Mumps epidemiology in Germany 2007-11

A Takla (taklaa@rki.de)1,2,3, O Wichmann1, C Klinc4, W Hautmann4, T Rieck1,5, J Koch1 1. Immunization Unit, Robert Koch Institute, Berlin, Germany 2. Postgraduate Training for Applied Epidemiology (PAE), Berlin, Germany 3. European Programme for Intervention Epidemiology Training (EPIET), Stockholm, Sweden 4. Bavarian Health and Food Safety Authority (LGL), Oberschleißheim, Germany 5. Charité – University Medical Center, Berlin, Germany


Introduction
Mumps is a vaccine-preventable viral disease, typically characterised by swelling of the parotid glands, but can also cause severe complications like orchitis, meningitis, encephalitis or pancreatitis [1,2].The disease usually occurs among children and in the prevaccine era, the annual reported mumps incidences in Western European countries ranged between 100 and 600 per 100,000 inhabitants [3].With the availability of a live-attenuated mumps vaccine since the 1960s [4], disease incidence dramatically decreased in countries with mumps vaccination programmes [5,6].Over the past years, however, numerous reports from different countries with long-established vaccination programmes have been published about extensive mumps outbreaks that occurred predominantly among vaccinated children, adolescents, and young adults [7][8][9][10][11][12][13][14][15].Until 2013, mumps has been notifiable in Germany only in the five Eastern federal states (EFS) of former East Germany.During the years 2001 to 2011, annually reported incidence for all EFS ranged between 0.26 and 0.78/100,000 [16].There was no mandatory notification system in place in the remaining 11 Western federal states (WFS) of former West Germany.However, if an outbreak occurs in an institutional setting, the 'German Infection Protection Act' requires the institution to immediately inform the district health authority.A recent comprehensive survey suggested an increase in the number of mumps outbreaks over the past 10 years [17].The largest recorded outbreak occurred in the WFS of Bavaria in 2010/11.Voluntary (ad-hoc) reporting was set up during the outbreak period and identified 295 cases.
Although mumps incidences have been available for EFS for the past 10 years, incidences cannot be extrapolated to the 11 WFS due to historical differences in vaccination schedules and coverage rates.While West Germany had recommended one dose of mumps vaccine as part of the routine childhood vaccination schedule from 1976 onwards, East Germany had no mumps vaccination programme in place until reunification in 1990.Since 1991, two doses (in the second and sixth year of life) have been recommended throughout the reunified Germany, since 2001 with the first dose given at 11-14 months and the second dose at 15-23 months of age [18]; the vaccine strain predominantly used is Jeryl Lynn.Although routine mumps vaccination was adopted in EFS 15 years later than in WFS, vaccination coverage rates at school entry have been substantially higher in EFS since introduction of coverage monitoring in 1998 (Figure 1) [19].
In the absence of a countrywide mandatory mumps notification requirement until March 2013 we used billing data of the Associations of Statutory Health Insurance Physicians (ASHIP) for outpatients as an alternative data source.Approximately 85% of the population living in Germany is covered by statutory health insurances (total population in Germany in 2011: 81.8 million; WFS: 69.0 million, EFS: 12.8 million), and mumps is a disease usually treated on an outpatient basis (in EFS 2001-11: 94%).The aim of our study was to use countrywide ASHIP data (i) to estimate ambulatory mumps incidences and describe mumps-related demographics countrywide and separately for EFS and WFS, (ii) to estimate incidence and describe demographics for the outbreak in Bavaria 2010/11, and (iii) to compare the number of cases and demographics identified in the ambulatory ASHIP dataset with corresponding figures from the mandatory notification system in EFS and ad-hoc notification during the 2010/11 outbreak in Bavaria.

Definitions
For ASHIP data, a mumps case was defined as a person diagnosed with a mumps-related International Classification of Diseases (ICD-10) code (B26.0orchitis,B26.1-meningitis, B26.2-encephalitis, B26.3pancreatitis, B26.8-'other complication', B26.9-'no complication'), and for notification data as a person fulfilling the mumps case definition (clinical case, i.e. more than two days of one-or two-sided parotidal swelling without any other apparent cause, and/or clinical case with epidemiological link and/or clinical case with laboratory confirmation) [20].Incidence based on ASHIP data was defined as number of outpatient cases per 100,000 statutory health-insured [21], whereas incidence based on the mandatory notification system was defined as number of outpatient cases per 100,000 inhabitants in Germany [22].The German federal states of Brandenburg, Mecklenburg-Western Pomerania, Saxony, Saxony-Anhalt, and Thuringia were classified as EFS; Baden-Württemberg, Bavaria, Berlin (comprising of the former Eastern and Western part of the city), Bremen, Hamburg, Hesse, Lower Saxony, North Rhine-Westphalia, Rhineland-Palatinate, Saarland, Schleswig-Holstein as WFS.The Bavarian outbreak period ranged from the third quarter of 2010 to the second quarter of 2011, the non-outbreak period were the remaining quarters in the years 2007 to 2011.

