A new concept and a comprehensive evaluation of SYSMEX UF-1000i flow cytometer to identify culture-negative urine specimens in patients with UTI.

Urinary tract infections (UTIs) are among the most common bacterial infections in men and urine culture is gold standard for diagnosis. Considering the high prevalence of culture-negative specimens, any method that identifies such specimens is of interest. The aim was to evaluate a new screening concept for flow cytometry analysis (FCA). The outcomes were evaluated against urine culture, uropathogen species and three conventional screening methods. A prospective, consecutive study examined 1,312 urine specimens, collected during January and February 2012. The specimens were analyzed using the Sysmex UF1000i FCA. Based on the FCA data culture negative specimens were identified in a new model by use of linear discriminant analysis (FCA-LDA). In total 1,312 patients were included. In- and outpatients represented 19.6% and 79.4%, respectively; 68.3% of the specimens originated from women. Of the 610 culture-positive specimens, Escherichia coli represented 64%, enterococci 8% and Klebsiella spp. 7%. Screening with FCA-LDA at 95% sensitivity identified 42% (552/1312) as culture negative specimens when UTI was defined according to European guidelines. The proposed screening method was either superior or similar in comparison to the three conventional screening methods. In conclusion, the proposed/suggested and new FCA-LDA screening method was superior or similar to three conventional screening methods. We recommend the proposed screening method to be used in clinic to exclude culture negative specimens, to reduce workload, costs and the turnaround time. In addition, the FCA data may add information that enhance handling and support diagnosis of patients with suspected UTI pending urine culture [corrected].

Introduction

Urinary tract infections (UTIs) are among the most common infections in community and hospitalized patients with more than 175 million UTI incidences worldwide [1]. UTI is caused by pathogenic microorganisms which induce infection and an inflammatory response with presence of leukocyturia and erythrocyturia. However, UTIs are often harmless and self-eradicated [2]. Nevertheless, UTIs are associated with high morbidity and costs and especially among patients with diabetes [3]. In the United States UTIs account annually for more than 7 million physician visits, more than 1 million emergency department visits and more than 100,000 hospitalizations. The estimated annual cost for treatment is calculated to more than US $1 billion and the indirect cost estimated to approximately $1.6 billion [4, 5]. Furthermore, 15% of all antibiotics prescribed to outpatients and data from some European countries also suggest a similar rate [6, 7].

In a large German study on patients with diabetes mellitus type 2 the cost was estimated to €316 per UTI event with a total increased cost of €3916 per patient in the UTI group compared with the non UTI group [3].

For diagnosis of UTI symptoms such as urgency, dysuria, frequent urination, back pain, leukocyturia and a positive nitrite test are considered reliable indicators [2, 8]. However, urine culture is the gold standard for diagnosis of UTI but is laborious and moderately costly, with a turnaround time of 24 to 48 h [9].

In clinical laboratories, urine specimens are among the most commonly encountered and approximately a quarter to more than half are considered culture negative [9, 10]. For this reason, any screening method that identifies and excludes urine specimens with non-significant bacteriuria would be of great interest [11-13].

Flow cytometry analysis (FCA) is a promising method to identify and enumerate bacteria, leukocytes, erythrocytes and other particles in urine. The second generation automated FCA instrument, the Sysmex UF-1000i (Medical Electronics, Kobe, Japan) has a separate detection channel for bacteria with improved sensitivity (SE) and specificity (SP) [11-15].

Using FCA, we have recently evaluated the inflammatory response of leukocytes and erythrocytes in urine for different pathogens in patients with suspected UTI [9]. Based on the dataset we now aimed to set up and assess a new screening model to identify and rule out culture negative urine specimens in patients with suspected UTI prior to culture. The prerequisites for the screening model are high SE to prevent specimens with significant bacteriuria (SBU) from being erroneously classified as negative (i.e. false negatives) and a relatively high SP to prevent unnecessary culturing. The present screening model was evaluated against three conventional screening methods [11, 16, 17].

Material and methods

Collection of urine specimens

A prospective consecutive multicenter study was conducted during January and February 2012 which analyzed urine specimens from in- and outpatients. The specimens were collected in non-preservative tubes, stored and transported at ≤6 °C to the Department of Clinical Microbiology at the University Hospital of Umeå for analysis. All specimens were from the county of Västerbotten, Sweden.

Urinalysis

All specimens underwent FCA and urine culture within 3 h of arrival to the laboratory. Specimens that arrived after 4 PM were analyzed with flow cell analysis (FCA) and stored at +6 °C until cultured the following morning. Excluded were specimens from pregnant women, urinary catheter and those that lacked complete FCA data.

Flow cytometry analysis was performed using the UF-1000i instrument (Sysmex, TOA Medical Electronics, Kobe, Japan), supported by the Sysmex software version of 00–15 [9]. The screening model was built on the observed bacterial and leukocyte counts.

Urine culture

Gram-negative and Gram-positive uropathogens were identified by Brilliance™ UTI agar (Oxoid Ltd., Basingstoke, UK) biochemical tests, and urine culture was performed as previously described [9].

Species identification

Isolates were identified in specimens with presence of ≥106 colony forming units/L (CFU/L) and those with mixed flora (with both gram negative and gram positive bacteria) with a dominating pathogen (i.e. bacterial count at least 10 times higher than any other species) [9].

Significant bacteriuria

SBU was defined in accordance with European guidelines (at ≥106 CFU/L of an uropathogen with acute uncomplicated cystitis) [7] and was the gold standard in the present article [18, 19]. In patients with <108 (≥106 and 107 CFU/L) and where the presence or absence of UTI symptoms could not be determined, the specimens were classified indeterminant.

Evaluations were also done when SBU was defined as ≥107 or ≥108 CFU/L of a uropathogen, respectively, irrespective of presence of UTI symptoms.

Ethical approval

All procedures were in accordance with institutional and national ethical standards and the Helsinki declaration.

