Regional differences in the incidence of lymphocytic and collagenous colitis over time

Abstract Background In microscopic colitis (MC), the incidence has increased over the last decades. The aim of the present study was to determine the incidence of lymphocytic (LC) and collagenous colitis (CC) in the county Skåne (Scania), southern Sweden, during the period 2010–20 with focus both on the temporal and spatial variations. Methods The MC diagnosis was retrieved from the biopsy registries at the Departments of Pathology. Established diagnostic criteria (increased lymphocyte count, inflammation in lamina propria and in CC a collagen band) were used for diagnosis. Age, gender, date for diagnosis and municipality of residence were retrieved for all patients. Results In total 1985 patients could be identified with a mean age of 62.9 years (SD 15.7) whereof 1415 were women. The incidence for CC was stable with a total age-standardized rate (ASR) per 100 000 person-years of 6.34, (range 4.6–8.1). In LC the ASR was 7.90 (range 1.7–15.2) but increased markedly 2015–20 reaching 15.2 in 2019. Also, the northwest part of the region showed significantly higher ASR:s of LC during the last part of the decade in comparation to the whole region. Conclusions The incidence of CC was stable during the period while LC differed substantially in a way that indicates that it most probably must be two different disease entities. In LC, in view of the marked and rapid increase, although no definitive explanation could be found, causative environmental factors could be contemplated, why further studies are indicated.


Introduction
Microscopic colitis (MC) is a chronic inflammatory colonic disease that causes non-bloody diarrhoea mainly in women over 60 years [1].MC includes two subtypes: Lymphocytic colitis (LC) and collagenous colitis (CC).The two types are almost undistinguishable clinically but differs slightly histopathologically in the colonic mucosa; except for an inflammatory infiltrate in lamina propria and increased intraepithelial lymphocytes (LC), CC also demonstrates a thickened subepithelial collagenous layer [2].
CC was first described in Malmö 1976 by Lindstrom and LC was introduced a decade later by Lazenby [3,4].Initially, MC was considered rare.Epidemiological studies from the 80s and 90s from Europe and the USA demonstrate incidence rates less than 10 cases per 100,000 person-years.Thereafter, increasing numbers of both CC and LC have been reported.A recent nationwide study from Denmark reported the highest incidence numbers worldwide in 2011 (32.3 MC cases per 100,000 person-years) [5].An increased awareness of the disease and increased endoscopic activity as well as increased exposure of known and yet unidentified risk factors is probably the reasons for this sharp increase in incidence.
The etiology and pathophysiology of MC remains largely unknown.Emerging evidence suggests a multifactorial process involving an abnormal mucosal immune response to luminal factors in genetically predisposed individuals [6].Well described risk factors are gender, age, smoking and drug exposure, i.e. non-steroidal anti-inflammatory drugs (NSAID), proton-pump inhibitors (PPIs) and selective serotonin reuptake inhibitors (SSRIs) [1].Alcohol [7,8], hormone replacement therapy (HRT) consumption [9,10] and previous gastrointestinal infections [11] seem to be risk factors as well, but with weaker evidence.One study has also suggested an inverse association to population density, but this has not been validated in other studies yet [5].
The increasing incidence and geographical variations indicate that environmental factors contribute to the pathogenesis.A mapping of these fluctuations in time and space could give some information about possible contributing factors.The purpose of this study was to investigate the overall incidence of MC in Skåne (Scania) 2010-2020, as well as the spatial and temporal variations in incidence.

Catchment area/population demographics
Region Skåne with its 33 municipalities is the southernmost part of Sweden and is a mixed urban rural region with a total population approx. of 1.4 mill people year 2020.The population concentration is highest in the western part, 53% of the Skåne population live in one of the ten municipalities adjacent to the western coastline.Thirty percent is of foreign background (born abroad or born in Sweden with both parents born abroad) [12].Life expectancy in Skåne is similar to Sweden in total, 84.3 years for females and 80.8 years for men.Forty percent of the females and 30% of the men between 16 and 74 years have a college degree or higher which is the third highest numbers in Sweden [13].As in the rest of Sweden, the healthcare system in Skåne is tax-funded and offered to the whole population and is assumed to offer equal access to health care for all residents.

