More high-risk patients but stable levels of healthcare-associated infections and antibiotic use: findings from national point prevalence survey conducted in 2017 and 2022 in the Paris region hospitals

Abstract

Background

Monitoring healthcare-associated infections (HAI) and antibiotic use (ABU) in hospitalized patients is essential to control infections and reduce the inappropriate use of antibiotics. This study aimed to describe trends in the prevalence of HAI and ABU between 2017 and 2022 within a group of healthcare facilities in the Paris region.

Methods

The study focused on data collected during point-prevalence surveys (PPS) conducted in hospitals in 2017 and 2022. Only facilities participating in both surveys and departments enrolling at least 10 patients were included in the analysis. Patient characteristics, HAI, and ABU were compared between 2017 and 2022, excluding COVID-19 infections in 2022, except in cases of co-infection with an HAI. Multivariate regression analysis was performed to measure the prevalence of HAI and ABU.

Results

Overall, 79 facilities were included representing 19 211 patients in 2017 and 16 266 in 2022. The overall HAI prevalence remained stable at 6.7%, despite an increase in patient risk factors, including higher proportions of immunocompromised patients (17.5% vs. 16.2%, p = 0.001) and greater use of invasive devices (48.1 vs. 42.6%, p < 0.001). Pneumonia prevalence increased in 2022 (1.3% vs. 1.1%, p = 0.03). In paediatric units, HAI prevalence decreased markedly (1.1% vs. 4.2%, p < 0.001). The prevalence of patients receiving antibiotics was also stable (20.3% in 2017; 21.1% in 2022). In 2022, monotherapy was more frequent (73.8% vs. 71.0%, p = 0.001), and intravenous administration increased (60.0% vs. 53.7%, p < 0.001). The use of fluoroquinolones and aminoglycosides have declined (5.8% vs. 8.3% and 1.6% vs. 3.0%, p < 0.001). The prevalence of patients receiving antibiotics declined in gynaecology-obstetrics (10.8% vs. 14.0%, p = 0.021). Factors significantly associated with HAI and ABU included gender, McCabe score, immunosuppression, cancer and vascular catheterization. After adjustment for these factors, HAI prevalence was significantly lower in 2022 than in 2017 (prevalence ratio 0.91, p = 0.033).

Conclusions

In this group of Paris-area hospitals, the stability in HAI in 2022 compared with the 2017 PPS, despite an increase in patient risk factors, may reflect an improvement in infection prevention practices. The overall prevalence of ABU remained stable but changes in molecular class indicated that prescribing practices improved. Efforts must continue with the support of mobile teams specialising in infection control and prevention and antibiotic stewardship.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13756-026-01804-w.

Phone number: +33 1 87 05 32 40.

Background

Healthcare-associated infections (HAI) and antimicrobial resistance (AMR) infections represent a major threat to the quality and safety of care in hospitalized patients. HAI are known to have an impact on morbidity, longer hospital stays, and mortality [1–4]. AMR is already a major public health challenge, with an estimated 5 million deaths attributed to it in 2019, and is expected to become an even greater threat in the coming years [5, 6]. The World Health Organization (WHO) has included it among its ten public health priorities [7]. It is well established that the use of antibiotics must be restricted to appropriate cases (confirmed bacterial infections, or where there is a strong clinical suspicion of a bacterial infection) and to use antibiotics with the narrowest possible spectrum of activity whilst maintaining the same level of efficacy, as antibiotic consumption is undoubtedly the major determinant of antibiotic resistance [8].

Monitoring is an essential component of efforts to improve healthcare practices, both in terms of infection prevention and the optimal use of antibiotics [9]. The WHO has identified it as a priority in its global action plan on antimicrobial resistance [10]. In hospitals, in addition to longitudinal surveillance within networks, point prevalence surveys (PPS) allow for the rapid collection of standardized data on the types and the prevalence of major infections, as well as selected information on the quality of antibiotic prescribing.

In France, a national point prevalence survey has been conducted every five years since 1996 to estimate the prevalence of HAI and antibiotic use (ABU) [11]. The most recent survey, conducted in 2022, included 151 676 patients from 1 155 healthcare facilities. This represented 42% of all healthcare facilities nationwide, as participation in the survey was voluntary. Thus, national prevalence estimates are derived using statistical weighting to account for non-participating healthcare facilities and ensure representativeness. In 2022 the prevalence of HAI (excluding COVID-19) was 5.7% [95 CI: 5.4–5.9], a rate that did not significantly differ from those reported in the 2012 and 2017 surveys [12]; while antibiotic use reached 16.2% [95 CI: 15.7–16.8] in 2022, representing a slight increase compared with 15.1% [95 CI: 14.2–16.1] in 2017.

The Paris region (Île-de-France), the most densely populated in France, contains the country’s largest concentration of hospital facilities and is characterized by marked socioeconomic disparities [13]. Monitoring data indicate that AMR levels are higher than the national average, both in the community and in hospital [14, 15]. The study aimed to analyse changes in the prevalence of HAI and antibiotic use in healthcare facilities that participated in both the 2017 and 2022 surveys.

Methods

Design and setting

The PPS is conducted by the French public health agency (Santé Publique France, SPF). All private and public healthcare facilities are encouraged to participate. This national point prevalence study is part of the European Point Prevalence Survey coordinated by the European Centre for Disease Prevention and Control (ECDC), to which France contributes data [16]. The last two PPS were conducted on a given day between May 15 and June 30, 2017 and 2022.