Statistical analysis

Urine specimens analyzed with FCA and urine culture were used to derive a decision rule that based on FCA data determines which specimens should be cultured. Screening methods based on bacterial counts (BC) and white blood cell counts (WBC) have been suggested by Jolkkonen, Manoni and De Rosa [11, 16, 17]. Here, specimens are cultured if either BC > a or WBC > b, where a and b are the method specific cut-off values: Jolkonnen: a = 405, b = 16; Manoni: a = 125, b = 40; De Rosa: a = 170, b = 150.

We suggest the FCA-LDA approach that uses linear discriminant analysis (LDA) on the FCA variables BC and WBC [20]. The FCA-LDA rule was derived using data from specimens that were concluded to be significant bacteriuria or non-significant bacteriuria and with BC < 5000. A LDA-model was fitted to the log-transformed FCA-values, assuming homoscedasticity and proportional priors, which resulted in a decision rule: [ln (BC) > α + βln(WBC)]. The intercept α was tuned so that the rule had the desired sensitivity (SE). The decision rule is a line, where specimens above the line are cultured (Fig. 1).

Fig. 1

Plot of 1,312 urine specimens with significant, non-significant and indeterminant bacteriuria and the classification criteria or the different screening methods examined. Specimens above the classification line were predicted as positives and send to culture. P (blue color) = “positive specimens", i.e. specimens with significant bacteriuria when urinary tract infection was defined according to European guidelines. N (red colour) = “negative specimens", i.e. specimens with non-significant bacteriuria when urinary tract infection was defined according to European guidelines. Q (green colour) = “indeterminant/questionable” specimens, i.e. specimens with indeterminant significant bacteriuria when urinary tract infection was defined according to European guidelines (mainly specimens that lacked information regarding UTI symptoms), number of bacteria (Bact = Y-axis) and white blood cells (WBC = X- axis) estimated by flow cell instrument per uL. Red lines represent the cut off for linear discriminant analysis at: 98% sensitivity (FCA-LDA98), 95% sensitivity (FCA-LDA95) and 90% sensitivity (FCA-LDA90). Dark blue color represents the cut-off values according to De Rosa et al. (cut off: bacteria 170, white blood cells 150). Green color represents the cut-off values according to Manoni et al. (cut off: bacteria 125, white blood cells 40). Light blue color represents the cut-off values according to Jolkkonen et al. (cut off: bacteria 405, white blood cells 16)

The methods were evaluated in terms of their SE and specificity (SP), negative and positive predictive values (PPV and NPV), the number of cultured specimens and the relative cost (RC) compared to the standard procedure. RC was calculated under the assumption that the running cost of culture is five times as high as running FCA. The evaluations were performed for different definitions of UTI.

In a similar style as described above a LDA-model, based on BC and the variable B-FSC (the forward scatter / “particle size”), was derived to determine if the specimens’ bacteria were gram positive or gram negative. This resulted in a classification rule: [ln (BC) > α + β ln(B-FSC)] where specimens were predicted gram positive if the inequality was valid and gram-negative otherwise.

Results

In total, 1587 urine specimens were eligible for FCA and urine culture. Two hundred seventy-five specimens (17.3%) were excluded: 125 urinary catheters, 128 specimens from pregnant women, 18 that lacked clinical information and four specimens with candidauria. In total, 1312 specimens were enrolled in the present study (Fig. 1).

Among genders, 68.3% (p < 0.001) of the specimens originated from women with a mean age of 58.9 years (standard deviation 25.5) versus 31.7% and 61.7 years (20.8) for men. Outpatients represented 79.4% and inpatients 20.6%. The demographic information of the dataset is presented in Table 1.

Table 1

Demographic data of 1,312 patients and urine specimens enrolled in the study

Characteristic	Totala	Womena	Mena
Age (mean)	58.9	57.6	61.7
Outpatients (%)	79.4	82.4	72.3
Specimens included, n (%)	1312	896 (68)	416 (32)
Growth distribution at culture
> 108 colony forming units per liter	447 (34)	359 (80)	88 (20)
107−108 colony forming units per liter	370 (28)	271 (73)	99 (27)
106−107 colony forming units per liter	288 (22)	193 (67)	95 (33)
No growth	207 (16)	73 (35)	134 (65)
SBUb , n (%)	473	380 (80)	93 (20)
Gram-negative	405 (86)	337 (83)	68 (17)
Escherichia coli	337 (71)	288 (85)	49 (15)
Klebsiella spp.	29 (6)	24 (83)	5 (17)
Other	39 (8)	25 (64)	14 (36)
Gram-positive	68 (14)	43 (63)	26 (37)
Staphylococcus spp.	24 (5)	17 (71)	7 (29)
Enterococcus spp.	32 (7)	18 (56)	14 (44)
Other	12 (2)	8 (67)	4 (33)
Indeterminant SBUb, n (%)	98	68 (69)	30 (31)
Gram-negative	74 (75)	50 (69)	24 (31)
Escherichia coli	54 (55)	37 (69)	17 (31)
Klebsiella spp.	13 (13)	9 (69)	4 (31)
Other	7 (7)	4 (57)	3 (43)
Gram-positive	24 (25)	18 (76)	6 (24)
Staphylococcus spp.	2 (2)	1 (50)	1 (50)
Enterococcus spp.	15 (15)	10 (67)	5 (33)
Other	7 (7)	7 (100)	0 (0)
Non significant SBUb, n (%)	741	448 (60)	293 (40)
Gram-negative	10 (1)	7 (70)	3 (30)
Escherichia coli	2 (0)	1 (50)	1 (50)
Klebsiella spp.	3 (0)	2 (66)	1 (33)
Other	5 (1)	4 (80)	1 (20)
Gram-positive	29 (4)	24 (83)	5 (17)
Staphylococcus spp.	5 (1)	2 (40)	3 (60)
Enterococcus spp.	4 (1)	4 (100)	0 (0)
Other	20 (3)	18 (90)	2 (10)
Mixed flora Gram-negative	202 (27)	148 (73)	54 (27)
Mixed flora Gram-positive	293 (39)	196 (67)	97 (33)
No growth	207 (28)	73 (35)	134 (65)

aThe percentage (%) represents the figures within each sub-group. The columns Women and Men represent the relative distribution between the genders

bSignificant bacteriuria according to European guidelines (SBU), indeterminant according to European guidelines and non-significant bacteriuria according to European guidelines

When UTI was defined according to European guidelines, 36.0% (473/1312 specimens) had significant bacteriuria, 56.5% (741/1312 specimens) had non significant bacteriuria and 7.5% (98/1312 specimens) were classified as indeterminant, due to lack of clinical information.

