The Cost of Quality Control Procedures In the Clinical Laboratory

Transcription

1 The Cost of Procedures In the Clinical Laboratory JOHN TYDEMAN, PH.D., J. IAN MORRISON, M.A., DAVID F. HARDWICK, M.D. FRCP(C), AND PAUL A. CASSIDY, M.A. Tydeman, John, Morrison, J. Ian, Hardwick, David F., and Cassidy, Paul A.: The cost of quality control procedures in the clinical laboratory. Am J Clin Pathol 77: 1982; The impact of quality control procedures on the workload and the cost of the clinical laboratory during the last decade is explored. Quality control procedures are shown to represent a relatively constant share of test procedures for acute care admissions. The effect of automation on quality control testing in the laboratory has been to reduce the workload units per quality control test and thus to reduce the relative share of total laboratory costs incurred by quality control. The need to assess changes in the cost of quality control testing against any change in quality of laboratory output is emphasized. (Key words: Cost effectiveness; Quality control; Clinical laboratory cost analysis) SUBSTANTIAL INCREASES in the volume of data produced by clinical laboratories 4 ' 5 ' 8,9 " and widespread concern about the variation in the quality, i.e., accuracy, precision and reproducibility of such data, within and between laboratories, 2,20 has led to a growing interest in laboratory quality control procedures. 2,3,7,10,12,20 Whitehead 19 has suggested that a "basic task of the clinical laboratory is to provide reliable data on the composition of specimens obtained from patients as an aid to the diagnosis and treatment of disease (and that) the regular provision of reliable data requires close attention to quality control techniques." Eilers 7 has defined quality control as a management tool consisting of four parts: 1. setting quality standards; 2. appraising conformance to these standards; 3. acting when the standards are not met; and 4. planning for improvements in the standards. These steps form a total quality control system which ensures that data produced meets agreed upon and appropriate levels of precision, accuracy, and reproducibility. Such quality control systems involve a wide range of activities including collection of quality control data, statistical analysis, instrument maintenance, staff education and monitoring, and development of analytical methodologies. 12,13,15 Received March 26, 1981; received revised manuscript and accepted for publication August 24, Supported in part by grants from the Mr. and Mrs. P. A. Woodward Foundation and the Ministry of Health, Government of British Columbia. Address reprint requests to Dr. Morrison: The University of British Columbia, Faculty of Medicine, Department of Pathology 2075 Westbrook Mall, Vancouver, BC Canada V6T 1W5. Institute for the Future, Palo Alto, California, Department of Pathology, University of British Columbia and Department of Pathology, Vancouver General Hospital, British Columbia, Canada The focus of the quality control literature has been primarily a technical and methodological one. Although some writers have encouraged a more critical appraisal of the purpose and effectiveness of quality control, 1 little attention has been paid to the impact of quality control procedures on laboratory workload and costs. This paper will explore the change in both volume and cost of quality control procedures (or tests) for acute care admissions* in a major teaching hospital laboratory during the last decade. These changes will be assessed against a surrogate measure of the technological change (or increase in automation) that has occurred during that time period. For the purpose of this analysis, quality control procedures for acute care admissions are defined as those tests performed in the laboratory for inpatients, over and above tests initiated by physicians. Quality control procedures include tests performed to calibrate instruments or monitor instrument drift; repeat or duplicate tests; and tests performed to assess the validity of analytical techniques or control reagents; not included are tests repeated by order of the physician or research tests. Throughout this paper the terms quality control procedures and quality control tests will be used interchangeably. Once quality control procedures have been defined, Statistics Canada laboratory workload unitsf can be used to measure the contribution of quality control procedures to overall laboratory workload. Quality control procedures are a laboratory-driven dimension * The laboratory under study also performed a wide range of procedures on referred-in specimens and hospital outpatients. These procedures and the share of total laboratory quality control procedures associated with them have been excluded from the analysis by prorating and removing the share of quality control work load units that these procedures incurred. t The term "workload unit" refers to the Statistics Canada' 1 laboratory work-load units allocated to laboratory tests. The unit value of the procedure is a measure of the overall labor input necessary to perform all aspects of a test where one unit equals one min of labor input /82/0500/ American Society of Clinical Pathologists 528

