Skip to main content

Derivation and validation of a nomogram for predicting nonventilator hospital-acquired pneumonia among older hospitalized patients

Abstract

Background

Currently, there is no effective tool for predicting the risk of nonventilator hospital-acquired pneumonia (NV-HAP) in older hospitalized patients. The current study aimed to develop and validate a simple nomogram and a dynamic web-based calculator for predicting the risk of NV-HAP among older hospitalized patients.

Methods

A retrospective evaluation was conducted on 15,420 consecutive older hospitalized patients admitted to a tertiary hospital in China between September 2017 and June 2020. The patients were randomly divided into training (n = 10,796) and validation (n = 4624) cohorts at a ratio of 7:3. Predictors of NV-HAP were screened using the least absolute shrinkage and selection operator method and multivariate logistic regression. The identified predictors were integrated to construct a nomogram using R software. Furthermore, the optimum cut-off value for the clinical application of the model was calculated using the Youden index. The concordance index (C-index), GiViTI calibration belts, and decision curve were analysed to validate the discrimination, calibration, and clinical utility of the model, respectively. Finally, a dynamic web-based calculator was developed to facilitate utilization of the nomogram.

Results

Predictors included in the nomogram were the Charlson comorbidity index, NRS-2002, enteral tube feeding, Barthel Index, use of sedatives, use of NSAIDs, use of inhaled steroids, and "time at risk". The C-index of the nomogram for the training and validation cohorts was 0.813 and 0.821, respectively. The 95% CI region of the GiViTI calibration belt in the training (P = 0.694) and validation (P = 0.614) cohorts did not cross the diagonal bisector line, suggesting that the prediction model had good discrimination and calibration. Furthermore, the optimal cut-off values for the training and validation cohorts were 1.58 and 1.74%, respectively. Analysis of the decision curve showed that the nomogram had good clinical value when the threshold likelihood was between 0 and 49%.

Conclusion

The developed nomogram can be used to predict the risk of NV-HAP among older hospitalized patients. It can, therefore, help healthcare providers initiate targeted medical interventions in a timely manner for high-risk groups.

Peer Review reports

Introduction

Hospital-acquired pneumonia (HAP) is a common healthcare-associated infection (HAI) that affects approximately 1% of all hospitalized patients [1,2,3]. Nonventilator hospital-acquired pneumonia (NV-HAP) and ventilator-associated pneumonia (VAP) are the two subtypes of HAP [4]. The former (NV-HAP) accounts for more than 60% of HAP cases [2]. Nonventilator hospital-acquired pneumonia has been linked to longer hospital stays, extensive use of intensive care services, higher medical costs, and a high risk of hospital mortality [5, 6]. Several unmodifiable and modifiable risk factors, such as age (> 65 years), surgery, enteral feedings, lack of mobility, malnutrition, high blood glucose, inhibition of gastric acid, and central nervous system depressants, that predict the occurrence of NV-HAP have been previously profiled [7]. Evidence from a previous systematic review indicates that interventions such as improved oral hygiene, dysphagia screening, and early mobilization targeting potential modifiable risk factors might reduce the risk of NV-HAP [8]. Although NV-HAP is preventable and controllable, the widespread nature of the disease among clinical departments necessitates the search for various preventive interventions targeted at inpatients [9, 10]. The incidence of NV-HAP increases with age, and it has been found that older patients have a higher mortality rate and a worse prognosis than younger patients [9, 11]. Therefore, identifying populations at a high risk of NV-HAP, especially in older hospitalized patients, is crucial, as it will allow early initiation of interventions to prevent the occurrence of NV-HAP. Furthermore, early and accurate identification of high-risk groups will lead to the design of effective policies and preventive measures for NV-HAP.

To date, few risk prediction tools for NV-HAP in older hospitalized patients have been developed. A nomogram is a visualization method for complex mathematical models that combines multiple risk factors. This method provides accurate and individualized risk estimates for patients and presents them intuitively [12, 13]. Therefore, the present study aimed to develop and validate a simple nomogram and a dynamic web-based calculator for predicting NV-HAP risk among older hospitalized patients.

Methods

Study design and participants

The current study adopted a retrospective cohort study design. A total of 15,420 consecutive older hospitalized patients admitted to Wenzhou People's Hospital (a 1500-bed teaching hospital in Zhejiang, China) between September 2017 and June 2020 were recruited. Patients with a stay length of < 48 h and those who had received mechanical ventilation during hospitalization were excluded from the study. In the case of repeated admissions for a patient, only the first admission was considered. The current study was approved by the Ethics Review Committee of Wenzhou People's Hospital (IRB no: WRY2018070). The same committee waived the requirement for informed consent due to the retrospective nature of the study.

