Sputum Color: Potential Implications for Clinical Practice

Allen L Johnson MD, David F Hampson MD, and Neil B Hampson MD

BACKGROUND: Respiratory infections with sputum production are a major reason for physician
visits, diagnostic testing, and antibiotic prescription in the United States. We sought to determine
whether the simple characteristic of sputum color provides information that impacts resource
utilization such as laboratory testing and prescription of antibiotics. METHODS: Out-patient
sputum samples submitted to the microbiology laboratory for routine analysis were assigned to one
of 8 color categories (green, yellow-green, rust, yellow, red, cream, white, and clear), based on a key
made from paint chip color samples. Subsequent Gram stain and culture results were compared to
sputum color. RESULTS: Of 289 consecutive samples, 144 (50%) met standard Gram-stain criteria
for being acceptable lower-respiratory-tract specimens. In the acceptable Gram-stain group, 60
samples had a predominant organism on Gram stain, and the culture yielded a consistent result in
42 samples (15% of the 289 total specimens). Yield at each level of analysis differed greatly by color.
The yield from sputum colors green, yellow-green, yellow, and rust was much higher than the yield
from cream, white, or clear. CONCLUSIONS: If out-patient sputum is cream, white, or clear, the
yield from bacteriologic analysis is extremely low. This information can reduce laboratory process-
ing costs and help minimize unnecessary antibiotic prescription. Key word: respiratory, infection,
sputum color, antibiotics, laboratory, prescription, microbiology, Gram stain, predominant organism,
bacteria. [Respir Care 2008;53(4):450 – 454. © 2008 Daedalus Enterprises]

Introduction prescriptions by ambulatory physicians in the United States

are for colds, upper-respiratory-tract infections, and bron-
Cough is one of the most common symptoms for which chitis,4 minimizing the unnecessary prescription of antibi-
medical evaluation is sought, accounting for approximately otics for respiratory illness is an important aspect of con-
30 million physician visits annually in the United States.1 trolling the emergence of antibiotic-resistant bacteria.3
Most acute cough syndromes are due to colds, upper-res- The presence of sputum production may prompt pa-
piratory-tract infections, or acute bronchitis. Despite the tients to request antibiotic therapy. Many patients believe
fact that multiple randomized trials have demonstrated that that antibiotics are effective for acute respiratory infec-
antibiotics provide no benefit for the treatment of these tions if purulence is present.5 Physician beliefs may be
illnesses,2 a majority of these patients are prescribed an- similar. Evidence-based guidelines for the identification of
tibiotics.3 Because a large proportion of the total antibiotic clinically important infection in patients with uncompli-
cated acute cough illness have minimized the role of spu-
tum characteristics.6 Nonetheless, productive cough, espe-
Allen L Johnson MD is affiliated with The Polyclinic, Seattle, Washing- cially with purulent sputum, correlates with antibiotic
ton. David F Hampson MD is affiliated with St Mary Medical Center, prescription in acute respiratory infections.2,7
Walla Walla, Washington. Neil B Hampson MD is affiliated with the One strategy for limiting or targeting antibiotic pre-
Section of Pulmonary and Critical Care Medicine, Virginia Mason Med- scription is to send sputum for microbiologic analysis,
ical Center, Seattle, Washington.
either to demonstrate to the patient that a substantial in-
The authors report no conflicts of interest related to the content of this fection is not present or to identify an organism when
paper. antibiotic treatment is deemed necessary. Sending speci-
Correspondence: Neil B Hampson MD, Section of Pulmonary and Crit-
mens from all patients with productive cough, however,
ical Care Medicine, Virginia Mason Medical Center, H4-CHM, 1100 would be costly. To increase the efficiency of this ap-
Ninth Avenue, Seattle WA 98101. E-mail: neil.hampson@vmmc.org. proach we sought to determine whether the simple char-

450 RESPIRATORY CARE • APRIL 2008 VOL 53 NO 4

 SPUTUM COLOR: POTENTIAL IMPLICATIONS FOR CLINICAL PRACTICE

acteristic of sputum color can predict the results of sputum

analysis. If color correlates with sputum Gram stain and
culture results, we reasoned that sputum color could dis-
criminate which specimens to send for microbiologic test-
ing and allow physicians to counsel patients regarding the
likelihood of important respiratory infection.

