The most common sources of data for surveillance of health events are health service providers, ranging from all health service providers in the case national statutorily notifiable diseases, to generalist health services (e.g. primary care services), specialised health services (e.g. sexually transmitted infection clinics, maternity services), or diagnostic service providers (e.g. microbiology laboratories). Some sources will provide a broader population perspective than others (e.g. microbiology laboratory reporting of hepatitis B infection is likely to provide information about cases acquired through all modes of transmission, whereas reporting from sexually transmitted infection clinics is likely to exclude a significant proportion of cases that not acquired through sexual contact). If it is important to be able to produce outputs that are expressed as population-based rates, which allow comparison between populations, particularly at the local population level, then it is necessary to use sources that draw from a well defined population base (e.g. general practitioners in the UK serve defined populations for which age-sex registers exist and can be used for calculating rates).

Other potential sources of surveillance data on disease or health status include disease registries, occupational health records, community services, emergency services, and screening programmes (including not only programmes such antenatal screening for HIV, syphilis or rubella, and screening of occupational groups such as military recruits, but also screening of blood donations). 'Over the counter' sales of medicines and rehydration solutions have more recently been shown to be potentially useful sources of data for surveillance, providing an early indication of community outbreaks [1]. The increasing uptake of the Internet also raises the possibility of basing surveillance on direct electronic reporting by members of the public.

Surveillance may also focus on exposures and hazards in the environment, in which case potential data sources will include veterinary services, environmental health services, water company quality testing records, and air quality monitoring records.

Expression of surveillance outputs in the form of a population rate or in the form of rates within population subgroups is required if the burden of morbidity and mortality between different populations and between different groups within a population is to be compared. Defining the population from which cases are drawn by surveillance systems is not always easy. While denominators are readily obtained for surveillance based on reporting by primary care physicians that serve a well defined, and documented, resident population, the same is not the case for surveillance based on cases drawn from emergency departments in hospitals serving cities with large transient populations, including workers who commute in on a daily basis and tourists. Defining the denominator for laboratory-based surveillance systems can also be a problem, particularly for specialist or reference microbiology services, where the population served by the referring clinical services may not be known or may change over time (as the pattern of referral of specimens may change according to contractual or other factors). Appropriate denominator data may also be required where the numerator may be influenced by the opportunity for case identification. For example, in the UK, where hospitals are compared with respect to rates of MRSA infection, the denominator takes into account the average number of patients in the hospital over the time period, and denominator data are also collected on the total number of blood cultures taken during the observation period.


1. Debjani Das, K. Metzger, R. Heffernan, S. Balter, D. Weiss, F. Mostashari. Monitoring Over-The-Counter Medication Sales for Early Detection of Disease Outbreaks --- New York City. MMWR 2005; 54(Suppl):41-46