respiratory investigation 52 (2014) 280 –287
Contents lists available at ScienceDirect
Respiratory Investigation journal homepage: www.elsevier.com/locate/resinv
Original article
Mortality and severity evaluation by routine pneumonia prediction models among Japanese patients with 2009 pandemic influenza A (H1N1) pneumonia Yuji Fujikura, M.D.a,n, Shuichi Kawano, M.D.a, Yuji Kouzaki, M.D.a, Masahiro Shinoda, M.D.b, Yu Hara, M.D.a, Masaharu Shinkai, M.D., Ph.D.b, Soichiro Kanoh, M.D., Ph.D.a, Akihiko Kawana, M.D., Ph.D.a a
Division of Infectious Diseases and Respiratory Medicine, Department of Internal Medicine, National Defense Medical College, 3-2 Namiki, Tokorozawa, Saitama 359-8513, Japan b Respiratory Disease Center, Yokohama City University Medical Center, 4-57 Urafunecho, Minami-ku, Yokohama, Kanagawa 232-0024, Japan
ar t ic l e in f o
abs tra ct
Article history:
Background: Influenza-related pneumonia, referred to as influenza pneumonia, was
Received 18 February 2014
reported relatively more frequently during a recent influenza pandemic in 2009. The
Received in revised form
validity of adapting routine pneumonia severity prediction models for various types of
17 April 2014
pneumonia is unclear.
Accepted 23 April 2014
Methods: We conducted a nationwide survey to evaluate influenza pneumonia among adult
Available online 23 June 2014
patients in Japan. Questionnaires were sent to physicians working in departments of
Keywords: Influenza Influenza pneumonia Pneumonia severity index CURB-65 A-DROP
respiratory medicine at 2491 hospitals. Both the outcome and pneumonia severity, using invasive positive pressure ventilation (IPPV) as an indicator, were evaluated by routine pneumonia severity index (PSI), CURB-65 (confusion, urea, respiratory rate, blood pressure, and ageZ65 years), and A-DROP (age, dehydration, respiration, disorientation, and blood pressure). Results: Data collected from 320 patients with influenza pneumonia, including 25 cases (7.8%) of death and 43 (13.4%) of IPPV, were analyzed. Although all routine prediction models showed that higher mortality tended to be associated with a higher risk class/grade, the actual mortality rates were higher than predicted. The risk class of mortality calculated by the PSI was influenced by pneumonia patterns. Although pneumonia severity was similarly predicted, the types of pneumonia also affected severity in all prediction models. A-DROP showed the highest accuracy on receiver operating characteristic analysis for both mortality and severity. Conclusions: CURB-65 and A-DROP are fair predictors of mortality regardless of pneumonia patterns. However, the current pneumonia prediction models may underestimate the severity and appropriate site of care for patients with influenza pneumonia. & 2014 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.resinv.2014.04.003 2212-5345/& 2014 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved.
281
respiratory investigation 52 (2014) 280 –287
1.
2.2.
Introduction
Influenza has a worldwide distribution and causes increased morbidity and mortality each year [1]. Pneumonia develops in some patients with influenza, which can lead to influenzarelated death. Influenza-related pneumonia, referred to as influenza pneumonia, was reported relatively more frequently in the influenza pandemic of 2009, with significant morbidity and mortality worldwide [2–6]. Severity evaluation and prediction models have been established to manage patients with pneumonia, make clinical decisions, and improve outcome, especially in cases of community-acquired pneumonia (CAP). The major guidelines for CAP evaluation and prediction models include the pneumonia severity index (PSI) from the Infectious Diseases Society of America [7] and CURB-65 (confusion, urea, respiratory rate, blood pressure, and age Z65 years) from the British Thoracic Society [8]. In Japan, the A-DROP (age, dehydration, respiration, disorientation, and blood pressure) scoring system from the Japanese Respiratory Society [9] is usually adapted for patients with CAP. These models are validated and recommended as clinical guidelines [10]. On the other hand, some reports on the recent influenza pandemic suggested that routine prediction models such as the PSI and CURB-65 failed to predict the need for mechanical ventilation or to estimate mortality [11–13]. In the 2009 influenza pandemic, some patients with severe influenza pneumonia presented with viral pneumonia with poor outcome, which is not typical of the usual form of CAP. Three manifestations of pneumonia associated with influenza are well described [14]; therefore, the validity of adapting routine prediction for various types of pneumonia is unclear. In a previous study, we collected and analyzed data from 346 cases of influenza pneumonia, including 27 fatal cases [15]. The present study uses the same dataset to evaluate each routine pneumonia prediction model and the effects of influenza pneumonia among adult patients in Japan.
2.
Materials and methods
2.1.
Study design and data collection
In a retrospective cohort study of adult patients with influenza pneumonia, questionnaires were sent to physicians working in departments of respiratory medicine at 2491 hospitals (4200 beds) across Japan. Data were collected by mail. This study conformed to the principles of the Declaration of Helsinki and was approved by the Institutional Review Board (Approval no. 778, 26/4/2010).
Abbreviations: CAP,
community-acquired pneumonia; IPPV,
Study definition
The influenza pandemic period was defined as April 1, 2009– April 30, 2010. Patients were given a diagnosis of pandemic influenza (H1N1) 2009 by rapid immunofluorescence influenza diagnostic kits or reverse transcription polymerase chain reaction. From the epidemiological data, it is generally accepted that rapid diagnostic kit results showing influenza A infection indicated H1N1 infection during this pandemic period [16]. Pneumonia was diagnosed by an attending physician based on respiratory symptoms and the findings of physical and radiological examinations. Influenza pneumonia is classified into three categories according to the definition by Louria et al. [14]: (1) pure influenza viral pneumonia (i.e., primary influenza pneumonia where pneumonia is caused only by the influenza virus and no bacterial pathogens are detected), (2) secondary bacterial pneumonia (i.e., influenza is complicated by secondary bacterial pneumonia that develops once the influenza symptoms have resolved), and (3) mixed viral and bacterial pneumonia (i.e., the influenza virus and bacterial pneumonia occur concurrently).
2.3.
Questionnaires
Age (in 10-year brackets), sex, and comorbid conditions were recorded. As described in the study definition, all pneumonia cases were classified as pure influenza viral pneumonia, secondary bacterial pneumonia, mixed viral and bacterial pneumonia, or miscellaneous according to the attending physician's diagnosis. Invasive positive pressure ventilation (IPPV) use, treatment options, and outcome data were also collected.
2.4.
Pneumonia severity evaluation
The severity of influenza pneumonia was estimated using routine PSI, CURB-65, and A-DROP. In all cases, age was expressed as the median of each class. Patients were categorized into five classes by the PSI score [7] and into the following three groups according to the CURB-65 score: “group 1: mortality low” with a score of 0 or 1; “group 2: mortality intermediate” with a score of 2; and “group 3: mortality high” with a score of Z3 [8]. The A-DROP system is based on the following five clinical features: age (A), dehydration (D), respiration (R), orientation (O), and blood pressure (P). According to this system, cases are classified as “mild” when none of the five criteria are met, “moderate” when one or two of the criteria are met, “severe” when three of the criteria are met, or “extremely severe” when four or five of the criteria are met. Cases of shock or altered mental status are regarded as extremely severe [9].
invasive positive pressure ventilation; PSI,
pneumonia severity index;
ROC, receiver operating characteristic n Corresponding author. Tel.: þ81 4 2995 1211; fax: þ81 4 2995 1497. E-mail addresses: [email protected] (Y. Fujikura), [email protected] (S. Kawano), [email protected] (Y. Kouzaki), [email protected] (M. Shinoda), [email protected] (Y. Hara), [email protected] (M. Shinkai), [email protected] (S. Kanoh), [email protected] (A. Kawana).
