Introduction

The growing spread of antibiotic resistance (AMR) is a significant threat for global public health [1]. Among European countries, 75% of infections directly attributable to AMR were estimated to be healthcare-associated [2]. The prevention of healthcare-associated infections (HAIs) is one of the main strategies to tackle AMR and is an achievable goal. Evidence-based infection prevention and control (IPC) interventions could prevent up to 70% of observed HAIs [3]. AMR rates in Italy are among the highest in Europe, and HAIs cause a significant burden [4]. A study based on 2015 data found Italy was one of the European countries with the highest rates of infections due to AMR pathogens. The same study estimated one-third of all deaths attributable to AMR in Europe occurred in Italy [2]. The 2017 ECDC (European Centre for Disease Prevention and Control) “Country visit to Italy to discuss AMR issues” highlighted critical issues in management and coordination [5].

The Italian National plan to combat AMR was published in 2017 and updated in 2022 [6]. The plan promotes evidence-based IPC strategies, and underlines the importance of evaluating the impact of interventions on HAI rates. Short-term objectives include the implementation of nationally coordinated and integrated IPC and antimicrobial stewardship strategies, as well as the development of IPC programs that are sustainable and applicable in different health care settings. The plan outlines IPC strategies coordinated at national and regional levels, detailing roles, responsibilities, and required resources [6].

Italy has a universal National healthcare system with a multi-level governance structure. The system aims to provide equal access to standardized levels of health services, regardless of income or geographic location. National health policies and priorities are under the responsibility of the central government (Ministry of health) which determines annual funding and controls the allocation of resources. However, the organization and delivery of health services, including the implementation of the National action plan to combat AMR, is decentralized and entrusted to the country’s 21 regions and autonomous provinces [7]. Regional AMR policies, plans, and programs are enacted, leading to different levels of implementation of the National action plan to combat AMR within the same National healthcare system.

In 2018, following the publication of the first National action plan to combat AMR, the Italian Centre for disease control and prevention (Centro Nazionale per la prevenzione e il controllo delle malattie, CCM, Ministry of health) financed the SPiNCAR (Supporto al Piano nazionale per il contrasto all’antimicrobico resistenza) project. Within this project, a self-assessment tool was developed, aiming provide a common methodological framework to monitor the level of implementation of the National action plan to combat AMR at regional and local level. The tool promotes continuous improvement by allowing regional governments and trusts to evaluate strengths and weaknesses of their programs against defined standards [8].

This study reports data from the baseline data collection through SPiNCAR, conducted at national level in 2023, with a delay due to the COVID-19 pandemic. The aim of this study was (a) to assess the level of implementation of the National action plan to combat AMR among Italian regions and (b) to validate the SPiNCAR tool through correlation with other relevant indicators.

Methods

In this study, regional-level responses collected through SPiNCAR were related to data collected through the third ECDC point prevalence survey (PPS) of HAIs and antimicrobial use in acute care hospitals, aggregated at regional level, and antibiotic consumption data retrieved from the Italian Medicine agency (Agenzia Italiana del farmaco, AIFA) national report [9]. Only regions participating in both SPiNCAR and PPS data collections were included in analyses. All sources were linked at the region/autonomous province level. SPiNCAR refers to implementation status as of 31 December 2022; the PPS was conducted between 1 November and 31 December 2022; antimicrobial consumption refers to the 2022 calendar year.

Results

This analysis included results from 13 regions (Fig 1). Regional-level results per considered variable are summarized in Table 1. Among regions participating in both SPiNCAR and PPS, 48,754 non-COVID-19 patients from 340 acute-care hospitals were included in the PPS. Of these, 4011 had an active HAI on the day of the survey, resulting in an overall HAI prevalence of 8.23 (95% confidence interval, CI 7.98–8.47), with substantial regional variation (median 8.12%, range 3.94–11.46). On the day of the PPS, 21,174 patients received an antimicrobial, resulting in an overall antimicrobial use prevalence of 43.43 (95% CI 42.99–43.87), showing a narrower regional spread (median 43.94%, range 39.61–50.54). Antimicrobial susceptibility testing results were available for 2257 isolates sampled from active HAIs. Among isolates with available antimicrobial susceptibility testing results, 612 were resistant to first-level AMR markers, resulting in an overall AMR composite index of 27.12 (95% CI 25.29–29). The AMR composite index displayed the widest heterogeneity (overall 27.12%, 95% CI 25.29–29.00; median 24.44%, range 10.00–48.61), suggesting marked differences in resistance burden across regions. Across participating regions, implementation was highest for governance and surveillance, while evaluation/performance assessment, education/training, and stakeholder alliance showed the lowest scores (Table 1). Overall SPiNCAR scores ranged from 32 to 82, indicating substantial heterogeneity.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. SPiNCAR scores, healthcare-associated infection (HAI) prevalence, antibiotic use prevalence, antimicrobial resistance (AMR) composite index, and antimicrobial consumption results for 13 regions of Italy, 2022.

