Journal of Antimicrobial Chemotherapy (2000) 46, Suppl. S1, 15–22
JAC
The ECO•SENS Project: a prospective, multinational, multicentre epidemiological survey of the prevalence and antimicrobial susceptibility of urinary tract pathogens—interim report Gunnar Kahlmeter* Department of Clinical Microbiology, Central Hospital, S-351 85 Växjö, Sweden The ECO•SENS Project is the first international survey to investigate the prevalence and susceptibility of pathogens causing community-acquired, uncomplicated urinary tract infections (UTIs) in women. At 240 centres in 17 countries, female patients presenting with symptoms of uncomplicated UTIs were asked to provide a urine sample for testing for the presence of leucocytes and bacteria. The bacteria were identified and their susceptibility to 12 antibiotics commonly used in the treatment of UTIs was determined. The objective of the survey was to collect 5000 urine samples to obtain approximately 3500 isolates of defined uropathogens. This interim report includes the results from 1960 urine samples, 75% of which contained a uropathogen. Escherichia coli accounted for the majority (80%) of uropathogens isolated in all 17 countries. The rates of resistance among E. coli strains were: ampicillin and sulphamethoxazole, 30%; trimethoprim alone or with sulphamethoxazole, 15%; nalidixic acid, 6%; ciprofloxacin, 3%; amoxycillin–clavulanic acid, mecillinam, cefadroxil, nitrofurantoin and fosfomycin, < 2%. The use of ampicillin, sulphonamides and trimethoprim alone or with sulphamethoxazole needs to be reconsidered. The seemingly rapid increase in quinolone resistance among communityacquired E. coli in some of the countries gives cause for concern.
bial Resistance Working Group of the European Federation of Pharmaceutical Industries’ Associations. Important elements have been identified as part of a possible strategy to combat resistant bacteria. These include initiatives to encourage better practices in the use of antibiotics, to provide updated knowledge of antibiotic consumption and antimicrobial resistance and to advance research programmes aimed at a better understanding of these issues (the Copenhagen Recommendations).2 To assess the prevalence of antimicrobial resistance, many national and international surveillance programmes have already been launched, most of which have focused on comparing resistance problems in general and between countries. These include the WHO Antimicrobial Resistance Monitoring Programme, the European Antimicrobial Resistance Surveillance System (EARSS), the Hospitals in Europe Link for Infection Control through Surveillance (HELICS), European Group on Nosocomial Infections and the Alexander Project on pathogens in lower respiratory tract infections. In the area of urinary tract infections (UTIs),
Introduction Infectious diseases continue to threaten human health. Numerous antimicrobial agents introduced over the last 50 years have contributed significantly to the control of infection. However, development of resistance as a result of the wide and increasing use of these agents in medicine, veterinary practice and farming has become one of the major concerns for public health. With increased public attention in recent years, action has been taken by several groups to provide a framework for combating antimicrobial resistance. These issues were discussed at a conference held in Copenhagen in September 1998 and organized on the initiative of the European Union Chief Medical Officers. The World Health Organization (WHO) has described the implications of resistance in a recent report.1 Several dedicated committees and working groups have been established, including the European Union Scientific Steering Committee on Antimicrobial Resistance, the WHO Antimicrobial Resistance Monitoring and Containment Network and the Antimicro-
*Corresponding author. Tel: ⫹46-470-58-74-77; Fax: ⫹46-470-58-74-55; E-mail:
[email protected]
15 © 2000 The British Society for Antimicrobial Chemotherapy
G. Kahlmeter uropathogens have shown a slow but steady increase in resistance to several antibiotics over the last decade. Escherichia coli and other Enterobacteriaceae have become less susceptible to commonly used antibiotics such as ampicillin, amoxycillin, co-amoxiclav, sulphonamides, trimethoprim, co-trimoxazole and, in some geographical areas, fluoroquinolones.3,4 Therefore, there is a need to reassess appropriate first-choice treatment of uncomplicated UTIs. No international multicentre survey of antimicrobial resistance among community-acquired bacteria causing uncomplicated UTIs has yet been performed. The present project is such a multinational, multicentre survey investigating the prevalence and antimicrobial susceptibility of uropathogens isolated from women with communityacquired UTIs. Sixteen European countries and Canada were included.
urinalysis (Bayer Corporation, Diagnostics Division, Elkhart, IN, USA); results were recorded as ‘negative’, ‘trace’, ⫹, ⫹⫹ or ⫹⫹⫹, as instructed by the manufacturer. A Uricult dip-slide (Orion Diagnostica, Espoo, Finland) was then prepared according to the manufacturer’s instructions and forwarded by courier to a central laboratory for identification and susceptibility testing of bacteria.
