ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, July 2005, p. 3021–3024 0066-4804/05/$08.00⫹0 doi:10.1128/AAC.49.7.3021–3024.2005 Copyright © 2005, American Society for Microbiology. All Rights Reserved.
Vol. 49, No. 7
Progressive Increase in Antimicrobial Resistance among Invasive Isolates of Haemophilus influenzae Obtained from Children Admitted to a Hospital in Kilifi, Kenya, from 1994 to 2002 J. Anthony G. Scott,1,2* Salim Mwarumba,1 Caroline Ngetsa,1 Salome Njenga,1 Brett S. Lowe,1 Mary P. E. Slack,3 James A. Berkley,1 Isaiah Mwangi,1 Kathryn Maitland,1,4 Mike English,1,5 and Kevin Marsh1,2 Wellcome Trust/Kenya Medical Research Institute, Centre for Geographic Medicine Research—Coast, Kilifi, Kenya,1 and Nuffield Department of Clinical Medicine2 and Department of Paediatrics,5 Oxford University, John Radcliffe Hospital, and Health Protection Agency Haemophilus Reference Unit, WHO Collaborating Centre for Haemophilus influenzae, John Radcliffe Hospital, Headington,3 Oxford, and Department of Academic Paediatrics, Imperial College, London,4 United Kingdom Received 29 September 2004/Returned for modification 14 November 2004/Accepted 1 March 2005
Etest susceptibilities to amoxicillin, chloramphenicol, and trimethoprim-sulfamethoxazole of 240 invasive isolates of Haemophilus influenzae cultured from children in rural Kenya were 66%, 66%, and 38%, respectively. Resistance increased markedly over 9 years and was concentrated among serotype b isolates. In Africa, the increasing cost of treating resistant infections supports economic arguments for prevention through conjugate H. influenzae type b immunization. Among children in sub-Saharan Africa, Haemophilus influenzae type b (Hib) causes 20% of cases of radiologically confirmed pneumonia and 40% of cases of meningitis (10, 11). Meningitis and severe pneumonia are treated with chloramphenicol and benzylpenicillin, while mild pneumonia is treated with trimethoprim-sulfamethoxazole, ampicillin, or amoxicillin. Data on susceptibility to these antibiotics are scarce (7, 9, 13, 15), and alternative drugs are considerably more costly. The Global Alliance on Vaccines and Immunization (GAVI) has offered financial support for conjugate Hib vaccine introduction, but response has been slow, as health ministries are cautious about a long-term commitment to an expensive vaccine. Although the costs of the vaccine are easily calculated, the costs of managing H. influenzae disease in hospitals are much less apparent and, in the face of chloramphenicol resistance, may be considerable. Here we define the antimicrobial resistance patterns of H. influenzae for one district in Kenya in order to indicate effective alternative therapies that might be used in such calculations. The study population comprised stored frozen isolates of Haemophilus influenzae cultured from blood or cerebrospinal fluid (CSF) from pediatric inpatients at the Kilifi District Hospital between January 1994 and December 2002. Before August 1998, cultures of CSF and blood were initiated on clinical suspicion of meningitis or sepsis at any point during admission. From August 1998 onwards, blood cultures were taken from all pediatric inpatients upon admission (1); lumbar puncture was undertaken systematically for patients with impaired consciousness, meningism, prostration, or seizures (other than febrile seizures) and for all children ⱕ60 days old with suspicion
of sepsis. No case of epiglottitis was seen during the study. Hib conjugate vaccine was introduced to Kenya in November 2001, and Hib disease is now uncommon, but the vaccine has not yet been introduced into neighboring Ethiopia, Sudan, or Tanzania. Before June 1998, blood was cultured in brain heart infusion broth (Oxoid, Basingstoke, United Kingdom) for 7 days at 37°C under 5% CO2 with obligatory subcultures at 2 and 7 days. Subsequently, BacTec Peds-Plus medium (Becton Dickinson, NJ) was incubated for 5 days and subcultured as indicated by the BacTec instrument. Broth medium was subcultured to 7% horse blood agar and chocolate agar. CSF samples were cultured on the same medium. Haemophilus species were identified