Article

Urinary tract infection in children: a review of its pathogenesis and risk factors M

Twaij

Mohammed

Twaij, MB, ChB, DCH, Staff Paediatrician, Paediatric Department, East Surrey Hospital, Redhill, Surrey RH1 5RH,

England

Key words Dysuria; frequency; reflux nephropathy (RN); urinary tract infection (UTI); vesico-ureteric reflux (VUR) Introduction

beginning of the 1900s there was approximately 20% mortality in infants hospitalised for pyelonephritis. With modem management, intensive diagnostic procedures and the help of current antibiotic therapy the mortality nowadays is almost negligible. However, the problem that remains is having to deal with the consequences of UTIs, i.e. reflux nephropathy, kidney scarring, progressive renal failure, hypertension, complications during pregnancy and end-stage renal disease. The true incidence of UTI in children in the UK is not known, although reports from general practices and health districts suggest an incidence of 1.7 to 12.1 per 1,000 boys and 3.1 to 30.7 At the

Abstract Urinary tract infection (UTI) is one of the commonest bacterial infections seen in children, ranking second only to those of the respiratory tract. A steady flow of urine, complete emptying of the bladder and various host defences provide the principal means of protection. Urinary stasis, on the other hand, provides an opportunistic environment for bacterial multiplication; several other factors also contribute to UTI, such as urinary obstruction, vesico-ureteric reflux (VUR) and other urinary tract malformations. UTI, especially in the presence of VUR, may lead to scarring of the kidney and progressive renal disease. It is vital, therefore, to diagnose such anomalies at an early stage, the overall aim being to identify those at high risk so that measures may be taken to prevent serious complications and preserve renal function. Those at low risk of UTI, however, should not be over investigated; this is in order to keep invasive procedures down to a minimum. Whenever possible, the management of high risk cases is best discussed between paediatrician, microbiologist, radiologist, urologist and the general

practitioner.

per 1,000 girls (Benador et al,1997). Recent advances in diagnostic imaging techniques have led to considerable improvement in the early detection of renal abnormalities and their management. On account of current ante-natal scanning one can with confidence detect some renal tract abnormalities at an early stage, e.g. hydronephrosis, mega-ureter, duplex renal systems and multicystic dysplastic

kidney (Figure 1).

Pathogenesis of urinary tract infections Gram-negative aerobic bacilli, usually of bowel origin, are the main cause of most UTIs; they obtain access to the urinary tract by way of the ascending route via the urethra. Escherichia coli accounts for over 60% of UTIs (Table 1); these have now been identified as a large genetically heterogeneous group (Zhang et a4 2000). Other commonly found bacteria are those of the Proteus spp (common in boys and also in the presence of renal stones), Klebsiella spp, Pseudomonas spp and Enterococcus spp. Less common micro-organisms include Staphylococcus epidennidis and Staphylococcus aureus Oodal and Winberg, 1987; Buys et at 1994). In some parts of the world Mycobacteriutn tuberculosis may be isolated from urine; confirmation requires special culture media. Viruses, too, cause UTIs - and include mumps, herpes simplex, vaccinia, adenoviruses and measles. Fungal UTIs (funguria) are rare, usually caused by Candida spp, but may also be due to Cryptococcus neofonnans and Aspergillus spp, leading to a wide variety of presentations ranging from asymptomatic candiduria to clinical sepsis (Sobel and Vazquez, 1999). The pathogenesis of UTIs depends on complex host-parasite interacthe extent of urethral colonisation and bacterial in the urine. Whilst in the urinary tract, certain bacteria are able to growth adhere to the mucosal surface, thereby resisting the mechanical effects of voiding, hence causing infection. The severity of such infection depends upon the virulence of the micro-organism and integrity of the host defence system. There are various specific recognised virulence factors within bacte-

tions, which determine

220

ria that determine include presence of

pathogenicity; these K antigens, capsular polysaccharide and glycoprotein fimbriae,

mediators - and the lipopolysaccharide content of their outer cell membranes. An example of a virulent organism in this category is the uropathogenic E. coli; this has the ability to enhance colonisation of the uroepithelium, induce inflammatory reaction and influence the epithelial cytokine response. Following the initial interaction of bacteria with bladder mucosa, there will be a surge in white cell proliferation, causing an increase of neutrophils and interleukins in the urine (Agace et at 1993; Tullus et as

