Hospital admission after primary care consultation for community-onset lower urinary tract infection: a cohort study of risks and predictors using linked data

Background Urinary tract infections (UTIs) are a common indication for antibiotic prescriptions, reductions in which would reduce antimicrobial resistance (AMR). Risk stratification of patients allows reductions to be made safely. Aim To identify risk factors for hospital admission following UTI, to inform targeted antibiotic stewardship. Design and setting Retrospective cohort study of East London primary care patients. Method Hospital admission outcomes following primary care consultation for UTI were analysed using linked data from primary care, secondary care, and microbiology, from 1 April 2012 to 31 March 2017. The outcomes analysed were urinary infection-related hospital admission (UHA) and all-cause hospital admission (AHA) within 30 days of UTI in primary care. Odds ratios between specific variables (demographic characteristics, prior antibiotic exposure, and comorbidities) and the outcomes were predicted using generalised estimating equations, and fitted to a final multivariable model including all variables with a P-value <0.1 on univariable analysis. Results Of the 169 524 episodes of UTI, UHA occurred in 1336 cases (0.8%, 95% confidence interval [CI] = 0.7 to 0.8) and AHA in 6516 cases (3.8%, 95% CI = 3.8 to 3.9). On multivariable analysis, increased odds of UHA were seen in patients aged 55–74 years (adjusted odds ratio [AOR] 1.49) and ≥75 years (AOR 3.24), relative to adults aged 16–34 years. Increased odds of UHA were also associated with chronic kidney disease (CKD; AOR 1.55), urinary catheters (AOR 2.01), prior antibiotics (AOR 1.38 for ≥3 courses), recurrent UTI (AOR 1.33), faecal incontinence (FI; AOR 1.47), and diabetes mellitus (DM; AOR 1.37). Conclusion Urinary infection-related hospital admission after primary care consultation for community-onset lower UTI was rare; however, increased odds for UHA were observed for some patient groups. Efforts to reduce antibiotic prescribing for suspected UTI should focus on patients aged <55 years without risk factors for complicated UTI, recurrent UTI, DM, or FI.


INTRODUCTION
Urinary tract infections (UTIs) are one of the commonest indications for antibiotic prescriptions in primary and secondary care, and reduction in prescribing for UTI would likely have a significant impact on overall antibiotic consumption and rates of antimicrobial resistance (AMR).Estimates suggest that the average person consults their GP 5.5 times per year and that 1%-3% of all GP consultations are for UTI symptoms. 1,2][5] Studies suggest that a proportion of uncomplicated UTIs are self-limiting, with up to 50% of women being symptom free without antibiotic treatment at 7 days. 6,7ntibiotic use is also associated with the development of AMR, gastrointestinal side effects, and complications such as Clostridium difficile colitis. 8,9Delay or avoidance of antibiotic treatment may therefore be preferable for certain patients.Trials involving young women have found that approximately two-thirds of patients recover with symptomatic treatment rather than antibiotics, but that this strategy was associated with a higher symptom burden and, in one of the studies, more cases of pyelonephritis. 10,11Such approaches may reduce antibiotic consumption, but data on patient outcomes are necessary to inform their acceptability and safety.A recent study using electronic health records (EHRs) found that the probability of sepsis was higher following consultations for UTI than for respiratory tract or skin infections, and that the risk of sepsis was higher among older adults. 12Other analyses of large datasets using EHRs have found conflicting results in older adults. 13,14he aim of this study was to obtain an accurate estimate of the risk of adverse outcomes following lower UTI in primary care in patients aged ≥16 years, in order to identify patients for whom antibiotic treatment could be safely delayed or avoided as a means of antibiotic stewardship.

METHOD
This was a retrospective cohort study using primary care EHRs linked to secondary care and microbiology data.

Data source
The study cohort was created using a primary care database of East London e694 British Journal of General Practice, September 2023 general practices developed and managed by the Clinical Effectiveness Group (CEG), part of Queen Mary University of London.The data were deterministically linked to Secondary Uses Services (SUS, managed by NHS Digital) secondary care data, and to microbiology data (urine and blood cultures) from Barts Health NHS Trust as per the supplementary material.
Consultations for lower UTI were identified through Read codes, antibiotic prescriptions for UTI, and positive urine cultures with relevant uropathogens.This study employed a modified version of the Read code lists used by previous similar studies (see Supplementary Table S2). 13,14irst-line antibiotics used to treat lower UTI are frequently unlinked to a diagnostic code, so UTI consultations were also identified through prescriptions for nitrofurantoin, trimethoprim, fosfomycin, and pivmecillinam. 15Relevant uropathogens are shown in Supplementary Table S3.UK guidelines recommend sending urine cultures only in certain situations including complicated UTI, treatment failure, or where antibiotic resistance is suspected. 16o ensure that the data set did not include patients with upper UTI, consultations where a Read code for upper UTI was recorded within +/-3 days of a positive urine culture were excluded (see Supplementary Table S4).Distinct episodes of UTI were identified using a 30-day washout period, with any consultations within that period considered part of the same episode.Any consultations outside the washout period were considered a new episode.Any consultation occurring within the washout period was considered an ongoing episode and excluded from the analysis (see Supplementary Figure S1).Patients could have >1 UTI episode during the study period.