Identification of MC cases
All residents in Sweden have their own Personal Identity Number (PIN), given at birth or at immigration.The numbers correspond to date of birth and is also encoded for gender.The PIN can be used to identify individuals within the health care registries.This identification was used together with the Systematized Nomenclature of Medicine (SNOMED) system for biopsies at the Dept of Pathology to identify all patients with LC and CC through their specific codes (M40600 for CC and M47170 for LC).
The validity of MC cases in the Swedish Pathology registries has recently been evaluated with a positive predictive value of 95% [14].In this study, all new MC-cases in Skåne during the period January 2010 until December 2020 were identified and the data set was compared with data from 2000 to 2009, and duplicates were eliminated to assure that only new cases were included.

Identification of population and population density
Statistics Sweden (SCB) offers an open data base were population by municipalities, regions and the whole country as well as age and sex can be achieved.SCB also hosts data concerning populations density by organization the population in Demographical Statistical Areas (DSA).DSA is divided in three main categories based on population concentration: A (rural), B (mixed, i.e. small towns) and C (urban areas).Skåne comprises 789 DSAs, where 15% constitute category A, 14% B and 71% C [12].

Environmental factors: food and water contamination, gastroenteritis incidence and prescriptions of drugs
Data on water and food contamination was retrieved from the Swedish Food Agency (which collects reports about food poisoning from all municipalities in Sweden on an annual basis), the Department for Prevention of infectious diseases Skåne (supplies data on cryptosporidium infections) and from the local water supply and sewerage company in north-west Skåne (analyses chemical compounds and bacteria).Data on gastroenteritis incidence in north-west Skåne 2016-2019 was retrieved from the regional health care data bases based on the ICD-10 codes (ICD-10 A04,7, 8 or 9).Data on prescriptions of PPI, statins, HRT and SSRI was collected from the local registries at the hospital that in turn retrieves data from the National Prescribed Drug Register that started registering individual data in July 2005.

Statistics
Crude and age-specific incidence rates for gender, age, calendar year, municipality of residence and DSA-category were calculated by dividing the total number of cases by the total number of person-years of observation, multiplied by 100,000.Two-sided exact confidence intervals (CI) were based on a Poisson distribution.To account for different demographics in the different municipalities, age-standardized rates (ASR) were calculated by multiplying the age-specific rates with weights from the 2020 Swedish population.For ASR, standard error (SE) and two-sided CI were calculated based on binominal approximation.In a sensitivity analysis, SE and CI also were calculated on Poisson approximation, with no different results.
To calculate the relative risk of disease between two ASRs, standardized rate ratio (SRR) was calculated by dividing ASR1 by ASR2.95% CIs were calculated with Smiths formula [15], based on the SEs retrieved from the previous calculations on ASR for each group.
Differences in age at diagnosis was calculated with Student's T-test.
Linear regression analysis was used to determine whether an increase in ASR was significant.Statistical calculations were performed with IBM SPSS Statistics editor version 25.
A p-value below 0.05 was considered to be significant.

Ethics
This study was approved by the Swedish Ethical Review Authority 2020-06631 and 2021-01554.Since the present study was strictly register-based informed consent was not required.Waiver of informed consent was approved by the Swedish Ethical Review Authority.

Cases
In the registry at the Dept of pathology 2000 cases with MC could be found during the period 2010-20.However, 15 cases could not be identified in the patient data bases in Region Skåne.The most common reason for this is that the PIN number is not yet complete due to recent immigration.Consequently, 1985 cases with first-time diagnosis of MC during the entire study period (2010-2020) could be included, of which 71% were women.LC was more common (n = 1105, 56%), than CC (n = 880, 44%).Of the total cohort 53% of the female cases and 61% of the male cases had LC.The demographics can be seen in Table 1.

