Solitary Blood Cultures as
a
Quality
Assurance Indicator Ron B. Schifman, M.D.,*† Calvin L. Strand, M.D.,‡ Egon Braun, M.D.,‡ Anthony Louis-Charles, M.D.,‡ Ronald P. Spark, M.D.¶ and Mary L. Fried, Ph.D.¶ &dag er; Department of Clinical Pathology, Tucson VA Medical Center and University of Arizona, College of Medicine, Tucson, Arizona 85723;‡ Department of Pathology, Beth Israel Medical Center, Mount Sinai School of Medicine, New York, New York 10003; and ¶ Department of Pathology, Tucson Medical Center, Tucson, Arizona 85712 For patients with suspected bacteremia, at least two separate blood cultures are recommended to achieve maximum sensitivity and to properly interpret results. Since a single blood collection may signify an improper procedure with serious consequences if the diagnosis of blood stream infection is missed, we investigated this problem with studies at three teaching hospitals (A, B, and C) and by a survey of 38 other hospitals. The incidence of solitary blood cultures ranged from 1 to 99% (median 26%) at the surveyed institutions. Among the cases investigated at hospitals B and C, between 10 and 30% of solitary blood cultures were not clinically indicated, while most of the others were caused by the physician not knowing that one culture was insufficient or by failure to complete the diagnostic plan. Focused concurrent intervention at hospital B was associated with reductions in solitary blood cultures from 40.0 to 24.6% (p = 0.045) and a decline in those not indicated from 38.1 to 12.5% (p = 0.192). Global educational efforts at hospital A were associated with a decrease in solitary blood culture rates from 52 to 37% (p = 0.016). These results show that blood culture practice varies widely among institutions in spite of consensus recommendations for proper specimen collections. We estimate that, nationwide, up to 18,000 etiologic diagnoses of bacteremia are missed annually because of this problem. Monitoring institutional solitary blood cultures is recommended as a test utilization indicator and as the basis for improving blood culture practice.
The blood culture is one of the most important and thoroughly studied procedures performed by the clinical laboratory. Multiple factors can impact test performance. For example, there is a direct association between test sensitivity and volume of blood cultured. Another important determinant of blood culture sensitivity as well as result interpretation is the number of blood cultures performed per septic or febrile episode (1). The sensitivity of blood culture results will
increase from approximately 80% when a single 20-ml sample is cultured to 99% when a set of three separate 20-ml specimens are processed (2). Collection of multiple samples also aids with the interpretation of blood culture results (3). For example, growth detected in one of several blood specimens suggests intermittent bacteremia or contamination, while a different condition, such as endocarditis, is suggested when the same microorganism is continuously isolated from
multiple specimens. Aronson and Bor (3) provide the following recommendations on the use of blood cultures to diagnose bacteremia: (a) a single blood culture is rarely, if ever sufficient, (b) two specimens should be obtained when the anticipated pathogen is different from the usual contaminating flora, (c) three should be obtained when likelihood bacteremia is high or continuous bacteremia is suspected, and (d) four or more are suggested when anticipated pathogen is also a common contaminant or when the patient has received antimicrobial therapy. There is widespread consensus about collecting multiple blood culture specimens. Among 178 clinical microbiology laboratories surveyed in 1986, all recommended collecting more than one sample for blood culture, and 79% recommended three collections per day (4). Thus, under most circumstances a solitary blood culture indicates an improper diagnostic procedure that may have detrimental outcome if bacteremia is not recognized and the etiologic diagnosis is not made. We evaluated solitary blood cultures as a quality indicator of blood culture utilization at three different hospitals and surveyed 38 additional institutions.
METHODS AND STUDY DESIGN * To whom requests for reprints should be addressed at Tucson VA Medical Center, Clinical Pathology (113), South 6th Avenue, Tuscon, AZ 85723.
Data were prospectively collected at three teaching hospitals (referred to as A, B, and C) ranging in size 132
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
133 from 325 to 934 beds in 1988 and 1989. Cases were identified in the laboratory by examining accession logs. Blood cultures are collected by house staff at hospitals A and B and by a phlebotomy service at hospital C. The phlebotomy service was not instructed to automatically collect multiple specimens unless specifically ordered by the physician. A blood culture was defined as a sample of blood obtained in an aseptic fashion from a single venipuncture and inoculated into media that supports microbial growth. A solitary blood culture was defined as a single blood collection during any 24-hour period. All three laboratories recommended that at least two blood specimens be cultured per septic episode. Special blood cultures for fungi and mycobacteria were not included in the study. Pediatric blood cultures were excluded from analysis in hospital C, but not A. Hospital B does not provide care for pediatric patients. Emergency room collections were assigned to the outpatient
category. Intervention procedures were conducted in hospitals A and B to improve blood culture practice. The Pathology department at hospital A redesigned the requisition form to include procedures for optimum culturing practice and distributed a memorandum describing appropriate blood culturing recommendations to the entire medical staff (Fig. 1). A repeat survey of 100 consecutive patients having blood cultures was conducted one year after the first surveillance period,
Fig.
