If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence to Jong Hun Kim, MD, PhD, Division of Infectious Diseases, Department of Internal Medicine, CHA Bundang Medical Center, CHA University, 59 Yatap-ro, Bundang-gu, Seongnam, 13496, Republic of Korea.
Marcus Institute for Aging Research, Hebrew SeniorLife, Boston, MA, USADivision of Gerontology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA, USA
Risk stratification tools are useful to provide appropriate clinical care for older patients with pneumonia. This study aimed to compare a Frailty Index (FI) with pneumonia severity measures, CURB-65, and the Pneumonia Severity Index (PSI), for predicting mortality and persistent disability after pneumonia.
Design
Single-center prospective cohort study.
Setting and Participants
The study included 190 patients aged ≥65 years who were hospitalized with pneumonia at a university hospital in Korea between October 2019 and September 2020.
Methods
At admission, a 50-item deficit-accumulation FI (range: 0-1), CURB-65 (range: 0-5), and PSI (range: 0-395) scores were calculated. The outcomes were death and a composite outcome of death or decline in ability to perform daily activities and physical task 6 months later.
Results
The median age was 79 years (interquartile range: 74-85), and 70 (36.8%) patients were women. The patients who died (n = 53) had higher FI (median, 0.46 vs 0.20; P < .011), CURB-65 score (median, 3 vs 2; P = .001), and PSI score (median, 149 vs 116; P < .001) than those who did not. The C-statistics (95% confidence intervals) for 6-month mortality were 0.69 (0.61-0.77) for the FI, 0.62 (0.53-0.71) for CURB-65, and 0.71 (0.62-0.79) for the PSI (P = .019). The C-statistics for the 6-month composite outcome were 0.73 (0.65-0.81) for the FI, 0.64 (0.55-0.73) for CURB-65, and 0.69 (0.60-0.77) for the PSI (P = .096). The C-statistics improved when the FI was added to CURB-65 (from 0.64 to 0.74; P = .003) and to the PSI (from 0.69 to 0.75; P = .044) for the composite outcome.
Conclusions and Implications
Measuring frailty provides additive value to widely used pneumonia severity measures in predicting death or persistent hospitalization-associated disability in older adults after pneumonia hospitalization. Early recognition of frailty may be useful to identify those who require in-hospital and post-acute care interventions for functional recovery.
Its annual incidence is 3.1 to 8.4 individuals per 1000 population, with a mortality rate ranging from 7.8% in patients aged ≥65 years to 15% in those aged ≥90 years.
Predicting mortality at the time of presentation is crucial to determine the site of care and provide appropriate clinical management. Currently, there are 2 widely used measures of pneumonia severity, namely, CURB-65 (Confusion of new onset; elevated blood Urea nitrogen >7 mmol/L; Respiratory rate ≥30 breaths per minute; Blood pressure <90/60 mmHg; and age ≥65 years)
and the Pneumonia Severity Index (PSI), which uses demographic and clinical information on comorbidities, physical examination findings, vital signs, and laboratory and radiographic findings.
Both scores are used to predict 30-day survival and determine the level of care for patients with pneumonia.
These measures, developed from adult patients, do not consider frailty, a common geriatric condition that increases the risk of adverse health outcomes after an acute illness in older adults.
Frailty is defined as a vulnerable state characterized by decrease in physiologic reserve that results from aging-related changes and accumulation of chronic conditions.
Previous research showed that frailty provides additional value to disease-specific severity measures in predicting mortality after acute medical admissions and surgical procedures.
Compared with pneumonia severity measures, which were developed to predict short-term mortality, frailty assessment may allow clinicians to predict older patients at high risk for longer-term mortality and persistent functional impairment after pneumonia hospitalization. A longer course of rehabilitation, supportive services, and palliative care may be necessary for patients with frailty.
we hypothesized that frailty would improve prediction of death or persistent disability after pneumonia hospitalization when combined with a pneumonia severity score.
Methods
Study Design and Participants
A prospective cohort study was conducted at a university hospital in Korea, with 201 patients aged ≥65 years who were hospitalized with pneumonia between October 2019 and September 2020. We screened 292 patients, and 91 were excluded for the following reasons: (1) patient refused (n = 58); (2) the research team was unavailable (n = 14); (3) informed consent could not be obtained (n = 14); or (4) there was a change in diagnosis after admission (n = 5). During the study period, patients diagnosed with the coronavirus disease were sent to government-designated hospitals, and only those with negative results were admitted. At 6 months, 11 patients were lost to follow-up and 190 patients were included for the current analysis. Our study was approved by the Institutional Review Board, Seoul, Korea, and a written informed consent was obtained from all patients or their caretakers.