1994). Fimbriae are rigid hair-like projections of bacterial cells that recognise specific receptor sites and have the ability to agglutinate red blood cells (Blanco et A 1997). The presence of P-fimbriae on E. coli is a significant determinant of virulence - especially in the case of upper UTI. The polysaccharide and glycoprotein mediators secreted by the capsule enhance adherence to epithelial cells, reduce phagocytic activity and compromise the effects of antibiotics. The lipopolysaccharides of the outer cell membrane can prevent complement activation and inhibit adhesion to macrophages. The capsular K antigens (five different types are recognised) are responsible for about 70% of cases of pyelonephritis in children. Other virulence factors are adherence to uroepithelial cells, production of haemolysin and colicin, the ability to acquire iron - and resistance to serum bactericidal activity (Rushton, 1997). UTI is accompanied by a cytokine response in the infected child with increased intracellular secretion of interleukin 6 (IL-6) and interleukin 8 (IL-8).

Clinical

sionally, by adenovirus type 11 et al, 1996). Risk factors for infections

and 12

(Lee

urinary tract

Considerable clinical evidence has established several risk factors contributing to UTIs. It is essential, therefore, to identify those children who have UTIs in association with progressive deterioration of renal function. Detection of anomalies such as renal tract obstruction, hypoplastic or dysplastic kidneys is very important, in order that measures for preserving renal function and preventing consequential further renal damage can be taken. Children with such malformations may have poor renal function from the time of birth. Although urinary stasis is so often the predisposing basis of UTI there may sometimes be other contributory factors; these will be elaborated

upon in the

succeeding paragraphs (Table 2). In addition, there are other aspects, not necessarily related directly to the urinary tract

but, nevertheless, predisposing to UTI;

these include bad personal hygiene, the use of bubble baths, tight clothing, sexual abuse, sexual activity and pregnancy (Table 3).

Vesico-ureteric reflux VUR is the retrograde flow of urine from bladder to ureter (Figure 2) and is a common anomaly of the renal tract with a prevalence of around 1 to 2% in normal children - but up to 30 to 40% following the first episode of UTI. There are four grades of VUR; the majority of cases are mild (grades I and II) and these usually resolve spontaneously in 80% of patients. The more severe cases (grades III and IV) may, however, cause sig-

nificant kidney scarring and deterioration of renal function; these are usually referred to as ’reflux nephropathy’ (RN).

symptoms

this review is not intended to include the symptomatology of UTI in detail, a brief discussion thereon at this point will be useful. The clinical presentation depends essentially on the age of the child. Younger children usually have nonspecific symptoms - such as pyrexia, irritability, vomiting and failure to thrive. Older children have more specific symptoms -

Although

namely, dysuria, urinary frequency, wetting, supra-pubic and/or loin pain and haematuria, especially in the

case

of acute haem-

orrhagic cystitis caused by E. coli and, occa221

Primary VUR, which may be inherited as autosomal dominant trait, is due to an anatomical defect or delayed maturation of the vesico-ureteric junction (Elder et al, 1997; Feather et al, 2000), whereby the ureter is inserted laterally and obliquely into the bladder, thus permitting retrograde flow of urine. Other suggested causes of primary VUR include abnormal innervations of the an

vesico-ureteric

junction,

abnormalities of

collagen and/or musculature of the ureteric wall. Genetic factors too, may have a determinant role in the development of idiopathic VUR. Reflux is present in over 25% of siblings bom to parents with VUR - and also in 36% of siblings with VUR. Secondary VUR, on the other hand, results from raised intra-vesical pressure

caused by urinary outflow obstruction, neuropathic bladder, constipation and occasionally follows trauma and surgical procedures. Children with a duplex renal system (Figure 3) may also have reflux - usually of the lower moiety, whereas obstruction is more likely to involve the upper moiety. When both ureters open into the bladder in close apposition, resembling a double-barrelled gun, reflux may affect both ureters. The mechanism of kidney damage in children with UTI in the presence of VUR has been postulated as being due to intrarenal reflux (IRR); this may be expressed in brief as UTI + VUR + IRR = RN. IRR is the backflow of urine from the renal pelvis into the renal papilla when calyceal pressure is high enough to reverse urine flow in the collecting tubules. IRR occurs in the compound papillae that are usually found in the poles of the kidney; it does not occur in the simple papillae where slit-like openings prevent the backflow of urine. Compound papillae are normally present in 75% of human kidneys, but only become significant when there is high pelvi-calyceal pressure resulting from outflow obstruction or highpressure reflux. It has been suggested that patients with RN have raised serum cytokines; this, in turn, would indicate an immunological basis Qutley et al, 2000). It has also been shown that kidney scarring is more related to the nature of the infection rather than to VUR per se (Linshaw,

1999).