1. The
and six months after the duced.
new
requisition
was
intro-
The Pathology department at hospital B conducted focused concurrent interventions during a five-month period. A memorandum reporting each solitary blood culture occurrence was sent to the responsible attending physician within seven days of the event. In addition, the physician was asked to explain why the solitary blood culture was performed or if it was indicated. A pathologist conducted a retrospective chart review of patients having solitary blood cultures at hospital C to assess indications. An additional 38 institutions were surveyed with a questionnaire distributed to pre-registrants at two national meeting workshops (1988 and 1989) sponsored by the American Society of Clinical Pathologists. Participants were asked to review retrospectively the last 50 or more patients having blood cultures at their institution and compute the frequency of solitary collections as defined above. Demographic information about these institutions was not collected. Finally, we derived an equation (the Missed Bacteremia Equation) that approximates the number of missed bacteremias, based on the expected decrease in test sensitivity caused by solitary blood culturing
(see the Appendix). RESULTS The number of solitary blood cultures obtained at the study hospitals is shown in Table 1. At hospital
redesigned requisition form
for blood cultures.
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
134 Table 1
B declined from 40.0 to 24.6% (X2 4.06, p 0.045) between the first and last study month, while those not clinically indicated fell from 38.1 to 12.5% (Fisher exact test, p 0.192). Results of the questionnaire were distributed to workshop registrants (Fig. 2). The median solitary blood culture prevalence was 26%. Results derived from the missed bacteremia equation are also shown (Fig. 3). =
Number of Blood Culture Specimens Collected at Three Hospitals
=
=
DISCUSSION of 100 patients cultured were pediatric, and of these eight had solitary cultures performed. However, with pediatric blood cultures excluded, the solitary adult blood culture rate was still high (48.9%). More than three blood cultures were collected within 24 hours in three (3%) patients from hospital A and one (0.8%) in hospital C (hospital B was not evalu-
A, eight seven
ated). The percentage of positives (excluding contaminants) isolated from patients having solitary, two, and greater than two blood cultures was 11.4%, 21.5%, and 33.3%, respectively, in hospital A. These differences are not statistically significant (X2 2.72, df 2, p = 0.258). The solitary blood culture frequency in hospital A declined from 52 to 35% (X2 5.88, p 0.016) after intervention. Focused, concurrent intervention in hospital B produced 53 responses (59% response rate). Attending physicians indicated that cultures were not indicated in 16 (30.2%) cases. Of the 37 clinically indicated cultures, attending physicians indicated that 14 (37.8%) were solitary because house staff did not =
=
There is
a
trend in the United States toward im-
proving the quality and cost effectiveness of health care by standardizing diagnostic procedures and developing medical practice guidelines (5). However, striking variations in practice exist among particular types of hospitals and geographic regions (6). Variations in practice patterns and test utilization may occur for a variety of reasons but should move toward uniformity when consensus standards are developed
=
=
know that this was insufficient. All blood cultures in this category were from inpatients. Ten (27.0%) solitary blood cultures were attributed to a failure by the physician to follow through on the diagnostic plan, and eight of these were associated with patients from either the emergency room or clinic. Other reasons for solitary blood cultures included change in patient status (three), technical problem with collecting a second specimen (three), testing performed as follow-up after therapy for bacteremia (three), specimens obtained simultaneously from two separate venipuncture sites and counted as a single collection (two), and patients who refused phlebotomy or were discharged before another could be taken (two). Although hospital C had a lower solitary blood culture rate, the reasons were similar and included physicians who did not understand that one was insufficient (three), patients who left or expired (three), failure to follow through on plan (two), change in patient status (one), and inappropriate indication (one). The proportion of solitary blood cultures in hospital
Fig.
2. Cumulative distribution of
rates in 38
solitary blood culture
surveyed hospitals.
Fig. 3. Effect of solitary blood culture frequency on missed bacteremias assuming 80% sensitivity for a 10 ml sample volume.
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
135 based on sound scientific principles and convincing clinical studies. There also must be effective monitoring systems that provide information about the sources of variation that can be used as a basis for
improvement (7). The diagnosis of bloodstream always made by blood culture.
infections is nearly There is consensus among infectious disease and clinical microbiology specialists for optimal blood culture testing procedures that has been validated from numerous clinical studies (1-3, 8). This includes the recommendation that multiple blood specimens be obtained to achieve the best sensitivity and to interpret results properly. This study, among others (9), shows that the existence of a consensus standard does not necessarily mean that it will be followed in practice. Indicators of appropriate laboratory testing are an important component of assessing the quality of patient care (10). While past studies have focused on difficulties related to excessive test usage (10-12), solitary blood cultures more often signify underutilization of a diagnostic laboratory procedure. We were unable to assess the overall quality of blood culture practice by using solitary blood cultures as an indicator of proper test utilization, since the clinical appropriateness of nonsolitary blood cultures was not evaluated, nor did we know the incidence of bacteremic patients among those not having a blood culture. However, in another study of 344 febrile patients, Makadon et al. (13) found that house officers performed blood cultures within 24 hours of the onset of fever in only 52% of febrile patients and failed to collect blood cultures promptly in 10% of eventually bacteremic patients. Solitary specimens were collected in 20% of the patients tested. These data support the notion that blood cultures, even when properly col-
lected,
are
underutilized.
