Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Shortcuts in cassava processing and risk of dietary cyanide exposure in Zaire

Mayambu Banea, Nigel H. Poulter, and Hans Rosling



Konzo is a paralytic disease that has been attributed to shortcuts in cassava processing resulting in dietary cyanide exposure. Through a combination of anthropological and chemical methods, this study in Zaire provides support for such an aetiology. Observations as well as focus group and individual interviews revealed that women in the konzo-affected Bandundu region, when in a hurry to sell cassava, shorten soaking to only one night, whereas a period of four nights was strictly adhered to in an unaffected neighbouring region. A village-based processing experiment revealed that apparently small differences in the sequence and length of pounding and sun-drying of short-soaked roots considerably change the amounts and types of cyanogens that remain in the cassava flour. Elucidation of the chemical mechanisms involved now makes it possible to estimate the risk for cyanide exposure from different shortcuts and also to give appropriate advice to prevent toxic effects from cassava.


Editor's note

This paper provides another example of the usefulness of a qualitative anthropological approach to determining behaviours that are not necessarily revealed by formal questionnaires. (See the Editor's Note to the article by Ríos et al. on page 115 above.)



Cassava is grown for its starchy roots, which constitute the main staple food in many sub-Saharan countries. The edible leaves also provide an important supply of nutrients in many cassava-consuming populations [1]. Bitter cassava cultivars, grown for their high yields, may contain considerable amounts of the cyanogenic glucoside linamarin, together with smaller amounts of a similar substance called lotaustralin [2]. To avoid dietary cyanide exposure, the glucosides must be removed before consumption through an effective processing method. During processing, the root tissue is disintegrated, facilitating endogenous enzymic hydrolysis of the glucosides to yield cyanohydrins. These rapidly dissociate to volatile hydrogen cyanide at a pH above 6 [3]

In the Bandundu region of Zaire, bitter cassava roots constitute the almost exclusive staple food. The roots are used only after fermentation by soaking in water, which is known to be an effective processing method [4]. Shortcuts in processing that result in high dietary cyanide exposure have been implicated in the causation of the paralytic disease konzo, which is common in central and southern Bandundu [5]. Konzo is an upper motor neuron disease that results in permanent spastic paralysis of both legs. It has an abrupt onset, and dry season epidemics mainly among children and women have been reported from Mozambique [6, 7] and Tanzania [8, 9]. The name originates from a local designation in the first affected population in southern Bandundu [10].

A recent epidemiological study in central Bandundu revealed an association between konzo and the frequent use of shortcuts in soaking which started in 1975, when cassava became a major cash crop after the construction of a new highway to Kinshasa, the capital of Zaire [5]. In contrast, no shortcuts in processing and no outbreaks of konzo have been reported from gas-Zaire, a region that has supplied Kinshasa with cassava for many decades. Detailed information on shortcuts used in processing is difficult to obtain by conventional household interviews, since this practice is due to food shortage and poverty and is thus considered taboo. We therefore decided to use qualitative methods to document the shortcut methods used [ 11 ].

The aims were first to identify the different shortcut methods in cassava processing used in konzo-affected villages in Bandundu and in unaffected villages in gas-Zaire, and second to estimate the content of cyanogens in flour obtained from these practices and hence the potential dietary exposure to cyanide.


Study area and methods

Bandundu and gas-Zaire regions are the main producers of cassava and other foods for Kinshasa [1214]. People in the two regions speak several local languages but have the same lingua franca, Kikongo, as well as a relatively similar cultural background.

Cassava-processing surveys were conducted at the end of the dry season in September 1989.

In the villages studied in Bandundu region, 26 women from konzo-affected and 24 from unaffected households were randomly selected from a population census record for individual interviews. In addition, a focus group interview [11] with 10 women was undertaken in each village, with the participating women chosen randomly from the remaining households of both types. Finally, participatory observation and unstructured interviews on the harvesting, processing, and consumption of cassava were carried out over three days, assessing one woman from each type of household in her daily work.

In each unaffected village studied in gas-Zaire, 10 women were chosen randomly for household interviews and another 10 for a focus group interview and participatory observation. Unstructured interviews were carried out while assessing a woman during five days of agricultural and household work.

A cassava-processing experiment was conducted in the konzo-affected village of Lumbi [5] in Bandundu in September 1990. Two local women were asked to process cassava roots by three identified shortcut methods under the supervision of the investigators. Fifteen flour specimens were obtained from each shortcut method. They were collected in double polyethylene bags, immediately frozen at -10° C, and sorted prior to analysis. The contents of cyanogenic glucosides, cyanohydrins, and hydrogen cyanide in the flours were determined by an enzymatic assay [151 specifically developed for these cyanogens.