Methods

Over a 10 month period, consecutive sputum samples

from out-patients were categorized by color when they
were submitted to the microbiology laboratory for Gram
stain and culture. Each sample was assigned one of 8 color
choices by the accepting microbiology technician, using a
standard color key. The key was a sheet of paper to which
paint chip color samples (Pittsburgh Paints, Pittsburgh,
Pennsylvania) were attached and numbered 1 through 8.
The colors and paint chip number were white (Pittsburgh
Paints number 2537), cream (2484), yellow (2251), yel-
low-green (3362), green (4452), rust (4252), red (7157), Fig. 1. Flow chart of the analysis of 289 out-patient sputum sam-
and clear. The corresponding color number was recorded ples.
by the technician in a log kept separate from the sample
during subsequent processing. in 70% (42/60). The yield of a maximally informative
Sputum Gram stains and culture were performed on result from all samples was therefore 15% (42/289).
each specimen via standard technique. A sample was be- The distribution of sputum colors is shown in Table 1.
lieved to represent lower-respiratory-tract secretions and Cream, yellow, and yellow-green were the most common,
defined as “acceptable” when the Gram stain had both accounting for 75% (217/289) of all samples. Colors in
⬍ 10 squamous epithelial cells and ⱖ 25 polymorphonu- which more than half of the samples demonstrated an ac-
clear white blood cells per high-power field. On Gram ceptable Gram stain were green, yellow-green, yellow, and
stain an organism was deemed “predominant” if it was in rust (see Table 1 and Fig. 2). The overall yield of a sub-
a quantity at least as great as the mixed respiratory flora. sequent predominant organism on Gram stain with cor-
The subsequent culture result was defined as correlating responding culture from this group of colors was 24%
with the Gram stain if the organism(s) identified on culture (35/135). This compares to a 5% yield (7/144) from the
was/were consistent with the morphology of the predom- color group that included clear, white, cream, and red.
inant organism(s) seen on Gram stain. Positive and negative predictive values for each color and
The most recent medical record note for each patient level of analysis are also listed in Table 1.
from whom sputum was submitted was reviewed by the Color did not correlate with specific bacteria. For ex-
investigators to determine which samples were obtained ample, among 30 acceptable-quality sputum samples that
from a patient taking an antibiotic of any kind. grew Haemophilus influenzae, 12 were yellow-green, 11 yel-
Descriptive statistics and sensitivity/specificity testing low, 4 cream, 1 white (3%), 1 green (3%), and 1 rust. Among
were used for data analysis. The study was approved by the 12 acceptable specimens that grew Moraxella catarrhalis,
the institutional review board of Virginia Mason Medical 4 were yellow-green, 3 yellow, 2 rust, 2 clear, and 1 cream.
Center and was deemed exempt with regard to the require- Traditional descriptions of sputum colors for some specific
ment for informed consent. organisms versus their prevalence in this study are listed in
Table 2.
Results With regard to current antibiotic therapy, 62%
(180/289) of the patients were not taking an antibiotic at
A total of 289 sputum samples were analyzed (Fig. 1). the time sputum was obtained, 20% (58/289) were tak-
Of these, 50% (144/289) were of acceptable quality and ing an antibiotic, and antibiotic status could not be de-
thought to be from the lower respiratory tract. Among the termined in 18% of cases (51/289). Among those with
acceptable samples, 42% (60/144) demonstrated a pre- acceptable-quality versus unacceptable-quality Gram
dominant organism on Gram stain. Subsequent culture iden- stains, antibiotic therapy was absent in 62% (90/144)
tified an organism consistent with that seen on Gram stain versus 63% (91/145), present in 19% (28/144) versus

RESPIRATORY CARE • APRIL 2008 VOL 53 NO 4 451

 SPUTUM COLOR: POTENTIAL IMPLICATIONS FOR CLINICAL PRACTICE

Table 1. Sputum Sample Data

Acceptable Gram Stain Plus

Acceptable Gram Stain Plus
Acceptable Gram stain Predominant Organism on Gram Stain
Predominant Organism on Gram Stain
Plus Corresponding Culture
Sputum Color
(n) Positive Negative Positive Negative Positive Negative
Predictive Predictive Predictive Predictive Predictive Predictive
n n n
Value Value Value Value Value Value
(%) (%) (%) (%) (%) (%)

Green (3) 3 of 3 100 51 3 of 3 100 80 2 of 3 67 86

Yellow-green (54) 37 of 54 69 54 20 of 54 37 83 18 of 54 33 90
Rust (13) 8 of 13 67 51 4 of 13 31 80 3 of 13 23 86
Yellow (75) 44 of 75 59 53 18 of 75 24 80 12 of 75 16 86
Red (2) 1 of 2 50 50 0 of 2 0 79 0 of 2 0 92
Cream (88) 39 of 88 44 48 9 of 88 10 75 4 of 88 5 81
White (25) 6 of 25 24 48 2 of 25 8 78 1 of 25 4 84
Clear (29) 6 of 29 21 47 4 of 29 14 78 2 of 29 7 85

Fig. 2. Categorization of 289 sputum samples into 8 color categories and 3 subcategories: (1) acceptable Gram stain, (2) predominant
organism on that Gram stain, and (3) culture result corresponded to the morphology of the organism seen on the Gram stain. * There were
zero red samples in the subcategories predominant organism on Gram stain and corresponding culture.