We found that solitary blood culture collections are common, while the number of excessive specimen collections (greater than three per patient) was infrequent. These data challenge the perception that excessively repetitious blood cultures are a significant utilization problem. In fact, inappropriate test utilization is more likely to be recognized by the solitary blood culture that is clinically not indicated-a problem accounting for 30.6 and 10% of solitary blood cultures surveyed at hospitals B and C, respectively, that without intervention would undoubtedly have been more prevalent in hospital B. This finding helps confirm the impression of others (14) that blood cultures are sometimes collected unnecessarily. Information obtained from quality indicators is useful only as far as it can be used to understand the sources of error and to make knowledgeable decisions
that improve practice and prevent problems in the future (7, 15). Many problems encountered at hospital B were preventable. Two of the most common problems (not knowing that a single culture was insufficient and tests not indicated) could be potentially corrected by educational efforts. The intervention procedure conducted at hospital B showed that focused, concurrent feedback was partially successful as an educational aid. Similarly, global communication with the medical staff and alterations in the test requisition form were moderately effective at hospital A. Previous studies have also demonstrated the value of employing educational activities and changes in the requisition form to improve test utilization, but these have primarily dealt with test overutilization (10-12). Intervention studies also show that beneficial effects are transient unless the process is continuously done. Blood culture specimens were obtained by a phlebotomy service in study hospital C but not A or B, and this may be an important variable accounting for the lower prevalence of solitary cultures observed at hospital C. A phlebotomy service at hospital B would have undoubtedly helped prevent the occurrence of 65% of solitary blood cultures caused by failure of the house staff to recognize the requirement for multiple collections and their failure to follow through with additional specimen collections. The risk of misutilizing the blood culture procedure can be substantial when the etiologic diagnosis of bacteremia is not made and antimicrobial therapy is given without benefit of in uitro susceptibility testing of the blood culture isolate. Failure to recognize blood stream infections because of improper test usage also can be potentially costly to institutions that are reimbursed by third-party payers according to disease severity. For example, the 1988 base (inlier) reimbursement for DRG 419 (pyrexia of unknown origin with urinary tract infection) is $3044, while DRG 416 (E. coli septicemia with urinary tract infection, over age 17) is $5054-a difference of $2010. The appropriate number of blood culture samples to collect has not been studied in all clinical circumstances. For example, it was observed that solitary blood cultures were usually performed on pediatric patients in hospital C’s intensive care unit, and these cases were excluded. No studies have assessed the sensitivity and interpretive value of collecting single versus multiple cultures on newborns. Sepsis in newborns is generally associated with higher levels of bacteremia than adults, and single blood cultures may therefore have greater sensitivity in infants (16). However, sensitivity may be compromised by the low volume of blood obtainable in this age group (17). Similarly, it is unclear how many specimens should be
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
136 collected when blood cultures are collected in outpatients, follow-up after therapy for bacteremia, and in other circumstances (14, 18). In addition, some solitary blood cultures are not preventable. For example, the patient may refuse, expire, or leave before another sample can be obtained. These factors must be considered when interpreting the data, and it may be unrealistic to expect solitary blood cultures to be reduced below a level of 5 to 10%, a frequency that probably will vary depending on patient mix, data interpretation, institutional resources, and culture practices. The median solitary blood culture rate that we observed in this study is similar to the results obtained (21.5% median incidence) among 500 laboratories in a study conducted by the College of American Pathologists in the summer of 1989 (19). In another study conducted in 1986, 110 of 185 laboratories (59%) reported bacteremia rates of between 6 and 15% of
patients cultured (4). Thus, assuming a 10% bacteremia rate and solitary blood culture frequency of 26%, estimate that as many as 5.25 bacteremias are missed per 1000 patients cultured nationally, or about 5% of bacteremic patients. This will vary among institutions in relation to both the solitary blood culture frequency, prevalence of bacteremic patients (Fig. 2), and indications for testing. Presuming a yearly national bacteremia rate of 350,000 (18) and solitary blood culture frequency of between 22 and 26%, we estimate that an etiologic diagnosis may be missed in up to 18,000 bacteremias per year. The prevalence of this problem may be underestimated since rates for septicemia are rising (20), and it has been shown that laboratory diagnosis of bacteremia in febrile patients is not consistently done (13). The model (see the Appendix) used for this estimate may underestimate missed bacteremias if the sample volume is below 20 ml (8). The number of missed bacteremias will be overestimated by the frequency of solitary blood cultures collected for inappropriate indications, a situation in which the pretest likelihood of a positive result will be low. Likewise, missed bacteremias will be higher than the equation indicates if bacteremic patients are not cultured. The model is theoretical and not outcome based. Additional quality assurance studies to evaluate the genuine impact of our findings would require assessments to identify a solitary blood culture incident in which a bacteremic patient has increased mortality, morbidity, length of stay, or costs that can be attributed to delay or failure
cultures are not optimally utilized at the current time. Surveillance of solitary blood cultures with appropriate intervention when problems are detected should be considered for implementation into quality assurance programs to improve blood culture practices. References 1. Weinstein MP, Reller LB, Murphy JR, Lichtenstein KA. The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults: I. Laboratory and epidemiologic observations. Rev Infect Dis
1983;5:35-53. 2.
