bs_bs_banner
ORIGINAL ARTICLE
Development and validation of a clinical risk score for predicting drug-resistant bacterial pneumonia in older Chinese patients HON MING MA,1 MARGARET IP,2 JEAN WOO1 AND DAVID SC HUI1 1
Department of Medicine and Therapeutics and 2Department of Microbiology, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong, SAR, China
ABSTRACT Background and objective: Health care-associated pneumonia (HCAP) and drug-resistant bacterial pneumonia may not share identical risk factors. We have shown that bronchiectasis, recent hospitalization and severe pneumonia (confusion, blood urea level, respiratory rate, low blood pressure and 65 year old (CURB-65) score ≥3) were independent predictors of pneumonia caused by potentially drug-resistant (PDR) pathogens. This study aimed to develop and validate a clinical risk score for predicting drug-resistant bacterial pneumonia in older patients. Methods: We derived a risk score by assigning a weighting to each of these risk factors as follows: 14, bronchiectasis; 5, recent hospitalization; 2, severe pneumonia. A 0.5 point was defined for the presence of other risk factors for HCAP. We compared the areas under the receiver-operating characteristics curve (AUROC) of our risk score and the HCAP definition in predicting PDR pathogens in two cohorts of older patients hospitalized with non-nosocomial pneumonia. Results: The derivation and validation cohorts consisted of 354 and 96 patients with bacterial pneumonia, respectively. PDR pathogens were isolated in 48 and 21 patients in the derivation and validation cohorts, respectively. The AUROCs of our risk score and the HCAP definition were 0.751 and 0.650, respectively, in the derivation cohort, and were 0.782 and 0.671, respectively, in the validation cohort. The differences between our risk score and the HCAP definition reached statistical significance. A score ≥2.5 had the best balance between sensitivity and specificity. Conclusions: Our risk score outperformed the HCAP definition to predict pneumonia caused by PDR pathogens. A history of bronchiectasis or recent hospitalization is the major indication of starting empirical broad-spectrum antibiotics. Key words: bacteria, drug resistance, older people, pneumonia, risk score. Correspondence: Hon Ming Ma, Department of Medicine and Therapeutics, 9/F Clinical Science Building, Prince of Wales Hospital, 32 Ngan Shing Street, Shatin, Hong Kong, SAR, China. Email: [email protected] Received 12 October 2013; invited to revise 3 November 2013; revised 26 November 2013; accepted 23 December 2013 (Associate Editor: Marcos Restrepo). © 2014 Asian Pacific Society of Respirology
SUMMARY AT A GLANCE This prospective cohort study shows that risk stratification with a clinical risk score is better than the HCAP definition in predicting drugresistant bacterial pneumonia. The presence of bronchiectasis or recent hospitalization is the most important indication of starting empirical broad-spectrum antibiotics.
Abbreviations: AUROC, area under the receiver-operating characteristics curve; CAP, community-acquired pneumonia; CAPD, chronic ambulatory peritoneal dialysis; CI, confidence interval; CUHK, Chinese University of Hong Kong; ESBL, extended-spectrum β-lactamase; GNB, Gram-negative bacilli; HCAP, health care-associated pneumonia; MRSA, methicillinresistant Staphylococcus aureus; NHAP, nursing home-acquired pneumonia; NHCAP, nursing and health care-associated pneumonia; NPV, negative predictive value; PD, peritoneal dialysis; PDR, potentially drug-resistant; PPV, positive predictive value; PWH, Prince of Wales Hospital; RRT, renal replacement therapy.
INTRODUCTION There is continuous debate about the concept of health care-associated pneumonia (HCAP) because of its imprecision in predicting community-acquired pneumonia (CAP) caused by potentially drugresistant (PDR) pathogens with subsequent excessive use of broad-spectrum antibiotics. Risk stratification has been advocated to guide physicians in starting empirical broad-spectrum antibiotics by clinical risk scoring tool.1–3 Older people constituted most of the patients hospitalized with CAP.4 We previously showed that bronchiectasis, recent hospitalization and severe pneumonia (confusion, blood urea level, respiratory rate, low blood pressure and 65 year old (CURB-65) score ≥3) were independent predictors of drugresistant bacterial pneumonia in older Chinese patients hospitalized with pneumonia. Each of these factors carried a differential risk.5 In contrast to our common belief, nursing home residence and recent antibiotic use were not the risk factors in our previous Respirology (2014) 19, 549–555 doi: 10.1111/resp.12267
550 study. Thus, the concept of HCAP may not be entirely applicable in Hong Kong where the prevalence of pneumonia caused by PDR pathogens is low. It remains uncertain what risk factors are indicative of empirical broad-spectrum antibiotics. Moreover, risk factor profiles have geographical variations because of the differences in the endemicity of drug-resistant pathogens and local antibiogram. The aim of this study was to develop and validate a clinical risk score to predict drug-resistant bacterial pneumonia among older people in Hong Kong.
