ACADEMIA
AND
CLINIC
Preoperative Screening: Value of Previous Tests David S. Macpherson, MD; Rita Snow, MD; and Richard P. Lofgren, MD, MPH
Objective: To determine the frequency of tests done in the year before elective surgery that might substitute for preoperative screening tests and to determine the frequency of test results that change from a normal value to a value likely to alter perioperative management. Design: Retrospective cohort analysis of computerized laboratory data (complete blood count, sodium, potassium, and creatinine levels, prothrombin time, and partial thromboplastin time). Setting: Urban tertiary care Veterans Affairs Hospital. Patients: Consecutive sample of 1109 patients who had elective surgery in 1988. Measurements and Main Results: At admission, 7549 preoperative tests were done, 47% of which duplicated tests performed in the previous year. Of 3096 previous results that were normal as defined by hospital reference range and done closest to the time of but before admission (median interval, 2 months), 13 (0.4%; 95% CI, 0.2% to 0.7%), repeat values were outside a range considered acceptable for surgery. Most of the abnormalities were predictable from the patient's history, and most were not noted in the medical record. Of 461 previous tests that were abnormal, 78 (17%; CI, 13% to 20%) repeat values at admission were outside a range considered acceptable for surgery (P < 0.001, frequency of clinically important abnormalities of patients with normal previous results with those with abnormal previous results). Conclusions: Physicians evaluating patients preoperatively could safely substitute the previous test results analyzed in this study for preoperative screening tests if the previous tests are normal and no obvious indication for retesting is present.
Annals of Internal Medicine. 1990;113:969-973. From the University of Minnesota and Minneapolis Veterans Affairs Medical Center, Minneapolis, Minnesota; and the University of Pittsburgh and the Pittsburgh Veterans Affairs Medical Center, Pittsburgh, Pennsylvania. For current author addresses, see end of text.
A reoperative testing is done to detect conditions that could alter perioperative management of the patient. Despite evidence that selective test ordering on the basis of history and physical examination is an appropriate method to evaluate surgical patients (1-9), it is common practice to screen all preoperative patients for unexpected clinically important laboratory abnormalities. The reasons that nonselective screening tests continue to be done are unknown. Physicians ordering preoperative tests may be unfamiliar with the results of studies on preoperative testing; however, educational interventions to decrease test ordering by both medical and surgical housestaff have had limited success (10, 11). Physicians may feel the history and physical examination are too insensitive to detect conditions that affect surgical management, and, therefore, screen all patients in order not to miss unsuspected abnormalities. They may feel nonselective screening decreases the risk of a malpractice suit, despite suggestions that nonselective screening actually increases legal culpability (12). Although the cost per test is low for most preoperative tests, the aggregate cost is substantial (13). Furthermore, nonselective testing frequently results in falsepositive findings that can lead to invasive and dangerous evaluations of abnormal results. It has been estimated that preoperative testing and evaluation cost more than 30 billion dollars annually, of which 60%, or 18 billion dollars, could be saved if tests were done selectively (12). We sought a practical method to decrease presurgical testing of elective surgical patients. Computerized display of previous test results had led to a decrease in test ordering in both an academic primary care medicine clinic (14) and an emergency room (15). If previous test results were readily available to physicians seeing patients before surgery, and test values rarely changed from a normal value to a value likely to affect perioperative management, previous tests might be safely substituted for preoperative screening tests. This might reassure practitioners that rescreening before surgery for occult but clinically important laboratory abnormalities is unnecessary and lead to fewer test orders. We hypothesized that many patients having elective surgery have accessible previous test results and that unexpected changes in test values from a normal range to a value likely to affect perioperative management are rare. To test these hypotheses, we reviewed data on patients who had elective surgery at our institution and determined the frequency of tests done in the year before elective surgery and the frequency of clinically important changes in these values. Methods Hospital discharge records of the Minneapolis Veterans Affairs Medical Center were used to identify patients who had
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12 Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
969
Table 1. Surgical Procedures Reviewed the Value of Preview Tests Surgical Procedure (ICD9 Discharge Code)* Unilateral repair of indirect inguinal hernia (5302) Unilateral repair of direct inguinal hernia (5301) Unilateral repair of inguinal hernia not otherwise specified (5300) Excision of hemorrhoids (4946) Total general surgical procedures Insertion of intraocular lens prosthesis at cataract extraction (1371) Other extracapsular extraction of lens (1359) Phacoemulsification and aspiration of cataract (1341)t Total ophthalmologic procedures Transurethral prostatectomy (602) Radical prostatectomy (605) Insertion or replacement of internal prosthesis of penis (6495) Total urologic procedures Total knee replacement (8141) Release of carpal tunnel (0443) Other total hip replacement (8159) Total hip replacement with use of methyl methacrylate (8151) Other replacement of head of femur (8162) Total orthopedic procedures Other septoplasty (2188) Local excision or destruction of intranasal lesion (2131) Alveoloplasty (245) Excision or destruction of other lesion of external ear (1829) Myringoplasty (194) Total otorhinolarygologic or oral surgery procedures Overall totals
To
Investigate Number of Procedures 52 65 17 26 160 342 3 6 351 207 43 37 287 78 59 52 21 2 212 21 17 23 15 23 99 1109
* Ninth revision. t This category includes patients who had phacoemulsification and aspiration without insertion of lens prosthesis. Patients who had phacoemulsification and aspiration of cataract with lens prosthesis insertion are included in ICD9 1371. t Delete subtotal values (160, 351, 287, 212, and 99).