2 Vol. 77 No. 5 COST OF QUALITY CONTROL PROCEDURES Table I. Cost and Utilization for -79, Pathology Laboratory, Vancouver General Hospital 529 Laboratory Cost* ( Millions) (000's) Average Length of Stay Days Tests Performedf Procedures ,381,576 1,309,628 1,429,485 1,539,812 1,758,080 1,773,836 1,950,741 2,213,291 2,267, , , , , , , , , ,363 Percentage increase over period - 255% 64% 70% * Laboratory Costs include labor (professional, technical and clerical) and consumables but do not include capital equipment and building depreciation, indirect cost and other overhead. t Including specimen collection, quality control, repeats, and standards. of laboratory costs in that, although the purpose is to assure the user clinician that laboratory data is reliable, quality control procedures are normally initiated in the laboratory. In this analysis the total variable cost of quality control procedures for acute care admissions (labor and consumables) was derived by pro-rating laboratory costs per acute care admission on the basis of the relative share of laboratory workload units that quality control procedures represent. This assumes that the cost per workload unit is constant for inpatient and quality control tests, which is a valid assumption for the labor cost component since a workload unit is a direct measure of labor input. This assumption is less likely to be valid for certain consumable costs associated with quality control procedures, for example control kits and reagents in clinical chemistry. Therefore, the following analysis explores the effect of changes in the consumables cost of quality control procedures relative to the consumables cost for all inpatient and quality control testing. More detailed cost accounting mechansims such as those suggested by Krieg and colleagues 14 and Muzzillo 16 can provide more accurate measures of quality control procedure costs at the work station level. In this analysis, however, the focus is at the level of the total laboratory. Intensity of Testing During the last decade at Vancouver General Hospital, (VGH), laboratory cost (in current dollars) has increased by approximately 250 per cent; the number of tests performed per acute care admission increased by 65 per cent and the number of tests defined as standards or quality control tests (for acute care admissions) increased by 70 per cent. During this time however the number of acute care admissions has declined by 20% and the average length of stay remained relatively constant (Table 1). During this time the proportion of total tests that were generated for quality control purposes has remained relatively constant (Table 2). Even in the area of clinical chemistry, where the degree of automation is higher and the overall standards of precision and accuracy are greater and more readily administered, there has not been any noticable trend in quality control share of tests performed. On the surface, then, any increase in volume of quality control tests due to automation seems to be offset by the fact that increased mechanization allows for a higher overall volume of tests to be performed. The net effect is that the quality control share of tests has not changed appreciably. This also suggests that any increase in propensity to perform quality control tests because of internal laboratory or external accreditation standards has again been offset by the overall "production" benefits of automation. A more insightful picture can be obtained by analyzing the change in laboratory workload units for quality control tests. Over time, the impact of automation and improved laboratory instrumentation has been: (1) to modify the menu of tests performed, and (2) to permit more tests to be performed for the same expenditure of manpower, i.e., to reduce the complexity of the tests performed. To gauge the extent of this reduced test complexity, it is possible to focus on the Statistics Canada workload units allocated for pathology tests. Workload units per test have been declining over time for total tests and quality control tests (Table 3). A trend

3 530 TYDEMAN ET AL. A.J.C.P. May 1982 Table 2. Tests as a Share of all Tests, -79 Pathology Laboratory, Vancouver General Hospital as a Percentage Share of All Tests in Clinical Chemistry as a Percentage Share of all Clinical Chemistry line was fitted to both series, and its was found that while both sets of workload units per test have been declining over time, those for quality control tests have declined at a faster rate. The slope of the trend line for all tests, U, = t, is significantly less than the slope of the trend line for quality control tests, V, = t at the 99 per cent level, where U, and V, are the numbers of workload units per test for total tests and quality control tests in year t respectively. Thus, while quality control procedures as a share of total tests has remained relatively stable, the workload units required to perform these quality control tests have declined significantly. Cost of Procedures The