E. coli was the most predominant uropathogen, representing 71.2% (337/473) of the uropathogens followed by Enterococcus faecalis 6.8%, Klebsiella pneumoniae 6.1% and coagulase-negative staphylococci 5.1% (Tables 1 and 2).

Table 2

Outcome of the four screening methods of 1,312 urine specimens examined with flow cytometry analysis when UTI was defined according to European guidelines

Groupa	Bacteria	SBU			Methods evaluatedb
					Jolkkonen			Manoni			De Rosa			FCA-LDA98			FCA-LDA95			FCA-LDA90
		Posc	Neg	Q	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC
EC	Escherichia coli	337	2	54	99	0	16	99	0	15	98	0	23	99	0	20	97	0	38	93	100	70
KC	Acinetobacter spp.	0	1	0		0	0		0	0		100	1		0	0		100	1		100	1
KC	Citrobacter spp.	9	0	0	100		0	100		0	100		0	100		0	100		0	89		1
KC	Citrobacter freundii	1	0	0	100		0	100		0	100		0	100		0	100		0	100		0
KC	Enterobacter spp.	2	1	2	100	0	1	100	0	1	100	0	1	100	0	1	100	0	1	100	0	1
KC	Other Klebsiella spp.	1	0	1	100		1	100		1	100		1	100		1	100		1	100		1
KC	Klebsiella oxytoca	8	1	1	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0	100	0	1
KC	Klebsiella pneumoniae	20	2	11	95	100	8	100	50	6	100	50	7	95	100	8	95	100	9	95	100	9
KC	Pantoea spp.	1	0	0	100		0	100		0	100		0	100		0	100		0	100		0
KC	Aggregated	42	5	15	98	40	10	100	20	8	100	40	10	98	40	10	98	60	12	95	60	14
PR	Morganella morganii	2	1	2	100	100	1	100	100	1	100	100	1	100	100	1	100	100	2	100	100	3
PR	Proteus mirabilis	9	1	2	100	0	0	100	0	0	100	100	1	89	0	1	89	100	2	89	100	3
PR	Proteus vulgaris	2	1	0	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0
PR	Providencia rettgeri	1	0	0	100		0	100		0	100		0	100		0	100		0	100		0
PR	Aggregated	14	3	4	100	33	1	100	33	1	100	67	2	93	33	2	93	67	4	93	67	6
PS	Pseudomonas aeruginosa	11	0	0	91		1	100		0	100		0	91		1	91		1	91		1
PS	Other Pseudomonas spp.	0	0	1			0			0			0			0			0			0
PS	Aggregated	11	0	1	91		1	100		0	100		0	91		1	91		1	91		1
ST	CoNS	13	4	0	100	0	0	100	25	1	100	25	1	100	25	1	100	25	1	85	25	3
ST	Staphylococcus aureus	8	1	1	88	0	1	100	0	0	100	0	0	100	0	0	75	0	2	63	100	5
ST	Staphylococcus saprophyticus	3	0	1	100		0	100		0	100		0	100		0	100		0	67		1
ST	Aggregated	24	5	2	96	0	1	100	20	1	100	20	1	100	20	1	92	20	3	75	40	9
EN	Enterococcus faecalis	28	4	10	93	0	6	96	0	4	93	0	6	96	0	5	82	25	11	71	75	17
EN	Enterococcus faecium	4	0	5	100		2	100		1	100		2	100		2	100		3	75		6
EN	Aggregated	32	4	15	94	0	8	97	0	5	94	0	8	97	0	7	84	25	14	72	75	23
SR	Alpha-hemolytic streptococci	3	6	0	67	17	2	67	17	2	67	17	2	67	17	2	67	17	2	67	33	3
SR	Gemella haemolysans	1	0	0	100		0	100		0	100		0	100		0	100		0	100		0
SR	Streptococcus Group A	0	0	1			0			0			0			0			0			1
SR	Streptococcus Group G	0	0	3			0			0			0			0			1			2
SR	Streptococcus Group C	1	0	3	100		0	100		0	100		0	100		0	100		0	0		2
SR	Aggregated	5	6	7	80	17	2	80	17	2	80	17	2	80	17	2	80	17	3	60	33	8
GB	Streptococcus agalactiae	6	14	0	83	36	6	100	29	4	100	36	5	100	50	7	67	64	11	50	79	14
OT	Diphtheroid rod	1	0	0	100		0	100		0	100		0	100		0	100		0	100		0
OT	Haemophilus influenzae	1	0	0	100		0	100		0	100		0	100		0	100		0	100		0
OT	Aggregated	2	0	0	100		0	100		0	100		0	100		0	100		0	100		0
NE	Culture negative	0	207	0		64	132		64	132		77	160		80	165		87	181		96	198
MP	Mixed flora G+	0	293	0		34	100		37	107		43	127		45	133		57	166		78	229
MN	Mixed flora G-	0	202	0		31	62		27	55		41	82		36	73		59	119		79	160
All	Total	473	741	98	98	41	339	99	41	330	98	51	420	98	52	421	95	65	552	90	83	732

SBU significant bacteriuria, SE sensitivity, SP specificity, NC not cultured, i.e. specimens screened as culture negative, spp species, CoNS coagulase negative staphylococci

aGroup = bacteria or bacterial groups. EC = E. coli; KC = Klebsiella/Citrobacter group; PR = Proteus group; PS = Pseudomonas group; ST = Staphylococcal group; SR = Streptococci group; GB = Streptococcus agalactiae (GBS); OT = other bacteria; NE = culture negative; MP = mixed Gram-positive flora; MN = mixed Gram-negative flora; All = total

bMethods evaluated: Jolkkonen et al., Manoni et al., De Rosa et al., linear discriminant analysis at 98, 95 and 90% sensitivity, respectively (FCA-LDA98, FCA-LDA95, FCA-LDA90)

cPos = SBU when urinary tract infection was defined according to European guidelines; Neg = non SBU according to European guidelines; Q = indeterminant (questionable) significant bacteriuria according to European guidelines

FCA-LDA decision rules, using BC and WBC as the predictors and controlling the SE at 90%, 95% and 98%, respectively, were derived, i.e.