Population
Patients aged ≥16 years registered at the approximately 100 GP surgeries across Tower Hamlets and Newham, London, who consulted their GP for UTI between 1 April 2012 and 31 March 2017 were eligible for inclusion in the study.Patients were excluded from the study if: • there were no data available for sex, age, Index of Multiple Deprivation (IMD) score; • they were registered for <12 months prior to their first episode (to allow for identification of comorbidities); • there were <30 days' follow-up data available (unless death occurred within that period); • they were admitted to hospital on the day of their UTI episode; and • they were discharged from hospital in the 30 days prior to their UTI episode.
Patients who had multiple episodes of UTI were included and entered the cohort

How this fits in
Reductions in antibiotic prescribing for urinary tract infections (UTIs) would reduce overall antibiotic consumption and antimicrobial resistance.Previous studies on prescribing using routinely collected data have previously not included microbiology data.Linkage of microbiology data to primary care and secondary care data is a novel element of this study that allowed the authors to strengthen the outcomes of urinary infection-related hospital admission in a way that has not been previously described.This study found that antibiotic stewardship efforts for UTIs should be targeted at younger patients without specific risk factors.at the start of their first episode.Patients left the cohort at the earliest of these dates: death, change of practice, or end of the study period.

Putative risk factors
A number of risk factors (see supplementary material for rationale and definitions) were examined as variables in the analysis, including: • demographics: age, sex, ethnicity, IMD quintile (obtained by linking individual patient's postcode to Lower Layer Super Output Area); • non-treatment with antibiotics within +/-7 days; • risk factors for complicated UTI: structural abnormalities, chronic kidney disease (CKD), urinary catheterisation; • recurrent UTI; • comorbidities: cancer, diabetes mellitus (DM), heart failure, hypertension, urinary incontinence (UI), faecal incontinence (FI), and obesity; • antibiotic prescriptions in the last 6 months (see supplementary material); and • season of the year.

Outcomes
The primary outcome was urinary infection- related hospital admission (UHA) in the 30 days from the start of an episode.This definition included International Classification of Diseases, 10th Revision (ICD-10) codes related to UTI (upper UTI, sepsis, and bloodstream infection; Supplementary Table S9), and urine and blood cultures positive for relevant organisms within 2 days of hospital admission.The secondary outcome was all-cause hospital admission (AHA) in the 30 days following an episode.

Statistical analyses
Descriptive statistics were used to summarise the clinical and demographic characteristics.Continuous variables were summarised using median and interquartile range (IQR), and categorical variables using absolute numbers and proportions.Crude associations (odds ratios [ORs]) were estimated between each of the variables (risk factors) and the outcomes using generalised estimating equations (GEEs) with a logit link and an exchangeable correlation structure to account for multiple episodes per patient.Huber-White sandwich estimators were used to calculate 95% confidence intervals (95% CIs).A final multivariable adjusted model was fitted using GEEs, including all predictors with a P-value <0.1 in the univariable analysis.Age was included as a categorical variable in the multivariable model, as this was felt to be most informative for primary care prescribing decisions.

Figure 2. Forest plot of multivariable analysis of odds of UHA. CKD = chronic kidney disease. DM = diabetes mellitus. UHA = urinary infection-related hospital admission. UTI = urinary tract infection.
e696 British Journal of General Practice, September 2023 Older patients may be more likely to be treated with antibiotics because of concerns around progression to sepsis, 17 so the possibility of an interaction between antibiotic treatment within +/-7 days and age on the outcome of UHA was explored.A model was run including an interaction term between age and antibiotic treatment, looking for a significant Wald P-value for the interaction coefficients.The Quasi Information Criterion (uQIC) was used to assess the model fit, using a difference in uQIC of ≥10 to signify a statistically significant improvement (where a lower number indicates a better fit).All data cleaning and analyses were performed using the statistical software R (version 3.6.1)for Windows.Generalised estimating equations were fitted using geepack (version 1.2-1).