Quantitative assessment On the day of receipt, the Uricult was incubated at 37°C at the central laboratory. After 16–20 h it was compared visually with a chart developed by Orion Diagnostica for quantitative assessment of the dip-slides, and the number of cfu/mL recorded (⬍103, ⭓103 to ⬍105, or ⭓105). Urine samples were classified as positive or negative in accordance with the guidelines for testing issued by the Infectious Diseases Society of America:5 a positive test was defined as a urine sample containing between ⭓103 and ⬍105 cfu/mL and presence of pyuria, or a sample containing ⭓105 cfu/mL irrespective of pyuria. Pyuria was defined as ‘trace’ or more as determined with the Multistix 2 method.6 For the purpose of this interim report, all isolates with bacterial counts of ⭓103 cfu/mL were included, irrespective of the presence of pyuria.
Materials and methods Collaborating centres A total of 505 centres in 16 European countries (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Luxembourg, The Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and the UK) and Canada were recruited for this project. Of these, 240 centres enrolled patients. Wherever possible the centres within a country were chosen to represent two or three geographically separate areas.
Identification All bacteria occurring in urine samples at ⭓103 cfu/mL were identified by their biochemical reaction profile using API identification products (bioMérieux, Marcy-l’Etoile, France). In case of mixed cultures no more than two bacteria were identified. Bacteria occurring at ⬍103 cfu/mL were not identified. Identified bacteria were stored on microbeads and frozen at –70°C. The following bacteria were classified as uropathogens: E. coli, Klebsiella spp., Proteus spp., Enterobacter spp., Citrobacter spp., other Enterobacteriaceae, Staphylococcus saprophyticus, enterococci and Pseudomonas spp. In this report the pathogens are grouped into the following categories: E. coli; other Enterobacteriaceae (Klebsiella, Proteus, Enterobacter and Citrobacter spp. and other Enterobacteriaceae); S. saprophyticus; and other pathogens (enterococci and Pseudomonas spp.).
Patients Female patients between 18 and 65 years of age with symptoms of uncomplicated UTI were eligible for inclusion. Patients who had had symptoms for ⬎7 days, those who had had more than three episodes of UTI in the previous 12 months and those who had received antibiotics during the 2 weeks before the onset of symptoms were excluded. Patients with upper UTI, pregnant patients and patients with urinary tract abnormalities or other complicating factors were also excluded. The severity of symptoms (i.e. frequency, dysuria, urgency and suprapubic pain) was rated on a four-point scale and only patients with a total symptom score of ⭓2 were included. Recruitment of a total of 5000 patients was planned for the project to collect approximately 3500 defined uropathogens. Each country was asked to recruit between 300 and 350 patients. Results from approximately 2000 patients are included in this interim report.
Antimicrobial susceptibility testing The antimicrobial susceptibility of bacteria was determined using the disc diffusion method as described by the Swedish Reference Group for Antibiotics (SRGA)7,8 (see also the SRGA website at http://www.srga.org). The medium used was IsoSensitest agar (Oxoid, Basingstoke, UK). Inhibition zone diameters were measured to the nearest millimetre with a slide gauge. E. coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853 were used as control strains. Test results were only accepted if the
Urine sampling procedures Patients were asked to provide a freshly voided midstream urine sample. Immediately after sampling, the urine was tested for leucocytes using a Multistix 2 reagent strip for 16
ECO•SENS Project: interim report inhibition zone diameters of the control strains were within performance range. All bacteria were tested against the following antimicrobial agents: ampicillin, amoxycillin–clavulanic acid, mecillinam, cefadroxil, trimethoprim and sulphamethoxazole (alone and in combination), ciprofloxacin, nalidixic acid, nitrofurantoin, fosfomycin and gentamicin. Breakpoint values for qualitative interpretation of inhibition zone diameters were based on the recommendations of SRGA [see their website (URL given above) and Table I].