3. Aronson
MD, Bor DH. Blood cultures. Ann Intern Med 1987;106:246-253.
4. Kehl KSC. Results of survey of blood culture methods. Clin Microbiol Newsletter 1986; 8:117-123, 8:127-134. 5. Brook RH. Practice guidelines and practicing medicine: are
they compatible? JAMA 1989;262:3027-3030. MR, Kosecoff J, Park RE, et al. Does inappropriate use explain geographic variation in the use of health care services? A study of three procedures. JAMA 1987 ;258 :2533-
6. Chassin
we
to make
a
etiologic diagnosis.
We conclude that solitary blood cultures are a useful quality indicator of test utilization. Based on a nationwide sample of 41 hospitals, it appears that blood
Washington JA II. Blood cultures: principles and techniques. Mayo Clin Proc 1975;50:91-98.
2537. 7. Berwick DM. Continuous improvement as an ideal in health care. N Engl J Med 1989;320:53-56. 8. Washington JA, Ilstrup DM. Blood cultures: issues and controversies. Rev Infect Dis 1986;8:792-802. 9. Lomas J, Anderson GM, Dominick-Pierre K, et al. Do practice guidelines guide practice? The effect of a consensus statement on the practice of physicians. N Engl J Med 1989;321:13061311. 10. Spiegel JS, Shapiro MF, Berman B, Greenfield S. Changing physician test ordering in a university hospital. An intervention of physician participation, explicit criteria, and feedback. Arch Intern Med 1989;149:549-553. 11. Winkel P, Statland BE. Assessing cost savings when unnecessary utilization of laboratory tests can be abolished. Am J Clin Pathol 1984;82:418-423. 12. Berwick DM, Coltin KL. Feedback reduces test use in a health maintenance organization. JAMA 1986;255:1450-1454. 13. Makadon HJ, Bor D, Friedland G, et al. Febrile inpatients: house officer’s use of blood cultures. J Gen Intern Med
1987;2:293-297. 14. Neu HC. Cost effective blood cultures: Is it possible or impossible to modify behavior? Infect Control 1986;7:32-33. 15. Batalden PB, Buchanan ED. Industrial models of quality improvement, in Providing Quality Care, Goldfield N, Nash DB, eds. Philadelphia, American College of Physicians, 1989, 133159. 16. Durbin WA, Szymaczak EG, Goldmann DA. Quantitative blood cultures in childhood bacteremia. J Pediatr 1978;92:778-780. 17. Neal PR, Kleiman MB, Reynolds JK, et al. Volume of blood submitted for culture from neonates. J Clin Microbiol
1986;24:353-356. Bryan CS. Clinical implications of positive blood
cultures. Clin Microbiol Rev 1989;2:329-353. 19. Schifman RB, Bachner P. Blood Culture Utilization: Data Analysis and Critique. Q-Probes 89-05A. College of American 18.
Pathologists, Northfield, Illinois, 1990. 20. Centers for Disease Control. Increase in national hospital discharge survey rates for septicemia: United States, 1979-1987. MMWR 1990;39:31-34.
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
137 APPENDIX
SEN (blood culture sensitivity) 1 - (SBC/5) MB (number of bacteremic patients missed) (BPT/SEN) - BPT =
=
The Missed Bacteremia Equation is an equation that approximates the number of missed bacteremias, based on the expected decrease in test sensitivity caused by solitary blood culturing. The estimated incidence of missed bacteremias is calculated as the difference between the theoretical bacteremia rate (assuming 100% sensitivity) and the actual measured bacteremia rate (excluding contaminants). The model assumes that sensitivity is 80% for a single specimen and 100% when two or more specimens are collected. Thus, the sensitivity is reduced by 1% for each 5% increment in the solitary blood culture frequency.
SBC BPT
=
=
solitary blood culture incidence (between 0.00 and 1.00) Bacteremic patients (excluding contaminants)
Example: An institution’s solitary blood culture patients during the year is 18%. A total
rate for adult
of 145 bacteremias
were
detected
during
that year.
