Sei sulla pagina 1di 4

Reliability of Criteria for Predicting

Persistent or Recurrent Sepsis


H. Harlan Stone, MD; Audra A. Bourneuf, RN; Lenora D. Stinson, RN

\s=b\ We reviewed the charts of 2,567 patients from 11 prospec- the termination of antibiotic therapy,1 prompted a more
tive clinical trials of antibiotic therapy for surgical infection to detailed examination of patient data that were obtained
identify reliable predictors of sepsis eradication. Particular from several controlled antibiotic trials.
attention was paid to temperature, blood cell counts, renal and
PATIENTS AND METHODS
hepatic function tests, arterial gases, and clotting factors,
both at the termination of parenteral antibiotic administration The charts of 2,567 patients with various types of surgical
as well as at patient discharge from the hospital. On the infection were reviewed. Each patient had been a participant in one
discontinuation of antibiotic therapy, sepsis recurred in 19% of of 11 prospective clinical trials of antibiotic efficacy. The following
the patients who had a normal rectal temperature, in 3% of the antibiotics had been administered: a cephalosporin alone, 1,100
patients if the rectal temperature and WBC count were normal, patients; cefamandole plus erythromycin, 60 patients; an amino-
but in no patient when both the temperature and WBC count glycoside alone, 743 patients; and an aminoglycoside plus either
were normal and the differential blood smear contained less
clindamycin or metronidazole, 664 patients.
than 73% granulocytes and less than 3% immature forms. Some form of intra-abdominal infection had been treated in 1,419
Rates for recurrent sepsis, once antibiotic therapy was discon- patients (Table 1). Of these, 1,389 patients underwent either formal
tinued for more than 48 hours, were 8%, 2%, and 0%, respec- laparotomy or the extraperitoneal drainage of an intra-abdominal
tively, for the same criteria at hospital discharge. abscess. An additional 889 patients had some type of soft-tissue
(Arch Surg 1985;120:17-20) sepsis, of which 202 could be classified as having synergistic
infectious gangrene. A miscellaneous infection with an extra-
often arises as to when the administration peritoneal location was the indication for antibiotic therapy in 222
The parenteral
question
of antibiotics can be safely discontinued in patients, while in 37 patients the infection could never be proved by
positive culture or operative exposure.
patients who are being treated for peritonitis or some other A significant associated disease was present in 1,787 patients.
type of surgical infection. Should a search first be made for The diseases included diabetes in 399 patients, obesity in 102
some residual focus of infection? Should therapy be changed
to another antibiotic regimen? Similarly, at the time of
patients, malnutrition in 97 patients, kidney disease in 84 patients,
cancer in 71 patients, liver disease in 38 patients, and some process
anticipated discharge of the patient from the hospital, there involving the reticuloendothelial system in 23 patients.
is still the uncertainty as to whether the infectious process Excluded from more detailed study were 381 patients, for the
has been eradicated completely. Can the patient be dis¬ following reasons: extraperitoneal or non-soft-tissue sepsis, 222
charged with confidence that sepsis will not recur? patients; unproved infection, 37 patients; death while receiving
To date, fever has been the prime guide, yet this criterion parenteral antibiotic, 43 patients; and obvious recurrent sepsis
has not been uniformly reliable. Recent documentation of before antibiotic therapy had been discontinued, 79 patients.
the predictive value of the peripheral WBC count, taken at Accordingly, the charts of the remaining 2,186 patients were
selected for more careful review to delineate the recurrence of
infection and potential prediction by various signs that sepsis
Accepted for publication Aug 16, 1984. would probably reappear. Rectal temperature, total and differen¬
From the Department of Surgery, University of Maryland School of tial WBC counts, renal (serum creatinine and serum urea nitrogen
Medicine, Baltimore. levels) and liver (serum bilirubin and SGOT levels) function test
Read before the Fourth Annual Meeting of the Surgical Infection Society,
Montreal, April 30, 1984. results, blood gases (arterial Po2 and pH), and specific clotting
Reprints not available. factors (prothrombin time and platelet count) were the areas given