analysis can now be extended to explore the cost implications of quality control procedures. To provide a constant base for comparison, laboratory costs per acute care admission (in constant or consumer price Table 3. Change in Complexity of Laboratory Tests for -79 Vancouver General Hospital Workload Units per Workload Units per Test for Admission Test index adjusted dollars) have been usedj (Table 4). As part of an overall study of costs in the pathology laboratory, 18 four key factors that drive laboratory costs per acute care admission were identified: 1) wage rates and payments for reagents measured as laboratory costs per FTE minute; 2) productivity of personnel measured as workload units per FTE minute of labor input; 3) impact of automation and technology measured as change in complexity of testing, i.e., workload units per test; and 4) intensity of testing (measured as tests per acute care admission). To some extent the changes in costs of quality control procedures are induced changes with the exception of those due to automation and technology. Wage rates and productivity gains (or losses) are factors relating to the total endeavor, including quality control testing. Intensity of testing, or physician demand, is directed specifically towards testing of patient specimens; the generation of quality control tests is a secondary consequence of this activity. The general findings of the cost analysis may be summarised as follows: (1) total laboratory costs and associated costs of quality control procedures per acute care admission have been increasing in real terms. Quality control costs, however, are not rising as quickly as total laboratory costs. Denoting Y, as the total laboratory costs (in ) per acute care admission and Z, the cost of quality control procedures per acute care admission (in ) in year t, we find that the linear regressions Y, = t and Z, = t are both significant at the 99 per cent level. The pragmatic interpretation is that total laboratory costs per acute care admission in constant dollars have been increasing by 8.5 per annum while costs of quality control procedures per acute care admission have only been increasing by 1.3, or 15 per cent of the total increase in laboratory costs. The second hypothesis, that the slope of the two regression lines are equal, is rejected at the 99 per cent level in favor of the alternative that quality control costs have been increasing at a significantly slower rate than total laboratory costs. (2) The cost of quality control procedures as a share of total laboratory costs has been decreasing significantly over the time period. Setting X, = quality control share of total cost in period t and t The use of the Consumer Price Index as a deflationary factor helps indentify relative changes in cost over time. The use of the general CPI is questionable as a deflator to establish real costs for the laboratory sector, but its use was considered valid for the purpose of relative comparison within the sector.

4 Vol. 77 No. 5 COST OF QUALITY CONTROL PROCEDURES 531 fitting a linear regression over time, we see that X, = t where the linear regression coefficient, is significant at the 99 per cent level. Thus, on the average, the quality control share of total laboratory costs has declined by 1.3 percent per annum. (3) Quality control costs per test have declined over time whereas inpatient (i.e., all other acute care admission tests) have increased in cost. This result follows from the finding that the number of workload units per test for quality control tests have decreased over time at a greater rate than the increase in cost per workload unit (Table 5). Partitioning total acute care tests into quality control procedures and inpatient or non-quality control procedures, it is seen that costs per test have tended to diverge sharply (Fig. 1). Setting Q, equal to the cost of quality control procedures per acute care admission in year t and I t to the cost of other testing procedures per acute care admission and fitting linear regressions over time we find that Q, = t and I, = t suggesting that on average inpatient tests significant at the 99 per cent level are increasing at about 0.03 per test per annum, while the cost of quality control tests are decreasing at about the same rate. The Cost of Consumables The calculated costs of quality control procedures are based on the assumption that the cost per workload unit is constant for quality control and non-quality control tests. Some evidence within the laboratory suggests that the consumable component of cost per workload unit is higher for quality control tests than for all other tests for acute care admissions. In particular, Bercz 6 identifies that for procedures that require considerable quality control, e.g., radio immunoassay, the consumables cost can amount to 50 per cent of the cost of the'testing effort, compared to an average consumables cost share of 25 per cent for the total laboratory. 