Association between abdominal hernia and the risk of subsequent dementia

Abstract Objective Matrix metalloproteinases (MMPs) may play a role in the pathophysiology of neurodegenerative disease and hernia formation. This retrospective cohort study was designed to assess whether there is an association between hernia and the risk of dementia. Materials and Methods Patients (≥45 years) with hernias were identified between 2000 and 2008 from a longitudinal claims data of one million beneficiaries from Taiwan's National Health Insurance program. A control group of patients with comparable distributions of sex, age, socioeconomic status, urbanization, and medical comorbidities without hernia were chosen for matching in a ratio of 1:1. Patients previously diagnosed with dementia were excluded. Follow‐up ended on December 31, 2013. Incidence rate of dementia was compared between patients with hernias and those without. Cox proportional hazards models were used to estimate hazards relative to those of the control group. Results After matching, there were 4,784 hernia and 4,784 nonhernia patients. Hernia patients showed a higher incidence rate and hazard ratio of dementia than those in nonhernia group (8.82 vs. 7.19/1,000 person‐years; adjusted hazard ratio [aHR], 1.24; 95% CI, 1.07 to 1.45; p < .01). Advanced age (p < .0001), hypertension (p = .0139), head injury (p = .0003), and stroke (p = .041) were found to be risk factors for dementia, while patients with high socioeconomic status (p < .01) and history of coronary artery disease (p = .0292) were unlikely to develop dementia in our cohort study. Conclusion Patients with hernias were associated with a higher incidence of dementia than those without. Our finding should be validated in further prospective studies with larger samples.


| INTRODUC TI ON
Dementia, which is one of the most common neurodegenerative diseases, is characterized by its slow progressive impairment of memory together with at least one more cognitive domain (Caselli, 2003). Nonvascular dementia (e.g., Alzheimer's disease [AD]), which accounts for 80%-90% of cases, is the most common type followed by vascular dementia and Lewy body dementia (Morris, 2003). The prevalence of dementia tends to increase with aging, while a growing body of evidence supported a variety of disorders (e.g., diabetes, hypertension, ischemic heart disease, chronic obstructive pulmonary disease, heart failure, depression, dyslipidemia, head injury, and stroke) as risk factors (Kohler, Buntinx, Palmer, & Akker, 2015;Poblador-Plou et al., 2014;Qiu, Xu, & Fratiglioni, 2010;Sahathevan, Brodtmann, & Donnan, 2012). It was estimated that the number of demented patients will double every 20 years, reaching 65.7 million by the year 2030 (Prince et al., 2013), thereby imposing substantial social and financial burdens on global health services.

Multiple neuropathologic processes involved in neurodegen-
erative and vascular diseases may contribute to the development of dementia. Previous studies have proposed an important role of matrix metalloproteinases (MMPs) in precipitating pathological changes in the brain and the blood-brain barrier (e.g., bloodbrain barrier leakage, neuroinflammation, and neurodegeneration; Kim & Joh, 2012;Mroczko, Groblewska, & Barcikowska, 2013;Rempe, Hartz, & Bauer, 2016;Rosenberg, 2009;Weekman & Wilcock, 2016). For example, elevated levels of MMP-9 as well as changes in the MMP/tissue inhibitors of MMPs (TIMP) balance in the plasma of AD patients were found (Lorenzl et al., 2003;Lorenzl, Buerger, Hampel, & Beal, 2008). Another study demonstrated that imbalances between TIMPs and MMPs are involved in blood-brain barrier breakdown and are implicated in the human immunodeficiency virus (HIV) type 1-associated neurocognitive disorders (Xing et al., 2017).
The pathogenesis underlying abdominal wall hernia is multifactorial, and current evidence suggests that the formation of abdominal hernia may be associated with disturbances in extracellular matrix turnover and connective tissue metabolism (Antoniou et al., 2011;Antoniou, Antoniou, Granderath, & Simopoulos, 2009;Franz, 2008;Henriksen, Yadete, Sorensen, Ågren, & Jorgensen, 2011). For instance, MMPs (e.g., MMP-1, MMP-2, MMP-9, MMP-13) appear to be involved in inguinal hernia or recurrent inguinal hernia development (Antoniou et al., 2011(Antoniou et al., , 2009. Considering that the pathogenesis for both hernia and dementia may be associated with MMPs, there may be a missing link between hernias and dementia. We hypothesize that adult patients with hernias may be at an increased risk for dementia. To test this hypothesis, we conducted a nationwide, population-based cohort study to examine the relationship between middle-aged (i.e., 45 years or older) patients with hernias and dementia.

