Role of bacteria

No bacterial pathogen can be isolated from the stools of the majority of children with gastroenteritis who are admitted to hospital in a developed country, but in developing countries over 50% of children may have bacterial isolations.

Table 6.1 indicates the percentage bacterial isolation from the stools of children with gastroenteritis admitted to children's hospitals in Britain, Australia, Indonesia and Bangladesh.

The percentage isolation of enteropathogenic Escherichia coli is far higher in the Indonesian series than the British or Australian reports, while the percentage isolation of salmonella is higher in the Australian series than in the other reports. This may reflect the popularity of frozen poultry in Australia. Some authorities no longer report the finding of the classical strains of enteropathogenic E. coli, for example this was not done in the Bangladeshi report. Others have not studied enterotoxigenic E. coli. Thus a total comprehensive survey comparing the aetiology in developing and developed communities is lacking. The high prevalence of bacterial pathogens in the Indonesian study related to the remarkably high level of isolation of enteropathogenic E. coli. In any event the high percentage of bacterial isolation in the Indonesian series is related to the prevalence of malnutrition and poor hygiene in the community. This is well illustrated by the observations of Gracey (1973) in young malnourished aboriginals with acute diarrhoea in Australia, where there is a similar prevalence of malnutrition and poor hygiene. He found that from a group of 251 such patients, 47 excreted a serotype of enteropathogenic E. coli, 17 a species of salmonella and 25 a shigella. A similar high isolation rate of bacterial pathogens in another developing community was reported by Maiya et al. (1976), who found recognized bacterial pathogens in the stools of 66% of a group of children under 2 years of age with acute gastroenteritis in Southern India, observed overa 12-month period.

Infection with individual bacterial pathogens will be discussed in more detail later, but the mere isolation of a known bacterial pathogen from the stool of a child with acute gastroenteritis does not establish that it is the causative agent of the syndrome. Feldman, Bhat and Kamath (1970) in South India have shown a dissociation between the pattern of clinical illness and the pattern of isolation of bacterial pathogens in an impressive epidemiological survey of pre-school Indian children. However, such isolation does provide good presumptive evidence of a cause and effect relationship at the time of an outbreak of infective diarrhoea when other children are found to excrete the same pathogen, but care has to be taken when interpreting the significance of the results of stool culture taken from an individual child with acute diarrhoea and vomiting.

There are important differences in the manner in which these bacterial pathogens may produce their toxic effects. Salmonellae penetrate the mucosa relatively deeply but do not produce an enterotoxin whereas Shigella shiga produces a powerful enterotoxin which may cause ileal hypersecretion, almost to the same degree as that which occurs in cholera. It is believed that it is this penetration or invasion of the mucosa by salmonellae, and some strains of enteropathogenic E. coli and most shigellae, that permits the endotoxin they produce to exert its toxic effect.

Thus salmonellae, shigellae and some strains of E. coli may invade the intestinal epithelial cell and multiply within the mucosa. Animal studies have greatly helped our understanding of the pathogenetic mechanisms involved. Formal et al. (1976) have shown that in both salmonellosis and shigellosis bacterial invasion of the colonic mucosa does occur. This is associated with acute inflammatory reaction and mucosal damage and, in turn, there is abnormal colonic salt and water transport. In the animals studied they have also found that the jejunum is in a net secretory state despite the absence of bacterial invasion or morphological abnormalities in the jejunum. In these circumstances the diarrhoea produced by invasive organisms results from the inability of the colon to reabsorb the increased volume of fluid entering the colon from the small intestine.

Enterotoxins are synthesized within the bacterial cell body and are elaborated into broth cultures containing intact bacteria, whereas endotoxins are associated within the bacterial cell wall and so are not found in broths unless there is damage or destruction of the bacteria. Enterotoxins have classically been shown to be produced by Vibrio cholerae but also by some strains of E. coli and food poisoning strains of Staphylococcus aureus and Clostridium perfringens as well as Shigella shiga.

One of the most fascinating observations in this field in recent years has been the demonstration that cholera enterotoxin produces its hypersecretory effect via 3′-5′ adenosine monophosphate (cyclic AMP). Field (1971) and others have shown that cholera enterotoxin increases intestinal levels of cyclic AMP. It also has been shown to activate adenyl cyclase, the enzyme that converts adenosine triphosphate (ATP) into cyclic AMP. It is this activation of adenyl cyclase which accounts for the fluid hypersecretion by the gut.

Invasive enteropathogens such as salmonellae as well as enterotoxin-producing bacteria like Vibrio cholerae can produce elevated levels of adenylate cyclase activity. This could result from stimulation by prostaglandins synthesized locally. Indomethacin, a potent inhibitor of prostaglandin synthesis, has been shown in the experimental animal to abolish adenylate cyclase activation and fluid secretion induced by Salmonella typhimurium (Gianella et al., 1975). Duguid and Gillies for the first time in 1957 showed that Shigella dysenteriae, enteropathogenic E. coli and Enterobacter species could adhere to human intestinal mucosal cells and also that surface filaments called fimbriae or pili were the adhesive structure.

The attachment is via surface proteins or adhesins. The ability of bacteria to adhere to the intestinal mucosa is an essential prerequisite for colonization of the small intestine and for pathogenicity. McNeish et al. in 1975 showed for the first time the relevance of bacterial adhesion to human disease. They showed that diarrhoeagenic E. coli could adhere to human foetal intestine whereas non-entero-pathogenic strains of E. coli did not. Species specificity was shown inasmuch that E. coli which adhered to human small intestine did not adhere to brush border preparations from the small intestine of guinea-pigs, rabbits, calves and pigs. The fact that adhesion of enteropathogenic E. coli was mediated by a 60-megadalton plasmid was later established (Williams et al., 1978). A plasmid is an extra-chromosomal genetic element. Knutton et al. (1984) have developed a practical in vitro assay for adhesion. Adhesion is clearly an important early event for colonization by bacteria of the intestine.

The importance of this is eloquently shown by the example of enterotoxigenic E. coli which cause life-threatening diarrhoea when they are adhesive but are completely non-pathogenic when they are not adhesive. A host factor may also be important as has best been shown in piglets where certain litters of piglets had brush border membranes to which K88 enterotoxigenic E. coli were unable to adhere to (Sellwood et al., 1975).

So, to summarize, two mechanisms have been recognized whereby bacterial pathogens may produce the syndromes of acute gastroenteritis. These are toxin production and mucosal invasion. The particular organisms may thus be designated toxigenic or invasive. However a vital factor in pathogenicity is preliminary bacterial adhesion.

Sometimes bacteria not ordinarily regarded as pathogens may be associated with outbreaks of diarrhoea. These include Pseudomonas aeruginosa, especially in premature infants, and some strains of klebsiella. This could relate to transfer of plasmids from known pathogens to these organisms.

Bacterial pathogens, not previously known to be aetiological agents for acute gastroenteritis were identified in the 1970s. These are Yersinia enter ocolitica, Campylobacter (Vantrappen et al., 1977; Tauwers, De Boeck and Butzler, 1978) and Aeromonas (Gracey et al., 1982).

Role of viruses

As mentioned earlier it has been known for some years now in developed countries that in only a minority of children with acute gastroenteritis can bacterial pathogens be isolated from the stools. The role of viruses in the remaining children with nonbacterial gastroenteritis was uncertain until the early 1970s.

Exciting new work was reported from the United States by Blacklow and his colleagues in 1972 who studied an outbreak of winter vomiting disease among school children in Norwalk, Ohio. They found that bacteria-free stool filtrates derived from people infected during this outbreak led to acute vomiting and diarrhoea when administered to human volunteers. The infectious agent in this outbreak was then established by means of immune electron microscopy to be a virus which has come to be known as the Norwalk agent or virus (Kapikian et al., 1972).

Bishop and her colleagues (1973) from Melbourne using the electron microscope found virus particles in the epithelial cells of duodenal mucosa obtained on small intestinal biopsy from 6 out of 9 children with gastroenteritis (Figure. 6.16 ). The virus particles were found more readily in small intestinal biopsies obtained early after the onset of symptoms in these infants and less readily in those who had a longer history at the time of biopsy.

Bishop and her colleagues in 1974 went on to use electron microscopy of negatively stained faecal extracts to reveal these particles in 11 out of 14 children aged less than 3 years who had acute non-bacterial gastroenteritis. Independently, in Birmingham, Flewett, Bryden and Davies (1973) found these particles and called them rotavirus from the Latin rota, a wheel (Figure. 6.1 ). Reports of similar particles in the stools of children came rapidly from around the world. It is now clear that rotavirus infection is the most common cause of acute diarrhoea in developed communities. Whilst in developing communities it is only enterotoxigenic E. coli infections which can rival rotavirus as significant identifiable aetiological agents for acute gastroenteritis.

Population based prospective studies in North America have shown annual rates of diarrhoea for the first year of life to be 0.82-0.86 per child for an area of southern Michigan (Koopman et al., 1984) and 1.05 per child for a region of Winnipeg (Gurwith et al., 1981). Rotavirus was identified in 19% and 31% respectively of these studies.

The technique for examining negatively stained faecal extracts with the electron microscope for the presence of virus particles has now been widely applied in the investigation of diarrhoeal states in children with gastroenteritis, and a number of viruses have been recognized by their morphological appearance (Madeley et al., 1975).

At Queen Elizabeth Hospital for Children, at present as well as rotavirus, adenovirus, astrovirus, Norwalk-like agents, calici virus, small round unstructured viruses and coronavirus are all recognized by their characteristic features on electron microscopy (Table 6.2 ). While discussing the possible viral aetiology of gastroenteritis, it is important to recall that acute diarrhoea may accompany a number of viral illnesses. In developing communities, where measles is often a severe illness and a common problem, acute diarrhoea is a frequent manifestation of the malady (Axton, 1986). In developed communities, diarrhoea is usually a less important manifestation of measles but there are occasional children in whom severe diarrhoea accompanies measles.

Giles, Monif and Hood (1970) reported post-mortem findings in a child who developed severe diarrhoea with measles, followed by sudden death. The ilecolitis diagnosed on histopathological section was attributed to the measles virus.