Thus:
example, of 150 bacteremic patients, about 5 (or 3.6%) would be expected to have been undetected because of a lowered sensitivity caused by an 18% solitary blood culture rate. In this
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
a
Quality
Assurance Indicator Ron B. Schifman, M.D.,*† Calvin L. Strand, M.D.,‡ Egon Braun, M.D.,‡ Anthony Louis-Charles, M.D.,‡ Ronald P. Spark, M.D.¶ and Mary L. Fried, Ph.D.¶ &dag er; Department of Clinical Pathology, Tucson VA Medical Center and University of Arizona, College of Medicine, Tucson, Arizona 85723;‡ Department of Pathology, Beth Israel Medical Center, Mount Sinai School of Medicine, New York, New York 10003; and ¶ Department of Pathology, Tucson Medical Center, Tucson, Arizona 85712 For patients with suspected bacteremia, at least two separate blood cultures are recommended to achieve maximum sensitivity and to properly interpret results. Since a single blood collection may signify an improper procedure with serious consequences if the diagnosis of blood stream infection is missed, we investigated this problem with studies at three teaching hospitals (A, B, and C) and by a survey of 38 other hospitals. The incidence of solitary blood cultures ranged from 1 to 99% (median 26%) at the surveyed institutions. Among the cases investigated at hospitals B and C, between 10 and 30% of solitary blood cultures were not clinically indicated, while most of the others were caused by the physician not knowing that one culture was insufficient or by failure to complete the diagnostic plan. Focused concurrent intervention at hospital B was associated with reductions in solitary blood cultures from 40.0 to 24.6% (p = 0.045) and a decline in those not indicated from 38.1 to 12.5% (p = 0.192). Global educational efforts at hospital A were associated with a decrease in solitary blood culture rates from 52 to 37% (p = 0.016). These results show that blood culture practice varies widely among institutions in spite of consensus recommendations for proper specimen collections. We estimate that, nationwide, up to 18,000 etiologic diagnoses of bacteremia are missed annually because of this problem. Monitoring institutional solitary blood cultures is recommended as a test utilization indicator and as the basis for improving blood culture practice.
The blood culture is one of the most important and thoroughly studied procedures performed by the clinical laboratory. Multiple factors can impact test performance. For example, there is a direct association between test sensitivity and volume of blood cultured. Another important determinant of blood culture sensitivity as well as result interpretation is the number of blood cultures performed per septic or febrile episode (1). The sensitivity of blood culture results will
increase from approximately 80% when a single 20-ml sample is cultured to 99% when a set of three separate 20-ml specimens are processed (2). Collection of multiple samples also aids with the interpretation of blood culture results (3). For example, growth detected in one of several blood specimens suggests intermittent bacteremia or contamination, while a different condition, such as endocarditis, is suggested when the same microorganism is continuously isolated from
multiple specimens. Aronson and Bor (3) provide the following recommendations on the use of blood cultures to diagnose bacteremia: (a) a single blood culture is rarely, if ever sufficient, (b) two specimens should be obtained when the anticipated pathogen is different from the usual contaminating flora, (c) three should be obtained when likelihood bacteremia is high or continuous bacteremia is suspected, and (d) four or more are suggested when anticipated pathogen is also a common contaminant or when the patient has received antimicrobial therapy. There is widespread consensus about collecting multiple blood culture specimens. Among 178 clinical microbiology laboratories surveyed in 1986, all recommended collecting more than one sample for blood culture, and 79% recommended three collections per day (4). Thus, under most circumstances a solitary blood culture indicates an improper diagnostic procedure that may have detrimental outcome if bacteremia is not recognized and the etiologic diagnosis is not made. We evaluated solitary blood cultures as a quality indicator of blood culture utilization at three different hospitals and surveyed 38 additional institutions.
METHODS AND STUDY DESIGN * To whom requests for reprints should be addressed at Tucson VA Medical Center, Clinical Pathology (113), South 6th Avenue, Tuscon, AZ 85723.
Data were prospectively collected at three teaching hospitals (referred to as A, B, and C) ranging in size 132
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
133 from 325 to 934 beds in 1988 and 1989. Cases were identified in the laboratory by examining accession logs. Blood cultures are collected by house staff at hospitals A and B and by a phlebotomy service at hospital C. The phlebotomy service was not instructed to automatically collect multiple specimens unless specifically ordered by the physician. A blood culture was defined as a sample of blood obtained in an aseptic fashion from a single venipuncture and inoculated into media that supports microbial growth. A solitary blood culture was defined as a single blood collection during any 24-hour period. All three laboratories recommended that at least two blood specimens be cultured per septic episode. Special blood cultures for fungi and mycobacteria were not included in the study. Pediatric blood cultures were excluded from analysis in hospital C, but not A. Hospital B does not provide care for pediatric patients. Emergency room collections were assigned to the outpatient
category. Intervention procedures were conducted in hospitals A and B to improve blood culture practice. The Pathology department at hospital A redesigned the requisition form to include procedures for optimum culturing practice and distributed a memorandum describing appropriate blood culturing recommendations to the entire medical staff (Fig. 1). A repeat survey of 100 consecutive patients having blood cultures was conducted one year after the first surveillance period,
Fig.
