Discussion

This study has demonstrated that compliance with local prescribing guidelines in patients with colonic diverticulitis was poor. There is a paucity of existing literature specifically addressing antibiotic compliance in complicated diverticulitis. A multicenter retrospective cohort study assessed guideline compliance in the management of uncomplicated diverticulitis. Compliance (no antibiotics) varied significantly between centers, ranging from 37% to 96%. ^{Reference Dalby, Orru and Sundh13} A smaller study assessing the implementation of an antibiotic-free treatment protocol in a single center also found poor rates of compliance (60%) and no differences in complications ^{Reference Azhar, Kulstad, Pålsson, Kurt Schultz, Lydrup and Buchwald14} . This study demonstrates similarly poor rates of guideline compliance in complicated diverticulitis. It has shown complicated disease was the only statistically significant variable associated with non-compliance. Meanwhile, no differences in premorbid patient characteristics were observed between groups. It was also notable that a large proportion of those admitted with diverticulitis and prescribed antibiotics had uncomplicated disease (290/445). Multiple studies have demonstrated that antibiotics or even hospitalization are unnecessary for these patients. ^{Reference Daniels, Unlu and de Korte7,Reference Chabok, Pahlman, Hjern, Haapaniemi, Smedh and Group8} This further suggests that routine prescription of broad-spectrum antibiotics for hospitalized diverticulitis patients is unnecessary.

Different guidelines do exist internationally and between institutions. Some advocate for upfront use of broader spectrum antibiotics, including PIT, particularly for patients presenting with sepsis or septic shock. One such example is The Surgical Infection Society Revised Guidelines on the Management of Intra-Abdominal Infections (SIS Guidelines). ^{Reference Mazuski, Tessier and May17} It is important to note that local susceptibilities must be considered, and rates of antimicrobial resistance differ. Even so, the SIS Guidelines advocate for broad-spectrum agents for higher-risk patients based on clinical severity and patient comorbidities. It is evident in this study, that although complicated presentations were associated with more non-compliant prescribing, patient comorbidities were not. Particularly within the complicated subgroup, compliant prescribing was not associated with statistically significant differences in biochemical markers of disease severity. This suggests that although the complicated subgroup was more likely to receive non-compliant antibiotics, disease severity in this subgroup did not impact prescribing. Similarly in the uncomplicated subgroup, there were no statistically significant differences in CRP, leukocytosis, or CCI. This contrasts with some existing literature, where non-compliance to guidelines was associated with higher CRP values and more comorbidities. ^{Reference Dalby, Orru and Sundh13,Reference Azhar, Kulstad, Pålsson, Kurt Schultz, Lydrup and Buchwald14} Interestingly, the vast majority of those non-compliantly prescribed PIT were stepped down to oral AMC. This further reflects that antibiotic choice was more driven by clinician familiarity rather than clinical necessity.

The non-compliant use of broader spectrum antibiotics was not associated with improved 90-day outcomes. In fact, non-compliance was in total associated with more frequent 90-day recurrence and need for surgery. This is likely because patients with complicated diverticulitis received more non-compliant antibiotics. Accounting for this, the complicated subgroup experienced essentially identical 90-day rates of recurrence, surgery, and mortality—regardless of compliance. A retrospective study investigated the impact of antibiotic duration on the outcome of non-operatively managed diverticulitis. That study included a variety of antimicrobial regimens. It was noted that the only statistically significant risk factor for treatment failure was the size of the diverticular abscess, not the choice or duration of antibiotics. ^{Reference Khaoudy, Sabbagh, Brac and Regimbeau9} This study has replicated those results to some extent and found no differences in the rates of treatment failure based on antibiotic selection (compliant or otherwise) in patients with complicated diverticulitis. Interestingly, within the uncomplicated subgroup, non-compliance was associated with higher rates of 90-day recurrence, although there were no differences in need for surgery or mortality. It is unclear from the data collected as to why this may have been. Other markers of disease severity (and likelihood of relapse), including the clinician’s subjective assessment, may have driven non-compliant prescribing.

The 90-day mortality rates in this study are comparable to the 2022 Cochrane Review examining the use of antibiotics in uncomplicated diverticulitis. Three randomized control trials in that review with follow-up periods of up to 12 months had overall mortality rates of approximately 4.6%. ^{Reference Dichman, Rosenstock and Shabanzadeh18} Another trial examining short course antibiotics for source-controlled intra-abdominal infection on the other hand demonstrated mortality rates of around 1% at 30-day follow-up. ^{Reference Sawyer, Claridge and Nathens19} The slightly higher mortality rates in the present study likely reflects the older age group and higher burden of comorbidities among that group of patients.