18 To test the impact of this hypothesis of higher consumables cost it was assumed that the average cost of consumables per workload unit for quality control procedures is 100 per cent higher than for all other acute care admissions tests. The effect of this change on cost of quality control procedures and the quality control share of laboratory costs is shown in Table 6 where hypothesis A assumes constant consumables cost for all workload units and hypothesis B assumes 100 per cent greater consumables cost for quality control workload units. Thus, although this change in assumption generates an increase in quality control costs and an increase in relative share of costs for any given year, the trend still shows that the cost of quality control procedures as a share of all laboratory costs is decreasing. Table 4. Laboratory Cost per Admission -79 Vancouver General Hospital Cost per Acute Care Admission Current Dollar Consumer Price Index All Index, Canada * 178.7* 195.2* Cost per Acute Care Admission Constant Dollars * The CP1 figures correspond to the period April I -March 31 as in Vancouver General Hospital changed from a calendar year to a fiscal year. Net Impact of Procedures Finally, the effects of the four basic cost factors-wage rates, productivity, automation, and intensity-are compared for total laboratory costs and the cost of quality control procedures using a first order difference model. 18 Perhaps the most interesting observation is the fact that the net effects of productivity gains and changes in automation have tended to drive costs down, whereas the increase in rea} wage rates for laboratory personnel over and above average weekly earnings, and real increases in the cost of consumables have tended to drive costs up. As expected, increased demand for tests has had the effect of increasing laboratory costs per acute care admission (Table 7). The most striking observation is the first order effect of automation. In dollar terms the contribution that automation has made to cost of quality control procedures per acute care admission is 65 per cent of the overall contribution that automation has made to the reduction in total laboratory costs. Yet at the same time, Table 5. Cost per Test and Cost per Workload unit for, Pathology Laboratory, Vancouver General Hospital Cost per Workload Unit () All Tests Cost Per Test () Inpatient Tests Quality Control Tests

5 532 TYDEMAN ET AL. A.J.C.P. May "o 160 Q R o> o ^ 1.20 o u 1.00 \ q ^» Cost/in ost/inpatient lest \,,.o Cost/q.c. test V ' i i i FIG. 1. Cost per test for inpatient and quality control tests -. the quality control share of total costs is less than 25 per cent (Table 6) and the quality control share of the change in total costs is only 17 per cent. This shows the relatively greater effect automation has had on the costs of quality control testing as opposed to laboratory testing as a whole. When the costs of quality control procedures are adjusted for a 100 per cent above average consumables cost (Hypothesis B, Table 7), the findings are not substantially altered. The effect of automation is, however, even more striking: 78 per cent of the overall contribution of automation to reduced laboratory cost is attributable to quality control procedures. Conclusion Quality control procedures represent a substantial proportion of laboratory work load and costs. It is estimated that total laboratory costs in the U. S. exceed 20 billion. This suggests that the cost of quality control procedures is in the order of 6-8 billion. The reliability of laboratory generated data depends on an appropriate level of quality control in the laboratory, and with increasing volumes of data production, an increase in relative volume of quality control procedures might have been expected. Equally, increasing automation and mechanization of laboratory procedures might have been expected to reduce the relative share of quality control testing. If either of these trends have occurred, their effects appear, from this analysis, to be self-cancelling as the quality control share of all tests has remained relatively constant. This analysis also shows that automation in the laboratory has tended to drive down the complexity of quality control tests and to reduce the relative cost of quality control tests per acute care admission. Overall, the quality control share of total laboratory testing has remained constant, but when transformed into workload units or into relative costs, the quality control component has tended to decline. The effects noted here are for a very large laboratory where increments in work load would be expected to reduce the marginal cost of quality control. In a smaller laboratory, however, where batch sizes are smaller and possibly suboptimal for quality control procedures, in- Table 6. Cost of Procedures Under Varying Hypotheses for Cost of Consumables, -79, Vancouver General Hospital Hypotheses A* Hypotheses Bt Laboratory Costs Per () Estimated Costs of Procedures Per () Share of Laboratory Costs % Estimated Costs of Procedures Per Acute () Share of Laboratory Costs * Assumes consumables cost per workload is constant for all types of tests. } Assumes consumables cost per workload unit for quality control tests is 100 per cent higher than for all other tests.

