Remote versus in-person consultations for acute cystitis: antibiotic treatment and clinical outcomes, a retrospective cohort study

Summary

Remote consultations for acute cystitis increased markedly during the COVID-19 pandemic, with a simultaneous reduction in the annual total acute cystitis contacts.

Antibiotic treatment within 3 days was less common following remote consultations for cystitis, compared with in-person consultations. Second-line treatment was more common following remote consultations. Repeat contact for cystitis within 14 days was also more common following remote consultations.

Strengths and limitations

All activity by GPs and OOH services must be reported to KUHR Database for reimbursements, ensuring a rich and complete dataset.²⁸ Registry data still has inherent limitations that are important to be aware of when interpreting our findings. Diagnoses and reimbursement codes can deviate from what occurred in the consultation. The rate of some events may therefore be underestimated, and the rate of others overestimated.

Biases are reduced when using routinely and systematically collected data, although we as researchers might introduce systematic errors when we define variables such as illness episode duration, and chronic morbidities, besides the present acute UTI. It is difficult to assess the direction and magnitude of these biases. A longer episode duration definition would risk grouping separate episodes into one. A shorter duration would potentially underestimate the number of repeat contacts and thereby lead to loss of information on the tail-end of the illness episodes. We believe the effect of these limitations on the observed total trends and correlations in our results to be comparatively minor, considering that we are using a complete national dataset.

Comparison with existing literature

A Dutch study by Nielen et al on defining illness episodes from registry data is relevant to our endeavour to define acute cystitis episodes to identify and compare modes of index consultations (remote versus in-person).²⁹ In that study, it was suggested to use an 8-week contact and antibiotic-free interval for cystitis and half of this interval (4 weeks) as episode duration. We tried to make this suggested model work with our individually linked national and complete dataset but found that using different intervals for washout and episode duration led to episodes erroneously overlapping. Numerous studies have documented that uncomplicated cystitis, the most common presentation of cystitis, is expected to resolve within 1 week.^1,4 Ultimately, we chose to construct our own episode definition for acute cystitis: washout of a minimum 2-week UTI contact and systemic antibiotic-free interval, episode start marked by an index consultation for cystitis and then a two-week episode duration.

The current understanding of the development of cystitis is that the ascending route of uropathogens is via the urethra into the bladder.³⁰ One would then not necessarily expect an airborne pandemic and associated infection control measures to affect the incidence of UTIs. Among common infections during the COVID-19 pandemic, the incidence of UTIs was observed to be uniquely stable in one study from England, while the incidence of all other investigated infectious diseases fell.³¹ We found an overall decrease in the annual consultations for acute cystitis during the COVID-19 pandemic. Patients’ physician-seeking behaviour changed during the pandemic,³² and more patients with acute cystitis may have chosen a wait-and-see approach. A well-established risk factor for acute cystitis is sexual intercourse.³³ A systematic review on the sexual health of women from 18 countries during the COVID-19 pandemic found a reduced frequency of sexual intercourse during the pandemic.³⁴ If there was a similar trend in Norway this could contribute to the observed annual decrease of cystitis consultations in our study.

Several previous articles report higher antibiotic treatment rates for remote than for in-person consultations for infections.^19,20,35 Conversely, we found that remote consultations for cystitis during the pandemic led to less antibiotic treatment overall within 3 days, but more repeat contacts during the subsequent 14 days. More repeat contacts could be interpreted in two ways; either the threshold for a repeat contact to the GP is lower following a remote consultation, or the acute cystitis was not adequately treated. The average time to repeat contact following index remote consultations was 4.4 days (SD = 3.9), which is after the time for the onset of maximum effect of the index antibiotic therapy (3 days) and repeat contacts could therefore either be an expression of therapy failure or a check-up to see if the infection had resolved. When antibiotics were dispensed following a remote consultation, they were more often second-line.

Repeat contact occurred in 23.3% of all acute cystitis episodes (Table 3), with 43.3% of these repeat contacts leading to antibiotic treatment, a finding that implies that in our study around 10% or less of acute cystitis episodes required repeat antibiotic treatment. This is similar to findings in the registry study from Denmark that found repeat antibiotic treatment in<10% of uncomplicated cystitis episodes.⁵ We did, however, not only examine uncomplicated cystitis, but also the acute cystitis episodes in our study included pregnant women, women with multimorbidity, and older people, which were adjusted for in the analyses.

