Diagnostic Tools in Patients With a Suspected Urinary Tract Infection in Primary Care

Background

The latest report from the Staten Serum Institute (SSI)1 showed that primary health care in Denmark represents 90% of all prescribed defined daily dose(DDD) of antibiotics, and that a growing consumption of antibiotics is running alongside increasing resistance to E. coli, specifically ESBL-producing E. coli (Extended Spectrum Beta Lactamase Producing E. coli) .

The main cause of E. coli infection at primary care level is a Urinary Tract Infection (UTI) which is one of the most frequent reasons for consultation in general practice in Denmark. Interestingly, there is no consensus yet about the most effective diagnostic approach to identify those patients that will benefit from antibiotic therapy.

Currently in Denmark, a wide variety of diagnostic approaches can be identified in primary care settings (i.e. only history taking, dipsticks, microscopy, culture and susceptibility testing in practice). Nonetheless, there is not conclusive evidence yet about the diagnostic validity of all these approaches and the association with appropriate prescription of antibiotics(i.e. correct decision to prescribe and correct choice of antibiotics reflecting the bacterial cause and resistance) and impact on patients' clinical and microbiological cure.

For example, some studies that only include uncomplicated UTIs have shown that symptoms alone have a very low predictive value leading to overuse of antibiotics2,3. However, Richards et al after comparing the response to antibiotics of women with symptoms of UTI and a negative dipstick test result, concluded that empirical antibiotic use guided only by symptoms was a reasonable option4.

Furthermore, microscopy and more recent diagnostic tools such as a culture performed in primary care settings have not been included in the latest pragmatic validation studies5,6. Phase-contrast microscopy is widely used in primary care in Denmark, although the very few studies on the subject show wide variation in sensitivity (60-100%) and specificity (49-100%) of urine microscopy to predict significant bacteriuria.

Besides susceptibility testing at primary care level is only done in Denmark and the quality control of the performance of these tests differs across the country. MIKAP (the Danish agency in charge of the surveillance of the quality of the microbiological tests performed in general practices) supervises only some Danish Regions and has not authority on the quality control of the tests carried out in the capital region.

In order to curb the development of resistance bacteria strains in patients with urinary tract symptoms and infections, we need to increase our knowledge of the validity of the different diagnostic approaches and its impact on proper use of antibiotics and recovery across the heterogeneous group of patients seeking health care in primary care.

Objectives

The central research question will be answered throughout the following specific objectives:

1. To assess the diagnostic validity (i.e. sensitivity and specificity, positive and negative predictive value) of different diagnostic approaches in ruling in or ruling out a UTI in different sub-population groups in general practice.
2. To describe the impact of different diagnostic approaches on the appropriateness of antibiotic therapy (i.e. decision to prescribe and choice of antibiotic).
3. To describe the impact of different diagnostic approaches on clinical and microbiological cure.
4. To measure the prevalence of ESBL-resistant E.coli in patients with urinary tract symptoms in general practice.

Materials and Methods:

Design and setting: Prospective observational study with comparison of different diagnostic approaches to gold standard in general practice.

Data Collection: Before the index consultation, the participating GPs will be asked to complete a background questionnaire about experience, organization of the service and access to different diagnostic tools.

The day of the index consultation, when the GP types the ICPC code for dysuria (U01) or frequency (U02), as a main motive of consultation, a pop-up will automatically appear with a questionnaire about the anamnesis, diagnostic approach, treatment and other relevant decisions made during the consultation; this questionnaire will be directly linked to DAK-E/DAMD database (Danish General Practice database aimed at improving quality in the delivery of primary medical care services). Afterwards, the patient will also be asked to provide a urine sample which will be sent to the Staten's Serum Institute (SSI) for culture, which will be used as the reference gold standard.

The following 14 days after the index consultation, the patients will complete a symptom diary (attached). They will be asked to rate some symptoms each day until recovery (or for 7 days if symptoms are ongoing) on a 3 point scale from "nothing " to "a lot". After 14 days patients will have to send a control urine sample. Patients with a persistently positive culture will be followed up after 1and 3 months.

Further information about patient background regarding co-morbidities, previous hospitalizations, and previous use of antibiotics will be gathered from national registers such as The Danish national patients register and The Danish national prescription register.

Biobank: 10 µl of urine will be stored following the safety and hygiene procedures of the Staten's Serum Institute microbiological laboratory only for those urine samples positive for ESBL-producing E. coli. After detection of a positive resistant strain, the samples will be anonymized and stored for a period of 2 years after completion of the current study. We plan to carry out further microbiological analysis of the resistant bacteria in collaboration with experts in proteonomics and genomics within the bigger UC-CARE projects. The collection of the urine sample does not represent any risk for the patients as it follows the normal procedure in general practice, it means, the patients is requested to collect a random urine sample in a plastic cup at the practice facility and 14 days after at home.