Candidate predictors

The following candidate predictors were collected: demographic data (age, sex), Nutritional Risk Screening 2002 (NRS-2002), Barthel Index, Morse Fall Scale, other nosocomial infections, season of admission, Charlson comorbidity index (CCI), and admission category. Furthermore, the patient's personal history (drinking status, smoking status), comorbidities (chronic obstructive pulmonary disease (COPD), swallow disability, stroke, diabetes mellitus, peptic ulcer disease, moderate or severe renal disease, liver disease, congestive heart failure, and solid tumor), clinical procedures (central venous catheter, indwelling urinary catheter, surgery, parenteral nutrition, and enteral tube feeding), in-hospital medications (antacids, sedatives, nonsteroidal anti-inflammatory drugs (NSAIDs), systemic steroids, inhaled steroids, and anticoagulant) and laboratory values (blood urea nitrogen, albumin, C-reactive protein, lymphocytes, white blood cells, and hemoglobin) were collected. The burden of comorbid diseases was described using the CCI [14]. The NRS-2002 [15], Barthel Index [16], and Morse Fall Scale [17] were used to assess the nutritional status, performance in activities of daily living, and risk of falling among inpatients, respectively.

To determine the effect of time on NV-HAP, the "time at risk" was also included as a potential predictor. For patients with NV-HAP, "time at risk" was calculated as the NV-HAP diagnosis date minus the admission date. For patients without NV-HAP, the "time at risk" was determined as the total length of stay. The variable "time at risk" was categorized into three subgroups based on tertile points, ≤ 6, 7–11, and ≥ 12 days. Laboratory data were collected from the first-time examinations at admission. During the "time at risk" period, clinical procedures, other nosocomial infections, and in-hospital medications were also collected.

Outcome

The data on NV-HAP were drawn from the Xinglin system, which is a real-time nosocomial infection surveillance system [18]. The system can collect data from multiple hospital databases [including Hospital Information System (HIS), Laboratory Information System (LIS), and Remote Installation Service systems (RIS)] in real time. In addition, the system can report automatically screened infection warnings, such as information on clinical signs, positive bacterial culture, and elevated biochemical and inflammatory indexes.

For nosocomial infection cases, clinicians made an initial diagnosis, which was then confirmed by a senior infection control practitioner. In case of a discrepant diagnosis between the two sides of the diagnosis, a joint discussion was held to reach a consensus. NV-HAP was defined according to the 2018 version of the Chinese guidelines for diagnosing and treating hospital-acquired pneumonia and ventilator-associated pneumonia in adults [19]. The definition was also consistent with the guidelines issued by the American Thoracic Society [20].

Feature selection and construction of a prediction model

The least absolute shrinkage and selection operator (LASSO) method was used to select optimal predictive features from the aforementioned 38 candidate predictors in the training cohort [21]. The LASSO method can reduce highly dimensional data and provides a better-fitting model. The optimal value of the penalty parameter λ was determined through a tenfold cross validation. Variables with nonzero coefficients in the fitted LASSO model were considered significant predictors. To obtain an integrated nomogram, a stepwise feature selection algorithm with the Akaike information criterion (AIC) and multivariate logistic regression model were used to filter the predictors [22]. The regression coefficients of the selected independent variables were used to construct a simple nomogram for predicting NV-HAP risk in older hospitalized patients.

Evaluating the model's performance

The performance of the model in predicting NV-HAP risk was expressed in terms of discrimination and calibration. Discrimination of the prediction model refers to the ability to distinguish between individuals developing and not developing the outcome (NV-HAP) [23]. The concordance index was used to validate the nomogram's discrimination ability (C-index; equal to the area under the receiver operating curve). The value of the C-index varied from 0.5 to 1.0, with values > 0.75 indicating relatively good discrimination.

The calibration of a prediction model refers to testing the agreement between the predicted and actual observed risk [23]. The GiViTI calibration belts were plotted to calibrate the nomogram [24]. The deviation between the predicted and observed probabilities was detected using the 0.95 confidence band of the calibration curve and calibration test. The 95% CI did not cross the bisector, which indicated a statistically significant deviation from the predicted probabilities. A P value > 0.05 in the calibration test demonstrated that there was no evidence of poor fit in the nomogram.

