Records of psychiatric morbidity in general practice: the National Morbidity Surveys

Traditionally, epidemiologists have been concerned with counting the dead. Now, particularly when they are studying the natural history of psychiatric disorders, epidemiologists are putting more of their time and effort into the estimation of rates of incidence and prevalence of diseases and linking changes in these rates with environmental and genetic influences. What is the best way of obtaining the relevant epidemiological information? The best, but not necessarily the most convenient or cheapest, approach is to undertake a large-scale purpose-built sample survey, using fully trained personnel and adequately validated methods. What are the alternatives? Either one can make use of existing general-purpose surveys, such as the General Household Survey (see, for example, HMSO, 1973), or one can aim to obtain important information from routinely collected medical records. The latter can either be hospital-based or (in the UK, at least) based on the records of general practitioners. The National Morbidity Surveys arose naturally from the setting-up of the National Health Service. Since then, most of the population of the UK has been registered with a general practitioner (GP), who represents a common channel through which individuals may obtain medical advice and care; his records offer a potentially unique opportunity to assess the rate of incidence and prevalence of psychiatric disorder in the community. This idealistic and optimistic view of the value of a GP's records was clearly expressed by J. A. Charles in the foreword to what was in effect the pilot study for the First National Morbidity Survey (Logan, 1953): 'Avoiding similitudes and comparisons one can say that medical record keeping, scrupulously careful, brief and purposeful, is necessary for the study of the measure and movement of disease, for providing information as to the success or failure of the methods of treatment, and for the acquisition of knowledge as to the broad economics of practice.' Charles also writes of the recording physician (Logan, 1953): 'His work as a recorder requires, within limits, the accuracy of the scientific observer; he relates the sequence of events as does the historian; like the man of affairs he can never escape the need of some form of accountancy'. It is the purpose of this brief review to assess, in the light of what is now known about psychiatric problems in the community, whether such optimism has been justified by the results of the first two National Morbidity Surveys (Logan & Cushion, 1958; Logan, 1960; GRO, 1962; HMSO, 1974, 1979), and to ask whether the data which they have provided have any real value to those wishing to study the epidemiology of psychiatric disorders. Although she excludes the routine collection of medical records from her definition of'surveys', and also avoids discussion of psychiatric disorders, Cartwright (1983) has recently provided a valuable critique of the scope and methods of conducting health surveys. In the chapter on general measures of health and sickness she provides a particularly useful assessment of the Survey of Sickness (Logan & Brooke, 1957). This survey was started during the second World War (1943) to assess how the stress caused by wartime conditions affected the health of the people in the UK. Despite a few methodological problems, the study was successful in illustrating how useful epidemiological information could be obtained from the large-scale surveys of this type. Nevertheless, the survey-method was abandoned in favour of using routine medical records as the major source of information on morbidity (Logan, 1953). This appears to have been due to the medical profession's misguided belief that doctors' routinely collected records are a better source of information on sickness than are the results of well-designed sample surveys.