rates over many years need to be utilised to determine baseline as well as excess mortality levels. In our study 'Atypical' bacteria are a mortality figures on their own over the past few years contributed little to our knowledge of the extent of influenza. common cause of An epidemic of influenza is diagnosed on both virological criteria (the proof of the presence of influenza virus) together community-acquired with epidemiological criteria (based on the presence of these nonspecific indicators). A number of definitions of an pneumonia in hospitalised epidemic or an epidemic threshold have been devised. For example, the rise of the monthly incidence of influenza-like adults illness beyond 400 per 100 000 inhabitants'6 or the isolation G. Maartens, S. J. Lewis, C. de Goveia, C. Bartie, • of influenza from at least 10% of submitted samples,7 or an excess of cases of influenza-like illness and nonspecific D. Roditi, K. P. Klugman acute respiratory illness for 2 consecutive weeks above the epidemic threshold" Objectives. To assess the proportion of cases of community· The success of the influenza surveillance programme acquired pneumonia caused by 'atypical' bacteria, inclUding depends directly on the interest and enthusiasm of the the recently discovered Chlamydia pneumoniae, and to sentinel doctors and the programme is an example par compare the clinical, radiographic and laboratory features of excellence of how primary care physicians and biomedical patients with and without 'atypical' bacteria. laboratories can co-operate and collaborate in a particularly Methods. A prospective serological study was carried out important preventive medical venture. on consecutive adult pneumonia patients from July 1987 to July 1988. Acute and convalescent sera were tested in Our sincere thanks to all the sentinel doctors who assisted batches for antibodies against Legionella pneumophila so enthusiastically with the Witwatersrand Viral Watch serogroup 1, C. pneumoniae, Chlamydia psittaci, Coxiella Surveillance Programme, as well as the principals and burnetii (phase-2 antigen) and Mycoplasma pneumoniae secretaries of all the schools included in the absenteeism (lgG and IgM). Records and chest radiographs were surveillance programme. We would also like to thank Ms examined retrospectively. Magda de Beer of the Department of Health, Housing and Results. Acute and convalescent sera were available from Urbanisation of the City of Johannesburg, and Or J. Lundie 113 patients. The records of 4 patients could not be traced and Sr E. Joss for providing mortality data for the two and 17 patients did not fulfil the inclusion criteria. Thirty-two homes for the aged studied. of these 92 patients (35,9%) were found to be infected with REFERENCES 'atypical' bacteria. The two most common organisms were 1. Hayle F. \'Yickramasinghe C. The case for life as a cosmic phenomenon. Nature C. pneumoniae (20,7%) and L. pneumophila (8,7%). There. 1986; 322; 509-511. were no differences in the clinical and radiographic features 2. Douglas R. Prevention, management and control of Influenza: a mandate for the 1980s. Am J Med 1987; 82: 1-3. of patients with and without 'atypical' bacteria. Clinicians 3. Ghendon Y. Influenza surveillance. Bull World j-Iealth Organ 1991; 69: 509-515. prescribed erythromycin or tetracyclines with equal 4. Schoub BD, Johnson S, McAnerney JM, Martin E, Dos Santos IL. Laboratory studies of the 1984 influenza epidemic on the Witwatersrand. S Atr Med J 1986; frequency in the two groups. 70: 815-818. Conclusions. 'Atypical' bacteria, especially C. 5. Assaad F, Cockburn WC, Sundaresan TK. Use of excess mortality from respiratory diseases in the study of influenza. Bull World Health Organ 1973; 49: 219-233. pneumoniae, are a common cause of community-acquired 6. Paul Glezen W, Decker M, Joseph SW, Mercready RG jun. Acute respiratory disease associated with influenza epidemics in Houston, 1981-1983. J Infect Dis pneumonia in adults in South Africa. This is the first 1987; 155: 1119-1126. demonstration of an aetiological role of C. pneumon.iae in 7. Hannoun C, Dab W, Cohen JM. A new influenza surveillance system in France: the lIe-de-France 'GAOG'. 1. Principles and methodology. Eur J Epidemio/1989; 5: this country. We confirmed the finding of other studies that 285-293. there are no clinical, radiographic or laboratory features 8. Public Health Laboratory Service Standing Advisory Committee on Influenza. Influenza surveillance. 1972-1975. J Hyg 1977; 78: 223-233. characteristic of 'atypical' bacterial infection in hospitalised 9. Snacken A, lion J. Van Casteren V, et al. Five years of sentinel surveillance of patients. This has major implications for therapy, as these acute respiratory infections (1985-1990). The benefits of an influenza eany warning system. Eur J Epidemiol 1992; 8: 485-490. organisms respond to erythromycin and tetracyclines, but· 10. Wand Health Organisation. Report of WHO/GEIG informal consultation on the ~-Iactam standardisation and improvement of Influenza surveillance, Monaco, 25 September not to antibiotics. 