Urinary tract obstruction The pathophysiological effects of bladder outflow obstruction depend on the nature, level, extent and speed of onset. Urinary tract obstruction can occur at any level from the urethra up to the calyceal system of the kidney. High-grade ureteral obstruction of early onset in foetal life may result in renal dysplasia and poor kidney function. Less chronic ureteral obstruction in late foetal life, or shortly after birth, may result in hypertrophy and dilatation of the upper collecting system and ureter. UTIs usually complicate obstruction, hence causing further renal damage. Symptoms of obstruction include hesitancy, poor urinary stream, stop-start voiding, terminal dribbling and a feeling of incomplete bladder emptying. Some of the causes of renal tract obstruction are discussed in the next section. 222

Mechanical facton

Constipation Constipation is a common, often underrecognised, problem in children; it predisposes to UTI as a result of bladder dysfunction caused by a distended rectum compressing the bladder and its neck. The condition may thus result in partial or even complete bladder outflow obstruction leading to recurrent UTIs, voiding disorder, detrusor instability, urinary incontinence and VUR. Clinical evidence suggests an association between constipation, recurrent UTIs and delayed resolution of reflux (Koff et al, 1998). The diagnosis is based on history and clinical examination; radiological examination is not always necessary (Figure 4). The management of constipation should be part of the treatment of children with recurrent UTIs, VUR and dysfunctional voiding. Successful management of constipation will result in normal bladder function and cessation of infection in the majority of cases

(Loening-Baucke, 1997).

Posterior urethral valve Congenital posterior urethral valves (pUV s) occur only in boys and are important factors leading to urinary tract obstruction (Silverman and Kuhn, 1993). PUVs cause urinary outflow obstruction with subsequent increased intra-vesical pressure, VUR (Figure 5) and bladder wall thickening. Prolonged obstruction in utero often results in dysplastic kidneys with impaired renal function. Ante-natal scans usually show bilateral hydronephrosis and hydroureter - with possible dysplastic kidney changes depending on severity of the PUV. The diagnosis is confirmed postnatally by performing a

the trigone of the bladder; however, narrow-necked ones are usually located near the ureteric orifice. Diverticuli are mostly congenital - but may be acquired, occurring in association with PUV and neuropathic bladder; the latter often resolve following correction of the underling problem. Diverticuli may remain undiagnosed unless symptoms arise as a result of UTI, VUR and obstruction of the ureter or bladder neck. Renal calculi The incidence and type of renal calculi vary in different parts of the world - and also at different sites within the renal tract itself. The geographical distribution of urolithiasis seems to be changing according to socioeconomic status. Calculi of the upper renal tract are more common in developed countries, whereas those in the bladder are more frequent in economically under-developed

countries (Trinchieri, 1996). In the UK urinary calculi affect one to two children per 1,000,000 (Ghazali et al, 1973) each year; they mainly affect the upper urinary tract (Figure 7) and are usually associated with UTIs caused by urease-producing bacteria. Stones may remain clinically silent until they

begin to move down the ureter, causing spasmodic pain. There are two types of renal calculi in children - primary and secondary. The primary type is usually of renal origin; amongst its causes are idiopathic hypercalciuria with normocalcaemia, hypercalciuria with hypercalcaemia, resorptive hypercalcaemia (mainly as a result of prolonged immobilisation), renal tubular acidosis and the use of carbonic anhydrase inhibitor compounds. Secondary urinary calculi, however, are rare and develop mainly from complex nidus formation where the

cysto-urethrogram. Ureterocele Ureterocele is cystic dilatation of the intravesical portion of the ureter, girls being more affected than boys - in a ratio of 3-4:1. The left ureter is affected more frequently than the right, the condition being bilateral in about 10% of cases. Ureterocele may be part of a single system, but occasionally is associated with a duplex renal system affecting the upper pole ureter (Figure 6). Diverticuli of the bladder A bladder diverticulum is a herniation of the lining mucosa caused by weakness of the detrusor muscle, often occurring in association with elevated intra-vesical pressure. The majority of diverticuli originate from 223

detrusor areflexia

hyper-reflexia. Bladder dysfunction and voiding or

disorders

urine is infected in the presence of urinary tract obstruction. Renal calculi themselves can lead to ureteral obstruction, obstructive uropathy and kidney damage.