age and mean total symptom score for the patients. Owing to the limited number of patients from some countries, groups of countries and selected individual countries are compared. For the purpose of this report, all bacteria collected from samples with ⭓103 cfu/mL are included, irrespective of the presence of leucocytes. Patients in the different geographical areas were generally similar with regard to age and total symptom score. However, Canadian patients were somewhat younger, and patients from Spain and Portugal had a lower total symptom score. The distribution of uropathogens in the different geographical areas is given in Table III. Only 102 urine samples (5.2%) were culture negative. Another 395 samples (20.2%) contained non-pathogens. Of the 1463 samples (74.6%) that contained uropathogens, E. coli was the most common pathogen in all areas, accounting for
Results This report presents preliminary results from 1960 patients who entered the survey in the period 19 January 1999 to 10 January 2000. Table II shows the country of origin, mean
Table I. Antimicrobial breakpoint values for resistance of Enterobacteriaceae and other Gram-negative enteric bacilli
Antimicrobial agent
SRGA zone diameter breakpoint (mm)
Ampicillin Amoxycillin–clavulanic acida Mecillinam Cefadroxil Trimethoprim Sulphamethoxazolea Trimethoprim–sulphamethoxazole Nalidixic acida Ciprofloxacin Nitrofurantoin Fosfomycin Gentamicin a
g/disc 10 20/10 10 30 5 100 1.25/23.75 30 10 100 200 30
⭐11 ⭐14 ⭐13 ⭐13 ⭐13 ⭐12 ⭐13 ⭐18 ⭐19 ⭐13 ⭐32 ⭐17
Corresponding SRGA MIC breakpoint (mg/L) ⭓16 ⭓16 ⭓16 ⭓16 ⭓8 – ⭓64 – ⭓2 ⭓64 ⭓64 ⭓4
Tentative breakpoint values.
Table II. Demographic data
Country or countries Austria, Germany and The Netherlands Belgium, France, Luxembourg and Switzerland Ireland and UK Denmark, Finland, Norway and Sweden Canada Greece Portugal and Spain All areas
Number of patients
Number of centres
Age, years mean (S.D.)
Total symptom score mean (S.D.)
383
48
39.7 (14.0)
6.3 (2.6)
369
51
41.4 (12.7)
7.4 (2.1)
426 244
36 21
38.9 (13.3) 39.2 (15.2)
7.1 (2.4) 6.3 (2.2)
211 158 169 1960
5 9 16 186
34.4 (12.1) 41.1 (13.5) 42.3 (12.6) 39.6 (13.6)
7.6 (2.5) 6.0 (1.5) 5.1 (1.7) 6.7 (2.4)
17
G. Kahlmeter Table III. Distribution of isolates by geographical area. Data are number (%) of isolates Countriesa Pathogens
all
1
2
3
4
5
6
7
All pathogens 1463 (74.6) 280 (73.1) 294 (79.7) 303 (71.1) 197 (80.7) 149 (70.6) 109 (69.0) 131 (77.5) E. coli 1163 (79.5) 226 (80.7) 234 (79.6) 246 (81.2) 173 (87.8) 120 (80.5) 72 (66.1) 92 (70.2) other Enterobacteriaceae 186 (12.7) 35 (12.5) 35 (11.9) 34 (11.2) 11 (5.6) 15 (10.1) 29 (26.6) 27 (20.6) S. saprophyticus 53 (3.6) 7 (2.5) 14 (4.8) 10 (3.3) 9 (4.6) 7 (4.7) 3 (2.8) 3 (2.3) other pathogens 61 (4.2) 12 (4.3) 11 (3.7) 13 (4.3) 4 (2.0) 7 (4.7) 5 (4.6) 9 (6.9) Non-pathogens 395 (20.2) 96 (25.1) 54 (14.6) 84 (19.7) 39 (16.0) 36 (17.1) 49 (31.0) 37 (21.9) Negative cultures 102 (5.2) 7 (1.8) 21 (5.7) 39 (9.2) 8 (3.3) 26 (12.3) 0 (0.0) 1 (0.6) Total 1960 383 369 426 244 211 158 169 a
1, Austria, Germany and The Netherlands; 2, Belgium, France, Luxembourg and Switzerland; 3, UK and Ireland; 4, Denmark, Finland, Norway and Sweden; 5, Canada; 6, Greece; 7, Portugal and Spain.