one of the surgical procedures listed in Table 1. These surgical procedures were selected because they were common, usually elective in nature, generally the sole reason for hospitalization, and not commonly done as an outpatient. A potential disadvantage of this patient identification procedure is that it misses patients who are admitted for elective surgery but discharged without surgery because of abnormal presurgical screening tests. To determine the frequency of this event, we conducted a 1-month audit of patients discharged from the surgery wards in 1989. Of 260 patients discharged from the surgical wards during the time of the audit, no patient was discharged without surgery because of an abnormal preoperative screening test. For each patient identified by discharge record as having had elective/surgery, we recorded laboratory test results (Table 2) from 1 year before admission to the day after admission. These laboratory tests were chosen for study because they are commonly done for preoperative evaluation, have ranges outside of which alteration in management is recommended (4), and are contained in the computerized laboratory database. The results of laboratory tests obtained for inpatients and outpatients are 970
readily available to practitioners on the wards and in the outpatient clinics. Data are stored for most patients for over 1 year. The only laboratory panels done at our institution are the complete blood count and the electrolyte panel (sodium, potassium, chloride, and bicarbonate). All other blood analyses are ordered by individual test. We assumed that tests ordered within the first 2 days of hospitalization were for the purpose of preoperative screening. Patients were excluded from analysis if the admission testing was ordered through the emergency room (suggesting that the procedure was not elective) or if the admission testing was not ordered through one of the surgical wards (indicating that the patient was hospitalized for reasons other than elective surgery and that the testing done during the first 2 days of hospitalization may not have been done for presurgical screening). The location of admission testing was determined through coding on the laboratory test result. The frequency of a clinically important change in test value was determined by counting the number of tests that changed from the hospital reference range to a value that was abnormal enough to affect surgical management, previously defined as the "action limit range" by Kaplan and colleagues (4) (Table 2). Tests were counted in two ways. First, each value from the complete blood count (that is, leukocyte count, hemoglobin, and platelet count), and the electrolyte panel (that is, sodium and potassium) was counted as one test. Second, each complete blood count and electrolyte panel was counted as one test. In both analyses, the prothrombin time, partial thromboplastin time, and creatinine were counted as one test each. We reviewed the medical records of patients who had a clinically important change in test values. A test abnormality was counted as noticed by the physicians or nurses caring for the patient if it was mentioned in the progress notes, discharge summary, or anesthesia record, or if it was circled on the laboratory record. One of the investigators reviewed the medical record to determine the patient's recent medical history and whether any perioperative complications occurred. The 95% confidence intervals for differences in percentages were calculated in a standard manner, and the chi-square test was used to test differences in categorical variables. Results In 1988, 1323 patients had elective surgery of the types listed in Table 1. Of this total, 214 patients were excluded because admission laboratory testing had been ordered through the emergency department or a nonsurgical ward, leaving 1109 patients for analysis. The mean age of the included patients was 66.2 ± 10.6 (SD) years. The frequency of screening at admission for each type of test is indicated in Figure 1, top. If each patient had each test, 8872 tests would have been done. Overall, 7549 (85%) admission tests were done, which indicates that selective screening is probably not being
Table 2. Ranges of Laboratory Test Leukocyte count, x 109IL Hemoglobin, gIL Platelet count, x 109IL Prothrombin time, s Partial thromboplastin time, s Sodium, mmollL Potassium, mmollL Creatinine, /xmol/L
Test Values
Hospital Action Limit Range* Reference Range 4.5-11.0 120-170 130-380 11-13 21-35 136-143 3.7-5.2 70-120
3.0-15.0 100-180 115-800 10-13 22-40 130-150 3.2-5.8 < 120
* Values outside the action limit range would probably result in alteration in management (4).
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12
Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
practiced at our institution. The proportion of patients having laboratory testing in the year before surgery is presented in Figure 1, bottom. Of 7549 admission tests, 3557 (47%) repeated tests that were done in the year before admission for surgery. The median interval from the most recent previous test to admission test was 2 months. Of 3096 most recent previous results that were normal as defined by our hospital's reference range, 13 admission values exceeded the action range (0.4%; CI, 0.2% to 0.7%). These action range abnormalities consisted of 1 elevated leukocyte count, 6 elevated prothrombin times, 1 elevated potassium level, and 5 mildly elevated creatinine levels (Table 3). A review of the records with values that changed showed that most had clinical indications that the result would have changed (Table 3), and only 1 was noted preoperatively. All potentially serious changes were readily explained by the patient's medical history. The patient with the elevated leukocyte count was on corticosteroid therapy. The patient with a markedly elevated prothrombin time had been started on warfarin, and the patient with the elevated potassium value was on peritoneal dialysis for chronic renal failure. One perioperative complication occurred in the group with changed test results but the complication was not related to the changed test value (a patient whose creatinine had risen from 84 to 122 /xmol/L had a perioperative myocardial infarction while having a total hip replacement). Of the 461 most recent previous values that were abnormal as defined by the hospital reference range, 78 (17%; CI, 13% to 20%) exceeded the action range at admission. The frequency of action range abnormalities at admission was significantly higher in those patients whose most recent previous value was abnormal (17%) compared with those whose previous value was normal (0.4%; CI for the difference, 14.9% to 18.1%; P < 0.001). Of 3992 tests done at admission without the benefit of previous tests, 63 (1.6%; CI, 1.2% to 2.0%) exceeded the action range. The frequency of action range abnormalities was significantly higher in those patients who had no previous result (1.6%), compared with those whose previous result was normal (0.4%; CI for the difference, 0.67% to 1.65%; P < 0.001). Because physicians order the hemoglobin, leukocyte count, and platelet count together as a complete blood count and order sodium and potassium determinations as part of an electrolyte panel, we repeated the analysis, counting the complete blood cell count and electrolyte panel each as one test and the prothrombin time, partial thromboplastin time, and creatinine as one test each. When we analyzed the data in this manner, 0.8% of admission tests or test panels contained values that were outside the action range if the previous test or panel result was normal; 18% exceeded the action range if the previous test or panel result was abnormal; and 2.6% exceeded the action range if no previous test result was available. Discussion We sought a practical approach to decrease indiscriminate preoperative screening, hypothesizing that
Figure 1. Top. The percentage of patients being screened at admission with the specified test. If each patient had the specified test, then 100% would be depicted by the bar. Bottom. The percentage of tests that were done both within 1 year before admission as well as after admission for surgery. FIT = partial thromboplastin time.
