Presentation of infection in older patients--a prospective study.

Abstract Background. Traditional wisdom suggests that infections in older patients have atypical presentation, including blunted febrile response. Data are scarce. Design. We analyzed data from a prospectively collected database on presentation of infection in 4,308 patients, and compared the presentation of older patients (≥ 75 years) versus adults (< 75 years). Settings. Single tertiary medical center. Participants. Patients admitted with suspected bacterial infection during 2002–2004 and 2010–2011. Measurements. We evaluated clinical presentation on day of admission, including vital signs and laboratory parameters. Results. No difference in fever values as a presenting sign of infection was found between older patients and adults (median fever 38.3°C, interquartile range [IQR] 37.4–39.0°C; and 38.4°C, IQR 37.3–39.0°C, respectively, P = 0.08). Median leukocyte count was significantly higher in older patients (median 11.60, IQR 8.30–15.72 in older patients; 10.84, 7.50–15.00 in adults, P < 0.001). Presentation with septic shock, acute renal failure, and reduced consciousness was significantly more common in older patients. These findings were also consistent in the subgroups of bacteremic patients and patients with microbiologically documented infection. Conclusion. Elevated fever and leukocytosis were found to be at least equally common in older patients compared to younger adults as part of the presentation of infection.

Multiple explanations for the increased rate of infections among older patients have been suggested, including co-morbid illnesses, exposure to instrumentation and procedures, institutionalization, immunosenescence, malnutrition, and poor performance status (5). Higher morbidity and mortality due to infections in older patients compared to younger patients have been attributed in part to the diagnostic challenge in this population (3,7).
Th e traditional wisdom is that presentation of infection in older patients is diff erent than in younger patients and that older patients tend to have fewer symptoms (3), e.g. fever may be absent or blunted in 20% -30% of older patients with serious infection (8). In addition, non-specifi c manifestations such as falls, altered mental status, anorexia, urinary incontinence, or generalized weakness are considered common and oft en only presenting signs of infection (2,3,5).
Data of these observations are scarce. Clinical studies comparing infection presentation in older patients versus younger patients are few and most of them small and retrospective (9 -16). Most of them report lower rates of fever and leukocytosis as presenting signs in older patients, and higher rates of altered mental status and renal failure. Th e question whether bacterial infections may have atypical presentations in older patients is of utmost importance: it dictates the threshold for starting antibiotics in old people with atypical presentations, and drives antibiotic consumption and resistance.
We aimed to assess diff erences in infection presentation between patients Ն 75 years old versus Ͻ 75 years old, hospitalized with suspected bacterial infection or in whom infections should have been suspected and included in a prospectively collected database.

Patients
We analyzed a database of patients from Rabin Medical Center, Beilinson Campus in Israel, collected from six departments of internal medicine (240 beds). Patients were enrolled as part of a three-phase study (two observational and one interventional) designed to evaluate the eff ectiveness of TREAT, a decision support system for antibiotic treatment of common bacterial infections in medical inpatients. Data were collected between June and December 2002, May and November 2004, and between May 2010and April 2011. Th e local research ethics committee approved the study protocol.

Inclusion and exclusion criteria
We included all patients fulfi lling systemic infl ammatory response syndrome diagnostic criteria (17); patients with a focus of infection; patients with shock compatible with septic shock; patients with febrile neutropenia; patients prescribed antibiotics (not for prophylaxis); and patients from whom blood cultures were drawn, regardless of whether the suspected infection was acquired in the community or in the hospital.
We excluded human immunodefi ciency virus-positive patients; solid-organ or bone-marrow transplant recipients; children aged less than 18 years; those with suspected travel infections or tuberculosis; and pregnant women. Patients fulfi lling inclusion criteria were prospectively identifi ed by daily chart review. Data collected at infection presentation included: demographic details; background conditions; predisposing conditions for infection and devices; general and focal signs and symptoms; and routine laboratory data. Data concerning use of beta-blockers are missing.

Defi nitions
We compare adults (age 18 -74 years) to older patients (age 75 years or older). Fever was defi ned as any measurement of Ն 38 ° C on the day of inclusion, and hypothermia as any measurement of Ͻ 36 ° C on the day of inclusion. Leukocytosis was defi ned as white blood cells (WBC) Ն 12,000 cells/ μ L and leukopenia as Յ 4,000 cells/ μ L.
Septic shock was defi ned as systolic blood pressure Ͻ 90 mmHg. Acute renal failure was defi nes as a rise in serum creatinine of Ͼ 0.5 mg/dL if the baseline creatinine was Ͻ 2 mg/dL; and Ͼ 1 mg/dL if the baseline was Ͼ 2 mg/dL.
Microbiologically documented infection (MDI) was defi ned as an infection with a pathogen defi ned by a positive culture or a determination of an antigen or positive serology or PCR deemed as signifi cant (and not contamination or colonization) by consensus of two clinicians.