The main steps in the normal and shortcut methods of processing for the various storable products are outlined in figure 1, and the products are shown in figure 2. In both areas, all the processing methods included fermenting the roots by soaking in water; they differed in the number, sequence, and length of the subsequent steps. Shortcuts were identified only in the Bandundu villages.

FIG. 1. Sequence and length of the main steps in normal and shortcut methods of processing cassava roots


Normal methods


In the Bandundu villages, the roots are normally peeled directly after harvesting and soaked for three nights in fish ponds or some other standing water.

To make boulettes, water is extracted from the soaked roots by squeezing, and they are pounded into a mash. Further water is removed by squeezing portions of the mash by hand into 5-10-cm balls, which are then sun-dried for at least four days. These boulettes are stored until needed, when they are pounded into flour. This was the main processing method used for generations until 1975.

Cossettes were introduced when cassava became a cash crop in 1975. They are obtained by sun-drying soaked, whole roots for four days. They are stored until sold, or pounded into flour on the day of consumption.

Miettes are made by pounding wet roots into 210-mm pieces, which are then spread on a mat for one day to sundry, before being pounded into flour.

All the processed cassava-root products in the konzo-affected villages are ultimately pounded into a flour which is consumed as fufu, a glutinous stiff porridge obtained by mixing the flour with boiling water.


In the Bas-Zaire villages, processing is always started by soaking unpeeled roots in a fish pond for four to five nights or in river water for six to ten days.

To make the dominant staple food, chikwangue, the soaked roots are squeezed in both hands to remove water and stored in the house for one night. The following day, the roots are pounded, fibres are removed, and the resulting mash is wrapped in leaves to form 10-20-cm packs, which can be stored after cooking for up to one week before consumption.

Cossettes, boulettes, and miettes are made as in Bandundu, except for an additional night of soaking followed by peeling.


Shortcut methods

The focus groups in the villages in Bandundu stated that shortcuts in processing were frequently used. These interviews consistently confirmed that soaking was often reduced from three nights to only one or two, and that this had been true since 1975, when cossettes were first produced in this area. Earlier, roots had always been soaked for four to five nights to make boulettes or miettes. After the construction of a new highway to Kinshasa in 1975, traders came with lorries to buy cassava. They requested the women to make cossettes from roots soaked for only two or three nights because boulettes and well-soaked cossettes would break up during transportation.

The workload and a desire to sell as much cassava as possible during periods of relatively favourable prices were the main reasons for the frequent use of still shorter soaking periods. Of the 50 women interviewed, 58% said the reason for shortcuts was to sell more cassava, 28% said it was to gain time, and 23"/o cited the traders' demand. A slight shortening of soaking time was undoubtedly initially promoted by traders; but now some traders in fact do not want to buy roots that have been soaked for only one or two nights, claiming that this is the major cause of the low quality of Bandundu cossettes.

The women interviewed individually stated that soaking time differed for roots sold to traders and those used for household consumption (table 1). However, the focus group interviews revealed that during dry seasons most families often consumed short-soaked roots intended for sale. It was obvious that the women were aware that this practice caused the acute intoxication frequently experienced within hours after eating such food. The predominant symptoms included vertigo, headache, vomiting, and diarrhoea.

TABLE 1. Normal soaking time for cassava roots for sale and for domestic consumption as reported by the women interviewed


(N = 50)

(N = 20)

























TABLE 2. Criteria reported by the women interviewed for determining sufficiency of soaking of cassava roots (percentages mentioning each criterion)

(N = 50)
(N= 20)
Some roots floating 90 65
Bubbles appearing from
50 60
Softness of roots 26 15
Minimum number of
days attained
0 96


The women in Bas-Zaire used a minimum soaking time as a major criterion to indicate when the cassava roots had been soaked sufficiently, whereas none of the women in Bandundu mentioned such a rule (table 2). All the gas-Zaire women soaked roots for four nights or more. The focus group interviews revealed that four nights was considered the absolute minimum for soaking.


Processing experiment

At 9 a.m. two women uprooted 100 roots from four-year-old plants of the most bitter cassava cultivar grown. Only one to three roots were obtained from each plant, since some had been harvested in earlier years according to local selective harvest practices. The harvested roots were peeled immediately, the upper and lower tips were cut off, and at 11 a.m. they were placed to soak in a fish pond containing stagnant water, which had a diurnal temperature variation of 22° to 27° C. The roots were removed from the water 24 hours later and placed in separate piles according to their suitability for making cossettes and miettes.

