Evaluation of the draft guidelines proposed by EMA and FDA for the clinical diagnosis of acute uncomplicated cystitis in women

Despite numerous publications, there is still no generally accepted strategy regarding the clinical diagnosis of acute uncomplicated cystitis (AC). The updated guidelines of the Infectious Diseases Society of America (IDSA) and the European Society for Microbiology and Infectious Diseases (ESCMID) mainly consist of recommendations about the treatment of AC and not the diagnosis [1]. These guidelines were limited to the treatment of AC and pyelonephritis in premenopausal, non-pregnant women with no known urological abnormalities or comorbidities. In addition, the authors noted that postmenopausal women or those who have well-controlled diabetes mellitus in the absence of urological sequelae may be considered as having uncomplicated UTIs (uUTIs) by some experts, but a discussion of specific management of these groups was outside the scope of the guidelines.

In the last update of these guidelines of the European Association of Urology (EAU) from 2019, AC is defined as acute, sporadic or recurrent cystitis limited to non-pregnant women with no known relevant anatomical and functional abnormalities within the urinary tract or comorbidities [2]. According to the EAU guidelines, the diagnosis of AC can be made with a high probability based on a focused history of lower urinary tract symptoms (dysuria, frequency and urgency) and the absence of vaginal discharge or irritation.

The definition of UTIs in a broader sense is presented in the updated German National Clinical Practice S3 Guideline [3]: UTIs may be classified as uncomplicated in the absence of relevant functional or anatomical abnormalities in the urinary tract, with no relevant renal functional impairment and any relevant concomitant disease that could aggravate the UTIs or condition, which could increase the risk of development of serious complications. Simple cystitis in this regard, may represent no additional health problem for the woman with stable diabetes mellitus, whereas any kind of pyelonephritis, whether earlier defined as uncomplicated or complicated, could interfere with her metabolic balance and could lead to severe complications. It becomes obvious today that a simple general classification of UTIs into uncomplicated and complicated UTIs is far too rough. Therefore, a more differentiated stratification of UTIs with the deeper consideration of risk factors was proposed earlier [4].

Recently, the US Food and Drug Administration (FDA) and European Medicines Agency (EMA) have proposed “draft” guidelines for the clinical diagnosis of patients with AC for further discussion:

We aimed to reassess the diagnostic values of these proposed draft guidelines using the Acute Cystitis Symptom Score (ACSS) which was validated in several languages [7‐10].

On the access date, the e-USQOLAT database contained information about 911 female respondents from seven different countries (Fig. 1). Respondents are allocated to the groups of patients (with AC) and controls (without AC) according to the final diagnosis of the treating physician.

A total number of 517 respondents from four countries matched all the inclusion criteria and could be selected for further data processing and analysis (Fig. 1). Missing results of urinalysis accounted for the majority of mismatches in the inclusion criteria (360; 39.52% of total). Only 39 of excluded respondents had no sufficient questionnaire data (4.28% of total).

The age of the population included in the study ranged from 15 to 87 years with the following averages: median (interquartile range—IQR) − 30.50 (24.00; 40.00), mean ± SD − 34.38 ± 13.71. The group of controls consisted of 232 (44.87%) respondents with a median age (IQR) − 31.00 (25.00; 40.00), a mean age ± SD - 33.94 ± 12.25, ranging from 15 to 73 years. Two hundred eighty-five (55.13%) respondents in the group of patients had a median age (IQR) of 30.00 (24.00; 41.00), a mean age ± SD − 34.74 ± 14.80, ranging from 18 to 87 years old. The process of selection of the study population and essential demographic data are presented in Fig. 1 and Table 1.

Linear model fit analysis for “diagnostically significant grades” of pyuria revealed values of ≥ 25 WBC/µL for dipstick analysis and > 8000 WBC/mL for urine microscopy according to Nechiporenko [13] to have a statistically significant positive relationship with the diagnosis of AC: sensitivity − 0.85 [95% CI = 0.80; 0.89], specificity − 0.72 [0.66; 0.78], PPV − 0.79 [0.74;0.84], NPV − 0.80 [0.74; 0.85], crude DOR − 14.77 [9.57; 22.80], Youden index − 0.57 [0.46; 0.67].

The median number of positive symptoms for controls was 1 with IQR of 0–3 and differed significantly non-significant (p < 0.001) from that for patients, which was 5 with IQR of 4–6 (Fig. 2).

