Herbal products are used by human immunodeficiency virus (HIV)-infected individuals regardless of safety or efficacy concerns. In this study, we examined the
microbiological quality of herbal preparations marketed to HIV-infected individuals. A convenience sample (N = 24) of herbal products was obtained from retailers
in Nairobi, Kenya in 2007. Petrifilm plate count methods were used to estimate total aerobic bacteria (APC), coliform, Escherichia coli, Staphylococcus
aureus and yeast and mould counts. APC counts ranged from an estimated 1.5 × 101 colony forming units (CFU)/g to
7.1 × 108 CFU/g. Total and faecal coliform counts ranged from an estimated <10 CFU/g to
3 × 106 CFU/g.
E. coli load ranged from <10 CFU/g to 5 × 101 CFU/g and S. aureus counts ranged from an estimated <10
2.5 × 103 CFU/g. Yeast and mould counts ranged from an estimated <10 CFU/g
to 9 × 104 CFU/g. An evaluation
using the World Health Organization limits for medicinal herbs found a percentage of samples to contain microorganisms above allowable limits: 33% (APC), 50% (coliforms)
and 33% (yeast and moulds). A total of 67% of samples contained S. aureus loads above the United States Pharmacopeia standard. We suggest that the introduction
of quality-control measures and safe handling practices for the selling of medicinal herbs and botanicals in Kenya would be beneficial in reducing the potential health
risks for immunocompromised consumers of these products.
Over 33 million people are infected with the human immunodeficiency virus worldwide.1 Sub-Saharan Africa and developing countries in
general have the highest
HIV prevalence and new HIV infections.1 In addition to costly and inaccessible antiretroviral drugs, sub-Saharan Africa has a high
patient to doctor ratio
(40 000:1). These factors lead to the increased use of medicinal herbs, especially for HIV-related
complications.2,3 In fact, numerous authors, including
the World Health Organization (WHO), report that 30% – 70% of HIV-infected individuals worldwide use herbal products.
Although consumers consider herbal products to be safe, microbial contamination in medicinal herbs is a concern, especially for HIV-infected individuals as a result of
their compromised immunity.7 Medicinal herbs frequently contain microorganisms indigenous to the soil and plants where they are grown;
sanitation in primitive harvest and postharvest conditions allows their survival.8,9 A South African study reported
contamination of herbal products with
Bacillus spp., Enterobacteriaceae spp., Salmonella spp., Staphylococcus
aureus, Penicillium and Aspergillus.10
Moreover, elevated levels of bacterial and fungal contaminants, such as Penicillium spp., Aspergillus and Fusarium, have been observed in herbs and
The purpose of this study was to examine the microbiological quality of herbal products marketed to HIV-infected individuals in Kenya and to evaluate the safety of
products using microbial contamination limits set by the WHO.12 The United States Pharmacopeia (USP)
standard13 was used for S. aureus
because there is no applicable WHO standard. Two established WHO microbial load limits were used: Limit I for herbal products to which boiling water is added in
order to prepare herbal teas and infusions and Limit II for herbal products prescribed for internal use with no heat treatment.12 To our
knowledge, no other
study has investigated the microbiological quality of herbal products used by HIV-infected individuals in Kenya.
Sample collection and description
Laboratory work was registered and approved by the Institutional Biosafety Committee of the University of Arkansas (Fayetteville, AR, USA). From May to June 2007, a
convenience sample of 24 different medicinal herbs was purchased from herbal product retailers in Nairobi, Kenya. Individual samples were not randomly collected because
(1) certain parts of Nairobi are inaccessible for security reasons, (2) some herbalists had only liquid mixtures, concoctions or extracts which were unsuitable for
sampling because of shipping restrictions and (3) some retailers refused to sell their products to us for reasons unknown. For 20 of the samples, duplicates were bought
at different time points; for the remaining four samples, duplicates were not available because the retailer either ran out of the product or was unavailable to sell a
second sample. Samples were purchased from the following markets: Kibera, Kawangware, Shaurimoyo, along Ngong Road, Kamukunji, Ngara, Westlands, Kangemi and downtown
Nairobi. To acquire samples for this study, one of us posed as a relative of a HIV-infected person. Herbal retailers or practitioners were requested to recommend herbal
products for HIV-infected individuals. Using aseptic techniques, the herbal samples were packaged in labelled sterile bags and shipped to the University of Arkansas;
each bag was treated as an experimental unit. Samples were mainly ground dry mixtures or pastes. No botanical identification of the samples was carried out as this was
beyond the scope of our experiment.
