Hypothesis / aims of study
Hypothesis: Although uncomplicated urinary tract infections (UTI) in women are very common, with a high cost to public health, high morbidity, and a significant impact on quality of life, there are few parameters to predict the risk of recurrence. Cai et al [1] created the nomogram called LUTIRE (Lower Urinary Tract Infection Recurrence Risk), which analyzes six risk factors and establishes a probability of recurrence. The 6 factors that compose the nomogram were given points related to their weight in obtaining the probability of recurrence. In descending order of relevance these are: a) treatment of asymptomatic bacteriuria in the previous year (50 points for previous treatment, zero score for absence of treatment); b) intestinal function (when constipated, 50 points, 30 for diarrhea, and no score for normal habit); c) type of pathogen involved (gram-negative 39 points and gram-positive zero points); d) number of infections in the previous 12 months before inclusion in the study (≥ 3 episodes received 39 points and 2 episodes received a zero score); e) number of different sexual partners in the previous year (if ≥ 3, 35 points, two partners, 19 points, one or no sexual activity, zero points); f) hormonal status (if in menopause, received 12 points, fertile, zero points).To calculate the probability, in the original study it was necessary to establish the score of each variable on a scale and, from the sum of the points of each variable, the total score and the corresponding probability of recurrence of urinary infection were obtainedThis nomogram was validated in an Italian population, proving useful in determining the risk of recurrence Aims of study: To validate the LUTIRE nomogram to predict the risk of reinfection in a Brazilian female population with a history of recurrent uncomplicated UTIs.
Study design, materials and methods
The sample calculation was performed based on the area under the ROC curve (AUC) found in the external validation group of the Italian nomogram composed of 373 women. In this study the AUC was 0.85 and the recurrence rate of UTIs at 12 months was 33.9%. Considering a type I error (α) of 0.05, a type II error (β) of 0.20, the null hypothesis value of the AUC of 0.50, and the AUC of the Italian nomogram of 0.85; the minimum sample size calculated was 21 women in the recurrence group and 21 women in the non-recurrence group. However, considering that the 12-month UTI recurrence rate in the reference study was 33.9%, the minimum sample size calculated was 21 women in the group who had a recurrence of urinary tract infection and 41 women in the group with no recurrence, totaling 62 patients. The LUTIRE nomogram was applied to 81 women aged between 18 and 65 years, who were screened from a private urological clinic and presented with uncomplicated recurrent UTIs. The inclusion criteria were: women aged equal to or over 18 years and under 65 years who presented recurrent uncomplicated UTIs with two or more previous episodes of cystitis in the previous six months, according to the criteria of Wagenlehner et al.[2] The clinical framework was confirmed by symptoms and positive uroculture (count at or above 105 CFU per milliliter of medium jet urine) with an antimicrobial susceptibility test collected before treatment started. The patients were included after a review of the database electronic records, using the CID-10 version, code N39.0, in the period between January 2014 and December 2016. From the review of the digital clinical files (a retrospective longitudinal study), the recurrence probabilities of the patients were calculated using the nomogram. The variables were obtained through information on the 12 months prior to the study. After inclusion and establishment of predictive probability of recurrence, the patient histories were followed for one year when asymptomatic, (outcome: no reinfection); or until recurrence (outcome). The outcome was compared to the probability originally established in the nomogram and the accuracy of the instrument was calculated. The continuous variables were evaluated by the Mann-Whitney U test and T-test; and the qualitative variables by the Chi-squared test. Subsequently, the six nomogram variables were included in a logistic regression model and the time for recurrence was assessed by Kaplan-Meier curve. An ROC curve (Receiver Operator Characteristic) was constructed to graphically demonstrate the sensitivities and specificities of the different odds of recurrence of a urinary tract infection calculated by the nomogram. Thus, the best cut-off point and area under the ROC curve (ASC) were determined with their respective accuracy.
Results
The mean age of the patients was 42.8 years. Fifty-seven women (70.37%) presented recurrence. The independent variables with statistical significance in the univariate and multivariate analyses were gram-negative bacteria (OR 18.38, p= 0.03897) and number of UTIs in the previous 12 months (OR 25.11, p= 0.00006). The accuracy of the nomogram was 82.6% (95% CI = 72.5-90.1). Women who presented a calculated recurrence probability greater than 40% had a 20.64 times greater chance of recurrence of UTI in 12 months than those who had a probability less than or equal to this value (table 1). The Brier score was 0.1971 ± 0.183 and ranged from 0.0344 to 0.5613, indicating good predictive ability of the nomogram.
Figure 1 illustrates the sensitivity and specificity for different values of calculated probability obtained by the nomogram to predict the recurrence of a UTI over a 12-month period. Considering a type I error (α) of 0.05, a type II error (β) of 0.20, the null hypothesis value of the AUC of 0.50, and the AUC obtained by the nomogram of 0.826, the minimum sample size was 24 women in the recurrent group and 24 women in the non-recurrent group. Therefore, the minimum sample size was reached to obtain a power of 80%.
Interpretation of results
To our knowledge, this is the first study to validate the nomogram in a population different from that analyzed by Cai et al [1]. One of the main aspects related to the use of predictive instruments and their implications in clinical practice is their superiority to clinical judgment when making decisions on the actions to be taken. The LUTIRE nomogram demonstrated high accuracy when applied in a Brazilian population with the purpose of discriminating the women with the greatest risk of recurrence of UTIs. Our results are similar to those of Cai et al.[1], with an accuracy of 82.6%, compared to the accuracy in the Italian cohort, which was 85%. The external validation of the nomogram by a different group is one of the fundamental steps for the study of implications in clinical practice, raising its classification to level 2 in the scale of studies of hierarchy of predictive systems [3]. Determination of patients with a higher probability of recurrence may be useful in the management of this condition, especially in relation to treatment dilemmas, such as the increased bacterial resistance to antimicrobials and the costs involved in the treatment.