All inpatient units were considered, including acute care, rehabilitation care and long-term care units. All patients present before 8 a.m. on the day of the survey were included. Data collection was carried out by trained investigators following the national methodology with a reference guide providing standardized definitions [17, 18]. Data were collected through discussions with the physician responsible for the patient in the unit and from patient medical records. Individual characteristics (age, sex, date of admission) and various infection risk factors such as McCabe score (which classifies the severity of underlying medical conditions), immunosuppression, solid cancer or hematologic malignancy, and invasive devices were recorded. For the McCabe score, categories 1 (fatal illness within five years) and 2 (ultimately fatal disease within a year) have been grouped together because this assessment is relatively subjective and both issues have the same outcome, in opposition to category 0 (non-fatal disease) [19]. Midlines were not included among the possible invasive procedures in the 2017 survey, unlike in the 2022 survey. Data on HAI included the site of infection and any microorganisms that had been identified. Data on antibiotic use included dosage, route of administration, indication, duration, and whether the treatment had been escalated or de-escalated. Antibiotics were classified according to the WHO Anatomical Therapeutic Chemical (ATC) classification system [20].

Furthermore, a general questionnaire on the organisation of the facility was completed once by the healthcare professional responsible for carrying out the survey, indicating the number of beds, annual admissions and annual patients-days, types of units (medicine, surgery, intensive care unit, gynaecology-obstetrics, psychiatry, rehabilitation, or long-term care), legal status (private, public, private non for profit), type of establishment (public teaching, military-teaching, non-teaching-public, non-teaching-private, cancer specialized, psychiatric, rehabilitation centre, long term-care) and location. All data were entered into an electronic Case Report Form (eCRF) provided by SPF, where each patient file was anonymized using a code number.

The Paris region is the region with the largest population in France, the youngest, and with the greatest economic inequalities [13]. It comprises 470 healthcare facilities with 31 000 beds, roughly half allocated to medicine, surgery, and obstetrics and the other half to rehabilitation hospitals [21]. Approximately, half of the beds for medicine, surgery, gynaecology-obstetric are provided by the teaching hospital system.

Change from 2017 to 2022

Data were obtained from Île-de-France facilities participating in both PPS (2017 and 2022), including only wards with at least 10 patients over the two surveys. As in the analysis carried out by SPF, all HAI were included, whether they were acquired in the facility during the current stay or a previous stay, or in another healthcare setting. COVID-19 infections were excluded from the 2022 analysis except in cases of co-infection with HAI. Comparisons of antibiotic use were performed for selected antibiotic and classes or grouped according to their spectrum of activity and common use.

Statistical analysis

Patient characteristics, HAI prevalence, antibiotic use and antimicrobial resistance data were compared between 2017 and 2022 according to facility type and size, type of activity and patient risk factor. Categorical variables were reported as percentages and analyzed using the χ² test or Fisher’s exact test depending on expected frequencies. Continuous variables were presented as medians [1st − 3rd quartiles] and compared using the Wilcoxon–Mann–Whitney test. Univariable analyses were performed using logistic regression to assess associations between patient characteristics and HAI. Variables of interest were then included in multivariable Poisson regression models with a log link to estimate associations between patient characteristics (age category, gender, immunosuppression, cancer, vascular catheterization, McCabe score, and year of study) and HAI. Results were expressed as prevalence ratios and 95% confidence intervals. The same approach was applied to assess associations between patient characteristics and antibiotic use. Missing data were handled using multiple imputation by chained equations (MICE), generating five imputed datasets. Logistic regression was used for binary variables (immunosuppression and cancer), polynomous regression for categorical variables (age category), and classification and regression trees for the McCabe score. Variables without missing data were not imputed. A statistical significance level of p < 0.05 was applied. The analyses were performed using R^® 4.3.2 and STATA 11^® software.

Results

A total of 35 477 patients were included in the analysis (19 211 in 2017 and 16 266 in 2022), representing nearly 65% of the regional database. Patients were hospitalized across 79 facilities with the majority in the public sector (72%). Most patients were admitted to public teaching hospitals, non-teaching public hospitals with over 300 beds and non-teaching private hospitals with over 100 beds (respectively 35, 28 and 17%). In both years, most patients were hospitalized in adult medicine (40%, followed by rehabilitation and surgical sectors (17 and 16%, see the supplementary Table 1). Among paediatric patients, nearly half were under 1 year of age (334/789, or 42% in 2017 and 357/743, or 48% in 2022). This distribution remained unchanged between 2017 and 2022.