According to the ACSS data, the most common symptom among the entire study population was urinary frequency (72.92%). It included 47.84% of controls and 93.33% of patients. Whereas the majority of controls experienced “mild” urinary frequency (81/111 = 72.97%), “moderate” or “severe” values of the symptom were more “specific” for the group of patients (189/266 = 71.05%) (Table 2).

Figures 3 and 4, respectively, represent the prevalence, DOR and Youden’s index of the six “typical” symptoms and their severity, used in the ACSS questionnaire. All six symptoms had a significant positive association with a positive outcome (PO), i.e. diagnosis of AC. It also was verified that not only the presence of the symptoms but also their severity is important for the diagnosis (Fig. 4). More detailed results of the analysis of different diagnostic values of these symptoms and their severity are given in Supplementary Tables 1, 2, 3.

Scoring the symptoms into 0 (no symptom), 1 (mild), 2 (moderate), and 3 (severe) revealed for controls a median symptom score of 1 with IQR of 0–4 which significantly differed from that for patients: 10 with IQR of 7–13 (p < 0.001) (Fig. 5).

ROC curve analysis revealed the largest area under the curve (AUC) for the summary score of the “typical” domain of the ACSS (AUC [95% CI] = 0.93 [0.91; 0.95]), in descending order followed by dysuria (0.85 [0.82; 0.88]), urination urgency (0.85 [0.82; 0.88]), sense of incomplete bladder emptying (0.79 [0.75; 0.83]), suprapubic pain (0.74 [0.70; 0.78]), and visible blood in urine (0.63 [0.60; 0.67]) (Fig. 6).

Sensitivity and specificity (average [95% CI]) for the different proposed approaches of diagnosing AC are the following:

If the cut-off value of the ACSS is combined with positive pyuria, then the specificity and sensitivity change to 0.96 [0.93; 0.98] and 0.73 [0.67; 0.78], respectively.

Pyuria by itself had a reasonable sensitivity (0.85 [0.80; 0.89]) and specificity (0.72 [0.66; 0.78]) (Suppl. Table 2).

Diagnostic values of different numbers and scores of symptoms with or without considering pyuria are presented in Supplementary Tables 2 and 3. Graphical representation of the different diagnostic proposals by FDA, EMA, and ACSS is given as Supplementary Figs. 1, 2, 3, 4, 5, 6, 7, 8, 9.

Urinary tract infections (UTIs) are among the most widespread infectious diseases in general practice [17], with 80% of cases classified as uUTIs. Although current guidelines recommend antibiotics as the first choice of treatment for the acute phase [2, 18], several prospective randomized, placebo-controlled studies comparing antibiotic and non-antimicrobial symptomatic therapeutic modalities have been performed [19‐22]. Results of these studies were compelling enough for the updated German Clinical Guidelines [18] to encourage the use of the non-AB symptomatic treatment in selected cases of acute lower uUTIs with mild-to-moderate symptoms.

Since AC can be considered a benign infection without general risk of aggravation of UTI or serious complications, mainly the clinical diagnosis with or without point of care urinalysis (such as pyuria) and longer term follow-up with patient-reported clinical outcome (e.g. for at least 4 weeks after end of treatment) should become the main inclusion and outcome criteria of future studies. This would also better correspond to the general recommendations and everyday practice, making urine culture unnecessary, with the exception of specific situations, such as (a) suspected pyelonephritis, (b) symptoms not resolving within about 1 week or recurring within 4 weeks after the completion of treatment; (c) atypical symptoms; (d) pregnancy [2].