Clinical usage

Decision curve analysis (DCA) was performed to assess the clinical utility of the developed nomogram [25]. Decision curve analysis (DCA) is a tool for assessing the potential benefits of a risk prediction model once applied in clinical practice. It is calculated as shown in the formulae below.

$${\text{Net}}\;{\text{benefit}} = {\text{true}}\;{\text{positive}}\;{\text{rate}} - {\text{false}}\;{\text{positive}}\;{\text{rate}} \times \frac{{P_{{\text{t}}} }}{{1 - P_{{\text{t}}} }}$$

where Pt represents the threshold probability at which the expected benefit of intervention-all-patients is equal to the expected benefit of intervention-none.

In addition, the highest Youden index was used to obtain the best cut-off value for clinical application (sensitivity + specificity − 1).

Statistical analysis

The present study was reported as per the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) statement [26]. All the included patients were randomly divided into the training (n = 10,796) and validation (n = 4624) cohorts at a ratio of 7:3 [27, 28]. The training cohort was used for model training and estimation of model parameters, whereas the validation cohort was used to test the model and evaluate its benefits. Nonnormally distributed continuous data were analyzed with the Mann–Whitney U test and expressed as medians (inter-quartile range, IQR). Categorical variables were expressed as numbers (%) and compared using the chi-square (and Fisher's exact) test. Patients with missing data were excluded from this study because the proportion of cases with missing data was minimal (< 5%), and it was considered to be missing at random [29, 30]. All data were statistically analyzed using R software (version 3.6.1; https://www.R-project.org). A two-tailed P value of ≤ 0.05 was considered to be statistically significant.

Results

Characteristics of the study cohort

A total of 15,420 participants were included in the final analysis (Additional file 1: Fig. S1). A total of 10,796 and 4624 participants belonged to the training cohort and validation cohort, respectively. In the training cohort, 153 cases were NV-HAP, and 10,643 cases were non-NV-HAP. In the validation cohort, 72 cases were NV-HAP, and 4552 cases were non-NV-HAP. The percentage of patients with NV-HAP in the training and validation cohorts was 1.4% and 1.6%, respectively, and the difference was not statistically significant (P = 0.507). The baseline characteristics of patients in the training and validation cohorts are presented in Table 1.

Table 1 Baseline characteristics of the study population

Feature selection

Among the 38 variables collected, 17 potential predictors were identified in the training cohort (Fig. 1a, b). They included the Charlson comorbidity index, Hb, NRS-2002, stroke, admission category, central venous catheter, indwelling urinary catheter, surgery, enteral tube feeding, Barthel Index, Morse Fall Scale, other nosocomial infections, use of sedatives, use of NSAIDs, use of inhaled steroids, use of anticoagulant, and "time at risk". The predictors showed nonzero coefficients in the LASSO regression model.

Fig. 1
figure 1

Variable selection using the LASSO binary logistic regression model. a Profiles of the LASSO coefficients for the 38 candidate variables. b Optimal penalization coefficient (λ) selection in the LASSO model using tenfold cross-validation via minimum criteria. Note: the left vertical line represents the minimum error, and the right vertical line represents the one standard error of the minimum criteria (1-SE criterion). LASSO least absolute shrinkage and selection operator

Prediction model development

To further identify independent predictors of NV-HAP, a multivariable logistic regression model with a two-way stepwise strategy was adopted. Finally, the model including CCI, NRS-2002, Enteral tube feeding, Barthel Index, use of sedatives, use of NSAIDs, use of inhaled steroids, and "time at risk" had the lowest AIC, hence the best goodness of fit (Table 2). Therefore, we used these eight independent predictors to construct a novel nomogram model for predicting the probability of NV-HAP (Fig. 2).

Table 2 Logistic analysis of each factor's ability in predicting the risk of NV-HAP
Fig. 2
figure 2

The nomogram for predicting the risk of NV-HAP in older hospitalized patients. Note The NV-HAP risk nomogram was developed with the predictors including CCI, NRS-2002, enteral tube feeding, Barthel Index, use of sedatives, use of NSAIDs, use of inhaled steroids, and “time at risk”. NV-HAP nonventilator-associated hospital-acquired pneumonia; CCI Charlson comorbidity index; NRS nutritional risk screening

Meanwhile, the R package shiny was used to create a visual and operational dynamic web-based calculator for the developed model. The user can conveniently obtain the prediction probability of NV-HAP by entering or selecting a variable in the graphical user interface (https://predction.shinyapps.io/DynNomapp/). For instance, using the dynamic nomogram for NV-HAP, the probability of NV-HAP in an older hospitalized patient with CCI = 8, NRS-2002 = 3, no enteral tube feeding, Barthel Index = 95 (slight dependency), use of sedatives, use of NSAIDs, use of inhaled steroids, and "time at risk" = 7 days was estimated to be 15.6% (Additional file 1: Fig. S2).