Downloaded from www.archsurg.com at McMaster University, Health Sciences Library, on December 3, 2009
special attention. The gross appearance of an involved wound and Table 1.—Infected Patients Studied
the volume of purulence discharged each day from that wound were
also used for assessment. Infection No. of Patients
RESULTS Peritonitis
Perforative/suppurative appendicitis 274
Prediction at Discontinuation Perforated gastroduodenal ulcer 244
of Antibiotic Therapy Cholecystitis/cholangitis 153
Colonie diverticulitis 101
Rectal temperatures taken at the time of termination of Bowel gangrene 126
Bowel perforation 47
antibiotic therapy proved to be unreliable sole predictors Pelvic inflammatory disease 42
for excluding the possibility of recurrent sepsis (Table 2). Miscellaneous generalized peritonitis 79
Subtotal 1,066
Infection at the original locus reappeared in 19% of the Intra-abdominal abscess
patients despite a normal temperature (37 ±0.5 °C, Peritoneal 326
mean ± SD). If fever was present, however, 65% of the Pancreatic 27
Subtotal 353
patients had recurrent sepsis. Soft-tissue sepsis
The WBC count was considerably more accurate, but still Perirectal abscess 291
was not an absolute guide (Table 3). Counts above the Polymicrobial necrotizing infection 202
Wound soft-tissue infection 396
normal range (8±2.5xl03/cumm, mean ± SD) did predict a Subtotal 889
significantly greater likelihood of sepsis exacerbation, ie, Miscellaneous infections
No infection or infection never proved
222
37
67%. Nevertheless, 6% of the patients experienced a recur¬ Total 2,567
rence of the original or a specifically related infection
despite a normal WBC count.
The percent of granulocytes in the differential count Table 2.—Likelihood of Recurrent
provided no greater accuracy (Table 4). If granulocytes Sepsis and Rectal Temperature*
constituted less than 60% of the total WBC count, then
recurrence of sepsis developed in 5% of the patients. When Rectal Sepsis
Temperature, °C No. of Recurred,
granulocytes constituted 61% to 80% of the count, the risk of (Mean ± SD) Patients No. of Patients Incidence, %
recurrent infection was 23%. However, if immature gra¬ 37±0.5 2,132 408 19.1
nulocyte forms (eg, band cells, metamyelocytes, and pro- 38 ±0.5 51 32 62.7
myelocytes) alone were considered, the differential count 39±0.5 3 3 100.0
became an exceedingly reliable tool (Table 5). In the ab¬
Total 2,186 443 20.3
sence of any immature granulocytic stages, sepsis never
"The predictions were made at the time of the discontinuation of parenteral
recurred. If there were three or less immature granulocytes antibiotic therapy.
per 100 WBCs that were counted, the likelihood of recurrent
infection was only 0.2%. In contrast, if four or more
immature forms were seen, recrudescence of sepsis fol¬ Table 3.—Likelihood of Recurrent Sepsis and WBC Count*
lowed in 54% of the patients. WBC Count, Sepsis
Use of the preceding signs in an additive fashion demon¬ x103cumm No. of Recurred,
strated the reappearance of infection in 19% of the patients (Mean ± SD) Patients No. of Patients Incidence, %
who had a normal rectal temperature, in 3% of the patients <5.5 7 2 28.6
who had a normal rectal temperature as well as a normal 8 ±2.5 1,666 97 5.8
WBC count, and in no patient with both of the preceding 13±2.5 476 309 64.9
values in the normal range plus less than 3% of the WBCs on >15.5 37 35 94.6
a differential count being immature granulocytes. Total 2,186 443 20.3
The gross appearance of the wound involved in the *The predictions were made at the time of the discontinuation of parenteral
infectious process proved to be a highly subjective sign. In antibiotic therapy.
addition, terms used to describe the exposed tissues were
seldom uniform and led to considerable confusion. The serum creatinine, serum urea nitrogen, serum bilirubin,
recorded or estimated volume of pus that drained during a and SGOT levels, arterial blood Poz and pH, or prothrombin
given day was a more concrete sign. At the time of the time. The platelet count served more to indicate a per¬
discontinuation of the administration of parenteral antibiot¬ sistence of sepsis, not a later recurrence, as other signs of
ics, infection recurred in only 3% of the patients if the infection immediately returned on the discontinuation of
volume of pus that discharged from the wound during the the administration of antibiotics when the platelet count
preceding 24 hours was less than 10 mL. Volumes between was low. More specifically, platelet counts lower than
10 and 100 mL during the same interval were attended by an 60 x lOVcu mm reliably predicted a rapid recrudescence of
11% recurrence rate. If more than 100 mL of pus drained infection in 91% of 163 patients whose infections were so
during the previous day, sepsis reappeared in 82% of the evaluated. In contrast, the recurrence rate was only 16% if
patients. the platelet count was greater than 60 103/cu mm at the
No consistent statistically significant trends could be time when administration of parenteral antibiotics was
found for predicting exacerbating infection with respect to stopped.