Samples were tested for aerobic plate count (APC), coliform count, Escherichia coli count, S. aureus count, and yeast and mould count
(YM). Standard laboratory methods and procedures, consistent with the American Public Health Association’s guidelines, were
used.14,15 Methods for using Petrifilm™ were followed according to instructions from 3M Microbiology
Products (3M Microbiology
Products, St. Paul, MN, USA). PetrifilmTM methods are reproducible, practical and accurate.16
Using aseptic techniques, 1 g of each herbal product sample was mixed with 99 mL Butterfield’s phosphate buffer (Weber Scientific
Hamilton, NJ, USA) and homogenised for 2 min at 200 rpm (Stomacher® 400, Seward Ltd, London, UK). To obtain a pH between 6.6
and 7.2, either 1 M sodium hydroxide or hydrochloric acid was added before serial dilutions were done. Aliquots of 1 mL for each dilution
were plated in duplicate on Petrifilm™ media (3M Company, St. Paul, MN, USA) for APC, E. coli, S. aureus counts and YM. A
Quebec dark field colony counter was used to manually count the colony forming units (CFU) per gram.
The negative control for each sample was Butterfield’s phosphate buffer while positive controls were S. aureus (ATCC No. 25923),
E. coli (ATCC No. 25922), Rhizopus nigricans (ATCC No. 14038) (all obtained from Presque Isle Cultures, Erie, PA, USA), and
Saccharomyces cerevisiae (ATCC No. 2601; Quality Technologies LLC, Newbury Park, CA, USA).
Liquid samples were not included in this study because of shipping regulations restricting their transport. It is conceivable that liquid samples
and dry samples are contaminated with different microorganisms because of the difference in moisture content and preparation methods. Although the
samples in this study were not randomly collected, they are representative of herbal products available in different areas of Nairobi at the time,
but it is acknowledged that they may not reflect a possibly larger range of contamination issues. However, the difficulty in acquiring samples made
the use of a convenience sample a necessity.
Microbial counts were transformed to common logarithmic values for statistical analysis. SAS® was used for all statistical
analysis.17 A significance level of 0.05 was used for all tests. An analysis of variance was used to test for differences between
means for each of the microorganisms. A t-test was used to find significant differences between USP or WHO limits and sample microbial
counts when sample means exceeded the established microbial limits. The degree of variability of microorganisms was evaluated and is presented in
box plots; correlation coefficients were calculated to assess combination patterns in the occurrence of the five microorganisms. An overlay plot
was constructed to graphically illustrate microorganisms found in each of the 24 samples.
Aerobic bacteria were detected in all 24 samples. APC was the highest count (CFU/g) in 96% of samples, when compared to the other four
microorganisms (Figure 1). Further analysis demonstrated that 4% of samples had aerobic bacteria levels above the WHO allowable limits for teas
and infusions, and 33% of the samples were above the WHO allowable limit for internal use (Table 1). APC counts ranged from an estimated 1.5
× 101 CFU/g to 7.1 × 108 CFU/g with sample I and sample X containing the lowest and
respectively (Figure 1).
Summary of results showing the distribution of microorganisms (colony forming units/g) in 24 samples of medicinal herbs marketed to HIV-infected individuals in Kenya.
A plot showing the number of microorganisms contained in 24
samples, labelled A–X, of medicinal herbs marketed to HIV-infected individuals
Coliforms were detected in 96% of the samples (Figure 1). Coliform counts ranged from an estimated <10 CFU/g to
3 × 106 CFU/g and were the second highest in both CFU counts per gram and occurrence (Figure 1). Of the samples,
13% had coliform levels above the WHO allowable limits for teas and infusions, and 50% had counts above the WHO allowable limit for internal use
(Table 1). E. coli was detected in 25% of the samples (Figure 1), ranging from <10 CFU/g to
5 × 101 CFU/g; all E. coli counts were lower than the WHO limits (Table 1). S. aureus
counts ranged from an estimated <10 CFU/g to 2.5 × 103 CFU/g, with 67% of samples containing
S. aureus in quantities above that recommended by USP (Table 1). Yeast and mould counts were detected in all 24 samples tested
(Figure 1) and ranged from an estimated <10 CFU/g to 9 × 104 CFU/g. Further analysis showed that 13% of
samples had YM levels above WHO allowable limits for teas and infusions and 33% of the samples contained levels above the WHO allowable limit for
internal use (Table 1). No attempt was made to identify yeasts and moulds to the species level, as this was beyond the scope of this study.
Side-by-side box plots show the distribution of microorganisms within the 24 samples (Figure 2). The distributions of E. coli and
S. aureus are highly skewed to the right and left, respectively. Several outliers are shown for APC, coliforms and E. coli
Distribution of microorganisms in 24 samples of medicinal herbs
marketed to HIV-infected individuals in Kenya.