282
2.5.
respiratory investigation 52 (2014) 280 –287
Outcome measures and statistical analysis
The primary measures were used to evaluate validity in terms of mortality by the routine prediction models (PSI, CURB-65, and A-DROP) for influenza pneumonia and the influence of pneumonia patterns. Secondary measures were used to evaluate pneumonia severity by using IPPV as an indicator [17]. The effect of pneumonia patterns on mortality or severity was also analyzed. Frequency analysis, with results expressed as percentages, was used for the descriptive statistics of nonparametric data. We performed the Shapiro–Wilk test for normality and the Kruskal–Wallis test for age-group comparisons. The Chi-square test was used to analyze pneumonia severity, use of mechanical ventilation, and outcomes, and the Fisher exact test was used to analyze small samples. The Cochran– Armitage test was used for trend analysis, and multiple logistic regression models were performed to analyze risk estimation. Po0.05 was considered significant. All statistical analyses were performed using SPSS version 21.0.0.2 (IBM Corporation, Armonk, NY) and R version 2.15.3 (R Foundation for Statistical Computing, Vienna, Austria) software.
3.
Results
3.1.
Patients
We collected data on 346 cases of influenza pneumonia from 994 physicians (overall questionnaire response rate, 39.9%). Patient characteristics, comorbid conditions, treatments, and outcomes are described elsewhere. We excluded 26 cases from the analysis because of insufficient data to enable a severity calculation (Fig. 1).
3.2.
Patient background, symptoms, and findings
Of the remaining 320 cases, pure influenza viral pneumonia was accounted for in 94 patients (29.4%), mixed viral and bacterial pneumonia in 127 (39.7%), secondary bacterial pneumonia in 53 (16.6%), and miscellaneous in 46 (14.4%). In addition, the dataset included 43 cases of IPPV (13.4%) and 25 cases of death (7.8%). The in-hospital mortality rate was 8.6%. Patient background, symptoms, and findings for each group (excluding the miscellaneous group) are given in
Table 1. Pure influenza viral pneumonia tended to affect young, obese patients (21.3%) presenting with a nonproductive cough. In these patients, detailed laboratory investigations revealed low white blood cell counts, and chest X-rays showed ground-glass opacity with diffuse distribution. Microbiological tests were performed in 196 (61.6%) patients, and Streptococcus pneumoniae was detected as the major agent in 67 (20.9%). Other detected pathogens included Hemophilus influenzae (n¼ 10, 3.1%), methicillin-sensitive Staphylococcus aureus (n ¼6, 1.9%), Mycoplasma pneumoniae (n¼ 4, 1.3%), Moraxella catarrhalis (n ¼4, 1.3%), Pseudomonas aeruginosa (n ¼4, 1.3%), Klebsiella pneumoniae (n¼ 2, 0.6%), Streptococcus viridans (n¼ 2, 0.6%), Enterobacter cloacae (n ¼1, 0.3%), and Enterococcus faecalis (n¼ 1, 0.3%).
3.3.
The number of patients and deaths and the observed and predicted mortality rates in each class/group according to the PSI, CURB-65, and A-DROP are shown in Fig. 2A–C, respectively. The PSI, CURB-65, and A-DROP classes/groups correlated well with one another (data not shown). All routine prediction models showed that higher mortality was related to a higher class/grade on the Cochran–Armitage tendency test (po0.001); however, the actual rate of mortality was higher than that predicted from the routine CAP prediction models [7,8,18]. The receiver operating characteristic (ROC) curve for mortality in each prediction model (by class/score) is shown in Fig. 4A. The highest accuracy, as evaluated by the area under the curve, was 0.820 for A-DROP. Multiple logistic regression analysis of PSI risk classes I versus II and I versus III revealed no significance between odds ratios. PSI evaluation of the types of pneumonia showed that their effect on mortality was independent (p¼ 0.027), especially in pure influenza viral pneumonia (Table 2).
3.4.
Severity evaluation by each prediction model
Fig. 3A–C shows the number of IPPV cases and proportion of IPPV cases in each class/group according to the PSI, CURB-65, and A-DROP, respectively. With regard to mortality, the use of IPPV tended to be greater in higher classes/grades on Cochran–Armitage tendency testing (po0.001). However, IPPV was used even in relatively mild classes/grades. The ROC curve for IPPV is shown in Fig. 4B. A-DROP had the highest area under the curve, of 0.778. In multiple logistic regression models, the need for IPPV was affected by the type of pneumonia, particularly pure influenza viral pneumonia, in all prediction models (Table 3). The calculations for each prediction model for each type of pneumonia are detailed in Fig. S1.
4.
Fig. 1 – Study patients.
Mortality evaluation by each prediction model
Discussion
This study shows that the current routine prediction models for CAP underestimate the true severity and mortality rates of influenza pneumonia. Although mortality tends to be increased in higher classes/grades, the overall mortality rates are higher than predicted [7,8,18]. The findings suggest that
respiratory investigation 52 (2014) 280 –287
283
Table 1 – Patient background, symptoms, and laboratory findings in each pneumonia type.
Patient background Age (years old, average) No underlying medical conditions Chronic respiratory diseases Chronic heart diseases Chronic renal diseases Chronic liver diseases Cerebrovascular disease Hematological diseases Diabetes mellitus Collagen vascular diseases Malignancy Obesity Symptom/physical examination Productive cough Nonproductive cough Dyspnea Crackles on physical examination Laboratory test/chest X-ray findings WBCo4000/μL PaO2o60 Torr or SpO2o90% Ground glass opacity Consolidation Extent of shadow more than 2/3 area of one lung
Pure influenza viral pneumonia n ¼ 94
Secondary bacterial pneumonia n ¼ 53
Mixed viral and bacterial pneumonia n ¼127
p Value
47 38 (40.4)
57 15 (28.3)
58 37 (29.1)
o0.001 0.154
14 5 1 2 0 4 7 2 3 20
(14.9) (5.3) (1.1) (2.1) (0) (4.3) (7.4) (2.1) (3.2) (21.3)
24 3 1 1 3 0 6 0 1 2
(45.3) (5.7) (1.9) (1.9) (5.7) (0) (11.3) (0) (1.9) (3.8)
43 20 6 4 15 1 11 3 1 8
(33.9) (15.7) (4.7) (3.1) (11.8) (0.8) (8.7) (2.4) (0.8) (6.3)
o0.001 0.004 0.073 0.395 o0.001 0.064 0.726 0.529 0.151 0.001
27 40 55 49
(29.0) (43.0) (59.1) (52.7)
34 8 15 18
(64.2) (15.1) (28.3) (34.0)
80 15 62 60
(63.0) (11.8) (48.8) (47.2)
o0.001 o0.001 0.002 0.090
21 43 58 44 41
(22.6) (46.2) (69.9) (53.0) (50.0)
1 17 10 44 6
(1.9) (32.1) (19.2) (84.6) (11.8)
7 52 42 94 31
(5.5) (40.9) (33.9) (75.8) (25.4)
o0.001 0.247 o0.001 o0.001 o0.001
Data are expressed as n (%) unless otherwise indicated. Abbreviations: WBC, white blood cell; PaO2, arterial partial pressure of oxygen; and SpO2, oxygen saturation by pulse oximetry.