previous results might safely substitute for preoperative screens in elective surgical patients. We found that almost one half of admission tests duplicated tests done in the year before surgery and that changes in test values from a hospital reference range to values that would affect perioperative management were very rare. Further, the rare abnormal values at admission were predictable from the patient's medical history. Despite studies suggesting that preoperative testing in all patients has a low yield (1-9, 12, 16), a high percentage of patients at our institution are screened before surgery with several laboratory tests; this finding is consistent with other recent studies (17, 18). Educational efforts to decrease test ordering have been largely unsuccessful (5, 6), although administrative restriction of testing has been effective (10, 11). Using previous tests as substitutes may be an effective approach to minimize test use because it may decrease uncertainty in physicians who are uncomfortable about having no test results available before surgery. By substituting previous values, physicians would have results, albeit not current, on all patients who had been tested earlier. Readily accessible previous laboratory data decrease outpatient test ordering by internists (14). In an emergency department, a trend toward decreased test ordering by surgeons was also found when previous tests were accessible (15). We suspect that providing previous laboratory data to surgeons and anesthesiologists in the more stable setting of elective surgery would result in significantly fewer tests. Our findings provide the scientific data necessary to
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12 Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
971
Table 3. Abnormal Laboratory Test
Values and History of Patients Whose Test Results Value
9
Leukocyte count, x 10 IL Prothrombin time, s
Potassium, mmollL Creatinine, /jonol/L
16.3 16.9 13.5 13.1 13.1 13.1 13.1 5.9 130 130 122 122 122
Clinical History Corticosteroid use Warfarin use Prostate cancer Bleeding gums Cirrhosis End-stage renal disease Obstructed voiding Obstructed voiding Obstructed voiding Obstructed voiding
develop a practice guideline regarding preoperative testing. Implementing such guidelines has proved difficult and remains a challenge (19). Developing organizational strategies to minimize repeated testing, such as providing a printout of all recent test results to physicians when their patients are admitted for elective surgery, may decrease repeat testing. Proof that this approach would result in decreased test ordering requires a clinical trial. If physicians seemed unwilling to change test ordering behavior with this approach, external pressures or incentives, such as denial of reimbursement for duplicated laboratory tests if no indication for retesting is noted, may be more effective. Our findings underscore previous work indicating that nonselective preoperative screening yields few abnormal results (4). Even in patients without previous tests who were admitted for elective surgery at our institution, the frequency of values abnormal enough to affect perioperative management was only 1.6%; however, this frequency is probably too high for physicians to abandon screening. We found that 17% of patients whose previous tests were abnormal had admission test values likely to affect surgical management, whereas only 0.4% whose previous tests were normal had results that were abnormal enough to affect perioperative management. This finding is consistent with another recent report indicating that previous values are strongly predictive of subsequent abnormal results (20). Our findings support a practical method to target patients for presurgical screening. Those patients whose previous results are abnormal should be retested; those whose earlier values are normal should not be tested unless obvious reasons for retesting exist. Our findings have a second practical application. Because preoperative evaluations are being shifted from the inpatient to the outpatient setting, long intervals can occur between preoperative assessment and surgery. Our results suggest that preoperative laboratory screening can be done safely at the outpatient level at a considerable interval before surgery and need not be repeated when patients are admitted, if preadmission test results are normal. The longest allowable interval between testing and surgery requires comment. The frequency distribution of the interval between the most recent test and admission test was skewed with more tests done within a few months of surgery. The median interval was 2 months 972
Changed Surgical Procedure Alveoplasty Cataract extraction Cataract extraction Cataract extraction Alveoplasty Inguinal herniorrhaphy Cataract extraction Cataract extraction Transurethral resection of prostate Transurethral resection of prostate Hip arthroplasty Transurethral resection of prostate Transurethral resection of prostate
but 70% were done within 4 months of admission. Although the estimate of the percentage of tests older than 4 months changing to a value likely to affect perioperative management was no different from the overall percentage of tests that changed, the smaller number of previous tests older than 4 months reduces our confidence that the frequency of new abnormalities is rare when the previous test is quite old. For this reason, we recommend that tests more than 4 months old be used with caution. Limitations of this study also require comment. The population studied was elderly and predominantly male; extrapolation of our findings to women and to younger age groups should be done with caution. We know of no reason, however, to suspect that women have more frequent changes in test results and we suspect that younger persons, because of lower prevalence of disease, are even less likely to have significant changes. Computerized laboratory databases with easy access to previous results are not available to all physicians, limiting the generalizability of our findings to other health care settings. Computerized laboratory services are becoming more common, however, and are available nationally in Veterans Affairs medical facilities. We have no data regarding the relative stability of our population's source of health care. Other populations may have a less stable source of health care leading to less accessibility to previous tests; however, studies of health maintenance organizations indicate the annual attrition rate is low (21, 22). Our patient identification procedure may have led to inclusion of some patients who were truly having nonelective surgery (for example, herniorrhaphy for incarcerated hernia or hip arthroplasty for hip fracture). We believe these patients are even more likely to have a changed test value, and, for this reason, feel their inadvertent inclusion would strengthen our conclusion that changes in tests values are rare. Finally, our findings should not be extrapolated to other usual preoperative screening tests (such as urinalysis, chest roentgenogram, or electrocardiogram) without further investigation. The potential savings realized by adopting a strategy of substituting previous tests for preoperative screening tests could be substantial. Assuming that 30 billion dollars is spent annually for preoperative screening (7) and evaluation in the United States and that 80% of tests are for nonemergency surgery, then 24 billion dollars is
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12
Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
spent screening before elective surgery. In Minneapolis, Minnesota, approximately 35% of the charges for the usual preoperative screening tests (that is, in addition to the tests analyzed in this study—urinalysis, electrocardiogram, and chest roentgenogram) is attributable to the tests we studied. Therefore, we estimate that, nationally, 8.4 billion dollars is spent annually on preoperative screening with the tests analyzed in this study. Our data indicate that 40% of patients have previous normal test results; however, we viewed the previous results as individual tests, not as panels often done for preoperative screening. Most physicians order screening complete blood counts and electrolyte panels. (At our institution, only complete blood cell counts and electrolyte panels are available; all other tests are ordered individually.) Abnormalities of any individual test of the panel should prompt reordering, and physicians therefore may reorder the entire panel (although reordering individual tests would suffice). Viewing the data in the context of panels, of 4524 individual tests or test panels ordered at admission, 1651 (36%) previous tests or panels were normal and therefore did not require duplication after admission. An unknown percentage of these patients with these previous test results probably have clear indications for repeat determinations but we estimate this figure to be low and conservatively will assume it to be 10% of all tests in patients having elective surgery. Thus, we estimate that 26% of preoperative screening tests could be eliminated, resulting in a possible annual cost savings of 2.2 billion dollars. Even gross errors in these estimates in a direction that reduces savings would still probably result in an annual savings of more than 100 million dollars. We conclude that a high percentage of patients have tests in the year before elective surgery that could substitute for preoperative screening tests and that changes in test values that would affect surgical management are rare. Clinicians who feel uncomfortable about not screening preoperatively could use the results of the previous tests. If the previous results are normal and there has been no change in the patient's clinical state, then repeat testing is not necessary. Patients having elective surgery could safely be tested preoperatively, if indicated, up to 4 months before the procedure. Strategies to implement practice guidelines based on the findings of this study need to be developed. The actual reduction in preoperative screening achieved by providing physicians with previous tests would be best determined by a clinical trial. Grant Support: By a grant from the Department of Veterans Affairs.