Prediction model validation

The C-index of the nomogram in the training cohort was 0.813 (95% confidence interval: 0.774–0.853) (Fig. 3a), whereas that in the validation cohort was 0.821 (95% confidence interval: 0.764–0.879) (Fig. 3b). These results suggested that the developed prediction model had good discrimination.

Fig. 3
figure 3

ROC curves of the nomogram. a The training cohort. b Validation Cohort. Note the x-axis represents the false-positive rate of the risk prediction. The y-axis indicates the true-positive rate of the risk prediction. ROC receiver operating characteristic; AUC area under the curve

The 95% CI region of the GiViTI calibration belt in the training and validation cohorts did not cross the diagonal bisector line (P = 0.694, P = 0.614; respectively) (Fig. 4a, b). This suggested that the developed prediction model had strong concordance performance in the two data sets.

Fig. 4
figure 4

The GiViTI calibration belt for the nomogram. a The training cohort. b Validation Cohort. Note The 80% CI and 95% CI calibration belt are plotted in light and dark gray, respectively. The red diagonal line is the reference line indicating perfect calibration

Determination of optimal cut-off values for the nomogram

At the maximum Youden index, the optimal cut-off values in the training and validation cohorts were 1.58 and 1.74%, respectively. The developed model was applied to predict NV-HAP in the studied cohorts (Additional file 1: Table S1). The following results were obtained in the training cohort: specificity (81.94%), sensitivity (69.93%), accuracy (81.77%), positive predictive value (PPV) (5.27%), negative predictive value (NPV) (99.48%), positive likelihood ratio (PLR) (3.87%), negative likelihood ratio (NLR) (0.37%), and diagnostic odds ratio (DOR) (10.55%). For the validation cohort, the following results were obtained: specificity (82.36%), sensitivity (70.83%), accuracy (82.18%), positive predictive value (PPV) (5.97%), negative predictive value (NPV) (99.44%), positive likelihood ratio (PLR) (4.02%), negative likelihood ratio (NLR) (3.54%), and diagnostic odds ratio (DOR) (11.34%).

Clinical usefulness of the nomogram

The results of the decision curve analysis of the risk nomogram for the training and validation cohorts are shown in Fig. 5a, b, respectively. At a likelihood range of between 0 and 49%, the nomogram showed better performance in predicting NV-HAP risk compared to intervention-all-patients or intervention-none strategies. For example, in the training cohort, the standardized net benefit was approximately 60% at the 1.58% probability threshold.

Fig. 5
figure 5

Decision curve analysis for the nomogram. a The training cohort. b Validation Cohort. Note The y-axis represents the standardized net benefit. The thick red solid line is the nomogram to predict NV-HAP risk. The thin red solid line represents the 95% credible interval. The black solid line represents the assumption that all patients had no NV-HAP. The gray solid line represented the assumption that all patients had NV-HAP

Discussion

Nonventilator hospital-acquired pneumonia (NV-HAP) is a major safety issue among hospitalized patients and leads to high treatment costs, mortality rates, and longer hospital stays. Therefore, it is important to develop risk prediction models that can accurately identify high-risk groups to facilitate the early implementation of preventive interventions. Researchers have developed tools for measuring the risk of NV-HAP. For instance, Evans [31] developed an NV-HAP risk assessment tool based on risk factors reported by previous studies. In his assessment tool, each risk was weighted depending on the specific nature of the research and the number of authors who proposed the risk factors rather than using rigorous mathematical procedures. In their study, they conducted a hypothetical risk assessment, and thus further evaluation of the proposed risk factors and the weighting method is required. In a separate study, Wolfensberger et al. [32] developed and validated an automated classification algorithm based on radiological procedure (X-ray or CT scan) criteria to distinguish patients 'not at risk' from patients 'at risk' of developing NV-HAP. However, the primary aim of this automated classification algorithm was to reduce the manual surveillance workload rather than predicting the development of NV-HAP. Currently, there is no effective tool for assessing the risk of NV-HAP in patients, especially older hospitalized patients. Therefore, the current study aimed to develop and validate a simple-to-use nomogram and to establish an online calculator for the early prediction of NV-HAP among older hospitalized patients.