In adults, Sheeky, Artenstein and Green (1964) have studied small intestinal mucosal morphology using biopsy in a variety of viral illnesses and they have found a significant incidence of abnormal findings in a number of viral illness including infectious hepatitis and measles.

The current pattern for identification of stool pathogens both bacterial and viral for in-patients at Queen Elizabeth Hospital is indicated in Table 6.3 .

Hypernatraemia dehydration

Hypernatraemia is defined as a concentration of sodium in the serum of 150 mmol 1⁻¹ or greater.

This type of dehydration will be discussed in more detail as it has a higher mortality than the other types of dehydration. In addition, clinically significant damage to the central nervous system occurs more commonly and there are more hazards involved in its management. Finberg in the USA in 1970 found that in his experience isonatraemic dehydration accounted for about 65% of dehydrated infants with gastroenteritis, hypernatraemia for 20-25%, and hyponatraemia for approximately 10%. However, in 1973 he found that the frequency of hypernatraemic dehydration had risen to 28–40% of the annual number of admissions of infants with gastroenteritis seen in New York.

In London, hypernatraemia was found in only 14% of children with acute gastroenteritis at the Queen Elizabeth Hospital in 1973. In fact, following publication of the Department of Health and Social Security's Report Present-day Practice in Infant Feeding in 1974, and revised subsequently, which described the hazards of high solute milks, there has been a remarkable fall in the prevalence of hypernatraemic dehydration. This change relates to the replacement of high-solute milks by low-solute milks for most infants under 6 months in Britain. The only way an infant can have high solute milk today is if he is given undiluted cows' milk. Manuel and Walker-Smith (1980) observed a fall in the number of children admitted with hypernatraemic dehydration to the gastroenteritis unit at the Queen Elizabeth Hospital, and this trend has since accelerated. This change is illustrated in Table 6.9 . It is noteworthy that there were deaths in the early years but none since 1977. Indeed the disappearance of hypernatraemic dehydration as a cause of death has been reported in Sheffield, from December 1976 (Sunderland and Emery, 1979) and in the Royal Hospital for Sick Children, Glasgow, from 1977 (Arneil and Chin, 1979). The relative importance of warnings concerning the overconcentration of feeds and the switch from high-solute to low-solute milks leading to this change is emphasized differently by the two groups. Whichever factor is more important it is clear that reduction in the solute load fed to infants in Britain has dramatically reduced the incidence of hypernatraemic dehydration and has been a major factor in further lowering infant mortality from gastroenteritis (Davies et al., 1977).

In this type of dehydration there are two physiological disturbances present, namely a loss of body water and a maldistribution of water between the major compartments of body water. This variety of dehydration is sometimes incorrectly equated with hyperosmolar dehydration, but although hypernatraemia is the most frequent cause of hyperosmolality in infants, high blood glucose levels, for example, may also be the cause, and hyperosmolar dehydration may occur in diabetes mellitus without hypernatraemia. Some of the factors predisposing to hypernatraemic dehydration are listed in Table 6.10 .

Finberg (1973) has reviewed hypernatraemic dehydration in detail and has observed that the relative preservation of the circulation and the early presence of neurological symptoms are hallmarks of the disorder. The volume of the extracellular fluid tends to be maintained to a greater degree in hypertonic dehydration when compared with isotonic dehydration and so signs of circulatory collapse appear later. Their appearance is ominous because of the degree of water deficit by then apparent. In addition, the usual skin criteria for diagnosing dehydration are not accurate, the skin having a characteristic doughy consistency. The fontanelle is typically not sunken, and in many cases may even bulge. This type of dehydration tends to occur more often in obese infants, being related to previous high solute intake. There is a risk in such infants that the signs of dehydration may be underestimated because of their obesity. The state of consciousness is usually depressed, and some infants with a severe degree of hypernatraemic dehydration may have convulsions, which may be related to brain shrinkage, a drop in CSF pressure causing haemorrhage in the central nervous system. Cerebral thromboses may also sometimes occur.

Renal tubular necrosis and hyocalcaemic tetany are less common complications but hyperglycaemia may often accompany hypernatraemia and may lead to an incorrect diagnosis of diabetes mellitus being made. Such a phenomenon is transient and tends to occur when there is coexistent severe acidosis. Insulin should not be given because lowering of blood glucose levels may enhance the risks of developing cerebral oedema during treatment, as correction of the fluid and electrolyte balance may be too rapid.

Hyponatraemia and hypokalaemia

These biochemical findings are more common in those infants who are already severely malnourished when they develop gastroenteritis. It is known that there is a predisposition for such infants to develop hypotonic dehydration. Abdominal distension and absent or markedly diminished bowel sounds strongly suggest K⁺ depletion.

It is important to realize that the aetiology of gastroenteritis and the water and electrolyte disturbances that occur have a profound geographical variation. Observations made in one country or even in one region may not be appropriate to another country or to another region within the same country. These patterns change with time.

Hypertonic dehydration used to be common in the United States, as mentioned earlier. A high incidence of hyponatraemia and hypokalaemia has been reported from many places including the West Indies, Africa and among Australian aborigines. Kingston (1973) found that more than 80% of infants in Liberia with gastroenteritis had hypotonic dehydration. He has emphasized the importance of recognition of hypokalaemia in these circumstances and its appropriate treatment. Kingston believes that continued intake of small amounts of breast milk at frequent intervals while diarrhoea continues is an important factor in the infants he described, as the sodium and potassium contents of breast milk are so low (Table 6.11 ). He considers that a different pattern of water and electrolyte disturbance is seen when gastroenteritis occurs in breast-fed infants, as distinct from formula fed children. Table 6.11 indicates the amounts of Na⁺ and K⁺ found in breast and cows' milk but the range for breast milk is very wide.

The amount of Na⁺ and K⁺ in breast milk is far less than the average stool electrolyte losses (see Table 6.13 ) and hence the net loss of Na⁺ and K⁺ is relatively in excess of the net loss of water in breast-fed infants with gastroenteritis. This is a most unusual finding in Western countries where gastroenteritis is so uncommon in breast-fed infants. These considerations have important implications in the management of gastroenteritis, which will be discussed later.

Acidosis

Various theories have been put forward to explain the aetiology of the metabolic acidosis that occurs in infants with severe gastroenteritis accompanying clinical dehydration. At present it seems that excessive intestinal losses of bicarbonate has been shown to be a major cause of their acidosis. Blood gas estimation together with the measurement of blood pH are useful investigations in a child who is severely dehydrated.

Acute renal failure

When this occurs as a complication of gastroenteritis, it may be due to oligaemia, producing reversible failure of renal function. This responds well to intravenous fluids. Much less commonly, it may be due to acute renal tubular necrosis. The prognosis of such a complication is often not good but it may respond to appropriate management.

Diet therapy

The principles of this management are to stop all solids and to replace the child's milk feeds with a glucose electrolyte solution in infants or clear fluids in older children given hourly or two-hourly for 24 hours. Usually, vomiting and diarrhoea stop under this regimen. If this does not happen, the child is often admitted to hospital for observation because the child has a severe form of gastroenteritis and may require intravenous fluids, or there is some other malady, or possibly his mother has not been giving him the oral regimen as recommended.

There is no general agreement about the precise formulation for a glucose electrolyte solution but it is clear that whatever is given the aim should be to provide a total fluid intake greater than about 20% of the child's usual fluid requirements so that the input of fluid will keep pace with the fluid loss.

Table 6.12 indicates the theoretical fluid requirements in ml kg⁻¹ at various ages throughout childhood.

Dietary management: oral rehydration therapy

A wide range of oral replacement fluids has been recommended for children with gastroenteritis. Table 6.13 shows a comparison of the electrolyte content of diarrhoeal fluid in childhood gastroenteritis.

Weil has laid down certain guiding principles in relation to the oral feeding referred to earlier, but it is apparent from Table 6.14 that very widely ranging solutions in relation to electrolyte content are used. Glucose has been recommended for many years, e.g. by Darrow and his colleagues in 1949, and evidence has now been produced to show that there are good reasons for using it. There is coupled absorption and transport for glucose and sodium in the small intestinal mucosa and the absorption of sodium and water is consequently greatly stimulated by luminal glucose. Hirschhorn and his colleagues (1973) and others have found that glucose-electrolyte mixtures given by mouth or by intubation greatly diminish the intestinal loss of fluids and electrolytes in children with cholera.

This type of therapy has been extended to infants with acute diarrhoea due to other causes, with great success. It has led to a great fall in mortality from acute diarrhoea. It is known as oral rehydration therapy or ORT. In the developing world the WHO or UNICEF solution (WHO-ORS) is most widely used. ORT is discussed in more detail in Appendix II. Its sodium and potassium composition is indicated in Table 6.14 together with other solutions.

The sodium content of this solution is within the range of the sodium content found in diarrhoeal fluid but it is higher in electrolyte content than the other solutions. Unless large volumes are taken or free water as well there is a risk of hypernatraemia with this solution, except in communities where hyponatraemia is common, e.g. communities where there is a high incidence of protein-calorie malnutrition. Also, if diarrhoea ceases immediately, too much sodium can be given, especially in small infants. However, in view of the declining importance of hypernatraemic dehydration in developed countries, owing to widespread use of low solute milks and better education, this risk may not now be so great.

Before the advent of the concept of oral rehydration therapy, oral half-strength Darrow's solution was widely used in developing countries and continues to be used successfully in some countries. Electrosol tablets when dissolved produce an adequate electrolyte solution for oral use, but as tablets they do carry the risk of being given in excess dosage. Half-strength saline contains no potassium and significant potassium loss does occur in gastroenteritis, but in emergencies this solution flavoured with a little orange juice may be very useful until other solutions can be obtained. There are risks of giving soft drinks in a casual manner as their electrolyte content and osmolality varies greatly (Table 6.15 ). A glucose-electrolyte mixture (GEM) was used at Queen Elizabeth Hospital for children from 1952 to 1980. Its composition was based on one devised by the Medical Research Council in 1952. It was similar to the first solution used by Darrow.

