Background
Life-threatening infections present major challenges for health systems in the developing world. Managing these infections requires accurate data on the prevalence of specific causative organisms and their antimicrobial susceptibility profiles. However, microbiologic surveillance data providing this crucial information are scarce. In Malawi, a typical sub-Saharan African country, routine microbiologic culture and sensitivity testing are not performed due to lack of personnel, equipment and financial resources [
1]. Instead, antimicrobial therapy is empirical and a small repertoire of antimicrobials is overused. This approach, although relatively inexpensive, leads to the emergence of antibiotic resistance and hence sub-optimal clinical outcomes. Antimicrobial resistance in Malawi may also be occasioned by uncontrolled medical and veterinary use of antibiotics by the private sector and the community.
Antimicrobial resistance, an emerging global public health threat, is increasing worldwide and has been reported in Malawi [
2‐
5]. Of particular concern is the reported resistance to many of the first-line antibiotics for treating bacteraemia and other life-threatening infections [
6]. However, some of these initial studies are now outdated, had inadequate sample sizes, or were retrospective. Consequently, the problem of antimicrobial resistance requires a more complete description [
1,
6‐
8].
At Kamuzu Central Hospital (KCH), Lilongwe, Malawi we prospectively studied patients suspected of having an infection in order to identify their causative organism and the antimicrobial susceptibility using disk diffusion method. This study was conducted to inform empirical treatment choices and also to generate hospital-wide baseline data.
Patients and study site
We conducted a cross-sectional study at KCH, a 750-bed government medical centre serving Lilongwe district (population 1,897,167) that also provides referral services to the central region (population 5,491,034) [
9]. The study period, from July 2006 to December 2007, covered two dry seasons and one and a half rainy seasons. During evaluation for admission, patients presenting with a suspected infection were evaluated using appropriate microbiologic procedures. Specimens were only collected at admission, and were not collected from those seen on out-patient basis. This report includes specimens from blood, cerebrospinal fluid (CSF), joint fluid, bone biopsy, or pus collections.
We enrolled patients with pneumonia, meningitis, bone and joint infections, peripartum infections and other febrile clinical presentations (Table
1). Patients who were already on antibiotics
1, too sick
2 or unwilling to undergo study procedures were excluded from the study. Due to logistical constraints, patients presenting at night and on weekends were also excluded. All laboratory analysis was conducted in the University of North Carolina (UNC) Project laboratory at KCH. Institutional Review Boards in both Malawi and at UNC-Chapel Hill approved the study and considered the collection of biological samples for diagnostic purposes to be the standard of care.
Table 1
Characteristics of Study Population (N = 2056)
Overall | | 2056 | 848 (43.3) | 1110 (56.7) | 26.0 ± 16.7 |
Paediatrics | | 543 | 254 | 218 | 4.4 ± 4.9 |
| Febrile child | 391 (89.5) | 192 (90.6) | 166 (87.8) | |
| Neonates | 46 (10.5) | 20 (9.4) | 23 (12.2) | |
Medicine | | 1213 | 506 | 692 | 34.8 ± 12.0 |
(adults) | Meningitis | 294 (25.1) | 133 (27.4) | 157 (23.3) | |
| Pneumonia | 360 (30.7) | 116 (23.9) | 240 (35.7) | |
| Fever/sepsis | 518 (44.2) | 236 (48.7) | 276 (41.0) | |
Obs/Gyn | | 127 | 0 | 127 | 24.7 ± 6.2 |
| Meningitis | 9 (7.5) | 0 | 9 (7.5) | |
| Pneumonia | 7 (5.8) | 0 | 7 (5.8) | |
| Fever/sepsis | 14 (11.7) | 0 | 14 (11.7) | |
| Prepartum sepsis | 8 (6.7) | 0 | 8 (6.7) | |
| Puerperal sepsis | 46 (38.3) | 0 | 46 (38.3) | |
| Gynaecological sepsis | 36 (30.0) | 0 | 36 (30.0) | |
Surgery | | 166 | 89 | 73 | 21.6 ± 15.8 |
(adults and children) | Osteomyelitis | 9 (5.7) | 3 (3.6) | 5 (7.2) | |
| Necrotising fasciitis | 3 (1.9) | 2 (2.4) | 1 (1.5) | |
| Abdominal Abscess | 84 (53.5) | 42 (50.6) | 40 (58.0) | |
| Burn | 8 (5.1) | 3 (3.6) | 5 (7.2) | |
| Septic arthritis | 53 (33.8) | 33 (39.8) | 18 (26.1) | |
Prior to study implementation, information and training sessions were held with the staff of Medicine, Pediatrics, Surgery, and Obstetrics and Gynecology departments. Content of the training sessions included selection criteria for patient identification and appropriate sample collection techniques. Interpretation and use of study results were discussed at clinical meetings.