Table IV. Distribution of bacteria (%) by age group Patient age group (years)
All pathogens E. coli other Enterobacteriaceae S. saprophyticus other pathogens Non-pathogens Negative cultures Total number of samples
18–50
51–65
P
1063 (73.3) 851 (80.1) 116 (10.9) 49 (4.6) 47 (4.4) 301 (20.7) 87 (6.0) 1451
400 (78.6) 312 (78.0) 70 (17.5) 4 (1.0) 14 (3.5) 94 (18.5) 15 (2.9) 509
0.018 0.39 ⬍0.001 ⬍0.001 0.43 0.27 0.008
without any resistance mechanisms to the respective drug are equally sensitive irrespective of their geographical origin. Table VI shows the antimicrobial resistance figures for E. coli in the individual countries or country groups. In all areas, resistance to sulphamethoxazole and ampicillin was higher (average 30%) than resistance to other drugs; the highest figures were from Spain and Portugal and the lowest from the Scandinavian countries. Rates of resistance to trimethoprim and trimethoprim–sulphamethoxazole were 10–15% in all areas except Spain and Portugal, where it was nearly 35%. Quinolone resistance, as judged by resistance to nalidixic acid and ciprofloxacin, was low in most areas but very high in Spain and Portugal. Resistance to antimicrobial agents used only for UTIs (mecillinam, nitrofurantoin and fosfomycin) was low in all areas. Figure 3 shows a correlation analysis of the inhibition zone diameters for amoxycillin–clavulanic acid and mecillinam for the 1163 E. coli isolates. It is evident that, in most cases, decreased sensitivity to one drug is associated with an equivalent decrease in sensitivity to the other drug.
1163 isolates or 80% of all isolated uropathogens. Other Enterobacteriaceae, such as Proteus, Klebsiella, Enterobacter and Citrobacter spp. were isolated more frequently in samples from the southern European countries than in the northern part of Europe (P ⬍ 0.001). S. saprophyticus was found more frequently in the younger age group (P ⬍ 0.001) and Enterobacteriaceae other than E. coli were more common in the older age group (P ⬍ 0.001) (Table IV). There was a good correlation between bacteriuria and the presence of leucocytes (P ⬍ 0.001) (Table V). As E. coli is the most important uropathogen, the remainder of this report will consider only this organism. Figures 1 and 2 show the distributions of the E. coli inhibition zone diameter for 12 antimicrobial agents. Each chart shows the composite result of all 1163 E. coli isolates from all geographical areas. An analysis (not shown) of the zone distributions of the different geographical areas demonstrated that the median zone size for the part of each distribution containing the most sensitive organisms was identical in the various areas. This indicates that organisms 18
ECO•SENS Project: interim report Table V. Distribution of bacteria (%) by presence of leucocytes Presence of leucocytes, no. of samples (%)
All pathogens E. coli other Enterobacteriaceae S. saprophyticus other pathogens Non-pathogens Negative cultures
positive
negative
1375 (94.0) 1101 (94.7) 174 (93.5) 52 (98.1) 48 (78.7) 361 (91.4) 79 (77.5)
88 (6.0) 62 (5.3) 12 (6.5) 1 (1.9) 13 (21.3) 34 (8.6) 23 (22.5)
Total number of samples 1463 1163 186 53 61 395 102
Table VI. Escherichia coli antimicrobial resistance (%) by geographical area Countriesa Antimicrobial agent Ampicillin Amoxycillin– clavulanic acid Mecillinam Cefadroxil Trimethoprim Sulphamethoxazole Trimethoprim– sulphamethoxazole Nalidixic acid Ciprofloxacin Nitrofurantoin Fosfomycin Gentamicin a
all 1 2 3 4 5 6 7 (n ⫽ 1163) (n ⫽ 226) (n ⫽ 234) (n ⫽ 246) (n ⫽ 173) (n ⫽ 120) (n ⫽ 72) (n ⫽ 92) 29.9
27.0
23.1
41.1
16.8
30.8
22.2
54.3
2.1 1.0 2.3 15.6 30.3
1.8 1.3 2.2 14.6 30.1
0.9 0.4 1.3 15.0 30.3
3.7 1.6 1.6 16.7 36.2
2.9 0.6 4.0 11.0 17.9
2.5 1.7 1.7 10.8 23.3
1.4 1.4 4.2 13.9 20.8
0.0 0.0 3.3 33.7 54.3
14.6 6.4 2.9 1.4 0.4 0.9
13.7 4.4 2.2 0.9 0.0 0.9
13.7 6.4 2.6 1.3 0.4 0.4
15.0 2.0 1.2 0.0 0.4 0.4
8.7 2.9 0.0 1.2 0.6 0.0
11.7 0.8 0.0 1.7 0.8 0.0
12.5 6.9 2.8 2.8 1.4 1.4
34.8 35.9 19.6 5.4 0.0 5.4
See Table III.
