Study selection and basic characteristics of included studies

A total of 5242 studies have been identified in the literature search. After removing duplication, 4130 titles and abstracts have been screened. 4096 studies were deleted after reading the titles and abstracts based on the inclusion and exclusion criteria. We obtained and examined a total of 34 full-text articles. Out of these, 6 articles were deemed ineligible for inclusion in this meta-analysis. Ultimately, our analysis comprised 28 studies. The flowchart illustrating the process of study selection is presented in Fig. 1. The basic characteristics of 28 included studies are shown in Supplementary Table 3. A total of 30 groups were extracted from the 28 studies. The incidence analysis of the MDRPI encompassed a sample population of 117,624 individuals across 28 studies. The study duration spanned from 2013 to 2022, while the publication period ranged from 2014 to 2023. The literature sources include Netherland, Norway, America, Canada, Australia, New Zealand, Brazil, China, Jordan, Japan, Turkey, Korea, and the United Kingdom. There were 11 cross-sectional studies, 2 retrospective cohort studies, and 15 prospective cohort studies in this review.

Fig. 1

The flowchart illustrating the process of study selection

Overall incidence of MDRPI

The overall MDRPI incidence was 19.3% (95% CI 13.5%–25.2%, n = 30, I2 = 99%) (Table 1). Figure 2 illustrated a forest plot depicting the overall MDRPI incidence across different arms and the entire study population.

Table 1 Stratification of MDRPI incidence estimates by moderator variablesFig. 2

The forest plot depicting the overall MDRPI incidence

MDRPI incidence by geographical regions

The incidence of MDRPI was 48.3% (95% CI 20.8–75.7%, n = 2, I2 = 94%) in South America, 21.9% (95% CI 14.3–29.6% n = 17, I2 = 98%) in Asia, 17.3% (95% CI 12.7–21.9% n = 2, I2 = 62%) in Europe, 13.0% (95% CI 5.0–21.1% n = 5, I2 = 96%) in Oceania, 3.6% (95% CI 0.0–8.5% n = 3, I2 = 92%) in North America, and 3.1% (95% CI 1.7–5.1% n = 1) in North America and Oceania (Table 1 and Fig. 3). A statistically significant difference was observed among the subgroups (p < 0.01). In the Asian region, Turkey emerged as the most frequently reported country, exhibiting an incidence rate of 30.6% (95% CI 16.7–44.6%) across 7 studies conducted (Table 1 and Fig. 4).

Fig. 3

The forest plot depicting MDRPI incidence by continents

Fig. 4

The forest plot depicting MDRPI incidence by countries

MDRPI incidence by age

MDRPI incidence was 20.0% (95% CI 13.4–26.6%, n = 26, I2 = 99%) in the adult population (aged ≥ 18), 6.9% (95% CI 2.8–11.0%, n = 2, I2 = 66%) in the child population (aged < 18) and 23.8% (95% CI 16.6–30.9%, n = 2, I2 = 72%) in the mixed population (p < 0.01) (Table 1 and Fig. 5).

Fig. 5

The forest plot depicting MDRPI incidence by age

MDRPI incidence by study years

The incidence of MDRPI in patients was 14.4% (95% CI 3.6–25.2%, n = 2, I2 = 92%) from 2010 to 2015, increased to 20.2% (95% CI 12.0–28.4%, n = 16, I2 = 99%) between 2016 and 2020, and further rose to 25.2% (95% CI 12.0–38.5%, n = 7, I2 = 98%) after the year of 2020. The incidence of MDRPI for the five studies that did not mention study time was 10.7% (95% CI 0.0–23.9%, n = 5, I2 = 98%). The incidence demonstrated a numerical increase in patients over the observed time period; however, no statistically significant differences were observed among the subgroups (p = 0.39; see Table 1 and Fig. 6).

Fig. 6

The forest plot depicting MDRPI incidence by study years

MDRPI incidence by population source

A significant difference in MDRPI incidence was observed among subgroups based on the population source (p < 0.01). Incidence of MDRPI was 19.0% (95% CI 12.4–25.6%, n = 17, I2 = 98%) in the ICU, 11.0% (95% CI 2.8–19.2%, n = 6, I2 = 97%) in hospitalized patients, 32.0% (95% CI 13.8–50.2%, n = 6, I2 = 97%) in special inpatient groups, and 0.6% (95% CI 0.6–0.7%, n = 1) in large database (Table 1 and Fig. 7).

Fig. 7

The forest plot depicting MDRPI incidence by population source

MDRPI incidence by gender

The MDRPI incidence of men was 17.0% (95% CI 9.0–25.0%, n = 13, I2 = 95%), and 17.7% (95% CI 8.0–27.4%, n = 13, I2 = 95%) of women. There was no statistical difference in gender subgroups (p = 0.91) (Table 1 and Fig. 8).

Fig. 8

The forest plot depicting MDRPI incidence by gender

Results for meta-regression analysis

In a univariate meta-regression analysis, South American continents, ICU or special inpatient group, and study years 2016–2020 or > 2020 were related to higher MDRPI incidence (p = 0.0004, p = 0.0404, p < 0.0001, p = 0.0045 and p < 0.0001, respectively) and child was associated with lower incidence of MDRPI (p = 0.0018) (Table 2). Multivariate meta-regression showed that continent from South America (p = 0.0457) and population source from special inpatient group (p = 0.0255) were significantly associated with higher incidence of MDRPI (Table 2).

Table 2 Meta-regression analysis for incidence of MDRPIMedical devices and risk factors of MDRPI

21 articles mentioned about medical devices that cause MDRPI. Among all the devices, nasogastric and tracheal tubes are the predominant medical devices associated with MDRPI, as indicated by 17 articles highlighting their causative role. Devices such as oxygen masks, neck immobilization devices, pulse oximeters, and orthopedic instruments were also mentioned as common contributors to MDRPI. Risk factors associated with MDRPI were identified in 12 articles. Among all the risk factors, length of hospital stay as a risk factor for MDRPI was mentioned in 8 articles. Other risk factors cited included multiple medical devices, male sex, poor initial health score, mechanical ventilation, multiple medications, vasoactive drug infusions, old age, history of cardiovascular diseases, administration of vasopressors, postural ulcers, chronic liver disease, hemoglobin less than 9.0, Lower Braden scores and having skin edema (Supplementary Table 3).

Publications bias and sensitivity analysis

The Egger test showed publication bias involving the overall incidence of MDRPI (p < 0.0001). The sensitivity analyses showed little change after estimating the effect of each study (Supplementary Fig. 1).