Its composition is indicated in Table 6.14. Its use was stopped for economic reasons in 1980 when a commercially available solution Dioralyte (Armour Pharmaceuticals) was introduced. The glucose electrolyte solution (GEM) used at Queen Elizabeth Hospital contained some calcium and magnesium because it had been noted that some infants with acute diarrhoea had low plasma levels of these electrolytes. In relation to more recent formulations it is interesting to observe the relatively high glucose composition of 50 g glucose/litre, the low sodium level and the use of citrate. This solution was used very successfully for almost 30 years.

However, it was not used for children who were 5% dehydrated or more who were given intravenous fluids. This mixture was a ready-made solution and once opened it had a short ‘shelf life’ because of the risk of bacterial contamination. Hence inpatients only were given this solution at Queen Elizabeth Hospital whereas outpatients were given an oral sucrose-electrolyte solution known as out-patient electrolyte solution (OPEM). The basic electrolyte solution was supplied in a concentrated form and diluted five times with water at home. Then sucrose was added at home. It was thus hoped to avoid bacterial contamination.

In view of the theoretical advantages of glucose in cholera toxin induced diarrhoea and the experience of its use in diarrhoeal states not due to cholera, by Hirschorn and colleagues (1973), as well as the theoretical drawback of using a surcrose solution because of the documented low sucrase levels found in the small intestinal mucosa in acute gastroenteritis (Barnes and Townley, 1973), on two occasions a clinical comparison of the use of glucose and sucrose additions to a basic electrolyte mixture in the outpatient management of acute gastroenteritis in children was undertaken (Rahilly et al., 1976; Hutchins et al., 1979).

In this latter study it was found that when the osmolality of the solutions as made up by the mothers was assessed it was remarkably variable. This could be detrimental leading to osmotic diarrhoea if hyperosmolar solutions were used. In fact Ironside (1973) had cautioned against the unrestricted use of glucose in home-made solutions because of the risk of making up a hypertonic solution and so aggravating the clinical situation. To investigate this further a study was undertaken of the variations in osmolality of samples of oral solutions as provided by parents (Hutchins et al., 1980) (Figure. 6.13 ). The following solutions were compared, first parents diluted a concentrated outpatient electrolyte mixture (glucose OPEM) and then added glucose or sucrose (sucrose OPEM), next parents added sucrose at home to an already diluted electrolyte mixture, then this solution as prepared by pharmacy staff was studied and finally a sachet of Dioralyte (based upon sodium chloride and dextrose compound powder BPC) to which water was added at home was studied.

All mixtures except those prepared by the pharmaceutical staff were often made up with marked inaccuracy. It was only with solutions made up with glucose that dangerously high levels of osmolality were reached. For sucrose even when inaccurately made up the levels of osmolality barely exceeded the correct value for the glucose solutions.

These two studies at the Queen Elizabeth Hospital were carried out on outpatients attending casualty during the winter months, 2 years apart. Both showed that, in those given GEM (27% and 32% respectively in the two studies) more children experienced failure with out-patient management and required admission compared with those given OPEM (10% and 18%). As social reasons often contributed to failure in outpatient management, these studies can only be cited as evidence that a sucrose electrolyte solution appeared to be at least as good as glucose in the outpatient management of acute gastroenteritis in a developed community.

The second study (Hutchins et al., 1978) did, however, make it clear that it was easier to make an error when the solution was made up with glucose and could lead to unacceptably high levels of osmolality in some instances. Such high levels could contribute to outpatient management failure. The author commends the use of a sucrose electrolyte solution with its lower osmolality, ready availability, and economic advantages, both in developed and also in developing countries, from which encouraging results have been reported, from Indonesia (Moenginah et al., 1975) and also from Calcutta (Chatterjee et al., 1977). A plea was made for the use of sucrose on the basis of these observations but commercial formulations have largely continued to use glucose.

One other issue the study of Hutchins et al. addressed was the concentration of sodium. However as all solutions were low in sodium content none were associated with dangerously high levels. When high content sodium sachets, such as the WHO formulation are used, this is a risk (Cutting, 1979).

Since 1980 commercial formulations have been used at Queen Elizabeth Hospital. Initially a sachet preparation (Dioralyte) was used. This has the advantages of compactness and convenience over an already made up solution but as seen in Figure. 6.13 it does carry the occasional risk of being made up inaccurately and so producing a dangerously high osmolality. Therefore a study was made of a ready-to-use glucose electrolyte solution (Dextrolyte, Cow & Gate) in the outpatient management of infants with acute gastroenteritis (Trounce et al., 1985). This showed this solution was readily accepted by parents and children. Twenty-two samples of the mixture as actually given to the baby were brought for analysis. The osmolality ranged from 291 to 308 mosm l⁻¹ with a mean of 298 mosm l⁻¹. Thus in no case was there a dangerously high level. A disadvantage of ready to feed solutions is the risk of bacterial contamination once the bottle is opened. At present such solutions are presented in 100 ml bottles and parents are advised to discard the unused portion. If such solutions are refrigerated even when the bottle is opened and used again the risk of bacterial contamination is minimal (Santosham et al., 1982). These workers showed that if the solution is stored at 4°C after the introduction of 1000 colonies of E. coli it will remain sterile 24 hours later, whereas if it is stored at room temperature the E. coli will, not surprisingly, grow freely.

Experimental animal work from Hamilton et al. (1976) provides a theoretical basis for the view that there is no specific benefit in using a glucose electrolyte solution when there is viral gastroenteritis. Using a transmissable gastroenteritis virus (TGE) to study gastroenteritis in pigs, they have been able to study glucose-stimulated sodium transport which in fact falls to its lowest level at the peak of the diarrhoea. Thus it seems probable that in viral gastroenteritis, carbohydrate electrolyte solutions merely provide calories and are unlikely to have a specific effect on sodium absorption. These studies do show the virtue of a lower glucose level as currently in WHO-ORS solution (see Appendix II).

Dietary management: bottle-fed infants

In the United Kingdom for many years it has been the practice after 24 hours of a glucose electrolyte mixture (GEM) to gradually reintroduce the infant's usual feeding formula or cows' milk back into his diet by a system known as regrading. The speed of this process of regrading will depend upon the individual infant's circumstances; in particular, the severity of his vomiting and diarrhoea during the acute episode and the persistence or otherwise of such symptoms. Regrading is usually done in quarter-strength increments, i.e. starting with quarter-strength dilution of the milk feeding with the glucose electrolyte mixture moving up to full strength in quarters (Table 6.16 ) at intervals of 24 hours. This is a slow regrade. When the increments occur at 6 or 12 hours this is known as a fast regrade. The rationale for this conventional approach is based upon the following points:

• •

  The perceived need to provide a short period of relative bowel rest by giving dilute feed of increasing strength following the acute onset at a time when the small intestinal mucosa is known to be damaged (at least in the case of viral diarrhoea, see page 223) and is then recovering.

• •

  To prevent a significant return of symptoms, i.e. relapse, by rapidly returning to a lactose containing feed which might result in osmotic diarrhoea in view of the state of temporary lactase deficiency related to the small intestinal mucosal damage present at this time or due to malabsorption of some other nutrient by the damaged mucosa.

• •

  To prevent food (i.e. cows' milk) sensitization by reducing antigen load at a time when the small intestinal mucosa is abnormal and there is evidence of increased antigen uptake (Gruskay and Cooke, 1955) and increased permeability to sugar (Ford et al., 1985). Both factors predispose to increased antigen entry to the mucosa and so to sensitization.

Once the infant is on full-strength formula again, solids may be reintroduced to his diet. It is important for the mother and indeed for the doctor to remember that some looseness of stools may persist for several days after gastroenteritis and that it may take time for the stools to return to normal.

This established approach has been challenged from two directions. First, by studies which question the necessity for such regrading; and second, by those who consider that food intolerance is so common a sequel to acute gastroenteritis that a lactose-free hypo-allergenic feed should be given at this time.

Both these challenges need to be considered. However, it must be made clear that the concept of regrading has arisen in developed communities. Here nutritional reserves are generally sufficient to offset the detrimental effects of actue episodes of diarrhoea as the children are usually previously well nourished. Furthermore the type of infant feeding has changed in recent years to adapted formulae that are less sensitizing than doorstep milk.

First considering the need for regrading, as long ago as 1948 Chung performed detailed studies on food absorption in infants with acute gastroenteritis. He showed that duration of diarrhoea was not prolonged in infants continuing to receive full strength feeds compared with those placed on a starvation regime. However his views were not generally accepted in practice. More recently, others have challenged the need to regrade (Rees et al., 1979; Dugdale et al., 1982; McDowell and Evans Jones, 1985; Haque et al., 1983). There have been warnings from the developing world that dilute feeds may lead to malnutrition but, as stated earlier, only developed countries are being discussed here.

Placzek and Walker-Smith (1984) at Queen Elizabeth Hospital studied this problem in 48 children less than 18 months of age admitted with acute gastroenteritis (see Table 6.20 for clinical features). Complicated progress, i.e. relapse, occurred in 7 (30%) where no regrade was used and only 1 (4%) where a regrade was used. Relapse was most often due to lactose intolerance but was due to severe vomiting in two. However, the complications entirely occurred in children under the age of 9 months. This risk of relapse was confined to children under the age of 9 months. Thus it was considered that in children older than 9 months gradual reintroduction of milk was not necessary. However, for those aged under 9 months there was a significantly greater risk of relapse due mainly to temporary lactose intolerance associated with rotavirus infection. A similar result was obtained in Chester (McDowell and Evans-Jones, 1985) although they interpreted their results differently (Walker-Smith, 1986).

The numbers in this study were still relatively small and only concerned inpatients. Therefore another study was undertaken at Queen Elizabeth Hospital of 68 babies admitted with acute gastroenteritis under 6 months of age (Armistead et al., 1987); 39 were admitted to Hospital and 29 attended a daily outpatient clinic. Three groups were studied: a control group (I) regraded in the conventional way; group II, no regrades; group III, regraded onto the hypoallergenic whey hydrolysate formula Alfaré (Table 6.17 ). There was no significant difference in relapses between the three groups.