RESULTS
A total of 169 524 UTI episodes (in 86 561 patients) were included in the study (Figure 1).The majority of episodes (n = 132 094; 77.9%) occurred in female patients, and the median age was 43 years (IQR 31-60; see Supplementary Table S1).Factors associated with antibiotic treatment within +/-7 days of the episode were female sex, older age, increased socioeconomic deprivation, prior antibiotic treatment, risk factors for complicated UTI, and all other comorbidities examined.
Of the 169 524 UTI episodes, AHA occurred in 6516 episodes (3.8%, 95% CI = 3.8 to 3.9).On multivariable analysis adjusting for age, sex, antibiotic treatment, and all variables associated with the outcome on multivariable analysis, factors associated with increased odds of AHA included older age, increased socioeconomic deprivation, prior antibiotic exposure, risk factors for complicated UTI, and all comorbidities examined apart from recurrent UTI (Table 2 and Figure 3).

Strengths and limitations
The strengths of this study include the assessment of microbiological and clinical outcomes to determine the proportion of admissions directly attributable to UTI, rather than simply assessing all-cause admission, which is potentially misleading.
The study used a novel database including a large cohort of patients with individual-level data on primary care consultations, antibiotic prescriptions, hospital admissions, and microbiology data, allowing investigation of risk factors for adverse outcomes in a way that has not been done previously.Recognising that coding for clinical consultations is not always complete, a wide range of indicators were used to capture UTI episodes.The limitations of this study are common to many studies using routinely collected data; the data were collected in short consultations that are focused on delivering clinical care.Prescriptions for UTI-specific antibiotics were used to identify consultations for UTI that did not include a diagnostic Read code.Although nitrofurantoin, pivmecillinam, and fosfomycin are not used for other indications, trimethoprim may rarely be used to treat respiratory tract or skin and soft tissue infections, which may have led to misclassifying consultations for another infection as one for UTI.Positive urine cultures may have represented asymptomatic bacteriuria; however, positive urine cultures accounted for only 5% of the included consultations (Figure 1).Furthermore, UK guidance is to send a urine culture from primary care only in cases where complicated infection is suspected so most urine cultures should be from patients who are symptomatic. 16his study aimed to include only community-onset cases and so did not capture attendances or prescriptions from urgent care centres, accident and emergency (without admission), or outpatient clinics.A proportion of included episodes may therefore have been healthcare-associated.Patients with a Read code for a suspected sexually transmitted infection (STI) were not specifically excluded, so it is possible that symptoms were due to STI rather than UTI.However, Read codes included: 'Urine culture' (62%), 'Suspected UTI' (22%), 'Urinary tract infection, site not specified' (8%), and 'Recurrent UTI' (3%).Additionally, none of the antibiotics used to identify consultations are used to treat STIs in the UK.In the UK, the majority of STIs are managed in sexual health clinics and a study of conditions treated with antibiotics in primary care in England found that only 6.7% of genitourinary conditions were genital in origin. 15he proportion of male patients who were not treated was high, which was surprising given that UTIs in male patients are considered complicated and will usually warrant treatment.It is possible that a significant proportion of the male consultations were catheter urine specimens, which the GP did not treat as they felt a catheter-associated UTI (CAUTI) was clinically unlikely.Very few Read codes for CAUTI were identified, there is no ICD code for CAUTI, and urine cultures frequently do not specify the specimen type.An attempt to mitigate this was made by identifying urinary catheters through prescriptions for devices and accessories recorded in the 6 months preceding the episode, but this will not have captured all CAUTIs.As CAUTI is one of the commonest healthcare-associated infections, this highlights the importance of improving coding of catheter use in primary care.It is also possible that a proportion of men had dysuria that the GP thought was an STI, so a urine culture was sent for completeness and the patient was referred to a sexual health clinic for treatment.
The authors of the study also acknowledge the risk of residual confounding as a study limitation.Non-treatment within +/-7 days was associated with reduced odds of UHA.This is surprising and suggests that the effect of confounding by indication in this study may be stronger than the protective effect of antibiotics in treated patients.There were systematic differences between treated and non-treated patients; non-treated patients were generally younger, had fewer comorbidities, and had less prior antibiotic exposure than treated patients, and this result may simply reflect healthier patients.On examining the data for an interaction between age and antibiotic treatment, no interaction was found (data not shown).
There is evidence that asymptomatic bacteriuria and symptomatic UTI in pregnancy is associated with increased risk of pyelonephritis. 18Very few codes related to UTI in pregnancy were identified so it was not possible to examine this as a risk factor.

Table 2 . Multivariable analysis of odds of AHA
The funders played no direct role in designing or conducting the study.This publication presents independent research funded by the NIHR and the Wellcome Trust.The views expressed are those of the authors and not necessarily those of the Wellcome Trust, the NHS, the NIHR, or the Department of Health and Social Care.This research was funded in whole, or in part, by the Wellcome Trust (reference: 206441/Z/17/Z).For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript version arising from this submission.
org.uk/ethics/) identified that research ethics approval was not required for this study as all data were pseudonymised and presented in aggregate form.HRA approval was received on 25 January 2018 (IRAS project ID: 226836; Research Ethics Committee reference: 18/HRA/0502).