Background.—In 1986 and 1989, the Centers for Disease Control and Prevention sponsored institutes on Critical Issues in Health Laboratory Practice. It was noted during the institutes that physician's office laboratories were a rapidly emerging site for clinical laboratory testing, yet no comprehensive data were available regarding the practice of clinical laboratory medicine in physician's office laboratories. As a mechanism to begin addressing this void, the Centers for Disease Control and Prevention added questions on clinical laboratory practice to the National Ambulatory Medical Care Survey, a national probability sample of ambulatory care provided by office-based physicians. Data were collected for survey years 1989, 1991, 1993, and 1994.
Methods.—Each survey was conducted among a nationally representative, random sample of office-based physicians who provide ambulatory patient care. Sample physicians were enlisted using both mail and telephone contacts. Clinical laboratory data were obtained via telephone by trained field representatives. Weighted univariate and multivariate analyses were performed on responses from each of the 4 survey years. Analyses were repeated after combining survey responses from years 1989 and 1991 and 1993 and 1994 as representative of physician's office laboratory practices before and after implementation of the Clinical Laboratory Improvement Amendments of 1988 (CLIA ’88) final rule in 1992.
Results.—Quality laboratory practice indicators showed significant increases during the study interval, with implementation of the CLIA ’88 final rule in 1992 playing a pivotal role. Relative to 1992, enrollment in proficiency testing programs increased from 32.4% to 52.7% (P < .001), use of daily quality control samples increased from 79.2% to 89.0% (P < .001), and use of daily quality control with written instructions for action following a questionable quality control result (quality control with action step documentation) increased from 62.6% to 77.2% (P < .001). The presence of a medical technologist or technician in the office laboratory was also significantly and independently associated with each of the quality indicators. Although the percentage of physician's offices performing on-site testing decreased from 56% to 45% during the survey interval, overall testing volume appeared unchanged.
Conclusions.—The quality of clinical laboratory practice in physician's office laboratories improved during the study interval (1989–1994) as measured by the quality indicators used in the study. The association of this improvement with implementation of the CLIA ‘88 final rule and the presence of a trained laboratory professional in the testing site indicate the importance of minimum practice standards and professional expertise in ensuring use of quality laboratory practices. Overall test volume appeared to be stable despite a decreased proportion of physician's offices at which on-site testing was performed.
In 1996, Americans made more than 734 million visits to physician's offices in the United States, with approximately one third of those visits involving 1 or more laboratory tests.1 Despite a recent plateau in physician's office laboratory (POL) testing volume,2 there were between 571 million and 899 million laboratory tests performed in POLs in 1996 (Steindel et al, National Inventory of Clinical Laboratory Testing Services, 1997). The expansion of the POL testing market dates back to the 1970s, when the technological advances of miniaturization and the availability of computer chips enabled the development of small, affordable analyzers designed specifically for POLs. The convenience of on-site testing and its economic practicability led to a boom in testing in the physician's office.3 As POL testing volume grew, a parallel interest emerged in the laboratory community regarding the quality of testing in this largely unregulated group of testing sites. Limited, regional regulation of POLs evolved as some state legislatures instituted governmental oversight of POLs. However, the content and scope of the regulations varied from one state to another, as did the sites covered by the regulations.4
In 1986 and 1989, the Centers for Disease Control and Prevention (CDC) sponsored 2 institutes on Critical Issues in Health Laboratory Practice.5,6 During these forums, leaders of the clinical laboratory community from both the public and private sectors noted that despite the fact that POLs were a rapidly emerging site for clinical laboratory testing, no comprehensive data were available regarding the quality of the practice of clinical laboratory medicine in POLs. As a mechanism to begin addressing this void, the CDC added questions on clinical laboratory practice to the National Ambulatory Medical Care Survey (NAMCS). Since NAMCS is a national probability sample survey of ambulatory care provided by office-based physicians, the survey provided an ideal sampling frame and an opportunity to collect data about POL practices.