However, no child developed lactose intolerance. Relapses were due to a return of diarrhoea without excess stool reducing substances. This study differed from the earlier study as there were no cases of previous failure to thrive and there were fewer severely dehydrated.

In communities where malnutrition is common it is important that there should not be any prolonged reduction of caloric intake following gastroenteritis and it is vital that a starvation diet for a long interval should be avoided. Except during acute water and electrolyte disturbance the diet of malnourished children should not be unduly restricted even though increased feedings may accentuate diarrhoea. Provided this does not lead to dehydration and further electrolyte disturbances in these children, it is important to continue with an oral feed containing an adequate amount of calories. Such malnourished infants tolerate further caloric restriction very poorly and diarrhoea may even become worse with the untoward consequences. This is a quite different situation from that of the previously well-nourished child who develops gastroenteritis.

Intravenous therapy

Intravenous therapy for dehydration in man was first used by a Scottish general practitioner, Dr Thomas Latta, in 1832, with some success, but it was a long time before this treatment became accepted. Gamble (1953) reviewed the subsequent history of this technique, and it was indeed notably Gamble, and also Darrow and colleagues, who in the 1940s and 1950s put safe intravenous therapy in childhood on the therapeutic map. The subject has been well reviewed by Weil (1969) and Finberg (1970).

Fortunately, in the developed world, despite the continuing importance of acute gastroenteritis, the numbers of children with severe dehydration requiring intravenous fluids is falling. In 1973, at the Queen Elizabeth Hospital, only 60 from the 353 children admitted with acute gastroenteritis required intravenous fluids i.e. 16% (Gribbin et al., 1976). The most recently published figure of 1.4% from the UK (Cutting et al., 1986) is even lower. Nevertheless, the situation still may be urgent. In the developing world severely dehydrated children requiring intravenous fluids continue to be a common problem.

There are three important phases in intravenous therapy; namely resuscitation, rehydration and maintenance. These will be outlined first of all for children with isotonic dehydration.

Investigations

Serum electrolytes, blood urea and total serum protein levels should be estimated as a matter of urgency in all children who are shocked, and within 24 hours in all children who are given intravenous fluid, for evidence of any electrolyte level disturbances, especially hypernatraemia, and as a confirmation of the state of dehydration. The levels of blood urea and total protein are sensitive indices of dehydration. Other useful investigations include haemoglobin and blood gas estimation when signs of acidosis are present.

Management of complications of fluid, electrolyte and acid-base balance

Viral gastroenteritis syndromes

The clinical associations described here are based largely upon experience of gastroenteritis at Queen Elizabeth Hospital since 1973. The hospital is situated in an area of socioeconomic deprivation with a large number of children of immigrants and a high incidence of gastroenteritis, with between 400 and 500 admitted most years and many more children seen as outpatients. The clinical associations which are described concern observations made on in-patients. However, Weindling, Walker-Smith and Bird (1980), reported the results of an investigation of 58 infants with gastroenteritis treated as outpatients at Queen Elizabeth Hospital and 30 controls with diarrhoea. The study was conducted in the autumn of 1977.

Rotavirus was found in 7% compared with 12% in-patients at the same time. Adenovirus was found in 3.5% of outpatients and 4% of in-patients. Neither virus was identified in controls.

Rotavirus

Bishop and her cplleagues (1973) used electron microscopy to demonstrate for the first time the presence of rotavirus particles in the enterocytes of six small intestinal biopsies from children with acute gastroenteritis.

Since then, rotavirus gastroenteritis has come to be recognized as the most important cause of acute diarrhoea in children between 6 months and 2 years of age in industrialized countries. In developing communities only enterotoxigenic E. coli rivals rotavirus in importance in most reports, although studies from such communities do vary in their assessment of the relative importance of rotavirus in the causation of acute diarrhoea. Diagnosis is usually based on electron microscopy of negatively stained stools as first reported by Flewett et al. (1973) or by enzyme-linked immunosorbent assay (ELISA) analysis of stools. This latter technique is now widely used diagnostically.

Pathophysiology

Mavromichalis et al. (1977) found rotavirus in intestinal aspirate of 6 out of 8 children with gastroenteritis who had rotavirus in their stools. All six had abnormal xylose absorption, whereas the two without rotavirus in the lumen had normal xylose tests suggesting that small intestinal dysfunction, presumably due to mucosal damage, occurred when the virus was present in the lumen.

To date, little is known in man of the mechanism of diarrhoea in rotavirus gastroenteritis but Hamilton et al. (1976) have studied an animal model using a virus in piglets. This is a corona virus, known as the transmissible gastroenteritis virus (TGE). It produces massive diarrhoea some 16–40 hours after infection, with high faecal electrolyte levels. At 40 hours they found decreased sodium and water flux, decreased mucosal activities of disaccharidases and sodium, potassium-ATPase, but normal adenyl cyclase activity. Unlike enterotoxigenic diarrhoeas such a cholera, under these experimental conditions, sodium flux failed to respond to glucose. Ferguson and Snodgrass (1978) have gone on to study infection of 1-day-old germ-free lambs with rotavirus. They found a state of accelerated epithelial cell turnover with functionally immature enterocytes clothing the villi. If a similar situation exists in rotavirus gastroenteritis in man, it is clear that the pathophysiology may be very different from the toxigenic bacterial diarrhoeas.

Age range

Rotavirus has been found principally in the stools of children with gastroenteritis from the neonatal period up to 5 years but most are under 2 years of age. Most studies have reported a peak between 6 months and a year (Kapikian et al., 1976; Rodriquez et al., 1977), but Shepherd et al. (1975) found 43.3% of their children under 6 months.

Although the disease is commonest in young children it may also occur in adults. Rotavirus has, indeed, been found in the stools of adults with acute gastroenteritis (Von Bonsdorff et al., 1976). The virus may also occur in adults who are symptom free (Tallet et al., 1977). It is often found in the families of infected children, i.e. it is a contagious virus.

Sex ratio

As in earlier reports of non-bacterial gastroenteritis in which viral studies were not done (Gribbin, Walker-Smith and Wood, 1976; Tripp, Wilmers and Wharton, 1977), most studies of rotavirus gastroenteritis report that more boys are affected than girls particularly in younger children (Table 6.7). The reason for this male predominance is not yet clear.

Incubation period

The incubation period appears to be about 48–72 hours (Shepherd et al., 1975).

Clinical features

Most studies have described a sudden onset with vomiting as the first symptom, often accompanied by diarrhoea but sometimes preceding it by several hours (Shepherd et al., 1975). Diarrhoea is typically acute in onset and often watery in character with usually less than ten stools each day. However on occasion, there may even be no diarrhoea, with vomiting as the only symptom.

Initial clinical reports of rotavirus gastroenteritis described a relatively mild-self-limiting illness with dehydration, usually less than 5% present in about half or less of those children admitted to hospital.

Pyrexia is a common but not a constant feature. Although Champsaur et al. (1984) describe a significant association between fever, diarrhoea, vomiting and the presence of rotavirus in the stools. A concurrent upper respiratory tract infection has been documented in 42% in one report (Carr et al., 1976) and 29% in another (Rodriquez et al., 1977). Most clinical reports from developed countries have described a mild self-limiting illness with dehydration, usually less than 5%, present in about half or less of those admitted to hospital. Rodriquez et al. (1977) described dehydration in 83% of children with rotavirus gastroenteritis compared with 40% in children who did not have rotavirus in their stools. But dehydration was mostly mild to moderate. It has, however, become clear that rotavirus gastroenteritis may on occasion be associated with a much more severe illness, and death has been reported from Canada (Carlson et al., 1978) and Australia (Bishop et al., 1973). Fatal rotavirus gastroenteritis does not appear to have been documented yet in Britain although a child brought dead to Queen Elizabeth Hospital following a diarrhoeal illness had rotavirus found in his gut lumen at post-mortem. Nevertheless, it now is clear that rotavirus may be a major cause of diarrhoea-related infantile death (Carlson et al., 1978). Vesikari et al. (1983) have indeed estimated that one million rotavirus associated deaths may occur annually.

An initial study of 30 children with rotavirus gastroenteritis from the Queen Elizabeth Hospital for Children, London during the winter of 1974–1975, found a relatively mild illness (Shepherd et al., 1975). Ryder et al. in Bangladesh (1976) have indeed found rotavirus in the stools of 12 out of 22 children, with severe dehydration due to diarrhoea, the degree of dehydration being comparable with that found in cholera, in their experience.

In southern India, Maiya et al. (1976) found no difference in the clinical presentation of children with gastroenteritis in whom rotavirus was detected as compared with those in whom a bacterial pathogen was identified or no agent recognized. In rural Bangladesh when rotavirus infections were compared with most gastrointestinal infections due to other aetiological agents, rotavirus was more often found to be associated with dehydration and hospital admission. Only cholera led more often to dehydration (Black et al., 1980).

So although rotavirus gastroenteritis seems most often to cause a relatively mild illness in developed communities, it may on occasion cause more severe disease and even death. Rotavirus is a significant cause of death in the developing world.

Accompanying pathogens

Any account of the severity of the clinical disease must try to take note of the possible role of accompanying pathogens, bacterial or viral. Several reports have described a more serious illness, when bacterial pathogens, especially enteropathogenic E. coli, are found in the stools at the same time (Shepherd et al., 1975; Carr et al., 1976). Madeley et al. (1975) have found rotavirus and other bacterial and viral pathogens occurring in different stool samples from the same infants and have found it difficult to determine whether one or all were significant pathogens.

Duration of illness

The duration of illness ranges between 5 days to 3 weeks (Shepherd et al., 1975) but usually lasts only 8 days (Tallett et al., 1977). Most children present within 5 days of onset if they come to hospital.

Biochemistry and haematology

An elevated blood urea on hospital admission has been described (Rodriquez et al., 1977). Hypernatraemia may occur but is not characteristic. When there is dehydration it is likely to be isotonic. Lymphocytosis may be found and also neutropenia (Phillips, A.D., unpublished observations).

Stool microscopy

Characteristically, leucocytes are not present in the stools although they may be seen in some patients, 18% in one series (Rodriquez et al., 1977).