Downloaded from www.archsurg.com at McMaster University, Health Sciences Library, on December 3, 2009
Table 4.—Likelihood of Recurrent Sepsis and Table 7.—Likelihood of Recurrent
Differential WBC Count by Total Granulocytes* Sepsis and WBC Count*
Total WBC Count, Sepsis
Granulocytes, Sepsis xKP/cumm No. of Recurred,
% of Total No. of Recurred, (Mean SD) ± Patients No. of Patients Incidence, %
Count Patients No. of Patients Incidence, % <5.5 162 0 0.0
<50 43 2 4.7
8±2.5 693 14 2.0
51-60 629 31 4.9
13±2.5 252 70 27.8
61-70 804 116 14.4
>15.5 16 15 93.8
71-80 633 218 34.4
Total 1,123 99 8.8
>81 77 76 98.7
'The predictions were made at the time of patient discharge from the
Total 2,186 443 20.3 hospital.
*The predictions were made at the time of discontinuation of parenteral
antibiotic therapy.
Table 8.—Likelihood of Recurrent Sepsis and
Differential WBC Count by Total Granulocytes*
Table 5.—Likelihood of Recurrent Sepsis and
Differential WBC Count by Immature Granulocytes* Total
Granulocytes, Sepsis
Immature % of Total No. of Recurred,
Granulocytes, Sepsis Count Patients No. of Patients Incidence, %
% of Total No. of Recurred, <50 37 0 0.0
Count Patients No. of Patients Incidence, % 51-60 233 2 0.9
126 0 0.0
61-70 391 38 9.7
1-3 1,248 3 0.2
71-80 456 54 11.6
4-6 414 89 21.5
>80 6 5 83.3
381 334 87.6
Total 1,123 99
>10 17 17 100.0
Total
*The predictions were made at the time of patient discharge from the
2,186 443 20.3 hospital.
*The predictions were made at the time of discontinuation of parenteral
antibiotic therapy.
Table 9.—Likelihood of Recurrent Sepsis and
Table 6.—Likelihood of Recurrent
Differential WBC Count by Immature Granulocytes*
Sepsis
and Rectal Temperature* Immature
Granulocytes, Sepsis
Rectal Sepsis % of Total No. of Recurred,
Temperature, No. of Recurred, Count Patients No. of Patients Incidence, %
C (Mean ± SD) Patients No. of Patients Incidence, % 265 0 0.0
37.0 ±0.5 1,034 83 8.0 1-3 522 6 1.1
38.0 ±0.5 89 16 18.0 4-6 291 55 18.9
Total 1,123 99 8.8 7-9 44 37 84.1
*The predictions were made at the time of patient discharge from the >10 1 1 100.0
hospital. Total 1,123 99 8.8