Patient characteristics

In 2022 as in 2017, 46% of patients were men with a median age of 68 (in 2022: [Q1 = 45 - Q3 = 82]). The proportion of immunocompromised patients was significantly higher in 2022 than in 2017 (17.5% vs. 16.5% in 2017, p = 0.001), as was the proportion of patients with a disease considered fatal within one or five years, according to the McCabe score (38.6% vs. 35.7%, p < 0.001, Table 1). The day of the survey, patients were hospitalized for a median of 8 days [Q1 = 3 - Q3 = 29] in 2022 (no difference with 2017) and 18% had undergone surgery since admission. The proportion of patients with an invasive medical device was significantly higher in 2022 than in 2017 (48.1% vs. 42.6%, p < 0.001), particularly for vascular access (46.0% vs. 40.4%, p < 0.001) and peripheral vascular catheter (32.1% vs. 26.9%, p < 0.001). The proportion of patients with a peripheral inserted central catheter (PICC) were also significantly higher in 2022 (1.7% vs. 1.1%, p < 0.001). Among patients with catheters, most had only one catheter (94%). The proportion of patients with peripheral vascular catheter (70% vs. 67%, p < 0.001) and PICCs (3.5% vs. 2.5%, p < 0.001) increased, while the proportion of patients with implantable venous access port, subcutaneous catheter, central veinous catheter (excluding PICC) and arterial catheter significantly declined. The proportion of intubated patients did not vary between the two years (1.3% in 2017 as in 2022, p = 0.745).

Table 1.

Patient characteristics and evolution of prevalence of healthcare infections and antibiotic use. Significant p-values are shown in bold

				Prevalence of HAI (%)				Prevalence of antibiotic use (%)
	2017	2022	P value	2017	2022	p-value	Trend	2017	2022	p-value	Trend
Total	19 211	16 266		6.7	6.7	0,989		20.3	21.1 426	0.075
Man	8 834 (46.0)	7 437 (45.7)	0.620	7.6	7.3	0.480		23.9	23.8	0.854
Woman	10 377 (54.0)	8 829 (54.3)		5.4	5.5	0.789		17.2	18.8	0.005	↗
Age ≥ 65 years	10 580 (55.1)	9 044 (55.6)	0.319	7.4	7.1	0.436		19.7	21.3	0.006	↗
Immunodeficiency	3 062 (16.2)	2 801 (17.5)	0.001	14.0	11.8	0.014	↘	38.4	37.1	0.314
McCabe score			< 0.001
0 (non fatal disease)	11 011 (64.3)	9 213 (61.4)		3.7	4.0	0.295		16.7	18.5	< 0.001	↗
1 or 2 (fatal disease within 5 or 1 year)	6 101 (35.7)	5 796 (38.6)		11.3	10.0	0.029	↘	27.9	25.4	0.002	↘
Progressive cancer
None	15 231 (81.5)	12 957 (81.4)		5.1	5.0	0.750		18.0	18.5	0.232
Solid cancer	2 791 (14.9)	2 426 (15.2)		10.5	10.3	0.853		25.4	25.8	0.715
Hematologic malignancy	659 (3.5)	538 (3.4)		20.8	18.2	0.265		54.5	55.4	0.752
Invasive device	8 175 (42.6)	7 823 (48.1)	< 0.001	11.7	10.5	0.024	↘	36.3	35.1	0.136
Vascular catheter	7 769 (40.4)	7 489 (46.0)	< 0.001	11.7	10.6	0.028	↘	37.0	35.9	0.163
Peripheral venous catheter	5 169 (26.9)	5 217 (32.1)	< 0.001	9.9	8.6	0.022	↘	38.5	37.3	0.223
Arterial catheter	275 (1.4)	204 (1.3)	0.149	23.3	32.8	0.020	↗	53.1	58.3	0.254
Subcutaneous catheter	895 (4.7)	707 (4.3)	0.158	8.7	5.9	0.036	↘	14.7	11.2	0.036	↘
Central venous catheter	751 (3.9)	597 (3.7)	0.241	27,8	26.6	0.624		57.7	52.9	0.083
Venous umbilical catheter	12 (0.1)	11 (0.1)	0.849	0	0	–		41.7	36.4	> 0.999
PICC	202 (1.1)	280 (1.7)	< 0.001	25.3	23.9	0.740		55.0	46.4	0.065
Implantable venous access port	1 064 (5.5)	848 (5.2)	0.177	12.0	11.3	0.632		33.8	32.5	0.553
Midline	–	138 (0.8)		–	31.2	–		–	61.6
Urinary catheter	1 790 (9.3)	1 615 (9.9)	0.052	17.4	19.2	0.183		39.0	40.7	0.315
Intubation	252 (1.3)	207 (1.3)	0.745	26.2	29.5	0.435		59.1	53.1	0.198
Surgery since admission	3 465 (18.0)	2 980 (18.3)	0.490	10.8	11.2	0.542		27.1	27.6	0.645

Healthcare associated infection

The proportion of patients with at least one HAI (excluding COVID-19) remained stable between 2017 and 2022 (1026/16 266 patients or 6.3% [95 IC: 6.0–6.7] in 2022, vs. 1231/19 211 or 6.4% [95 CI: 6.1–6.8] in 2017). As a patient may have several HAI, when excluding COVID-19, the point prevalence of all HAI was 6.7% both year (1090/16 266 in 2022 and 1288/19 211 in 2017).

Urinary tract infections, pneumonia, surgical site infections, and bloodstream infections constituted the four main HAI sites in 2017 and 2022, accounting each year for around 70% of all documented infections amongst all HAI (789/1090 = 72.4% in 2022 and 878/1288 = 68.2% in 2017). The prevalence of pneumonia was significantly higher in 2022 (218/16 266 = 1.3% vs. 206/19 211 = 1.1%, p = 0.03), whilst the proportion of intubated patients among patients with healthcare-associated pneumonia was similar (37/218 = 17.0% in 2022 and 34/206 = 16.5% in 2017, p = 0.90; Fig. 1). The prevalence of other infections was stable. The median time between hospital admission and HAI onset was 15 days [7–30] in 2022, with no difference compared with 2017.