Infectivity

It appears to be a highly contagious disease, apparently spread by the faecal-oral route. Massive quantities of virus may be passed in the stools some 24–72 hours after infection. Outbreaks may occur in families and in institutions. In one study, 35% of parents of children with rotavirus gastroenteritis had serological or stool evidence of rotavirus infection (Rodriquez et al., 1977). It is also clear that adult staff members may play a role in transmitting rotavirus infections in children's wards.

Rotavirus in neonates

A high incidence of rotavirus in the stools of asymptomatic neonates has been reported from Sydney (Murphy, Alberry and Crewe, 1977) and St. Thomas's Hospital, London (Chrystie et al., 1975). In one study, Murphy et al. found no less than 50% of neonates by the age of 3–4 days were excreting virus. Only 24% of these had diarrhoea.

Despite the finding of rotavirus in asymptomatic neonates, Murphy et al. found that significantly more neonates with diarrhoea excreted the virus than in the symptom-free group. It does, in fact, seem clear that rotavirus can be a cause of gastroenteritis in the neonate, as suggested by Bishop et al. (1976). Quite severe symptoms have been observed in premature babies in some special care baby units (Bryden et al., 1982). The role of rotavirus antibodies in protecting the infants against rotavirus infection is uncertain. A clue to why some with stool rotavirus do not get clinical disease is provided by Murphy et al., who found that children who had in their stools viruses coated with an ‘antibody-like’ material did not get diarrhoea. It is known from the work of Matthews et al. (1976) that breast milk and cows' milk contain non-antibody virus inhibitor and this may play a role too. So neonates appear to be relatively immune to serious rotavirus infection. Rotavirus has now been described to occur endemically among neonates in hospital nurseries in a number of countries as well as England and Australia, namely the United States and Sweden (Grillner et al., 1985). In these studies rotavirus infection occurred throughout the year without any obvious seasonal variation. Another feature is the difficulty in eradicating the virus from nurseries. This may relate to the role of the staff.

Hjelt et al. (1985) have looked at rotavirus infection in the nursing staff in paediatric wards; 36% of staff in the infant/toddler wards had rotavirus. All members of the staff had rotavirus IgA or IgG antibody. As it is known that rotavirus is present in the hand-washings of children with rotavirus (Samadi et al., 1983), this study shows that nursing staff may contribute to the spread of rotavirus gastroenteritis in paediatric wards via contaminated hands.

Spread of rotavirus

A community-based study of the spread of rotavirus in 47 families in New Zealand (Grimwood et al., 1983) showed that the primary mechanism in the spread of rotavirus infection is through intrafamilial contacts. Thirty-six out of 78 adults and children exposed to children suffering from rotavirus infection developed the disease in the following 4 weeks. In contrast, none of the 53 adults and children exposed to children who had non-rotavirus gastroenteritis developed diarrhoea.

The spread of rotavirus infection has been further investigated in a prospective study using the ELISA technique in children admitted to hospital in Holland (Walther et al., 1983). The children were examined weekly for rotavirus. There was a high incidence of hospital acquired infection and even asymptomatic rotavirus excretors were identified. In another study in Paris (Champsaur et al., 1984) using stool electron microscopy and ELISA, asymptomatic excretion of rotavirus was often found to occur in children of 1–6 months of age (50%) and in those of 7–24 months (26%). The authors found only a statistical association with the diarrhoea-fever-vomiting syndrome. They subdivided rotavirus excretors into:

• •

  Rotavirus gastroenteritis.

• •

  Asymptomatic infection, i.e. rotavirus with serological response but no symptoms.

• •

  Rotavirus carriers.

All these studies pose problems for the relevance of rotavirus found in the stool to the aetiology of diarrhoea in an individual child. This probably relates to the stability of the rotavirus in the environment. It is probably spread as an aerosol to the environment during the change of nappies. Thus the main reservoir of infection is the babies themselves. In the study of Grillner et al. (1985) one single electropherotype belonging to subgroup 1 of the human rotavirus was found throughout. Hjelt et al. (1985) have studied 25 healthy Danish mothers and their breast-fed infants. All mothers had detectable concentrations of rotavirus IgA and IgG in their serum at time of lactation. These levels did not change during the period of breast feeding.

All infants had rotavirus IgG in serum by 3–4 days but none had serum rotavirus IgA or ScIg. Rotavirus IgG in the serum of infants correlated with that of their mothers. Few samples of duodenal juice from the infants contained rotavirus ScIg and IgA. On the other hand, about 80% of the infants stools samples contained rotavirus ScIg and IgA. These disappeared from the infants stool after cessation of breast feeding. So it may be inferred that infants receive rotavirus IgG through the placenta and receive rotavirus ScIg and IgA in constant amounts via breast milk during lactation. Thus any protectional effect of breast feeding requires frequent breast feeding and is limited to the period of lactation.

Diagnostic techniques

A number of techniques have been used. First and foremost has been electron microscopy of stools using negative staining, which has been used in many centres. This is a relatively simple technique provided an electron microscope is available. Second is the technique of counterimmunoelectrophoresis. This is a technique to detect viral antigen in the stool (Spence, Farrel and Bourchard, 1975; Middleton, Petric and Hewitt, 1976). It appears to be about as sensitive as the electron microscope technique. Third is serum serology using a complement fixation technique. This was developed by Kapikian, Cline and Mebus (1975). With rotavirus stool antigen, these workers showed a rise in serum antibody titre with a complement fixation test using acute and convalescent sera. A related antigen, Nebraska calf rotavirus, is more readily available. Fourth is an indirect immunofluorescent antibody technique (Davidson et al., 1975b). This is a difficult technique but one that seems to be slightly more sensitive than the complement fixation test. Fifth is the ELISA technique.

Several studies have recently looked at the comparative efficiency of various techniques for the detection of rotavirus antigens in the stools. Miotti, Eiden and Yolken (1985) compared two enzyme immunoassays (Bio Enza Bead and Rotazyme) with a latex agglutination assay (Rotalex) using solid-phase enzyme immunoassay as the reference. No false positive assay was found. All three commercial assays accurately detected rotavirus in the stools from children with rotavirus gastroenteritis but there was a decreased level of efficiency later in the course of the illness.

Another study (Julkunen et al., 1985) found latex agglutination test to be the recommended test for stool screening for rotavirus infection. Chernesky et al. (1985) have modified Rotazyme to produce Rotazyme 11. This technique achieved 99.4% sensitivity compared with electron microscopy. This was regarded as a significant advance.

Another study (Knisley et al., 1986) compared Rotazyme 11 and other polyclonal antibody-based assay systems with a monoclonal assay system which they found to be more sensitive than other assay systems. Yet another paper (Pai, Shahrabadi and Ince, 1985) studied Rotalex and compared it with direct electron microscopy and Rotazyme 1. They found it to be much faster and cheaper than electron microscopy or Rotazyme. It was claimed to be available even in the doctor's office for rapid diagnosis. It is not possible at present to adjudicate on all these techniques for field studies but what is clear is that for hospital studies the electron microscope is the technique of choice as it not only accurately shows the presence of rotavirus but accurately diagnoses other virus particles too.

Delayed recovery

In the first detailed report of the clinical features of rotavirus gastroenteritis, Shepherd et al. (1975) described temporary sugar intolerance in 2 of 30 children. In a prospective study of sugar intolerance at Queen Elizabeth Hospital of 200 children with acute gastroenteritis (Trounce and Walker-Smith, 1985) 45 children had rotavirus gastroenteritis. 15 (45%) developed sugar intolerance, 9 (20%) were monosaccharide intolerant and 6 (15%) were disaccharide intolerant. In this relatively small study there was only a statistical association between rotavirus infection and sugar intolerance. This was not found with any other virus or bacterial pathogen although the numbers were small. The glucose electrolyte solution used at the time of the study was much higher than that presently used so its importance may have been higher at that time than today. An association between sugar malabsorption and rotavirus would in fact be expected in view of the damage to microvilli in infected enterocytes (Bishop et al., 1973). In addition, it would appear theoretically possible that severe and extensive infection could temporarily damage a high proportion of lactase activity of the enterocytes and, perhaps, also cause a transient surface block to sugar absorption by temporarily damaging the brush border membrane which would cause a block in absorption to all sugars, as proposed by Walker-Smith, Kamath and Stobo (1972). The clinical severity would depend on the number of enterocytes damaged. Finally, Holmes et al. (1976) have suggested that lactase is the receptor and uncoating enzyme for rotavirus. This would mean that infants with their high lactase levels may be more vulnerable, helping to explain the age of most affected individuals.

Immunity

Little is known yet about immunity to rotavirus infection. Repeat infection with rotavirus in childhood has rarely been reported (Fonteyne, Zusis and Lambert, 1978; Simhon et al., 1981). In the former report a different type was responsible for the second infection but not in the latter. Reinfection was also reported in the Santa Maria Caque study. Antibody levels tend to rise with age, reaching adult levels at two years (Kapikian et al., 1976). Children with pre-existing antibody have become infected, however. Infection with one subgroup may not protect against another. Infants aged 6-24 months appear more susceptible to infection regardless of the presence of antibody in their sera. Lower prevalence rates of rotavirus infection under 6 months may be attributable to immunity acquired from the mother. Bortulussi, Szymanski and Hamilton (1974) found normal amounts of IgA in intestinal secretions during acute illness and in convalescence.

Protection against rotavirus

It is clear that there is an urgent need to provide protection against rotavirus. Vesikari et al. (1983) have tested an oral live rotavirus vaccine RIT 4237 of bovine origin for immunogenicity and safety in adults and subsequently in children. They also showed in a randomized double-blind placebo controlled trial, the ability of this vaccine to protect against natural rotavirus infection in children. A single dose of RIT 4237 was given to 178 infants aged 8–11 months (Vesikari et al., 1984). During 5 months observation after vaccination two of the 86 vaccine recipients and 18 of the 92 placebo recipients had rotavirus diarrhoea lasting more than 24 hours which was clearly a significant result. No side-effects attributable to the vaccine were observed. The two children who were not protected were primary vaccine failures as they had no detectable serum antibody responses after vaccination. It remains to be seen whether this vaccine will provide protection against all subgroups of human rotavirus. The epidemic in Finland, at the time when protection was found, was caused by subgroup 2 rotaviruses.

