Commentary: The iceberg revisited

The epidemiological concept of the ‘iceberg’ of disease1—what is visible and what lurks below the surface—has been durable and useful. As of September 2012, the paper had been cited 1380 times2 and as the figure shows, citations associated with directly relevant papers have remained remarkably steady for 50 years. The metaphor of the iceberg is a valuable communications aid, immediately grasped by everyone. It is relevant to surveillance, measurement of population health, measuring the burden of illness, screening, needs assessment, health services planning and much else, for instance understanding selection bias. It clarifies the relationship between clinical epidemiology (which deals only with the visible part) and population-based epidemiology. My paper grew from a germ of the idea when I was a visiting fellow from Australia, spending a mind-expanding year mentored by J.N. (‘Jerry’) Morris in the MRC Social Medicine Research Unit, located in those days in an office and laboratory building behind the London Hospital Medical College. Jerry taught me a great deal of epidemiology by engaging me in revision of data for a second edition of his wonderful little book Uses of Epidemiology.3 There was a table in the first edition on ‘Completing the clinical picture’—demonstrating variation of the observed numbers affected by particular conditions according to the method of observation and the stage of the disease at which observations are made. It occurred to me as a former general practitioner that the numbers in this table would have considerable dramatic impact if they were projected onto the population of an ‘average’ general practice. There is greatest interest in the submerged part of the iceberg—disease precursors, preclinical conditions and very early clinical stages, that if identified and appropriate interventions initiated, might nip in the bud an otherwise relentlessly progressive and ultimately lethal disease process. This has become an integral part of teaching in family medicine programmes and other front-line aspects of health care. In the 50 years since the paper was published, great progress has been made in many aspects of clinical medicine, notably screening tests, diagnostic imaging, genomics and methods of detecting and managing inborn errors of metabolism. We can map the genome of apparently healthy people who will eventually develop Huntington’s disease, and girls and women at unusually high risk of developing aggressive breast cancer. Challenging philosophical and ethical problems have arisen since the health care system became capable of detecting precursors and early stages of these and other dangerous and often fatal conditions.4 The concept has been applied to examine the natural history of many conditions and suggest early diagnostic interventions. Consider depression and dementia. Incipient forms may be poorly defined clinically and could be detected earlier in their natural course by screening questionnaires. Perhaps some forms where early intervention might be beneficial could be differentiated. We need to distinguish bipolar disorder that often responds to medication but can be life-threatening without treatment, from adolescent angst and ‘reactive’ depression associated with bereavement, loss of job, etc. Alzheimer’s disease—dementia that often begins comparatively early in middle age with pathognomonic histopathology—differs from other forms of dementia with onset usually later in life, secondary to cerebral vascular disease, repeated head injuries, chemical poisoning etc. Some of these conditions are preventable or amenable to treatment if detected early, others not so much at the present state of medical knowledge.5 After the paper was published, I was invited to speak on the concept in university departments and community clinics all around the USA where I lived for a year soon afterwards. At that time, perhaps because of the company I kept, there was considerable emphasis on use of the ‘iceberg’ concept to evaluate the quality of medical care, by comparing numbers of observed cases at specified disease stages with the numbers expected according to theoretical models. Of course it is essential when comparing ‘observed’ and ‘expected’ numbers to bear in mind wide confidence limits and selective factors that might influence a particular clinic’s population. One methodological detail bothered me occasionally. When I was asked whether the numbers in the key table in the paper were based on incidence or prevalence data, I gave an evasive answer or just said ‘yes’—they were based on incidence or prevalence data. This may be a slightly messy flaw in the methods I used to derive the numbers of conditions or cases of particular diseases that might be expected to be present or to arise over the course of a year in a hypothetical ‘average’ general practice, or a clinic’s population. At a pragmatic level, I don’t think it matters much. It’s the concept that matters, not precise numbers of specified conditions. Number of publications referring to ‘The Iceberg’, 1963–2012 Source:http://apps.webofknowledge.com/CitationReport.do?product=WOS&search_mode=CitationReport&SID=2DFD2DFEM6C@n2L@FmI&page=1&cr_pqid=6&viewType=summary (14 February 2013, date last accessed) Another perspective on the ‘iceberg’ concept arose in discussions at community clinics and prepaid health care programmes. Administrators and managers suggested that the numbers of ‘expected’ cases of certain conditions could be compared with the numbers observed over a period of years, thereby providing a measuring instrument to evaluate the efficacy of screening procedures and routine clinical examination of patients. Although this might be possible, I would be cautious about suggesting use of the numbers from a theoretical model in any practical setting. Even the numbers from a large database such as the population of a prepaid health care plan covering many millions, as in California and the New England states, are subject to selection bias and random variation. I doubt whether valid conclusions could be based on comparison of 1 year’s experience in such a population with another, or comparison of clinic populations in different countries or regions. Such comparisons would be interesting, and some useful inferences might be derived from them but not to evaluate the efficacy of screening procedures or clinical care. After Jerry Morris had reviewed and approved my final text without suggesting any further changes, I sent it to Sir Theodore (‘Robbie’) Fox, the eminent editor of the Lancet. He accepted it without changing the text in any way, but did make one important modification. My title was ‘Completing the clinical picture in general practice’. Robbie Fox made that a subtitle and added as the title the word that has identified the concept ever since: ‘The iceberg’. We are all in his debt for this: my title was cumbersome and forgettable, whereas the vivid metaphor of the iceberg has proved to be memorable and appropriate. I am grateful to Natalia Abraham, Karen Trollope Kumar, Ian McDowell and Bob Spasoff for help with this paper. Conflict of interest: None declared.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.