the more-fulminant patterns of pneumonia associated with severe hypoxemia and acute respiratory distress syndrome were more common in the 2009 influenza pandemic. The data also suggest that influenza pneumonia is not typical in patients with CAP. As described in previous reports, in addition to patients with respiratory complications, those without any underlying medical conditions were also commonly observed in the 2009 pandemic [19–22], and severe influenza was found in young patients [2,3,5,15,23]. The patient background characteristics in the present study are similar to those reported previously. We classified influenza pneumonia into three types according to the definition by Louria et al. [14]. Although judgment of pneumonia type was left to each attending physician, pure influenza viral pneumonia seemed to be deduced in young patients with a nonproductive cough, ground-glass opacity on chest X-ray in the diffuse lung or bilateral lungs, and leukocyte depression. Pure influenza viral pneumonia is typically considered rare in interpandemic periods [24]; however, the pattern for pure influenza viral pneumonia is clearly distinguishable from that for typical bacterial pneumonia, and the presence of more than a few cases of pure influenza viral pneumonia was strongly suggested. In the primary measurement, higher classes/grades were associated with increased mortality. The same tendency was
noted even when prediction models were adapted for each type of pneumonia in the detailed analysis. Because the current routine prediction models are weighted by advanced age, the severity of pneumonia might have been underestimated. Although similar phenomena could occur with atypical pneumonia [25,26], atypical pathogens, especially M. pneumoniae and Chlamydophila pneumoniae, are associated with a generally mild clinical course and low mortality [27,28]. On the other hand, severe pure influenza viral pneumonia was more frequently observed during this pandemic period [29,30], providing a possible explanation for the poor performance of the routine models. Although the type of pneumonia influences the PSI and because we failed to calculate the odds ratio for mild influenza pneumonia, both CURB-65 and A-DROP were found to be fair models because they were not affected by pneumonia patterns. In addition, A-DROP showed the highest accuracy for evaluating severity and mortality on ROC analysis. As a secondary measurement, the trend between class/ grade and IPPV indication was similar to that of mortality. On the other hand, in the detailed analysis that entailed adapting prediction models for each type of pneumonia, the opposite trend was seen when adapting CURB-65 to assess IPPV. However, the small sample size, especially in the upper class/grade, made it difficult to determine significance. Notably, in the severity analysis, the group classified as mild
284
respiratory investigation 52 (2014) 280 –287
Fig. 2 – Observed and predicted mortality according to (A) pneumonia severity index (PSI), (B) CURB-65 (confusion, urea, respiratory rate, blood pressure, and age Z 65 years), and (C) A-DROP (age, dehydration, respiration, disorientation, and blood pressure). Solid lines indicate mortality in this study, and dotted lines show predicted mortality from other reports. *po0.001 indicates a tendency of higher mortality in higher-severity classes/grades (by the Cochran–Armitage test).
pneumonia included cases of severe pneumonia requiring IPPV. Moreover, pure influenza viral pneumonia was found to affect the indications for IPPV. Pneumonia prediction models were originally used to decide on the appropriate site of care [9,10,31]. However, the routine prediction models underestimate the true severity of pneumonia and may, therefore, not always be helpful when choosing the appropriate site. Interestingly, type of pneumonia, especially pure influenza viral pneumonia, was an independent risk factor for severity but not for mortality by the routine prediction models. It is generally accepted that pure influenza viral pneumonia exhibited a severe clinical course in the 2009 pandemic, but it was not associated with higher mortality in the present study. In fact, the data show a relatively low inhospital mortality rate (8.6%) for pandemic influenza pneumonia compared with that reported in other studies (6.0– 15.2%) [29,32–34]. Other reports mentioned that one of the factors contributing to the deaths may have been delays in and fewer cases of oseltamivir therapy [3,35]. Our previous report revealed a higher proportion of anti-influenza drug use
in Japan, which could have contributed to the lower mortality rate [15]. Thus, early and more widespread use of antiinfluenza viral drugs is proposed as one way to reduce mortality [20,36]. Several limitations of this study should be noted. First, the questionnaire was retrospective and voluntary, and the pneumonia pattern diagnosis depended on the attending physician's clinical judgment; thus, the true proportion of viral/bacterial pneumonia was not necessarily reflected by the data. However, many cases were reported and included various patterns of influenza pneumonia. The data appeared to include more than a few cases of pure influenza viral pneumonia, which influenced pneumonia severity and indicated IPPV treatment. Second, age was categorical and not provided as detailed numerical data (in line with the guidance of the Institutional Review Board), which led to inaccurate calculations in each prediction model. However, categorical data made the questionnaires easier to answer and may have improved the response rate. Finally, as described in a previous report [15], recall bias was likely a
285
respiratory investigation 52 (2014) 280 –287
Table 2 – Multiple logistic regression models for mortality evaluation. p Value
Odds ratio [95% CI*]
PSI I vs. II I vs. III I vs. IV I vs. V Types of pneumonia
0.247 0.104 0.003 o0.001 0.027
3.9 [0.4–38.1] 6.7 [0.7–67.0] 23.2 [2.9–187.1] 105.3 [10.5–1055.1] –
CURB-65 Mild vs. intermediate Mild vs. high Types of pneumonia
o0.001 0.001 0.124
6.7 [2.5–18.1] 10.9 [2.8–42.0] –
A-DROP Mild vs. moderate Mild vs. severe Mild vs. extremely severe Types of pneumonia
0.012 o0.001 o0.001 0.125
13.8 [1.8–106.3] 101.7 [10.1–1023.1] 104.4 [9.7–1123.8] –
Abbreviations: PSI, pneumonia severity index; CURB-65, confusion, urea, respiratory rate, blood pressure, and age Z 65 years; and A-DROP, age, dehydration, respiration, disorientation, and blood pressure. n CI, Confidence interval.
Fig. 3 – Proportion of invasive positive pressure ventilation (IPPV) according to (A) pneumonia severity index (PSI), (B) CURB-65, and (C) A-DROP. Pneumonia severity was evaluated using IPPV as an indicator. *po0.001 indicates a tendency of frequent IPPV usage in higher-severity classes/grades (by the Cochran–Armitage test).
286
respiratory investigation 52 (2014) 280 –287
Table 3 – Multiple logistic regression models for severity evaluation by IPPV usage. p Value
Odds ratio [95% CI*]
PSI I vs. II I vs. III I vs. IV I vs. V Types of pneumonia
0.015 0.022 o0.001 o0.001 o0.001
13.7 [1.7–112.3] 13.1 [1.5–117.7] 72.0 [9.1–571.4] 116.1 [11.3–1197.6] –
CURB-65 Mild vs. intermediate Mild vs. high Types of pneumonia
o0.001 0.001 o0.001
8.5 [3.5–20.6] 9.3 [2.5–34.3] –
A-DROP Mild vs. moderate Mild vs. severe Mild vs. extremely severe Types of pneumonia
o0.001 0.007 o0.001 0.002
17.5 [4.1–75.3] 18.5 [2.3–152.0] 256.4 [30.2–2173.8] –
Abbreviations: IPPV, invasive positive pressure ventilation. CI, Confidence Interval.
n
that actual mortality is underestimated and that the mortality rates observed were higher than predicted. A-DROP showed the highest accuracy, but all models underestimated severity and the appropriate site of care for patients with influenza pneumonia; therefore, more-appropriate models for influenza pneumonia are required.
Conflict of interest This study received no financial or material support. The authors have no conflicts of interest.
Appendix A.
Supplementary information
Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.resinv. 2014.04.003.