Data management: All data will be entered into a database which will be linked to different national registers to obtain detailed data about background information at patient level. Data will be screened for data entry errors and extreme values using tables, plots and specific commands of the SAS 9.03.

Variables: considering the multidimensional and multivariable models that will have to be constructed in order to accurately answer the objectives, the following variables with the corresponding data source will be included:

Patient level

Patients' demographics (gender, age), dysuria, frequency, supra-pubic pressure, polyuria, Fever (>38oC), trembling, urine offensive smell, gynecological complaints (vaginal discharge, pruritus), low back pain, blood in urine. Number of days since first symptom started. Diagnostic test:, diagnose, treatment, severity of symptoms, bothersomeness, impact on daily activities. Previous hospitalization within the last year: diagnosis, discharge date. Consumption of antibiotics within the last year.

GP level Number of consultations during office hours due to dysuria and/or frequency as a main motive of consultation, GPs' demographics, number of patients in the GPs' list, point-of-care test for urine examination: type of point-of-care tests available in the practice, brand and technical specifications for each test. Attendance to courses about performance of the tests.

Statistical analysis: Data will be analyzed using SAS 9.03. A comparison of proportions will be performed using the chi-square test. For the first two objectives, the result of the urine culture will be used as the gold standard, a cut-off value of >=103cfu/ml, or pure growth of E.coli, and > 10 4 for more unusual organisms or mixed growth will be used to define a positive culture,and we will calculate sensitivity, specificity, negative predictive value and positive predictive value for each diagnostic approach against the gold standard.

Regarding the third objective prescription of antibiotics and choice of the right antibiotic will be operationalized as binary dependent variables in a logistic regression model, while for the last two objectives number of days until symptomatic and microbiological cure will be analyzed as continuous dependent variables in a multiple linear regression model. Depending on the number of cases included in the one month and three months follow-up of microbiological cure, a Poisson regression model will be developed.

The propensity score matching technic will be employed to adjust for pre-test imbalances in the different groups of treatment approaches in order to produce unbiased estimates of the tests effect on the main outcomes and create participant characteristic balance among the groups.

Sample calculation Different samples sizes will be needed to compare proper use of antibiotics between the 16 diagnostic profiles. Due to the observational design of the study, we do not know yet with certainty the distribution of the diagnostic profiles, so at least there are two samples sized we are aiming at. The first one aims to see a clinically relevant 10% difference in proper use of antibiotics when comparing the diagnostic profiles that affect the decision during the consultation versus the diagnostic profiles that affect the decision after the consultation.

The assumptions are: a) the 4 diagnostic profiles that influence the decision during the consultation are use in 60% of the patients, b) the correct decision is taken in 60% of the patients in this group, while c) the correct decision during the consultation is taken only in 40% of the patients in which any of the remaining 12 diagnostic profiles are performed. We will need 60 practices recruiting 15 patients each (n=900). Intra-class correlation =0.2, α=0.05 and β=0.2.

The second sample size relates to the detection of a prevalence of ESBL-producing E. Coli between 0 to 6% in patients with a suspected UTI in primary care. Our assumptions for the sample calculation are as follows: a) 60% of patients with symptoms do not have a positive culture, b) 80% of the patients with a positive culture are infected with a E.Coli strain. Therefore, we need 1500 patients with a suspected UTI attending primary care to detect a prevalence of ESBL-resistant E.coli between 0 to 6% with a 95% confidence interval.

Ethics

The study does not represent any risk for the patients due to the observational design. It means the diagnostic process and the treatment will be register as it is done in every day practice. Participation in the study implies for the patient that they have to fill in a diary about symptoms for 7 days and they should be willing to be contacted 3 days and 14 days after by the personnel from the practice to know whether they have filled in the diary, sent it to the research unit and sent the second sample to the microbiological department. Consequently, the course of the symptoms/disease is not affected by the fact of participating or not in the study.

All the personnel at each practice will be informed about the study, so as soon as the patient contacts the practice due to the symptoms listed at the inclusion criteria section, they will receive the first information about the study.

The first contact person (i.e. secretary, nurse, general practitioner) will ask the patient whether they are interested in participating in a project that aims to improve the management of patients with a suspected urinary tract infection. If the person is interested in participating in the study, then he/she will be given a leaflet with information about the study. This leaflet can be read while waiting for the consultation.

If the patient has further questions, they will be answered during the consultation to guarantee privacy. The patient is welcome to have a companion, in case they consider it is necessary in order to make a well-informed and thorough decision. However, due to the fact that these symptoms are acute and non-life-threatening complaints, in every day practice the patient uses to come alone. It means, to guarantee that the patient makes a non-influenced decision, the personnel is instructed to emphasize to the patients that they have as much time as the want to think about, they can ask questions and they can be sure that they refusal to participating in the project won't change the normal diagnostic process and treatment.