Incidence of MC based on age
For the entire study period, the mean age at diagnosis of MC was 62.9 years, range 4-95.For CC, the mean age was slightly higher than in LC (65.1(a-c)).

Incidence of MC based on calendar period
The incidence for CC was stable with a total ASR per 100,000 person-years of 6.3, (range 4.6-8.1).In LC the ASR for the whole period was 7.9 per 100,000 person-years (range 1.7-15.2) but increased significantly 2015-20 reaching 10.7 per

Incidence of MC based on population density
The incidence of MC including the subtypes did not differ significantly when the population was divided into different population density-groups (Table 2).

Incidence of MC based on municipality of residence
The ASR for MC for the whole decade varied between the municipalities with a maximum level in Båstad (at the coast in NW Skåne) reaching 20.1 per 100,000 person-years, and a minimum level in Bjuv (inland in NW Skåne) at 8.2 per 100,000 person-years, see Figure 4.As can be seen in Figure 3 the incidence in MC was mainly dependent on a marked increase in LC during the second part of the decade.The north-western part of the region excelled in high LC incidence 2015-2020.This part of Skåne consists of four municipalities with in total 469,270 inhabitants.All these municipalities had significantly or almost significantly higher ASR:s; (varying from 15.4 to 26.1) compared to the whole region 2015-2020 (10.7 per 100,000 person-years).See Figures 4 and 5.The LC incidence peaked 2019 where Helsingborg municipality (298,580 inhabitants in 2020) had 50 new LC cases as compared to between two and five per year during 2010-14.Båstad municipality had the highest crude incidence reaching 79.6 per 100,000 the same year.However, since Båstad is a small municipality (30,754 inhabitants), the absolute number of new cases was only 12.

Colonoscopies with biopsies in Skåne
The total number of colonoscopies with biopsies during the follow-up were 88,955.The mean number per year was 10,102 (range 9589-10,839).During the pandemic year 2020, 9589 colonoscopies were performed, which is the lowest number during the time period.During the whole period, 22.3 MC-cases were discovered per 1000 colonoscopies.The lowest rate was during 2014, where only 7.3 cases per 1000 colonoscopies were found.This is in line with the low number of the MC cases that year.See Figure 6.

Contamination in food and water and onset of LC
The patient group that developed LC in North-West Skåne 2016-19 consisted of 212 patients (mean age 60 years [range 5-92], 69% women).In view of the vast increase in incidence over a limited time period and within a small area, information about contamination in water or food was retrieved.During the study period, only two municipalities in North-West Skåne reported affected cases with a peak in 2013 with ten cases.The local water supply and sewerage company in North-West Skåne did not have any reports on any major water contamination during the study period either.Of the 212 patients that developed LC in north-west Skåne, 18 patients were diagnosed with gastroenteritis one year before their disease onset.

Medication and onset of LC
The time span from the first prescription of PPIs, statins, HRTs in women and SSRIs until disease onset for LC in the cohort in North-West Skåne was evenly distributed.No accumulation of prescriptions in close relation to disease onset could be noticed.In the cohort of 212 patients, 80 had PPIs, 43 statins, 34 HRT, and 42 SSRI.
The mean duration between the first prescription and LC diagnosis were as follow: PPI 4.7 years (range 0.0-9.0),statins 5.6 (range 0.6-8.9),HRT 5.6 years (range 0.6-9), and SSRI 4.8 years (range 0.3-9).Of these, PPI may be sold over the counter, and it is not possible to correct for this.