An effective vaccine for rotavirus should have an important effect upon the very high mortality rates for children from gastroenteritis in the developed world (De Zoysa and Feachem, 1985). As already pointed out, bovine (RIT 4237) vaccine has been shown to be safe and effective against rotavirus gastroenteritis in Finland (Vesikari et al., 1985).

However, it has had a disappointing lack of effectiveness in developing communities (De Mai et al., 1986), e.g. in Rwanda. This failure could be due to over-attenuation of the virus and interference by other enteric viruses affecting children in developing communities where multiple viral infection is common. Another rotavirus vaccine derived from rhesus monkey rotavirus RMV (MMU-18006) has been studied in Venezuela as it is more immunogenic (Flores et al., 1987). The protective efficiency of RMV vaccine against rotavirus diarrhoea was 68% of 247 infants aged 1–10 months studied for up to 1 year. More interestingly, for the entire study group, vaccine efficiency was 100% against the most severe rotavirus diarrhoeal episodes. Most rotavirus strains detected in the study belonged to rotavirus subgroup 2 which includes serotypes 1, 3 and 4. Protection afforded was limited to subgroup 2 viruses.

This study represents a most encouraging result and clearly further extensive field studies are indicated.

Rotavirus-like agents

A number of rotavirus strains which are antigenically distinct from rotavirus causing human gastroenteritis have been recognized. These are called rotavirus-like agents (RVLA). They have been isolated from a wide range of animals and also man. These non-group A rotaviruses have been tested for in infants and adults with gastroenteritis by oral inoculation of infant rats with stool specimens from 16 patients. Six faecal specimens, but no other disease controls, gave rise to infection in the rats (Eiden, Vonderfecht and Yolken, 1986). All these rats had antigenic evidence of infection as determined by immunofluorescent staining of their gastrointestinal epithelium.

Vonderfecht et al. (1985) studied an enzyme immunoassay inhibition assay for the detection of rat rotavirus-like agent in intestinal and faecal specimens from both rats and humans with diarrhoea. Four of six human faecal specimens that induced diarrhoea in suckling rats were positive for this assay system.

This rotavirus-like agent has a similar RNA migration pattern to a rotavirus-like agent described in adults in China. This virus, now known as ADRV, has been further studied by counterimmunoelectrophoresis (CIE) as a method for detecting antibodies to this virus from healthy adult subjects in China, Hong Kong and Australia. No outbreaks of diarrhoea due to this virus have been yet reported from Hong Kong or Australia yet 18% and 15% respectively were found in two outbreaks in Junzhou, China where this virus was first identified (Tao et al., 1985). This antibody was also found in 47% house rats, 36% pigs and 17% wistar rats. These observations thus raise the possibility of animal origin for this virus infection in man. Flewett (1986) has pointed out that a good study to determine on the one hand whether animal rotaviruses do infect young children or on the other whether human rotaviruses infect young animals is badly needed.

Clinical features

Clinical features

Non-cultivatable or enteric adenoviruses have been associated with outbreaks of diarrhoea and vomiting in children (Richmond et al., 1979) more so than adenoviruses which can be grown in culture. There has been a long-standing uncertainty about an aetiological role for adenoviruses in gastroenteritis. There are reports, however, of death related to adenovirus where it was found in the mucosa of a child who died from gastroenteritis (Whitelaw et al., 1977). In addition adenovirus has been detected by electron microscopy in the mucosa of a child who suffered a sudden unexpected death (Phillips, 1982).

Further evidence that adenovirus may cause infantile gastroenteritis (Figure. 6.19 ) comes from a report that an outbreak of gastroenteritis was due to a newly identified subgroup of adenovirus designated as F and closely related to type 40 adenovirus (Chiba et al., 1983).

Clinical features

Coronavirus was sought in the stools of all children admitted to Queen Elizabeth Hospital with acute gastroenteritis between 1st January 1980 to 1st August 1983; 18 . of 2387 stool specimens (0.6%) had the coronavirus in 15 children by electron microscopy. Coronavirus was detected in stools more frequently in the spring and summer months, 12 out of 15 cases occurring between April and August. In six patients, other enteric pathogens were also present: two had Shigella, one entero-pathogenic E. coli, one rotavirus, and two Giardia lamblia.

It was found predominantly in older children. Only two out of 15 were less than 1 year old. Eleven were older than 2 years. This is a striking contrast to rotavirus gastroenteritis. There was a high incidence of children of immigrants (12 out of 15 (80%) compared with the usual rate of 52% for children with gastroenteritis admitted to Queen Elizabeth Hospital). In eight out of fifteen cases (53%) there had been recent travel abroad, either to the Indian subcontinent, or to Africa, compared with 5.3% in the gastroenteritis unit as a whole, i.e. the most important feature of these patients was the high incidence of those who had recently returned from tropical or subtropical areas. The most likely explanation for this finding is that it reflects the high incidence of coronavirus excretion in the areas from which they have returned; in other words, a transplanted third world population is being examined. Although it remains unexplained why these children are asymptomatic whereas coronavirus is found so often in asymptomatic children in India.

Only six patients had no other medical conditions or associated intestinal pathogens. These children had an episode of acute self-limiting diarrhoea of mild severity only. Two patients were pyrexial, although the temperature never rose above 37.9°C. No reducing substances were found in the stools of any child with coronavirus. Further work is required to determine the importance of this virus.

Clinical features

From among 1248 stool specimens examined during the 2-year period, February 1979 to February 1981 at Queen Elizabeth Hospital astrovirus was detected by electron microscopy in 42 stool specimens from 28 children (Nazer, Rice and Walker-Smith, 1983). Diarrhoea was an invariable clinical feature. Vomiting occurred in 18 children, abdominal pain in 7, and mild dehydration in 5, i.e. less than 5% dehydration. Fourteen of the children with acute diarrhoea were admitted directly to the gastroenteritis unit. The other fourteen children developed their acute diarrhoea some time after hospital admission. Although the excretion of astrovirus was associated with mild gastroenteritis, the presence of other enteric pathogens in 16 of the 28 children limited the degree to which the clinical symptoms could be attributed to astrovirus alone.

Excretion of astrovirus alone was found to be less common in summer than in winter. Excretion of astrovirus with rotavirus showed a winter peak.

Examination of serial stool specimens from some children showed that astrovirus excretion occurred for at least 7 days (Kurtz et al., 1977). In relation to the date of onset of diarrhoea, astrovirus was detected for at least 8 days in three children while in another the stool became negative for astrovirus 4 days after onset of diarrhoea. Transient monosaccharide intolerance lasting 1–2 days occurred in 18 children, and cows' milk protein intolerance requiring milk elimination for several months was a sequel in three children.

This experience then confirms that astrovirus occurs in hospital acquired infections and sporadic gastroenteritis. It has been detected in school outbreaks too and in nursing home outbreaks. On balance it seems probable that astrovirus is a human enteric pathogen causing mild diarrhoea and vomiting in children.

Shigellosis

Chantemesse and Widal in France, in 1888, were the first to associate the shigellae with dysentery but a more complete description was given in Japan by Shiga in 1898. Shigellosis is an acute enterocolitis due to infection with an organism of the shigella group. The term shigellosis is not synonymous with bacillary dysentery, which is infection with Shigella shiga. Other Shigellae are 5. sonnei, S. flexneri and S. boydii. Shigellosis is a highly contagious infection, requiring a very low infecting dose. Transmission is by person-to-person contact and food/water contamination by human faecal matter, man being the only natural host although flies may serve as vectors.

Salmonellosis

There are more than 1000 types of this Gram-negative bacillus that make up the genus salmonella. Infection with these organisms may cause a variety of illness in animals and in man.

Infection may result in several distinct clinical entities including acute gastroenteritis, septicaemia, focal infection, and the enteric fevers. The enteric fevers, caused by S. typhi and S. paratyphi are exclusive to man and present primarily as septicaemic disorders. In contrast, the other types of salmonella which are zoonoses most commonly result in disease localized to the gastrointestinal tract, i.e. acute gastroenteritis, although septicaemia with metastatic focal infection including meningitis, osteomyelitis, pneumonia, endocarditis, endopthalmitis and cholecystitis may occur, particularly in infancy.

Many infections with salmonellae in children arise from animal sources via contamination of food but food may also be contaminated by the infected stools or urine of a person who is excreting salmonellae. The risk of hospital cross-infection with salmonella infections is considerable, because of this infectivity of stools and urine. Small numbers of organisms can cause illness in a compromised host but very large numbers may have no effect in a healthy naturally immune host (Curry, 1976).

Salmonella infections are common in farm animals, particularly poultry. The increasing use of frozen foods, especially poultry, in developed countries may be an important factor in the observed trend of an increase of salmonellosis in such countries (Fenner, 1971). While infection with salmonella has been considered of greatest importance to children in developing countries salmonella food poisoning is a disease of modern western society.

Interestingly, there was a higher percentage of children with gastroenteritis who were found to have salmonellae isolated from their stools in Sydney (see Table 6.1), where the trend to increased poultry consumption is more advanced, than in the East End of London where children had salmonellae isolated. However, in fact there is now an overall trend in England and Wales of increasing reports of salmonella infections. This relates to an increase in food poisoning incidents due to serotypes other than S. typhimurium and from 1980 to 1984 an increase in incidents due to S. typhimurium. There was one notable outbreak in 1985 when S. ealing was found in dried milk powder from one manufacturer which probably originated from one cow in a herd supplying milk to the factory; 62% of reported cases were in infants less than 1 year of age and one died.

Incorrect cooking methods, for example inadequate cooking of incompletely thawed frozen poultry, are responsible for such infections. Salmonella is endemic in Britain and many foods both home-produced and imported are contaminated. Table 6.20 indicates some recent outbreaks. It clearly is a serious matter with death resulting in some cases. The incubation period is 12-36 hours.