20% (29/145), and unknown in 18% (26/144) versus Few prior studies have specifically examined the role of
17% (25/145). Among the patients with an acceptable sputum color. In a study of in-patients at a veterans hos-
quality Gram stain, a predominant organism on stain, pital, yellow was the only sputum color that correlated
and a correlating culture, 64% (27/42) were not on an- with good or fair quality on Gram stain of the specimen.10
tibiotics, 17% (7/42) were taking an antibiotic, and an- In a study of patients who presented with exacerbations of
tibiotic status was unknown in 19% (8/42). moderate-to-severe chronic obstructive pulmonary disease,
sputum samples were categorized by color and then com-
Discussion pared to culture results.11 Gram-positive organisms were
isolated more frequently from white-gray sputum, and
The presence of sputum, especially when discolored, is Gram-negative organisms were most frequently from yel-
commonly interpreted by both patients and physicians to low sputum. The importance of these findings is difficult
represent the presence of bacterial infection and an indi- to interpret, however, because that study did not indicate
cation for antibiotic treatment. In reality, purulence pri- whether the cultured organisms correlated with the Gram
marily occurs when inflammatory cells or sloughed mu- stain findings. Since sputum cultures in every color cate-
cosal epithelial cells are present, and it can result from gory yielded an average of more than one organism per
either viral or bacterial infection.8,9 As this study demon- specimen, it is probable that more than just the predomi-
strates, sputum color is most useful to maximize the like- nant organism(s) on Gram stain were reported. The im-
lihood that a sputum specimen will yield useful informa- portance of correlating culture results with Gram-stain find-
tion on microbiologic analysis. ings has long been emphasized.12

452 RESPIRATORY CARE • APRIL 2008 VOL 53 NO 4

 SPUTUM COLOR: POTENTIAL IMPLICATIONS FOR CLINICAL PRACTICE

Table 2. Sputum Colors of Specific Organisms Grown From not to submit those samples for analysis unless there is a
Acceptable-Quality Specimens, as Compared to Traditional high suspicion of bacterial infection. Alternatively, since
Descriptions of Sputum Produced by Those Organisms
the clinician may be responding to the patient’s report and
Traditional Sputum Colors may not actually see the sputum color, laboratories could
Organism (n) Description of Found in the limit processing of such samples in the same way they do
Sputum Color Present Study with unacceptable Gram stains. Further, if the sputum is
Klebsiella pneumoniae (2) Red “currant jelly” Red 0 cream, white, or clear there should be other evidence to
Yellow-green 2 support the clinical likelihood of substantial infection be-
Streptococcus pneumoniae (6) Rusty Rust 0 fore prescribing antibiotics.
Yellow-green 4 Third, we saw no relationship between the yield of spu-
Yellow 1 tum analysis and current antibiotic use. Patients who were
Cream 1 taking an antibiotic were as likely as those not on antibi-
Pseudomonas aeruginosa (9) Green Green 1 otics to produce sputum of acceptable Gram-stain quality,
Yellow-green 4 to demonstrate a predominant organism on stain, and to
Yellow 3 yield a corresponding culture. Though this suggests that a
Rust 1 clinician should not be dissuaded from analyzing sputum
from a patient taking an antibiotic if suspicion of infection
is sufficiently high, it also raises the question of patient
Sputum color has also been used to determine the need compliance with out-patient antibiotic therapy.
for antibiotic therapy in exacerbations of chronic obstruc-
tive pulmonary disease. In one study, samples were di- Limitations
vided into “mucoid” (opaque or milky color) versus “pu-
rulent” (various shades of green).13 Sputum cultures were Limitations of the present study relate largely to the lack
more likely to grow bacteria when purulent and were of information collected about the individual patients from
thought to indicate the need for antibiotic treatment. whom the samples were obtained. Disease processes that
This study presents several important findings. First, were responsible for their sputum production are unknown.
there was a clear relationship between sputum color and It is possible, for example, that sputum characteristics differ
acceptability of the Gram-stain quality (see Fig. 2). The between those suffering from conditions such as bronchitis,
Gram stain is recommended as a means of grading sputum pneumonia, asthma, cystic fibrosis, and bronchiectasis. In
adequacy on the basis of presence of epithelial cells (which addition, no attempt was made to correlate the specific anti-
represent oropharyngeal contamination) and polymorpho- biotics being taken with the culture results. If the cultures
nuclear white blood cells (which represent lower-respira- yielded organisms that should have been sensitive to the an-
tory-tract inflammation). Several guidelines have been pro- tibiotic the patient was taking, the clinician would consider
posed for evaluating Gram-stain quality, with different the possibility of medication noncompliance. Finally, in con-
combinations and cutoffs, but none is considered clearly ditions with chronic lower-respiratory-tract colonization and
superior.14 In the present study more than half of the spec- sputum production, such as cystic fibrosis, the clinical utility
imens with yellow, green, or rust colors demonstrated ac- of information gained from a high-quality sputum sample is
ceptable Gram stains, according to the criteria chosen (see sometimes difficult to interpret.
Fig. 2). Conversely, white or clear specimens were accept-
able less than 25% of the time. Conclusions
Second, the association between color and microbio-
logic findings was even more striking. One presumably
Objective assessment of out-patient sputum color has
sends sputum to the laboratory to determine the presence/
the potential to (1) yield substantial cost savings by elim-
absence and identities of organisms. A predominant or-
inating the processing of low-yield specimens and (2) re-
ganism was seen on an acceptable Gram stain in only 10%
duce unnecessary antibiotic prescription when there is not
of the cream, white, or clear specimens, and a correspond-
more clinical evidence of bacterial infection than cream,
ing bacterial species was grown on subsequent culture in
white, or clear sputum. It will be necessary to test these
only a fraction of those cases. Many microbiology labo-
concepts prospectively in a defined patient population to
ratories do not proceed with sputum culture if the Gram
determine the clinical utility of sputum color grading.
stain is of unacceptable quality. Since cream, white, or
clear sputum specimens accounted for half of the samples
in this study, substantial cost savings could be achieved by REFERENCES
educating physicians about the rare utility of cream, white, 1. Irwin RS, Baumann MH, Bolser DC, Boulet LP, Braman SS,
or clear sputum specimens, and by encouraging physicians Brightling CE, et al. Diagnosis and management of cough, executive