The questions to be added to the survey were developed in 1987 for implementation in the 1989 survey. The quality indicators for the study were selected based on accepted clinical laboratory practice principles and included the topics of proficiency testing (PT), quality control (QC) and quality assurance (QA), and personnel qualifications. The timing of the implementation of the laboratory practice survey questions became serendipitous when Congress enacted the Clinical Laboratory Improvement Amendments of 1988 (CLIA ’88). Based on the principle of site-neutral test quality, CLIA ’88 brought all sites that perform clinical laboratory testing under a universal set of minimum practice standards for the first time in the nation's history. Since the CLIA ’88 final rule was not implemented until 1992, the NAMCS provided a unique opportunity to obtain data about clinical laboratory practice in POLs both before and after CLIA ’88 implementation, thus allowing some measure of the impact of CLIA ’88 on clinical laboratory practice in this milieu. This study reports the findings from the NAMCS laboratory practice questions for survey years 1989, 1991, 1993, and 1994. Also reported are findings of a composite analysis in which survey responses from years 1989 and 1991 and 1993 and 1994 were combined to represent POL practices before and after implementation of the CLIA ’88 final rule in 1992. The study examines other variables that may have affected clinical laboratory practice during the study interval in an effort to discern the impact of CLIA ’88 independent of other factors in the POL milieu.
The data analyzed in this report were collected in 4 National Center for Health Statistics NAMCSs conducted from March 1989 through December 1994. During this period, the surveys collected information about testing practices in respondents' POLs.
The surveys were conducted using a 3-stage sampling design.7 The first stage involved probability sampling of counties (or county equivalents) used in the National Health Interview Survey. The second stage involved probability sampling of office-based physicians (principally engaged in patient care and not federally employed) selected from the master files of the American Medical Association and the American Osteopathic Association. All eligible physicians were stratified into 15 medical specialties. Sampling weights for the aforementioned strata were computed by the National Center for Health Statistics based on the population sizes in the American Medical Association and American Osteopathic Association master files. In the third stage of the sampling plan, the physician sample was divided into 52 random subsamples, each of which was randomly assigned to a week of the year. During the assigned week, data were collected on a systematic random sample of patient visits to the respondent physician. Both mail and telephone contacts were used to enlist participants. Physicians who did not respond initially were contacted again by mail or telephone. The data were collected from either the sampled physicians or designated members of their staffs.
All variables on the surveys related to laboratory testing were included in the statistical analysis. The information gathered included questions regarding testing personnel, test volume, enrollment in PT, and QC practices for 19 specific laboratory tests. The variables are listed in Table 1. The 6 tests most frequently cited as being performed on site (urinalysis, urine pregnancy test, hemoglobin, hematocrit, glucose, and occult blood) can all be performed using procedures that are categorized by the CLIA ’88 regulations as either waived or under the purview of a provider-performed microscopy certificate. The remaining 13 tests specifically named on the survey are all categorized as either moderately or highly complex test procedures. To note this distinction, the tests were grouped into 1 of 2 categories: “simple” if they met the former description and “complex” if they met the latter description. On analyzing each of the 4 surveys, we noted a natural division between the responses from the surveys before 1992 and after 1992. Since 1992 was also the year the CLIA ’88 final rule was implemented, this categorization was used in subsequent rounds of analysis.
All population estimates and hypothesis tests were computed using the weighting scheme noted herein. For univariate analyses, percentages were compared using normal theory tests of proportions, and medians were compared using Wilcoxon signed rank test. Logistic regression techniques were used to construct all multivariate models, yielding adjusted odds ratios. All statistical computations were performed using SAS statistical software.8 Hypothesis tests with P values less than .05 were considered to be statistically significant.
In 1989, 1991, 1993, and 1994, there were 1432, 1869, 1616, and 1474 eligible survey respondents, respectively. The 4 most frequent physician specialties represented in the surveys were general and family practice (18%), internal medicine (14%), obstetrics and gynecology (9%), and pediatrics (8%). The order and approximate relative frequencies were the same before and after 1992 and closely mirror the specialty distribution of the physician population.
The percentage of physicians with on-site testing was 55% before 1992 and 46% after 1992. Of the physicians who were in solo practice, 48% performed on-site laboratory testing before 1992 and 37% did so after 1992. By specialty, physicians who most frequently reported on-site testing were urologists (78% and 83% before and after 1992, respectively), general and family practitioners (75% and 74% before and after 1992, respectively), pediatricians (80% and 75% before and after 1992, respectively), obstetricians and gynecologists (73% and 69% before and after 1992, respectively), and internists (69% and 60% before and after 1992, respectively). Family and general practitioners, internists, pediatricians, and obstetricians and gynecologists comprised 48% of all surveyed physicians but accounted for 70% of all physicians performing on-site laboratory testing.