ASHIP data structure and data included in the analysis
Statutory health insurance physicians send their reimbursement claims for provided ambulatory medical services, based on the ICD-10 code, to their corresponding regional ASHIP on a quarterly basis.Our dataset included the patient's anonymous unique identifier (ID), sex, month/year of birth, district/state of residence, state of billing physician's practice, ICD-10 code, quarter/year of diagnosis, reliability of diagnosis (suspected, confirmed, excluded or recovered), and type of diagnosis (current state, previous state, unknown or not provided).

ASHIP data cleaning
We only included confirmed diagnoses that were labelled as 'current state' for the calculation of incidences.To limit the dataset to a single diagnosis per unique ID, we used the following algorithm: For data from Bavaria, Rhineland-Palatinate (2007-11) and regional parts of North Rhine-Westphalia (2008-11), step 3 had to be omitted as the transmitted data did not routinely contain information on 'type of diagnosis'.

Mandatory and ad-hoc notification data
We retrieved mandatory mumps notification data reported from EFS through the German electronic surveillance system 'SurvNet' [23] that is routinely used by public health authorities to anonymously report information on inpatient and outpatient cases with notifiable diseases to the national level.The 'SurvNet' system was also used by district health authorities for the voluntary ad-hoc mumps reporting during the outbreak in Bavaria 2010/11.Notification datasets contained information on age, sex, date of disease onset, notification week, district/state of residence, vaccination status, and whether the case required hospitalisation.

Statistical analysis
We used chi-square test to test differences in incidences and proportions and Poisson regression to determine trends in incidence rate ratios (IRR).P values were defined as statistically significant if <0.05.Statistical analysis was performed by using Stata version 12.1 (StataCorp, Texas, US).

Estimation of countrywide and regional incidences based on ASHIP data
Of the 32,330 confirmed cases, 15,000 (46.4%) were male; for 212 (0.7%) no sex was specified.For the years 2007 to 2011, overall countrywide mumps incidence was 10.  ).The mean annual incidence of orchitis was higher in WFS than in EFS (0.72/100,000 versus 0.12/100,000 male cases; p<0.005).The proportion of orchitis complications among male cases by age group and geographic region is displayed in Figure 2.

Description of the 2010/11 outbreak in Bavaria based on ASHIP data
Between 2007 and 2011, Bavarian physicians billed 6,111 confirmed outpatient mumps cases to the   The proportion of claimed male cases was higher during the outbreak period than in the non-outbreak period (52.0%versus 47.5%; p<0.005), and a comparison of age-specific incidences indicated that the outbreak affected mainly the age group of 15-34 year-olds (Figure 3).Furthermore, the proportion of orchitis complications was significantly higher during the outbreak period compared to the non-outbreak period (17.7% versus 11.7%; p<0.005); all other proportions of complications showed no significant differences (data not shown).