Correlation coefficients were calculated to investigate patterns in the occurrence of the different microorganisms within samples. Counts of APC
in each sample were positively correlated to counts of coliforms (r = 0.65; 95% confidence interval = 0.33–0.83). No other significant
correlations were observed between the numbers of the microorganisms in the samples (data not shown).
Aerobic bacteria occur naturally in plant-based products and are the most commonly used quality parameter for assessing the hygiene status of food
and drug samples.18 The WHO APC limit provided for herbal products made for internal use is 5 log units/g, and over one third of the
samples in this study had APCs higher than this limit. This finding implies that conditions during harvest or postharvest were unsanitary. Storage
tends to decrease APC counts in dried foods19; therefore variation in counts between samples may be as a result of differences in
Coliforms are found mainly in soil, vegetation and the faeces of warm-blooded animals.9 Previous studies have reported high coliform
counts, up to 3.4 × 104 CFU/g, in other botanical products such as chamomile.20 In this study,
we observed that
50% and 13% of the samples failed to meet the WHO requirement of less than 3 log units/g for teas and 4 log units/g for internal use. High
coliform counts are an indication of poor hygiene and a lack of sanitation measures. Contamination results from several sources, including improper
cleaning procedures and open air drying of plant materials which results in contamination with soils and dust.21
Although none of the samples in this study contained E. coli counts above the WHO standards of 2 log units/g for teas or
infusions and 1 log unit/g for internal use, 25% of the samples were contaminated with E. coli at low levels. Detection of
E. coli confirms faecal contamination, which is directly associated with unsanitary conditions. Because heat treatment of herbal
products results in a considerable reduction in the viable counts of E. coli,19 consumer education is critical to prevent
foodborne illnesses among consumers.
The presence of S. aureus in food and herbal products is associated with unsanitary human handling. Botanicals frequently
undergo human handling as they are collected, cleaned, pooled, dried, packed and dispensed. S. aureus was detected in 67% of samples, at
levels that were above 1 log unit/g. Other studies in Nigeria and South Africa have reported similar S. aureus
Fungal contamination of herbal products chiefly occurs during a slow drying process, because of inadequate drying or during postharvest storage
if relative humidity is high and temperatures are favourable.21 In this study, while all samples were contaminated with fungi at
different concentrations, over one third of the products did not fall within acceptable microbial limits for internal use and 13% failed to fall
within the limit for teas or infusions. Such contaminated products may cause fungal infections or other serious health complications as a result
of mycotoxin accumulation from toxin-producing fungi such as Aspergillus parasiticus and Aspergillus flavus.21 In fact,
several studies have reported the presence of mycotoxins in botanical preparations.9,23
Although S. aureus is infrequently found in medicinal herbs, our and previous findings suggest that there is a need to develop
microbial limits for S. aureus for countries that rely on WHO quality standards for herbal products. Moreover, there is a need
to re-evaluate quality-control standards for herbal and traditional medicines to reduce the health risks associated with microbial contamination
in developing countries.
Training in food safety procedures may improve the safety of the products and reduce health risks to consumers. It should be noted that during
sample collection in this study, several retailers recommended not using boiling water and encouraged the addition of the botanical to food or its
use as a ’paste’ by adding it to water or milk. Although boiling water will not kill all the microorganisms or spores in a botanical
mixture, it may significantly reduce the microbial load.
As a safety precaution, botanicals are not recommended for infants because of the risk of infant botulism, which
is caused by Clostridium botulinum spores. Although Clostridium spp. were not tested in this study, given the microbial
numbers seen in this project, and the prevalence of C. botulinum in medicinal herbs,15 the possibility of infant botulism is
a concern. Clostridium spp., Listeria spp., Shigella spp., and Bacillus cereus should be tested for in future
studies, as these are common contaminants of herbs and spices.
In addition, dates of manufacture and expiration were unknown in 98% of the samples in this study. Consequently it was not possible to estimate
the effects of storage on microbial loads in the samples, although storage is known to influence microbial activity and counts.19
The use of herbal or botanical products is part of the health-care culture in Kenya. When standard pharmaceuticals are not readily available or
affordable, consumers use botanical products as their only option. Microorganism contamination of these products was found to frequently exceed
the accepted limits. However, given the environmental conditions and how they were dispensed, the levels of contamination could have been greater.
We recommend that quality-control measures and safe handling practices be established for medicinal herbs or botanicals sold in Kenya. Health-care
providers and public health professionals should be aware that these products are being sold to HIV-infected individuals and that they may serve as
a source of infection and cause potential health risks to these immunocompromised individuals.