Baseline Assessments
At admission, study clinicians evaluated sociodemographic characteristics, admission source (nursing home vs home), vital signs, body mass index, medical comorbidities, and self-reported functional status. Self-reported preadmission functional status was assessed using a questionnaire including 21 items: 7 activities of daily living (feeding, dressing, grooming, ambulating, transferring, bathing, and toileting), 7 instrumental activities of daily living (doing housework, making telephone calls, using transportation or driving, shopping, cooking, taking medications, and managing money), and 7 activities in the Nagi and Rosow-Breslau scales (pulling or pushing a large object, lifting 5 kg, walking up and down a flight of stairs, walking 1 km, writing or handling small objects, reaching arms above shoulder, and stooping, crouching, or keeling).
was calculated using 50 items from the baseline assessment—25 comorbidities, polypharmacy (≥5 prescription drugs), self-reported ability to perform 21 activities listed above, weight loss >5 kg in the past year, body mass index <21, and serum albumin level <3.5 g/L). Based on FI (range: 0-1), patients were classified into robust (<0.15), prefrail (0.15-0.24), mild to moderately frail (0.25-0.44), and severely frail (≥0.45) categories. CURB-65 (range: 0-5) and PSI (range: 0-395) scores were calculated using information collected at admission.
Study clinicians (CMP, WSK) conducted telephone interviews with patients or their proxies to evaluate the above-listed functional status at 1, 3, and 6 months after baseline assessment. A disability score (range: 0-21) was calculated as the total number of activities requiring another person's assistance. Our study outcomes were (1) death and (2) a composite outcome of death or functional decline, defined as any worsening ability to perform the 21 activities between baseline and at 6 months.
Statistical Analysis
The analysis for death included all 190 patients, and the analysis for the composite outcome of death or functional decline at 6 months included 159 patients who did not have the maximum level of disability on admission. The baseline characteristics of patients who experienced the outcome and those who did not were compared between using t test, Wilcoxon rank-sum test, or Fisher exact test. We examined the association of each risk score (FI, CURB-65, and PSI) as a continuous variable (after standardization) as well as a categorical variable with the outcomes using logistic models to adjust for age and sex. The discriminatory ability of each risk score and their combinations (FI and CURB-65, and FI and PSI) was evaluated using C-statistics. The C-statistics were compared using 1000 bootstrap resampling. A 2-sided P value of <.05 was considered statistically significant. Analysis was performed using Stata, version 16 (StataCorp, LLC, College Station, TX).
Results
Characteristics of Study Population
The median age was 79 years (interquartile range: 74, 85), 70 patients (36.8%) were women, and 35 patients (18.4%) were admitted from nursing homes (Table 1). The patients who died (n = 53) had higher FI (median, 0.46 vs 0.20; P < .001), CURB-65 (median, 3 vs 2; P = .001), and PSI scores (median, 149 vs 116; P < .001) than those who did not. Of the 159 patients at risk for functional decline, 103 (64.8%) patients developed composite outcomes at 6 months—53 (51.5%) died and 63 (61.2%) experienced functional decline. Compared to patients who did not develop the composite outcome, those who did had higher FI (median, 0.30 vs 0.16; P < .001) and PSI (median, 127 vs 107; P < .001) scores. The prevalence of activities of daily living dependency and instrumental activities of daily living dependency, and hospital length of stay were greater among those who experienced death or the composite outcome (Table 1).
Table 1Characteristics of Patients with Pneumonia by Outcome Status at 6 Months
Thirty-one patients who had the maximum level of disability on admission were excluded from the composite outcome analysis.