FCA-LDA98: ln(BC) > 10.46–0.65 ln (WBC)

FCA-LDA95: ln(BC) > 8.22–0.65 ln (WBC)

FCA-LDA90: ln(BC) > 7.07–0.65 ln (WBC)

Recall that a specimen is cultured if the inequality is fulfilled. The FCA-LDA-rules were compared to the bivariate decision rules suggested by Jolkkonen, Manoni and De Rosa [11, 16, 17]. The six decision rules and their outcome are shown in Fig. 1.

Evaluation when SBU was defined according to European guidelines

Overall, when the FCA-LDA98 decision rule was used and UTI was defined according to European guidelines (at ≥106 CFU/L of an uropathogen with acute uncomplicated cystitis) [7] then the screen resulted in 98% SE (464 true positive specimens), 52% SP (385 true negative specimens) and 32% (421 specimens) were identified as culture negative (Fig. 2, total data). The corresponding numbers for the more restrictive inclusion rules FCA-LDA95 and FCA-LDA90 were (95%, 65%, and 42%) and (90%, 83%, and 56%), respectively (Fig. 2, Table 2). In comparison, the screening methods by Jolkkonen, Manoni and De Rosa had similar SE as for FCA-LDA98; however, the Jolkkonen’s and Manoni’s methods had lower SP (41%) compared to FCA-LDA98 (52%) and De Rosa (51%). The outcome of the six decision rules are presented in Table 2. In general, the screening methods had relative high SE for E. coli (0–3% above average SE for the examined methods) and Klebsiella/Citrobacter groups (KC-group 0–5% above average SE). Relative low SE was found for the enterococci group (EN group −1 to −18% below average SE) and the streptococci group (SR group −18 to −30% below average SE for the examined methods ; Table 2).

Fig. 2

Outcome of the 1312 urine specimens examined by flow cytometry analysis and samples sent to culture after screening using linear discriminant  analysis at 90, 95 and 98% sensitivity when significant bacteriuria was defined according to European guidelines. Blue arrows: represent the outcome of linear discriminant analysis at 98, 95 and 90% sensitivity, respectively (FCA-LDA98, FCA-LDA95, FCA-LDA90) when urinary tract infection was defined according to European guidelines. G+ = gram positive species. G– = gram negative species. 1SBU = significant bacteriuria, blue-colored boxes; Q-SBU = indeterminant (questionable) SBU, green-colored boxes; Non SBU, red-colored boxes. 2Number of specimens send to culture when FCA-LDA selection was applied. 3 Specimens send to culture was classified as Gram negative or Gram positive bacteria

Screening at SBU defined according to European guidelines yielded higher SE and SP than UTI defined at ≥107 and ≥108 /CFU/L (Table 3).

Table 3

Outcome of the four screening methods to rule out culture negative urine specimens among patients with suspected UTI

SBUa	Measures (%)	Methods evaluatedb
		Jolkkonen	Manoni	De Rosa	FCA-LDA98	FCA-LDA95	FCA-LDA90
European Guidelines	SE	98	99	98	98	95	90
	SP	41	41	51	52	65	83
	PPV	51	52	56	56	63	77
	NPV	96	98	98	98	95	93
	NCc	26	25	32	32	42	56
	RCd	94	95	88	88	78	64
≥107 CFU/L	SE	94	96	94	94	90	81
	SP	42	42	53	52	67	84
	PPV	56	56	61	61	68	80
	NPV	91	93	92	92	89	85
≥108 CFU/L	SE	99	100	100	99	98	95
	SP	38	38	48	48	62	81
	PPV	44	44	49	49	56	71
	NPV	99	100	100	99	99	97

SE sensitivity, SP specificity, PPV positive predictive value, NPV negative predictive value

aSBU = Definition of significant bacteriuria according to European guidelines, ≥107 and ≥108 colony forming units per liter in urine at culture

bComparing screening methods according to Jolkkonen et al., Manoni et al., De Rosa et al. Flow cell analysis-linear discriminant analysis at 98% sensitivity (FCA-LDA98), FCA-LDA95 = at 95% sensitivity, and FCA-LDA90 = at 90% sensitivity

cNC = not cultured (%): specimens identified as culture negative by the screening method

dRC = relative cost of the screening method (%) compared to the gold standard procedure (100%) when all specimens are cultured. A calculated cost of, e.g. 78%, represents a 22% cost reduction when FCA screening was included in its cost

At FCA-LDA98, FCA- LDA95 and FCA- LDA90 screening 9, 23 (9 + 14) and 47 (23 + 24) specimens proved false negative, respectively, representing 0.7, 1.7 and 3.6% of the specimens. Information related to the false negative specimens is presented in Table 4. For the FCA-LDA95 screen, seven of the false negative specimens had 108 CFU/L, eight 106 and 108 CFU/L each. Patients 2, 3, 7, 9, 10, 16 had all low WBC- and elevated bacterial counts indicating asymptomatic bacteriuria in these patients. Patient 8 had “high” WBC and red blood cell (RBC) counts but low bacterial counts despite presence of Proteus mirabilis at 108 CFU/L at culture. Of the 47 false negative specimens, 72% (34/47) originated from women.

