581488

research-article2015

JLAXXX10.1177/2211068215581488Journal of Laboratory AutomationIalongo et al.

Original Report

Total Automation for the Core Laboratory: Improving the Turnaround Time Helps to Reduce the Volume of Ordered STAT Tests

Journal of Laboratory Automation 1­–8 © 2015 Society for Laboratory Automation and Screening DOI: 10.1177/2211068215581488 jala.sagepub.com

Cristiano Ialongo1, Ottavia Porzio2, Ilio Giambini1, and Sergio Bernardini1,2

Abstract The transition to total automation represents the greatest leap for a clinical laboratory, characterized by a totally new philosophy of process management. We have investigated the impact of total automation on core laboratory efficiency and its effects on the clinical services related to STAT tests. For this purpose, a 47-month retrospective study based on the analysis of 44,212 records of STAT cardiac troponin I (CTNI) tests was performed. The core laboratory reached a new efficiency level 3 months after the implementation of total automation. Median turnaround time (TAT) was reduced by 14.9±1.5 min for the emergency department (p < 0.01), reaching 41.6±1.2 min. In non–emergency departments, median TAT was reduced by 19.8±2.2 min (p < 0.01), reaching 52±1.3 min. There was no change in the volume of ordered STAT CTNI tests by the emergency department (p = 0.811), whereas for non–emergency departments there was a reduction of 115.7±50 monthly requests on average (p = 0.026). The volume of ordered tests decreased only in time frames of the regular shift following the morning round. Thus, total automation significantly improves the core laboratory efficiency in terms of TAT. As a consequence, the volume of STAT tests ordered by hospital departments (except for the emergency department) decreased due to reduced duplicated requests. Keywords total laboratory automation, turnaround time, cardiac troponin I, STAT, core laboratory

Introduction Since its advent in the mid-1960s, the automation of clinical chemistry processes has changed the face of the laboratory in the fashion of a true industrial revolution.1,2 The ability to efficiently perform repetitive tasks, the possibility to accomplish simple actions combined in elaborate series, and finally the ease of simultaneously managing multiple samples on different instruments are the obvious reasons of such affirmation. Since then, its growing success has pushed automation to an even higher level, culminating in a functional continuum known as total laboratory automation (TLA).3,4 In this kind of “automated pipeline” that integrates the pre-analytics (check-in, sorting, centrifugation, and aliquoting) and the postanalytics (storage and disposition) with the analytics senso strictu, every operation is managed by a software system that also mediates the intervention of the human operator. The term TLA tends to denote a virtualized management of the actual processing workflow, rather than a fully mechanized ensemble devoid of human staff. The transition to TLA technology (or, rather, philosophy) demands a considerable effort in terms of economic resources, infrastructural interventions, and a staff’s cultural leap.5–7

This makes it of paramount importance to evaluate its impact on the laboratory system. Computer-based simulations can aid in forecasting the most appropriate system setup8,9 before attempting any real implementation. However, the actual effectiveness of the system can be assessed only through factual evidence gathered in the field. In a time when spending reviews have become a constraint for public health, the justification of such efforts to achieve efficiency improvements through TLA by concrete results seems to be mandatory.10 The most relevant outcome shown in the recent past has been represented by the relationship between turnaround time (TAT) reduction and improved critical care services.11,12 However, current financial demands 1

Department of Laboratory Medicine, Tor Vergata University Hospital of Rome, Rome, Italy 2 Department of Experimental Medicine and Surgery, University of Rome Tor Vergata, Rome, Italy Received Nov. 10, 2014. Corresponding Author: Dr. Med. Cristiano Ialongo, Department of Laboratory Medicine, Tor Vergata University Hospital of Rome, Viale Oxford 81, 00135 Rome (RM), Italy. Email: [email protected]

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

2

Journal of Laboratory Automation 

Table 1.  Inclusion Criteria. 1. TAT: not less than the minimum sufficient time to complete the workflow (check-in, sorting, centrifugation, delivery to the chemical analyzer, assaying, and validation of the result) 2.  Check-in: done in the ordinary morning shift (from Monday to Saturday, from 6:00 to 13:59, no holidays) 3.  Test result: within the analytical test range (no dilution, 0.015–40 ng/mL) 4.  Operation: only carried out in strict accordance with standard operating procedures (e.g., not a CTNI test ordered after sample delivery to the laboratory, or analyzer operated as stand-alone because of a system’s failure) To minimize the contribution of confounding data to the statistical analysis, a set of validity requirements was stated to include only meaningful CTNI test records. Thereby excluded were those records that could have biased the estimation of percentiles. CTNI, cardiac troponin I, TAT, turnaround time.