Requests for Reprints: David Macpherson, MD, Section of General Internal Medicine, Pittsburgh Veterans Affairs Medical Center (11 A), University Drive C, Pittsburgh, PA 15240. Current Author Addresses: General Internal Medicine, (11 A), University Drive C, Dr. Snow: Group Health d'Alene, ID 83814.
Drs. Macpherson and Lofgren: Section of Pittsburgh Veterans Affairs Medical Center Pittsburgh, PA 15240. Northeast, 1111 Ironwood Drive, Coeur
References 1. Eisenberg JM, Clarke JR, Sussman SA. Prothrombin and partial thromboplastin times as preoperative screening tests. Arch Surg. 1982;147:48-51. 2. Suchman AL, Mushlin AI. How well does the activated partial thromboplastin time predict postoperative hemorrhage? JAMA. 1986;256:750-3. 3. Rohrer MJ, Michelotti MC, Nahrwold DL. A prospective evaluation of the efficacy of preoperative coagulation testing. Ann Surg. 1988; 208:554-7. 4. Kaplan EB, Sheiner LB, Boeckmann AJ, et al. The usefulness of preoperative laboratory screening. JAMA. 1985;253:3576-81. 5. Turnbull JM, Buck C. The value of preoperative screening investigations in otherwise healthy individuals. Arch Intern Med. 1987;147: 1101-5. 6. Robbins JA, Mushlin A. Preoperative evaluation of the healthy patient. Med Clin North Am. 1979;63:1145-55. 7. Charpak Y, Blery C, Chastang C, et al. Usefulness of selectively ordered preoperative tests. Med Care. 1988;26:95-104. 8. Johnson H, Knee-loli S, Butler TA, Munoz E, Wise L. Are routine preoperative laboratory screening tests necessary to evaluate ambulatory surgical patients? Surgery. 1988;104:639-45. 9. Catchlove BR, Wilson MR, Spring S, Hall J. Routine investigations in elective surgical patients. Med J Austr. 1979;2:107-10. 10. Williams SV, Eisenberg JM. A controlled trial to decrease the unnecessary use of diagnostic tests. J Gen Intern Med. 1986;1:8-13. 11. Schroeder SA, Myers LP, McPhee SJ, et al. The failure of physician education as a cost containment strategy. JAMA. 1984;252:225-30. 12. Roizen MF. The compelling rationale for less preoperative testing [Editorial]. Can J Anaesth. 1988;35:214-5. 13. Maloney TW, Rogers DE. medical technology: a different view of the contentious debate over costs. N Engl J Med. 1979;301:1413-9. 14. Tierney WM, McDonald CJ, Martin DK, Hui SL, Rogers MP. Computerized display of past test results. Ann Intern Med. 1987;107: 569-74. 15. Wilson GA, McDonald CJ, McCabe GP. The effect of immediate access to a computerized medical record on physician test ordering: a controlled clinical trial in the emergency room. Am J Public Health. 1982;72:698-702. 16. Campbell IT, Gosling P. Preoperative biochemical screening [Editorial]. Br Med J. 1988;297:803-4. 17. Mozes B, Lubin D, Modan B, Ben-Bassat I, Gitel SN, Halkin H. Evaluation of an intervention aimed at reducing inappropriate use of preoperative blood coagulation tests. Arch Intern Med. 1989;149: 1836-8. 18. Gortmaker SL, Bickford AF, Mathewson HO, Dumbaugh K, Tirrell PC. A successful experiment to reduce unnecessary laboratory use in a community hospital. Med Care. 1988;26:631-42. 19. Goldman L. Changing physicians' behavior: the pot and the kettle [Editorial]. N Engl J Med. 1990;322:1524-5. 20. Tierney WM, Martin DK, Hui SL, McDonald CJ. Using clinical data to predict abnormal serum electrolytes and blood cell profiles. J Gen Intern Med. 1989;4:375-83. 21. Welch WP. HMO enrollment and Medicaid: survival analysis with a Weibull function. Med Care. 1988;26:45-52. 22. Hennelly VD, Boxerman SB. Disenrollment from a prepaid group plan. Med Care. 1983;21:1154-67.
