Q J Med 2002; 95: 197-198
© 2002 Association of Physicians
Editorial |
Case reports, case series and systematic reviews
Large numbers of reports of single cases and case series drop through the letterbox of the QJM's editorial office. A glance back through the journal's archive shows why. We have carried on publishing these sorts of descriptive studies long after the editors of journals that picked up the torch for evidence-based medicine fell out of love with them.According to the NHS Centre for Reviews and Dissemination, the randomized controlled trial is top of the hierarchy of what counts as reliable evidence for clinical decision-making. Case series are at the bottom, contained in the rubric: opinions of respected authorities based on clinical experience; descriptive studies; and reports of expert committees. But, despite this lowly position, there are many instances where valuable knowledge has come from someone taking the trouble to write up cases that are out of the ordinary. Two modern classics are the reports of Pneumocystis carinii pneumonia and mucosal candidiasis in previously healthy homosexual men1 and of hepatocellular adenomata in women taking oral contraceptives.2 When such reports are published, they alert other doctors and may stimulate further investigation. Where a mechanism can be devised for collating accounts of unusual events, as for instance in the UK's yellow card scheme for the reporting of suspected adverse drug reactions, case reports turn into a system for population surveillance.
There is another circumstance in which authors should write, and journals should publish, descriptive accounts of case series. This is in rare conditions where population-based studies or treatment trials are difficult or impossible to organize. Lachmann and colleagues' article on the treatment of amyloidois in unicentric Castleman's disease in this issue (see pp 211–18) is an example. What's more, this style of report often makes more enjoyable reading than accounts of randomized controlled trials that several journals now insist must conform to the procrustean requirements of the Consolidated Standards of Reporting Trials (CONSORT) checklist and flow-chart.
On the other hand, it must be admitted that descriptive studies have serious limitations. One is that retrospective reviews of case notes are rarely complete. Who can say whether the outcomes of the missing cases might have been very different? Another is that quirks in the way that unusual cases get referred make it hard to feel confident in generalizing from the experience of one centre. A third, as Grimes and Schulz have pointed out,3 is that without a comparison group, causal inferences about temporal associations need to be treated with deep suspicion.
At best, descriptions of case series act as catalysts for further investigation by methods that are more systematic. At worst, they can cause useful treatments to be abandoned (think of recent events concerning the MMR vaccine) or potentially harmful procedures to be adopted (remember past obstetric enthusiasm for routine fetal monitoring in pregnancy). Sometimes they construct diseases of doubtful validity but remarkable longevity. Forty years ago, Elwood, in a survey of more than 4000 people, showed that the presence of dysphagia, post-cricoid web and iron-deficiency anaemia in the same person occurred no more often than would be expected from the background prevalence of the three conditions.4 Tellingly, he also showed that agreement between radiologists over whether a post-cricoid web was present on a barium swallow was only slightly better than chance.5 Yet, cases of Patterson-Kelly syndrome continue to find their way into print.
Systematic reviews of observational studies are much harder to carry out and interpret than systematic reviews of randomized controlled trials. The main difficulties lie in coping with the diversity of study designs used by investigators, and the biases inherent in most observational studies. Methods are still being developed and argued over,5 but it is already clear that applying an evidence-based approach to traditional descriptive studies is useful.
So what should authors and readers expect from the QJM? No prizes for guessing that we are becoming unenthusiastic about reports of single cases. Even so, where they are of exceptional interest and written concisely, we may offer publication in the correspondence columns. Reports of case series will be given a warmer welcome, particularly if the circumstances are such that it would be unreasonable to demand a more systematic approach. (At the same time, the condition that they describe must not be so rare that few readers of the journal will ever encounter a case.) We encourage the submission of systematic reviews and meta-analyses that are directed at questions of clinical relevance. The investigators will need, however, to have been rigorous in their methodology and to have synthesized a useful amount of evidence. Reviews that conclude that there have been few studies, all of poor quality, and that further research is needed make poor reading.
References
1. Gottlieb MS, Schroff R, Schanker HM, Weisman JD, Fan PT, Wolf RA, Saxon A. Pneumocystis carinii pneumonia and mucosal candidiasis in previously healthy homosexual men: evidence of a new acquired cellular immunodeficiency. N Engl J Med1981; 305:1425–31.[Abstract]
2.
Rooks JB, Ory HW, Ishak KG, Strauss LT, Greenspan JR, Hill AP, Tyler CW Jr. Epidemiology of hepatocellular adenoma: the role of oral contraceptive use. JAMA1979; 242:644–8.
3. Grimes DA, Schulz KF. Descriptive studies: what they can and cannot do. Lancet2002; 359:145–9.[Web of Science][Medline]
4. Elwood PC, Jacobs A, Pitman RG, Entwistle CC. Epidemiology of the Patterson-Kelly syndrome. Lancet1964; ii:716–19.
5.
Elwood PC, Pitman RG. Observer error in the radiological diagnosis of Patterson-Kelly webs. Br J Radiol1966; 39:587–9.
6. [http://www.consort-statement.org/MOOSE.pdf]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||