require looking with almost the same emphasis at results of an economic nature. In the light of the discussion so far, the present article attempts to investigate the effectiveness of the implementation of TLA on the core laboratory of the Tor Vergata University Hospital of Rome, Italy, a tertiary care structure with 380 beds and 3.5 million clinical chemistry tests per year. Particularly, our aim was also to investigate the relationship between a more efficient (automated) laboratory service and the number of STAT tests ordered.

Materials and Methods Data The records of cardiac troponin I (CTNI) test requests were collected by querying the resident laboratory information system (LIS) ModuLab Version 2.2.07 (Systelab Technologies, Barcelona, Spain) for STAT orders incoming from the emergency department (ED) and the other nonEDs (NEDs) between June 1, 2010, and April 30, 2014. This data set was used to compute the frequency distribution of TAT (calculated according to the laboratorial process definition)13 and the volume of incoming orders for each month of the study. No statistical procedure for outlier removal was performed. However, to limit the influence of confounding data, a set of inclusion criteria was stated to exclude potentially aberrant records from the data set (see Table 1 for details).

Laboratory Setup Before automation, one stand-alone Siemens Vista 1500 (Siemens Healthcare, Milan, Italy) with a dedicated centrifuge was used to perform the sole STAT tests, whereas two other Siemens Vista 1500s were devoted to routine activity. The pre-analytics were hand-operated by two laboratory technologists, together with sample receipt, sorting, and transportation. The implementation of TLA consisted of the integration of the chemical analyzers to a 24 m long linear single-track

FlexLab conveyor belt (Inpeco, Lugano, Switzerland), with an online bulked input module (BIM) for automated checkin, an input–output module (IOM) for sample sorting, two centrifuges, a tube decapper, an automated specimen aliquoter, and a tube sealer (see Supplementary Fig. S1). The middleware Nemo (Inpeco) was used for process control and results validation. The same Siemens Vista 1500 analyzer was used before and after automation implementation to run STAT samples. The validation was carried out by a single pathologist with a dedicated computer terminal, and results were electronically reported. With respect to STAT testing, in our laboratory a different level of processing priority was assigned to the incoming requests according to the ordering department. Thus, two distinct workflows were set. For EDs, samples check-in and centrifugation [3500 relative centrifugal force (rcf)× 5 min] were performed off-line, and the tubes manually loaded through the IOM; in contrast, for NEDs, samples were fully processed in-line (centrifugation is at 3500 rcf × 5 min).

Data Analysis For TAT, the 50th (TAT50) and the 90th (TAT90) percentiles of each monthly distribution were taken to monitor laboratory efficiency.14,15 Thus, four distinct data series were obtained as follows: ED50 and ED90 for ED TATs, and NED50 and NED90 for NED TATs. Each series encompassed 47 longitudinal observations. For the volume of ordered tests (VOT), the number of monthly requests was taken as aggregate data (entire morning shift, 6:00–13:59) or as stratified according to time frames (TFs) of 60 min (ranging from TF6 for the 6:00–6:59 interval to TF13 for the 13:00–13:59 interval). Segmented regression analysis (SRA) was used for data modeling. Particularly, by means of SRA, a three-segment broken-stick model was fitted as follows: The first segment (S1) represented the data before TLA, whereas a second segment (S2) and a third segment (S3) represented the data after TLA. The S1–S2 knot [or change point 1 (CP1)] was constrained at the 11th month, corresponding to the TLA implementation. In contrast, the S2–S3 knot (or CP2) was chosen in

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

3

Ialongo et al.