Although guideline non-compliance does not necessarily equate to inappropriate prescribing, it would nevertheless be expected to carry a higher rate of inappropriate prescribing, with various downsides. The association between antibiotic exposure and the emergence of resistant organisms at both the individual and ecological levels is well established. ^{Reference Holmes, Moore and Sundsfjord20} Although important, this risk is often not appreciated in the immediate clinical context. To this end, this study has also been able to demonstrate more readily appreciable harm in inappropriate prescribing. Namely, it was identified that prolonged antibiotic prescribing beyond clinical necessity was commonplace and represented a significant portion of non-compliant prescription. Although not an outcome measured in this study, the association between longer duration and rates of adverse effects of antibiotics has been repeatedly demonstrated in other studies. One study found a 2.4-times increase in relative risk of antibiotic adverse effects with as few as 3 more days of additional exposure. ^{Reference de Wijkerslooth, Boerma and van Rossem21} With further regards to antibiotic exposure, patients in this study on average received 5 days of additional oral therapy, irrespective of intravenous duration. Several studies have demonstrated short-course total treatment is non-inferior in complicated intra-abdominal infections. ^{Reference Sawyer, Claridge and Nathens19,Reference Montravers, Tubach and Lescot22} Therefore, the seemingly standard practice of 5 days of oral antibiotic is likely unnecessary. Furthermore, although PIT was heavily prescribed in the non-compliant group, this study has demonstrated relatively low rates of P. aeruginosa, with two (0.4%) bloodstream infections and seven (1.6%) intra-abdominal cultures. The utility therefore of routine empirical anti-pseudomonal therapy appears weak.

Although assessing the factors influencing antibiotic prescription was outside of the scope of this study, some conjectures can be made. Prescribing decisions likely reflect prior training, institutional and alternative guidelines, as well as established practice patterns. There is also interesting existing literature in sociology and economics which examine some possibilities. The pressure of mitigating perceived short-term risks of infection may outweigh the responsibility of appropriate prescribing and longer-term stewardship. Surveys have also demonstrated the desire to appease senior colleagues can be of great influence within antibiotic decision-making. ^{Reference Broom, Broom, Kirby and Adams23} Peer behavior and perceived normal prescribing practices within an organization have also been correlated with such decision-making. ^{Reference Carlsson, Jacobsson, Lampi and Rönnerstrand24}

To the best of the authors’ knowledge, this study is unique in assessing the rates of compliance to local antibiotic guidelines. Analysis of guideline compliance with respect to patient outcomes also appeared novel. That said, there are several notable limitations to this study. Firstly, a portion of oral antibiotic data was not available to the authors. However, the use of subsequent oral antibiotics was not the focus of this study. The available data effectively reflected that patients were switched to narrower-spectrum antibiotics despite receiving broad-spectrum treatment in hospital, suggesting again that familiarity more than necessity may be the reason behind many prescriptions. The small proportion of absent oral antibiotic data did not impact on the ability to assess guideline compliance of initial management. Secondly, the authors chose to only assess the broadest spectrum agent prescribed by the treating team to present a relatively simple overview of prescribing practices. The authors observed numerous instances of broadening treatment without clear clinical justification. Future studies analyzing exact proportions of patients’ antibiotic courses with regards to compliance would be of interest. Particularly investigating the rates of broadening or narrowing therapy and the justifications of such decisions. Thirdly, the definition used for complicated diverticulitis differs slightly from that of the American Gastroenterological Association, which includes presence of abscess, peritonitis, obstruction, stricture and/or fistula as markers of complicated diverticulitis. ^{Reference Peery, Shaukat and Strate4} Although this may limit generalizability of findings, the authors felt using the standard Australian antimicrobial guidelines definition best emulated local clinical practice. Including positive blood culture among the definition for complicated diverticulitis is relatively unique, which was felt to be a biologically plausible consequence of greater disease burden. Fourthly, granular clinical markers of disease severity (such as heart rate, blood pressure) were not captured. This may have limited the study’s ability to find additional associations with non-compliance. Finally, the study’s retrospective nature inevitably introduced bias. It was also limited to only observing statistically significant differences without being able to definitively assign causation. The period of follow up was also relatively short, which may overlook morbidity/mortality beyond the 90-day time frame.