In each department, one lead staff member was identified to facilitate study implementation, including assurance that all eligible patients were identified. A study assistant was engaged full-time to track specimens from around the hospital to the laboratory, return laboratory results to the clinicians, manage study supplies and provide general logistical support to the study. We conducted monthly review meetings with the lead staff members to discuss study progress and address logistical problems.
Discussion and Conclusions
We conducted a large, cross-sectional microbiologic study of patients with presumed infections seen at a large medical centre in Malawi. We observed that the common bacterial causes of bacteraemia differ between children and adults, and that S. aureus is extremely common in blood, abscesses, joints, and bone. We also observed widespread antimicrobial resistance to commonly used antibiotics.
Our findings highlight the variation in bacteremia etiology, both geographically and temporally, within sub-Saharan Africa [
8,
12,
13]. In our setting, gram positive bacteria, especially
S. aureus and
S. pneumoniae, were the most common cause of bacteraemia and infections of other anatomic sites. An earlier study in neonates in Malawi observed a predominance of
Salmonella and other gram negative bacilli in patients with bacteremia [
12]. We also isolated
S. aureus more commonly than was observed elsewhere [
7,
14]. However, in The Gambia,
S. aureus and
S. pneumoniae caused almost three quarters of bacteraemia cases in an urban hospital [
15]. In addition, this is a population which did not receive the pneumococcal conjugate vaccine.
Our observation that
S. aureus was the most commonly isolated organism must be interpreted cautiously. Robust organisms, such as
S. aureus and
Salmonella, are more likely to be isolated than fastidious ones [
16]. Thus, the prevalence of the robust organisms could possibly be biased upwards. However, we frequently isolated
S. pneumoniae, a less robust organism, which suggests that our aerobic culture techniques were adequate.
Although we had relatively few CSF cultures, the results were highly informative. C. neoformans was a common cause of meningitis, second only to S. pneumoniae. The high prevalence of Cryptococcus was most likely due to a high prevalence of HIV infection in the patient population at KCH, where up to three-quarters of medical inpatients are HIV infected. We can not comment on the frequency of tuberculous meningitis since culture for that organism was not attempted in this study population. We also observed very few cases of Haemophilus influenzae meningitis or bacteraemia, possibly due to the introduction of routine childhood H. influenzae b vaccination in 2002. Neisseria meningitidis also was not recovered reflecting this as a rare cause of meningitis in this population.
We found widespread bacterial resistance to almost all of our commonly used antibiotics. Our findings contrast starkly with the limited resistance observed in Malawi between 1996 and 2001, although resistance to first line antibiotics for bacteremia was observed over a decade ago [
6,
12]. Of particular concern is high incidence of bacteraemia caused by non-Typhi
Salmonella, which has increased with development of multi-drug resistance [
17]. In our setting,
Salmonella is a common gram negative bacterium infecting blood and joints and it was susceptible to gentamicin and ceftriaxone in vitro. The problem of fluoroquinolone resistance in
Salmonella enterica serotypes other than Typhi is not well studied in Africa.