| Data source
The National Health Insurance (NHI) program, which had a coverage rate of approximately 99% under mandatory enrollment, was initiated in 1995 for the purpose of providing comprehensive healthcare for residents and citizens in Taiwan. The present study adopted

| Identification of study patients and control group
Patients aged 45 years and older with the diagnosis of abdominal hernias (ICD-9-CM codes 550.xx-553.xx) between January 1, 2000, and December 31, 2008, were included as the hernia group ( Figure 1).
The database was searched for the study population to identify any hospitalized event with hernia as one of the discharge diagnoses and used outpatient claims to identify any visit for hernia. Patients were classified as having hernia if they had two or more outpatient visits with a hernia diagnostic code or at least one hospital admission with a diagnostic code of hernia within 365 calendar days. The first date of outpatient visit or hospital admission date that met the selection criteria for hernias, whichever came first, was considered the index date. Patients who had a history of dementia (ICD-9-CM codes 290, 294.1, and 331.0, respectively) diagnosed before or on the index date, died before index date as well as those with missing information (e.g., age or sex) were excluded. Besides, patients with a history of acquired immunodeficiency syndrome were also excluded.
For each patient with a hernia, one insured patient without a history of hernia or diseases (i.e., dementia or acquired immunodeficiency syndrome) before and on the index date were randomly assigned to the nonhernia group. Frequency matching with age, sex, urbanization level, socioeconomic status, and all comorbidities at ratio of 1:1 was adopted for the case and comparison cohort.

| Study outcomes and follow-up
Our primary outcomes were pooled incidence rates of dementia defined as at least two outpatient visits or one hospital admission with related diagnosis codes, in patients with or without hernias. All participants were followed till the occurrence of dementia, termination of study follow-up (i.e., December 31, 2013), or death (i.e., withdrawal from the insurance system), whichever came first.

| Statistics
Because this was a retrospective study, we did not a priori power calculations. Baseline characteristics were presented as mean (standard deviation) or numbers (percent), and group difference was measured by t test or Pearson χ 2 test for continuous and categorical variables, respectively. The incidence rate of dementia was defined as the total number of dementia diagnosis divided by the total sum of follow-up years (i.e., per 1,000 person-years), and the cumulative incidence curves were compared by Kaplan-Meier method with the difference being tested by log-rank test between hernia and comparison cohort. Cox proportional hazard model was applied in the study to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) of primary outcomes. For multivariate models, the study adjusted age, sex, and any significant variables in crude model for the outcome. All statistical analyses were performed with SAS 9.4 software (SAS Institute Inc.) and R software (R Foundation for Statistical Computing). Statistical significance was defined as a two-tailed p-value <.05.

| RE SULTS
The study included 4,784 newly diagnosed hernia participants, and 4,784 nonhernia participants after matching, and the demographic F I G U R E 1 Flow chart for selection of study patients and follow-up process. AIDS, acquired immunodeficiency syndrome characteristics were shown in Table 1. There were no differences in distribution of age group, sex, socioeconomic status, urbanization level, and all comorbidities between two cohorts. There was higher prevalence in age group (46-64 years, 54.93% vs. 54.72%), male (both 82.44%), socioeconomic status (NTD < 20,000, both 57.4%), urbanization level (level 2, both 30.2%) in hernia, and comparison cohort. The TA B L E 1 Comparison of demographics and comorbidity between hernia patients and controls hypertension, head injury, and stroke were found to be significant risk factors for dementia, while a high socioeconomic status and history of coronary artery disease appeared to be protective against the development of the disorder in our cohort study (Table 2). Table 3 presented the subgroup analysis of dementia risk between hernia and nonhernia cohort. Compared with nonhernia cohort, participants who diagnosed with hernia had significant higher risk to develop dementia in male gender, socioeconomic status of <20,000, urbanization level 1, with and without congestive heart failure, without hypertension, with comorbidities of diabetes, depression, CAD, head injury, stroke, COPD, CKD, hyperlipidemia, obesity, smoking, and alcoholism-related disease. Figure   showed the Kaplan-Meier curve for the cumulative risk of dementia for hernia and nonhernia cohort, where a significant difference was observed between the two cohorts (p = .0075 by log-rank test, Figure 2). Abbreviations: ARD, alcoholism-related disease; CAD, coronary artery disease; CCI, Charlson Comorbidity Index; CHF, congestive heart failure; CI, confidence interval; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; HR, hazard ratio; IR, incidence rates, per 1,000 person-years; PY, person-years. a Adjusted hazard ratio: mutually adjusted for gender, age, and baseline comorbidities in Cox proportional hazard regression. Despite current advances in the diagnosis of mental diseases, diagnosis of dementia remains challenging to physicians in primary care and is prone to be delayed or missed (Boustani et al., 2005).

