Article Text

PDF

Improving access to diagnostics: an evaluation of a satellite laboratory service in the emergency department
  1. P Leman1,
  2. D Guthrie2,
  3. R Simpson2,
  4. F Little3
  1. 1Emergency Department, St Thomas’ Hospital, London, UK
  2. 2Pathology Directorate, St Thomas’ Hospital
  3. 3Health Economist, Colchester, UK
  1. Correspondence to:
 Dr P Leman
 Emergency Department, Fremantle Hospital, Alma Street, Fremantle 6160, Australia; peter.lemanhealth.wa.gov.au

Abstract

Objectives: To measure the impact of a satellite laboratory upon laboratory result turnaround times and clinical decision making times.

Design: A prospective cohort study, the intervention group had blood tests sent Monday to Friday 12 noon to 8 pm and the control group had blood tests sent outside these hours. The data were collected over a six week period before the laboratory was opened, and a subsequent six week period.

Setting: An urban teaching hospital emergency department.

Participants: 1065 patients requiring blood tests.

Main outcome measure: Time from the blood sample being sent to the laboratory to the results being available on the clinician’s computer.

Results: The time to haematology (blood count) results in the intervention group decreased by 47.2 minutes (95% CI 38.3 to 56.1, p<0.001) after the laboratory was opened. The corresponding control group times were unchanged (0.6 minutes; −13.8 to 15.0, p = 0.94). Similar sized differences were also seen for haemostasis (d-dimer) testing 66.1 (41.8 to 90.4) minutes compared with −14.2 (−47.1 to 18.7) and chemistry 41.3 (30.3 to 52.2) compared with −4.2 (−17.4 to 8.9) testing. Decisions to discharge patients were significantly faster (28.2 minutes, 13.5 to 42.8, p<0.0001) in the intervention group after the laboratory was opened (controls; −2.6 minutes −27.0 to 21.7). No change was seen with decisions to admit patients. There was a trend for earlier laboratory results modifying intravenous drug or fluids orders, or both (p = 0.06)

Conclusion: A comprehensive satellite laboratory service is an important adjunct to improve the timeliness of care in the emergency department.

Statistics from Altmetric.com

Improving access to diagnostics is an essential component of the health services strategy to improve access to emergency care.1 In particular it is recognised that the time taken to obtain blood results for patients in the emergency department is a rate limiting step that can prevent them being seen, treated, and either admitted or discharged within four hours.1 Within the NHS this four hour standard needs to be achieved for 90% or all patients attending emergency departments by April 2003, and 100% by April 2004.2

Much has been written about the role of some near patient testing kits that can be used by clinicians to improve access to pathology results.3 This is used most frequently to rule in or out specific diseases such as acute coronary syndromes (troponin T)4–6or pulmonary emboli (d-dimer).7,8 Other examples, are the use of more broad near patient testing techniques (I-Stat) monitors.9,10 There are drawbacks to this however, and previous studies have not always shown a significant benefit of near patient testing. This has been mainly attributable to the lack of the full range of tests being available, the perceived inaccuracy/reliability of the results obtained that require clinicians to regularly perform quality assurance,11,12 and that other clinical factors also determine timings in an episode of care.13 In general, it is also more expensive.14,15

We therefore decided to evaluate the utility of a satellite laboratory, managed, controlled, and staffed by the pathology department, but situated in the heart of the emergency department. The diagnostic tests were performed by state registered biomedical scientists and the analysers used were similar to those found in a main hospital laboratory.

METHODS

Using a process improvement model we formed a key group of stakeholders who would oversee the changes, these came from the emergency department and the two main laboratory services (chemistry and haematology). Seconded people from other key areas including infection, haemostasis (a separate department in our hospital) and user groups were convened to determine how the changes would be implemented. A bank of tests was created that were both feasible to perform and may have an impact upon clinical decision making (see box). A storage room within the resuscitation area was converted to a laboratory (requiring modifications in climate control, plumbing, electrical, and IT conduits, etc). Laboratory equipment was leased from manufacturers. The main criteria when choosing the analysers being that the results had to be identical to those generated in the main laboratory, the equipment had to fit into the space available, and interface with the main laboratory and hospital IT systems. Once installed a vigorous quality assurance programme was implemented and the equipment went through thorough acceptance testing over several months.