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12 Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
973
AND
CLINIC
Preoperative Screening: Value of Previous Tests David S. Macpherson, MD; Rita Snow, MD; and Richard P. Lofgren, MD, MPH
Objective: To determine the frequency of tests done in the year before elective surgery that might substitute for preoperative screening tests and to determine the frequency of test results that change from a normal value to a value likely to alter perioperative management. Design: Retrospective cohort analysis of computerized laboratory data (complete blood count, sodium, potassium, and creatinine levels, prothrombin time, and partial thromboplastin time). Setting: Urban tertiary care Veterans Affairs Hospital. Patients: Consecutive sample of 1109 patients who had elective surgery in 1988. Measurements and Main Results: At admission, 7549 preoperative tests were done, 47% of which duplicated tests performed in the previous year. Of 3096 previous results that were normal as defined by hospital reference range and done closest to the time of but before admission (median interval, 2 months), 13 (0.4%; 95% CI, 0.2% to 0.7%), repeat values were outside a range considered acceptable for surgery. Most of the abnormalities were predictable from the patient's history, and most were not noted in the medical record. Of 461 previous tests that were abnormal, 78 (17%; CI, 13% to 20%) repeat values at admission were outside a range considered acceptable for surgery (P < 0.001, frequency of clinically important abnormalities of patients with normal previous results with those with abnormal previous results). Conclusions: Physicians evaluating patients preoperatively could safely substitute the previous test results analyzed in this study for preoperative screening tests if the previous tests are normal and no obvious indication for retesting is present.
Annals of Internal Medicine. 1990;113:969-973. From the University of Minnesota and Minneapolis Veterans Affairs Medical Center, Minneapolis, Minnesota; and the University of Pittsburgh and the Pittsburgh Veterans Affairs Medical Center, Pittsburgh, Pennsylvania. For current author addresses, see end of text.
A reoperative testing is done to detect conditions that could alter perioperative management of the patient. Despite evidence that selective test ordering on the basis of history and physical examination is an appropriate method to evaluate surgical patients (1-9), it is common practice to screen all preoperative patients for unexpected clinically important laboratory abnormalities. The reasons that nonselective screening tests continue to be done are unknown. Physicians ordering preoperative tests may be unfamiliar with the results of studies on preoperative testing; however, educational interventions to decrease test ordering by both medical and surgical housestaff have had limited success (10, 11). Physicians may feel the history and physical examination are too insensitive to detect conditions that affect surgical management, and, therefore, screen all patients in order not to miss unsuspected abnormalities. They may feel nonselective screening decreases the risk of a malpractice suit, despite suggestions that nonselective screening actually increases legal culpability (12). Although the cost per test is low for most preoperative tests, the aggregate cost is substantial (13). Furthermore, nonselective testing frequently results in falsepositive findings that can lead to invasive and dangerous evaluations of abnormal results. It has been estimated that preoperative testing and evaluation cost more than 30 billion dollars annually, of which 60%, or 18 billion dollars, could be saved if tests were done selectively (12). We sought a practical method to decrease presurgical testing of elective surgical patients. Computerized display of previous test results had led to a decrease in test ordering in both an academic primary care medicine clinic (14) and an emergency room (15). If previous test results were readily available to physicians seeing patients before surgery, and test values rarely changed from a normal value to a value likely to affect perioperative management, previous tests might be safely substituted for preoperative screening tests. This might reassure practitioners that rescreening before surgery for occult but clinically important laboratory abnormalities is unnecessary and lead to fewer test orders. We hypothesized that many patients having elective surgery have accessible previous test results and that unexpected changes in test values from a normal range to a value likely to affect perioperative management are rare. To test these hypotheses, we reviewed data on patients who had elective surgery at our institution and determined the frequency of tests done in the year before elective surgery and the frequency of clinically important changes in these values. Methods Hospital discharge records of the Minneapolis Veterans Affairs Medical Center were used to identify patients who had
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12 Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
969
Table 1. Surgical Procedures Reviewed the Value of Preview Tests Surgical Procedure (ICD9 Discharge Code)* Unilateral repair of indirect inguinal hernia (5302) Unilateral repair of direct inguinal hernia (5301) Unilateral repair of inguinal hernia not otherwise specified (5300) Excision of hemorrhoids (4946) Total general surgical procedures Insertion of intraocular lens prosthesis at cataract extraction (1371) Other extracapsular extraction of lens (1359) Phacoemulsification and aspiration of cataract (1341)t Total ophthalmologic procedures Transurethral prostatectomy (602) Radical prostatectomy (605) Insertion or replacement of internal prosthesis of penis (6495) Total urologic procedures Total knee replacement (8141) Release of carpal tunnel (0443) Other total hip replacement (8159) Total hip replacement with use of methyl methacrylate (8151) Other replacement of head of femur (8162) Total orthopedic procedures Other septoplasty (2188) Local excision or destruction of intranasal lesion (2131) Alveoloplasty (245) Excision or destruction of other lesion of external ear (1829) Myringoplasty (194) Total otorhinolarygologic or oral surgery procedures Overall totals
To
Investigate Number of Procedures 52 65 17 26 160 342 3 6 351 207 43 37 287 78 59 52 21 2 212 21 17 23 15 23 99 1109
* Ninth revision. t This category includes patients who had phacoemulsification and aspiration without insertion of lens prosthesis. Patients who had phacoemulsification and aspiration of cataract with lens prosthesis insertion are included in ICD9 1371. t Delete subtotal values (160, 351, 287, 212, and 99).