The roots intended for cossettes were sun-dried for four days. At 4 p.m. on the fourth day, the still slightly wet cossettes were divided into 15 groups. Each group was pounded and sieved (with difficulty) to give 3 dl of flour.

The smaller and softer roots, preferred for miettes, were randomly divided into two equal piles. The first pile was divided into 15 equal groups of four to five root pieces. Each group was pounded, and the pieces obtained were formed into balls that were squeezed by hand to remove liquid. Thereafter' the 2-10-mm pieces were spread on a mat for six hours to sun-dry. At sunset, the still slightly wet pieces from each group were pounded and sieved (with difficulty) to obtain 3 dl of flour.

The second pile of roots was soaked in water for a further 21 hours, removed at 8 a.m., and divided into 15 groups. The roots were by then soft enough to be squeezed by hand to remove most of the liquid before being pounded into 2-10-mm pieces. These pieces dried fully when spread on a mat in the sunshine for eight hours and were subsequently easily pounded and sieved into 3 dl of flour.

Table 3 shows that the glucoside content was high in the flour from the miettes from roots soaked for two nights, while the cyanahydrin content was high in the flour from the cossettes, which had been soaked for only one night. Surprisingly, glucoside levels were low and cyanohydrins only moderately high in the flour from the miettes from roots soaked for one night. Moisture content was high in the flour from the cossettes and the miettes from roots soaked one night but low in that from the miettes from roots soaked two nights.



Social and economic factors

This study supports earlier observations of the importance of understanding post-harvest cassava technology in Africa, from which several principal findings have emerged. Each population usually uses several different processing methods, the choice of technique depending on the variety of cassava and the end use of the product [16] Processing can render roots even from toxic cassava varieties safe for consumption [4]. Different methods are often used in neighbouring areas of the same country [17], and the shortcuts used are of greater importance to resulting dietary cyanide exposure than is the type of processing method in regular use [18]

Our findings in Zaire furthermore support the causative role for konzo of the frequent use of shortcuts in cassava processing, resulting in high cyanogen contents in flours and hence high dietary cyanide exposure [5-9]. Although few women admitted in individual interviews that their families consumed cassava soaked fewer than three nights, the qualitative methods, focus groups interviews, observations. and unstructured interviews convincingly revealed that in times of intensive sales most women in the affected villages use cassava that has been soaked for two nights or even only one night. Since the women know that this implies a risk of acute intoxication, and since the assumed association between konzo and the consumption of short-soaked roots has been communicated by health staff and local authorities. short soaking is regarded as a taboo subject, which explains why valid information is difficult to obtain. This is possibly the reason that an earlier study of konzo in Bandundu failed to identify any association with the consumption of insufficiently processed roots [19].

The reason for processing shortcuts in the konzo affected villages is intensive sale of cassava roots, which are still exclusively produced and processed by women in very poor households. The roots are soaked for a short time when women are in a hurry to obtain cash and are thus willing to make a tradeoff with respect to a risk of intoxication.

Despite intensive enquiries, we found no indication of processing shortcuts in gas-Zaire. This finding is in accordance with the higher quality and price of gas-Zaire cossettes in Kinshasa [20]. A better market integration as a consequence of better infrastructure, diversified agricultural production. and less severe poverty may explain the adherence to the traditional soaking time in this area. A contributing factor is probably also the fact that the daily cassava staple food is chikwangue, a steam-boiled product that requires well-soaked soft roots. In chikwangue-eating communities shortcuts are thus not possible without changing the normal food preparation, unlike areas where the staple is fufu, a stiff porridge made from sun-dried boulettes, cossettes, or miettes.

The participatory observations revealed that our earlier interview studies had underestimated the use of miettes in the konzo-affected villages in Bandundu [5] The reason is probably that, even traditionally, miettes were made only when households had an urgent need for food. Observation also revealed that shortcuts in processing, in addition to reduced soaking, included pounding cossettes and miettes before they were sufficiently dried. Our experience thus indicates that collecting valid information on cassavaprocessing practices in communities with food security problems requires a combination of qualitative and quantitative field research methods.

Analysis of the experimental flours

The underlying hypothesis in the processing experiments was that the length of soaking and drying would be inversely related to cyanogen levels in the flour. The risk of dietary cyanide exposure cannot, however, be evaluated from the type of processing method used, as shown by the processing experiment. Surprisingly low levels of cyanogens, close to the level considered safe by FAD/WHO (1() mg cyanide equivalents per kilogram dry weight [21] ), were found in flour from miettes soaked only one night, whereas products soaked and dried for longer periods still had high levels. These results may be explained by differences in the factors controlling the breakdown of the glucosides and cyanohydrins.