Our findings show that increased use of remote consultations for acute cystitis in Norwegian primary care during the pandemic led to less initial antibiotic treatment. Contributing to this observation could be that Norwegian primary care is organised as a publicly funded RGP-scheme and has well-defined antibiotic guidelines, which in general advise against antibiotic treatment via remote consultations.²⁵ Another important factor is the continuity of the patient–physician relationship following the RGP scheme, which has proven to reduce the use of OOH services, acute hospital admissions, and mortality.³⁶ Continuity in the patient–physician relationship has also been suggested to reduce inappropriate antibiotic prescriptions.³⁷ The choice of consultation modality is at the discretion of the patient, which highlights the need for increased patient guidance, to ensure safe use of remote consultations.¹² More second-line treatment following index remote consultations could be an expression of uncertainty about the severity of the UTI in question on the part of the GP, leading to prescription of what could be perceived as more effective treatment, second-line antibiotics, just in case. Repeat contacts after index remote consultations resulted in proportionally more antibiotics than repeat contacts following in-person consultations (44.7% versus 43.1%, P<0.001) (Table 3). Cystitis episodes initiated by remote consultations and followed up with secondary remote consultations were characterised by a comparable proportion of antibiotic treatment overall, but more repeat contacts and more second-line antibiotics.

In our study, repeat contacts for pyelonephritis within 14 days occurred in 1.7% of all acute cystitis episodes, in line with the registry study from Sweden, which found that about 1% of untreated uncomplicated cystitis episodes were followed by pyelonephritis within 30 days.⁶ We found a comparatively stronger association between an in-person consultation and subsequent contact for pyelonephritis. We interpret this as possible selection bias; patients who had the early hallmarks of pyelonephritis may have preferred in-person consultation rather than to be evaluated by remote consultation. Another possibility is that the threshold for seeking an in-person consultation could be higher than for a remote consultation, which introduces a delay in the index contact for acute cystitis. Aligning with this view is that repeat contacts occurred later following in-person (mean = 5.2 days, SD = 3.7), than remote consultations (mean = 4.4 days, SD = 3.9 days) and that when a repeat contact occurred, index remote consultations were more often followed by a remote consultation, compared with index in-person consultations (29.0% versus 5.0%, respectively, Table 4). Acute cystitis itself occurs before a patient meets with the physician and our earliest datapoint is the index contact, when the patient already has had an infection or infectious-related symptoms for some time.

The acute cystitis episodes in this study were not strictly compatible with the classic definition of uncomplicated acute cystitis; for instance, we chose to include all females aged ≥16 years and included pregnancy as an adjustment variable, rather than being an exclusion criterion. The age of majority in matters of health is 16 years in Norway and this is the earliest age patients can use remote consultations by themselves, which marked a natural inclusion point for our study, comparing remote and in-person consultations. Defining pregnancy in registry-based studies is challenging but important, as acute cystitis frequently occurs in women of childbearing age. Acute cystitis in pregnant women is, however, not uncomplicated and Norwegian guidelines recommend prompt antibiotic treatment.²⁵ A Norwegian study combined data from three national registries for the years 2009–2013 and found 421 201 pregnancies in total or around 84 240 pregnancies per year.³⁸ This divided by the average annual Norwegian female population aged ≥16 years during the same period (1 979 333 women), as included in our study,³⁹ gives around 4.3% of Norwegian women being pregnant, per year. Our definition of pregnancy (Table 1) arguably gave a reasonable estimate of 4.0% of women with acute cystitis aged ≥16 years being labelled as pregnant (Table 3).

A slightly higher proportion of patients using remote consultations were pregnant or had multimorbidity (5.7% and 16.9%), compared with (3.9% and 13.8%, respectively) for in-person consultations. One possible explanation could be an increased worry for contracting COVID-19 during the height of the pandemic in these patient groups or perhaps the convenience of remote consultations, or both. These findings underscore how remote consultations can expand access to care.

Implications for research and practice

Our findings are influenced by the specific conditions in the Norwegian primary care system and the unique period the COVID-19 pandemic represented. A relatively low rate of antimicrobial resistance and the Norwegian RGP scheme are two challenges to the generalisability of our findings to primary care in other countries. Further validation of our suggested method to define pregnancy in ICPC2 registry-based studies (Table 1) could be an interesting avenue for future research.

The increased use of remote consultations has endured in the years following the pandemic.⁴⁰ The Norwegian Directorate of Health now aims to establish an online GP service and to mandate by law that RGPs offer remote consultations.¹⁸ Remote consultations can expand access to care, but our findings suggest some important limitations that policymakers should be aware of, including the observed increased dispensation of second-line antibiotic treatment.