LASSO and multivariate logistic regression analyses were conducted to identify the key risk factors for NV-HAP in older hospitalized patients. Consequently, CCI, NRS-2002, enteral tube feeding, Barthel Index, use of sedatives, use of NSAIDs, use of inhaled steroids, and "time at risk" were identified as the major predictors and were used to develop a predictive nomogram. The constructed nomogram showed good discrimination and strong concordance performance in predicting NV-HAP. Similarly, the results of the DCA plots revealed that the nomogram was more effective in predicting NV-HAP than intervention-all-patients or intervention-none strategies. This implied that the developed nomogram can be implemented in clinical practice.

It was also found that patients with 12 or more days of "time at risk" experienced a nearly fourfold increased risk of NV-HAP compared with those who had a "time at risk" of less than six days. A longer "time at risk" usually means a longer period of bed rest and hospitalization. Previous studies have shown that hospital stay duration is an independent predictor of healthcare-acquired infections [33, 34]. Therefore, we hypothesize that "time at risk" may help predict NV-HAP occurrence by reflecting the length of hospital stay of the patients. In addition, four medical interventions (enteral tube feeding, use of sedatives, NSAIDs, and inhaled steroids) were significantly associated with a high risk of NV-HAP in older hospitalized patients. In particular, the use of feeding tubes increased oral colonization by pathogenic organisms, causing bacteria to migrate to the lungs via the tube [7, 35]. In addition, this procedure can raise gastric volume and pressure, hence increasing the risk of gastric reflux and pulmonary aspiration [7, 35]. Several previous studies have also reported that the use of sedatives, NSAIDs, and inhaled steroids increases the risk of HAI. These drugs exert their therapeutic effects by regulating immune functions [36,37,38,39].

The Barthel Index, the CCI, and the NRS-2002 were used to assess performance in daily living activities, the burden of comorbid diseases, and nutritional status, respectively. These assessment tools were found to be independent predictors of NV-HAP. A lower Barthel Index indicates poor performance in daily activities, which may reduce respiratory secretion clearance and contribute to the development of NV-HAP [9]. Malnutrition and underlying diseases have been identified as risk factors for HAIs [40,41,42]. Therefore, the application of the aforementioned eight risk factors in the current model is plausible and theoretically grounded.

Currently, NV-HAP is diagnosed using imaging examinations and clinical feature changes [3, 43]. However, when symptoms appear, it is usually too late to intervene, which results in poor clinical outcomes. Using the current model, healthcare workers can implement early interventions for high-risk elderly inpatients. Interventions such as improved oral hygiene, dysphagia screening, and early mobilization should be implemented more aggressively for high-risk individuals, and this may reduce the risk of NV-HAP. Moreover, the online version of the nomogram provided a more intuitive and convenient prediction of the risk of NV-HAP disease in patients.

This study has some limitations. First, the study adopted a retrospective design and was performed in a single center, which limits the generalizability of the findings. Furthermore, although internal validation was conducted to ensure the robustness of the developed nomogram, external validation could not be conducted. Therefore, further research is needed to validate the nomogram using an external cohort. Second, the study selected a few readily available risk factors and medical interventions, leaving out others such as the risk across the continuum of care and indicators of blood gas analysis [7]. Third, although each case of NV-HAP was extensively assessed and co-confirmed by a doctor and a senior infection control practitioner to avoid misclassification bias in the current investigation, misdiagnosis and missed diagnosis may still have occurred. Finally, even though various assessment tools, such as the Barthel Index, Morse Fall Scale, and NRS-2002, were conducted by specially trained nurses, the existence of measurement bias cannot be ruled out because the screening was performed by different nurses.

Conclusions

In conclusion, a novel nomogram combining CCI, NRS-2002, enteral tube feeding, Barthel Index, use of sedatives, use of NSAIDs, use of inhaled steroids, and "time at risk" was constructed in the present study. The nomogram showed good discrimination and calibration and hence can help healthcare workers estimate the risk of NV-HAP disease in older hospitalized patients. This will promote the implementation of early intervention for high-risk groups. However, there is a need to conduct external validation of the proposed nomogram in larger populations.

Availability of data and materials

All data generated or analysed during this study are included in this published article and Additional file 1.