Yes (n = 53; 27.9%)
No (n = 137; 72.1%)
P Value
Yes (n = 103; 64.8%)
No (n = 56; 35.2%)
P Value
Age, y, median (IQR)
79 (74, 85)
82 (75, 85)
79.0 (73, 84)
.13
80 (75, 86)
77.5 (72, 81)
.041
Female
70 (36.8)
19 (35.9)
51 (37.2)
.80
40 (38.8)
25 (44.6)
.48
BMI, mean (SD)
22.0 (4.5)
19.6 (3.9)
22.9 (4.4)
<.001
22.1 (4.4)
23.0 (4.5)
.34
Nursing home resident
35 (18.4)
20 (37.7)
15 (11.0)
<.001
15 (14.6)
3 (5.4)
.08
Frailty Index, median (IQR)
0.28 (0.14, 0.50)
0.46 (0.28, 0.54)
0.20 (0.12, 0.45)
<.001
0.30 (0.16, 0.46)
0.16 (0.09, 0.20)
<.001
CURB-65, median (IQR)
2 (1, 3)
3 (2, 4)
2 (1, 3)
.001
2 (1, 3)
2 (1, 3)
.005
PSI, median (IQR)
123 (97, 157)
149 (117, 175)
116 (93, 141)
<.001
127 (99, 160)
107 (78, 125)
.001
Cardiovascular disease
52 (27.4)
17 (32.1)
35 (25.6)
.37
31 (30.1)
16 (28.6)
.84
Diabetes
66 (34.7)
15 (28.3)
51 (37.2)
.25
34 (33.0)
18 (32.1)
.91
COPD
29 (15.3)
5 (9.4)
24 (17.5)
.17
17 (16.5)
10 (17.9)
.83
Stroke
50 (26.3)
13 (24.5)
37 (27.0)
.73
23 (22.3)
9 (16.1)
.35
ADL dependency
90 (47.4)
37 (69.8)
53 (38.7)
<.001
50 (48.6)
9 (16.1)
<.001
IADL dependency
114 (67.1)
42 (79.3)
72 (52.6)
.001
66 (64.1)
17 (30.4)
<.001
Hospital LOS ≥ 15 d
72 (37.9)
32 (60.4)
40 (29.2)
<.001
43 (41.8)
10 (17.9)
.002
ADL, activities of daily living; BMI, body mass index; COPD, chronic obstructive pulmonary disease; IADL, instrumental activities of daily living; IQR, interquartile range; LOS, length of stay; SD, standard deviation.
Unless otherwise noted, values are n (%).
∗ Thirty-one patients who had the maximum level of disability on admission were excluded from the composite outcome analysis.
Associations of FI, CURB-65, and PSI With Death or Functional Decline at 6 Months
FI, CURB-65, and PSI were associated with mortality and the composite outcome at 6 months (Table 2). After adjusting for age and sex, the odds ratio (OR) [95% confidence interval (CI)] for death associated with a 1 standard deviation increase in each risk score was 2.04 (1.44-2.89) for FI, 1.65 (1.18-2.31) for CURB-65, and 2.31 (1.59-3.34) for PSI. The corresponding OR (95% CI) for the composite outcome was 2.81 (1.70-4.63) for FI, 1.59 (1.10-2.30) for CURB-65, and 1.99 (1.31-3.02) for PSI.
Table 2Six-Month Mortality and Composite Outcome According to the Severity of Frailty and Pneumonia
When analyzed in clinically meaningful categories (Table 2), the risk of mortality increased from 12.2% for the nonfrail group (reference) to 45.2% for the severely frail group (adjusted OR, 5.83, 95% CI 2.14-15.89). The risk of the composite outcome increased from 46.9% for the nonfrail group to 83.9% for the severely frail group (adjusted OR 6.80, 95% CI 2.10-22.06). Increasing pneumonia severity by CURB-65 was positively associated with death [low risk, 18.0%, to high risk, 37.2% (adjusted OR 2.57, 95% CI 1.09-6.05)] and the composite outcome [low risk, 52.0%, to high risk, 77.4% (adjusted OR 2.80, 95% CI 1.22-6.42)]. PSI was also associated with death [class I-III, 13.5%, to class V, 41.0% (adjusted OR 4.32, 95% CI 1.45-12.9)] and the composite outcome [class I-III, 44.4%, to class V, 79.4% (adjusted OR 3.89, 95% CI 1.52-10.00)].
Prediction of Death or Composite Outcome at 6 Months
The C-statistics (95% CI) in predicting the 6-month mortality were 0.69 (0.61-0.77) for FI, 0.62 (0.53-0.71) for CURB-65, and 0.71 (0.62-0.79) for PSI (P = .019) (Figure 1A). C-statistics improved considerably when the FI was added to CURB-65 (0.62 vs 0.70; P = .015) (Figure 1B) but only modestly when the FI was added to the PSI (0.71 vs 0.73; P = .311) (Figure 1C).