METHODS Derivation and validation cohorts We prospectively conducted two observational cohort studies at the Prince of Wales Hospital (PWH), which was the teaching hospital of the Chinese University of Hong Kong (CUHK). The derivation and validation cohorts were recruited from January 2004 to June 2005,5 and from October 2009 to September 2010, respectively.6 Older patients, aged ≥65years, were analysed if they presented with pneumonia. Patients were excluded if they had been hospitalized within the previous 14 days or were transferred from another hospital with suspected nosocomial infection. Demographics, premorbid functional status, comorbid illnesses, clinical parameters, blood tests, radiological features and clinical outcomes were recorded. In the derivation cohort, functional status was classified by the Modified Functional Ambulation Category as full independence (independent walker), partial dependence (walk with aids or under supervision) or total dependence (chairbound or bedridden). In the validation cohort, functional status was represented by the Katz index as full independence (score 6), partial dependence (score 3–5) or total dependence (score 0–2). Both studies were approved by the Conjoint CUHKPWH Clinical Research Ethics Committee. Study definitions Pneumonia was defined by the presence of respiratory symptoms and signs associated with abnormal infiltrates on chest radiographs, which were suggestive of lower respiratory tract infection. Risk factors for HCAP comprised hospitalization for ≥2 days in the preceding 90 days, residence in a nursing home or extended-care facility, home infusion (including antibiotic) therapy, chronic wound care, chronic dialysis within 30 days and immunosuppression. The definition of chronic dialysis was modified to include both peritoneal dialysis (PD) and haemodialysis because the former was the predominant mode of renal replacement therapy (RRT) in Hong Kong. Immunosuppression was a risk factor for PDR pathogens and was defined by the presence of ≥1 of the following conditions: acquired immunodeficiency syndrome, active solid or haematological malignancy undergoing chemotherapy or radiotherapy, neutropenia (
ORIGINAL ARTICLE
Development and validation of a clinical risk score for predicting drug-resistant bacterial pneumonia in older Chinese patients HON MING MA,1 MARGARET IP,2 JEAN WOO1 AND DAVID SC HUI1 1
Department of Medicine and Therapeutics and 2Department of Microbiology, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong, SAR, China
ABSTRACT Background and objective: Health care-associated pneumonia (HCAP) and drug-resistant bacterial pneumonia may not share identical risk factors. We have shown that bronchiectasis, recent hospitalization and severe pneumonia (confusion, blood urea level, respiratory rate, low blood pressure and 65 year old (CURB-65) score ≥3) were independent predictors of pneumonia caused by potentially drug-resistant (PDR) pathogens. This study aimed to develop and validate a clinical risk score for predicting drug-resistant bacterial pneumonia in older patients. Methods: We derived a risk score by assigning a weighting to each of these risk factors as follows: 14, bronchiectasis; 5, recent hospitalization; 2, severe pneumonia. A 0.5 point was defined for the presence of other risk factors for HCAP. We compared the areas under the receiver-operating characteristics curve (AUROC) of our risk score and the HCAP definition in predicting PDR pathogens in two cohorts of older patients hospitalized with non-nosocomial pneumonia. Results: The derivation and validation cohorts consisted of 354 and 96 patients with bacterial pneumonia, respectively. PDR pathogens were isolated in 48 and 21 patients in the derivation and validation cohorts, respectively. The AUROCs of our risk score and the HCAP definition were 0.751 and 0.650, respectively, in the derivation cohort, and were 0.782 and 0.671, respectively, in the validation cohort. The differences between our risk score and the HCAP definition reached statistical significance. A score ≥2.5 had the best balance between sensitivity and specificity. Conclusions: Our risk score outperformed the HCAP definition to predict pneumonia caused by PDR pathogens. A history of bronchiectasis or recent hospitalization is the major indication of starting empirical broad-spectrum antibiotics. Key words: bacteria, drug resistance, older people, pneumonia, risk score. Correspondence: Hon Ming Ma, Department of Medicine and Therapeutics, 9/F Clinical Science Building, Prince of Wales Hospital, 32 Ngan Shing Street, Shatin, Hong Kong, SAR, China. Email: [email protected] Received 12 October 2013; invited to revise 3 November 2013; revised 26 November 2013; accepted 23 December 2013 (Associate Editor: Marcos Restrepo). © 2014 Asian Pacific Society of Respirology
SUMMARY AT A GLANCE This prospective cohort study shows that risk stratification with a clinical risk score is better than the HCAP definition in predicting drugresistant bacterial pneumonia. The presence of bronchiectasis or recent hospitalization is the most important indication of starting empirical broad-spectrum antibiotics.