Fig. 1.

Main infection sites and prevalences (%)

Prevalence of HAI according to the type of healthcare facilities and ward and according to the demographic and care characteristics

The prevalence of infected patients did not differ significantly between the two years according to the category of facilities, except for the “medicine, surgery, gynaecology-obstetric with fewer than 100 beds” category, where it was lower in 2022 than in 2017. According to the type of ward, prevalence was also similar between the two years except in paediatrics where it decreased significantly in 2022 (1.1% vs. 4.2%, p < 0.001). HAI prevalence was significantly lower in 2022 among immunocompromised patients compared with 2017 (11.8 vs. 14.0%, p = 0.014) and among patients with a McCabe score of 1 or 2 (10.0% vs 11.3%, p = 0.029, Table I). Additionally, HAI prevalence among patients with invasive devices, particularly vascular catheters, decreased significantly in 2022 compared with 2017 (10.5% vs 11.7%, p = 0.024; 10.6% vs 11.7%, p = 0.028).

Characteristics of infected patients

No significant differences in demographic characteristics were observed between 2017 and 2022 within infected patients. However, more infected patients received antibiotics (94.5% vs. 91.1%, p = 0.002), had a PICC line catheter (6.5% vs. 4.1%, p = 0.01) and urinary catheter (30.2% vs. 25.4%, p = 0.01) in 2022. Conversely, the proportion of infected patients with a subcutaneous catheter was significantly lower (4.1% vs. 6.3%, p = 0.02).

Microbiology of healthcare infections

HAI were less frequently microbiologically documented in 2022 than in 2017 (951/1090 = 87.2% vs. 2017: 1179/1288 = 91.5%, p = 0.001). In 2022 the most isolated micro-organisms were E. coli (21.3%), S. aureus (12.2%), Enterococcus spp (11.0%), P. aeruginosa (8.0%) and K. pneumoniae (7.3%), with no significant difference compared with 2017 (see supplementary Table 2). The proportion of methicillin-resistant Staphylococcus aureus (MRSA) among S. aureus did not differ significantly between 2022 and 2017 (16/115 = 13.9% vs. 25/142 = 17.6%, p = 0.42), nor did the proportion of extended-spectrum beta-lactamases (ESBL; 60/391 = 15.3% vs. 87/479 = 18.2%, p = 0.27) or carbapenem-resistant Enterobacterales (4/382 = 1.0% vs. 3/479 = 0.6%).

Antibiotic use

Antibiotic therapy was administered to 3899 patients in 2017 and 3426 in 2022, corresponding to prevalences of 20.3% and 21.1% respectively, with no significant difference. When stratifying by ward type, no significant change was observed compared to 2017, except in obstetrics and gynaecology wards where it decreased from 14.0% to 10.8% (p = 0.021).

Prevalence of antibiotic use according to the demographic and care characteristics

Antibiotic prevalence was significantly higher in 2022 among patients aged 65 and over (21.3% in 2022 vs. 19.7% in 2017, p = 0.006, Table 1). Conversely, a significant decrease was observed among patients with a fatal McCabe score (i.e. equal to 1 or 2, from 27.9 to 25.4% in 2022, p = 0.002). Regarding care characteristics, antibiotic use remained largely stable, except in patients with subcutaneous catheters where it declined from 14.7% in 2017 to 11.2% in 2022 (p = 0.036).

Characteristics of patients receiving at least one antibiotic

In 2022, patients receiving antibiotics were slightly older (median 68 vs. 67 years, p = 0.023) and included fewer men (51.6% vs. 54.1% in 2017, p = 0.030). Regarding care characteristics, the use of invasive devices among patients receiving antibiotic increased significantly from 76.0% to 80.2% in 2022 (p < 0.001). This increase particularly concerned vascular catheters and more specifically peripheral venous catheters and PICCs. On the other hand, the use of subcutaneous catheters declined (3.4 to 2.3%, p = 0.006) as well as central venous catheters (11.1 to 9.2%, p = 0.008).

Indication

In both years the main indication for antibiotic use was curative (80% in 2022, Fig. 2a), with respiratory, digestive and urinary tract being the most frequently targeted sites. The proportion of antibiotics prescribed for treatment of community-acquired infections compared with those prescribed for HAI treatment remained stable (66% in 2017 and 68% in 2022, p = 0.2). In 2022, curative antibiotic treatments lasting ≥ 7 days declined (23% vs. 26%, p = 0.009). In case of modification of curative treatment (21% in 2017 and 24% in 2022), the proportion of treatment escalation increased significantly in 2022 (39.4% vs. 31.6% in 2017, p < 0.001). The proportion of de-escalation remained stable at 37% for both years. Medical and surgical prophylaxis accounted for 11% and 5% of antibiotic use in 2022 (13% and 6% in 2017, Fig. 2a). In 40.6% (124/305 in 2017) and 41.7% (101/242 in 2022) of cases, surgical prophylaxis lasted longer than 48 h.

Fig. 2.