Fig. 4 – Discriminatory power of each prediction model by receiver operating characteristic (AUC, area under the curve). (a) Mortality and (b) Severity.
factor. Nonetheless, because the study targeted almost 2500 respiratory departments at hospitals nationwide, the responses seem reasonable and should reflect the trends in pneumonia during the 2009 pandemic.
5.
Conclusions
Of the current pneumonia prediction models, both CURB-65 and A-DROP are fair predictors because they are not influenced by pneumonia patterns. However, it should be noted
r e f e r e n c e s
[1] Thompson WW, Shay DK, Weintraub E, et al. Mortality associated with influenza and respiratory syncytial virus in the United States. J Am Med Assoc 2003;289:179–86. [2] Dominguez-Cherit G, Lapinsky SE, Macias AE, et al. Critically Ill patients with 2009 influenza A (H1N1) in Mexico. J Am Med Assoc 2009;302:1880–7. [3] Perez-Padilla R, de la Rosa-Zamboni D, Ponce de Leon S, et al. Pneumonia and respiratory failure from swine-origin influenza A (H1N1) in Mexico. N Engl J Med 2009;29:680–9. [4] Dawood FS, Jain S, Finelli L, et al. Emergence of a novel swine-origin influenza A (H1N1) virus in humans. N Engl J Med 2009;360:2605–15. [5] CDC. Hospitalized patients with novel influenza A (H1N1) virus infection – California, April–May, 2009. Morb Mortal Wkly Rep 2009;58:536–41.
respiratory investigation 52 (2014) 280 –287
[6] CDC. Intensive-care patients with severe novel influenza A (H1N1) virus infection – Michigan, June 2009. Morb Mortal Wkly Rep 2009;58:749–52. [7] Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336:243–50. [8] Lim WS, van der Eerden MM, Laing R, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax 2003;58:377–82. [9] The Committee for the Japanese Respiratory Society Guidelines in the Management of Respiratory Infections. Guidelines for the management of community acquired pneumonia in adults, revised edition. Respirology 2006;11: S79–133. [10] Mandell LA, Wunderink RG, Anzueto A, et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of communityacquired pneumonia in adults. Clin Infect Dis 2007;1:S27–72. [11] Muller MP, McGeer AJ, Hassan K, et al. Evaluation of pneumonia severity and acute physiology scores to predict ICU admission and mortality in patients hospitalized for influenza. PLoS One 2010;5:e9563. [12] Pereira JM, Moreno RP, Matos R, et al. Severity assessment tools in ICU patients with 2009 influenza A (H1N1) pneumonia. Clin Microbiol Infect 2012;18:1040–8. [13] Bjarnason A, Thorleifsdottir G, Love A, et al. Severity of influenza A 2009 (H1N1) pneumonia is underestimated by routine prediction rules. Results from a prospective, population-based study. PLoS One 2012;7:e46816. [14] Louria DB, Blumenfeld HL, Ellis JT, et al. Studies on influenza in the pandemic of 1957–1958. II. Pulmonary complications of influenza. J Clin Invest 1959;38:213–65. [15] Fujikura Y, Kawano S, Kouzaki Y, et al. The (H1N1) 2009 pandemic influenza pneumonia among adult patients in Japan. Jpn J Infect Dis 2014;67:100–4. [16] Infectious Disease Surveillance Center (IDSC). Analysis of seasonal and pandemic (H1N1) 2009 influenza viruses isolated in 2009/10 season in Japan. Infect Agents Surveill Rep 2010;31:253–60. [17] Kohno S, Seki M, Takehara K, et al. Prediction of requirement for mechanical ventilation in community-acquired pneumonia with acute respiratory failure: a multicenter prospective study. Respir Int Rev Thorac Dis 2013;85:27–35. [18] Kohno S, Seki M, Watanabe A. Evaluation of an assessment system for the JRS 2005: A-DROP for the management of CAP in adults. Int Med 2011;50:1183–91. [19] Estenssoro E, Rios FG, Apezteguia C, et al. Pandemic 2009 influenza A in Argentina: a study of 337 patients on mechanical ventilation. Am J Respir Crit Care Med 2010;182:41–8. [20] Nguyen-Van-Tam JS, Openshaw PJ, Hashim A, et al. Risk factors for hospitalisation and poor outcome with pandemic A/H1N1 influenza: United Kingdom first wave (May– September 2009). Thorax 2010;65:645–51.
287
[21] Webb SA, Pettila V, Seppelt I, et al. Critical care services and 2009 H1N1 influenza in Australia and New Zealand. N Engl J Med 2009;361:1925–34. [22] Skarbinski J, Jain S, Bramley A, et al. Hospitalized patients with 2009 pandemic influenza A (H1N1) virus infection in the United States – September–October 2009. Clin Infect Dis 2011;52(Suppl. 1):S50–9. [23] The review meeting on measures against pandemic influenza (A/H1N1). Tokyo; 2010 [cited 2010]. Available from: 〈http://www.mhlw.go.jp/bunya/kenkou/ kekkaku-kansenshou04/dl/infu100331-02.pdf〉. [24] Treanor JJ. Influenza viruses, including avian influenza and swine influenza. In: Mandell GL, Bennett JE, Dolin R, Principles and practice of infectious diseases, 7th edition. Philadelphia, PA: Elsevier; 2010. p. 2265–88. [25] Cilloniz C, Ewig S, Polverino E, et al. Microbial aetiology of community-acquired pneumonia and its relation to severity. Thorax 2011;66:340–6. [26] Singanayagam A, Chalmers JD. Severity assessment scores to guide empirical use of antibiotics in community acquired pneumonia. Lancet Respir Med 2013;1:653–62. [27] Ruiz M, Ewig S, Marcos MA, et al. Etiology of communityacquired pneumonia: impact of age, comorbidity, and severity. Am J Respir Crit Care Med 1999;160:397–405. [28] von Baum H, Welte T, Marre R, et al. Mycoplasma pneumoniae pneumonia revisited within the German Competence Network for Community-acquired pneumonia (CAPNETZ). BMC Infect Dis 2009;9:62. [29] Cui W, Zhao H, Lu X, et al. Factors associated with death in hospitalized pneumonia patients with 2009 H1N1 influenza in Shenyang, China. BMC Infect Dis 2010;10:145. [30] Mauad T, Hajjar LA, Callegari GD, et al. Lung pathology in fatal novel human influenza A (H1N1) infection. Am J Respir Crit Care Med 2010;181:72–9. [31] Niederman MS. Making sense of scoring systems in community acquired pneumonia. Respirology 2009;14:327–35. [32] Bai L, Gu L, Cao B, et al. Clinical features of pneumonia caused by 2009 influenza A (H1N1) virus in Beijing, China. Chest 2011;139:1156–64. [33] Mulrennan S, Tempone SS, Ling IT, et al. Pandemic influenza (H1N1) 2009 pneumonia: CURB-65 score for predicting severity and nasopharyngeal sampling for diagnosis are unreliable. PLoS One 2010;5:e12849. [34] Kok J, Blyth CC, Foo H, et al. Viral pneumonitis is increased in obese patients during the first wave of pandemic A (H1N1) 2009 virus. PLoS One 2013;8:e55631. [35] Jain S, Kamimoto L, Bramley AM, et al. Hospitalized patients with 2009 H1N1 influenza in the United States – April–June 2009. N Engl J Med 2009;361:1935–44. [36] Higuera Iglesias AL, Kudo K, Manabe T, et al. Reducing occurrence and severity of pneumonia due to pandemic H1N1 2009 by early oseltamivir administration: a retrospective study in Mexico. PLoS One 2011;6:e21838.