Abbreviations: AUROC, area under the receiver-operating characteristics curve; CAP, community-acquired pneumonia; CAPD, chronic ambulatory peritoneal dialysis; CI, confidence interval; CUHK, Chinese University of Hong Kong; ESBL, extended-spectrum β-lactamase; GNB, Gram-negative bacilli; HCAP, health care-associated pneumonia; MRSA, methicillinresistant Staphylococcus aureus; NHAP, nursing home-acquired pneumonia; NHCAP, nursing and health care-associated pneumonia; NPV, negative predictive value; PD, peritoneal dialysis; PDR, potentially drug-resistant; PPV, positive predictive value; PWH, Prince of Wales Hospital; RRT, renal replacement therapy.
INTRODUCTION There is continuous debate about the concept of health care-associated pneumonia (HCAP) because of its imprecision in predicting community-acquired pneumonia (CAP) caused by potentially drugresistant (PDR) pathogens with subsequent excessive use of broad-spectrum antibiotics. Risk stratification has been advocated to guide physicians in starting empirical broad-spectrum antibiotics by clinical risk scoring tool.1–3 Older people constituted most of the patients hospitalized with CAP.4 We previously showed that bronchiectasis, recent hospitalization and severe pneumonia (confusion, blood urea level, respiratory rate, low blood pressure and 65 year old (CURB-65) score ≥3) were independent predictors of drugresistant bacterial pneumonia in older Chinese patients hospitalized with pneumonia. Each of these factors carried a differential risk.5 In contrast to our common belief, nursing home residence and recent antibiotic use were not the risk factors in our previous Respirology (2014) 19, 549–555 doi: 10.1111/resp.12267
550 study. Thus, the concept of HCAP may not be entirely applicable in Hong Kong where the prevalence of pneumonia caused by PDR pathogens is low. It remains uncertain what risk factors are indicative of empirical broad-spectrum antibiotics. Moreover, risk factor profiles have geographical variations because of the differences in the endemicity of drug-resistant pathogens and local antibiogram. The aim of this study was to develop and validate a clinical risk score to predict drug-resistant bacterial pneumonia among older people in Hong Kong.
METHODS Derivation and validation cohorts We prospectively conducted two observational cohort studies at the Prince of Wales Hospital (PWH), which was the teaching hospital of the Chinese University of Hong Kong (CUHK). The derivation and validation cohorts were recruited from January 2004 to June 2005,5 and from October 2009 to September 2010, respectively.6 Older patients, aged ≥65years, were analysed if they presented with pneumonia. Patients were excluded if they had been hospitalized within the previous 14 days or were transferred from another hospital with suspected nosocomial infection. Demographics, premorbid functional status, comorbid illnesses, clinical parameters, blood tests, radiological features and clinical outcomes were recorded. In the derivation cohort, functional status was classified by the Modified Functional Ambulation Category as full independence (independent walker), partial dependence (walk with aids or under supervision) or total dependence (chairbound or bedridden). In the validation cohort, functional status was represented by the Katz index as full independence (score 6), partial dependence (score 3–5) or total dependence (score 0–2). Both studies were approved by the Conjoint CUHKPWH Clinical Research Ethics Committee. Study definitions Pneumonia was defined by the presence of respiratory symptoms and signs associated with abnormal infiltrates on chest radiographs, which were suggestive of lower respiratory tract infection. Risk factors for HCAP comprised hospitalization for ≥2 days in the preceding 90 days, residence in a nursing home or extended-care facility, home infusion (including antibiotic) therapy, chronic wound care, chronic dialysis within 30 days and immunosuppression. The definition of chronic dialysis was modified to include both peritoneal dialysis (PD) and haemodialysis because the former was the predominant mode of renal replacement therapy (RRT) in Hong Kong. Immunosuppression was a risk factor for PDR pathogens and was defined by the presence of ≥1 of the following conditions: acquired immunodeficiency syndrome, active solid or haematological malignancy undergoing chemotherapy or radiotherapy, neutropenia (