Abbreviations

HAP:

Hospital-acquired pneumonia

HAIs:

Healthcare-associated infections

NV-HAP:

Nonventilator hospital-acquired pneumonia

VAP:

Ventilator-associated pneumonia

IQR:

Inter-quartile range

COPD:

Chronic obstructive pulmonary disease

ICU:

Intensive care unit

NSAIDs:

Non-steroidal anti-inflammatory drug

CCI:

Charlson comorbidity index

NRS:

Nutritional risk screening

BUN:

Blood urea nitrogen

ALB:

Albumin

CRP:

C-reaction protein

LY:

Lymphocyte

WBC:

White blood cell

Hb:

Hemoglobin

LASSO:

Least absolute shrinkage and selection operator

AIC:

Akaike information criterion

DCA:

Decision curve analysis

AUC:

Area under curve

CI:

Confidence interval

PPV:

Positive predictive value

NPV:

Negative predictive value

PLR:

Positive likelihood ratio

NLR:

Negative likelihood ratio

DOR:

Diagnostic odds ratio

References

  1. Magill SS, Edwards JR, Bamberg W, Beldavs ZG, Dumyati G, Kainer MA, et al. Multistate point-prevalence survey of health care-associated infections. N Engl J Med. 2014;370:1198–208.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  2. Magill SS, O'Leary E, Janelle SJ, Thompson DL, Dumyati G, Nadle J, et al. Changes in Prevalence of Health Care-Associated Infections in U.S. Hospitals. N Engl J Med. 2018;379:1732–44.

  3. Ji W, McKenna C, Ochoa A, Ramirez Batlle H, Young J, Zhang Z, et al. Development and Assessment of Objective Surveillance Definitions for Nonventilator Hospital-Acquired Pneumonia. JAMA Netw open. 2019;2:e1913674.

  4. de Miguel-Diez J, Albaladejo-Vicente R, Hernández-Barrera V, Ji Z, Lopez-Herranz M, Jimenez-Garcia R, et al. Hospital admissions for community-acquired, ventilator-associated and nonventilator hospital-acquired pneumonia in COPD patients in Spain (2016–2017). Eur J Intern Med. 2020;79:93–100.

    Article  PubMed  Google Scholar 

  5. Micek ST, Chew B, Hampton N, Kollef MH. A Case-control study assessing the impact of nonventilated hospital-acquired pneumonia on patient outcomes. Chest. 2016;150:1008–14.

    Article  PubMed  Google Scholar 

  6. Giuliano KK, Baker D, Quinn B. The epidemiology of nonventilator hospital-acquired pneumonia in the United States. Am J Infect Control. 2018;46:322–7.

    Article  PubMed  Google Scholar 

  7. Vignari M. Non-ventilator health care-associated pneumonia (NV-HAP): NV-HAP Risk Factors. Am J Infect Control. 2020;48:A10–3.

    Article  PubMed  Google Scholar 

  8. Mitchell BG, Russo PL, Cheng AC, Stewardson AJ, Rosebrock H, Curtis SJ, et al. Strategies to reduce non-ventilator-associated hospital-acquired pneumonia: a systematic review. Infect Dis Heal. 2019;24:229–39.

    Article  Google Scholar 

  9. Pássaro L, Harbarth S, Landelle C. Prevention of hospital-acquired pneumonia in non-ventilated adult patients: a narrative review. Antimicrob Resist Infect Control. 2016;5:43.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Giuliano KK, Penoyer D, Middleton A, Baker D. Original research: oral care as prevention for nonventilator hospital-acquired pneumonia: a four-unit cluster randomized study. Am J Nurs. 2021;121:24–33.

    Article  PubMed  Google Scholar 

  11. Li W, Ding C, Yin S. Severe pneumonia in the elderly: a multivariate analysis of risk factors. Int J Clin Exp Med. 2015;8:12463–75.

    PubMed  PubMed Central  Google Scholar 

  12. Balachandran VP, Gonen M, Smith JJ, DeMatteo RP. Nomograms in oncology: more than meets the eye. Lancet Oncol. 2015;16:e173–80.

    Article  PubMed  PubMed Central  Google Scholar 

  13. Park SY. Nomogram: an analogue tool to deliver digital knowledge. J Thorac Cardiovasc Surg. 2018;155:1793.

    Article  PubMed  Google Scholar 

  14. Charlson M, Szatrowski TP, Peterson J, Gold J. Validation of a combined comorbidity index. J Clin Epidemiol. 1994;47:1245–51.