We observed that the effects of pounding and the speed of drying varied between the elastic roots obtained after one night and the softer roots obtained after two nights of soaking. This can explain the differences in cyanogen levels. The breakdown of glucosides depends on the fermentation induced disintegration of cell tissue that brings the glucosides into contact with the enzyme, but glucoside levels remain high during the first two days of soaking. Pounding the wet roots considerably speeds up the breakdown of glucosides, but, when the pounded pulp is dried, the process appears to cease almost completely at a critical moisture level. This level is rapidly passed during the drying of the thin miettes produced from the relatively soft roots obtained after two nights of soaking. The resulting flour has low moisture but high glucoside levels.

TABLE 3. Analysis of flour specimens obtained by three shortcut processing methods

Cossettes 24 60 41.0±3.1 4.54±0.15
soaked 1 night 24 5 44.1±2.5 4.69 ±0. 11
soaked 2 nights 45 8 16.6±1.2 4.96±0.13
Cossettes 11±15 95±33 9±3 115±40
soaked 1 night 1±2 19±19 5±2 26±21
soaked 2 nights 179±43 3±1 3±1 185±43

a. Mean ±SD of 15 random samples for each method.
b. Milligrams of cyanide equivalents per kilogram of dry-weight flour.

In contrast, the flour from miettes obtained from roots soaked one night had almost three times as much moisture and much lower glucoside levels. This can be explained by the combination of sufficient disruption of the cellular structure during wet pounding and sufficient time for glucoside breakdown, since the miettes dried slowly and incompletely. At the end of the day the women had difficulty pounding and sieving the flour from these miettes, and further glucoside breakdown undoubtedly occurred also during these procedures.

The low levels of glucosides in the flour from cossettes can be explained by continued cell disintegration and glucoside breakdown throughout the four days of drying, since moisture never fell to low levels in these big root pieces.

The breakdown of cyanohydrins is spontaneous and rapid at pH levels above 7 and very slow at low pH levels [22]. However, in spite of the low pH caused by lactic acid fermentation, our results indicate that the cyanohydrins appear to break down rapidly when moisture decreases below a certain critical level at which hydrogen ions can no longer stabilize the molecule. Cyanohydrin levels were thus lowest in the dry flour from the miettes from roots soaked for two nights and highest in the flour from cossettes, which had the lowest pH and, despite long drying, also had a high moisture content. It can be assumed that cyanohydrins in short-soaked cossettes are lost during storage (and transportation) when the moisture level drops [22]. This may explain why toxic effects have not been reported by urban consumers.

Finally, the level of hydrogen cyanide was low in all the flours, and this volatile substance is further lost during boiling, whereas glucosides and cyanohydrins remain in the stiff porridge.

As has been shown in animals [23], cyanohydrins yield cyanide rapidly in the alkaline environment found in the small intestine [3, 22], whereas ingested glucoside breaks down only if suitable glucosidases are present in the gut flora [24].



In conclusion, several processing methods can render cassava safe for consumption by removing the large quantities of glucosides present as well as the resulting products of hydrolysis, the cyanohydrins and hydrogen cyanide. Shortcuts in processing can result in considerable dietary cyanine exposure in an apparently unpredictable way. This may even be the case when so-called lowcyanide varieties (defined as having less than 150 mg cyanide equivalents per kilogram of dry weight when harvested [25]) are used. The majority of these cyanogens must be removed by processing to reach the level of 10 mg per kilogram considered safe for consumption [21].

Elucidation of the chemical mechanisms involved now makes it possible to estimate the risk of cyanide exposure from cassava products from a careful description of the processing methods and shortcuts used, as well as to formulate advice to prevent the toxic effects. In areas such as Bandundu, the advice in nutrition education may be as simple as "Stick to grandmother's methods." In areas where bitter cassava varieties are new to the population, however, the preventive actions needed may be more complex [7].



This study was supported by the International Programme in Chemical Science at Uppsala University in Sweden and the Ministry of Public Health of Zaire. We are also grateful for assistance from local communities, the Passionist Sisters at the Lumbi Mission, and staff members at the institutions involved.



1. FAO. Utilization of tropical foods: roots and tubers. Rome: Food and Agriculture Organization, 1989.

2. Conn EE. Cyanogenic glycosides. J Agr Food Chem 1969;17:519-26.

3. Cook RD, de la Cruz EM. An evaluation of enzymatic and autolytic assays for cyanide in cassava (Manihot esculenta, Crantz) J Sci Food Agric 1982;33:1001-09.