Fig. 1Receiver operating characteristics curves showing the area under the curve for FI, CURB, and PSI for prediction of 6-month mortality and 6-month composite outcome. (A) The C-statistics (95% confidence intervals) for 6-month mortality were 0.69 (0.61-0.77) for the FI, 0.62 (0.53-0.71) for CURB-65, and 0.71 (0.62-0.79) for the PSI. For 6-month mortality prediction, (B) adding FI to CURB-65 yielded an increase in C-statistics from 0.62 to 0.70 compared with CURB-65 alone; and (C) adding FI to PSI mildly increased the C-statistic from 0.71 to 0.73. (D) The C-statistics for 6-month composite outcomes were 0.73 (0.65-0.81) for the FI, 0.64 (0.55-0.73) for CURB-65, and 0.69 (0.60-0.77) for the PSI. (E) Incorporating FI to CURB-65 increased the C-statistic from 0.64 to 0.74 compared with CURB-65 alone; and (F) adding FI to PSI modestly increased the C-statistic from 0.69 to 0.75 for prediction of the 6-month composite outcome.
The C-statistics for the 6-month composite outcome were 0.73 (0.65-0.81) for FI, 0.64 (0.55-0.73) for CURB-65, and 0.69 (0.60-0.77) for PSI (P = .096) (Figure 1D). The FI had a higher C-statistic than CURB-65 and the PSI, although these differences were not statistically significant. Adding the FI to each pneumonia severity measure significantly improved prediction: 0.64 for CURB-65 alone vs 0.74 for FI and CURB-65 (P = .003) (Figure 1E), and 0.69 for PSI alone vs 0.75 for FI and PSI (P = .044) (Figure 1F).
Discussion
Although each of FI, CURB-65, and PSI was associated with an increased risk of death or functional decline at 6 months in older adults hospitalized with pneumonia, we found that the FI combined with CURB-65 or PSI largely improved prediction compared with CURB-65 or PSI alone. To our knowledge, our study is the first to evaluate the additive value of FI to 2 widely used pneumonia severity indices in predicting death or persistent functional decline. Of the 3 indices, the simplest CURB-65 showed the worst predictive ability for mortality or the composite outcome; the difference in C-statistics between the FI and PSI was small (FI vs PSI: 0.69 vs 0.71 for mortality; 0.73 vs 0.69 for the composite outcome). Our findings that combining the FI and PSI was better than PSI alone suggest that the FI and PSI capture different risks in a complementary way.
Although several studies have evaluated the association of preadmission functional status or quality of life with the severity of pneumonia or mortality,
addition of Eastern Cooperative Oncology Group (ECOG) functional status score to the CURB-65 improved reclassification of high-risk patients. Sanz et al
also found that the combined PSI and Barthel Index provided a better prediction for mortality than the use of each index separately. In comparison, our study used an FI as a prognostic tool for death or functional decline. Although functional status is correlated with frailty, they are not identical; frailty is a measure of vulnerability and physiologic reserve for older adults,
which includes medical, functional, and nutritional domains. Therefore, the FI can help assess an older patient's recovery potential after pneumonia hospitalization. Through the early recognition of frailty during the hospitalization, clinicians can proactively intervene to maintain older patients' functional status and minimize dependence on caregivers.
Although PSI predicts both outcomes well, previous studies raised challenges in obtaining a broad range of diagnostic tests, including chest radiograph and arterial blood gas analysis, especially in the resource-poor settings.
Although a deficit-accumulation FI similarly require multiple (at least 30) components, it can be constructed with routinely collected variables from medical history and physical examination.
Our results suggest that the FI is likely to provide risk stratification as accurate as (when used alone) or better (when combined with CURB-65 or PSI) than the pneumonia severity indices.
Limitations of our study include potentially limited generalizability of our results from a single-center study of older Koreans. The absolute risk of death and functional decline may vary by health care systems and geographic areas. However, the relative improvement in C-statistics by adding frailty to pneumonia severity indices may not necessarily differ across health systems or geographical areas. Our study had insufficient power to detect a potentially clinically meaningful difference in C-statistics among FI and pneumonia severity indices for the composite outcome prediction. Lastly, because we only used a deficit-accumulation FI, the additional value of other frailty measures to pneumonia severity indices remains uncertain.
Conclusions and Implications
In conclusion, incorporating an FI in pneumonia risk stratification improve prediction of death or persistent functional impairment in hospitalized older patients with pneumonia. Early recognition of this vulnerable group at admission would help clinicians intervene early in the hospital course to prevent functional decline and develop appropriate post-acute rehabilitation care plan to restore function after discharge.
D.H.K. provides paid consultative services to Alosa Health, a nonprofit educational organization with no relationship to any drug or device manufacturers. He also receives research grants from the National Institute on Aging for unrelated projects. The other authors declare no conflicts of interest.
00818-5&id=gr1.jpg){kind=link}