Table 4

Data associated with the 47 false negative urine specimens present in linear discriminatory analysis at 98%, 95% and 90% sensitivity

Sensitivity (%)	Patient	Gender	Patient age (years)	Unit	Specimen	10n CFU/La	Bacteriab	Nitrit-testc	UTId Symptom	Comments / clinical information	RBCe	WBCf	Bacterial countsg
0.98	1	M	32	HCC	IC	7	α-hemolytic strm			IC	2	5	49
0.98	2	F	6	HCC	MSU	7	Escherichia coli		P	BSU	5	6	127
0.98	3	M	75	Hosp.	IC	6	Escherichia coli				10	4	101
0.98	4	M	73	HCC	MSU	6	Escherichia coli		P	Increased CRP, BSU	4	6	69
0.98	5	F	80	HCC	MSU	6	Escherichia coli			500 WBCh	22	18	122
0.98	6	M	91	HCC	MSU	6	Ent. faecalis		P	Urgency	4	5	38
0.98	7	F	78	HCC	MSU	8	Klebsiella pneum.				1	7	224
0.98	8	F	23	Hosp.	MSU	8	Proteus mirabilis			Smelly urine, increased WBCh	65	170	9
0.98	9	F	29	HCC	MSU	8M7	Ps. aeruginosa			Ante-natal clinic	36	8	172
0.95	10	F	54	HCC	MSU	7	Escherichia coli				37	11	343
0.95	11	M	60	HCC	MSU	7	Escherichia coli		P	Urgency	74	69	76
0.95	12	F	5	HCC	MSU	6	Escherichia coli		P	Cystit symptom	18	423	62
0.95	13	F	36	HCC	MSU	6	Escherichia coli	+			4	81	131
0.95	14	F	68	HCC	MSU	6	Escherichia coli		P	Urgency, 1 h BID	20	244	33
0.95	15	F	75	HCC	MSU	6	Escherichia coli		P	Urgency	7	334	40
0.95	16	M	69	Hosp.	MSU	8	Ent. faecalis				8	8	290
0.95	17	M	79	HCC	MSU	8M6	Ent. faecalis				110	98	179
0.95	18	F	89	Hosp.	MSU	7M6	Ent. faecalis		P	Urgency	12	65	152
0.95	19	F	6	HCC	MSU	7M6	Ent. faecalis		P	Frequency	5	20	357
0.95	20	F	79	Hosp.	MSU	8	Str. agalactiae		P	BSU, urgency	18	14	364
0.95	21	F	50	HCC	MSU	8	Str. agalactiae				10	32	220
0.95	22	F	89	Hosp.	MSU	7	S. aureus			100 WBCh	6	11	236
0.95	23	F	37	HCC	MSU	7	S. aureus		P	Urgency	7	29	414
0.90	24	F	29	HCC	MSU	6	Citrobacter spp.				4	144	348
0.90	25	M	70	Hosp.	IC	7	CoNS			Fever	108	79	1085
0.90	26	M	79	HCC	IC	7	CoNS			UTI symptom. Neurogenic urine bladder disease	70	2495	82
0.90	27	M	20	HCC	MSU	8M7	Escherichia coli		P	BSU	4	210	467
0.90	28	F	26	HCC	MSU	8	Escherichia coli			500 WBCh	4	190	234
0.90	29	M	30	HCC	MSU	8	Escherichia coli				1	89	471
0.90	30	F	36	HCC	MSU	8	Escherichia coli	−	P	Urgency,	26	55	1334
0.90	31	F	47	HCC	MSU	8	Escherichia coli		P	Urgency, BSU	2	10	7478
0.90	32	F	47	HCC	MSU	8	Escherichia coli				87	28	2122
0.90	33	F	74	HCC	MSU	8M6	Escherichia coli	+			5	20	3684
0.90	34	F	92	HCC	MSU	8	Escherichia coli			UTI?	26	122	1382
0.90	35	F	35	HCC	MSU	7M6	Escherichia coli			500WBCh, 25RBCh	43	893	101
0.90	36	F	48	HCC	MSU	7	Escherichia coli			500 WBCh, 50 RBCh	4	288	785
0.90	37	F	70	HCC	MSU	7	Escherichia coli	+			8	139	186
0.90	38	F	33	HCC	MSU	6	Escherichia coli		P	500 WBCh, 200 RBCh	14	1113	151
0.90	39	F	71	HCC	MSU	6	Escherichia coli		P	BID 4 h, BSU, Urgency	6	52	284
0.90	40	M	62	HCC	MSU	8	Ent. faecalis			Urgency	2	6	1913
0.90	41	M	72	HCC	MSU	8	Ent. faecalis			Fever	17	25	981
0.90	42	F	73	HCC	MSU	8M7	Ent. faecalis		P	Urgency	12	35	2449
0.90	43	F	56	Hosp.	MSU	8M6	Ent. faecium			Neutropenia	78	12	1116
0.90	44	F	82	HCC	MSU	8M6	Str. agalactiae		P	Urgency	42	49	2339
0.90	45	F	70	Hosp.	IC	8	S. aureus			Infection?	40	6	2440
0.90	46	F	26	HCC	MSU	8M6	S. saprophyticus		P		9	155	1119
0.90	47	F	60	HCC	MSU	8	Streptococci				14	273	410

M male, F female, HCC heath care centre, Hosp hospital, MSU mid stream urine, IC intermittent catheterized, α-hemolytic str alpha haemolytic streptococci, BSU burning sensation during urination, CoNS coagulase negative staphylococci

aM6, M7 represents mixed flora at 106, 107 / colony forming units/L, respectively 10n the numbers 6, 7, 8 represents 106, 107 and 108 colony forming units/L, respective of bacteria quantified at urine culture. M = mixed flora, i.e. 8M7 = specimens with 108 colony forming units/L of the dominating uropathogen and mixed flora 106 colony forming units/L.

bBacteria/uropathogen identified at culture

c + = positive test, − = negative test

dUTI (urinary tract symptoms): + P = present

eRBC = red blood cells counted in flow cytometry analysis × 106/L

fWBC = white blood cells counted in flow cytometry analysis × 106/L

gBacterial counts estimated in the flow cytometry analysis ×106/LhWBC = white blood cells, RBC = red blood cells estimated at clinic/health care centre

Evaluation when SBU was defined at ≥107 and ≥108 CFU/L

Overall, defining UTI at ≥107 CFU/L of an uropathogen, irrespective of presence of UTI symptoms, resulted in lower sensitivities (3–9%) and similar specificities in comparison when SBU was defined according to European guidelines (Tables 3 and 5). Defining UTI as ≥108 CFU/L (irrespective of presence of UTI symptoms) resulted in higher sensitivities (1–5%) and lower specificities (2–4%) in comparison to European guidelines (Tables 3 and 6).