one of the surgical procedures listed in Table 1. These surgical procedures were selected because they were common, usually elective in nature, generally the sole reason for hospitalization, and not commonly done as an outpatient. A potential disadvantage of this patient identification procedure is that it misses patients who are admitted for elective surgery but discharged without surgery because of abnormal presurgical screening tests. To determine the frequency of this event, we conducted a 1-month audit of patients discharged from the surgery wards in 1989. Of 260 patients discharged from the surgical wards during the time of the audit, no patient was discharged without surgery because of an abnormal preoperative screening test. For each patient identified by discharge record as having had elective/surgery, we recorded laboratory test results (Table 2) from 1 year before admission to the day after admission. These laboratory tests were chosen for study because they are commonly done for preoperative evaluation, have ranges outside of which alteration in management is recommended (4), and are contained in the computerized laboratory database. The results of laboratory tests obtained for inpatients and outpatients are 970
readily available to practitioners on the wards and in the outpatient clinics. Data are stored for most patients for over 1 year. The only laboratory panels done at our institution are the complete blood count and the electrolyte panel (sodium, potassium, chloride, and bicarbonate). All other blood analyses are ordered by individual test. We assumed that tests ordered within the first 2 days of hospitalization were for the purpose of preoperative screening. Patients were excluded from analysis if the admission testing was ordered through the emergency room (suggesting that the procedure was not elective) or if the admission testing was not ordered through one of the surgical wards (indicating that the patient was hospitalized for reasons other than elective surgery and that the testing done during the first 2 days of hospitalization may not have been done for presurgical screening). The location of admission testing was determined through coding on the laboratory test result. The frequency of a clinically important change in test value was determined by counting the number of tests that changed from the hospital reference range to a value that was abnormal enough to affect surgical management, previously defined as the "action limit range" by Kaplan and colleagues (4) (Table 2). Tests were counted in two ways. First, each value from the complete blood count (that is, leukocyte count, hemoglobin, and platelet count), and the electrolyte panel (that is, sodium and potassium) was counted as one test. Second, each complete blood count and electrolyte panel was counted as one test. In both analyses, the prothrombin time, partial thromboplastin time, and creatinine were counted as one test each. We reviewed the medical records of patients who had a clinically important change in test values. A test abnormality was counted as noticed by the physicians or nurses caring for the patient if it was mentioned in the progress notes, discharge summary, or anesthesia record, or if it was circled on the laboratory record. One of the investigators reviewed the medical record to determine the patient's recent medical history and whether any perioperative complications occurred. The 95% confidence intervals for differences in percentages were calculated in a standard manner, and the chi-square test was used to test differences in categorical variables. Results In 1988, 1323 patients had elective surgery of the types listed in Table 1. Of this total, 214 patients were excluded because admission laboratory testing had been ordered through the emergency department or a nonsurgical ward, leaving 1109 patients for analysis. The mean age of the included patients was 66.2 ± 10.6 (SD) years. The frequency of screening at admission for each type of test is indicated in Figure 1, top. If each patient had each test, 8872 tests would have been done. Overall, 7549 (85%) admission tests were done, which indicates that selective screening is probably not being
Table 2. Ranges of Laboratory Test Leukocyte count, x 109IL Hemoglobin, gIL Platelet count, x 109IL Prothrombin time, s Partial thromboplastin time, s Sodium, mmollL Potassium, mmollL Creatinine, /xmol/L
Test Values
Hospital Action Limit Range* Reference Range 4.5-11.0 120-170 130-380 11-13 21-35 136-143 3.7-5.2 70-120
3.0-15.0 100-180 115-800 10-13 22-40 130-150 3.2-5.8 < 120
* Values outside the action limit range would probably result in alteration in management (4).
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12
Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
practiced at our institution. The proportion of patients having laboratory testing in the year before surgery is presented in Figure 1, bottom. Of 7549 admission tests, 3557 (47%) repeated tests that were done in the year before admission for surgery. The median interval from the most recent previous test to admission test was 2 months. Of 3096 most recent previous results that were normal as defined by our hospital's reference range, 13 admission values exceeded the action range (0.4%; CI, 0.2% to 0.7%). These action range abnormalities consisted of 1 elevated leukocyte count, 6 elevated prothrombin times, 1 elevated potassium level, and 5 mildly elevated creatinine levels (Table 3). A review of the records with values that changed showed that most had clinical indications that the result would have changed (Table 3), and only 1 was noted preoperatively. All potentially serious changes were readily explained by the patient's medical history. The patient with the elevated leukocyte count was on corticosteroid therapy. The patient with a markedly elevated prothrombin time had been started on warfarin, and the patient with the elevated potassium value was on peritoneal dialysis for chronic renal failure. One perioperative complication occurred in the group with changed test results but the complication was not related to the changed test value (a patient whose creatinine had risen from 84 to 122 /xmol/L had a perioperative myocardial infarction while having a total hip replacement). Of the 461 most recent previous values that were abnormal as defined by the hospital reference range, 78 (17%; CI, 13% to 20%) exceeded the action range at admission. The frequency of action range abnormalities at admission was significantly higher in those patients whose most recent previous value was abnormal (17%) compared with those whose previous value was normal (0.4%; CI for the difference, 14.9% to 18.1%; P < 0.001). Of 3992 tests done at admission without the benefit of previous tests, 63 (1.6%; CI, 1.2% to 2.0%) exceeded the action range. The frequency of action range abnormalities was significantly higher in those patients who had no previous result (1.6%), compared with those whose previous result was normal (0.4%; CI for the difference, 0.67% to 1.65%; P < 0.001). Because physicians order the hemoglobin, leukocyte count, and platelet count together as a complete blood count and order sodium and potassium determinations as part of an electrolyte panel, we repeated the analysis, counting the complete blood cell count and electrolyte panel each as one test and the prothrombin time, partial thromboplastin time, and creatinine as one test each. When we analyzed the data in this manner, 0.8% of admission tests or test panels contained values that were outside the action range if the previous test or panel result was normal; 18% exceeded the action range if the previous test or panel result was abnormal; and 2.6% exceeded the action range if no previous test result was available. Discussion We sought a practical approach to decrease indiscriminate preoperative screening, hypothesizing that
Figure 1. Top. The percentage of patients being screened at admission with the specified test. If each patient had the specified test, then 100% would be depicted by the bar. Bottom. The percentage of tests that were done both within 1 year before admission as well as after admission for surgery. FIT = partial thromboplastin time.