L.K. thanks the School of Human Environmental Sciences, University of Arkansas, and the American Association of University Women (AAUW) for the
financial support that enabled her to complete her MS degree.
We declare that we have no financial or personal relationships which may have inappropriately influenced us in writing this paper.
L.K. performed the experiments and contributed to the writing of the manuscript. J.C.F. was the project leader and was in charge of the project
design and contributed to the writing and editing of the manuscript. E.E.G. was responsible for the experimental design, the statistical analysis
and the editing of the manuscript.
1. Joint United Nations Programme on HIV/AIDS (UNAIDS) and World Health Organization (WHO). AIDS epidemic update. Towards universal access:
Scaling up priority HIV/AIDS interventions in the health sector. Progress report 2009;UNAIDS/09.36E/JC1700E. Geneva: UNAIDS and WHO; 2009.
2. Langlois-Klassen D, Kipp W, Jhangri GS, Rubaale T. Use of traditional herbal medicine by AIDS patients in Kabarole District, Western Uganda.
Am J Trop Med Hyg. 2007;77(4):757–763.
3. Bodeker G, Carter G, Burford G, Dvorak-Little M. HIV/AIDS: Traditional systems of health care in the management of a global epidemic. J Altern
Comp Med. 2006;12(6):563–576.
4. Liu JP, Manheimer E, Yang M. Herbal medicines for treating HIV infection and AIDS. Cochrane Database Syst Rev. 2005;2:1–23.
5. Harnack JL, Rydell SA, Stang J. Prevalence of use of herbal products by adults in Minneapolis/St.Paul, Minn, Metropolitan Area. Mayo Clin Proc.
6. Zhang X. Traditional medicine: Definitions. WHO Fact sheet 2007 [homepage on the Internet]. c2008 [cited 2008 Dec 18]. Available from:
7. Ernst E. Complementary medicine – is it effective and safe? Drug Inf J. 1999;33:291–295.
8. Farkas J. Spices and herbs. In: Lund BM, Baird-Parker TC, Gould GW, editors. The microbiological safety and quality of food. Gaithersburg, MD:
Aspen Publishers, 2000; p. 897–918.
9. Candlish AAG, Pearson SM, Aidoo KE, Smith JE, Kelly B, Irvine H. A survey of ethnic foods for microbial quality and aflatoxin content. Food
Addit Contam. 2001;18(2):129–136.
10. Govender S, Du Plessis-Stoman D, Downing TG, Van de Venter M. Traditional herbal medicines: Microbial contamination, consumer safety and the
need for standards. S Afr J Sci. 2006;102(5):253–255.
11. Kneifel W, Czech E, Kopp B. Microbial contamination of medicinal plants. Planta Med. 2002;68:5–15.
12. World Health Organization. Quality control methods for medicinal plant materials. Geneva: World Health Organization, 1998; 24–35.
13. US Pharmacopeial Catalog [homepage on the Internet]. c2008 [cited 2008 Apr 24]. Available from:
14. Downes PF, Ito K, editors. Compendium of methods for the microbiological examination of foods. 4th ed. Washington DC: American Public Health
15. US Food and Drug Administration. Bacteriological analytical manual. Gaithersburg, MD: AOAC International, 1998; p. 12:01–15:08.
16. Blackburn CW, Baylis CL, Petitt SB. Evaluation of PetrifilmTM methods for enumeration of aerobic flora and coliforms in a wide range of foods.
Lett Appl Microbiol. 1996;22(2):137–140.
17. SAS. Version 9.1. Cary, NC: SAS Institute Inc.; 2003.
18. Czech E, Kneifel W, Kopp B. Microbiological status of commercially available medicinal herbal drugs – A screening study. Planta Med.
19. Montville JT, Mathews RK. Food microbiology: An introduction. Washington DC: American Society of Microbiology, 2005; p. 20–21.
20. Foote JC, Thompson LD, Wester DB. Microbiological quality of chamomile. J Am Diet Ass (Suppl). 2005;105:38.
21. Sharma A. Irradiation to decontaminate herbs and spices. In: Peter KV, editor. Handbook of herbs and spices. Cambridge, UK: Woodhead
Publishing, 2001; p. 60–73.
22. Adeleye IA, Okogi G, Ojo EO. Microbial contamination of herbal preparations in Lagos, Nigeria. J Health Popul Nutr.
23. Lutomski J, Kedzia B. Mycoflora of crude drugs: Estimation of mould contaminations and their toxicity. Planta Med. 1980;40:212–217.