Statistical analysis
Proportions were compared using the chi-square test, and continuous variables were compared using the Mann -Whitney U test (as most variables did not have a normal distribution). Analyses were performed using the Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL, USA).

Results
Overall 4,308 patients were included: 2,375 adults and 1,933 older patients.

Fever
Th e median fever in older patients (2,269 patients) was 38.3 ° C (interquartile range [IQR] 37.4 -39.0) and in adults (1,906 patients) 38.4 ° C (IQR 37.3 -39.0) ( P ϭ 0.080). A signifi cantly higher percentage of older patients (1.3%) had hypothermia (fever Ͻ 36 ° C) compared to adults (0.4%), but the number of patients presenting with hypothermia was low (24/1,906 older patients and 9/2,269 adults, P ϭ 0.002). Th ere was no signifi cant difference in the percentage of patients with normal temperature on presentation ( Ͻ 38 ° C and Ն 36 ° C) between older patients (31.7% of patients) and adults (32.2% of patients) ( P ϭ 0.716). Th ere was also no signifi cant diff erence in the percentage of patients with high temperature ( Ն 38 ° C) on presentation between older patients and adults (67.1% of older patients and 67.7% of adults, P ϭ 0.818).
Herat rate and blood pressure (Table I) Th ere was a signifi cant diff erence in the distribution of heart rate and blood pressure between age groups, with lower heart rate (HR) and diastolic blood pressure (DBP) and higher systolic blood pressure (SBP) in the older patients group. Data concerning use of beta-blockers are missing.

Other parameters assessed on presentation (Table I)
Normal level of consciousness was signifi cantly less common in older patients (54.6% of older patients compared with 85.2% of adults) ( P Ͻ 0.001). Parameters signifi cantly more common in older patients were presentation with dyspnea, septic shock, and acute renal failure. On the other hand, chills ( P Ͻ 0.001) and vomiting ( P ϭ 0.001) were signifi cantly more common in adults.
Analyzing separately patients with abnormal consciousness at presentation demonstrated elevated fever, leukocytosis, or septic shock in 89.6% of older patients and 86.5% of adults. Among patients with abnormal consciousness, but with normal fever and leukocytes, and no septic shock only 19.1% of older patients and 16.7% of adults had microbiologically documented infection. (Table II) Comparisons of laboratory values are presented in Table II. In summary, median values of leukocytes, creatinine, urea, glucose, and sodium were all signifi cantly higher in older patients, while median platelet counts were lower.
No diff erence was found in C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) between older patients and adults. Lactate levels were signifi cantly higher in older patients and albumin signifi cantly lower. Th ese data were available, however, for less than 50% of patients.

