
% West	15	15	16
Charlson Comorbidity Index
% 0	87	81	81
% 1	10	14	14
% ≥2	3	5	5
IBD Medication Exposure
% 5-ASA*	0	24	37
% Thiopurine	0	8	4
% Anti-TNF#	0	5	1
% Systemic Corticosteroids^	1	7	8
Meningitis
Unique patients	235	85	77
Incidence%	12.7	27.6	20.7
Incidence Rate Ratio (95% CI)		2.17 (1.69–2.78)	1.63 (1.26–2.11)

IQR: Inter-Quartile Range

^*5-Aminosalicylates: oral formulations only

^#TNF: Anti-Tumor Necrosis Factor-α

^{^}Defined as ≥ 2 week course of any oral systemic formulation

CI: confidence interval

^%calculated per 100,000 person-years

There were 85 CD patients, 77 UC patients and 235 non-IBD comparators who had a claim for meningitis associated with an ED or hospitalization. Incidence of a claim for meningitis requiring an ED visit or hospitalization was 27.6/100,000 person-years (95% CI: 22.3 – 34.2 per 100,000 person-years) for those with CD, 20.7/100,000 person-years (95% CI: 16.6 – 25.9 per 100,000 person-years) for those with UC and 12.7/100,000 person-years (95% CI: 11.2 – 14.5 per 100,000 person-years) for matched comparators (Figure 1). CD patients had an incidence rate ratio (IRR) of 2.17 (95% CI: 1.69 – 2.78) and UC patients had an IRR of 1.63 (95% CI: 1.26 – 2.11) compared with matched non-IBD comparators. Those in younger age categories had a higher rate of developing meningitis (Figure 2).

Figure 1:

Incidence rate of a claim for meningitis in the Quintiles IMS Legacy PharMetrics Adjudicated Claims Database between January 2001 and June 2016

Figure 2:

Incidence rate of meningitis by age category in the Quintiles IMS Legacy PharMetrics Adjudicated Claims Database between January 2001 and June 2016

In the IBD cohort, 25% of meningitis cases were coded as bacterial meningitis, 50% were coded as viral meningitis and the remainder were coded as other. In the non-IBD comparison cohort, 23% of meningitis cases were coded as bacterial, 55% were coded as viral, 2% as fungal and the remainder were coded as other.

NESTED CASE-CONTROL STUDY

Characteristics of the IBD cases who had a claim for meningitis compared with matched IBD controls who did not have a claim for meningitis are summarized in Table 2. IBD patients in the East were more likely to have a claim for meningitis. IBD patients who had a claim for meningitis were more likely to have to one or more co-morbidity. IBD patients who had a claim for meningitis were less likely to be treated with mesalamine. In a multivariable conditional logistic regression model adjusting for type of IBD, current age, sex, comorbidity index, region and IBD medications to determine associations for a meningitis claim, these trends held true (Table 3). IBD patients who had a Charlson Comorbidity Index ≥1 had a significantly higher odds of having a claim for meningitis (OR: 2.21, 95% CI: 1.76 – 2.77). Those who live in the Midwest and West had significantly lower odds of meningitis than those who live in the East. IBD patients who were treated with oral 5-ASA also had a significantly lower odds (OR: 0.40, 95% CI: 0.26 – 0.62) of having a claim for meningitis. These trends held true in sensitivity analyses in sub-sets of the population that had a claim for an IBD medication as well as those on immunosuppressive agents only.

Table 2:

Characteristics of inflammatory bowel disease (IBD) subjects with meningitis and matched IBD controls in the Quintiles IMS Legacy PharMetrics Adjudicated Claims Database between January 2001 and June 2016

	Meningitis IBD Cases	Non-Meningitis IBD Controls	p-value
N	162	281
% Female	62	63	0.92
% Crohn’s Disease Median current age in years (IQR)	53 55 (44–63)	52 57 (47–64)	0.92 0.10
Region			<0.01
% East	30	8
% Midwest	22	30
% South	35	31
% West	13	31
Charlson Comorbidity Index			<0.01
% 0	62	82
% 1	23	15
% ≥2	15	3
IBD Medication Exposure
% 5-ASA*	23	34	0.02
% Thiopurine	11	10	0.75
% Anti-TNF#	10	8	0.60
% Systemic Corticosteroids^	15	8	0.02

IQR: Inter-Quartile Range

^*5-Aminosalicylates: oral formulations only

^#TNF: Anti-Tumor Necrosis Factor-α

^{^}Defined as ≥ 2 week course of any oral systemic formulation

Table 3:

Adjusted odds of an inflammatory bowel disease patient having meningitis in the Quintiles IMS Legacy PharMetrics Adjudicated Claims Database between January 2001 and June 2016 in a nested case-control study

	aOR	95% CI
CCI# > 0	2.21	(1.76 – 2.77)
5-ASA*	0.40	(0.26 – 0.62)
Thiopurine	1.56	(0.88 – 2.76)
Anti-TNF#	0.79	(0.32 – 1.92)
Systemic Corticosteroids^	1.54	(0.99 – 2.40)
Region: East	Referent	Referent
Region: Midwest	0.68	(0.51 – 0.90)
Region: South	0.87	(0.67 – 1.12)
Region: West	0.64	(0.45 – 0.91)