Prevalence of patients on antibiotics according to the context of administration (2a) and top 6 of monotherapies (2b)

Molecules and classes of antibiotics

Each year, amoxicillin-clavulanic acid, ceftriaxone, sulfamethoxazole-trimethoprim, piperacillin-tazobactam and amoxicillin accounted for nearly half of all antibiotics administered (Table 2). In 2022, a significant increase was observed in the penicillin group, notably piperacillin-tazobactam (8.5% vs. 5.8% in 2017, p < 0.001) and cefotaxime (7.3% vs. 4.8 in 2017, p < 0.001) whereas the proportions of fluoroquinolones, antitubercular and aminoglycosides decreased. Changes over time in each molecule or family are shown by indication in Supplementary Fig. 1.

Table 2.

Changes over time in each administered antibiotic (classes and mainly molecules). Significant p-values are shown in bold

	2017 N = 5252	2022 N = 4467	p-value	Trend
Penicillins	1689 (32.16%)	1523 (34,09%)	0.043	↗
Amoxicilline-clavulanic acid	814 (15.50%)	666 (14.91%)	0.420
Piperacillin-tazobactam	303 (5.77%)	378 (8.46%)	< 0.001	↗
Amoxicillin	431 (8.21%)	365 (8.17%)	0.949
Cephalosporins	1155 (21.99%)	1041 (23,30%)	0.123
Ceftriaxone	540 (10.28%)	425 (9.51%)	0.207
Cefotaxime	251 (4.78%)	328 (7.34%)	< 0.001	↗
Cefazolin	151 (2.88%)	107 (2.40%)	0.143
Carbapenems	152 (2.89%)	127 (2.84%)	0.881
Meropenem	44 (0.84%)	85 (1.90%)	< 0.001	↗
Sulfonamides	460 (8.76%)	412 (9.22%)	0.424
Cotrimoxazole	458 (8.72%)	409 (9.16%)	0.453
Imidazole derivatives	366 (6.97%)	300 (6.72%)	0.623
Metronidazole	350 (6.66%)	299 (6.69%)	0.954
Fluoroquinolones	435 (8.28%)	258 (5.78%)	< 0.001	↘
Levofloxacin	110 (2.09%)	121 (2.71%)	0.048	↗
Ciprofloxacin	129 (2.46%)	64 (1.43%)	< 0.001	↘
Ofloxacin	174 (3.31%)	68 (1.52%)	< 0.001	↘
Macrolides and derivatives	247 (4.70%)	234 (5.24%)	0.225
Clindamycin	47 (0.89%)	74 (1.66%)	< 0.001	↗
Azithromycine	40 (0.76%)	44 (0.99%)	0.236
Spiramycine	77 (1.47%)	67 (1.50%)	0.891
Aminoglycosides	159 (3.03%)	78 (1.75%)	< 0.001	↘
Tetracyclins	40 (0.76%)	42 (0.94%)	0.337
Antitubercular	285 (5.43%)	161 (3.60%)	< 0.001	↘
Rifampicin	111 (2.11%)	77 (1.72%)	0.164
Glycopeptides	153 (2.91%)	120 (2.69%)	0.500
Vancomycin	146 (2.78%)	107 (2.40%)	0.235
Others	108 (2.06%)	163 (3.65%)	< 0.001	↗
Linezolide	29 (0.55%)	48 (1.07%)	0.004	↗
Daptomycine	25 (0.48%)	61 (1.37%)	< 0.001	↗

The category “Others” counts for fidaxomicin, colistin, fusidic acid, furans, fosfomycin, linezolid, tedizolid, daptomycin.

Number of antibiotics per patient and routes of administration

Monotherapies dominated both years and increased in 2022 (74% vs. 71%, p = 0.008), particularly in medicine (70 to 75%, p = 0.001) and paediatrics (66 to 76%, p = 0.049), while it decreased in gynaecology/obstetrics (82% vs. 89% in 2017). The five most commonly administered antibiotics in monotherapy remained the same although their relative distribution shifted in 2022 (Fig. 2b). Amoxicillin-clavulanic acid remained the most commonly molecule although its proportion decreased slightly (from 25% in 2017 to 23% in 2022, p = 0.098). Cefotaxime use remained lower than that of ceftriaxone and overall use of these injectable third generation cephalosporin (3GC) inactive against Pseudomonas aeruginosa remained stable at 11%.

Intravenous administration was more frequently used in 2022 than in 2017 (60.0% vs. 53.7%, p < 0.001) while oral administration decreased from 43% to 38% (not significant) and subcutaneous and other routes remained marginal. For the main antibiotics that can be administered either intravenously or orally (i.e. amoxicillin, amoxicillin-clavulanic acid, metronidazole, quinolones, and macrolides) the distribution of administration routes did not change significantly between the two years.

Trends in prevalence and risk factors

Risk factors associated with HAI as well as ABU were masculine gender (p < 0.001), age (p < 0.001, except for < 15 years for HAI and ≥ 85 years age group for ABU), McCabe score 1 or 2 (p < 0.001), immunosuppression (p < 0.001), solid cancer and hematologic malignancy (p < 0.001) and vascular catheterization (p < 0.001, Table 3). The imputed regression model adjusted for these various factors showed that the HAI prevalence significantly decreased in 2022 (prevalence ratio (PR) 0.91 [0.84–0.99], p = 0.033). The results were similar using modelling without imputation (N = 31 541, PR = 0.91 [0.83–0.99], p = 0.031).