Urine culture before and probably after treatment will remain important for epidemiological studies, and for studies including at least one antimicrobial therapy arm. However, the use of any defined significant bacteriuria as post hoc inclusion criterion is at least questionable. Nowadays, it is known that even under normal physiological condition, urine is not sterile [23]. The term “significant bacteriuria” was used in the past to differentiate between infection and contamination of a urine sample collected for analysis. When bladder urine from patients with unquestioned acute pyelonephritis was examined quantitatively, none contained less than 10⁵ colony forming units (CFU) of uropathogen per mL [24]. Bacteriuria of ≥ 10⁵ CFU/mL in adults was originally defined significant only for the diagnosis of pyelonephritis. In 1982, Stamm et al. [25] documented that the levels of ≥ 10⁵ CFU/mL of a pathogen in urine have a very high specificity (99%) but a very low sensitivity (51%) for the diagnosis of AC. Bacteriuria of ≥ 10² CFU/mL was suggested by the authors as the best diagnostic criterion (sensitivity, 95%; specificity, 85%). In 2013, Hooton et al. [26] confirmed that E. coli identified as low as 10¹–10² CFU/mL was sensitive and specific for the diagnosis of AC in symptomatic women. But still, about 20% of these symptomatic female patients were culture “negative” even when being tested for such low counts. Quantitative PCR (qPCR) for E. coli and S. saprophyticus finally demonstrated that almost all women with symptoms suggestive for UTIs and a “negative” culture still have an infection with E. coli [27]. Therefore, according to the German National S3 Guideline, the detection of E. coli in symptomatic women is predictive for a bacterial UTI, irrespective of the number of pathogens. In contrast, the presence of Enterococci and group B Streptococci in urine is not predictive for UTIs [3].

Hence, the use of a general definition for significant bacteriuria of ≥ 10⁵ CFU/mL as an inclusion criterion, may falsely exclude about half of the patients with a probable diagnosis of AC presented with the same symptoms. Therapeutic consequences drawn from such studies may have to be then restricted for this subgroup of patients. Therefore, we recommend considering all patients included with the same clinical criteria into a study as the main target population. Patients with bacteriuria of ≥ 10² or ≥ 10³ CFU/mL, in turn, should then only be considered as microbiologically evaluable patients. The same principles should be applied for outcome criteria, based on patient-reported outcome using a validated questionnaire at least up to 4 weeks after the end of therapy. Consideration of the elimination of bacteriuria as the main study aim is scientifically questionable, due to the findings that asymptomatic bacteriuria may probably be protective against recurrent UTI [28, 29]. It should, however, be registered as additional results of the study.

The analysis of 517 female respondents (patients and controls) has revealed that the diagnostic value of the ACSS cut-off value without urinalysis is at least as favourable as the draft proposals by FDA or by EMA. The most important advantage of the ACSS is that it could be used also in epidemiological studies or for self-diagnosis of the patient without the need for additional laboratory tests, such as urinalysis. For clinical interventional studies, however, the same threshold could be used as an inclusion criterion together with the evidence of pyuria and thus dramatically increasing the specificity.

Although it has been demonstrated that the scoring of the five first typical symptoms in the ACSS questionnaire are not much inferior to the six symptoms, including visible blood, we recommend to include further all six items in the typical domain, because visible hematuria in connection with typical urinary symptoms may be pathognomonic for acute hemorrhagic cystitis. It can also be an important differential sign. If visible hematuria persists after treatment, it needs a further careful investigation of the patient to exclude any other urological disease, such as bladder cancer. The Swiss guidelines have also included a recent onset of hematuria as one of the typical symptoms of AC besides frequency, urgency and dysuria with pyuria and bacteriuria of ≥ 10² CFU/ml [30].

The shortcoming of the study is mainly related to the design as a non-interventional, case–control study.

There are different laboratory methods defining pyuria in different countries. The dichotomized approach allocating pyuria into “significant” and “non-significant” allowed to bring the values together, thus reducing possible biases. The difference between pyuria (WBC ≥ 25/µL) tested in the current analysis and pyuria (WBC ≥ 10/µL) proposed by FDA and EMA remains open. The ratio of patients and controls in our study is 1:0.81, which is close to the optimal ratio of 1:1.

Because of the non-interventional approach, the study protocol could only be suggestive for the participating physicians, who were asked to follow the national and international guidelines for the diagnosis and treatment of women with AC. Therefore, variations of the management could only be minimized but not completely avoided.

The diagnostic values of the “draft” guidelines proposed by FDA and EMA were compared with the validated ACSS questionnaire. Not only the presence but also the severity of the symptoms (scoring) are important for an accurate diagnosis of AC. It could be shown that the diagnostic value of the ACSS, even without additional urinalysis, is at least as favorable for the clinical diagnosis of AC as the draft clinical proposals by FDA and EMA. Therefore, the ACSS can be recommended for epidemiological and interventional studies, and allows women for self-diagnosis of AC, which makes the ACSS also cost-effective for healthcare.