6 Vol. 77 No. 5 COST OF QUALITY CONTROL PROCEDURES 533 Table 7. Net Effects on Cost per Admission Due to Changes in Cost Factors* Vancouver General Hospital (Contribution of Changes in Factors'!") Net Change in Cost per Acute Admission Wage Rates Productivity Automation Intensity of Testing Hypothesis A% Total Tests Tests Tests as a (%) of All Tests Hypothesis fl Tests Tests as a (%) of All Tests r' * All costs are in constant terms. t Figures apply to net changes -79. t Assumes consumables cost per workload unit is constant for all types of tests Assumes consumables cost per workload unit for quality control tests is 100 per cent higher than for all other tests. crements in workload would probably yield an even greater relative reduction in the marginal cost of quality control testing. In small or large laboratories the question of whether the amount of quality control activity is excessive, optimal, or even adequate is difficult to answer. Further analysis is required to determine the degree of reliability required by the laboratory users and to ascertain whether the quality of laboratory output fails to meet, fulfills, or even exceeds these user requirements. Such an analysis may permit even more gains in laboratory efficiency and effectiveness through modification of quality control protocols. References 1. Annino JS: What does laboratory "quality control" really control? N Engl J Med ; 299(20): Aronsson T, Bjornstad P, Johansson SG, Leskinen R, Raabo E, De Verdier C-H: Inter-laboratory quality control with investigation of different methodological characteristics. Scand J Clin Lab Invest ; 38(0: Bell WN: Quality control in the automated central laboratory. J La State Med Soc ; 125(9): Benson ES: Clinical laboratory utilization and the cost of health care. Postgrad Med ; 63(4): Benson ES: Strategies for improved use of the clinical chemistry laboratory in patient care. In Benson ES, Rubin M, eds. Logic and economics of clinical laboratory use. New York: Elsevier/ North Holland Biomedical Press, : Bercz JP: Cost effectiveness of esoteric testing. Laboratory Management ; 13: Eilers PJ: Total quality control for the medical laboratory. South Med J 1969; 62: Grams RR: Clinical laboratories: influence of increased productivity and new laboratory instruments on health care. South Med J ; 69: Griner P, Liptzin B: Use of the laboratory in a teaching hospital. Ann Intern Med ; 75: Hamill RD: Quality control in the laboratory. Clin Toxocol ; 12(2): Hardwick DF, Morrison JI, Tydeman J, Cassidy PA, Chase WH: Laboratory costs and utilization: a conceptual framework for analysis and policy design. J Med Educ 1981; 56: International Federation of Clinical Chemistry. Approved recommendation () on quality control in chemistry. Part I: General principles and terminology IFCC Section () No. 3. In: Clinica Chimica Acta ; 98(1): 129 F-144F 13. Koenig AS, Day JC, Jollife LS, Sodeman TM, Alpert NL: Instrument quality assurance for test result quality and laboratory operation. Pathologist 1980; 34(3): Krieg AF, Israel M, Fink R, Shearer LK: An approach to cost analysis of clinical laboratory services. Am J Clin Pathol ; 69(5): Lott J A: Laboratory personnel: the most important aspect of quality control. Med Instrum ; 8: Muzzillo MJ: How to monitor lab costs. MLO Jan; 8: Statistics Canada: Canadian Schedule of unit values for clinical laboratory procedures, edition 18. Tydeman J, Morrison JI, Cassidy PA, Hardwick DF, Chase WH: Analyzing the factors contributing to raising laboratory costs. Working paper, p 91, Institute for the Future, Palo Alto, California, Whitehead TP: Quality control in clinical chemistry. New York, John Wiley & Sons, 20. Whitehead TP: Quality control techniques in laboratory services. Br Med Bull ; 30(3):