Some types of salmonella infection occur both in animals and in man whereas others such as typhoid fever due to Salmonella typhi occur only in man. The small intestine is significantly involved in typhoid fever as, characteristically, the lymph follicles of Peyer's patches are inflamed and as a result ulceration may occur. Uncommonly in children, but more commonly in adults, perforation and haemorrhage may result. However, both typhoid fever and paratyphoid fever are primarily septicaemic diseases that manifest primarily as systemic disease, but on occasion a child admitted to a gastroenteritis unit may have this organism cultured from his stool.

The other species of salmonella cause predominantly a gastrointestinal disorder, but occasionally blood stream infection may occur and in neonates this is more common than in older infants. Neonates may even develop meningitis due to a salmonella organism. Other situations where blood stream infection may occur include sickle cell disease and chronic granulomatous disease. In both conditions osteomyelitis may develop and indeed may be the clinical mode of presentation of the salmonella infection.

This vulnerability of the neonate to blood stream infection with salmonella may be related to low levels of IgM in the neonatal period. This is particularly likely as it is known that antibodies against the O antigen of Gram-negative organisms are chiefly found in IgM. This may also account for the susceptibility of young infants to infection with enteropathogenic strains of E. coll. Penetration of small intestinal, particularly ileal, or colonic mucosa by this ‘invasive’ organism results in fluid secretion and diarrhoea which may be mediated by prostaglandins released secondary to inflammation (Bachino, Suchy and Snyder, 1984).

Most salmonellae have two main antigens, the H or flagellar antigen and the O or somatic antigen. Some have a third antigen known as the Vi antigen, e.g. Salmonella typhi. These various antigenic components may be used to identify the particular species of salmonella. Their number, as mentioned earlier, is large but Salmonella typhimurium is the commonest encountered in most series. Table 6.21 indicates the types of salmonellae encountered in children with gastroenteritis at the Royal Alexandra Hospital in 1970.

Over 500 children presented to the Queen Elizabeth Hospital for Children in London with salmonella infections of any kind over a 12-year period from 1973 to 1984 (Figure. 6.23 ). A substantial increase in the total number of infections from 1978 can be seen but infection due to 5. typhi and S. paratyphi has not increased over this time. During this period no deaths were attributable to salmonella infection.

There has been a more than threefold increase in salmonella as a cause of inpatient gastroenteritis over the twelve years, from 1.9% in 1973 to 2.2% in 1977 and 7.4% in 1984. A similar trend was seen in the number of salmonella isolates in all stools examined at this hospital including inpatients and outpatients in 1973; 10 out of 4912 (0.2%) stools examined contained salmonella species. This increased to 26 out of 4900 (0.5%) in 1977 and 47 out of 7390 (0.63%) in 1984. Thus in addition to an absolute increase in the number of salmonella infections of any kind, this organism is now responsible for a larger proportion of both inpatient gastroenteritis and positive bacterial isolates in the stools.

Escherichia coli enteritis

This subject has been well reviewed by Boedeker and Sherman (1986). Adam, in Germany, in 1923, identified special types of lactose splitting strains of E. coli as a cause of severe infantile gastroenteritis (Braun, 1974) but this observation was not widely accepted, and up until 1945 shigellae and salmonellae were the only bacterial pathogens to have been established as causes of infantile gastroenteritis. In that year, Bray at the Hillingdon Hospital in London published observations he had made concerning the isolation of enteropathogenic strains of E. coli from infants with summer diarrhoea. It has subsequently been widely confirmed that certain strains of E. coli are enteropathogenic. The criteria for accepting such enteropathogenicity usually includes the consistent isolation of a particular strain in an epidemic as the predominant coliform in the stools of affected infants and its relative rarity in the general community. The antigenic structure of each strain is recorded and reference criteria are developed for future recognition.

Several types of antigen are assessed in identifying the strains of E. coli. These include the O, K and H antigens. It is the O antigen that is particularly important in identifying enteropathogenic strains of E. coli. This antigen is a heat-stable lipopolysaccharide associated with the bacterial cell wall. A number of B subtypes are recognized. The enteropathogenic strains known at present are as follows:

These strains of E. coli may colonize the small intestine, and their capacity to do so is an important factor in determining their enteropathogenicity. This appears, in turn, to be related to the ability of these organisms to adhere to the surface mucosa (McNeish et al., 1975).

Another critical factor in determining their enteropathogenicity appears to be the elaboration by some strains of E. coli of a diarrhoea-producing enterotoxin (Orskov et al., 1976). Enterotoxigenic E. coli have been shown to produce two types of toxin. First, there is the heat-labile (LT) enterotoxin that is antigenically and functionally similar to cholera toxin. It stimulates adenyl cyclase, as does cholera enterotoxin, and also produces an excess loss of water into the intestinal lumen, i.e. a secretory state. The second toxin is the non-antigenic heat-stable (ST) toxin whose mode of action is via stimulation of another intracellular mediator (cyclic GMP) (Field, 1978). Our understanding of these mechanisms of enteropathogenicity have been considerably expanded by animal work, notably that of Williams Smith (1976). It has now been shown from such studies that the ability of E. coli to colonize the gut, adhere to mucosal surface, and to produce enterotoxin can be transmitted from one organism to another by transmissible plasmids. These are extra-chromosomal genetic elements. Thus plasmids could transfer the potential to produce one of these capacities, e.g. enterotoxin production, to organisms not ordinarily regarded as pathogens. At present, enterotoxin can be demonstrated only by its effect on animal models or on cultured cells. The development of a simple laboratory procedure that would allow rapid recognition of such an enterotoxin would be a great advantage and might demonstrate the Enteropathogenicity of hitherto unsuspected organisms.

Another critical factor that may determine the enteropathogenicity of E. coli strains is the ability of the organisms to invade the epithelial cells of the intestine. Du Pont and his colleagues (1971) suggested that two different types of clinical illness could occur, enterotoxin-producing strain or an invasive strain. In the former, severe watery diarrhoea similar to that seen in cases of cholera occurred, whereas in the latter an illness similar to shigellosis occurred.

The relative importance of enteropathogenic, enterotoxin-producing and invasive strains of E. coli in Britain remains uncertain. Although enterotoxin producing strains of E. coli that cause a severe disease with profound fluid loss, sometimes referred to as cholera infantum, may occur such disease is far more common in the developing world than in developed communities. The best evidence that these mechanisms of toxin production and invasion are causes of human disease comes from studies of E. coli strains that cause traveller's diarrhoea in adults. In fact, most studies of classical enteropathogenic E. coli serotypes have failed to show that those strains, associated with sporadic or epidemic cases of infantile diarrhoea, possess either invasive or toxin-producing properties. Levine et al. (1978) have shown that three enteropathogenic strains (0127, 0128, 0142) given to adult volunteers all cause diarrhoea, but despite this there is no evidence of toxin production or invasiveness of these strains in extensive laboratory studies. There is evidence in fact that the classic E. coli enteropathogenic organisms produce their effects by a completely different mechanism described as entero-adherent E. coli of the cytotoxic variety (Clausen et al., 1982) (Figure. 6.25 ).

Neonatal infection with an enteropathogenic strain of E. coli is particularly dangerous and can spread very rapidly through a nursery of newborn infants and be associated with a very severe form of diarrhoea and even death. The significance of the isolation of such a strain during an epidemic of infantile gastroenteritis is quite clear but its isolation from the stool of the occasional symptom-free infant is uncertain. Nevertheless, the isolation of such a strain from the stools of a child with diarrhoea under the age of 2 years is usually regarded as a significant finding.

Thus, to say the least, from the viewpoint of the practising clinician the present position concerning diagnosis and recognition of infection due to strains of enteropathogenic E. coli is most unsatisfactory. It is clear that invasive and toxin-producing strains of E. coli do exist and are not detected by conventional serotyping. Conventional serotyping continues to be valuable (Boedeker and Sherman, 1986). The author recommends the continuation of the present routine serotyping of E. coli.

Ironside (1973) has speculated that the disease may be due to the temporary clinical upset that occurs when new strains of E. coli reach the gut of the infant. A similar change is considered to be one of the causes of ‘travellers’ diarrhoea’. He suggests that a ‘baby's trip to the local nursery is the bacteriological equivalent of the adult's trip to Egypt’. The incubation period for infections with these strains is believed to be 24 to 72 hours.

Table 6.1 indicates the isolation rate for these organisms in four hospitals in different countries. Within Britain itself there has been a change since 1953 when the MRC trial of prophylactic antibiotic found that 37% of children with gastroenteritis were infected with one of the only two serotypes then known to be enteropathogenic. This figure is higher than the present figure of 9.5% at the Queen Elizabeth Hospital and the figure reported by Ironside, Tuxford and Heyworth in Manchester in 1970, namely 4%. These findings suggest a continuing change in the aetiology of gastroenteritis. Infection with enteropathogenic E. coli is associated with a high incidence of the post-enteritis syndrome.

Yersiniosis

The genus Yersinia contains three Gram-negative cocobacillary organisms pathogenic to man, Y. pestis, the plague bacillus, Y. enterocolitica and Y. pesudotuberculosis. The term yersiniosis is usually reserved for disease caused by the last two.

Campylobacter enteritis

Campylobacter is a Gram-negative motile bacterium which was first described in 1913 (MacFadyean and Stockman) and was placed in the genus vibrio. In 1963 the alternative name Campylobacter was used as this term means curved rod in Greek, which describes the shape of these bacteria. The most common species are Campylobacter jejuni and Campylobacter coli which cause disease in man. More rarely Campylobacter fetus and Campylobacter laridis may cause infection in immunocompromised patients. It has been claimed that Campylobacter pyloris may cause gastritis. It was only in 1972 that Dekeyser et al. first associated Campylobacter infection with human disease. This was largely related to the fact that previous culture techniques were unsatisfactory.

It is now known to be a common cause of bacterial diarrhoea in the United Kingdom and is a common stool pathogen in acute gastroenteritis at Queen Elizabeth Hospital for Children at present (Table 6.20). Worldwide it is very common (Billingham, 1981). Transmission of infection from human to human is unusual and most often transmission is from animals to man. Human infection is a zoonosis. Sources of infection include frozen poultry, wild birds, milk and domestic pets (Robinson, 1981).