Contents lists available at ScienceDirect
Respiratory Investigation journal homepage: www.elsevier.com/locate/resinv
Original article
Mortality and severity evaluation by routine pneumonia prediction models among Japanese patients with 2009 pandemic influenza A (H1N1) pneumonia Yuji Fujikura, M.D.a,n, Shuichi Kawano, M.D.a, Yuji Kouzaki, M.D.a, Masahiro Shinoda, M.D.b, Yu Hara, M.D.a, Masaharu Shinkai, M.D., Ph.D.b, Soichiro Kanoh, M.D., Ph.D.a, Akihiko Kawana, M.D., Ph.D.a a
Division of Infectious Diseases and Respiratory Medicine, Department of Internal Medicine, National Defense Medical College, 3-2 Namiki, Tokorozawa, Saitama 359-8513, Japan b Respiratory Disease Center, Yokohama City University Medical Center, 4-57 Urafunecho, Minami-ku, Yokohama, Kanagawa 232-0024, Japan
ar t ic l e in f o
abs tra ct
Article history:
Background: Influenza-related pneumonia, referred to as influenza pneumonia, was
Received 18 February 2014
reported relatively more frequently during a recent influenza pandemic in 2009. The
Received in revised form
validity of adapting routine pneumonia severity prediction models for various types of
17 April 2014
pneumonia is unclear.
Accepted 23 April 2014
Methods: We conducted a nationwide survey to evaluate influenza pneumonia among adult
Available online 23 June 2014
patients in Japan. Questionnaires were sent to physicians working in departments of
Keywords: Influenza Influenza pneumonia Pneumonia severity index CURB-65 A-DROP
respiratory medicine at 2491 hospitals. Both the outcome and pneumonia severity, using invasive positive pressure ventilation (IPPV) as an indicator, were evaluated by routine pneumonia severity index (PSI), CURB-65 (confusion, urea, respiratory rate, blood pressure, and ageZ65 years), and A-DROP (age, dehydration, respiration, disorientation, and blood pressure). Results: Data collected from 320 patients with influenza pneumonia, including 25 cases (7.8%) of death and 43 (13.4%) of IPPV, were analyzed. Although all routine prediction models showed that higher mortality tended to be associated with a higher risk class/grade, the actual mortality rates were higher than predicted. The risk class of mortality calculated by the PSI was influenced by pneumonia patterns. Although pneumonia severity was similarly predicted, the types of pneumonia also affected severity in all prediction models. A-DROP showed the highest accuracy on receiver operating characteristic analysis for both mortality and severity. Conclusions: CURB-65 and A-DROP are fair predictors of mortality regardless of pneumonia patterns. However, the current pneumonia prediction models may underestimate the severity and appropriate site of care for patients with influenza pneumonia. & 2014 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.resinv.2014.04.003 2212-5345/& 2014 The Japanese Respiratory Society. Published by Elsevier B.V. All rights reserved.
281
respiratory investigation 52 (2014) 280 –287
1.
2.2.
Introduction
Influenza has a worldwide distribution and causes increased morbidity and mortality each year [1]. Pneumonia develops in some patients with influenza, which can lead to influenzarelated death. Influenza-related pneumonia, referred to as influenza pneumonia, was reported relatively more frequently in the influenza pandemic of 2009, with significant morbidity and mortality worldwide [2–6]. Severity evaluation and prediction models have been established to manage patients with pneumonia, make clinical decisions, and improve outcome, especially in cases of community-acquired pneumonia (CAP). The major guidelines for CAP evaluation and prediction models include the pneumonia severity index (PSI) from the Infectious Diseases Society of America [7] and CURB-65 (confusion, urea, respiratory rate, blood pressure, and age Z65 years) from the British Thoracic Society [8]. In Japan, the A-DROP (age, dehydration, respiration, disorientation, and blood pressure) scoring system from the Japanese Respiratory Society [9] is usually adapted for patients with CAP. These models are validated and recommended as clinical guidelines [10]. On the other hand, some reports on the recent influenza pandemic suggested that routine prediction models such as the PSI and CURB-65 failed to predict the need for mechanical ventilation or to estimate mortality [11–13]. In the 2009 influenza pandemic, some patients with severe influenza pneumonia presented with viral pneumonia with poor outcome, which is not typical of the usual form of CAP. Three manifestations of pneumonia associated with influenza are well described [14]; therefore, the validity of adapting routine prediction for various types of pneumonia is unclear. In a previous study, we collected and analyzed data from 346 cases of influenza pneumonia, including 27 fatal cases [15]. The present study uses the same dataset to evaluate each routine pneumonia prediction model and the effects of influenza pneumonia among adult patients in Japan.
2.
Materials and methods
2.1.
Study design and data collection
In a retrospective cohort study of adult patients with influenza pneumonia, questionnaires were sent to physicians working in departments of respiratory medicine at 2491 hospitals (4200 beds) across Japan. Data were collected by mail. This study conformed to the principles of the Declaration of Helsinki and was approved by the Institutional Review Board (Approval no. 778, 26/4/2010).
Abbreviations: CAP,
community-acquired pneumonia; IPPV,
Study definition
The influenza pandemic period was defined as April 1, 2009– April 30, 2010. Patients were given a diagnosis of pandemic influenza (H1N1) 2009 by rapid immunofluorescence influenza diagnostic kits or reverse transcription polymerase chain reaction. From the epidemiological data, it is generally accepted that rapid diagnostic kit results showing influenza A infection indicated H1N1 infection during this pandemic period [16]. Pneumonia was diagnosed by an attending physician based on respiratory symptoms and the findings of physical and radiological examinations. Influenza pneumonia is classified into three categories according to the definition by Louria et al. [14]: (1) pure influenza viral pneumonia (i.e., primary influenza pneumonia where pneumonia is caused only by the influenza virus and no bacterial pathogens are detected), (2) secondary bacterial pneumonia (i.e., influenza is complicated by secondary bacterial pneumonia that develops once the influenza symptoms have resolved), and (3) mixed viral and bacterial pneumonia (i.e., the influenza virus and bacterial pneumonia occur concurrently).
2.3.
Questionnaires
Age (in 10-year brackets), sex, and comorbid conditions were recorded. As described in the study definition, all pneumonia cases were classified as pure influenza viral pneumonia, secondary bacterial pneumonia, mixed viral and bacterial pneumonia, or miscellaneous according to the attending physician's diagnosis. Invasive positive pressure ventilation (IPPV) use, treatment options, and outcome data were also collected.
2.4.
Pneumonia severity evaluation
The severity of influenza pneumonia was estimated using routine PSI, CURB-65, and A-DROP. In all cases, age was expressed as the median of each class. Patients were categorized into five classes by the PSI score [7] and into the following three groups according to the CURB-65 score: “group 1: mortality low” with a score of 0 or 1; “group 2: mortality intermediate” with a score of 2; and “group 3: mortality high” with a score of Z3 [8]. The A-DROP system is based on the following five clinical features: age (A), dehydration (D), respiration (R), orientation (O), and blood pressure (P). According to this system, cases are classified as “mild” when none of the five criteria are met, “moderate” when one or two of the criteria are met, “severe” when three of the criteria are met, or “extremely severe” when four or five of the criteria are met. Cases of shock or altered mental status are regarded as extremely severe [9].
invasive positive pressure ventilation; PSI,
pneumonia severity index;
ROC, receiver operating characteristic n Corresponding author. Tel.: þ81 4 2995 1211; fax: þ81 4 2995 1497. E-mail addresses: [email protected] (Y. Fujikura), [email protected] (S. Kawano), [email protected] (Y. Kouzaki), [email protected] (M. Shinoda), [email protected] (Y. Hara), [email protected] (M. Shinkai), [email protected] (S. Kanoh), [email protected] (A. Kawana).