Neuropathic causes Neuropathic bladder results from many neurological disorders affecting the spinal cord and, to a lesser extent, the brain. There four types of neuropathic bladder

are

depending on site and level of the lesion affecting the CNS; these are supra-sacral, sacral, peripheral and suprapontine. The common causes of such a problem are spina bifida, sacral agenesis and injury to the spinal cord. Clinical presentation includes dysfunctional voiding and urinary incontinence - characterised by spontaneous

uninhibited detrusor contractions,

The pontine micturition centre in the brain controls voiding, by way of co-ordinating contraction and relaxation of the detrusor muscle and urethral sphincter. Parasympathetic anticholinergic nerves (S2-S4) innervate the detrusor muscle. Dysfunction of the anatomy and physiology of the bladder may result in disordered storage capacity and/or abnormal voiding. Bladder dysfunction and residual urine are probably the major factors in causing recurrent UTIs in the absence of VUR or renal tract obstruction gjdefelt et al, 1989). In a population of school girls with asymptomatic bacteriuria, about half had more than 5 mL of residual urine (Iindberg et al, 1975); they had a 75% recurrency rate for UTI within one year, compared with 17% in those having 0-5 mL residual urine. Emphasis is made on the frequent association of dysfunctional bladder and VUR in children with recurrent UTIs. The increased risk of UTIs and VUR in children with dysfunctional voiding results from inadequate bladder emptying; this, in turn, leads to a high residual urine volume, adding to the increased intra-vesical pressure created by uninhibited bladder contraction. Several forms of bladder dysfunction are recognised in children; they range from a small capacity unstable bladder to infrequent voiding with a large capacity bladder (but with poor voiding). There have been several reports suggesting a link between dysfunctional voiding and UTIs (Hansson, 1992).

Inflammation Inflammation of the bladder wall induces irritation of the mucosa that may affect sensation - causing an increased desire to void. This increased irritability may cause detrusor muscle instability with subsequent poor function, thus predisposing to recurrent UTIs (Mayo and Burns, 1990). Inflammation of the bladder is usually caused by infection, trauma (e.g. bladder catheterisation, introduction of foreign body via the urethra, etc), sexual abuse and the use of some cytotoxic medications (e.g. cyclophosphamide) ; in other parts of the world it may be caused by parasitic infestation (e.g. schistosomiasis). The inflammatory reaction of the bladder wall mucosa leads to reduction in functional capacity and/or hyperactivity 224

of the detrusor muscle, thus increasing intra-vesical pressure - with subsequent VUR.

Host defences excellent medium for bacterial growth; micro-organisms reach the bladder they multiply quickly, thus causing inflammation. The integrity of host defence, namely cellular and humoral immunity with other specific factors on the uroepithelial surface, together determine the severity of a disease (Funfstuck et ai, 1997). However, the bladder is able to resist bacterial invasion to some degree in two principal ways; Urine is

an

once

these

(i) mechanically - via regular empof the bladder, thus reducing the chance of colonisation, and (ii) through the protective effect of antibacterial factors secreted by the lining epithelial cells. In addition, there are biological defences (such as variable urine osmolality and the mucoid layer of Tamm-Horsfall protein) and immunological defences (such as presence of secretory IgA from the bladder lining and the phagocytic activity of macrophages). Failure of the immune system, as in longterm steroid therapy, immunocompromisation or HIV infection, predisposes patients to recurrent infections. are

tying

Uncircumcised male infants Uncircumcised male infants appear to be at an increased risk of UTI - at least in the first few months of life, compared with those who have been circumcised (Linshaw, 1999). There is an approximately nine-fold reduction in incidence of UTI during the first year of life when circumcision has been performed during the neonatal period (Schoen et al, 2000). Colonisation of the foreskin with bacteria, namely E. coli and Proteus spp, place these infants at risk of UTI (Wiswell et al, 1988).