    CAS  Article  PubMed  Google Scholar 

  15. Kondrup J, Rasmussen HH, Hamberg O, Stanga Z. Nutritional risk screening (NRS 2002): a new method based on an analysis of controlled clinical trials. Clin Nutr. 2003;22:321–36.

    Article  PubMed  Google Scholar 

  16. Mahoney FI, Barthel DW. Functional evaluation: the Barthel Index. Md State Med J. 1965;14:61–5.

    CAS  PubMed  Google Scholar 

  17. Chow SKY, Lai CKY, Wong TKS, Suen LKP, Kong SKF, Chan CK, et al. Evaluation of the Morse Fall Scale: applicability in Chinese hospital populations. Int J Nurs Stud. 2007;44:556–65.

    Article  PubMed  Google Scholar 

  18. Du M, Xing Y, Suo J, Liu B, Jia N, Huo R, et al. Real-time automatic hospital-wide surveillance of nosocomial infections and outbreaks in a large Chinese tertiary hospital. BMC Med Inform Decis Mak. 2014;14:9.

    Article  PubMed  PubMed Central  Google Scholar 

  19. Shi Y, Huang Y, Zhang T-T, Cao B, Wang H, Zhuo C, et al. Chinese guidelines for the diagnosis and treatment of hospital-acquired pneumonia and ventilator-associated pneumonia in adults (2018 Edition). J Thorac Dis. 2019;11:2581–616.

  20. Kalil AC, Metersky ML, Klompas M, Muscedere J, Sweeney DA, Palmer LB, et al. Management of adults with hospital-acquired and ventilator-associated pneumonia: 2016 clinical practice guidelines by the infectious diseases Society of America and the American Thoracic Society. Clin Infect Dis. 2016;63:e61-111.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Tibshirani RJ. Regression shrinkage and selection via the LASSO. J R Stat Soc Ser B Methodol. 1996;73:273–82.

    Article  Google Scholar 

  22. Akaike HT. A new look at the statistical model identification. Autom Control IEEE Trans. 1974;19:716–23.

    Article  Google Scholar 

  23. Shipe ME, Deppen SA, Farjah F, Grogan EL. Developing prediction models for clinical use using logistic regression: an overview. J Thorac Dis. 2019;11(Suppl 4):S574–84.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Nattino G, Finazzi S, Bertolini G. A new test and graphical tool to assess the goodness of fit of logistic regression models. Stat Med. 2016;35:709–20.

    Article  PubMed  Google Scholar 

  25. Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Mak Int J Soc Med Decis Mak. 2006;26:565–74.

    Article  Google Scholar 

  26. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ. 2015;350:g7594.

  27. Noorbakhsh J, Farahmand S, Foroughi Pour A, Namburi S, Caruana D, Rimm D, et al. Deep learning-based cross-classifications reveal conserved spatial behaviors within tumor histological images. Nat Commun. 2020;11:6367.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  28. Shi X, Xu L, Ma B, Wang S. Development and validation of a nomogram to predict the prognosis of patients with gastric cardia cancer. Sci Rep. 2020;10:14143.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  29. Wills A-MA, Pérez A, Wang J, Su X, Morgan J, Rajan SS, et al. Association Between Change in Body Mass Index, Unified Parkinson's Disease Rating Scale Scores, and Survival Among Persons With Parkinson Disease: Secondary Analysis of Longitudinal Data From NINDS Exploratory Trials in Parkinson Disease Long-term Study . JAMA Neurol. 2016;73:321–8.

  30. Futier E, Lefrant J-Y, Guinot P-G, Godet T, Lorne E, Cuvillon P, et al. Effect of individualized vs standard blood pressure management strategies on postoperative organ dysfunction among high-risk patients undergoing major surgery: a randomized clinical trial. JAMA. 2017;318:1346–57.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Evans S. Could a risk-assessment tool prevent hospital-acquired pneumonia? Br J Nurs. 2018;27:402–4.

    Article  PubMed  Google Scholar 

  32. Wolfensberger A, Jakob W, Faes Hesse M, Kuster SP, Meier AH, Schreiber PW, et al. Development and validation of a semi-automated surveillance system-lowering the fruit for non-ventilator-associated hospital-acquired pneumonia (nvHAP) prevention. Clin Microbiol Infect. 2019;25:1428.e7-1428.e13.

    CAS  Article  Google Scholar 

  33. Hur EY, Jin Y, Jin T, Lee S-M. Development and evaluation of the automated risk assessment system for catheter-associated urinary tract infection. Comput Inform Nurs. 2019;37:463–72.