282
2.5.
respiratory investigation 52 (2014) 280 –287
Outcome measures and statistical analysis
The primary measures were used to evaluate validity in terms of mortality by the routine prediction models (PSI, CURB-65, and A-DROP) for influenza pneumonia and the influence of pneumonia patterns. Secondary measures were used to evaluate pneumonia severity by using IPPV as an indicator [17]. The effect of pneumonia patterns on mortality or severity was also analyzed. Frequency analysis, with results expressed as percentages, was used for the descriptive statistics of nonparametric data. We performed the Shapiro–Wilk test for normality and the Kruskal–Wallis test for age-group comparisons. The Chi-square test was used to analyze pneumonia severity, use of mechanical ventilation, and outcomes, and the Fisher exact test was used to analyze small samples. The Cochran– Armitage test was used for trend analysis, and multiple logistic regression models were performed to analyze risk estimation. Po0.05 was considered significant. All statistical analyses were performed using SPSS version 21.0.0.2 (IBM Corporation, Armonk, NY) and R version 2.15.3 (R Foundation for Statistical Computing, Vienna, Austria) software.
3.
Results
3.1.
Patients
We collected data on 346 cases of influenza pneumonia from 994 physicians (overall questionnaire response rate, 39.9%). Patient characteristics, comorbid conditions, treatments, and outcomes are described elsewhere. We excluded 26 cases from the analysis because of insufficient data to enable a severity calculation (Fig. 1).
3.2.
Patient background, symptoms, and findings
Of the remaining 320 cases, pure influenza viral pneumonia was accounted for in 94 patients (29.4%), mixed viral and bacterial pneumonia in 127 (39.7%), secondary bacterial pneumonia in 53 (16.6%), and miscellaneous in 46 (14.4%). In addition, the dataset included 43 cases of IPPV (13.4%) and 25 cases of death (7.8%). The in-hospital mortality rate was 8.6%. Patient background, symptoms, and findings for each group (excluding the miscellaneous group) are given in
Table 1. Pure influenza viral pneumonia tended to affect young, obese patients (21.3%) presenting with a nonproductive cough. In these patients, detailed laboratory investigations revealed low white blood cell counts, and chest X-rays showed ground-glass opacity with diffuse distribution. Microbiological tests were performed in 196 (61.6%) patients, and Streptococcus pneumoniae was detected as the major agent in 67 (20.9%). Other detected pathogens included Hemophilus influenzae (n¼ 10, 3.1%), methicillin-sensitive Staphylococcus aureus (n ¼6, 1.9%), Mycoplasma pneumoniae (n¼ 4, 1.3%), Moraxella catarrhalis (n ¼4, 1.3%), Pseudomonas aeruginosa (n ¼4, 1.3%), Klebsiella pneumoniae (n¼ 2, 0.6%), Streptococcus viridans (n¼ 2, 0.6%), Enterobacter cloacae (n ¼1, 0.3%), and Enterococcus faecalis (n¼ 1, 0.3%).
3.3.
The number of patients and deaths and the observed and predicted mortality rates in each class/group according to the PSI, CURB-65, and A-DROP are shown in Fig. 2A–C, respectively. The PSI, CURB-65, and A-DROP classes/groups correlated well with one another (data not shown). All routine prediction models showed that higher mortality was related to a higher class/grade on the Cochran–Armitage tendency test (po0.001); however, the actual rate of mortality was higher than that predicted from the routine CAP prediction models [7,8,18]. The receiver operating characteristic (ROC) curve for mortality in each prediction model (by class/score) is shown in Fig. 4A. The highest accuracy, as evaluated by the area under the curve, was 0.820 for A-DROP. Multiple logistic regression analysis of PSI risk classes I versus II and I versus III revealed no significance between odds ratios. PSI evaluation of the types of pneumonia showed that their effect on mortality was independent (p¼ 0.027), especially in pure influenza viral pneumonia (Table 2).
3.4.
Severity evaluation by each prediction model
Fig. 3A–C shows the number of IPPV cases and proportion of IPPV cases in each class/group according to the PSI, CURB-65, and A-DROP, respectively. With regard to mortality, the use of IPPV tended to be greater in higher classes/grades on Cochran–Armitage tendency testing (po0.001). However, IPPV was used even in relatively mild classes/grades. The ROC curve for IPPV is shown in Fig. 4B. A-DROP had the highest area under the curve, of 0.778. In multiple logistic regression models, the need for IPPV was affected by the type of pneumonia, particularly pure influenza viral pneumonia, in all prediction models (Table 3). The calculations for each prediction model for each type of pneumonia are detailed in Fig. S1.
4.
Fig. 1 – Study patients.
Mortality evaluation by each prediction model
Discussion
This study shows that the current routine prediction models for CAP underestimate the true severity and mortality rates of influenza pneumonia. Although mortality tends to be increased in higher classes/grades, the overall mortality rates are higher than predicted [7,8,18]. The findings suggest that
respiratory investigation 52 (2014) 280 –287
283
Table 1 – Patient background, symptoms, and laboratory findings in each pneumonia type.
Patient background Age (years old, average) No underlying medical conditions Chronic respiratory diseases Chronic heart diseases Chronic renal diseases Chronic liver diseases Cerebrovascular disease Hematological diseases Diabetes mellitus Collagen vascular diseases Malignancy Obesity Symptom/physical examination Productive cough Nonproductive cough Dyspnea Crackles on physical examination Laboratory test/chest X-ray findings WBCo4000/μL PaO2o60 Torr or SpO2o90% Ground glass opacity Consolidation Extent of shadow more than 2/3 area of one lung
Pure influenza viral pneumonia n ¼ 94
Secondary bacterial pneumonia n ¼ 53
Mixed viral and bacterial pneumonia n ¼127
p Value
47 38 (40.4)
57 15 (28.3)
58 37 (29.1)
o0.001 0.154
14 5 1 2 0 4 7 2 3 20
(14.9) (5.3) (1.1) (2.1) (0) (4.3) (7.4) (2.1) (3.2) (21.3)
24 3 1 1 3 0 6 0 1 2
(45.3) (5.7) (1.9) (1.9) (5.7) (0) (11.3) (0) (1.9) (3.8)
43 20 6 4 15 1 11 3 1 8
(33.9) (15.7) (4.7) (3.1) (11.8) (0.8) (8.7) (2.4) (0.8) (6.3)
o0.001 0.004 0.073 0.395 o0.001 0.064 0.726 0.529 0.151 0.001
27 40 55 49
(29.0) (43.0) (59.1) (52.7)
34 8 15 18
(64.2) (15.1) (28.3) (34.0)
80 15 62 60
(63.0) (11.8) (48.8) (47.2)
o0.001 o0.001 0.002 0.090
21 43 58 44 41
(22.6) (46.2) (69.9) (53.0) (50.0)
1 17 10 44 6
(1.9) (32.1) (19.2) (84.6) (11.8)
7 52 42 94 31
(5.5) (40.9) (33.9) (75.8) (25.4)
o0.001 0.247 o0.001 o0.001 o0.001
Data are expressed as n (%) unless otherwise indicated. Abbreviations: WBC, white blood cell; PaO2, arterial partial pressure of oxygen; and SpO2, oxygen saturation by pulse oximetry.