RESPIRATORY CARE • APRIL 2008 VOL 53 NO 4 453

 SPUTUM COLOR: POTENTIAL IMPLICATIONS FOR CLINICAL PRACTICE

summary: ACCP evidence-based clinical practice guidelines. Chest 8. Robertson AJ. Green sputum. Lancet 1952;1(1):12–15.
2006;129(1 Suppl):1S–23S. 9. Heald A, Auckenthaler R, Borst F, Delaspre O, Germann D, Matter
2. Hall KK, Philbrick J, Nadarni M. Evaluation and treatment of acute L, et al. Acute bacterial nasopharyngitis: a clinical entity? J Gen
bronchitis at an academic teaching clinic. Am J Med Sci 2003; Intern Med 1993;8(12):667–673.
325(1):7–9. 10. Flournoy DJ, Davidson LJ. Sputum quality: can you tell by looking?
3. Gonzales R, Steiner JF, Sande M. Antibiotic prescribing for adults Am J Infect Control 1993;21(2):64–69.
with colds, upper respiratory tract infections, and bronchitis by am- 11. Allegra L, Blasi F, Diano PL, Consentini R, Tarsia P, Confalonieri
bulatory care physicians. JAMA 1997;278(100):901–904. M, et al. Sputum color as a marker of acute bacterial exacerbations
4. McCaig LF, Hughes JM. Trends in antimicrobial drug prescribing of chronic obstructive pulmonary disease. Respir Med 2005;99(6):
among office-based physicians in the United States. JAMA 1995; 742–747.
273(3):214–219. 12. Heineman HS, Chawla JK, Lofton WM. Misinformation from spu-
5. Mainous AG, Zoorob RJ, Oler MJ, Haynes DM. Patient knowledge tum cultures without microscopic examination. J Clin Microbiol 1977;
of upper respiratory infections: implications for antibiotic expecta- 6(5):518–527.
tions and unnecessary utilization. J Fam Pract 1997;45(1):75–83. 13. Stockley RA, O’Brien C, Pye A, Hill SL. Relationship of sputum
6. Metlay JP, Kapoor WN, Fine MJ. Does this patient have community- color to nature and outpatient management of acute exacerbations of
acquired pneumonia? Diagnosing pneumonia by history and physical COPD. Chest 2000;117(6):1638–1645.
examination. JAMA 1997;278(17):1440–1445. 14. Wong LK, Barry AL, Horgan SM. Comparison of six different cri-
7. Mainous AG III, Hueston WJ, Eberlein C. Colour of respiratory teria for judging the acceptability of sputum specimens. J Clin Mi-
discharge and antibiotic use. Lancet 1997;350(9084):1077. crobiol 1982;16(4):627–631.

454 RESPIRATORY CARE • APRIL 2008 VOL 53 NO 4