Figure 1.  Segmented regression analysis (SRA) of ED50 (A), ED90 (B), ED VOT (C), NED50 (D), NED90 (E) and NED VOT (F) data series. Dashed vertical lines represent the change points (CP1 and CP2, if present), and the dotted horizontal line represents the regression line; each segment of the model is labeled as S1, S2, and S3 (if present).

correspondence with the month (>11th month) whose value maximized the adjusted R2 (aR2) of the segmented model. For each segmented model, the reliability was assessed by the Durbin–Watson statistics (DW), the probability of Shapiro– Wilk’s test for normality (SW), and the probability of the constant variance test (CVT). To assess the effect of automation on laboratory efficiency, the segmented model of the percentilebased TAT indicator (pTAT) data series underwent univariate analysis of covariance (ANCOVA). Particularly, simple contrast was used to assess the difference in mean level between the S1 (pre-TLA) and S3 (the true post-TLA) of each pTAT. To assess the effect of automation on VOT, the segmented model of data series underwent multivariate analysis of covariance (MANCOVA). Simple contrasts were used to assess the difference in mean level between S1 (preTLA) and S3 (true post-TLA). The same procedure was applied on stratified data to assess the difference in the mean level between the S1 (pre-TLA) and S3 (true postTLA) of each TF. Data were represented as the mean and standard error (SE), and the statistical significance was assessed for p < 0.05. All calculations were performed with SPSS 20.0 (IBM, Armonk, NY).

Results Data Set A total of 143,539 STAT CTNI requests were found in the LIS database, with 44,212 records (30.8%) considered as valid in accordance with the stated criteria (18,318 from EDs and 25,849 from NEDs).

ED50 The SRA showed that CP2 = 22±1.1 months (p < 0.01) (Fig. 1A). The ordinary least squares regression model proved to be reliable because no serial correlation was shown (DW = 1.687), nor were found a significant departure from normality (SW p = 0.464) and homoscedasticity (CVT p = 0.624) of residuals. The ANCOVA of the segmented model showed that the pre-TLA (S1) level was 56.6 ±1 min (p < 0.01), with a trend not significantly different from zero (b = 0.1±0.2 min, p = 0.507). After TLA, in the first phase (S2), the trend significantly changed by −1±0.2 min per month (p < 0.01). At equilibrium (S3), the mean level was 41.6±1.2 min (p < 0.01), with the trend not significantly different from pre-TLA

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

4

Journal of Laboratory Automation 

NED50 The SRA showed that CP2 = 15±0.6 months (p < 0.01) (Fig. 1D). The ordinary least squares regression model proved to be reliable because no serial correlation was shown (DW = 1.761), nor were found a significant departure from normality (SW p = 0.667) and homoscedasticity (CVT p = 0.235) of residuals. The ANCOVA of the segmented model showed that the pre-TLA (S1) level was 71.9±1.7 min (p < 0.01), with a trend not significantly different from zero (b = −0.3±0.3 min, p = 0.341). After TLA, in the first phase (S2), the trend significantly changed by −6.8±1.4 min per month (p < 0.01). At equilibrium (S3), the mean level was 52.±1.3 min (p < 0.01), with the trend not significantly different from that of pre-TLA (slope change = 0.1±0.3, p = 0.632) (see Fig. 2A). The mean difference before and after TLA (at equilibrium) was −19.8±2.2 min (p < 0.01) (see Fig. 2B).

NED90

Figure 2. (A) The pre–total laboratory automation (TLA) (S1) and post-TLA (S3) mean levels of percentile-based TAT indicators (pTATs) are displayed according to the ordering department; and (B) the post-to-pre TLA (S3–S1) mean difference of pTATs is displayed with markedly significant results (*p < 0.01), ordered according to their magnitude.

(slope change = −0.2±0.2, p = 0.315) (see Fig. 2A). The mean difference before and after TLA (at equilibrium) was −14.9± 1.5 min (p < 0.01) (see Fig. 2B).

The SRA showed that CP2 = 15±0.6 months (p < 0.01) (Fig. 1E). The ordinary least squares regression model proved to be reliable because no serial correlation was shown (DW = 1.792), nor were found a significant departure from normality (SW p = 0.532) and homoscedasticity (CVT p = 0.830) of residuals. The ANCOVA of the segmented model showed that the pre-TLA (S1) level was 126.3±3.8 min (p < 0.01), with a trend not significantly different from zero (b = −0.24±0.64 min, p = 0.707). After TLA, in the first phase (S2), the trend significantly changed by −11.3±3.1 min per month (p < 0.01). At equilibrium (S3), the mean level was 86.6±2.9 min (p < 0.01), with the trend not significantly different from that of pre-TLA (slope change = −0.8±0.7, p = 0.907) (see Fig. 2A). The mean difference before and after TLA (at equilibrium) was −39.7±4.7 min (p < 0.01) (see Fig. 2B).

