Sample size calculation
A total number of 29 nursing homes was found to be necessary in order to obtain a sufficiently accurate estimate of the national incidence rate of an infectious disease in the Dutch population of nursing home residents. This number was calculated assuming random sampling without replacement from the total number of nursing homes in the country (330), using the average number of residents per nursing home (175) and an estimate of the standard deviation (based on pilot data) of the number of cases per nursing home, and by requiring the 95% confidence interval for the true incidence rate to have a width of about 0.02.
Initial recruitment of the nursing homes took place via the regional networks of nursing homes for education of elderly care physicians. Alongside an announcement of the set-up of the network was published in journals: Dutch Journal for Elderly Care Medicine (readers: Elderly care physicians), the Dutch Journal for Hygiene and Infection prevention (readers: infection control nurses) and the Dutch Infectious Disease bulletin (readers: Infection control physicians appointed at Municipal Health Services). Furthermore, a website with information on the SNIV network (www.SNIV.net) was launched.
We provided detailed documentation about the SNIV network to each nursing home that was interested to participate and an on the site presentation was planned to inform the manager, elderly care physicians and nurse practitioner. Afterwards, the manager signed a form in which they oblige themselves to participate in the network until further notice, but preferably for a year.
Only nursing homes with more than 50 residents could participate in the network. When a participating nursing home wished to participate with a second location with less than 50 beds this was allowed. In addition, an effort was made to recruit nursing homes from all parts of the Netherlands. Nursing homes with only a revalidation function were excluded from the network.
Protocols and procedures
The standardized form for weekly data collection and the methodology of the surveillance was previously tested in two pilot studies performed in a subset of eight nursing homes in 2008. In the first pilot, the feasibility of systematic data collection via a paper registration form was tested. Main recommendations were: to collect data via a web-based application to facilitate the registration process, to send reminders to enhance complete data collection and paper registration forms should stay available at each ward to facilitate internal data collection. In the second pilot, data registration via a web-based internet application was tested. The second pilot was used to further fine tune the digital registration form and process. In each participating nursing home a Private contact person was appointed who was responsible for weekly data collection. One confidential login and password were e-mailed to each of the contact persons. Since January 2009, full data registration started and all weekly data are collected online for all nursing homes.
Privacy of residents and nursing homes and safe data collection was an important issue in the set-up of the network and was addressed in a data regulation form. Only aggregated data will be published, no individual nursing home data. Weekly count data is not traceable to individual residents.
An advice committee meets up at least once a year to give the project team advice on which infectious diseases to survey. All Dutch parties involved in infection prevention and control were asked to designate a representative.
Feed back reports are sent twice a year to the nursing homes that participate in SNIV. This report with an overview of individual results versus the national results provides an indication whether their infection incidence is above or below the national average. During the influenza season weekly national incidence data from SNIV is published on the website of the Dutch Institute for Public Health and the Environment. Once a year a meeting is organised for participants with state-of-the-art lectures on the infectious disease under surveillance. Four times a year, news letters are sent to all persons involved with an update of the national nursing home infection incidence and information about the infectious disease under surveillance and related projects. The newsletters are published on the website together with an overview of the data collected in each year.
Data collection and analysis
Each year general facility characteristics are gathered; age distribution of residents, resident mix (e.g. somatic, psychogeriatric), size of the facility, size of the wards, availability of private bathroom and/or toilet facilities, interchange of personnel between wards, influenza vaccination uptake among residents and personnel, and availability of infection control protocols. As from 2010, these data are collected online for all nursing homes.
To minimize the weekly workload for the participants a maximum of five infections can be registered in the surveillance. In focus groups with elderly care experts and infectious disease experts we discussed which infectious diseases should be under surveillance in this sentinel network. Of major importance were influenza-like illness, gastro-enteritis and lower respiratory tract infection, for which count data are registered since January 2009. Urinary tract infections are under surveillance since 2011. All cause mortality is registered since January 2009.
Infections were registered based on clinical definitions conform medical practice in the nursing home setting. The definitions of the infectious diseases are stated in figure 1. For gastro-enteritis the definition was chosen that was previously used in nursing home setting for research into outbreaks of gastro-enteritis. For lower respiratory tract infection the definition was based on previous research on pneumonia in the elderly. For influenza-like illness the definition of the European Influenza Surveillance Network (EISN) was chosen. In addition nursing homes were asked to send in nose/throat swabs of residents with influenza-like illness or acute respiratory tract infections for the weekly virological surveillance of influenza. In the second pilot we also evaluated the application of these definitions and the way the elderly care physicians diagnose infectious disease in their nursing home.
Weekly incidence rates were calculated by dividing the total number of cases in one week by the total number of residents in the participating nursing homes in that week (resident-weeks). 95% Confidence intervals were calculated for the annual incidence rates. To study trends in infections the running average technique was used and 5 weeks-running averages were calculated.