EFS: Comparison of ASHIP data with mandatory notification data
In .The significant incidence differences in individuals younger than 20 years as well as the significant decreasing trend over the study period correspond to increasing vaccination coverages seen at school entry in both EFS and WFS.However, two-dose coverage in EFS has been substantially higher since beginning of monitoring.Although two-dose vaccination coverage rates are approaching levels to reach herd immunity of at least 92% [25] in both parts of the country, previous lower rates and a lack of catch-up vaccination activities may have left a pool of susceptibles that account for the high incidences in children and young adults.A representative seroprevalence study, conducted in 2003 to 2006 among more than 13,000 individuals aged 0 to 17 years in Germany, revealed approximately 20-22% to be mumps IgG-negative or borderline-positive, with higher proportions in the WFS [26].However, if the present high vaccination coverage at school entry can be sustained, incidences among children can be expected to further decrease.
In contrast to EFS, we observed in WFS a significant increasing incidence trend among 20 to 29 year-olds, suggesting an age-shift over time.This observation is mirrored by the Bavarian outbreak 2010/11 where highest incidences were seen among 15 to 29 year-olds; the finding is further in line with recent outbreak reports from other countries with long-established vaccination programmes [10][11][12][13]15].One reason could be that suboptimal vaccination coverage rates combined with low circulation of wild virus have caused an accumulation of susceptibles in those age groups.This theory is supported by the German serosurvey [26], but also by 86,098 immunisation records from 10 to 12 year-olds in Bavaria in 1988 (born one to three years after adoption of mumps vaccination) that revealed mumps coverage rates of only 55% [27].Another relevant factor could be waning immunity.In the same serosurvey, authors identified significant waning effects already in 0 to 17 year-old immunised children [26], and those effects could be even more dominant in older age groups with longer time spans since last vaccination and/or only one vaccination dose.Furthermore, the proportion of vaccinated cases in the Bavarian outbreak increased with age in the group of 15 to 29 year-olds.Although we only had vaccination status information for 217 reported cases, age and sex distribution were comparable with findings from ASHIP data.The results could therefore be a hint that waning immunity may indeed play a role in those age groups in Germany.In contrast, breakthrough infections were not observed among young adult cases in EFS; all breakthrough infections occurred among children and adolescents and could therefore reflect the expected proportion of vaccine failure in populations with high vaccination coverage.
Mumps incidences for Germany based on ASHIP data are substantially higher than incidences for Europe published by the European Centre for Disease Prevention and Control: 3.5 per 100,000 in 2010 [28] and 3.2 per 100,000 in 2009 [29].However, the 27 countries differ in their mumps surveillance systems, case definitions and time points of routine mumps vaccine introduction [30,31], and comparisons are therefore difficult.
Comparison of the mandatory notification data with ASHIP data in EFS indicated severe underreporting, especially among adults.Underreporting could even be higher, as we only included cases coded as confirmed and diagnosed in an ambulatory practice.Our findings suggest that in EFS, where mandatory mumps surveillance has been in place since 1964 [24], physicians diagnose mumps, but fail to report cases to public health authorities.This has important implications as Germany has introduced nationwide mandatory reporting of mumps as of March 2013.To retain reliable mumps notification data, it is crucial not only to (re)inform physicians about their reporting duty, but to think of new strategies to ease reporting.One approach could be to develop computer tools that directly link the ICD coding of notifiable infectious diseases with a report to the health authorities.
There is an increasing use of electronic health records or claims data to estimate disease trends or disease burden, especially for admission or discharge ICD codes from hospitals, e.g. for rotavirus, gonorrhoea, or varicella [32][33][34].In Germany, ICD-10 codes have previously been used to assess the herpes zoster disease burden also in the outpatient sector [35], and to assess reporting completeness of notifiable disease surveillance systems [36,37].However, electronic health records or claims data cannot and should not replace a surveillance system, as the data are usually only available with a time lag of several months, precluding rapid containment actions in the event of an outbreak, and are primarily intended for documentation or reimbursement, not surveillance purposes.In the case of ASHIP data, important information such as vaccination status at diagnosis or required hospitalisation is lacking.Nevertheless, such data can be an important source to assess the magnitude of incidences, disease trends and underreporting, and their reliability to determine communicable disease trends and burden should be further explored, also for the outpatient sector.
Using ASHIP data to estimate mumps incidences had several limitations related to data structure.There are no standardised guidelines for physicians of when to code a case as suspected or confirmed.Aiming at a conservative estimate, we decided to only include confirmed diagnoses.This restriction, as well as limiting the dataset to one observation per patient (who may have had several physician consultations during one disease episode), reduced the initial dataset considerably during the cleaning process.Therefore, the extent of the reduction cannot be interpreted as an indicator for the ASHIP data quality per se.As laboratory confirmation is not required to code a case confirmed, the proportion of laboratory confirmations is unknown.Moreover, ASHIP data only cover ambulatory cases.However, since mumps has a very distinct clinical presentation and the vast majority of mumps patients remain outpatients or have consulted their ambulatory physician before complications may require hospitalisation, it can be assumed that our incidence estimates are a good reflection of the true disease incidence.
Our dataset did not include the ca.15% of the population with private health insurance.However, we used the ASHIP population as denominator to calculate incidences and do not expect that mumps vaccination coverage differs substantially between privately and statutory insured.Finally, as information on 'type of diagnosis' is routinely missing for Bavaria, Rhineland-Palatinate and parts of North Rhine-Westphalia, we could not exclude diagnoses coded as 'previous state', 'unknown' or 'not provided' in the data cleaning process.However, if proportions of reduction had been applied to these federal states as observed for the others (step 3: 26.9%; step 4: 13.6%), the total number of cases would have been reduced by only 899 (2.7%).