Aggregated VOT

ED90 The SRA showed that CP2 = 15±0.5 months (p < 0.01) (Fig. 1B). The ordinary least squares regression model proved to be reliable because no serial correlation was shown (DW = 1.878), nor were found a significant departure from normality (SW p = 0.237) and homoscedasticity (CVT p = 0.091) of residuals. The ANCOVA of the segmented model showed that the pre-TLA (S1) level was 103.3±4.4 min (p < 0.01), with a trend not significantly different from zero (b = −0.4±34 min, p = 0.237). After TLA, in the first phase (S2), the trend significantly changed by −8.7±1.6 min per month (p < 0.01). At equilibrium (S3), the mean level was 69.6±1.5 min (p < 0.01), with the trend not significantly different from that of pre-TLA (slope change = 0.2±35, p = 0.607) (see Fig. 2A). The mean difference before and after TLA (at equilibrium) was −27.7±2.5 min (p < 0.01) (see Fig. 2B).

The SRA gave no reliable three-segment model for ED VOT because CP2 was not statistically significant (p = 0.1816) (Fig. 1C), and thus a two-segment model (CP1 = 11) was evaluated by means of ANCOVA. Hence, the preTLA (S1) mean monthly level was found to be 443.5±20.9 CTNI requests (p < 0.01), and the post-TLA (S2) mean monthly level was 437.6±12.6 CTNI requests (p < 0.01). No significant trend was found in either S1 or S2 (p = 0.295 and p = 0.139, respectively). The mean level difference of −5.9±24.4 was not statistically significant (p = 0.811) (see Fig. 3A and 3B). The SRA showed, for NED VOT, the CP2 = 23±3.7 months (p < 0.01, DW = 1.547, SW = 0.97, CV = 0.70) (Fig. 1E). Before the TLA (S1), the mean monthly level was 675.4±30.3 CTNI requests (p < 0.01), with a trend not significantly different from zero (b = 8.4±5.1 requests per

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

5

Ialongo et al.

Figure 3. (A) Pre–total laboratory automation (TLA) (S1) and post-TLA [S2 or S3, according to the segmented regression analysis (SRA)] mean levels of the volume of ordered tests (VOT) are displayed according to the ordering department; and (B) post-to-pre TLA (S3–S2 or S3–S1) mean differences of VOT are displayed with markedly significant results (*p < 0.01).

month, p = 0.110). After TLA implementation, in the first phase (S2), the trend changed significantly (slope change = −18.3±6.8 requests per month, p = 0.01). At equilibrium (S3), the mean level was 559.7±40.1 CTNI monthly requests (p < 0.01), with a trend not significantly different than that before TLA (slope change = −8.5±5.3 requests per month, p = 0.120) (see Fig. 3A). The mean level difference was −115.7±50.1 CTNI monthly requests (p = 0.026) (see Fig. 3B).

Stratified VOT The stratified VOT showed a peculiar profile for both EDs and NEDs (see Fig. 4A and 4B). The segmented model undergoing MANCOVA was obtained by applying change points derived by the SRA of aggregate data. For NED VOT, a significant mean difference between pre- (S1) and postTLA at equilibrium (S3) was observed for TF8 (−24.8±6.5 CTNI monthly requests, p < 0.01), TF9 (−20.1±4.9 CTNI monthly requests, p < 0.01), TF10 (−8.3±4 CTNI monthly requests, p = 0.045), TF12 (−18.3±5.8 CTNI monthly

Figure 4.  Volume of ordered tests (VOT) stratified according to 60 min time frames (TFs) of the regular morning shift. (A) Data relative to the emergency departments (EDs); and (B) data relative to the non–emergency departments (NEDs).

requests, p < 0.01), and TF13 (−15.1±5.5 CTNI monthly requests, p < 0.01). In contrast, no significant difference was found for TF6 (−26.5±34.2 CTNI monthly requests, p = 0.442), TF7 (−24.5±43.3 CTNI monthly requests, p = 0.575), and TF11 (−1.8±5.7 CTNI monthly requests, p = 0.754) (Fig. 5A and 5B). A significant trend was not found in either the S1 or S3 segments (data not shown).