1. The
and six months after the duced.
new
requisition
was
intro-
The Pathology department at hospital B conducted focused concurrent interventions during a five-month period. A memorandum reporting each solitary blood culture occurrence was sent to the responsible attending physician within seven days of the event. In addition, the physician was asked to explain why the solitary blood culture was performed or if it was indicated. A pathologist conducted a retrospective chart review of patients having solitary blood cultures at hospital C to assess indications. An additional 38 institutions were surveyed with a questionnaire distributed to pre-registrants at two national meeting workshops (1988 and 1989) sponsored by the American Society of Clinical Pathologists. Participants were asked to review retrospectively the last 50 or more patients having blood cultures at their institution and compute the frequency of solitary collections as defined above. Demographic information about these institutions was not collected. Finally, we derived an equation (the Missed Bacteremia Equation) that approximates the number of missed bacteremias, based on the expected decrease in test sensitivity caused by solitary blood culturing
(see the Appendix). RESULTS The number of solitary blood cultures obtained at the study hospitals is shown in Table 1. At hospital
redesigned requisition form
for blood cultures.
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
134 Table 1
B declined from 40.0 to 24.6% (X2 4.06, p 0.045) between the first and last study month, while those not clinically indicated fell from 38.1 to 12.5% (Fisher exact test, p 0.192). Results of the questionnaire were distributed to workshop registrants (Fig. 2). The median solitary blood culture prevalence was 26%. Results derived from the missed bacteremia equation are also shown (Fig. 3). =
Number of Blood Culture Specimens Collected at Three Hospitals
=
=
DISCUSSION of 100 patients cultured were pediatric, and of these eight had solitary cultures performed. However, with pediatric blood cultures excluded, the solitary adult blood culture rate was still high (48.9%). More than three blood cultures were collected within 24 hours in three (3%) patients from hospital A and one (0.8%) in hospital C (hospital B was not evalu-
A, eight seven
ated). The percentage of positives (excluding contaminants) isolated from patients having solitary, two, and greater than two blood cultures was 11.4%, 21.5%, and 33.3%, respectively, in hospital A. These differences are not statistically significant (X2 2.72, df 2, p = 0.258). The solitary blood culture frequency in hospital A declined from 52 to 35% (X2 5.88, p 0.016) after intervention. Focused, concurrent intervention in hospital B produced 53 responses (59% response rate). Attending physicians indicated that cultures were not indicated in 16 (30.2%) cases. Of the 37 clinically indicated cultures, attending physicians indicated that 14 (37.8%) were solitary because house staff did not =
=
There is
a
trend in the United States toward im-
proving the quality and cost effectiveness of health care by standardizing diagnostic procedures and developing medical practice guidelines (5). However, striking variations in practice exist among particular types of hospitals and geographic regions (6). Variations in practice patterns and test utilization may occur for a variety of reasons but should move toward uniformity when consensus standards are developed
=
=
know that this was insufficient. All blood cultures in this category were from inpatients. Ten (27.0%) solitary blood cultures were attributed to a failure by the physician to follow through on the diagnostic plan, and eight of these were associated with patients from either the emergency room or clinic. Other reasons for solitary blood cultures included change in patient status (three), technical problem with collecting a second specimen (three), testing performed as follow-up after therapy for bacteremia (three), specimens obtained simultaneously from two separate venipuncture sites and counted as a single collection (two), and patients who refused phlebotomy or were discharged before another could be taken (two). Although hospital C had a lower solitary blood culture rate, the reasons were similar and included physicians who did not understand that one was insufficient (three), patients who left or expired (three), failure to follow through on plan (two), change in patient status (one), and inappropriate indication (one). The proportion of solitary blood cultures in hospital
Fig.
2. Cumulative distribution of
rates in 38
solitary blood culture
surveyed hospitals.
Fig. 3. Effect of solitary blood culture frequency on missed bacteremias assuming 80% sensitivity for a 10 ml sample volume.
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
135 based on sound scientific principles and convincing clinical studies. There also must be effective monitoring systems that provide information about the sources of variation that can be used as a basis for
improvement (7). The diagnosis of bloodstream always made by blood culture.
infections is nearly There is consensus among infectious disease and clinical microbiology specialists for optimal blood culture testing procedures that has been validated from numerous clinical studies (1-3, 8). This includes the recommendation that multiple blood specimens be obtained to achieve the best sensitivity and to interpret results properly. This study, among others (9), shows that the existence of a consensus standard does not necessarily mean that it will be followed in practice. Indicators of appropriate laboratory testing are an important component of assessing the quality of patient care (10). While past studies have focused on difficulties related to excessive test usage (10-12), solitary blood cultures more often signify underutilization of a diagnostic laboratory procedure. We were unable to assess the overall quality of blood culture practice by using solitary blood cultures as an indicator of proper test utilization, since the clinical appropriateness of nonsolitary blood cultures was not evaluated, nor did we know the incidence of bacteremic patients among those not having a blood culture. However, in another study of 344 febrile patients, Makadon et al. (13) found that house officers performed blood cultures within 24 hours of the onset of fever in only 52% of febrile patients and failed to collect blood cultures promptly in 10% of eventually bacteremic patients. Solitary specimens were collected in 20% of the patients tested. These data support the notion that blood cultures, even when properly col-
lected,
are
underutilized.