1991 (WHO/GEIG.RPT/ke/5). Geneva: WHO. 1991; 1-5. S Afr Med J 1994; 84: 678-682. 11. Mancini GM, Arangio-Ruiz G. Campitelli L. et al. Surveillance of influenza A and 8 viruses in Italy between 1984 and 1987. Eur J Epidemiol 1988; 4: 445-450. 12. World Health Organisation. Aecommended composition of Influenza virus vaccines for use in the 1990-1991 season. WkJy Epidemiol Aec 1990; 65: 53-60. 13. Wand Health Organisation. Recommended composition of Influenza virus vaccines for use in the 1991-1992 season. WkJy Epidemiol Rec 1991; 66: 57-64. Departments of Medicine and Medical Microbiology, Groote Schuur 14. World Health Organisation. Recommended composition of Influenza virus vaccines Hospital and University of Cape Town for use in the 1992-1993 season. WkJy Epidemiol Rec 1992; 67: 57-64. 15. Udwell OM. Sommerville T. Observations on the incidence and distribution of the G. Maartens. EC.P. (SA) common cold in a rural community during 1948 and 1949. J Hyg 1951; 49: 365­ 381. S. J. Lewis. M.RC.? (UX) 16. Fleming DM, Ayres JG. Diagnosis and patterns of Incidence of influenza-like illness C. de Goveia. M.MED. (MICROBIOL) and the common cold in general practice. J R CoJJ Gen Pract 1988; 38: 159-162. D. Roditi. M.MED. (MICROBIOL) Accepted 4 July 1994. Department of Microbiology, South African Institute for Medical Research and University of the Witwatersrand, Johannesburg C. Bartie. 8.SC. HONS K. P. KJugman. PH.D.• EE PATH. (SA) Volume 84 No. 10 October 1994 SAMJ SAMJ ARTICLES Atypical pneumonia is a clinical syndrome first described by All of the studies quoted so far were performed in Reimann in 1938.' There is no precise definition of this developed countries. A previous study from this institution22 syndrome! but the illness has a subacute onset with showed that 10% of cases were due to M. pneumoniae. constitutional symptoms and a dry cough. Radiographic Other 'atypical' bacteria were not tested for in that study. shadowing is characteristically non-segmental and more The aim of the current study was to assess the proportion of extensive than the physical examination would suggest. cases of community-acquired pneumonia due to M. Leukocytosis is absent or moderate. The course is generally pneumoniae, Legionella pneumophila, C. psittaci, C. benign. Mycoplasma pneumoniae is the most common pneumoniae and C. burnetii in a developing country. A aetiological agent of atypical pneumonia. Other bacteria that secondary objective was to compare the clinical, cause atypical pneumonia are Chlamydia psittaci, Legionella radiographic and laboratory features of patients with and spp. and Coxiella burnetii (the agent that causes Q fever)! without 'atypical' bacteria. These organisms share three characteristics: they are not visible on Gram stain, they do not grow on conventional culture media and they do not respond to penicillins or cephalosporins. Patients and methods There is a widespread belief that pneumonia caused by We evaluated consecutive adults with an admission these organisms can readily be differentiated from diagnosis of pneumonia from July 1987 to July 1988 at pneumonia due to conventional bacteria. However, the most Groote Schuur Hospital, a large community-based university common radiographic finding in pneumonia due to these hospital. Pneumonia was defined as an acute respiratory four 'atypical' bacteria is unilateral consolidation, often lobar illness with compatible shadowing on chest radiograph or segmental,3-8 and the clinical features are generally (performed on admission or within 48 hours). Patients with indistinguishable from those found in pneumonia caused by pulmonary tuberculosis or severe immunosuppression (e.g. conventional bacteria. 2.7,'0 in cases of AIDS) were excluded. Folders were Table I summarises four recent studies which show that retrospectively reviewed for demographic data, underlying 'atypical' bacteria cause a significant proportion of cases of illnesses, symptom duration, blood pressure values, community-acquired pneumonia in adults. The major respiratory rate, presence of confusion, laboratory features differences in these studies are the incidence of (P02, urea values, white cell count) and whether the chest mycoplasma pneumonia, which occurs in cyclical epidemics radiograph showed a segmental or non-segmental pattern every few years," and legionella pneumonia which varies of pulmonary shadowing. geographically and temporally." A fifth 'atypical' bacterium that causes pneumonia has Microbiological