Effect of TAT on VOT The simple linear regression showed that, in the S2 of the NED VOT series, a linear relationship (aR2 = 0.51, SW = 0.88, CVT = 0.48) between VOT and NED50 was found to be significant (ANOVA of the regression model p < 0.01). In particular, it was found that NED VOT = 172.5+(7.9 *NED50). (1)

Discussion The transition to total automation represents the greatest leap for a clinical laboratory. The adoption of a totally new philosophy of processing samples is, in our point of view, the most demanding change to face. In this regard, our analysis aimed to represent the transformation of the core

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

6

Journal of Laboratory Automation 

Figure 5. (A) The pre–total laboratory automation (TLA) (S1) and post-TLA (S3) mean level of stratified non–emergency department (NED) volume of ordered tests (VOT) is displayed according to 60 min time frames of the regular morning shift; and (B) the post-to-pre TLA (S3–S1) mean difference of stratified NED VOT is displayed with markedly significant results (*p < 0.01, **p = 0.045).

laboratory under different but complementary aspects of the efficiency: the promptness of the response to the new process, changes in the process capacities, and improvements in its control. To these qualitative attributes corresponded quantitative indicators, which were respectively the posteffect change point CP2, and the pTATs based on the 50th and 90th percentiles.16 By means of the analysis of these indicators, it can be seen that the effect of total automation was greater when the process control formerly had been worse. There are three points of evidence in this regard: First, higher percentilebased pTATs (ED90 and NED90) showed a more relevant reduction with respect to lower percentile-based pTATs (ED50 and NED50); second, the level reached by pTATs of NEDs under automation was even lower than what pTATs of EDs had before automation; and, third, the CP to equilibrium was reached sooner by the pTATs that represented a process under a looser control (ED90, NED90, and NED50). In our opinion, this all stems from the nature of the automated process. In fact, before total automation, the workflow management depended on the ability of the staff to execute different operations through a simple sequentiality.17 Hence, when a priority task was present (the incoming of an ED sample), all of the other tasks had to be queued, causing NEDs to have higher TAT than EDs. Afterward, with the automated system to manage different tasks in parallel, every process received the same level of control, so

that NEDs reached the efficiency that an unbalanced management formerly gave to EDs. With control not degraded by conflicting tasks, overtime processes were contained, and thus higher percentile-based pTATs were achieved. Almost the same response was exhibited by NED50 because, before TLA, the lower priority of NEDs corresponded to a looser control. The absence of a significant trend in the S3 for all pTATs showed the permanent nature of the change produced by total automation on efficiency. Did the improvement observed in TAT have an impact on the services depending on laboratory efficiency? In face of a TAT shorter than the clinically relevant goal of 45 min,18,19 quite surprisingly, the behavior of EDs in terms of ordered STAT CTNI tests remained practically unchanged. In contrast, NEDs’ monthly level of STAT CTNI orders significantly reduced by slightly more than 100 units. Review of the spectrum of VOT along the TFs of the regular morning shift (from 6:00 to 13:59; Fig. 5B) showed how the reduction was distributed in some of them (Fig. 5B). Notably, such TFs were grouped in two blocks: a first following the morning ward round (between 6:00 and 7:59, corresponding to TF6 and TF7), and a second following the TF that coincided with a 6 h cycle for a serial monitoring of CTNI20 (between 11:00 and 11:59, or TF11). In ED, the VOT was essentially generated by the randomness of outpatient admissions, so that such a functional relationship like that shown by improved laboratory efficiency with length of stay was not observed.11,12 Conversely, regarding both random and planned events of the NEDs with respect to inpatient monitoring, laboratory efficiency had a part in determining the systematic component of VOT. By the pattern we observed in TFs, we can conclude that an improved efficiency contained the tendency of physician to repeat or duplicate a STAT test due to nontimely reporting.21 Considering the reasons to order a CTNI test, it can be excluded that such a reduction may be attributed to a shortening of routine TATs.17 Hence, when regression was applied, we found that for a 1 min reduction in NED50, there were about eight saved STAT CTNI tests in the period between the implementation of TLA (S1) and the point of equilibrium for VOT (S2). Once at equilibrium (S3), there was a flat trend (data not shown). Notably, such a relationship was quite strong, as demonstrated by an aR2 = 0.51. Before concluding, we would highlight the strengths and limitations of our study. Total automation in our laboratory was reached by integrating the same analyzers that were formerly operated as stand-alone units, and therefore it was ideal to assess how workflow efficiency changed due to the automated operation. Because we had two parallel workflows for STAT samples, we were also able to show the effect of the automated processes in contrast to a somewhat hybrid operation. There are two distinct aspects to consider with regard to limitations. First, our conclusions are based on the observation of CTNI, and thus all generalizations of