However, the figure also shows that there were strains with other, non-linked, resistance mechanisms.
lowest proportion (66–70%). Other Enterobacteriaceae accounted for 13% of the isolates and S. saprophyticus for 2–5% (overall 4%). Enterobacteriaceae other than E. coli were more common in the older age group than the younger age group (18% and 11%, respectively) whereas the opposite was true for S. saprophyticus (1% and 5%, respectively). All susceptibility data presented in this report were generated in one laboratory using the methodology and breakpoint criteria of the SRGA (see website; URL given above). Figures 1 and 2 show the distribution of inhibition zone diameters for all the antibiotics tested against all collected E. coli isolates. Although the degree of resistance varied between countries, the profiles of the distributions were very homogeneous considering that the E. coli
Discussion The number of patients planned to be included in this survey (5000) should yield sufficient E. coli isolates for each of the 17 countries for a meaningful analysis of the results. The number of isolates collected in the first 1960 patients allows a preliminary overview of data for E. coli if the 17 countries are divided into seven geographical areas. As expected, E. coli was, by far, the most common pathogen (mean 80%, range 66–88%) among women with uncomplicated UTI. Portugal, Spain and Greece had the 19
G. Kahlmeter
Figure 1. Distribution of inhibition zone sizes for 1163 E. coli isolates with (a) ampicillin 10 g disc, (b) amoxycillin–clavulanic acid 20/10 g disc, (c) mecillinam 10 g disc, (d) cefadroxil 30 g disc, (e) fosfomycin 200 g disc and (f) gentamicin 30 g disc. The dotted lines indicate the breakpoints for resistance (from SRGA website: see text for URL).
isolates were collected from many different areas. The zone diameter breakpoints used to divide sensitive/intermediate strains from resistant strains were appropriate irrespective of the geographical origin of the E. coli. There were no problems of breakpoints dividing homogeneous biological populations. Where the resistance rates for the various antimicrobial agents can be compared with those from previous, geographically more limited, studies1,3,9–11 they suggest the same ranking of the antimicrobial agents, i.e. high resistance to sulphonamides, ampicillin, trimethoprim, trimethoprim–
sulphamethoxazole and low resistance to amoxycillin– clavulanic acid, mecillinam, cephalosporins, nitrofurantoin and fosfomycin. Resistance to quinolones is still low in most areas, but the data on resistance to nalidixic acid are a cause for concern, as is the degree of ciprofloxacin resistance in Portugal and Spain (36% and 20%, respectively). The intention of this survey was to describe the current prevalence of uropathogens and their resistance level but, when completed, it should also address the relationship between resistance development and the use of antimicrobial agents. Although there is an expectation that a high 20
ECO•SENS Project: interim report
Figure 2. Distribution of inhibition zone sizes for 1163 E. coli isolates with (a) trimethoprim 5 g disc, (b) sulphamethoxazole 100 g disc, (c) co-trimoxazole (1.25/23.75 g disc), (d) nalidixic acid 30 g disc, (e) ciprofloxacin 10 g disc and (f) nitrofurantoin 100 g disc. The dotted lines indicate the breakpoints for resistance (from SRGA website: see text for URL).