by colony morphology, Gram staining, dependence on X and V factors, satellitism, and serotyping (16). Quality control was provided by the United Kingdom National External Quality Assessment Service (www.ukneqas.org.uk). Antimicrobial susceptibilities were determined by Etest, a reliable and appropriate method for MIC determinations in developing countries (6, 14). A broth suspension with a 0.5 McFarland standard was evenly inoculated onto Haemophilus Test Medium agar (Oxoid, Basingstoke, United Kingdom). After drying, Etests for amoxicillin, cefotaxime, chloramphenicol, amoxicillin-clavulanate, and trimethoprim-sulfamethoxazole were placed on the medium, which was incubated at 35°C under 5% CO2 for 20 h. H. influenzae (ATCC 49247) was included in every batch as a control. MICs were read at the point of complete inhibition of growth. All Etests were done at the laboratories of the manufacturer, AB Biodisk, in Solna, Sweden. Although the Etest has not been approved by the NCCLS for Haemophilus testing, susceptibility breakpoints were taken from the 2003 NCCLS guidelines (12). Raw Etest readings that fell between 1-in-2 dilutions were rounded up for analysis. Beta-lactamase activity was detected by a chromogenic cephalosporin test (BR66A; Oxoid).
* Corresponding author. Mailing address: Wellcome Trust/KEMRI Centre for Geographic Medicine Research—Coast, P.O. Box 230, Kilifi, Kenya. Phone: 254 415 25453. Fax: 254 415 22390. E-mail:
[email protected]. 3021
3022
NOTES
ANTIMICROB. AGENTS CHEMOTHER.
TABLE 1. Resistance patterns of 240 invasive isolates of H. influenzae to five antibiotics determined by Etest Antibiotic
Amoxicillin Amoxicillin-clavulanate Cefotaxime Chloramphenicol Trimethoprim-sulfamethoxazole a b
Total no. of strains
237 239 240 240 240
Susceptible strains MIC
a
ⱕ1 ⱕ4/2 ⱕ2 ⱕ2 ⱕ0.5/9.5
Intermediate strains
Resistant strains
% of total
MIC
% of total
MIC
% of total
66 100 100 66 38
2
8.4 0 0 12 5.4
ⱖ4 ⱖ8/4 ⬎2 ⱖ8 ⱖ4/76
25 0 0 22 57
4 1/19–2/38
Minimum Maximum MIC50b MIC90b MIC MIC
0.125 0.094 0.004 0.19 0.047
⬎256 0.38 12 0.75 0.38 0.5 0.023 0.012 0.016 ⬎32 0.5 8 ⬎32 ⬎32 ⬎32
All MIC values are in g/ml. MIC50 and MIC90, MICs at which 50 and 90% of isolates, respectively, are inhibited.
Isolates were serotyped by latex agglutination with polyclonal rabbit antisera (Difco Laboratories, Michigan). Serotyping results were confirmed by PCR-based capsular genotyping using primers designed to amplify the capsule type-specific regions of the cap loci in each of the six capsular types of H. influenzae, a through f (5). Hypotheses of association were tested by the chi-square test and chi-square test for trend. Logistic models were developed using backward stepwise regression to examine risk factors (age, sex, human immunodeficiency virus status, year of admission (in three strata), diagnosis of lower respiratory tract infection or meningitis, source of isolate, serotype b, and resistance to other antibiotics) for resistance to each antibiotic separately. There were 245 inpatient episodes in which H. influenzae was isolated from blood or CSF. Two frozen isolates were irretrievable, two were not carried to Sweden, and two originated from the same patient in separate admissions 10 days apart. Of 240 episodes studied, 123 (51%) occurred in males and 124 (52%) among children aged 2 to 11 months; 116 (48%) strains were isolated from CSF; and 203 strains (85%) belonged to serotype b, while others were a (n ⫽ 6), c (n ⫽ 2), f (n ⫽ 6), or nontypeable (n ⫽ 22). Human immunodeficiency virus testing among 128 patients presenting after July 1998 revealed a seroprevalence of 15%. Antibiotic susceptibility patterns are shown in Table 1. Multiple resistance was common; 40% (94/236) of isolates were resistant or intermediate to at least two antibiotics and 28% (65/236) to three antibiotics. Of 79 isolates that were not susceptible to chloramphenicol, 65 (82%) also were not susceptible to amoxicillin. The clinical coverage of amoxicillin combined with chloramphenicol was 73% (172/237), but this declined to 50% (23/46) in 2001 and 32% (10/31) in 2002. All of the amoxicillin-resistant isolates and 19 of 