Table 3.

Univariate and multivariate analysis of risk factors for HAI and antibiotic prevalence. Significant p-values are shown in bold

Characteristic	N	HAI prevalence				Antibiotic prevalence
		Univariate		*Multivariate		Univariate		*Multivariate
		OR (95 CI)	p-value	PR (95 CI)	p-value	OR (95 CI)	p-value	PR (95 CI)	p-value
Gender	35 477
Man		ref		–		ref		–
Woman		0.71 (0.65, 0.77)	< 0.001	0.83 (0.76, 0.90)	< 0.001	0.70 (0.66, 0.74)	< 0.001	0.85 (0.81, 0.89)	< 0.001
Age	35 469
15–44		ref		–		ref		–
< 15		0.97 (0.74, 1.25)	0.8	0.88 (0.67, 1.13)	0.3	1.31 (1.16, 1.47)	< 0.001	1.09 (0.98, 1.21)	0.12
45–64		1.96 (1.68, 2.30)	< 0.001	1.27 (1.09, 1.49)	0.002	1.36 (1.25, 1.48)	< 0.001	0.99 (0.92, 1.07)	0.8
65–74		2.20 (1.88, 2.58)	< 0.001	1.33 (1.14, 1.57)	< 0.001	1.33 (1.22, 1.45)	< 0.001	0.97 (0.89, 1.05)	0.4
75–84		2.07 (1.77, 2.42)	< 0.001	1.40 (1.19, 1.64)	< 0.001	1.23 (1.13, 1.33)	< 0.001	1.01 (0.93, 1.09)	0.8
≥ 85		1.83 (1.57, 2.15)	< 0.001	1.38 (1.18, 1.63)	< 0.001	0.97 (0.89, 1.06)	0.5	0.95 (0.88, 1.03)	0.2
McCabe	32 121
Non fatal		ref		–		ref		–
Fatal		2.96 (2.70, 3.25)	< 0.001	1.73 (1.56, 1.92)	< 0.001	1.71 (1.62, 1.81)	< 0.001	1.09 (1.03, 1.15)	0.004
Immunodeficiency	34 868
No		ref		–		ref		–
Yes		2.79 (2.54, 3.06)	< 0.001	1.38 (1.22, 1.55)	< 0.001	2.92 (2.75, 3.10)	< 0.001	1.59 (1.49, 1.71)	< 0.001
Cancer	34 602
None		ref		–		ref		–
Solid cancer		2.17 (1.96, 2.41)	< 0.001	0.87 (0.77, 0.99)	0.032	1.54 (1.44, 1.65)	< 0.001	0.71 (1.49, 1.71)	< 0.001
Hematologic malignancy		4.57 (3.92, 5.32)	< 0.001	1.44 (1.21, 1.71)	< 0.001	5.45 (4.84, 6.13)	< 0.001	1.21 (1.09, 1.33)	< 0.001
Vascular catheter	35 477
No		ref		–		ref		–
Yes		4.50 (4.09, 4.98)	< 0.001	3.34 (3.03, 3.70)	< 0.001	6.01 (5.67, 6.38)	< 0.001	3.83 (3.63, 4.05)	< 0.001
Year	35 477
2017		ref		–		ref		–
2022		0.98 (0.90, 1.07)	0.7	0.91 (0.84, 0.99)	0.033	1.05 (1.00, 1.10)	0.075	0.96 (0.92, 1.01)	0.12

Associations between patient characteristics and health-care–associated infections (HAI) and antibiotic use (ABU). Univariable analyses were performed using logistic regression and are presented as odds ratios (OR) with 95% confidence intervals (CI). Multivariable analyses were conducted using Poisson regression with a log link to estimate prevalence ratios (PR) with 95% CIs. All variables listed were included in the multivariable models. Missing data were handled using multiple imputation by chained equations

Discussion

We assessed trends in the prevalence of HAI and ABU in more than 35 000 patients in 79 healthcare facilities in the Paris region, based on national PPS conducted in 2017 and 2022. These public and private institutions of varying sizes were located across all regional departments. Data were collected using a standardized methodology consistently applied for 30 years and compatible with the European prevalence survey.

The overall HAI prevalence was 6.7% in 2022. This rate appears to be higher than the rate recorded at national level (5.7%). However, within this group of 79 facility of Paris area, teaching hospitals and general hospitals with more than 300 beds are over-represented compared with the national population [12, 22]. Furthermore, as not all hospitals take part in the national survey, the French data are estimates, based on a sampling plan tailored to the actual demographic profile of the hospitals; it is therefore not methodologically possible to compare these two prevalences.

The 2022 HAI prevalence remained unchanged from 2017 within our group of hospitals. This stability is similar to that observed in other national surveys, starting with the survey in France (5.7% each year). In the same way, the European survey reported a prevalence of 6.5% in 2017 and 6.2% in 2022, excluding COVID (7.1% including it), whilst the Canadian survey also show that the HAI prevalence has remained stable in recent years (8.4% in 2017 and 8.1% in 2024) [16, 23]. Several studies have sought to measure the proportion of preventable HAI [24, 25]; it could be hypothesised that, as a result of the implementation of infection prevention and control (IPC) measures, the support from an IPC team, the regular conduct of PPS with feedback of the results and the establishment of surveillance networks, the preventable proportion of HAI has indeed been achieved by following these guidelines [9, 26]. Moreover, it is reasonable to assume that, in order to further reduce the HAI prevalence, preventive measures would need to be tailored to each patient’s level of risk [27].