Discussion
The present study verifies the ongoing increase of MC incidence that is reported from epidemiological studies over the world.Our results demonstrate current incident rates of MC in Skåne that is lower than reported from Denmark (MC 20.7 cases per 100,000 person-years 2001-2016 [5] and overseas in Olmsted, USA (25.8 cases per 100,000 person-years, 2011-2019 [16] but confirm the ongoing national increase in Sweden, something that Bergman et al. demonstrated in a   nationwide incidence study with observation time 1995-2015 [17].They observed steady increasing incidence rates for the whole observation time, where ASR for MC 2010-2015 was 10.7 cases per 100,000 person-years.In our study, the incidence in CC remained stable while LC increased substantially and significantly.Actually, we found the highest incidence rates in LC this far described in Sweden (15.2 cases per 100,000 person-years 2018-2019).
Our data indicate that LC is more common than CC, which is in line with the incidence studies from America, where Olmsted (2002-2010 and 2011-2019) and Calgary (2004)(2005)(2006)(2007)(2008) reported data similar to ours (CC:LC 0.8:1, 0.6:1 and 0.5:1) [16,18,19] as well as the previous Swedish nationwide study (67% LC) [17].Also, a meta-analysis from 2015 showed a higher rate of LC than in CC [1].On the other side, nationwide studies from Denmark, Netherlands and Iceland all describe CC as the predominant subtype [5,20,21].The explanation of these differences remains unclear, but the traditions on number of biopsies taken at colonoscopies may have an impact since the thickness of the collagenous band may vary in different parts of the large intestine.Hence, some cases of LC may be a "hidden" CC.However, more likely, LC that in general and in particular in our study, shows more varied incidence rates is probably influenced by a yet unidentified environmental risk factor in a higher extent compared to CC.Some previous investigations in the region have been carried out.In a smaller study with 198 CC patients, the incidence in Malmö of CC 2001-2010 was slightly less than ours (CC 5.4 per 100,000 inhabitants) [22].However, in a geographically nearby region, but with another nationality, the incidence rate of MC from Zealand (Denmark) was estimated to 27.5 cases per 100,000 person-years in 2010-2016 [23].Several putative risk factors were contemplated.Smoking and consumption of PPIs were both more frequent in Zealand in that study.
The age-specific rate was highest in age classes between 70 and 80 which emphasizes the well documented association between MC and high age.However, for the oldest inhabitants, the incidence rate decreases.This could at least partly be an effect of the fact older patients more seldom is investigated with endoscopy sometimes due to comorbidity that prevents such measures.Otherwise, this could indicate that there really could be a risk maximum at 70-80 years of age.
We can in this study confirm the risk factor of female gender with a more than double relative risk for MC in females compared to men.This is in line with previous studies [5,17,21].The association between MC and gender is stronger in CC (SRR 2.7:1) than in LC (SRR 2.0:1), which may be a result of that LC is more associated with environmental risk factors and CC could be more dependent on hormonal factors.
Few studies have investigated population concentration as a potential risk factor for MC.However, the difference in incidence rates between Iceland 1995-1999 (MC 9.2 cases per 100,000 person-years, 3 inhabitants per square kilometre) and Netherlands 2000-2012 (MC 3.4 cases per 100,000 personyears, 409 inhabitants per square kilometre) may indicate that [20,21].In line with this, the nationwide study from Denmark demonstrated higher incidence of MC in rural areas compared to densely populated [5,24].In order to investigate whether we could confirm this skewed distribution, we divided the cohort with MC in Skåne into three different groups; urban, small villages and rural areas based on statistical information from Statistics Sweden.However, no differences could be found in our cohort.Instead, the MC cases including the subtypes were evenly distributed across the three groups which indicates that population concentration does not seem to be a risk factor for MC.
Based on previous observations in Skåne regarding inflammatory bowel disease (IBD) where some municipalities had a high incidence of ulcerative colitis and low of morbus Crohn and vice versa, we also divided the cohort based on the 33 municipalities in Skåne [24].Interestingly, even though many municipalities presented stable age-standardized rates, some increased markedly during the last part of the 2010s, especially LC cases.For example, in the north-west part of Skåne several municipalities showed exceptionally high incident rates.Båstad municipality peaked at 79.6 LC cases per 100,000 person-years during 2019 (crude incidence rate), which to the best of our knowledge is the highest number reported in LC.The only resemblance could be the incidence of IBD in the Faroe Islands that also was shown to have a magnitude of almost 80 cases per 100,000 person-years 2010-2014 [25].In view of the rapid change in a few years some environmental effect that has an impact over that time span must be suspected.Pollution in the water could be a possible explanation.An epidemic with gastroenteritis should most probably have a shorter time span.However, all age-standardized rates instead declined during 2020.
The number of colonoscopies with biopsies were stable during the follow-up time.Least examinations were performed during 2020, probably as a consequence of the covid pandemic.However, the number of MC cases per 1000 colonoscopies during 2020 (21.5) was almost of the same magnitude as the number for the whole period (22.3).Since the health care system in the county is well organised with both nine hospitals as well as private endoscopists this should not cause any problems when referring a patient for investigation.Furthermore, when a patient cannot carry out the investigation in their local hospital they will be remitted elsewhere in the county.Consequently, the sharp increase of LC cases in north-west Skåne is probably not due to an increased number of endoscopies.Furthermore, if that was the case also CC would increase during the same time span.Finally, the physicians working at the Dept of Gastroenterology and Pathology in North-West Skåne were substantially the same over the 2010 decade without any major changes.Besides this aspect, the pathologists in Skåne examine all specimens from colonoscopies consecutively regardless of where in the county they were taken.
Recently an association between infection with cryptosporidium and an increased incidence of ulcerative colitis and MC was found in Northern Sweden [26].The sharp change in incidence in North-West Skåne could very well be caused by that kind of infection too.However, no association was found between cryptosporidium and LC.Furthermore, there was no accumulation of prescriptions of any disease promoting medicines during the period preceding the diagnosis making this assumption less likely.
The study has some limitations.Because it is a registrybased study it has not been possible to retrieve detailed information about other diseases, medication, or lifestyle related risk factors such as smoking on an individual basis.All these factors may contribute to the demonstrated variations.There are also some strengths.Skåne is a well-defined region where almost all patients attend their local hospital All biopsies are analysed at the department of Pathology and can be collected through available data bases.General information about the background population is freely available from open registries.
In conclusion, the incidence of CC was stable during the period while LC differed substantially in a way that indicates that it most probably must be two different disease entities with different pathogenetical mechanisms and/or dependence on different risk factors.The incidence rate of LC exceeds previous investigations and whether the incidence will continue to increase remains to be seen.In LC, in view of the marked and rapid increase, still not yet identified causative environmental risk factors such as water contamination or infectious agents could be contemplated, while further studies are indicated.