    Article  PubMed  Google Scholar 

  34. Isik O, Kaya E, Dundar HZ, Sarkut P. Surgical site infection: re-assessment of the risk factors. Chirurgia (Bucur). 2015;110:457–61.

    CAS  Google Scholar 

  35. Leibovitz A, Baumoehl Y, Steinberg D, Segal R. Biodynamics of biofilm formation on nasogastric tubes in elderly patients. Isr Med Assoc J. 2005;7:428–30.

    PubMed  Google Scholar 

  36. Kumar G, Adams A, Hererra M, Rojas ER, Singh V, Sakhuja A, et al. Predictors and outcomes of healthcare-associated infections in COVID-19 patients. Int J Infect Dis IJID Off Publ Int Soc Infect Dis. 2021;104:287–92.

    CAS  Google Scholar 

  37. Liu P, Li X, Luo M, Xu X, Su K, Chen S, et al. Risk factors for carbapenem-resistant klebsiella pneumoniae infection: a meta-analysis. Microb Drug Resist. 2018;24:190–8.

    CAS  Article  PubMed  PubMed Central  Google Scholar 

  38. Caroff DA, Szumita PM, Klompas M. The relationship between sedatives, sedative strategy, and healthcare-associated infection: a systematic review. Infect Control Hosp Epidemiol. 2016;37:1234–42.

    Article  PubMed  Google Scholar 

  39. Micallef J, Soeiro T, Jonville-Béra A-P. Non-steroidal anti-inflammatory drugs, pharmacology, and COVID-19 infection. Therapie. 2020;75:355–62.

    Article  PubMed  PubMed Central  Google Scholar 

  40. Wałaszek M, Kosiarska A, Gniadek A, Kołpa M, Wolak Z, Dobroś W, et al. The risk factors for hospital-acquired pneumonia in the Intensive Care Unit. Przegl Epidemiol. 2016;70(15–20):107–10.

    Google Scholar 

  41. Hu Q, Wang G, Ren J, Ren H, Li G, Wu X, et al. Preoperative prognostic nutritional index predicts postoperative surgical site infections in gastrointestinal fistula patients undergoing bowel resections. Medicine (Baltimore). 2016;95:e4084.

  42. Gamaletsou MN, Poulia K-A, Karageorgou D, Yannakoulia M, Ziakas PD, Zampelas A, et al. Nutritional risk as predictor for healthcare-associated infection among hospitalized elderly patients in the acute care setting. J Hosp Infect. 2012;80:168–72.

    CAS  Article  PubMed  Google Scholar 

  43. Baker D, Quinn B. Hospital acquired pneumonia prevention initiative-2: incidence of nonventilator hospital-acquired pneumonia in the United States. Am J Infect Control. 2018;46:2–7.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

This work was supported by the Medicine and Health Science and Technology Project of Wenzhou Health Bureau [Grant Numbers 2018B01]. The funder was not involved in the design of the study, collection, analysis, and interpretation of data or in writing the manuscript.

Author information

Authors and Affiliations

Authors

Contributions

LC and XYL conceived and designed the present study. ZHC contributed to the data collection, interpretation, statistical analysis, and drafting of the manuscript for the present study. ZQX, HMW, SCG, HHW, and JRJ contributed to data collection, statistical analysis, and revision of the manuscript. All authors read and approved the final manuscript for publication.

Corresponding authors

Correspondence to Xiuyang Li or Le Chen.

Ethics declarations

Ethics approval and consent to participate

The study was performed in accordance with the principles of the Declaration of Helsinki and was approved by the Wenzhou People's Hospital Ethics Review Committee (IRB no: WRY2018070). The Wenzhou People's Hospital Ethics Review Committee granted a waiver of informed consent due to the study's retrospective nature.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1. Fig.S1.

Flow chart of the study population. Fig. S2. Screenshot of the online tool used for the prediction of NV-HAP risk. Table. S1 Diagnostic efficacy of the nomogram model for estimating the risk of NV-HAP.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Chen, Z., Xu, Z., Wu, H. et al. Derivation and validation of a nomogram for predicting nonventilator hospital-acquired pneumonia among older hospitalized patients. BMC Pulm Med 22, 144 (2022). https://doi.org/10.1186/s12890-022-01941-z

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12890-022-01941-z

Keywords

  • Hospital-acquired pneumonia
  • Aspiration pneumonia
  • Infection prevention
  • Nomogram
  • Prediction model