level of consumption of these agents leads to the development of resistance, there are clearly major differences in resistance between countries, as shown here for E. coli, that cannot be explained solely by the pattern of antimicrobial use.12 For example, Sweden uses an increasing amount of quinolones but no sulphonamides and very little trimethoprim–sulphamethoxazole for community-acquired UTIs, whereas the opposite is true in neighbouring Denmark. In France and Belgium, fosfomycin has been in use for many years whereas the Scandinavian countries use it very little. In Sweden and Finland close to 20% of prescriptions for
uncomplicated UTIs are for mecillinam, but this drug is less often used outside Scandinavia. From a microbiological point of view and based on the results of this interim report it would seem reasonable to reconsider the use of sulphonamides, ampicillin and trimethoprim (alone or combined with sulphamethoxazole) for the treatment of uncomplicated UTI. The extensive use of quinolones appears to lead to problems with development of resistance.4,10 In this interim report, the number of E. coli isolates from Spain and Portugal was small, but the results seem to corroborate those of previous studies. 21
G. Kahlmeter
References 1. World Health Organization. (1999). Report on Infectious Diseases: Removing Obstacles to Healthy Development. WHO, Geneva. 2. Anon. (1998). The Copenhagen Recommendations. Report from the Invitational EU Conference on The Microbial Threat, Copenhagen, Denmark, September 1998. Ministry of Health, Ministry of Food, Agriculture and Fisheries, Denmark. 3. Winstanley, T. G., Limb, D. I., Eggington, R. & Hancock, F. (1997). A 10 year survey of the antimicrobial susceptibility of urinary tract isolates in the UK: the Microbe Base project. Journal of Antimicrobial Chemotherapy 40, 591–4. 4. Garau, J., Xercavins, M., Rodriguez-Carballeira, M., GomexVera, J. R., Coll, I., Vidal, D. et al. (1999). Emergence and dissemination of quinolone-resistant Escherichia coli in the community. Antimicrobial Agents and Chemotherapy 43, 2736–41. 5. Rubin, R. H., Shapiro, E. D., Andriole, V. T., Davis, R. J. & Stamm, W. E. (1992). Evaluation of new anti-infective drugs for the treatment of urinary tract infection. Infectious Diseases Society of America and the Food and Drug Administration. Clinical Infectious Diseases 15, Suppl. 1, S216–27.
Figure 3. Correlation between inhibition zone diameters of amoxycillin–clavulanic acid (20/10 g disc) and mecillinam (10 g disc) for E. coli.
6. Scheer, W. D. (1987). The detection of leukocyte esterase activity in urine with a new reagent strip. American Journal of Clinical Pathology 87, 86–93. 7. Olsson-Liljequist, B., Larsson, P., Walder, M. & Miörner, H. (1997). Antimicrobial susceptibility testing in Sweden. III. Methodology for susceptibility testing. Scandinavian Journal of Infectious Diseases Supplementum 105, 13–23.
Some of the antimicrobial agents used specifically for treating community-acquired UTIs, such as fosfomycin, mecillinam and nitrofurantoin, still exhibit surprisingly low resistance frequencies in all the countries investigated, despite many years of use. The fact that they are not used in hospitals and institutions may contribute to this. As these antimicrobial agents are not used as growth promoters in animal husbandry and are not closely related to other antimicrobials, they do not cause selection of resistance to drugs used in more serious infections.
8. Kahlmeter, G., Olsson-Liljequist, B. & Ringertz, S. (1997). Antimicrobial susceptibility testing in Sweden—quality assurance. Scandinavian Journal of Infectious Diseases Supplementum 105, 24–31. 9. Gupta, K., Scholes, D. & Stamm, W. (1999). Increasing prevalence of antimicrobial resistance among uropathogens causing acute uncomplicated cystitis in women. Journal of the American Medical Association 281, 736–8. 10. Kresken, M., Hafner, D., Mittermayer, H., Verbist, L., BergogneBerezin, E., Giamarellou, H. et al. (1994). Prevalence of fluoroquinolone resistance in Europe. Study Group ‘Bacterial Resistance’ of the Paul Ehrlich Society for Chemotherapy. Infection 22, Suppl. 2, S90–8.
Acknowledgements
11. Grüneberg, R. N. (1994). Changes in urinary pathogens and their antibiotic sensitivities, 1971–1992. Journal of Antimicrobial Chemotherapy 33, Suppl. A, 1–8.
The skilful technical assistance of Johanna Zwenson and Hanna Odén is gratefully acknowledged. This project is funded by Leo Pharmaceutical Products, Denmark.
12. IMS Health (Q3/1999). Maxims Database, London.
22