Odds Ratios (ORs) are adjusted for type of IBD, current age, sex, comorbidity index, region and IBD medications

^#Deyo modification of the Charlson Comorbidity Index

^*5-Aminosalicylates: oral formulations only

^#TNF: Anti-Tumor Necrosis Factor-α

^{^}Defined as ≥ 2 week course of any oral systemic formulation

Discussion

We describe the incidence of meningitis in patients with IBD. In this study of a large US administrative claims database cohort, we found that CD patients had a 2.17 times higher incidence of meningitis associated with a hospitalization or ED visit and UC patients had a 1.63 times higher incidence of meningitis compared with a matched non-IBD comparison cohort. In a nested case-control study of IBD patients with and without meningitis, we found that treatment with a 5-ASA agent was associated with a lower odds for a claim for meningitis, while having 1 or more co-morbidities was associated with a significantly higher odds for meningitis. Other immunosuppressive medications were not associated with a significantly increased risk of meningitis.

Meningitis is considered a significant neurological issue in IBD patients and an emergency to recognize.¹⁵ However, the epidemiology of meningitis in IBD patients has not been previously described. While the overall rate of meningitis associated with an ED visit or hospitalization is low (<1%), IBD patients in this cohort had a higher rate of meningitis than a matched non-IBD comparison cohort. The current data regarding meningitis in IBD patients is at the level of case reports.^16–19 These reports note that meningitis is seen in those patients on immunosuppression. The majority of these patients were on corticosteroids at the time of meningitis diagnosis. One case report of a patient with undiagnosed CD reported meningitis as a complication of an epidural abscess, a serious complication that can be associated with CD or CD treatment.²⁰ Meningitis has also been in reported in a CD patient who was receiving gut-selective immunosuppression with vedolizumab.²¹ Due to the small numbers of IBD patients on vedolizumab in our cohort, we were not able to assess vedolizumab exposure.

We found that treatment with mesalamine was a protective factor for the incidence of meningitis. Patients treated only with mesalamine are likely the ones with lower inflammatory burden; so it may be that either inflammatory burden or immunosuppression are risk factors for meningitis. In this cohort where more patients are treated with 5-ASAs than immunosuppressive agents, it is possible that immunosuppressive medications confer an increased risk for meningitis, but that was not captured due to the small number of cases. For this reason, we did a sensitivity analysis looking at any immunosuppressive medication; this did not show a significantly increased risk of meningitis associated with any class of immunosuppression. Our finding is supported by other large studies of biologic agents which do not identify meningitis as a common opportunistic infection.^{22, 23} Furthermore, the majority of patients of this cohort were not receiving immunosuppression. Therefore, it might be that IBD itself confers a greater risk for meningitis.

With the introduction of vaccines against pathogens that commonly cause bacterial meningitis, rates of bacterial meningitis are decreasing in the US, especially in those under the age of 18 years and those over the age of 60.⁶ However, vaccination coverage is suboptimal in adults between the ages of 19 and 64; in one report, only 23% of adults who were at higher risk for pneumococcal infections received the vaccine.²⁴ IBD patients receive fewer preventive health services than those without IBD.²⁵ Our study further highlights the need to discuss vaccine preventable infections in IBD patients. It is difficult to accurately identify subtypes of meningitis in administrative claims data, however, our study demonstrates that viral meningitis is common in both IBD and non-IBD patients. As GI physicians are often the primary providers for young IBD patients, it is imperative for practitioners caring for IBD patients to understand the higher risk of meningitis in this population. Furthermore, the risk for meningitis in IBD patients may be independent of immunosuppressive treatment.

There are certainly limitations to reporting the epidemiology of IBD related outcomes in administrative claims databases. We are not able to assess markers of disease severity or other known risk factors for meningitis, such as living conditions. Including a population with insurance and pharmacy benefits may certainly introduce selection bias. The age restriction due to the data source may also contribute to selection bias. As meningitis disproportionately affects those in resource limited settings and at the extremes of age,²⁶ our study may, in fact, be underestimating the outcomes. Additionally, the assessment of the type of meningitis is limited in an administrative claims cohort. However, the outcome of meningitis is rare enough in IBD patients that describing the epidemiology of the condition in a large population is valuable. Furthermore, by including a large number of patients from various practice and geographic settings across the U.S. over a long period our time, our study has broad external generalizability.

In conclusion, our results highlight the importance of recognizing that patients with IBD have an increased risk of meningitis. While the data source does not allow for accurately determining the etiology of meningitis, this is an opportunity to reinforce pneumococcal vaccination in IBD patients, which is already a guideline for all CD patients and UC patients on immunosuppressive regimens. It is also important to reinforce general meningococcal recommendations in young IBD patients, who are at particularly high risk for bacterial meningitis.