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

7

Ialongo et al. our results to other tests should be carefully evaluated. This could be particularly true when considering the VOT, because a different analyte with its own plan of monitoring could show a different result. Nevertheless, we feel that what we have presented may be representative of the effect that automation has on order duplication caused by delayed reporting. Second, we would underline that we were not able to find recent and strictly comparable experiences in the published literature. In the few significant papers that we could take into consideration,22–24 the results presented came from a too-short period of observation and did not result in statistical analysis of the data. We would finally remark that the length of the observation period was probably a crucial aspect in the analysis of TAT (causing most of the papers we found to provide just a simple description rather than a true data analysis). In fact, a main issue in comparing longitudinal data with a clear trend is the unsuitability of a basic grouped approach like a t-test.25 However, if the ordinary least squares regression is chosen, results may be unreliable because of autocorrelation that arises in data due to the way they are collected (serially) throughout time.26 In our case, a longer period of observation allowed us to recognize the presence of some change points in the trend that were used to break the data (by means of SRA), limiting the effect of serial correlation. Thus, with more “ordinary” data, the ANCOVA made it possible to statistically assess the differences in mean levels found in the data, also controlling for the presence of trends. More powerful techniques exist for such situations (e.g., the intervention analysis27); however, they are less familiar to most researchers in the field of clinical pathology and laboratory sciences, and therefore were not used.

Conclusions Total automation affects the core laboratory by creating more efficient and stable management of the workflow. By means of an improved TAT, the volume of STAT tests ordered by hospital departments (except the ED) is decreased due to reduced duplicated requests.

Abbreviations ACS acute coronary syndrome ANCOVA analysis of covariance aR2 adjusted R2 BIM bulked input module CP1 change point 1 CP2 change point 2 CTNI cardiac troponin I CVT constant variance test DW Durbin–Watson statistics ED emergency department IOM input–output module

LIS laboratory information system MANCOVA multivariate analysis of covariance min minutes NED non–emergency department pTAT percentile-based TAT indicator standard error SE SRA segmented regression analysis SW Shapiro-Wilks test turnaround time TAT TAT50 median-based TAT indicator TAT90 90th percentile–based TAT indicator TLA total laboratory automation VOT volume of ordered tests Acknowledgments The authors would thank all of the staff at the core laboratory for their collaboration and professionalism.

Declaration of Conflicting Interests The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding The authors received no financial support for the research, authorship, and/or publication of this article.

References 1. Boyd, J. Robotic Laboratory Automation. Science. 2002, 295, 517–518. 2. Armbruster, D. A.; Overcash, D. R.; Reyes, J. Clinical Chemistry Laboratory Automation in the 21st Century: Amat Victoria Curam (Victory Loves Careful Preparation). Clin. Biochem. Rev. 2014, 35, 143–153. 3. Hoffmann, G. E. Concepts for the Third Generation of Laboratory Systems. Clin. Chim. Acta. 1998, 278, 203–216. 4. Hawker, C. D. Laboratory Automation: Total and Subtotal. Clin. Lab. Med. 2007, 27, 749–770. 5. Markin, R. S; Whalen, S. A. Laboratory Automation: Trajectory, Technology, and Tactics. Clin. Chem. 2000, 46, 764–771. 6. Wing, A. K. Laboratory Automation and Optimization: The Role of Architecture. Clin. Chem. 2000, 46, 784–791. 7. Sarkozi, L.; Simson, E.; Ramanathan, L. The Effects of Total Laboratory Automation on the Management of a Clinical Chemistry Laboratory: Retrospective Analysis of 36 Years. Clin. Chim. Acta. 2003, 329, 89–94. 8. Godolphin, W.; Bodtker, K.; Wilson, L. Simulation Modelling: A Tool to Help Predict the Impact of Automation in Clinical Laboratory. Lab. Robot. Autom. 1992, 4, 249–252. 9. Hoffmann, G. E.; Slapansky, P.; Tlstak, R. Redesign Your Laboratory on a PC: New Tools for Simulation-Based Workflow Analysis. Lab. Robot. Autom. 1998, 10, 215–221. 10. Singh, S. R. Public Health Spending and Population Health: A Systematic Review. Amer. J. Prev. Med. 2014, 47 (5), 634–640. 11. Holland, L. L.; Smith, L. L.; Blick, K. E. Total Laboratory Automation Can Help Eliminate the Laboratory as a Factor