4. Bourdoux P, Seghers P, Mafuta M et al. Cassava products: HCN content and detoxification processes. In: Delange F, Iteke F, Ermans A, eds. Nutritional factors involved in the goitrogenic action of cassava. Ottawa: International Development Research Centre 1982:5158.

5. Tylleskär T, Banea M, Bikangi N, Fresco L, Persson L-A, Rosling H. Epidemiological evidence from Zaire for a dietary aetiology of konzo, an upper motor neuron disease. Bull WHO 1991:69:581-90.

6. Ministry of Health, Mozambique. Mantakassa: an epidemic of spastic paraparesis associated with chronic cyanide intoxication in a cassava staple area in Mozambique: 1, Epidemiology and clinical and laboratory findings in patients. Bull WHO 1984;62:477-84.

7. Essers AJA, Alsen P, Rosling H. Insufficient processing of cassava induced acute intoxications and the paralytic disease konzo in a rural area of Mozambique. J Food Ecol Nutr (in press).

8. Howlett WP, Brubaker GR, Mlingi N. Rosling H. Konzo, an epidemic upper motor neuron disease studied in Tanzania. Brain 1990; 113: 223-35.

9. Mlingi N, Kimatta S. Rosling H. Konzo, a paralytic disease observed in southern Tanzania. Trop Doc 1991 ;21 :24-25.

10. Trolli G. Paraplégie spastique épidémique. "konzo" des indigenes du Kwango. In: Trolli G, ed. Resume des observations réunies. au Kwango, au sujet de deux affections d'origine indéterminée. Brussels: Fonds Reine Elisabeth, 1938:1-36.

11. Scrimshaw SCM. Hurtado E. Rapid assessment procedures for nutrition and primary health care: anthropological approaches to improving programme effectiveness. Los Angeles. Calif, USA: UCLA Latin American Center. ]9X7.

12. Mwamba L, Goossens F, Minten B, Ntumba N. Analyse des prix des produits vivriers á Kinshasa pendant la période 19611989. Project "Commercialisation des produits agricoles". vol. 12. Kinshasa/Leuven: Direction des marches, prix et credits de campagne, Department d'Agriculture, 1989.

13. Fresco LO. Cassava in shifting cultivation: a systems approach to agricultural technology development in Africa. Amsterdam: Royal Tropical Institute, 1986.

14. Minten B. Goossens F, Ntumba N, Ndontoni B. Aspects généraux de l'agriculture dans la region du gasZaire. Project "Commercialisation des produits agricoles," vol. 15. Kinshasa/Leuven: Direction des marches, prix et credits de campagne, Department d'Agriculture, 1990.

15. O'Brien GM, Taylor AJ, Poulter NH. An improved enzymatic assay for cyanogens in fresh and processed cassava. J Sci Food Afric (in press).

16. Hahn SK. An overview of African traditional cassava processing and utilization. Outlook on Agriculture 1989;18:110-18.

17. Lancaster PA, Ingram JS, Lim MY, Coursey DG. Traditional cassava based foods: survey of processing techniques. Econ Bot 1982;36:12-45.

18. Ministry of Health, Mozambique. Mantakassa: an epidemic of spastic paraparesis associated with chronic cyanide intoxication in a cassava staple area in Mozambique: 2, Nutritional factors and hydrocyanic content of cassava products. Bull WHO 1984;62:485-92.

19. Carton H, Kayembe K, Kabeya, Odio, Billau A, Maertens K. Epidemic spastic paraparesis in Bandundu (Zaire). J Neurol Neurosurg Psych 1986;49:620-27.

20. Goossens F, Mwamba L, Castermans A, Minten B. Une comparaison des niveaux de prix des produits vivriers sur les principaux marches de Kinshasa. Project "Commercialisation des produits agricoles," vol. 5. Kinshasa/Leuven: Direction des marches, prix et credits de campagne, Department d'Agriculture, 1988.

21. Codex Alimentarius Commission. Report of the eighth session of the Codex Coordinating Committee for Africa. Cairo: FAO/WHO. 1988.

22. Fomunyam RT, Adegbola AA, Oke OL. The stability of cyanohydrins. Food Chem 1985;17:221-25.

23. Majak W, McDiarmid RE, Hall JW, Cheng K-J. Factors that determine rates of cyanogenesis in bovine ruminal fluid in vitro. J Animal Sci 1990;68: 1648-55.

24. Barrett MD, Hill DC, Alexander JC, Zitnak A. Fate of orally dosed linamarin in the rat. Can J Physiol Pharmacol 1977;55:134-36.

25. Hahn SK, Keyser J. Cassava: a basic food for Africa. Outlook on Agriculture 1985;14:95-99.

Contents - Previous - Next