20 intermediate isolates were beta-lactamase positive. Susceptibility to trimethoprim-sulfamethoxazole decreased progressively from 53% among children aged ⬍2 months to 13% for children aged ⱖ60 months (P ⫽ 0.02). For amoxicillin and chloramphenicol, the prevalence of resistance peaked in the third year of life. There was no association between resistance and sex. Resistance to amoxicillin, chloramphenicol, and trimethoprim-sulfamethoxazole increased markedly throughout the 9 years of the study (Fig. 1). The prevalence of resistance to amoxicillin was 39% (78/ 201) for serotype b isolates and 6% (2/35) for others; for chloramphenicol the proportions were 38% (78/203) and 8% (3/36), respectively; for trimethoprim-sulfamethoxazole they
were 66% (134/203) and 42% (15/36), respectively (P ⱕ 0.005 for each comparison). In univariate analyses no other factor was significantly associated with resistance to any of the antibiotics. In multivariable analyses resistance was strongly associated with time and with serotype b for each antibiotic (Table 2). Resistance to amoxicillin was strongly associated with resistance to chloramphenicol. All isolates that were resistant to either amoxicillin or chloramphenicol were also resistant to trimethoprim-sulfamethoxazole. Comments. Chloramphenicol is the mainstay of inpatient therapy for H. influenzae infections in children ⬎2 months old; amoxicillin is the most commonly used drug in outpatient therapy of mild pneumonia. This study shows a dramatic and progressive increase in chloramphenicol and amoxicillin resistance over 4 years that now greatly exceeds previous estimates. In Nairobi in 1998 to 1999, chloramphenicol resistance was estimated at 5% and amoxicillin resistance at 9% (7). In neighboring Ethiopia, only 1 of 28 H. influenzae isolates from children with meningitis in 1990 was resistant to chloramphenicol and 5 were resistant to ampicillin (15); in 1999, resistance to chloramphenicol and ampicillin was found in only 2 of 74 isolates from children with meningitis (9). Because of the paucity of culture facilities in the region, the selection of antibiotics for sick children in East Africa must be based on clinical rather than microbiological criteria (2). Potential therapeutic alternatives must also be selected within the constraints of the local health care budget; for example, throughout the period of this study, the annual Kenya government expenditure on health was only $9.50 per person (17). For meningitis, potential alternative parenteral drugs are cefotaxime, ceftriaxone, or amoxicillin-clavulanate. For a 15-kg child, the costs of 7-day courses of these antibiotics in Kenya are $108, $40, and $67, respectively, compared to $5.50 for the combination of chloramphenicol and benzylpenicillin. The costs could be minimized by strengthening laboratory services and targeting prescriptions after CSF Gram stain or susceptibility results are obtained. Chloramphenicol resistance rates of ⬃20% led clinicians in two other developing countries to switch to first-line therapy with ceftriaxone. In Malawi, only patients with gram-negative bacilli on a CSF Gram stain were given ceftriaxone (8). In Papua New Guinea, all patients received ceftriaxone, but those with susceptible isolates were switched back to chloramphenicol to save costs; cure rates improved from 29% to 91% (3). These appealing approaches are clearly valid only in hospitals that routinely perform lumbar punctures for meningitis, which are still rare in Kenya (4). The World Health Organization Regional Office for Africa
FIG. 1. Percentages of 240 H. influenzae isolates resistant to three commonly used antibiotics by year. (A) Amoxicillin; (B) chloramphenicol; (C) trimethoprim-sulfamethoxazole. Black shading, resistant isolates; grey, intermediate isolates; white, susceptible isolates. In each bar absolute numbers are shown for each category except where the frequency count was zero. Results of a 2 test for trend, merging intermediate with resistant cells, are as follows: (A) 61.7 (P ⬍ 0.00005); (B) 66.0 (P ⬍ 0.00005); (C) 31.06 (P ⬍ 0.00005). 3023
3024
NOTES
ANTIMICROB. AGENTS CHEMOTHER.