Table 5

Outcome of the four screening methods of 1,312 urine specimens examined with flow cytometry analysis when UTI was defined as growth of ≥107 colony forming units per liter at culture

Groupa	Bacteria	SBU		Methods evaluatedb
				Jolkkonen			Manoni			De Rosa			FCA-LDA98			FCA-LDA95			FCA-LDA90
		Posc	Neg	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC
EC	Escherichia coli	374	19	97	26	16	98	32	15	96	47	23	97	37	20	94	84	39	86	100	70
KC	Acinetobacter spp.	0	1		0	0		0	0		100	1		0	0		100	1		100	1
KC	Citrobacter spp.	8	1	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0	100	100	1
KC	Citrobacter freundii	1	0	100		0	100		0	100		0	100		0	100		0	100		0
KC	Enterobacter spp.	4	1	100	100	1	100	100	1	100	100	1	100	100	1	100	100	1	100	100	1
KC	Other Klebsiella	2	0	50		1	50		1	50		1	50		1	50		1	50		1
KC	Klebsiella oxytoca	10	0	100		0	100		0	100		0	100		0	100		0	90		1
KC	Klebsiella pneumoniae	29	4	86	100	8	90	75	6	86	75	7	86	100	8	83	100	9	83	100	9
KC	Pantoea spp.	1	0	100		0	100		0	100		0	100		0	100		0	100		0
KC	Aggregated	55	7	91	71	10	93	57	8	91	71	10	91	71	10	89	86	12	87	100	14
PR	Morganella morganii	4	1	100	100	1	100	100	1	100	100	1	100	100	1	75	100	2	50	100	3
PR	Proteus mirabilis	11	1	100	0	0	100	0	0	100	100	1	91	0	1	91	100	2	82	100	3
PR	Proteus vulgaris	3	0	100		0	100		0	100		0	100		0	100		0	100		0
PR	Providencia rettgeri	1	0	100		0	100		0	100		0	100		0	100		0	100		0
PR	Aggregated	19	2	100	50	1	100	50	1	100	100	2	95	50	2	89	100	4	79	100	6
PS	Pseudomonas aeruginosa	11	0	91		1	100		0	100		0	91		1	91		1	91		1
PS	Other Pseudomonas	1	0	100		0	100		0	100		0	100		0	100		0	100		0
PS	Aggregated	12	0	92		1	100		0	100		0	92		1	92		1	92		1
ST	CoNS	17	0	100		0	94		1	94		1	94		1	94		1	82		3
ST	Staphylococcus aureus	9	1	89	0	1	100	0	0	100	0	0	100	0	0	89	0	1	56	100	5
ST	Staph. saprophyticus	4	0	100		0	100		0	100		0	100		0	100		0	75		1
ST	Aggregated	30	1	97	0	1	97	0	1	97	0	1	97	0	1	93	0	2	73	100	9
EN	Enterococcus faecalis	38	4	87	25	6	92	25	4	87	25	6	89	25	5	76	50	11	63	75	17
EN	Enterococcus faecium	9	0	78		2	89		1	78		2	78		2	67		3	33		6
EN	Aggregated	47	4	85	25	8	91	25	5	85	25	8	87	25	7	74	50	14	57	75	23
SR	Alpha-hemolytic streptococci	9	0	78		2	78		2	78		2	78		2	78		2	67		3
SR	Gemella haemolysans	1	0	100		0	100		0	100		0	100		0	100		0	100		0
SR	Streptococcus Group A	1	0	100		0	100		0	100		0	100		0	100		0	0		1
SR	Streptococcus Group G	3	0	100		0	100		0	100		0	100		0	67		1	33		2
SR	Streptococcus Group C	4	0	100		0	100		0	100		0	100		0	100		0	50		2
SR	Aggregated	18	0	89		2	89		2	89		2	89		2	83		3	56		8
GB	Streptococcus agalactiae	18	2	72	50	6	83	50	4	78	50	5	67	50	7	44	50	11	28	50	14
OT	Diphtheroid rod	1	0	100		0	100		0	100		0	100		0	100		0	100		0
OT	Haemophilus influenzae	1	0	100		0	100		0	100		0	100		0	100		0	100		0
OT	Aggregated	2	0	100		0	100		0	100		0	100		0	100		0	100		0
NE	Culture negative/no	0	207		64	132		64	132		77	160		80	165		88	181		96	198
MP	Mixed flora G+	0	293		34	100		36	107		43	127		45	133		57	166		78	229
MN	Mixed flora G-	2	200	100	31	62	100	28	55	100	41	82	100	37	73	100	59	119	100	80	160
All	Total	577	735	94	42	339	96	42	330	94	53	420	94	53	421	90	67	552	81	84	732

SBU significant bacteriuria, SE sensitivity, SP specificity, NC not cultured (i.e. specimens screened as culture negative), spp. species, CoNS coagulase negative staphylococci

aGroup = bacteria or bacterial groups. EC = E. coli; KC = Klebsiella/Citrobacter group; PR = Proteus group; PS = Pseudomonas group; ST = Staphylococcal group; SR = Streptococci group; GB = Streptococcus agalactiae (GBS); OT = other bacteria; NE = culture negative; MP = Mixed grampositive flora; MN = mixed gramnegative flora; All = total

bMethods evaluated: Jolkkonen et al., Manoni et al., De Rosa et al., Flow cell analysis-linear discriminant analysis at 98% sensitivity (FCA-LDA98), FCA-LDA95 = at 95% sensitivity, FCA-LDA90 = at 90% sensitivity

cPos = SBU when urinary tract infection was defined as growth of ≥107 colony forming units per liter at culture