previous results might safely substitute for preoperative screens in elective surgical patients. We found that almost one half of admission tests duplicated tests done in the year before surgery and that changes in test values from a hospital reference range to values that would affect perioperative management were very rare. Further, the rare abnormal values at admission were predictable from the patient's medical history. Despite studies suggesting that preoperative testing in all patients has a low yield (1-9, 12, 16), a high percentage of patients at our institution are screened before surgery with several laboratory tests; this finding is consistent with other recent studies (17, 18). Educational efforts to decrease test ordering have been largely unsuccessful (5, 6), although administrative restriction of testing has been effective (10, 11). Using previous tests as substitutes may be an effective approach to minimize test use because it may decrease uncertainty in physicians who are uncomfortable about having no test results available before surgery. By substituting previous values, physicians would have results, albeit not current, on all patients who had been tested earlier. Readily accessible previous laboratory data decrease outpatient test ordering by internists (14). In an emergency department, a trend toward decreased test ordering by surgeons was also found when previous tests were accessible (15). We suspect that providing previous laboratory data to surgeons and anesthesiologists in the more stable setting of elective surgery would result in significantly fewer tests. Our findings provide the scientific data necessary to
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12 Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
971
Table 3. Abnormal Laboratory Test
Values and History of Patients Whose Test Results Value
9
Leukocyte count, x 10 IL Prothrombin time, s
Potassium, mmollL Creatinine, /jonol/L
16.3 16.9 13.5 13.1 13.1 13.1 13.1 5.9 130 130 122 122 122
Clinical History Corticosteroid use Warfarin use Prostate cancer Bleeding gums Cirrhosis End-stage renal disease Obstructed voiding Obstructed voiding Obstructed voiding Obstructed voiding
develop a practice guideline regarding preoperative testing. Implementing such guidelines has proved difficult and remains a challenge (19). Developing organizational strategies to minimize repeated testing, such as providing a printout of all recent test results to physicians when their patients are admitted for elective surgery, may decrease repeat testing. Proof that this approach would result in decreased test ordering requires a clinical trial. If physicians seemed unwilling to change test ordering behavior with this approach, external pressures or incentives, such as denial of reimbursement for duplicated laboratory tests if no indication for retesting is noted, may be more effective. Our findings underscore previous work indicating that nonselective preoperative screening yields few abnormal results (4). Even in patients without previous tests who were admitted for elective surgery at our institution, the frequency of values abnormal enough to affect perioperative management was only 1.6%; however, this frequency is probably too high for physicians to abandon screening. We found that 17% of patients whose previous tests were abnormal had admission test values likely to affect surgical management, whereas only 0.4% whose previous tests were normal had results that were abnormal enough to affect perioperative management. This finding is consistent with another recent report indicating that previous values are strongly predictive of subsequent abnormal results (20). Our findings support a practical method to target patients for presurgical screening. Those patients whose previous results are abnormal should be retested; those whose earlier values are normal should not be tested unless obvious reasons for retesting exist. Our findings have a second practical application. Because preoperative evaluations are being shifted from the inpatient to the outpatient setting, long intervals can occur between preoperative assessment and surgery. Our results suggest that preoperative laboratory screening can be done safely at the outpatient level at a considerable interval before surgery and need not be repeated when patients are admitted, if preadmission test results are normal. The longest allowable interval between testing and surgery requires comment. The frequency distribution of the interval between the most recent test and admission test was skewed with more tests done within a few months of surgery. The median interval was 2 months 972
Changed Surgical Procedure Alveoplasty Cataract extraction Cataract extraction Cataract extraction Alveoplasty Inguinal herniorrhaphy Cataract extraction Cataract extraction Transurethral resection of prostate Transurethral resection of prostate Hip arthroplasty Transurethral resection of prostate Transurethral resection of prostate
but 70% were done within 4 months of admission. Although the estimate of the percentage of tests older than 4 months changing to a value likely to affect perioperative management was no different from the overall percentage of tests that changed, the smaller number of previous tests older than 4 months reduces our confidence that the frequency of new abnormalities is rare when the previous test is quite old. For this reason, we recommend that tests more than 4 months old be used with caution. Limitations of this study also require comment. The population studied was elderly and predominantly male; extrapolation of our findings to women and to younger age groups should be done with caution. We know of no reason, however, to suspect that women have more frequent changes in test results and we suspect that younger persons, because of lower prevalence of disease, are even less likely to have significant changes. Computerized laboratory databases with easy access to previous results are not available to all physicians, limiting the generalizability of our findings to other health care settings. Computerized laboratory services are becoming more common, however, and are available nationally in Veterans Affairs medical facilities. We have no data regarding the relative stability of our population's source of health care. Other populations may have a less stable source of health care leading to less accessibility to previous tests; however, studies of health maintenance organizations indicate the annual attrition rate is low (21, 22). Our patient identification procedure may have led to inclusion of some patients who were truly having nonelective surgery (for example, herniorrhaphy for incarcerated hernia or hip arthroplasty for hip fracture). We believe these patients are even more likely to have a changed test value, and, for this reason, feel their inadvertent inclusion would strengthen our conclusion that changes in tests values are rare. Finally, our findings should not be extrapolated to other usual preoperative screening tests (such as urinalysis, chest roentgenogram, or electrocardiogram) without further investigation. The potential savings realized by adopting a strategy of substituting previous tests for preoperative screening tests could be substantial. Assuming that 30 billion dollars is spent annually for preoperative screening (7) and evaluation in the United States and that 80% of tests are for nonemergency surgery, then 24 billion dollars is
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12
Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
spent screening before elective surgery. In Minneapolis, Minnesota, approximately 35% of the charges for the usual preoperative screening tests (that is, in addition to the tests analyzed in this study—urinalysis, electrocardiogram, and chest roentgenogram) is attributable to the tests we studied. Therefore, we estimate that, nationally, 8.4 billion dollars is spent annually on preoperative screening with the tests analyzed in this study. Our data indicate that 40% of patients have previous normal test results; however, we viewed the previous results as individual tests, not as panels often done for preoperative screening. Most physicians order screening complete blood counts and electrolyte panels. (At our institution, only complete blood cell counts and electrolyte panels are available; all other tests are ordered individually.) Abnormalities of any individual test of the panel should prompt reordering, and physicians therefore may reorder the entire panel (although reordering individual tests would suffice). Viewing the data in the context of panels, of 4524 individual tests or test panels ordered at admission, 1651 (36%) previous tests or panels were normal and therefore did not require duplication after admission. An unknown percentage of these patients with these previous test results probably have clear indications for repeat determinations but we estimate this figure to be low and conservatively will assume it to be 10% of all tests in patients having elective surgery. Thus, we estimate that 26% of preoperative screening tests could be eliminated, resulting in a possible annual cost savings of 2.2 billion dollars. Even gross errors in these estimates in a direction that reduces savings would still probably result in an annual savings of more than 100 million dollars. We conclude that a high percentage of patients have tests in the year before elective surgery that could substitute for preoperative screening tests and that changes in test values that would affect surgical management are rare. Clinicians who feel uncomfortable about not screening preoperatively could use the results of the previous tests. If the previous results are normal and there has been no change in the patient's clinical state, then repeat testing is not necessary. Patients having elective surgery could safely be tested preoperatively, if indicated, up to 4 months before the procedure. Strategies to implement practice guidelines based on the findings of this study need to be developed. The actual reduction in preoperative screening achieved by providing physicians with previous tests would be best determined by a clinical trial. Grant Support: By a grant from the Department of Veterans Affairs.