Pathology investigations available in the satellite laboratory

Chemistry

  • Urea/creatinine and electrolytes

  • Liver function tests

  • Amylase

  • Calcium and phosphate

  • C reactive protein

  • Troponin T

  • Paracetamol and salicylate concentrations

Haematology

  • Full blood count

Haemostasis

  • d-dimer

The equipment installed included; Allegra 21 Centrifuge, Beckman Coulter CX5 Pro Chemistry Analyzer, Beckman Coulter Ac.T 5diff CP Haematology Analyzer, Roche cardiac reader (troponin T), and a Nyo Card Reader II (d-dimer).

The evaluation project was designed to assess the various stages of the patient journey and the timeframes around the blood sample processing. Thus, for patients attending the emergency department in whom blood tests were performed, data were collected on time of arrival, time to nurse/doctor assessment, time blood sample taken, time blood sample received in laboratory, time results posted on laboratory computer, time clinician aware of results, time decision made to either discharge or refer the patient for admission. Other information was also collected on grade of doctor initially seeing patient, data around other diagnostics required (for example, radiology), and whether the patient was first discussed with a more senior emergency department doctor before a decision was made. As a simple marker of the clinical impact of the blood test results the clinicians were asked to enter whether any intravenous therapy was either changed or started upon knowledge of the blood test results.

The data collection form was piloted in May 2002 and the modified form was used over a six week period in June and early July 2002 to collect the baseline data on samples that were wholly processed in the main laboratory. The satellite laboratory went on line in August and similar data were collected over a six week period in late September through early November 2002.

The satellite laboratory was only open between 12 00 to 20 00 Mondays to Fridays. This allowed us to use a before and after comparison of the equivalent “lab open” times (intervention group) and use the equivalent before and after data from the “lab closed” times (20 00 to 12 00 and weekends) as a direct control group. The data were entered into an MS Access database and subsequently analysed using SYSTAT software. Student’s, t tests were used to compare continuous data between groups. We used χ2 tests for 2×2 tables. The multiple regression analysis used stepwise backward regression for modelling the effect of the different blood test request upon timings (factors being excluded if p>0.15). The measured confounders regression did not exclude any factors in the final model.

RESULTS

A total of 1065 data forms on sample patients were collected (252 intervention group-before, 337 control group-before, 206 intervention group-after, and 270 control group-after). Not all forms had all fields completed and this is reflected in the given denominators in the relevant results. There was a large overall decrease in the processing time for samples processed in the satellite laboratory compared with the main laboratory (table 1). This was across the range of tests available, however there was no change in processing time when samples taken outside normal satellite laboratory hours were compared.

Table 1

 Laboratory turnaround time

We assessed two main clinical indicators of efficacy; time to make the decision to refer for admission, and time to make the decision to discharge (table 2) The results showed that discharge decisions occurred significantly earlier (28.2 minutes) and that no effect was seen on decision making times for patients referred for admission. Referred patients however had far shorter decision making times than discharged patients.

Table 2

 Effect of laboratory results on decision times

To account for other process changes over time, in addition to using an out of hours control group, we also looked at whether our overall system had improved (table 3). We found that in both the intervention and the control groups patients were seen by a doctor much more quickly after the laboratory had become operational. However, no improvement was noted in the time it took the doctors to send off the samples, once the patients were seen.

Table 3

 Process changes during study

In addition we found further factors that could be improved. The mean time from the results being available on the departmental computer and the doctor recording that they were aware of them was excessive, and did not significantly vary either between day and night or with the satellite laboratory open or closed. The mean was between 29 to 36 minutes for haematology results and 17 to 20 minutes for chemistry results.