Table 6

Outcome of the four screening methods of 1,312 urine specimens examined with flow cytometry analysis when UTI was defined as growth of ≥108 colony forming units per liter in urine at culture

Groupa	Bacteria	SBU		Methods evaluatedb
				Jolkkonen			Manoni			De Rosa			FCA-LDA98			FCA-LDA95			FCA-LDA90
		Posc	Neg	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC	SE	SP	NC
EC	Escherichia coli	311	82	100	20	16	100	18	15	100	28	23	100	24	20	100	48	39	98	77	70
KC	Acinetobacter spp.	0	1		0	0		0	0		100	1		0	0		100	1		100	1
KC	Citrobacter spp.	7	2	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0	100	50	1
KC	Citrobacter freundii	1	0	100		0	100		0	100		0	100		0	100		0	100		0
KC	Enterobacter spp.	2	3	100	33	1	100	33	1	100	33	1	100	33	1	100	33	1	100	33	1
KC	Other Klebsiella	1	1	100	100	1	100	100	1	100	100	1	100	100	1	100	100	1	100	100	1
KC	Klebsiella oxytoca	9	1	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0	100	100	1
KC	Klebsiella pneumoniae	20	13	95	54	8	100	46	6	100	51	7	95	54	8	95	62	9	95	62	9
KC	Pantoea spp.	1	0	100		0	100		0	100		0	100		0	100		0	100		0
KC	Aggregated	41	21	98	43	10	100	38	8	100	48	10	98	43	10	98	52	12	98	62	14
PR	Morganella morganii	2	3	100	33	1	100	33	1	100	33	1	100	33	1	100	67	2	100	100	3
PR	Proteus mirabilis	7	5	100	0	0	100	0	0	100	20	1	86	0	1	86	20	2	86	40	3
PR	Proteus vulgaris	3	0	100		0	100		0	100		0	100		0	100		0	100		0
PR	Providencia rettgeri	1	0	100		0	100		0	100		0	100		0	100		0	100		0
PR	Aggregated	13	8	100	13	1	100	13	1	100	25	2	92	13	2	92	38	4	92	63	6
PS	Pseudomonas aeruginosa	11	0	91		1	100		0	100		0	91		1	91		1	91		1
PS	Other Pseudomonas spp.	0	1		0	0		0	0		0	0		0	0		0	0		0	0
PS	Aggregated	11	1	91	0	1	100	0	0	100	0	0	91	0	1	91	0	1	91	0	1
ST	CoNS	10	7	100	0	0	100	14	1	100	14	1	100	14	1	100	14	1	100	43	3
ST	Staphylococcus aureus	6	4	100	25	1	100	0	0	100	0	0	100	0	0	100	25	1	83	100	5
ST	Staphylococcus saprophyticus	3	1	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0	67	0	1
ST	Aggregated	19	12	100	8	1	100	8	1	100	8	1	100	8	1	100	17	2	89	58	9
EN	Enterococcus faecalis	23	19	96	26	6	100	21	4	100	32	6	100	26	5	91	47	11	78	63	17
EN	Enterococcus faecium	4	5	100	40	2	100	20	1	100	40	2	100	40	2	100	60	3	75	100	6
EN	Aggregated	27	24	96	29	8	100	21	5	100	33	8	100	29	7	93	50	14	78	71	23
SR	Alpha-hemolytic streptococci	2	7	100	29	2	100	29	2	100	29	2	100	29	2	100	29	2	100	43	3
SR	Gemella haemolysans	1	0	100		0	100		0	100		0	100		0	100		0	100		0
SR	Streptococcus Group A	0	1		0	0		0	0		0	0		0	0		0	0		100	1
SR	Streptococcus Group G	0	3		0	0		0	0		0	0		0	0		33	1		67	2
SR	Streptococcus Group C	1	3	100	0	0	100	0	0	100	0	0	100	0	0	100	0	0	0	33	2
SR	Aggregated	4	14	100	14	2	100	14	2	100	14	2	100	14	2	100	21	3	75	50	8
GB	Streptococcus agalactiae	6	14	83	36	6	100	29	4	100	36	5	100	50	7	67	64	11	50	79	14
OT	Diphtheroid rod	1	0	100		0	100		0	100		0	100		0	100		0	100		0
OT	Haemophilus influenzae	1	0	100		0	100		0	100		0	100		0	100		0	100		0
OT	Aggregated	2	0	100		0	100		0	100		0	100		0	100		0	100		0
NE	Culture negative/no	0	207		64	132		64	132		77	160		80	165		87	181		96	198
MP	Mixed flora Gram+	0	293		34	100		36	107		43	127		45	133		57	166		78	229
MN	Mixed flora Gram-	2	200	100	31	62	100	28	55	100	41	82	100	37	73	100	59	119	100	80	160
All	Total	436	876	99	38	339	100	38	330	100	48	420	99	48	421	98	62	552	95	81	732

SBU significant bacteriuria, SE sensitivity, SP specificity, NC not cultured (i.e. specimens screened as culture negative), spp. species, CoNS coagulase negative staphylococci

aGroup = bacteria or bacterial groups. EC = E. coli; KC = Klebsiella/Citrobacter group; PR = Proteus group; PS = Pseudomonas group; ST = Staphylococcal group; SR = Streptococci group; GB = Streptococcus agalactiae (GBS); OT = other bacteria; NE = culture negative; MP = Mixed gram positive flora; MN = mixed gramnegative flora; All = total

bMethods evaluated: Jolkkonen et al., Manoni et al., De Rosa et al., Flow cell analysis-linear discriminant analysis at 98% sensitivity (FCA-LDA98), FCA-LDA95 = at 95% sensitivity, FCA-LDA90 = at 90% sensitivity

cPos = SBU when urinary tract infection was defined as growth of ≥107 colony forming units per liter at culture

Also at these screening methods SE had great impact on the proportion of specimens identified as culture negative (26–56%; Tables 2, 3, 5, and 6). For all methods a higher SE was observed for E. coli (0–5% above average SE), and lower SE for the enterococci (0 to −18% below average SE) and streptococci groups (0 to −25%) (Tables 2, 5, and 6).