and adhere to specific carbohydrate receptors of the P blood group antigen that are present on human uroepithelial cells (Kallenius et a.~ 1980; Leffler and Svan-

prominent in hotter weather. Frequent fluid intake will help to rinse the bladder, giving

borg Eden, 1980). Inadequate fluid intake Inadequate fluid intake will inevitably lead to concentration of urine and less frequent voiding, thus allowing bacterial multiplication. Concentrated urine also permits crystallisation of its contents, resulting in stone formation. The problem is less common in

UTI in children is one of the commonest clinical problems facing general practitioners (GPs) and paediatricians. Patients at high risk, such as those with VUR, renal tract obstruction or dilatation, should be identified and investigated thoroughly. Rapid diagnosis of UTI and early treatment, particularly in young children, is vital if irreversible renal damage is to be avoided. As a

recognise

colder climates but is, of course,

more

little time for bacteria

to

multiply.

Conclusions

Patients with blood group P1 Individuals of blood group P have cellular receptor sites for P-fimbriated E. coli; which predispose to pyelonephritis by way of the ascending route. Patients with phenotype P1, have more than 11 times the risk of attracting recurrent episodes of pyelonephritis with P-fimbriated E. coli than do those of phenotype P2 (Lomberg et a~ 1983). This is probably because P1 individuals express better receptors for P-fimbriated E. coli in the intestinal tract. It has been established that P-fimbriae are able to

225

result of advances in ante-natal diagnosis of renal tract anomalies it is now possible to identify high-risk infants with renal abnormalities. The commonest risk factor for UTI is VUR, which accounts for 35% of patients following a first episode of UTI. The management of UTI in children is under regular review; guidelines and protocols are constantly being updated - with the intention of modifying and reducing the number of investigations. The overall aim is to prevent serious renal damage and improve long-term prognosis. It is essential, therefore, that GPs, hospital doctors and other healthcare providers should follow up-to-date guidelines and be aware of the current diagnostic procedures available regarding exactly what is involved and how to interpret the results.

Acknowledgements The author wishes to thank Dr David Donaldson, Consultant Chemical Pathologist, East Surrey Hospital, Redhill, Surrey, for assistance in presenting this review.

References AGACE W W, HEDGES S R, CESKA M and SVANBORG C (1993). Interleukin-8 and the neutrophil response to mucosal gram-negative infecJ Clin Invest; 92, 780-785 tion. BENADOR D, BENADOR N, SLOSMAN D, MERMILLOD B and GIRARDIN E (1997). Are younger children at highest risk of renal sequelae after pyelonephritis? Lancet; 349,17-19 BLANCO M, BLANCOJ E, ALONSO M P, MORA A, BALSALOBRE C, MUÑOA F, JUÁREZ A and BLANCOJ (1997). Detection of PAP, SFA and AFA adhesin-encoding operons in uropathogenic Escherichia coli strains; relation with expression of adhesins and production of toxins. Res Microbiol ; 148, 745-755

atric vesicoureteral reflux guidelines panel summary report on the management of primary vesicoureteral reflux in children. J Urol; 157(5), 1846-1851 FEATHER S A, MALCOLM S, WOOLF A S, WRIGHT V, BLAYDON D, REID C J, FLINTER F A, PROESMANS W, DEVRIENDT K, CARTER J, WARWICKER P, GOODSHIP T H and GOODSHIP J A (2000). Primary, nonsyndromic vesicoureteric reflux and its nephropathy is genetically heterogeneous, with a locus on chromosome 1. Am J Hum Genet, 66(4), 1420-1425 FUNFSTUCK R, STEIN G (1997).

SMITHJ W, TSCHAPE H and Pathogenetic aspects of uncomtract infection: recent advances.

plicated urinary Clin Nephrol; 47(1),13-18

GHAZALI S, BARRATI T M and WILLIAMS D I (1973). Childhood urolithiasis in Britain. Arch Dis

child; 48, 291-295 HANSSON S (1992). Urinary incontinence in children and associated problems. Scand J Urol

Nephrol, 141, 47-55 JODAL U and WINBERGJ (1987). Management of children with unobstructed urinary tract infection. 1,647-656 Paediatr Nephrol; JUTLEY R S, YOUNGSON G G, EREMIN O and NINAN G K (2000). Serum cytokine profile in reflux nephropathy. Pediatr Surgery Int; 16(1-2), 6468