Multiple regression modelling was used to determine whether any particular measured confounders had an impact upon decision times. Although senior doctors were quicker this was not significant for either discharges (coefficient −21.8 minutes, p = 0.13) or admissions (coefficient −14.4 minutes, p = 0.23). The most important delays for discharges and admissions were waiting for non-standard imaging tests such as CT scans or ultrasound scans (discharged group, coefficient 53.9 minutes, p = 0.02; admitted group, coefficient 51.8 minutes, p⩽0.01). For discharged patients only, waiting to discuss a case with a more senior or other doctor (coefficient 19.9 minutes, p = 0.02).

A similar regression technique was used (though with a stepwise backwards technique to exclude non-significant relations) to model the various blood tests and their effect upon the outcome times. In the discharged patients the largest effects were seen with erythrocyte sedimentation rate (ESR) (coefficient 73.4 minutes, p = 0.03) and d-dimer (31.8 minutes, p = 0.09) before the laboratory was opened. In the intervention phase clotting screen (28 minutes, p = 0.03) was the most significant delay, this test still being performed in the main laboratory. In the patients who were referred the problems of waiting for amylase (coefficient 39.7 minutes, p<0.01) and other chemistry tests were removed with the satellite laboratory open, leaving clotting screen (coefficient 28.2 minutes, p = 0.06) as the only significant factor.

The final effect measured was used to determine if there was a clinical or quality of care improvement subsequent to earlier test result availability. We asked clinicians to record whether having earlier blood tests had any effect to change the current intravenous fluid or drug therapy. Thus receipt of the blood test result had to produce treatment changes. These included changes, such as earlier blood transfusion, addition of antibiotics, addition of potassium to intravenous fluids, etc. We found that twice as many patients (10.2% v 5.0%) in the intervention group were having treatment modified after blood test results, and this approached significance (p = 0.06). In the control groups there was no difference seen (p = 0.33).

Costs

We report the costs in general terms as this was a pilot project and revenue and capital costs may vary considerably if a long term satellite laboratory was designed. Furthermore, these costs would not necessarily be generalisable to other institutions.

At 2002/3 financial year prices the capital costs were zero as equipment was leased, the annual lease charge for the analysers being £31 998. The annual staffing costs to provide an eight hour online laboratory (12 hours of BMS time) five days a week are £48 788. Reagent costs were estimated at £24 871. The reagent costs are provided net, reflecting the decreased main laboratory reagent costs.

Actual laboratory use has increased significantly since the pilot phase reported here with more emergency department testing and the addition of blood testing of patients referred directly to inpatients teams (who were excluded during the evaluation phase). We estimate an annual throughput of about 10 000 patients, and given the various possible combinations of requested tests, the average marginal premium for satellite laboratory investigation is £12 per patient.

DISCUSSION

We have found that a satellite laboratory based in the emergency department can have a profound impact upon turnaround times for blood results. Previous studies have usually shown quicker results, but rarely has this flowed into quicker journeys for patients.13,16,17 This may have been attributable to most of these studies using small near patient testing kits that required the clinicians themselves to operate and maintain. There may be issues with validity and quality assurance with such kits that may have an impact upon the clinician’s decision to use the results. Furthermore, as a full panel of tests equivalent to the main laboratory is not provided by such hand held kits, there is still the inevitable delay until all the necessary results are known.

While the decision to discharge was brought forward because of earlier blood results the decision to refer was not. Intuitively this fits with the workings of our emergency department. Most admissions are readily identifiable from history and examination, and observation of clinical progress/response to treatment, the blood tests only confirming the diagnosis in most, although not all, cases. Thus the decision to refer for admission is often made before the results are available. From our data this was around half the time it took to make the decision to discharge. This latter group is clearly the target patient group as they wait the longest in the department.