Laboratory Testing and QA Practices
The percentage of physicians having any given test performed on site ranged from a high of 51% (urinalysis) to a low of 8% (rheumatoid arthritis latex agglutination) before 1992 and from 38% to 6% (for the same 2 tests) after 1992 (Table 2). Of the physicians who reported on-site performance of at least 1 of the 19 tests listed on the NAMCS, 21% reported a test menu limited to simple procedures before 1992 and 29% after 1992.
Physicians, nurses, medical assistants, medical technicians, medical technologists, physician assistants, and others all performed on-site testing. Before 1992, respondents used the following personnel to perform on-site testing in the percentages indicated: 30% nurses, 27% physicians, 20% medical technicians, 18% medical assistants, and 12% medical technologists. Similar results were observed after 1992 when the following personnel utilization pattern was reported: 27% nurses, 25% physicians, 20% medical technicians, 23% medical assistants, and 9% medical technologists.
Among respondents who reported on-site performance of at least 1 of the 19 tests listed on the NAMCS, daily QC (which was used in this report as an abbreviated notation for QC performed each day of patient testing) was performed with 79% of tests before 1992 and 89% after 1992. The use of daily QC, by test, ranged from 59% (urinalysis) to 94% (rheumatoid arthritis latex agglutination) before 1992 and from 78% (urinalysis) to 98% (prothrombin time) after 1992. As a general assessment of daily QC use before and after 1992, the number of tests where at least 90% of respondents indicated daily QC use was counted. Before 1992, this 90% threshold was exceeded for 5 of the 13 complex tests, whereas this threshold was exceeded for 10 of the complex tests after 1992. The 3 tests for which daily QC use was reported by less than 90% of respondents after 1992 were β-streptococcus rapid tests (86%), urine colony counts (85%), and gonorrhea cultures (84%).
Respondents were asked if the laboratory had written instructions to follow if QC suggested an error. A variable termed QC with action step documentation was created for purposes of this analysis if, for a given test or analyte, respondents indicated daily QC use along with written instructions to follow if QC suggested an error. Before 1992, 63% of the on-site tests had QC with action step documentation; after 1992, this percentage increased to 77% (Figure). Quality control with action step documentation varied by the specific test in question. Before 1992, QC with action step documentation ranged from 45% (urinalysis) to 85% (creatinine); after 1992, the range extended from 68% (urinalysis) to 94% (rheumatoid arthritis latex agglutination).
Association With Quality Improvement Practices
Significant changes in laboratory testing practices were observed relative to 1992 and specifically involved the type of medical practice, the professional training of the analyst, enrollment in PT, use of daily QC, and QC with action step documentation (Table 3). Comparing the rates before and after 1992, enrollment in PT increased from 32% to 53%, daily use of QC increased from 79% to 89%, and QC with action step documentation increased from 63% to 77%. All these changes were highly significant (P < .001). No significant change was noted before and after 1992 in the percentage of medical technicians or medical technologists performing on-site laboratory testing.
By univariate analysis, factors that were positively associated with enrollment in a PT program were as follows: performance of testing by a medical technician or technologist, performance of complex tests, nonsolo medical practice, timing of the survey relative to 1992, and the number of tests performed daily. Physicians enrolled in a PT program had a median of 20 tests performed daily, whereas those not enrolled had a median of 8 tests (P < .001). The PT enrollment rate also varied by the medical specialty of the practitioner (Table 4). After controlling for the confounding effect of medical specialty in a multivariate logistic regression model, all these factors were significantly and independently associated with enrollment in a PT program (Table 5).
The same factors associated with enrollment in a PT program were also significantly and independently associated with all the measures of QC and QA examined. The measures were as follows: daily use of QC for all 19 tests, the 6 simple tests, and the 13 complex tests (Tables 6 and 7) and QC with action step documentation for all 19 tests, the 6 simple tests, and the 13 complex tests (Tables 8 and 9).