KALLENIUS G, MOLLBY R, SVENSSON S B, et al (1980). The pK antigen as receptor for the of hemagglutination pyelonephritogenic Escherichia coli. FEMS Microbiol Lett; 7, 297-302 KOFF S A, WAGNER T T and JAYANTHI V R (1998). The relationship among dysfunctional elimination syndromes, primary vesicoureteral reflux and urinary tract infections in children. J Urol;

160(3Pt2),1019-1022 LEE H-J, PYO J-W, CHOI E-H, et al (1996). Isolation of adenovirus type 7 from the urine of children with acute hemorrhagic cystitis. Pediatr Infects

Dis J; 37 , 15 LEFFLER H and SVANBORG EDEN C (1980). Chemical identification of a glycosphingolipid receptor for Escherichia coli attaching to human urinaryl tract epithelial cells and agglutinating human erythrocytes. FEMS Microbiol Lett; 8, 127134

BUYS H, PEAD L, HALLETT R and MASKELL R (1994). Suprapubic aspiration under ultrasound guidance in children with fever of undiagnosed cause. BMJ; 308, 690-692

LIDEFELT K J, ERASMIE U and BOLLGREN I (1989). Residual urine in children with acute cystitis and in healthy children: assessment by sonogra-

PETERS C A, ARANT B S Jr, EWALT D H, HAWTREY C E, HURWITZ R S, PARROTT T S, SNYDER H M III, WEISS R A, WOOLF S H and HASSELBLAD V (1997). Pedi-

phy. 141, Urol; J 916-917 LINDBERG U, BJURE J, HAUGSTVEDT S and JODAL U (1975). Asymptomatic bacteriuria in schoolgirls. III: Relation between residual urine vol-

ELDERJ S,

226

ume

and

recurrence.

Acta Paediatr Scand;

64(3),

437-440

LINSHAW M A (1999). Controversies in childhood urinary tract infections. World J Urol; 17(6), 383-395 LOENING-BAUCKE V (1997). Urinary incontinence and urinary tract infection and their resolution with treatment of chronic constipation of childhood. Pediatrics; 100, 228-232 LOMBERG H, HANSSON L A, JACOBSON B, LEFFLER H and EDEN C S (1983). Correlation of p-blood group phenotype, vesicoureteral reflux and bacterial attachments in patients with recurrent

pyelonephritis. N Eng J Med; 308,1189-1192 HELLSTROM M, JODAL U, LEFFLER H, LINCOLN K and SVANBORG EDEN C (1994). Virulence-associated traits in Escherichia coli causing first and recurrent episodes of urine tract infection in children with or without vesicoureteric reflux. J Infect 150, 561-569 Dis; MAYO M E and BURNS M W namics in children. BJU; 65, 641

(1990). Urody-

RUSHTON H G (1997). Urinary tract infection in children: epidemiology, evaluation and management. Pediatr Clin North Am; 44(5),1133-1169 SCHOEN E J, COLBY C J and RAY G T (2000). Newborn circumcision decreases incidence and costs of urinary tract infections during the first year of life. Pediatrics ; 105(4Pt1), 789-793 SILVERMAN F N and KUHN J P (1993). Abnormalities of the pelvocaliceal system, ureter, bladder and urethra. In: Pediatric X-ray Diagnosis, p 1284. (Edited by JJ Caffey). St Louis: Mosby

SOBELJ D and VAZQUEZJ A (1999). Fungal infections of the urinary tract. World J Urol; 17(6), 410-414 TRINCHIERI A (1996). Epidemiology of urolithiasis. Arch Ital Urol Androl; 68(4), 203-249 TULLUS K, FITURI O, BURMAN G, WRETLIND B and BRAUNER A (1994). Interleukin-6 and ineterleukin-8 in the urine of children with acute pyelonephritis. Pediatr Nephrol; 8, 280284 WISWELL T E, MILLER G M, GELSTON H M

Jr and JONES S K (1988). Effect of circumcision status on periurethral bacterial flora during the first year J of Pediatr; 113, life 442-446 ZHANG L, FOXMAN B, MANNING S D, TALLMAN P and MARRS C F (2000). Molecular epidemiologic approaches to urinary tract infection: gene discovery in uropathogenic Escherichia coli. Infection and Immunity, 68(4), 2009-2015