We chose not to use total time in the department as our evaluation measure because of the multitude of additional factors (both within the emergency department and outside it) affecting this outcome parameter. The NHS Plan target of percentage of patients with total time in the department being less than four hours did vary between the study periods. It improved for majors patients from 36% to 50%, however minors patient also improved from 56% to 77% in the same period, where few had any blood tests taken. We believed that these reflect whole systems changes in the emergency department and hospital, thus affecting the validity of this as a valid outcome variable.

There is still room for improvement upon the times that we found. It is clear that even though you could save 40 to 50 minutes with the satellite laboratory, you may also theoretically decrease the equivalent time just by taking all blood tests upon arrival of the patient, rather than doing what most clinicians do—that is, take a history, perform an examination, and then think about what blood tests to do. We did not measure whether such a change would have any actual benefit and this clearly requires additional research. In addition there is a need to “push” the information to the clinicians once it is available. In our department, for many patients once the blood tests have been sent and initial treatment begun, the doctor starts seeing a new patient. They are therefore usually unaware when the blood results have been posted to the system. We use a rapid printer that prints out the blood tests immediately they are available, but it is usually the nursing staff who pick these up and then track down the doctor to act upon them. The main practical difficulty with other options (such as paging the doctor, using a tannoy system or phone call) is that they require the biomedical scientist (BMS) to perform this task for every individual result as it comes up, clearly extremely time consuming.

We do see a role for clinical BMSs within the emergency department not only to undertake, interpret, and report on the investigative procedures but also to ensure that the basic aspects of clinical laboratory procedures are followed by the medical and nursing staff in the emergency department, for example, checking that samples are correctly chosen, filled, and labelled. This may improve the quality of samples and subsequent laboratory analysis. However, additional BMS staff are expensive and this cost has to be compared with other options including stand alone near patient testing in further studies.

The results of our regression have pointed us to the need to expand our haematology/haemostasis suite of tests in the satellite laboratory. We have subsequently taken on board ESR and INR testing. We are also considering the use of malarial antigen test kits as we commonly run about three to five malaria films per day, the vast majority of positives being P falciparum from West African exposures.

The study was not designed to examine the issues around quality of care as well as timeliness of care. However, we did see a trend towards more changes or commencement of intravenous therapies subsequent to blood test availability. We believe that this reflects the clinicians providing more appropriate and individualised treatment earlier in the patient journey rather than providing a standard response to the clinical problem and referring onwards.

There are many limitations to this study that should be taken account of in interpreting the results. We elected not to use a purely randomised model looking at timings once the laboratory had opened for several reasons. Firstly, we received advice that this would require individual consent from patients, which we determined would probably greatly diminish the number of patients recruited because of the excess time it would take to recruit them for the medical staff. Furthermore, we were looking to obtain our results early as the initial funding for the service was only for six months (including set up time to build the laboratory) and we needed to provide a reasonably quick evaluation to continue the service. However, the data for the intervention group before the satellite laboratory was opened and the control group seem to provide adequate comparators. The use of the before/after analysis for the control group was used to take account of any changes over time that may have adversely influenced our findings. The main change we saw was that of improved time from arrival to first seeing the doctor. The control group has usefully illustrated that this occurred for all patients and did not affect the main study findings. The study was clearly not blinded to the clinicians and this may have affected how they acted or how they recorded their decision making times. Not all patients who were eligible for entry to the study were recorded and the missing data may bias the results—in particular there were fewer data from registrar and consultant seen patients than expected.

Overall we have found that the satellite laboratory now provides a useful service to the emergency department that has improved both the timeliness and quality of patient care. We realise that it is only part of a whole systems approach that is required to improve timeliness of care, and cannot resolve the issue of improving total time in the emergency department by itself.

REFERENCES

View Abstract

Footnotes

  • Funding: none.

  • Conflicts of interest: none declared.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.