The results of this study clearly demonstrate a change in clinical laboratory practice in the physician's office setting during the study period. There are several possible explanations for this observation. It is possible the increases in the utilization of the quality indicators of this study reflect a difference in the composition of POL practices over time (ie, that attrition led to a relative decrease in POLs with less than standard quality laboratory practices). However, inspection data and enrollment information from the Health Care Financing Administration neither clearly support nor refute this possibility. Another possible explanation for the observations of this study is that implementation of the CLIA ’88 regulations had a positive impact on clinical laboratory practice in the physician office setting as evidenced by the quality indicators for the study. This is an important finding, since the physician office setting is an important site for clinical laboratory services. During 1996, there were an estimated 892 million ambulatory care visits made in the United States, with visits to physician's offices predominating at 82%.1 In 1996, approximately one third or 294 million physician office visits were associated with 1 or more clinical laboratory procedures. The proportion of office visits associated with laboratory testing has remained stable during recent years.2 With more than 166 000 POLs of approximately 387 000 physician offices within the United States,9–11 it is apparent that POL testing is an important component of the ambulatory medical care delivery system in the United States. The importance of the role of POL testing in ambulatory care delivery led to interest in the quality of laboratory services provided in the physician office milieu and was the impetus for this study.
As with any complex, dynamic environment, the POL milieu is a challenge to study. The forces that influenced the POL environment during the study interval were multiple, diverse, and too numerous to be comprehensively itemized; however, several of the more prominent factors are noted here as examples. From a socioeconomic standpoint, physician offices were influenced by the increasing market share of managed care in the health care delivery system. Even offices not directly involved in managed care contracts are often indirectly influenced by the market forces created by the presence of managed care in a given market. The past decade has seen a shift toward multipractice mergers and consolidation of health care services.12–14 Both of these factors, managed care and the consolidation trend, are relevant to the structure and function of POLs. For example, managed care contracts can include stipulations regarding laboratory testing, often requiring testing to be sent to a specific laboratory with which the parent organization has contracted for laboratory services. Similarly, practice mergers either may lead to consolidation of laboratory services into a large, private laboratory or may involve contracting with an outside laboratory. The economic forces that influenced POLs during the study interval were significant in a broad sense even if not influential in specific cases.
Three new regulatory forces were also a prominent influence on POLs during the study interval. In 1987, the Occupational Safety and Health Act15 regulations were modified to expand coverage to all industries where employees are exposed to hazardous chemicals, including POLs. This expansion of coverage required POLs to establish a Hazard Communication Program and provide hazardous materials training for all employees.16 In addition, the Omnibus Reconciliation Act of 198917 included a provision known as Stark I, which barred self-referral arrangements for clinical laboratory services under the Medicare program. Prompted by a study conducted by the US Department of Health and Human Services,18 the law banned physicians from making referrals to laboratories for services for which Medicare would otherwise pay if the physician had an ownership or investment interest in the laboratory. The implications for large POLs in which several physicians or practices had an interest were significant.
The third piece of legislation directly affecting POLs during the study interval was CLIA ’88.19 The impact of CLIA ’88 on POLs was substantial,20–24 since CLIA ’88 established regulation of a previously largely unregulated sector of the clinical laboratory community. This study clearly demonstrates that the newly regulated community responded to the quality standards described in the CLIA ’88 regulations swiftly and in substantial numbers, yielding significant improvements in this key area of laboratory practice.
The POLs service a patient population that is predominantly characterized by its vast diversity in terms of age, type of morbidity, and severity of illness. Laboratory testing plays an integral role in diagnosis, screening, and monitoring the health of a diverse ambulatory care population. Testing specimens from such a diverse population requires a variable degree of analytical accuracy. Laboratory screening of healthy patients, for whom testing is part of an overall health maintenance and prevention program, does not require the same degree of analytical accuracy that is necessary for patients who are chronically and/or critically ill. This is largely due to differences in decision making and action thresholds in different patient populations. It is, therefore, in the best interest of both patients and physicians to deliver test results of a sufficient level of quality to meet all patients' needs a priori.25–29
Many factors contribute to the quality of a laboratory result. In the parlance of clinical laboratory science and in the legislative use of the term, quality refers to the accuracy and reliability (precision) of laboratory testing. However, there are clearly other important factors that contribute to a quality laboratory test, such as convenience, timeliness, validity, accessability, and report clarity. The quality indicators used in this study are derived from the more focused definition of quality but have definite applicability to many of the other factors that contribute to the overall quality of the clinical laboratory's product.