https://doi.org/10.1371/journal.pone.0347044.t001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Italian regions included in the study.

Regions participating in both SPiNCAR (Support for the National plan to combat antimicrobial resistance) and PPS (point prevalence survey of healthcare associated infections and antibiotic use among European acute-care hospitals), and therefore included in this study, are represented in dark blue. Italian regions not included in the study are represented in light blue. Regions not part of Italy are represented in grey. Created with MapChart.

https://doi.org/10.1371/journal.pone.0347044.g001

Results of correlation analyses are summarized in Tables 2–4. Selected significant correlations are depicted in Fig 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Results of correlation analyses, Spearman’s Rho (95% confidence intervals) – SPiNCAR scores vs. PPS indicators, Italy, 2022.

https://doi.org/10.1371/journal.pone.0347044.t002

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Results of correlation analyses, Spearman’s Rho (95% confidence intervals) – SPiNCAR scores vs. antimicrobial consumption data, Italy, 2022.

https://doi.org/10.1371/journal.pone.0347044.t003

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Results of correlation analyses, Spearman’s Rho (95% confidence intervals) – PPS vs. antimicrobial consumption data, Italy, 2022.

https://doi.org/10.1371/journal.pone.0347044.t004

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Scatter plots depicting selected significant results of correlation analyses, Italy, 2022.

Top row, left to right: SPiNCAR appropriate antimicrobial use score vs. antimicrobial use prevalence (p 0.046); AMR composite index vs. community consumption of systemic antimicrobials (p 0.003) [9]; SPiNCAR overall score vs. community consumption of systemic antimicrobials (p 0.048) [9]. Bottom row, left to right: SPiNCAR governance score vs. HAI prevalence (p 0.034); SPiNCAR surveillance and monitoring score vs. community consumption of systemic antimicrobials (p 0.012) [9]; SPiNCAR surveillance and monitoring score vs. antimicrobial use prevalence (p 0.046). Blue dots represent regional-level results. AMR: antimicrobial resistance; DDDs: defined daily dosages; HAI: healthcare-associated infections; PPS: point prevalence survey of HAIs and antibiotic use among European acute-care hospitals; pds: patient-days; SPiNCAR: support for the National plan to combat antimicrobial resistance.

https://doi.org/10.1371/journal.pone.0347044.g002

In uncorrected analyses, higher SPiNCAR governance scores were moderately associated with lower HAI prevalence (Spearman’s ρ = −0.59; p = 0.034). Higher surveillance/monitoring and appropriate antimicrobial use scores were each moderately associated with lower antimicrobial use prevalence (ρ = −0.56; p = 0.046 and ρ = −0.56; p = 0.047, respectively). The strongest association was observed for the SPiNCAR “evaluation of impact and implementation” domain, which showed a strong inverse correlation with antimicrobial use prevalence (ρ = −0.77; p = 0.002). In addition, community antibiotic consumption was inversely correlated with SPiNCAR surveillance/monitoring (ρ = −0.67; p = 0.012) and overall SPiNCAR score (ρ = −0.56; p = 0.048). Finally, the AMR composite index was strongly and positively correlated with community antibiotic consumption (ρ = 0.75; p = 0.003), and this was the only result remaining significant after Bonferroni correction.

Discussion

This was the first study reporting the level of implementation of the Italian National action plan to combat AMR across regions, assessed through a standardized tool. Monitoring implementation is crucial to integrate and coordinate strategies [8]. From a collaborative perspective, results of monitoring through SPiNCAR could facilitate sharing of best practices: regions with higher implementation levels could become references for those facing greater challenges [14].

