Health surveillance is an essential tool in monitoring health in an armed force and in helping to protect the health of service personnel. This study used a literature search and direct contacts with individual countries to identify and evaluate health surveillance mechanisms used by armed forces worldwide. The study identified several health surveillance mechanisms ranging from periodic health assessments of personnel to complex databases of medical data linked to demographic and other supporting data. Essential elements of an effective health surveillance system are outlined, including the requirement that systems are adequately supported and allow the routine monitoring of health at the population level consistently throughout an armed force and consistently during times of peace and during operations. Areas for further research and development include linking of data on hazardous exposures, jobs and the locations of personnel with medical data, and the follow-up of personnel beyond their military service.

he clinical picture of carbon monoxide poisoning is well known -sud-

Carbon monoxide is one of the most common toxins encountered in work settings, the gas being emitted in situations where there is incomplete combustion of carbon-containing substances. Its acute and chronic health effects have been well-documented. While identification of dangerous situations and evaluation of control measures are conducted by environmental monitoring, the body burden due to inhalation of carbon monoxide is measured by an individual's blood carboxyhaemoglobin level. Carboxyhaemoglobin level can be measured directly from a blood sample or, indirectly, by measuring the end-expired carbon monoxide level and using the charts provided to read the corresponding carboxyhaemoglobin level. As the end-expired method is not an intervention method, and is therefore easy to conduct, it is being used widely in epidemiological studies and it could also be used for individual measurements. This study presents a better statistical method for validating the end-expired method than the correlation method used and described in previous studies.