the more-fulminant patterns of pneumonia associated with severe hypoxemia and acute respiratory distress syndrome were more common in the 2009 influenza pandemic. The data also suggest that influenza pneumonia is not typical in patients with CAP. As described in previous reports, in addition to patients with respiratory complications, those without any underlying medical conditions were also commonly observed in the 2009 pandemic [19–22], and severe influenza was found in young patients [2,3,5,15,23]. The patient background characteristics in the present study are similar to those reported previously. We classified influenza pneumonia into three types according to the definition by Louria et al. [14]. Although judgment of pneumonia type was left to each attending physician, pure influenza viral pneumonia seemed to be deduced in young patients with a nonproductive cough, ground-glass opacity on chest X-ray in the diffuse lung or bilateral lungs, and leukocyte depression. Pure influenza viral pneumonia is typically considered rare in interpandemic periods [24]; however, the pattern for pure influenza viral pneumonia is clearly distinguishable from that for typical bacterial pneumonia, and the presence of more than a few cases of pure influenza viral pneumonia was strongly suggested. In the primary measurement, higher classes/grades were associated with increased mortality. The same tendency was
noted even when prediction models were adapted for each type of pneumonia in the detailed analysis. Because the current routine prediction models are weighted by advanced age, the severity of pneumonia might have been underestimated. Although similar phenomena could occur with atypical pneumonia [25,26], atypical pathogens, especially M. pneumoniae and Chlamydophila pneumoniae, are associated with a generally mild clinical course and low mortality [27,28]. On the other hand, severe pure influenza viral pneumonia was more frequently observed during this pandemic period [29,30], providing a possible explanation for the poor performance of the routine models. Although the type of pneumonia influences the PSI and because we failed to calculate the odds ratio for mild influenza pneumonia, both CURB-65 and A-DROP were found to be fair models because they were not affected by pneumonia patterns. In addition, A-DROP showed the highest accuracy for evaluating severity and mortality on ROC analysis. As a secondary measurement, the trend between class/ grade and IPPV indication was similar to that of mortality. On the other hand, in the detailed analysis that entailed adapting prediction models for each type of pneumonia, the opposite trend was seen when adapting CURB-65 to assess IPPV. However, the small sample size, especially in the upper class/grade, made it difficult to determine significance. Notably, in the severity analysis, the group classified as mild
284
respiratory investigation 52 (2014) 280 –287
Fig. 2 – Observed and predicted mortality according to (A) pneumonia severity index (PSI), (B) CURB-65 (confusion, urea, respiratory rate, blood pressure, and age Z 65 years), and (C) A-DROP (age, dehydration, respiration, disorientation, and blood pressure). Solid lines indicate mortality in this study, and dotted lines show predicted mortality from other reports. *po0.001 indicates a tendency of higher mortality in higher-severity classes/grades (by the Cochran–Armitage test).
pneumonia included cases of severe pneumonia requiring IPPV. Moreover, pure influenza viral pneumonia was found to affect the indications for IPPV. Pneumonia prediction models were originally used to decide on the appropriate site of care [9,10,31]. However, the routine prediction models underestimate the true severity of pneumonia and may, therefore, not always be helpful when choosing the appropriate site. Interestingly, type of pneumonia, especially pure influenza viral pneumonia, was an independent risk factor for severity but not for mortality by the routine prediction models. It is generally accepted that pure influenza viral pneumonia exhibited a severe clinical course in the 2009 pandemic, but it was not associated with higher mortality in the present study. In fact, the data show a relatively low inhospital mortality rate (8.6%) for pandemic influenza pneumonia compared with that reported in other studies (6.0– 15.2%) [29,32–34]. Other reports mentioned that one of the factors contributing to the deaths may have been delays in and fewer cases of oseltamivir therapy [3,35]. Our previous report revealed a higher proportion of anti-influenza drug use
in Japan, which could have contributed to the lower mortality rate [15]. Thus, early and more widespread use of antiinfluenza viral drugs is proposed as one way to reduce mortality [20,36]. Several limitations of this study should be noted. First, the questionnaire was retrospective and voluntary, and the pneumonia pattern diagnosis depended on the attending physician's clinical judgment; thus, the true proportion of viral/bacterial pneumonia was not necessarily reflected by the data. However, many cases were reported and included various patterns of influenza pneumonia. The data appeared to include more than a few cases of pure influenza viral pneumonia, which influenced pneumonia severity and indicated IPPV treatment. Second, age was categorical and not provided as detailed numerical data (in line with the guidance of the Institutional Review Board), which led to inaccurate calculations in each prediction model. However, categorical data made the questionnaires easier to answer and may have improved the response rate. Finally, as described in a previous report [15], recall bias was likely a
285
respiratory investigation 52 (2014) 280 –287
Table 2 – Multiple logistic regression models for mortality evaluation. p Value
Odds ratio [95% CI*]
PSI I vs. II I vs. III I vs. IV I vs. V Types of pneumonia
0.247 0.104 0.003 o0.001 0.027
3.9 [0.4–38.1] 6.7 [0.7–67.0] 23.2 [2.9–187.1] 105.3 [10.5–1055.1] –
CURB-65 Mild vs. intermediate Mild vs. high Types of pneumonia
o0.001 0.001 0.124
6.7 [2.5–18.1] 10.9 [2.8–42.0] –
A-DROP Mild vs. moderate Mild vs. severe Mild vs. extremely severe Types of pneumonia
0.012 o0.001 o0.001 0.125
13.8 [1.8–106.3] 101.7 [10.1–1023.1] 104.4 [9.7–1123.8] –
Abbreviations: PSI, pneumonia severity index; CURB-65, confusion, urea, respiratory rate, blood pressure, and age Z 65 years; and A-DROP, age, dehydration, respiration, disorientation, and blood pressure. n CI, Confidence interval.
Fig. 3 – Proportion of invasive positive pressure ventilation (IPPV) according to (A) pneumonia severity index (PSI), (B) CURB-65, and (C) A-DROP. Pneumonia severity was evaluated using IPPV as an indicator. *po0.001 indicates a tendency of frequent IPPV usage in higher-severity classes/grades (by the Cochran–Armitage test).
286
respiratory investigation 52 (2014) 280 –287
Table 3 – Multiple logistic regression models for severity evaluation by IPPV usage. p Value
Odds ratio [95% CI*]
PSI I vs. II I vs. III I vs. IV I vs. V Types of pneumonia
0.015 0.022 o0.001 o0.001 o0.001
13.7 [1.7–112.3] 13.1 [1.5–117.7] 72.0 [9.1–571.4] 116.1 [11.3–1197.6] –
CURB-65 Mild vs. intermediate Mild vs. high Types of pneumonia
o0.001 0.001 o0.001
8.5 [3.5–20.6] 9.3 [2.5–34.3] –
A-DROP Mild vs. moderate Mild vs. severe Mild vs. extremely severe Types of pneumonia
o0.001 0.007 o0.001 0.002
17.5 [4.1–75.3] 18.5 [2.3–152.0] 256.4 [30.2–2173.8] –
Abbreviations: IPPV, invasive positive pressure ventilation. CI, Confidence Interval.
n
that actual mortality is underestimated and that the mortality rates observed were higher than predicted. A-DROP showed the highest accuracy, but all models underestimated severity and the appropriate site of care for patients with influenza pneumonia; therefore, more-appropriate models for influenza pneumonia are required.
Conflict of interest This study received no financial or material support. The authors have no conflicts of interest.
Appendix A.
Supplementary information
Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.resinv. 2014.04.003.
Fig. 4 – Discriminatory power of each prediction model by receiver operating characteristic (AUC, area under the curve). (a) Mortality and (b) Severity.
factor. Nonetheless, because the study targeted almost 2500 respiratory departments at hospitals nationwide, the responses seem reasonable and should reflect the trends in pneumonia during the 2009 pandemic.