Strengths and implementation gaps were assessed at national and regional levels against a national benchmark, providing a roadmap to plan targeted interventions. Considering all participant regions, high levels of implementation were found in the areas of governance, surveillance, HAI prevention and control, and appropriate use of antimicrobials, while the areas with the lowest levels of implementation were performance assessment, education and training, and alliance among stakeholders. Scores for governance and surveillance were consistently high, whereas scores for performance assessment and alliance among stakeholders were consistently low across regions. The highest heterogeneity was found for the areas of HAI prevention and control and education and training.

This assessment highlighted significant room for improvement in multisectoral collaboration across human, animal, and environmental sectors, in a One Health perspective [15]. Awareness and engagement campaigns aimed at diverse stakeholders should be promoted. Successful interventions have been described aiming to promote appropriate antimicrobial use and involving citizens, pharmacists, and general practitioners [16,17]. To translate these priorities into measurable actions, the updated PNCAR 2022–2025 defines specific targets to be achieved by 2025, including at least a 10% reduction in community consumption of systemic antibiotics and at least a 20% reduction in the broad- to narrow-spectrum antibiotic consumption ratio; in pediatrics, it also targets a ≥ 30% increase in the amoxicillin to amoxicillin–clavulanate prescribing ratio. In addition, the plan sets a system-level milestone that by the second half of 2023 all Regions should adopt tools to monitor antimicrobial consumption and appropriateness and progress toward integration and interoperability of information flows to support corrective actions [6].

The second objective of this study was to provide an exploratory validation of SPiNCAR assessment against relevant indicators. Significant correlations were found for several domains. However, most associations were statistically significant only in uncorrected analyses; after Bonferroni adjustment, only the correlation between the AMR composite index and community antibiotic consumption remained significant, and all other correlations should be interpreted as exploratory.

The relation between SPiNCAR governance score and HAI prevalence underscores the importance of institutional commitment and coordination. As in any other field, the availability of a plan does not necessarily ensure that the plan is enacted [18]. Overarching National action plans to combat AMR have been described as ‘plans of plans’: they outline strategic objectives and list recommended actions, combining both new and existing activities enacted by different departments and sectors. For the successful implementation of the plan, clear structures and systems are required to resource and coordinate activities across disciplines [19].

Moderate inverse relations were found between SPiNCAR surveillance and monitoring score and both hospital and community antibiotic use. Surveillance data are essential to inform action at local, regional, and national levels, allowing to enact evidence-based interventions and guide strategy and policy [16]. Further, our study found a strong inverse correlation between SPiNCAR performance assessment and antimicrobial use prevalence, supporting the view of AMR as a dynamic issue which requires regular review and reporting for National action plans to be effective [18,20]. Locally-led follow-up and repeated assessments are essential to prioritize and refine interventions, in particular using standardized tools which allow to demonstrate evidence-based improvements [20].

Concerning antimicrobial stewardship initiatives, SPiNCAR appropriate antimicrobial use score significantly correlated with lower antimicrobial use prevalence in acute-care hospitals. A previous analysis of community antibiotic consumption in Italy found important regional disparities and highlighted the need for interventions to improve prescribing appropriateness targeted towards specific population subgroups [14]. Results of a systematic review of the impact of Japan’s National action plan on antimicrobial prescribing practices also suggest a positive impact in terms of promotion of antimicrobial stewardship activities and reduced antimicrobial consumption [21].

This study had several limitations. First, as a majority but not all Italian regions were included, and as participation was voluntary, we cannot exclude that only those with the highest interest in AMR and implementation levels participated, potentially leading to participation bias. Regions that chose to participate may differ systematically from those that did not. This may affect both the representativeness of national findings and the interpretation of the observed correlations. Responding regions may not reflect the performance or characteristics of non-responding regions, leading to a potential overestimation of the national implementation level. Further, data collection through SPiNCAR is performed as a self-assessment Other limitations pertaining to data collection include possible misunderstandings of the concepts and definitions underlying the SPiNCAR framework, and the fact that filling out additional criteria was only possible if all minimum criteria for the relevant standard were met, leading to a potential underestimation of additional scores. This approach avoids disadvantaging regions that did not assess optional items and limited heterogeneity arising from differing maturity levels across regions. Excluding additional criteria may have underestimated progress in regions implementing advanced practices not captured by minimum criteria. This could slightly compress observed variability at the upper end of performance within some domains.