Conclusions
ASHIP data proved a valuable alternative data source to estimate mumps incidences.The identified shift in age distribution, the vaccination status of reported cases, and serosurveys indicate that inadequate coverage (with less than two mumps vaccine doses) is the main reason for outbreaks und sustained mumps virus circulation in Germany.In 2010, the German Standing Committee on Vaccination (STIKO) recommended an additional MMR vaccination for persons born after 1970 with less than two measles vaccinations in their childhood [18], a recommendation that may simultaneously close some of the existing mumps vaccination gaps in adults.However, no catch-up vaccination activities for mumps have been initiated so far.In view of recent mumps outbreaks among adolescents and young adults and indications of waning immunity, the option of a third routine mumps vaccine dose is being discussed in the scientific literature [10,14,38].However, comprehensive data on the long-term effectiveness of two-dose mumps vaccination (e.g.measured during outbreaks among adolescents and young adults) and on the additional benefits of a third dose are lacking.
For the youngest age groups, efforts should focus on sustaining and even increasing the existing high vaccination coverage.In this respect, Finland has set an important example of how to successfully eliminate mumps by reaching and maintaining vaccination coverages of more than 95% [39].

Figure 1
Figure 1Mumps vaccination coverage rates for first and second dose at school entry (age 5-7 years) in Western and Eastern federal states, Germany, 1998-2010

Figure 2
Figure 2Mean annual mumps incidence among males and proportion of mumps-associated orchitis among all male mumps cases per age group, based on ambulatory statutory health insurance claims data, Germany 2007-2011 group (years) WFS | mumps incidence in males EFS | mumps incidence in males WFS | proportion of orchitis cases EFS | proportion of orchitis cases statutory health insurances.Of those, 1,995 (32.6%) were claimed during the outbreak period with the peak occurring in the first quarter of 2011 (n=752; 37.7%).

Figure 4
Figure 4Vaccination status per age group among mumps cases with known status, reported via the mandatory notification system, Eastern federal states,2007-2011 (n=284)

Figure 5
Figure 5Vaccination status per age group among mumps cases with known status, reported via the voluntary ad-hoc notification system, Bavarian outbreak 2010/2011 (n=217) (16,ata cleaning step 1, the number of diagnoses decreased to 136,142 (46,728 and 89,414), in step 2 to 49,746(16,516 and 33,230), and in step 3 to 40,819(16,516(not applicable) and 24,303).Step 4 limited the dataset to 32,330 confirmed mumps cases (11,330 and 21,000).Among the 86,396 diagnoses excluded in step 2, 36,957 (42.8%) observations were coded as suspected diagnoses that would, if included in the dataset, have accounted for an additional 29,514 suspected mumps cases after steps 3 and 4(10,406  and 19,108).

Table
Annual mumps incidence, incidence rate ratio and p value, by age group, based on ambulatory statutory health insurance claims data,Germany, 2007Germany,  -2011 (n=32,330)    (n=32,330) CI: confidence interval; IRR: incidence rate ratios.Significant IRR, 95%CI and p values are shown in bold.

Bavarian outbreak 2010/11: Comparison of ASHIP data with ad-hoc notification data
Title: Mumps incidence per age group based on ambulatory statutory health insurance claims data during the 2010/11 outbreak period (n=1,995) and non-outbreak period (n=4,116),Bavaria, 2007Bavaria,  -2011Point estimates showed 95% confidence intervals spanning a range of less than 0.05 (not shown in figure).
8%) were male.In total, 13.3 times more insurance cases were claimed than reported via the mandatory notification system.Stratified by age, 3,048 claimed ambulatory cases were adults (≥20 years) compared to 113 notified ambulatory cases (27.0-fold difference); among persons younger than 20 years, the difference was 5.8-fold (1,169 claimed versus 203 notified cases).DiscussionBecause countrywide mandatory mumps notification was not in place until 2013, we used mumps-related ICD-10 code diagnoses claimed through statutory Figure 3 Because routine mumps vaccination was introduced in the WFS in 1976, high prevalence rates of naturally acquired immunity are only found in those who were born in 1975 and later (in our dataset 32 to 36 years and older).During 1976-90, the WFS recommended one mumps vaccine dose, but comprehensive data on coverage rates from that period are not available.The two-dose mumps recommendation, introduced in WFS and EFS in 1991, has only targeted children born after 1990 (in our dataset 17 to 21 years and younger)