We found that solitary blood culture collections are common, while the number of excessive specimen collections (greater than three per patient) was infrequent. These data challenge the perception that excessively repetitious blood cultures are a significant utilization problem. In fact, inappropriate test utilization is more likely to be recognized by the solitary blood culture that is clinically not indicated-a problem accounting for 30.6 and 10% of solitary blood cultures surveyed at hospitals B and C, respectively, that without intervention would undoubtedly have been more prevalent in hospital B. This finding helps confirm the impression of others (14) that blood cultures are sometimes collected unnecessarily. Information obtained from quality indicators is useful only as far as it can be used to understand the sources of error and to make knowledgeable decisions
that improve practice and prevent problems in the future (7, 15). Many problems encountered at hospital B were preventable. Two of the most common problems (not knowing that a single culture was insufficient and tests not indicated) could be potentially corrected by educational efforts. The intervention procedure conducted at hospital B showed that focused, concurrent feedback was partially successful as an educational aid. Similarly, global communication with the medical staff and alterations in the test requisition form were moderately effective at hospital A. Previous studies have also demonstrated the value of employing educational activities and changes in the requisition form to improve test utilization, but these have primarily dealt with test overutilization (10-12). Intervention studies also show that beneficial effects are transient unless the process is continuously done. Blood culture specimens were obtained by a phlebotomy service in study hospital C but not A or B, and this may be an important variable accounting for the lower prevalence of solitary cultures observed at hospital C. A phlebotomy service at hospital B would have undoubtedly helped prevent the occurrence of 65% of solitary blood cultures caused by failure of the house staff to recognize the requirement for multiple collections and their failure to follow through with additional specimen collections. The risk of misutilizing the blood culture procedure can be substantial when the etiologic diagnosis of bacteremia is not made and antimicrobial therapy is given without benefit of in uitro susceptibility testing of the blood culture isolate. Failure to recognize blood stream infections because of improper test usage also can be potentially costly to institutions that are reimbursed by third-party payers according to disease severity. For example, the 1988 base (inlier) reimbursement for DRG 419 (pyrexia of unknown origin with urinary tract infection) is $3044, while DRG 416 (E. coli septicemia with urinary tract infection, over age 17) is $5054-a difference of $2010. The appropriate number of blood culture samples to collect has not been studied in all clinical circumstances. For example, it was observed that solitary blood cultures were usually performed on pediatric patients in hospital C’s intensive care unit, and these cases were excluded. No studies have assessed the sensitivity and interpretive value of collecting single versus multiple cultures on newborns. Sepsis in newborns is generally associated with higher levels of bacteremia than adults, and single blood cultures may therefore have greater sensitivity in infants (16). However, sensitivity may be compromised by the low volume of blood obtainable in this age group (17). Similarly, it is unclear how many specimens should be
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
136 collected when blood cultures are collected in outpatients, follow-up after therapy for bacteremia, and in other circumstances (14, 18). In addition, some solitary blood cultures are not preventable. For example, the patient may refuse, expire, or leave before another sample can be obtained. These factors must be considered when interpreting the data, and it may be unrealistic to expect solitary blood cultures to be reduced below a level of 5 to 10%, a frequency that probably will vary depending on patient mix, data interpretation, institutional resources, and culture practices. The median solitary blood culture rate that we observed in this study is similar to the results obtained (21.5% median incidence) among 500 laboratories in a study conducted by the College of American Pathologists in the summer of 1989 (19). In another study conducted in 1986, 110 of 185 laboratories (59%) reported bacteremia rates of between 6 and 15% of
patients cultured (4). Thus, assuming a 10% bacteremia rate and solitary blood culture frequency of 26%, estimate that as many as 5.25 bacteremias are missed per 1000 patients cultured nationally, or about 5% of bacteremic patients. This will vary among institutions in relation to both the solitary blood culture frequency, prevalence of bacteremic patients (Fig. 2), and indications for testing. Presuming a yearly national bacteremia rate of 350,000 (18) and solitary blood culture frequency of between 22 and 26%, we estimate that an etiologic diagnosis may be missed in up to 18,000 bacteremias per year. The prevalence of this problem may be underestimated since rates for septicemia are rising (20), and it has been shown that laboratory diagnosis of bacteremia in febrile patients is not consistently done (13). The model (see the Appendix) used for this estimate may underestimate missed bacteremias if the sample volume is below 20 ml (8). The number of missed bacteremias will be overestimated by the frequency of solitary blood cultures collected for inappropriate indications, a situation in which the pretest likelihood of a positive result will be low. Likewise, missed bacteremias will be higher than the equation indicates if bacteremic patients are not cultured. The model is theoretical and not outcome based. Additional quality assurance studies to evaluate the genuine impact of our findings would require assessments to identify a solitary blood culture incident in which a bacteremic patient has increased mortality, morbidity, length of stay, or costs that can be attributed to delay or failure
cultures are not optimally utilized at the current time. Surveillance of solitary blood cultures with appropriate intervention when problems are detected should be considered for implementation into quality assurance programs to improve blood culture practices. References 1. Weinstein MP, Reller LB, Murphy JR, Lichtenstein KA. The clinical significance of positive blood cultures: a comprehensive analysis of 500 episodes of bacteremia and fungemia in adults: I. Laboratory and epidemiologic observations. Rev Infect Dis
1983;5:35-53. 2.