TABLE 2. Logistic regression odds ratios for risk factors for resistance to amoxicillin, chloramphenicol, and trimethoprim-sulfamethoxazole Antibiotic
Risk factor
95% CIb
ORa Lower
Amoxicillin
Chloramphenicol
Trimethoprim-sulfamethoxazole
P valuec Upper
Isolated 1994–96 Isolated 1997–99 Isolated 2000–02
1.00 10.2 24.8
1.13 2.82
91.0 219
0.0004
Serotype b Chloramphenicol resistance
4.56 24.3
0.86 9.51
24.2 62.2
0.046 ⬍0.0005
Isolated 1994–96 Isolated 1997–99 Isolated 2000–02
1.00 4.89 17.5
0.55 2.09
43.5 146
0.0005
Serotype b Amoxicillin resistance
5.66 23.0
0.91 9.04
35.1 58.3
0.029 ⬍0.0005
Isolated 1994–96 Isolated 1997–99 Isolated 2000–02
1.00 1.59 5.27
0.76 2.53
3.29 11.0
⬍0.0005
Serotype b
3.03
1.35
6.82
0.0067
a
OR, odds ratio. CI, confidence interval. P values are derived from likelihood ratio tests comparing fitted to reduced models. Intermediate resistance is classified in this analysis as resistance. HosmerLemeshow goodness-of-fit 2 tests for each model yielded P values of 0.43 for amoxicillin, 0.104 for chloramphenicol, and 0.97 for trimethoprim-sulfamethoxazole. b c
has initiated bacteriological surveillance of meningitis in 22 centers throughout Africa (www.afro.who.int/hib/index.html). Through its Accelerated Development and Introduction Plan for pneumococcal vaccine, GAVI has also funded sentinel site surveillance for H. influenzae and Streptococcus pneumoniae in East Africa (www.netspear.org). Knowledge of resistance to commonly used antibiotics can determine the prescribing practice of individual hospitals, but it may also, as is the case here, strengthen the case for prevention through the Hib conjugate vaccine, a standard long since achieved throughout most of the developed world. This study is published with the permission of the director, Kenya Medical Research Institute, Nairobi. The study was supported by KEMRI and by the Wellcome Trust of Great Britain. J.A.B. held a Wellcome Trust training fellowship (053439); J.A.G.S. (061089) and M.E. (050563) hold Wellcome Trust career development fellowships; and K.M. holds a Wellcome Trust senior fellowship (061702). REFERENCES 1. Berkley, J. A., B. S. Lowe, I. Mwangi, T. N. Williams, E. Bauni, S. Mwarumba, C. Ngetsa, C. A. Hart, K. Maitland, M. English, K. Marsh, and J. A. Scott. 2005. Community acquired bacteremia amongst children admitted to a rural district hospital in Kenya. N. Engl. J. Med. 352:39–47. 2. Berkley, J. A., K. Maitland, I. Mwangi, C. Ngetsa, S. Mwarumba, B. S. Lowe, C. R. J. Newton, K. Marsh, J. A. G. Scott, and M. English. 2005. Use of clinical syndromes to target antibiotic prescribing in seriously ill children in malaria endemic area: observational study. BMJ 330:995. 3. Duke, T., A. Michael, D. Mokela, T. Wal, and J. Reeder. 2003. Chloramphenicol or ceftriaxone, or both, as treatment for meningitis in developing countries? Arch. Dis. Child. 88:536–539. 4. English, M., F. Esamai, A. Wasunna, F. Were, B. Ogutu, A. Wamae, R. W. Snow, and N. Peshu. 2004. Assessment of inpatient paediatric care in first referral level hospitals in 13 districts in Kenya. Lancet 363:1948–1953. 5. Falla, T. J., D. W. Crook, L. N. Brophy, D. Maskell, J. S. Kroll, and E. R. Moxon. 1994. PCR for capsular typing of Haemophilus influenzae. J. Clin. Microbiol. 32:2382–2386.