However, the stabilization of HAI rates seems paradoxical given the hospitalized population presents more infection risk factors. Nevertheless, this finding has already been noted elsewhere [28, 29]. In our multivariate regression, vascular catheterisation tripled HAI prevalence and confirms immunosuppression and cancer as known risk factors to be associated with infections. Nevertheless, we observed a decrease in the proportion of infected patients among immunocompromised individuals and/or those with vascular catheters. More generally, our multivariate model showed a slight but significant decrease in the prevalence of infections in 2022 while adjusting on these risk factors. These finding may reflect an improvement in preventive practices, such as better hand hygiene, which has genuinely improved as a result of promotional initiatives [30]. In addition, guidelines recommend reassessing on a daily basis whether it is appropriate to continue using the invasive device. However, the data collected here do not allow us to determine whether the median duration of use of a vascular or urinary catheter has decreased. Furthermore, the 2022 survey was conducted in the post-COVID-19 period, when preventive measures (e.g. increased use of hydroalcoolic solution) were likely more rigorously applied [31–33]. Also, it is possible that doctors have improved their targeting of immunocompromised patients, and that prevention among these patients has consequently increased [34, 35]. However, the absence of detailed medical records limits the interpretation. Finally, the cross-sectional design of the study limits the ability to establish a causal link between the device and the infection, as well as the reason for inserting a catheter.

No variation was observed in the prevalence of infections across different anatomical sites, with the exception of pneumonia, for which a slight increase was noted (from 1.1% to 1.3%). In this group of facilities, this increase in hospital-acquired pneumonia (HAP) does not appear to affect intubated patients. This finding is not reflected in the national and European prevalence surveys carried out in 2022 [16]. Unlike ventilator-associated pneumonia (VAP), there is no globally agreed definition or set of diagnostic criteria for non-ventilator-associated hospital-acquired pneumonia (NV-HAP); it is therefore difficult to track changes in its incidence [36]. A longitudinal analysis of trends over time carried out at a US veterans’ hospital reports that, following a period of decline in NV-HAP, the post-COVID period has been marked by a 25% rise in this incidence among patients who have not had COVID; the authors hypothesised a multifactorial aetiology, including in particular the increased burden of care for patients with Covid or the fact that the focus is primarily on preventing VAP, to the detriment of individual measures aimed at preventing NV-HAP [37]. These measures - such as early mobility programs, oral care, or aspiration precautions - are in fact associated with a reduced risk of NV-HAP, and it is necessary to step up their implementation [38]. The data collected in the French PPS are insufficient to support these hypotheses in our population.

Moreover, the prevalence of infected patients remained stable across most units, except in paediatrics where HAI prevalence decreased from 4.2% to 1.1%. Whilst studies suggest that the impact of the preventive measures introduced during the COVID-19 pandemic is mixed, depending on the methodology, the type of HAI monitored or the departments involved, several studies carried out exclusively in paediatric facilities nevertheless report a significant reduction in HAI during the COVID pandemic and the reduction in infections mainly concerned respiratory viral infections [39–41]. In two Canadian and American studies, the authors point out that IPC measures, such as mask-wearing and visiting restrictions, which were already in place every winter (prior to the pandemic), were already recognised as effective in reducing the transmission of viral infections in specific types of departments such as neonatology [42]. In light of this previous data, it can be assumed that staff at these units were also more aware of the impact of these measures on children’s morbidity and were therefore more diligent in their compliance, compared to other units.

From a microbiological perspective, the documentation of HAI decreased in 2022. Knowledge of the microorganisms isolated by anatomical site remains crucial for guiding empirical antibiotic therapy. For example, Enterococcus was the second most frequently isolated pathogen in nosocomial urinary tract infections, whereas 3GC- which are inactive against this germ- are typically used on a probabilistic basis. Furthermore, the levels of antimicrobial resistance in HAI were similar in both surveys, which is relatively reassuring. However, this observed stability must be interpreted with caution before concluding that it reflects a genuine epidemiological trend, as the number of strains is small and the estimate of the proportion is therefore more susceptible to random variations. In this group, in 2022, 23% of Enterobacterales strains isolated exclusively from HAI were resistant to 3GC. This rate is higher than that found in the national survey (15.9%), probably explained by the inclusion of patients with more severe conditions (teaching hospitals). The same year, regional annual surveillance data collected in Île-de-France reported that 11.5% of Enterobacterales from all clinical samples were ESBL producers; however hospitals participation differed (27% in 2022) [43]. In addition, studies conducted in the Paris region indicated ESBL carriage rates of 14% in 2019 and 17% in 2024 [44, 45].

The proportion of patients receiving antibiotics remained stable from 2017 to 2022. Here again, this stability is consistent with the national French trend (15.1% to 16.2%). It should be noted that the prevalence in this region as in France is relatively low compared with other European or North American countries (latest European average of 35.5% in 2022) [16, 46–48].