| D ISCUSS I ON
There is no previous study reporting the possible relationship between hernia and dementia. Taking into account the action of MMP, which is a possible link between the two clinical entities, the present study attempted to identify their possible association. The results showed a 1.24-fold increase in risk for subsequent development of dementia in patients with hernias compared to those without. These findings carried several striking clinical implications. First, in view of the potential risk for subsequent dementia, potential screening strategies may be implemented for early detection of the disease for high-risk patients (e.g., advanced age) with co-existing hernia.
Second, patients with hypertension, head injury, or stroke may also be chosen for dementia screening programs. Consistent with the findings of the present study, satisfactory cardiovascular risk factor control has been reported to delay or even prevent the development of dementia (Patterson et al., 2008). Third, MMP may be a potential target in the treatment of dementia, although the clinical significance remains to be elucidated.
Multivariate analysis showed that patients with advanced age (≥65 years), low socioeconomic status, hypertension, head injury, and stroke exhibited significantly higher risks of developing dementia, after adjusting for covariates. The findings support those from previous studies identifying these comorbidities as potential risk factors for dementia (Goldbourt, Schnaider-Beeri, & Davidson, 2007;Jorm & Jolley, 1998;Kohler et al., 2015;Poblador-Plou et al., 2014;Qiu et al., 2010;Sahathevan et al., 2012). However, the finding of coronary artery disease being protective against the development of dementia in the current study remains poorly explained. In addition, there was no significant gender difference in dementiariskin the present cohort (Table 2). While the finding was in concert with that of some previous studies (Jorm & Jolley, 1998;Yang et al., 2016), it was different from that of other studies reporting a higher risk of Alzheimer disease or dementia in women than in men (Gao, Hendrie, Hall, & Hui, 1998;Sun et al., 2014).
The major strength of the current investigation was the utilization of a nationwide population-based database as well as the longi-  (Ashrafi et al., 2019) would further diminish the study population and blemish the accuracy of risk assessment. Additionally, despite being a common clinical observation, the association between increase in abdominal pressure (e.g., chronic cough and constipation) and the development of hernias remains controversial (Ruhl, 2007). Although previous studies reported the presence of obstipation (Liem, van der Graaf, Zwart, Geurts, & van Vroonhoven, 1997) or chronic cough (Ruhl, 2007) may increase the risk of inguinal hernia in female patients, this association was not found in their male counterparts in a large-scale study (Ruhl, 2007). Also, taking into consideration possible bias arising from listing the diagnoses of "chronic cough" and "chronic constipation" as independent variables for analysis because of the lack of clear definitions, these conditions were not included for further evaluation.
Caution should be taken in the interpretation of our results due to several limitations in the current study. First, our study has the inherent limitations pertaining to the use of a retrospective database (e.g., selection bias and misclassification bias). Second, as the symptom of dementia may be unrecognized and the demented patients may not seek active medical care, some cases with dementia may be missed. However, it could be assumed that the probability of missing the diagnosis was the same for both groups in our study because of the large sample size. Third, even though socioeconomic factors and urbanization level of the enrollees have been considered in this study, levels of education (Karp et al., 2004;Sun et al., 2014) and social disengagement (Bassuk, Glass, & Berkman, 1999), which are known confounders for dementia, were not available in the NHIRD. In addition, previous occupation, physical activity, family history of hernia, and mobility are also potential confounders for hernia. Matching for these confounders would have improved the validity of our results. Fourth, information on laboratory studies, imaging findings, or symptom severity of dementia, which were inherent limitations of the insurance claims data, was unavailable for further analysis. Fifth, matching comorbidities for subject selection also meant reduction of sample size and increased risk for bias. Additionally, the incidence of dementia may vary in HIV-infected patients depending on whether antiretroviral therapy is used (Larussa et al., 2006). To avoid biasing our results, HIV-infected patients were excluded.
In conclusion, the results of the current nationwide, populationbased longitudinal study provided evidence in support of a temporal association between abdominal wall hernia and the risk of dementia irrespective of the subjects' age, sex, socioeconomic factors, urbanization, and comorbidities. Advanced age (i.e., ≥65 years), low socioeconomic status (<20,000), hypertension, head injury, and stroke showed significant associations with dementia. Despite these findings, the results of our study should be validated in further prospective studies with larger samples.

ACK N OWLED G M ENTS
This study is supported in part by Taiwan  Memorial Funds, Japan.

CO N FLI C T O F I NTE R E S T
All authors report no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
The authors' contributions are as follows. Kuo-Chuan Hung and Chia-Hung Kao involved in the conception and design of the study.
Chia-Hung Kao performed the administrative support. All authors collected the data, organized, performed the data analysis and interpretation, wrote the manuscript, and finally approved the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The dataset used in this study is held by the Taiwan