6. Jorgensen, J. H., A. W. Howell, and L. A. Maher. 1991. Quantitative antimicrobial susceptibility testing of Haemophilus influenzae and Streptococcus pneumoniae by using the E-test. J. Clin. Microbiol. 29:109–114. 7. Kariuki, S., J. Muyodi, B. Mirza, W. Mwatu, and J. J. D. Daniels. 2003. Antimicrobial susceptibility in community-acquired bacterial pneumonia in adults. East Afr. Med. J. 80:213–217. 8. Molyneux, E. M., A. L. Walsh, H. Forsyth, M. Tembo, J. Mwenechanya, K. Kayira, L. Bwanaisa, A. Njobvu, S. Rogerson, and G. Malenga. 2002. Dexamethasone treatment in childhood bacterial meningitis in Malawi: a randomised controlled trial. Lancet 360:211–218. 9. Muhe, L., and K. P. Klugman. 1999. Pneumococcal and Haemophilus influenzae meningitis in a children’s hospital in Ethiopia: serotypes and susceptibility patterns. Trop. Med. Int. Health 4:421–427. 10. Mulholland, K., S. Hilton, R. Adegbola, S. Usen, A. Oparaugo, C. Omosigho, M. Weber, A. Palmer, G. Schneider, K. Jobe, G. Lahai, S. Jaffar, O. Secka, K. Lin, C. Ethevenaux, and B. Greenwood. 1997. Randomised trial of Haemophilus influenzae type-b tetanus protein conjugate vaccine for prevention of pneumonia and meningitis in Gambian infants. Lancet 349:1191–1197. 11. Mwangi, I., J. Berkley, B. Lowe, N. Peshu, K. Marsh, and C. R. J. Newton. 2002. Acute bacterial meningitis in children admitted to a rural Kenyan hospital: increasing antibiotic resistance and outcome. Pediatr. Infect. Dis. J. 21:1042–1048. 12. National Committee for Clinical Laboratory Standards. 2003. Methods for dilution antimicrobial susceptibility tests for bacteria that grow aerobically; MIC Testing Supplemental Tables M100-S13 (M7). National Committee for Clinical Laboratory Standards, Wayne, Pa. 13. Rowe, A. K., M. S. Deming, B. Schwartz, A. Wasas, D. Rolka, H. Rolka, J. Ndoyo, and K. P. Klugman. 2000. Antimicrobial resistance of nasopharyngeal isolates of Streptococcus pneumoniae and Haemophilus influenzae from children in the Central African Republic. Pediatr. Infect. Dis. J. 19:438–444. 14. Rowe, A. K., B. Schwartz, A. Wasas, and K. P. Klugman. 2000. Evaluation of the Etest as a means of determining the antibiotic susceptibilities of isolates of Streptococcus pneumoniae and Haemophilus influenzae from children in the Central African Republic. J. Antimicrob. Chemother. 45:132–133. 15. Tetanye, E., D. Yondo, A. C. Bernard-Bonnin, P. F. Tchokoteu, I. Kago, M. Ndayo, and J. Mbede. 1990. Initial treatment of bacterial meningitis in Yaounde, Cameroon: theoretical benefits of the ampicillin-chloramphenicol combination versus chloramphenicol alone. Ann. Trop. Paediatr. 10:285– 291. 16. Topley, W. W. C., G. S. Wilson, A. Balows, and B. Duerden. 1997. Topley and Wilson’s microbiology and microbial infections: systematic bacteriology, 9th ed., vol. 2. Hodder Arnold, London, United Kingdom. 17. World Bank. 2002. World Development Indicators 2002. [Online.] www .worldbank.org/data/wdi2002/.