Discharge From Hospital


Prediction at *The predictions were made at the time of patient discharge from the
hospital.
Of the 2,186 patients with the potential for evaluation of
recurrent sepsis after hospital discharge, 1,063 were ex¬ whom leukocytosis persisted the recurrence rate was 32%
cluded. Sepsis had reappeared after discontinuation of the (Table 7).
administration of antibiotics, but before hospital discharge, When granulocytes constituted 60% or less of the differ¬
in 344 patients. Another 21 patients had died during that ential WBC count, infection at the primary or a related site
same interval. Finally, 698 patients failed to keep their recurred in 1% of the patients (Table 8). With a granulocyte
follow-up appointments. Thus, only 1,123 patients were count of 61% to 80%, the risk for recurrence was 11%. A
available for the assessment of the predictability of sepsis granulocyte count in excess of 80% almost guaranteed
eradication at the time of discharge from the hospital. No persisting infection, ie, an 83% recurrence rate. The per¬
patient left the hospital before 48 hours had elapsed since cent of immature granulocytes was equally useful for indi¬
discontinuation of parenteral antibiotic therapy. cating the likelihood of infection reappearing. When imma¬
If the patient's temperatures were normal at hospital ture granulocytes represented less than 3% of the total
discharge, the primary infection recurred in 8%, in contrast differential count, recurrence of infection was noted in only
with 18% if the patients were febrile (Table 6). The WBC 1% of the patients (Table 9). If immature forms constituted
count was more reliable, as sepsis recurred in only 2% of the 4% to 6% of the count, then recurrent sepsis developed in
patients if the total count was normal, while in patients in 19% of the patients. However, when the immature gra-