In a separate Malawi study, no fluoroquinolone resistance was reported among
Salmonella isolates [
17]. In a study from Senegal, approximately 5% of isolates showed low level fluoroquinolone resistance [
18]. We found none of the 25
Salmonella isolates tested to be nalixidic acid resistant although interestingly close to half (11/25) were nalidixic acid intermediate (Table
3). A previous study suggests
Salmonella strains with intermediate susceptibility by disk diffusion may have MICs in a range consistent with low level ciprofloxacin resistance [
19]. Unfortunately ciprofloxacin MICs were not available to determine if these nalidixic acid intermediate isolates did, in fact, have low level ciprofloxacin resistance. This issue deserves close scrutiny in the coming years.
Our findings of widespread antimicrobial resistance have a strong bearing on the current treatment choices. Malawi, like most resource poor countries, does not provide routine microbiologic culture and sensitivity testing [
3]. Instead, the Malawi National Treatment Guidelines provides for initiation of antimicrobial therapy empirically, based on clinical observation and history, which is often broad spectrum [
16,
20‐
23]. However, the development of antimicrobial resistance in this context can have major clinical consequences. In Tanzania, antimicrobial resistance was a significant risk for death in children with bacteremia [
23]. Antimicrobial resistance can also lead to treatment failures, prolonged drug treatment courses and protracted hospital admissions [
23].
Considering the potential consequences of increased antimicrobial resistance, our findings support a review of the current antibiotic choices in Malawi. Of the antibiotics we examined, only three are not widely used and show good susceptibility. Ceftriaxone and ciprofloxacin retain activity against gram negative bacteria, including Salmonella spp. Gram positive organisms, including S. aureus and S. pneumoniae, remain susceptible to clindamycin, and may also be susceptible to third-generation cephalosporins. Revision of the national treatment guidelines with inclusion of clindamycin, and ciprofloxacin or ceftriaxone, coupled with promotion of their prudent use, should improve patient outcomes and potentially save costs of providing clinical services.
This study represents an account of community-acquired infections in that it included only specimens collected at admission, and did not include subsequent specimens and specimens from patients who had already been on antimicrobial therapy prior to presentation. Unfortunately, we were unable to include patients who presented to the hospital outside of the daytime hours, due to the constraints of study personnel. In addition, the maternity ward was located outside the main hospital campus, limiting the enrolment of patients there. Thus, our study population was not a complete assessment of all infections presenting to KCH during the study period. Also clinical outcomes were not collected to triangulate our laboratory findings. More advanced microbiologic testing such as minimum inhibitory concentration (MIC) and extended spectrum beta-lactamases (ESBLs), which would have helped refine some of our basic findings, were not performed. Nonetheless, we were able to enrol a large number of patients with a diverse spectrum of clinical syndromes and also perform basic identification and antimicrobial susceptibility testing. Consequently, our results provide a large microbiologic assessment of bacterial infections in Malawi.
The widespread antimicrobial resistance in our study is concerning. Choices for empirical therapy in Malawi should be revised accordingly. Newer antibiotics, though more costly, must be used more widely, but judiciously, to limit the further spread of antimicrobial resistance. The establishment of a dependable, accurate and timely microbiological service and surveillance system along with optimal use of this service by the clinicians should considerably improve the care of patients. Finally, prudent use of antimicrobials should be promoted.
Competing interests
The authors declare that they have no competing interests. The study was partially funded by University of North Carolina at Chapel Hill Center for AIDS Research (CFAR) NIH funded program #P30 AI50410.
Authors' contributions
MHM was principal investigator, wrote first drafts of the study protocol and manuscript, contributed to laboratory and data analysis and interpretation. WCM performed the statistical analysis. IFH, PHG and MCH contributed to study conception and design, data interpretation, revising manuscript for intellectual content. GM and RC participated in data entry, cleaning and analysis. GJ, FM and DK participated in study concept and design. All authors read and approved the final manuscript.