Prediction of gram groups

Bacteria were predicted to their gram group according to their FCA results such that ln (BC) > −8.49 + 3.82 ln (B-FSC) were classified to be gram-positive otherwise to be gram-negative. In specimens sent to culture, 74–80% of the bacteria were correctly classified to their gram group (Fig. 2).

Calculation of the relative cost in a model including FCA screening found a 5–36% cost reduction compared with non-selective culture depending on the screening method used (Table 3). Implementation of the FCA-LDA95 estimated a 22% cost reduction when UTI was defined according to European guidelines (Table 3).

Discussion

UTIs are a common infection and one of the most commonly analyzed specimens in clinical microbiological laboratories. The aim of the present study was to present and evaluate a new model for identification of culture negative urine specimens identified by FCA.

The basis of the new screening model was to establish the cut off for bacterial and leukocyte counts based on LDA which differs from conventional screening models based on fixed cut off’s for bacteria- and WBC counts (Fig. 1) [11–13, 15–17, 20–22]. The new model was evaluated with respect to: SE, definitions of significant bacteriuria, different uropathogens, group of uropathogens, mixed flora and culture negative specimens and three conventional screening methods [12, 16, 17]. At evaluation the new LDA method proved superior to the Jolkkonen and Manoni methods but similar to that described by De Rosa et al. [17]. The prerequisite for the presented screening method was high SE since diagnosis was later confirmed by urine culture. Since the outcome for SBU defined according to European guidelines was superior compared with 107 CFU/L, irrespective of UTI symptoms. For this reason is the European guidelines recommended.

Currently, the instrument is mainly used for screening to identify and rule out culture negative specimens. However, screening can be done at different levels of SE, e.g. at lower SE for non hospitalized patients as UTIs are often harmless and self eradicating [2] or at higher SE (or omitted) in high risk patients in intensive care, immunosuppressed patients, pregnancy and those with pyelonephritis. In these patients, urine culture is recommended irrespective of the outcome of the FCA screening to avoid missed diagnosis in those deemed high risk. Since Streptococcus agalactiae is a potential threat to the unborn child and mother, we always recommend urine culture at pregnancy to identify colonization at low colony counts (<106 CFU/L).

Overall, we found that FCA-LDA95 provided a good balance between high SE and specimens excluded as culture negative (> 40%). At our laboratory, nearly 12,600 (of 30,000) specimens could be excluded from culture providing a significant reduction in workload, costs and turnaround time. The predicted cost reduction in the present model (5–36%) is in accordance with others [23]. In addition, FCA analysis adds important information to clinicians in identifying or rejecting patients for antimicrobial treatment. For those treated, the microorganisms Gram group adds further information for treatment success.

The microorganism’s Gram stain identity was predicted by use of B-FSC [24, 25] with an accuracy of 76% (74–80%) in the present study. Rod-shaped bacteria are predicted at higher accuracy than for cocci, >90% and 29%, respectively [26]. Use of NaOH-sodium dodecyl sulfate (SDS) has been proven to facilitate discrimination between Gram positive and Gram negative bacteria [27].

Excluded specimens (urinary catheters and those from pregnant women) were re-examined with a similar outcome as those included (data not shown). LDA screening seems therefore also to be useful for these specimens but further studies are warranted.

Of Among the false negative specimens patient 8 had “high” WBC counts and low bacterial counts despite presence of Proteus mirabilis at 108 CFU/L at culture. We speculate that the bacterial FCA-staining may have failed in this specimen.

The strength of the present study is the new screening approach, the large cohort and the comprehensive evaluations and against three conventional screening methods [11, 16, 17]. Also the outcomes are presented for different uropathogens, groups of bacteria, negative cultures and pinpoint the strength and weakness in the different methods as the impact of the definition of SBU.

Similar evaluations are, to our knowledge, not reported for SBU defined according to the European definition (at ≥106 CFU/L of an uropathogen with acute uncomplicated cystitis) [7]. In addition we also evaluated the outcome at ≥107 and ≥108 CFU/L irrespective of presence of UTI symptoms.

The distribution of uropathogens was in accordance with other studies [20, 28, 29] and sampling was carried out to minimize the destruction of RBC and WBC [30]. Also, the sampling approach possessed a very low risk for contamination with improved specimen quality.

A weakness is the cohort’s population structure with respect to gender, age, in−/outpatients, type of specimens, etc., which differs from other studies and explains the differences in outcome [11–13, 15–17, 19–22, 31]. However, the pathophysiology of UTI is expected to be quite similar despite some differences between species [9] and cohorts. However, the relatively large differences in outcome between studies is probably due to different cut off’s rather than differences between cohorts, which is supported in the present study where similar outcome was found for the present and De Rosa’s screening model. Also at an optimal cut off we expect a similar outcome, unless one method is superior. If so, a universal cut off can be implicated which significantly enhances implementation of the FCA screening method in clinic, but further studies are warranted.

Also, the delay from initiation until publication is a weakness but the distribution of uropathogens causing UTI is quite conserved over decades [2, 20, 28, 29, 32].

The present cohort may not reflect urine specimens from an average population but rather a selected cohort of specimens from patients with suspected UTI, pyelonephritis and failure or control after treatment. However, the specimens are representative of the urine samples analyzed at our laboratory.

Flow cytometry is a promising method to identify and rule out culture negative urine specimens prior to culture [11–13, 15–17, 19–22, 31]. The present screening model can be recommended in clinic for pre-screening of UTI specimens prior to culture to improve laboratory management of UTI specimens and service to clinicians. By an early identification of culture negative specimens, patients are excluded for unnecessary antimicrobial treatment and its associated side effects.

In summary, a new screening concept is presented based on LDA of FCA data that excludes 42% as culture negative urine specimens prior to culture when UTI was defined according to European guidelines. The present LDA screening method was superior or similar to three conventional methods and is recommended for clinical use to reduce workloads, costs, turnaround time and time to diagnosis.