3. Aronson
MD, Bor DH. Blood cultures. Ann Intern Med 1987;106:246-253.
4. Kehl KSC. Results of survey of blood culture methods. Clin Microbiol Newsletter 1986; 8:117-123, 8:127-134. 5. Brook RH. Practice guidelines and practicing medicine: are
they compatible? JAMA 1989;262:3027-3030. MR, Kosecoff J, Park RE, et al. Does inappropriate use explain geographic variation in the use of health care services? A study of three procedures. JAMA 1987 ;258 :2533-
6. Chassin
we
to make
a
etiologic diagnosis.
We conclude that solitary blood cultures are a useful quality indicator of test utilization. Based on a nationwide sample of 41 hospitals, it appears that blood
Washington JA II. Blood cultures: principles and techniques. Mayo Clin Proc 1975;50:91-98.
2537. 7. Berwick DM. Continuous improvement as an ideal in health care. N Engl J Med 1989;320:53-56. 8. Washington JA, Ilstrup DM. Blood cultures: issues and controversies. Rev Infect Dis 1986;8:792-802. 9. Lomas J, Anderson GM, Dominick-Pierre K, et al. Do practice guidelines guide practice? The effect of a consensus statement on the practice of physicians. N Engl J Med 1989;321:13061311. 10. Spiegel JS, Shapiro MF, Berman B, Greenfield S. Changing physician test ordering in a university hospital. An intervention of physician participation, explicit criteria, and feedback. Arch Intern Med 1989;149:549-553. 11. Winkel P, Statland BE. Assessing cost savings when unnecessary utilization of laboratory tests can be abolished. Am J Clin Pathol 1984;82:418-423. 12. Berwick DM, Coltin KL. Feedback reduces test use in a health maintenance organization. JAMA 1986;255:1450-1454. 13. Makadon HJ, Bor D, Friedland G, et al. Febrile inpatients: house officer’s use of blood cultures. J Gen Intern Med
1987;2:293-297. 14. Neu HC. Cost effective blood cultures: Is it possible or impossible to modify behavior? Infect Control 1986;7:32-33. 15. Batalden PB, Buchanan ED. Industrial models of quality improvement, in Providing Quality Care, Goldfield N, Nash DB, eds. Philadelphia, American College of Physicians, 1989, 133159. 16. Durbin WA, Szymaczak EG, Goldmann DA. Quantitative blood cultures in childhood bacteremia. J Pediatr 1978;92:778-780. 17. Neal PR, Kleiman MB, Reynolds JK, et al. Volume of blood submitted for culture from neonates. J Clin Microbiol
1986;24:353-356. Bryan CS. Clinical implications of positive blood
cultures. Clin Microbiol Rev 1989;2:329-353. 19. Schifman RB, Bachner P. Blood Culture Utilization: Data Analysis and Critique. Q-Probes 89-05A. College of American 18.
Pathologists, Northfield, Illinois, 1990. 20. Centers for Disease Control. Increase in national hospital discharge survey rates for septicemia: United States, 1979-1987. MMWR 1990;39:31-34.
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
137 APPENDIX
SEN (blood culture sensitivity) 1 - (SBC/5) MB (number of bacteremic patients missed) (BPT/SEN) - BPT =
=
The Missed Bacteremia Equation is an equation that approximates the number of missed bacteremias, based on the expected decrease in test sensitivity caused by solitary blood culturing. The estimated incidence of missed bacteremias is calculated as the difference between the theoretical bacteremia rate (assuming 100% sensitivity) and the actual measured bacteremia rate (excluding contaminants). The model assumes that sensitivity is 80% for a single specimen and 100% when two or more specimens are collected. Thus, the sensitivity is reduced by 1% for each 5% increment in the solitary blood culture frequency.
SBC BPT
=
=
solitary blood culture incidence (between 0.00 and 1.00) Bacteremic patients (excluding contaminants)
Example: An institution’s solitary blood culture patients during the year is 18%. A total
rate for adult
of 145 bacteremias
were
detected
during
that year.
Thus:
example, of 150 bacteremic patients, about 5 (or 3.6%) would be expected to have been undetected because of a lowered sensitivity caused by an 18% solitary blood culture rate. In this
Downloaded from ajm.sagepub.com at UNIV OF WESTERN ONTARIO on April 11, 2015