Figure 2 .
Figure 2. (a) Women and (b) men.age-standardized rate per 100,000 person-years in collagenous colitis and lymphocytic colitis during the period 2010-20 in women and men, respectively.

Figure 3 .
Figure 3. age-standardized rate per 100,000 person-years for collagenous colitis and lymphocytic colitis with 95% confidence interval during the period 2010-2020.

Figure 4 .
Figure 4. age-standardized rate per 100,000 person-years during the period 2010-20 divided in the 33 municipalities in Skåne.

Figure 5 .
Figure 5. age-standardized rates per 100,000 for lymphocytic colitis 2015-2020, illustrating the variation in incidence between different parts of Skåne.

Figure 6 .
Figure 6.cases of microscopic colitis per 1000 colonoscopies with biopsies in Skåne in relation to age-standardized rate per 100,000 person-years of microscopic colitis 2010-2020.
When stratifying for the subtypes, new cases of LC was most common in age class 80-84, in comparison to CC where age class 75-79 had at the highest risk (Figures 3 versus 61.0 years).Male patients were significantly older than women at diagnosis of MC (p < 0.01) The mean age for MC differed slightly between the years (highest 2013, 64.2 years, lowest 2011, 61.3 years) (data not shown).When calculating age-specific incidence rates, the age class 80-84 years had the highest risk of diagnosis.

Table 1 .
demographic characteristics of the Mc cohort.

Table 2 .
cases with Mc, cc and lc divided into three main categories based on population concentration.