5.
Conclusions
Of the current pneumonia prediction models, both CURB-65 and A-DROP are fair predictors because they are not influenced by pneumonia patterns. However, it should be noted
r e f e r e n c e s
[1] Thompson WW, Shay DK, Weintraub E, et al. Mortality associated with influenza and respiratory syncytial virus in the United States. J Am Med Assoc 2003;289:179–86. [2] Dominguez-Cherit G, Lapinsky SE, Macias AE, et al. Critically Ill patients with 2009 influenza A (H1N1) in Mexico. J Am Med Assoc 2009;302:1880–7. [3] Perez-Padilla R, de la Rosa-Zamboni D, Ponce de Leon S, et al. Pneumonia and respiratory failure from swine-origin influenza A (H1N1) in Mexico. N Engl J Med 2009;29:680–9. [4] Dawood FS, Jain S, Finelli L, et al. Emergence of a novel swine-origin influenza A (H1N1) virus in humans. N Engl J Med 2009;360:2605–15. [5] CDC. Hospitalized patients with novel influenza A (H1N1) virus infection – California, April–May, 2009. Morb Mortal Wkly Rep 2009;58:536–41.
respiratory investigation 52 (2014) 280 –287
[6] CDC. Intensive-care patients with severe novel influenza A (H1N1) virus infection – Michigan, June 2009. Morb Mortal Wkly Rep 2009;58:749–52. [7] Fine MJ, Auble TE, Yealy DM, et al. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl J Med 1997;336:243–50. [8] Lim WS, van der Eerden MM, Laing R, et al. Defining community acquired pneumonia severity on presentation to hospital: an international derivation and validation study. Thorax 2003;58:377–82. [9] The Committee for the Japanese Respiratory Society Guidelines in the Management of Respiratory Infections. Guidelines for the management of community acquired pneumonia in adults, revised edition. Respirology 2006;11: S79–133. [10] Mandell LA, Wunderink RG, Anzueto A, et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of communityacquired pneumonia in adults. Clin Infect Dis 2007;1:S27–72. [11] Muller MP, McGeer AJ, Hassan K, et al. Evaluation of pneumonia severity and acute physiology scores to predict ICU admission and mortality in patients hospitalized for influenza. PLoS One 2010;5:e9563. [12] Pereira JM, Moreno RP, Matos R, et al. Severity assessment tools in ICU patients with 2009 influenza A (H1N1) pneumonia. Clin Microbiol Infect 2012;18:1040–8. [13] Bjarnason A, Thorleifsdottir G, Love A, et al. Severity of influenza A 2009 (H1N1) pneumonia is underestimated by routine prediction rules. Results from a prospective, population-based study. PLoS One 2012;7:e46816. [14] Louria DB, Blumenfeld HL, Ellis JT, et al. Studies on influenza in the pandemic of 1957–1958. II. Pulmonary complications of influenza. J Clin Invest 1959;38:213–65. [15] Fujikura Y, Kawano S, Kouzaki Y, et al. The (H1N1) 2009 pandemic influenza pneumonia among adult patients in Japan. Jpn J Infect Dis 2014;67:100–4. [16] Infectious Disease Surveillance Center (IDSC). Analysis of seasonal and pandemic (H1N1) 2009 influenza viruses isolated in 2009/10 season in Japan. Infect Agents Surveill Rep 2010;31:253–60. [17] Kohno S, Seki M, Takehara K, et al. Prediction of requirement for mechanical ventilation in community-acquired pneumonia with acute respiratory failure: a multicenter prospective study. Respir Int Rev Thorac Dis 2013;85:27–35. [18] Kohno S, Seki M, Watanabe A. Evaluation of an assessment system for the JRS 2005: A-DROP for the management of CAP in adults. Int Med 2011;50:1183–91. [19] Estenssoro E, Rios FG, Apezteguia C, et al. Pandemic 2009 influenza A in Argentina: a study of 337 patients on mechanical ventilation. Am J Respir Crit Care Med 2010;182:41–8. [20] Nguyen-Van-Tam JS, Openshaw PJ, Hashim A, et al. Risk factors for hospitalisation and poor outcome with pandemic A/H1N1 influenza: United Kingdom first wave (May– September 2009). Thorax 2010;65:645–51.
287
[21] Webb SA, Pettila V, Seppelt I, et al. Critical care services and 2009 H1N1 influenza in Australia and New Zealand. N Engl J Med 2009;361:1925–34. [22] Skarbinski J, Jain S, Bramley A, et al. Hospitalized patients with 2009 pandemic influenza A (H1N1) virus infection in the United States – September–October 2009. Clin Infect Dis 2011;52(Suppl. 1):S50–9. [23] The review meeting on measures against pandemic influenza (A/H1N1). Tokyo; 2010 [cited 2010]. Available from: 〈http://www.mhlw.go.jp/bunya/kenkou/ kekkaku-kansenshou04/dl/infu100331-02.pdf〉. [24] Treanor JJ. Influenza viruses, including avian influenza and swine influenza. In: Mandell GL, Bennett JE, Dolin R, Principles and practice of infectious diseases, 7th edition. Philadelphia, PA: Elsevier; 2010. p. 2265–88. [25] Cilloniz C, Ewig S, Polverino E, et al. Microbial aetiology of community-acquired pneumonia and its relation to severity. Thorax 2011;66:340–6. [26] Singanayagam A, Chalmers JD. Severity assessment scores to guide empirical use of antibiotics in community acquired pneumonia. Lancet Respir Med 2013;1:653–62. [27] Ruiz M, Ewig S, Marcos MA, et al. Etiology of communityacquired pneumonia: impact of age, comorbidity, and severity. Am J Respir Crit Care Med 1999;160:397–405. [28] von Baum H, Welte T, Marre R, et al. Mycoplasma pneumoniae pneumonia revisited within the German Competence Network for Community-acquired pneumonia (CAPNETZ). BMC Infect Dis 2009;9:62. [29] Cui W, Zhao H, Lu X, et al. Factors associated with death in hospitalized pneumonia patients with 2009 H1N1 influenza in Shenyang, China. BMC Infect Dis 2010;10:145. [30] Mauad T, Hajjar LA, Callegari GD, et al. Lung pathology in fatal novel human influenza A (H1N1) infection. Am J Respir Crit Care Med 2010;181:72–9. [31] Niederman MS. Making sense of scoring systems in community acquired pneumonia. Respirology 2009;14:327–35. [32] Bai L, Gu L, Cao B, et al. Clinical features of pneumonia caused by 2009 influenza A (H1N1) virus in Beijing, China. Chest 2011;139:1156–64. [33] Mulrennan S, Tempone SS, Ling IT, et al. Pandemic influenza (H1N1) 2009 pneumonia: CURB-65 score for predicting severity and nasopharyngeal sampling for diagnosis are unreliable. PLoS One 2010;5:e12849. [34] Kok J, Blyth CC, Foo H, et al. Viral pneumonitis is increased in obese patients during the first wave of pandemic A (H1N1) 2009 virus. PLoS One 2013;8:e55631. [35] Jain S, Kamimoto L, Bramley AM, et al. Hospitalized patients with 2009 H1N1 influenza in the United States – April–June 2009. N Engl J Med 2009;361:1935–44. [36] Higuera Iglesias AL, Kudo K, Manabe T, et al. Reducing occurrence and severity of pneumonia due to pandemic H1N1 2009 by early oseltamivir administration: a retrospective study in Mexico. PLoS One 2011;6:e21838.