Requests for Reprints: David Macpherson, MD, Section of General Internal Medicine, Pittsburgh Veterans Affairs Medical Center (11 A), University Drive C, Pittsburgh, PA 15240. Current Author Addresses: General Internal Medicine, (11 A), University Drive C, Dr. Snow: Group Health d'Alene, ID 83814.
Drs. Macpherson and Lofgren: Section of Pittsburgh Veterans Affairs Medical Center Pittsburgh, PA 15240. Northeast, 1111 Ironwood Drive, Coeur
References 1. Eisenberg JM, Clarke JR, Sussman SA. Prothrombin and partial thromboplastin times as preoperative screening tests. Arch Surg. 1982;147:48-51. 2. Suchman AL, Mushlin AI. How well does the activated partial thromboplastin time predict postoperative hemorrhage? JAMA. 1986;256:750-3. 3. Rohrer MJ, Michelotti MC, Nahrwold DL. A prospective evaluation of the efficacy of preoperative coagulation testing. Ann Surg. 1988; 208:554-7. 4. Kaplan EB, Sheiner LB, Boeckmann AJ, et al. The usefulness of preoperative laboratory screening. JAMA. 1985;253:3576-81. 5. Turnbull JM, Buck C. The value of preoperative screening investigations in otherwise healthy individuals. Arch Intern Med. 1987;147: 1101-5. 6. Robbins JA, Mushlin A. Preoperative evaluation of the healthy patient. Med Clin North Am. 1979;63:1145-55. 7. Charpak Y, Blery C, Chastang C, et al. Usefulness of selectively ordered preoperative tests. Med Care. 1988;26:95-104. 8. Johnson H, Knee-loli S, Butler TA, Munoz E, Wise L. Are routine preoperative laboratory screening tests necessary to evaluate ambulatory surgical patients? Surgery. 1988;104:639-45. 9. Catchlove BR, Wilson MR, Spring S, Hall J. Routine investigations in elective surgical patients. Med J Austr. 1979;2:107-10. 10. Williams SV, Eisenberg JM. A controlled trial to decrease the unnecessary use of diagnostic tests. J Gen Intern Med. 1986;1:8-13. 11. Schroeder SA, Myers LP, McPhee SJ, et al. The failure of physician education as a cost containment strategy. JAMA. 1984;252:225-30. 12. Roizen MF. The compelling rationale for less preoperative testing [Editorial]. Can J Anaesth. 1988;35:214-5. 13. Maloney TW, Rogers DE. medical technology: a different view of the contentious debate over costs. N Engl J Med. 1979;301:1413-9. 14. Tierney WM, McDonald CJ, Martin DK, Hui SL, Rogers MP. Computerized display of past test results. Ann Intern Med. 1987;107: 569-74. 15. Wilson GA, McDonald CJ, McCabe GP. The effect of immediate access to a computerized medical record on physician test ordering: a controlled clinical trial in the emergency room. Am J Public Health. 1982;72:698-702. 16. Campbell IT, Gosling P. Preoperative biochemical screening [Editorial]. Br Med J. 1988;297:803-4. 17. Mozes B, Lubin D, Modan B, Ben-Bassat I, Gitel SN, Halkin H. Evaluation of an intervention aimed at reducing inappropriate use of preoperative blood coagulation tests. Arch Intern Med. 1989;149: 1836-8. 18. Gortmaker SL, Bickford AF, Mathewson HO, Dumbaugh K, Tirrell PC. A successful experiment to reduce unnecessary laboratory use in a community hospital. Med Care. 1988;26:631-42. 19. Goldman L. Changing physicians' behavior: the pot and the kettle [Editorial]. N Engl J Med. 1990;322:1524-5. 20. Tierney WM, Martin DK, Hui SL, McDonald CJ. Using clinical data to predict abnormal serum electrolytes and blood cell profiles. J Gen Intern Med. 1989;4:375-83. 21. Welch WP. HMO enrollment and Medicaid: survival analysis with a Weibull function. Med Care. 1988;26:45-52. 22. Hennelly VD, Boxerman SB. Disenrollment from a prepaid group plan. Med Care. 1983;21:1154-67.
15 December 1990 • Annals of Internal Medicine • Volume 113 • Number 12 Downloaded From: http://annals.org/ by a Penn State University Hershey User on 05/25/2015
973