Downloaded from www.archsurg.com at McMaster University, Health Sciences Library, on December 3, 2009
nulocyte count was greater than 7%, the incidence of However, the percent of granulocytes in that WBC count
recurrent infection escalated to 84%. and, more reliably, the percent of immature granulocytes
At the time of hospital discharge when the patients were are exceedingly valid indicators of a persisting sepsis that
off of antibiotic therapy for a minimum of 48 hours, if the will soon become clinically overt once parenteral antibiotic
rectal temperature was normal sepsis recurred in 8% of the coverage has been withdrawn. Use of these supplemental
patients. If in addition the WBC count was within the findings has provided a considerably more accurate guide
normal range, the recurrence rate fell to 2%. However, no for the earlier discontinuation of antibiotic therapy as well
patient had a recurrence of infection when the temperature as for a more expeditious search for the focus of a now-
and WBC count were normal and the differential WBC predictable recurrent infection.
count contained less than 3% immature granulocytes. This same approach can offer an even greater confidence
that infection has been eradicated and that the patient can
COMMENT
be discharged from the hospital with almost no chance of a
It has been routine in surgical practice to discontinue recurrent infection developing. However, should signs still
parenteral antibiotic therapy after five to ten days, unless suggest a strong possibility of sepsis reappearing, the
some sign of continuing infection dictates otherwise. To patient can be made aware of the risk and instructed in what
date, fever has been the criterion used by most surgeons, to look for and in what should be done.
yet it has by no means been an infallible prognosticator.1 References
The addition of a WBC count to the temperature chart
1. Lennard ES, Dellinger EP, Wertz MJ, et al: Implications of leuko-
increases the accuracy of clinical assessment, but this
cytosis and fever at conclusion of antibiotic therapy for intra-abdominal
combination also does not provide full proof of an infection. sepsis. Ann Surg 1982;195:19-24.
Discussion
E. Stan Lennard, MD, Seattle: The optimal duration of iron, and elevated levels of copper and ceruloplasmin. These
antibiotic treatment for infection has been very difficult to deter¬ responses are associated with the metabolism of injury and inflam¬
mine. This report has provided us with significant assistance in mation, and the degree of responses may be proportional to the
determining when antibiotic therapy can be stopped and who is at stimuli. Of great importance is the recognition that these re¬
risk for recurrent infection. sponses are elicited by considerably more than living bacteria.
In 1976 we wrote an intra-abdominal sepsis study protocol at the Endotoxin, exotoxin, viruses, fungi, antigen/antibody complexes,
University of Washington, Seattle, and defined one to two afebrile pyrogenic steroids, lymphokines, and certain synthetic agents
days as the criterion for discontinuing antibiotic therapy in pa¬ induce synthesis of this endogenous mediator. It is still not clear
tients who had been operated on and who had exhibited a carefully whether there is more than one mediator involved, or if molecular
defined clinical response. Two years ago, we reported our observa¬ heterogeneity exists in mediators with stereochemically specific
tions concerning predictors for persisting or recurrent infection in binding sites. The significance is that fever and/or leukocytosis do
the Annals of Surgery (1982;195:19-24). No intra-abdominal infec¬ not necessarily equal bacterial infection or pus.
tions recurred in afebrile patients who had normal WBC counts The trend toward restrictive use of antibiotics is good. The data
when antibiotic therapy was stopped, but one third of the afebrile presented here, and perhaps some of our own, will help to generate
patients with persisting leukocytosis manifested recurrent intra- algorithms for stopping antibiotic treatment that will ensure the
abdominal infection within two months of follow-up. A smaller safe treatment of our patients.
number of patients who were clinically well had persisting low- How many patients were followed up for at least 30 days, or
grade fevers, usually with an associated leukocytosis, when anti¬ as many as 60 days, and were late recurrences of infection seen
biotic therapy was stopped. Almost 60% of these patients man¬ during this longer time interval? We found abdominal recurrences
ifested recurrent intra-abdominal infections. In the majority of the in patients who were followed up postoperatively for two months
patients whom we reviewed, a blind continuation of antibiotic and recommended that patients at risk be followed up for that
therapy would have been potentially hazardous by delaying the length of time.
manifestation of a suppurative or necrotic process that ultimately Second, I noticed that there were patients who had total WBC
required surgical drainage. That same year, Doberneck and Mit- counts of less than 5,000/cu mm, and had an incidence of sepsis of
telman (Surgery, Gynecology, and Obstetrics 1982;154:875-879) approximately 30%. What percentage of band cells were seen in
correlated a prolonged duration of fever prior to drainage of those patients?
postoperative abdominal infection with a high redrainage rate and Dr Stone: Often, patients do not have WBC counts and fevers
increased mortality. that coincide. Indeed, it appears that these pyrogens and various
We are indebted to Kampschmidt, Robinson, Dinarello, and other stimulating factors are not the same. There may be a
Beeson, to name only a few researchers, who have provided data multiplicity of these influences.
that assist the interpretation of our observations and those pre¬ We were not able to obtain follow-up data on all of the patients.
sented in this review. There seems to be a mediator of inflammation We deleted those patients who were followed up for less than 30
that is actively released by bone marrow-derived fixed and circu¬ days, leaving a total of 1,123 patients who were followed up after
lating macrophages and monocytes. It seems to be a heat-labile discharge from the hospital.
glycoprotein with a molecular weight of approximately 15,000 Dr Lennard's observation regarding leukopenia was striking, in
daltons that is released from these cells when they are phago- that all patients who had leukopenia and who also had recurrent
cytically active. It has been called endogenous (or leukocyte) sepsis had immature granulocytic forms. This was absolutely
pyrogen, leukocyte endogenous mediator, colony-stimulating ac¬ consistent. The presence of an immature granulocyte in the smear
tivity, and (more recently) interleukin 1. A spectrum of responses of a patient who had leukopenia almost guaranteed that the patient
to this mediator includes fever, granulopoiesis, leukocytosis, re¬ would have recurrent sepsis.
lease of acute-phase proteins, decreased serum levels of zinc and

Downloaded from www.archsurg.com at McMaster University, Health Sciences Library, on December 3, 2009

Potrebbero piacerti anche