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

8

Journal of Laboratory Automation 

in Emergency Department Length of Stay. Amer. J. Clin. Pathol. 2006, 125, 765–770. 12. Holland, L. L.; Smith, L. L.; Blick, K. E. Reducing Laboratory Turnaround Time Outliers Can Reduce Emergency Department Patient Length of Stay: An 11-Hospital Study. Amer. J. Clin. Pathol. 2005, 124, 672–674. 13. Breil, B.; Fritz, F.; Thiemann, V.; et al. Mapping Turnaround Times (TAT) to a Generic Timeline: A Systematic Review of TAT Definitions in Clinical Domains. BMC Med. Inform. Decis. Mak. 2011, 11, 34. 14. Valenstein, P. N.; Emancipator, K. Sensitivity, Specificity, and Reproducibility of Four Measures of Laboratory Turnaround Time. Amer. J. Clin. Pathol. 1989, 91, 452–457. 15. Valenstein, P. Laboratory Turnaround Time. Amer. J. Clin. Pathol. 1996, 105, 676–688. 16. Hawkins, R. C. Laboratory Turnaround Time. Clin. Biochem. Rev. 2007, 28, 179–194. 17. Howanitz, P. J.; Steindel, S. J. Intralaboratory Performance and Laboratorians’ Expectations for Stat Turnaround Times: A College of American Pathologists Q-Probes Study of Four Cerebrospinal Fluid Determinations. Arch. Pathol. Lab. Med. 1991, 115, 977–983. 18. O’Gara, P. T.; Kushner, F. G.; Ascheim, D. D.; et al. 2013 ACCF/AHA Guideline for the Management of ST-Elevation Myocardial Infarction: A Report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J. Amer. Coll. Cardiol. 2013, 61, e78–e3140. 19. Anderson, J. L.; Adams, C. D.; Antman, E. M.; et al. 2012 ACCF/AHA Focused Update Incorporated into the ACCF/ AHA 2007 Guidelines for the Management of Patients with Unstable Angina/Non–ST-Elevation Myocardial Infarction:

A Report of the American College of Cardiology Foundation/ American Heart Association Task Force on Practice Guidelines. Circulation. 2013, 127, e663–e828. 20. Newby, L. K.; Jesse, R. L.; Babb, J. D.; et al. ACCF 2012 Expert Consensus Document on Practical Clinical Considerations in the Interpretation of Troponin Elevations: A Report of the American College of Cardiology Foundation Task Force on Clinical Expert Consensus Documents. J. Amer. Coll. Cardiol. 2012, 60, 2427–2463. 21. Howanitz, J. H.; Howanitz, P. J. Laboratory Results: Timeliness as a Quality Attribute and Strategy. Amer. J. Clin. Pathol. 2001, 116, 311–315. 22. Lam, C. W.; Jacob, E. Implementing a Laboratory Automation System: Experience of a Large Clinical Laboratory. JALA. 2012, 17, 16–23. 23. Brannon Thomas, L.; Borkowski, A. A.; Rubero-Lung, M.; et al. The Impact of Laboratory Automation on Performance Improvement. Fed. Pract. 2008, 25, 14–18. 24. Streitberg, G. S.; Bwititi, P. T.; Angel, L.; et al. Automation and Expert Systems in a Core Clinical Chemistry Laboratory. JALA. 2009, 14, 94–105. 25. Lagarde, M. How to Do (or Not to Do) … Assessing the Impact of a Policy Change with Routine Longitudinal Data. Health Policy Plan. 2012, 27, 76–83. 26. Wagner, A. K.; Soumerai, S. B.; Zhang, F.; et al. Segmented Regression Analysis of Interrupted Time Series Studies in Medication Use Research. J. Clin. Pharm. Ther. 2002, 27, 299–309. 27. Box, G. E. P.; Tiao, G. C. Intervention Analysis with Applications to Economic and Environment Problems. J. Amer. Stat. Assoc. 1975, 70, 70–79.

Downloaded from jla.sagepub.com at UNIV OF PITTSBURGH on November 15, 2015

Total Automation for the Core Laboratory: Improving the Turnaround Time Helps to Reduce the Volume of Ordered STAT Tests.

The transition to total automation represents the greatest leap for a clinical laboratory, characterized by a totally new philosophy of process manage...
631KB Sizes 0 Downloads 7 Views