Measurement accuracy is defined by the NCCLS as “the closeness of the agreement between the result of a measurement and a true value of the measurand.”30 A primary mechanism for assessing clinical laboratory accuracy is PT. The quality practice indicator used in this study to signify the active monitoring of overall laboratory accuracy was participation in PT. Proficiency testing is an external QC tool where simulated patient samples are tested by participating laboratories. Individual laboratory performance is assessed by comparison with either the collective performance of all participants or with a value determined by an accepted, definitive method. The scope of PT is limited to the analytic portion of the total testing process. Proficiency testing does not assess some important preanalytic and postanalytic steps that contribute to the overall accuracy of a reported test result.31–37 Further, although PT is relatively specific with regard to detecting systematic laboratory inaccuracies,31,33,38–41 it is not highly sensitive.39–44 Despite the limitations, however, PT is an accepted external measure of laboratory accuracy38, 45,46 and has been shown in certain settings to be a good indicator of routine laboratory performance.47,48
Laboratory precision refers to “the closeness of agreement between a series of measurements, under specified conditions, of a substance or biological product.”49 The laboratory's primary surveillance system for precision is internal QC.50–53 The QC materials are tested as patient samples. The QC results are subsequently compared with an established acceptable range as a measure of the degree to which the test system (ie, the equipment, the reagents, and the operator) is within control limits. The quality practice indicator used in this study to signify the active monitoring by the laboratory for systematic sources of error in the test system was use of daily QC. An unacceptable QC result should lead to some action on the part of the test system operator. Written documentation of possible action steps are part of a comprehensive QA program. For this study, use of daily QC and written documentation of action to be taken if QC results suggest an error (QC with action step documentation) was used as the quality indicator signifying an active QA program in the laboratory. An active QC program is considered essential to the quality practice of clinical laboratory medicine, 28,50,52,54–57 as is the use of a dynamic, comprehensive QA program.58–65
Examination of all 3 laboratory quality indicators shows marked increases in use after 1992. This type of response is not unprecedented in the medical community. In 1992, Congress enacted the Mammography Quality Standards Act “to establish the authority for the regulation of mammography services and radiological equipment.”66 Interim regulations went into effect in 1994 and in many ways mirrored the CLIA ’88 regulations in terms of establishing QA and QC program standards, equipment standards, personnel requirements, and patient test management system standards.19,67 By 1996, the Food and Drug Administration reported marked improvement in the quality of mammography facilities as measured by compliance with quality standards implemented in the Mammography Quality Standards Act regulations.68 What is demonstrated in the response of the medical community to the Mammography Quality Standards Act regulations and what we have shown with this study that examines the medical community's response to the CLIA ’88 regulations are as follows: (1) regulations are successful in increasing compliance with minimum quality standards, (2) the medical community responds relatively quickly to the implementation of such regulations, and (3) trained personnel facilitate compliance with minimum quality standards. The association noted in this study between personnel with specific education or training in the clinical laboratory sciences and various indicators of laboratory quality has been noted by other investigators.69–71
This study does not directly address some important aspects of quality, which were mentioned earlier in this discussion. However, if one accepts the tenet that newly focused attention on quality standards reflects an increased awareness of the role and importance of quality management, it would follow that improvement in the quality indicators used in this study may represent actual improvement in other quality factors or at least represent an open door to addressing other aspects of quality. Another issue that is important to consider and balance with attempts at quality improvement is access to services. Although this study does not directly address access, the stable testing volume during the study interval is an indication that overall access to services was not substantially affected by implementation of the CLIA ’88 regulations.
CLIA ’88 was implemented to improve the quality of clinical laboratory practice and thereby contribute to improved patient care. This study demonstrates that the frequency of laboratory quality indicators increased subsequent to CLIA ’88 implementation. The goal of allied health regulatory programs is the same as the ultimate goal of health care providers: to ensure that quality health care that meets the needs of each patient is delivered consistently and reliably. To that end, this study demonstrates that clinical laboratory regulation is making a contribution to this ultimate goal.
We acknowledge and thank Mr Fritz D. Robinett for his tireless assistance with the preparation of the manuscript.