It must be noted that an ecological, cross-sectional analysis cannot establish causality, and can lead to ecological bias. We used ecological-level Spearman correlations in order to analyse SPiNCAR results and external benchmarks (PPS indicators and antimicrobial consumption) at regional level. Spearman’s rho is appropriate in this context because it is non-parametric, does not assume linearity or normality, and is robust to outliers, which is advantageous with a small sample (n = 13) and potentially skewed indicators. As a construct (convergent) validity check, these correlations provide an interpretable, assumption-light assessment of whether regions with higher SPiNCAR implementation tend to show more favorable external indicators [22]. Considering correlation analyses, assumptions cannot be made on the direction of relations between investigated variables. In particular, based on our analyses we cannot ascertain whether positive results are due to pre-existing levels (and for how long) or to activities outlined by the National action plan. Pre-existing differences in regional healthcare capacity (e.g., staffing levels, availability of dedicated IPC/AMS teams, laboratory and IT resources, and surveillance infrastructure) may systematically influence SPiNCAR performance, with better-resourced regions more able to implement and document required activities—potentially yielding higher scores independent of underlying AMR/HAI burden. Results of our analyses are exploratory; we advise cautious interpretation. The different observation windows may have introduced temporal misalignment across indicators. Further, as the assessment period (PPS in late 2022 and SPiNCAR status at the end of 2022) overlapped with the COVID-19 pandemic, regional IPC capacity and surveillance systems may have been temporarily strengthened or, conversely, strained by staffing shortages and competing priorities. The pandemic may also have influenced antimicrobial prescribing patterns (e.g., changes in empiric use and case-mix) and the implementation and measurement of stewardship activities, potentially affecting SPiNCAR domain scores and contributing to residual confounding in the observed regional differences.

Finally, it was beyond the scope of our analysis to take into account the impact of confounders such as heterogeneity between regions/AP in terms of size, demography, and dedicated economic and human resources. We adopted unweighted regional correlations to preserve equal representation of each unit and to avoid results being driven primarily by a few large regions, which would be expected under population or bed-day weighting. Given this choice, we acknowledge that unweighted analyses may not reflect the national burden; a sensitivity analysis using population (or bed-days) as weights is required to assess the robustness of the main associations. In-depth analyses of political and decision-making processes would also provide a deeper understanding of regional AMR responses.

In conclusion, this study provides preliminary evidence of alignment between SPiNCAR scores and surveillance indicators, suggesting the validity of the SPiNCAR tool to monitor the level of implementation of the Italian National action plan to combat AMR. However, it must be noted that these results do not represent a full validation of the tool.

This study provided a national baseline for future reference and facilitated the comparison of different regional realities. Results are encouraging, but there is still much work to be done. Following the 2022–2025 PNCAR update, SPiNCAR was revised by updating the indicator set to align with the objectives and actions defined in the updated plan. In parallel, data-collection procedures and the web platform were refined to improve standardisation and usability, including: removal of the core/additional criteria distinction, introduction of a “not applicable” response option where relevant, addition of free-text note fields, expanded response formats beyond binary yes/no, and implementation of cascade questions. Criteria were also re-labelled from mandatory/optional to primary/secondary, where primary criteria are explicitly linked to PNCAR 2022–2025 indicators and secondary criteria reflect good practices inherited from the previous SPiNCAR version. Repeated assessments should be conducted to foster continuous improvement, particular focus should be dedicated to reducing regional inequalities and adopting an increasingly integrated approach among human, veterinary, and environmental health. The ongoing commitment of institutions, along with greater collaboration among regions, will be crucial to successfully addressing the global challenge posed by AMR and ensuring the sustainability of the actions taken in the long term.

Supporting information

Acknowledgments

The authors gratefully acknowledge personnel of Regional health authorities who collaborated in data collection. Members of the working group: Antonella Agodi, Luca Arnoldo, Francesco Auxilia, Martina Barchitta, Silvana Castaldi, Giuseppe Diegoli, Silvia Forni, Fabrizio Gemmi, Enrico Ricchizzi, Giancarlo Ripabelli, Giovanna Trambajolo, Elena Vecchi.