Subgroup of bacteremic patients (469 patients, Table I)
In the subgroup of bacteremic patients (236 elderly and 233 adults), no signifi cant diff erence was demonstrated between older patients and adults in the distribution of fever, SBP, and DBP. Heart rate was signifi cantly lower in the older patients ( P Ͻ 0.001). Among bacteremic older patients, 11.3% presented with septic shock compared with 4.6% of bacteremic adults ( P ϭ 0.010).
Among older bacteremic patients, 38.7% presented with acute renal failure, compared with 18.6% of bacteremic adults ( P Ͻ 0.001).    and older patients (age Ն 75 years). No signifi cant diff erence in the distribution of fever was demonstrated between groups among all patients, bacteremic patients, and MDI patients. No signifi cant diff erence was found in the percentage of patients presenting with elevated fever. Hypothermia at presentation was signifi cantly more common in older patients compared with adults, but overall it was a rare presentation (1.3% of older patients).
Heart rate was consistently lower in older patients among all groups analyzed, but data concerning use of beta-blockers, which may explain this observation, are missing. Systolic blood pressure was higher and diastolic blood pressure was lower in older patients in the entire cohort, but not in the bacteremia or MDI subgroups.
Data from all patients and from the analyzed subgroups showed higher rates of presentation with abnormal consciousness, dyspnea, septic shock, and acute renal failure among older patients. Chills and vomiting were more common in adults.
Signifi cantly higher median values of leukocytes, creatinine, and urea were measured in older patients in all the groups assessed. Median glucose and sodium values were higher in older patients in the entire cohort and among bacteremic elderly, and platelet levels were signifi cantly lower.
Several earlier clinical studies reported lower rates of fever and leukocytosis as presenting signs in older patients (Supplementary Table I to be found online at http://informahealthcare.com/doi/ abs/10.3109/07853890.2015.1019915), and higher rates of altered mental status and renal failure (10,11,13 -15). In a recent retrospective study (18) no signifi cant diff erence was demonstrated in rates of fever Ն 38.5 ° C between bacteremic adults and older patients. Chassagne et al. in their prospective study (9) did not fi nd a signifi cant diff erence in the frequency of fever or hypothermia between adults and older patients ( Ͼ 65 years). Reports on hypothermia in previous studies are also variable, including studies reporting no diff erence between older patients and adults (9,11), higher rates in older patients (10,12), and higher rates in adults (19).
Several studies have found lower baseline core temperature in older people (19 -22). Others suggested blunted hypothalamic pathways for fever elevation in older people (23) and lower amplitude of circadian-rhythm temperature fl uctuations (24). In a retrospective study a considerable number of patients presented without elevated absolute fever, but had a signifi cant change from baseline (19). Our data suggest no clinical diff erence in fever as the presenting sign of infection in older patients versus adults.
Concerning leukocytosis or leukopenia on presentation, some previous studies report no diff erence between adults and older patients (9,18), while others report higher rates of leukocytosis (11) or lower rates of leukopenia (10) in older patients.
Several other methods have been assessed for diagnosing infection in the elderly, including measurements of CRP, procalcitonin (25,26), and volume conductivity scatter parameters of leukocytes (27). All the above parameters were found to have variable sensitivity. Median CRP levels in our study were similar between older patients and adults. It should be emphasized that, in older patients with abnormal consciousness but no traditional signs of infection, rates of true MDI were low.
Our study limitations are missing data on several parameters of infection, including CRP, lactate, and albumin, and on use of beta-blockers.
Our study was probably biased towards showing less fever or other sepsis signs in older patients, because the overall cohort was defi ned by physician ' s clinical suspicion.
Diff erences in laboratory values among bacteremic patients are presented in Table II. In general, median values of leukocytes, creatinine, urea, glucose, and sodium were signifi cantly higher in older patients, while median platelet count was lower.
A signifi cantly higher percentage of bacteremic older patients had leukocytosis on presentation (123/

Subgroup of patients with MDI (1,005 patients, Table I)
In the subgroup of patients with MDI (494 elderly and 511 adults), no signifi cant diff erence was demonstrated between older patients and adults in the distribution of fever, SBP, and DBP. Heart rate was signifi cantly lower in older patients with MDI ( P Ͻ 0.001).
Among patients with MDI, normal consciousness was a significantly more common presentation in adults (78.5%) compared with older patients (45.6%) ( P Ͻ 0.001). Chills were also more common in adults (33.3%) compared with older patients (25.4%) ( P ϭ 0.014). Septic shock at presentation was signifi cantly more common in older patients with MDI (8.4%) compared with adults with MDI (3.9%) ( P ϭ 0.004), and so were dyspnea ( P ϭ 0.016) and acute renal failure ( P Ͻ 0.001).
Diff erences in laboratory values among patients with MDI are presented in Table II. In summary, median values of leukocytes, creatinine, and urea were signifi cantly higher in older patients, while no signifi cant diff erence was demonstrated for median values of platelets, glucose, and sodium.

Categories of fi nal diagnosis -entire cohort
Th e main categories of fi nal diagnosis are described in Table III. Generally, lower respiratory and urinary tract infections were signifi cantly more common among older patients ( P Ͻ 0.001 for each diagnosis). Among adults, neutropenic fever ( P ϭ 0.001), abdominal infections ( P Ͻ 0.001), and non-bacterial infections ( P Ͻ 0.001) were signifi cantly more common.
No diff erence in rates of non-infectious causes was demonstrated between adults and older patients ( P ϭ 0.505).

Discussion
We used data from a prospectively collected database to compare presentation of infection between adult patients (age Ͻ 75 years) temperature in older people, suggested above). However, previous studies described no diff erence in leukocytes and platelet values between older patients and younger patients (28,29). Some studies described elevated urea levels with advanced age, but most studies did not fi nd a relation between age and creatinine levels (29). Serum creatinine may remain stable with age despite decreases in the glomerular fi ltration rate (30).
In conclusion, in our analysis of a large prospective database, elevated fever and leukocytosis were found to be as common in older patients as in adults as part of the presentation of infection. It should be emphasized that, in older patients with abnormal consciousness but no traditional signs of infection, rates of true MDI were very low. Our fi ndings of typical presentation in older patients, including elevated fever and leukocytosis, may help in avoiding unnecessary use of antibiotics in older patients presenting without these signs.