Aeromonas infection

Aeromonas spp. are found in soil and natural water sources. They have been recognized for many years as a cause of disease in non-mammalian species including red leg disease of frogs and black rot in hen's eggs, but they have also been found in mammals viz, fish and pigs. Gracey et al. 1982 reported aeromonas-associated gastroenteritis in children. Further reports of aeromonas as a potential cause of acute diarrhoea in children have since been published (Kipperman et al., 1984; Ljungh et al., 1985; Nazer et al., 1986). Aeromonas has also been described as a cause of traveller's diarrhoea (Gracey et al., 1984; Taylor et al., 1985).

Terminology

Most often acute gastroenteritis is an acute self-limiting illness and after 24 hours of a glucose electrolyte solution followed by a regrade to the infant's usual cows' milk based feeding formula there is full recovery. Some looseness of stools may persist for a few days. However, there is no worsening of diarrhoea, return of vomiting or further risk of dehydration and weight gain is satisfactory, once there is adequate calorie intake. Thus there is uneventful recovery. In some infants recovery is delayed. Most often this may present as an intolerance to milk as this is reintroduced into the child's diet i.e. there is a failed re-regrade or a clinical relapse requiring the child to return to a glucose electrolyte solution or even require an intravenous infusion. When such delayed recovery occurs up to 2 weeks after the commencement of oral rehydration therapy it is likely to be part of the natural history of the gastrointestinal infection as most cases of gastroenteritis especially those of viral origin are associated with pathogen excretion in the stools for no more than 10 days to a fortnight. Hence during this period any clinical problem with relapse of symptoms may be embraced by the umbrella term delayed recovery. Although this term does extend beyond this period, once 2 weeks have elapsed then persistence of symptoms must be regarded as being part of the post-enteritis syndrome. This then must be regarded as a problem extending beyond the expected period of spontaneous recovery. It must be remembered, however, that in the cases of some bacterial syndromes of gastroenteritis, for example shigellosis, the natural history of the disease may often extend for several weeks.

Aetiology

The cause of post-enteritis enteropathy is uncertain but one explanation is that there is an allergic or hypersensitivity reaction, i.e. there may be an immunological explanation. Another cause may be that there is persistence of gastrointestinal infection beyond the normal span of self-limiting disease or acquisition of further infectious agents causing persistence of the small intestinal mucosal damage. Put another way post-enteritis enteropathy may be due to the following factors:

• •

  Persistent infection with the original infectious agent.

• •

  Reinfection with another pathogen or pathogens.

• •

  Sensitization of food antigens causing persistence of small intestinal mucosal damage.

Post-enteritis-enteropathy: infective enteropathy

Post-enteritis enteropathy may be due to persistent infection i.e. infection continuing to cause damage. The best example of this is provided by infection with enteroadherent E. coli of the cytotoxic variety producing small intestinal mucosal damage (Figure. 6.25). This may be very severe as seen as one variant of the intractable diarrhoea syndrome seen at Queen Elizabeth Hospital and described as Traveller's Diarrhoea with a Vengeance’ (Hutchins et al. 1982). This concerns children known to be well in the land of their birth (United Kingdom) but who develop severe diarrhoea when taken to their ancestral homeland (Indian subcontinent etc.) where they are exposed to a developing community where they are in contact with a heavily contaminated environment. Small intestinal biopsy in these children shows an enteropathy with clear evidence of infection e.g. neutrophils etc. and bacteria may be identified upon the surface of the mucosa adhering to the enterocytes in histological sections. However, the electron microscope gives the most revealing picture. These children have intractable diarrhoea and the damage has persisted long beyond the natural history of the acute infection.

A similar situation may occur in combined immunodeficiency syndromes where for example rotavirus infection may persist with an enteropathy. Chronic rotavirus infection is always associated with disturbed immune function.

The finding of enteropathy in a child with the post-enteritis syndrome may also be due to secondary or acquired infection. For example, Phillips has found astrovirus and adenovirus in the mucosa of children being biopsied for the post-enteritis syndrome who presumably have had intercurrent illnesses accounting for the virus in the abnormal mucosa.

Post-enteritis-enteropathy: food sensitive enteropathy

It is clear that some children with a post-enteritis enteropathy have a food sensitive enteropathy, most often cows' milk sensitive enteropathy. The question that must arise is why does this occur or even why don't all children who develop acute gastroenteritis develop this problem?

A partial explanation may be because there is an immunodeficiency state in those who do either locally in the small intestine or systemically.

In 8 of 20 cases of cows' milk sensitive enteropathy complicating acute gastroenteritis described by Harrison et al. (1976) at the time of diagnosis serum IgA levels were low rising to normal levels with recovery. Haidas et al. in a pilot study found that IgA numbers in the small intestinal mucosa were reduced in children diagnosed as milk-elimination responsive enteropathy, largely post-enteritis using a horseradish peroxidase method. Surprisingly, using an immunofluorescent technique Kilby et al. (1976) found IgA cell numbers to be increased. Kilby's children were more heterogeneous but it is not easy to resolve this difference. Nevertheless the possibility that there is a temporary IgA deficiency as suggested by Soothill (1976) still remains. A temporary opsonization defect has also been described. In addition immaturity of suppressor T-cell function has been suggested. The relative importance of all these immunodeficiency states must still be regarded as uncertain. Another partial explanation may relate to the immaturity of the small intestinal mucosa. The thinness of the mucosa in infants with cows' milk sensitive enteropathy as described by Maluenda et al. (1984) is in accord with this as well as the apparent limited ability of the crypt cells to respond to enterocyte loss. This immaturity may be important too in the ability of food antigens to adhere more readily to the microvillous membrane in immature animals. It is certainly true that post-enteritis food-sensitive enteropathy occurs particularly in young infants and is confined to infants less than 2 years of age.

Increased antigen entry may also play a part. There is certainly evidence that anatomical pathways do exist for antigen entry in the biopsies of children with post-enteritis enteropathy (Jackson et al., 1983). These may be shown by means of horseradish peroxidase uptake in biopsies from children with post-enteritis enteropathy (see Figure. 6.26). Thus there is morphological evidence that antigen entry can occur but what is critical is whether such entry is associated with sensitization. A key factor here may be the antigenicity or allergenicity of the food usually cows' milk fed at the time of acute gastroenteritis. There are both animal and clinical observations to suggest that allergenicity is of critical importance. Figure. 5.3 illustrates an hypothesis linking these factors. The declining importance of cows' milk sensitive enteropathy in Western societies may relate to reduced antigenicity of modern feeds.

Such a food-sensitive enteropathy may overlap with an infective enteropathy as they may both coexist. Walker-Smith et al. (1978) described 5 children with post-enteritis enteropathy who responded to a cows' milk-free diet yet who also had evidence of infection at the time of biopsy.

Malnutrition may play a role by interfering with the ability of the damaged mucosa to recover; however, the author believes that the importance of malnutrition as a cause of enteropathy has been exaggerated in the past (see Chapter. 13).

Diagnosis

When an infant with the post-enteritis syndrome has an abnormal mucosa the nature of the lesion as described above will influence diagnosis. When bacteria are present with the typical morphology of enteroadherent E. coli then the diagnosis is clear (see Figure. 6.25). If a child being fed cows' milk has an enteropathy with the features described in Chapter. 5 then cows' milk sensitive enteropathy is likely. If he then responds rapidly to a cows' milk-free diet with relief of diarrhoea and weight gain this is the likely diagnosis. Management is discussed in Chapter. 5.

Post-enteritis syndrome without an enteropathy

Turning attention now to the mechanism of chronic diarrhoea in children with this syndrome who do not have an enteropathy, the role of disturbed immunological function must also be considered. Avigad et al. (1978) have investigated the flora of the small intestine in children with the post-enteritis syndrome. They cultured duodenal juice and the homogenized duodenal mucosa from children with the post-enteritis syndrome, children with other causes of chronic diarrhoea, and a control group of children without diarrhoea. They found no significant difference in bacterial flora as measured by total bacterial flora within these three groups. The role of bacterial colonization of the small intestine and enteric disease was discussed briefly in Chapter. 1. Its role in this syndrome is not yet clear but Avigad et al. (1978) using an immunofluorescent technique were able to demonstrate mucosal antibodies against the bacteria grown from the luminal fluid or mucosa of the same children from children with the post-enteritis syndrome, and other causes of chronic diarrhoea. The finding of such antibodies was unrelated to the presence or absence of an enteropathy. Mucosal antibody to these non-pathogenic bacteria, obtained from the duodenal juice or homogenate culture, was present in 8 of 9 children with the post-enteritis syndrome regardless of whether an enteropathy was present. Antibody was also found in three children with other causes of chronic diarrhoea, but antibody was not present in the control group who did not have diarrhoea at the time of biopsy, except in one child who had previously had an enteropathy, and possibly one doubtful case who also had a previous enteropathy. The role of such small intestinal mucosal antibodies against non-pathogenic organisms cultured from the small intestine of these children is not clear in relation to the pathogenesis of chronic diarrhoea in children, but clearly warrants further study.

Phillips et al. (1978) have used the scanning electron microscope to study the surface of small intestinal lymphoid follicles obtained on proximal small intestinal biopsy which was morphologically normal (Figure. 6.29 ). Normally, no bacteria adhere to the surface of such follicles but as Figure. 6.29 shows there can be adhesion of rod-like bacteria to the surface, whereas bacteria were not adhering to the rest of the biopsy. This was a specimen taken from a child with post-enteritis syndrome shown to be producing antibody to rod-like bacteria. This preferential adhesion of bacteria to the surface of the lymphoid follicular region is interesting because there is now good evidence that antigen entry occurs through the epithelium overlying lymphoid follicles, and that immunoblast cells are primed at these sites prior to migration to the lamina propria of the gut mucosa where they secrete immunoglobulin (see Chapter. 1). Thus adherence of bacteria to the overlying epithelium of the lymph follicle may be related to antibody production and may have significance in the pathogenesis of this child's continuing diarrhoea.

Clearly, a great deal more research is required into the genesis of post-enteritis diarrhoea. This is an important subject both in the developed and developing communities, but in the latter it is a problem of enormous size.