The proportion of monotherapy increased significantly in 2022, suggesting that recommendations favouring monotherapies are becoming better adopted by clinicians. The decline in aminoglycoside use may reflect fewer indications for early dual therapy in severe infections. In obstetrics and gynaecology, overall antibiotic use decreased in 2022, while dual therapy increased at the expense of monotherapy, possibly reflecting the 2018 update of French guidelines on the management of membrane rupture [49]. There was a noticeable change in the choice of molecules: the reduced use of quinolones likely reflects safety warnings issued regarding adverse effects beyond the known impacts on the microbiota [50, 51]. More generally, this class of antibiotics has been markedly reduced in both hospitals and private practices in France [52]. Conversely, penicillin use increased in 2022. Overall, the use of amoxicillin–clavulanic acid decreased, consistent with recommendations to restrict this microbiota-selecting agent, including in medical prophylaxis (4.0% in 2017 vs. 2.5% in 2022). Finally, the increased use of therapeutic escalation in 2022 in cases of curative treatment may reflect efforts to limit the empirical use of antibiotics with greater ecological impact in non-severe cases [53, 54].

It is also worth noting that, compared with 2017, there was a decrease in antibiotic use in 2022 among patients with a McCabe score of 1 or 2, i.e., those whose general condition is impaired overall. This trend was also observed in the population of immunocompromised patients, but was not statistically significant. One might think that strategies for the antimicrobial stewardship (AMS) in immunocompromised patients are becoming more widespread within the medical community. In haematology, for example, the 2013 European guidelines recommend, under certain conditions, the early discontinuation of antibiotics following an episode of fever of unknown origin in a neutropenic patient [55]. These guidelines were already in place in 2017, but we know that it takes time for recommendations to be implemented in clinical practice after they are published [56].

The proportion of curative antibiotic treatments lasting more than seven days remains high (23% in 2022). Although, there has been a shift toward shorter treatment duration, severe infections—especially in hospitalized acute care patients—still require prolonged therapy [57, 58]. In addition, surgical antibiotic prophylaxis remains an area for improvement, as more than 40% of cases exceeded 48 h. We also observed an increase in parenteral antibiotics administration reaching 60% in 2022. Even if this prevalence remains below the European average of 80.3% [16], this finding should be interpretated in the context of the higher proportion of patients with vascular catheters, without being able to determine whether catheter insertion was driven by antibiotic therapy itself. In any case, this practice should alert healthcare providers as invasive devices are known to increase the risk of infection [59, 60].

Overall, these encouraging results – both in terms of maintaining a stable HAI prevalence amongst a population that is nevertheless at greater risk, and in terms of improvements in prescribing practices – also highlight the commitment of local support teams to implementing these programmes [61]. Furthermore, it has also been established that AMS and IPC measures are more effective when implemented together than when implemented separately [62].

Our surveys have several limitations. As prevalence studies, they cannot provide detailed information on patient case mix or specific risk factors for HAI and antibiotic prescribing. The lack of individual clinical or microbiological data prevents accurate assessment of prescription quality. Nevertheless, the data allow identification of major trends in relation to current recommendations. The 79 included facilities do not fully represent regional healthcare, where private facilities with fewer than 100 beds are more numerous. This is linked to voluntary participation in the national survey which is time-consuming and to our exclusion criteria of departments with fewer than 10 patients, particularly affecting intensive care units in private clinics. Thus, university hospitals and hospitals with over 300 beds are overrepresented. Differences in patient’ profile, organizations, human resources, and access to continuing education limit generalizability to the entire Paris region.

Conclusion

In conclusion, despite an increase in risk factors, the prevalence of HAI and antibiotic use remained stable between 2017 and 2022 in this group of Paris region hospital, as did the resistance of isolated strains from HAI infections. While adjusting for patients’ characteristics and vascular catheterization, HAI prevalence was actually significantly lower in 2022. This may reflect an improvement in IPC practices and highlights the importance of ensuring that these measures are maintained over time. Likewise, noticeable shifts in prescribing practices were observed toward a more appropriate use of antibiotics. These findings support the role of local IPC and AMS teams and encourage sustaining these efforts. Analysis of the 2027 survey will be essential to determine whether these trends persist.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ABU

Antibiotic use

AMR

Antimicrobial resistance

AMS

Antimicrobial stewardship

eCRF

Electronic Case Report Form

CI

Confidence Interval

ECDC

European Centre for Disease prevention and Control

3GC

Third generation cephalosporin

HAI

Healthcare-associated infection

IPC

Infection prevention and control

IQR

Interquartile range

Mice

Multivariate Imputation by Chained Equation

NV-HAP

Non-ventilator hospital-acquired pneumonia

OR

Odds ratio

PICC line

Peripherally inserted central catheter

PPS

Point prevalence survey

PR

Prevalence ratio

SPF

Santé Publique France

VAP

Ventilator-associated pneumonia

WHO

World Health Organization

Author contributions

FS: concept, study design, interpretation, writing and revision; RB: analysis and interpretation, and revision; BQ: analysis and interpretation, and revision; FL: interpretation and revision; CD: data collection; PA: interpretation, and revision; ES: analysis and interpretation, and revision; M-AB: interpretation, and revision. All authors read and approved the final manuscript.

Funding

No funding was received for this research project.

Data availability

Data from the 2017 and 2022 national prevalence surveys for the Île-de-France region are available on request to Santé Publique France.

Declarations

Ethics approval and consent to participate

The two investigations conducted by SPF were authorised by the competent authorities. The data were collected anonymously on an eCRF.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.