Variation in hospital pathology investigations

Background and aims

Diagnosis-Related Groups (DRGs) enable hospitals to be paid for the number and mix of patients they treat by reducing a large number of individual hospital patients into manageable and meaningful groups. This project examined the use of DRGs across six hospital sites in New South Wales to identify profiles of pathology requesting including test volume utilisation, test turn-around times, variation between clinicians, repeat testing rates and careset utilisation. Statistical and economic modelling was used to establish the relationship between the pathology requesting profiles and patient outcomes (e.g. length of stay in hospital, phlebotomy episodes, rates of re-admission); and resource utilisation. These indicators can be used as performance benchmarks when comparing between hospitals, and when examining temporal factors such as time of day, day of week or season, to monitor and enhance the quality use of pathology. The outcomes of this report have the potential to benefit a wide range of stakeholders within the healthcare system.

Aims

• Examine the use of Diagnosis-Related Group (DRG) and International Classification of Disease (ICD) codes to identify profiles of pathology requesting and compare performance across hospital and clinician levels.
• Undertake statistical and economic modelling to establish the relationship between pathology requesting profiles and patient outcomes (e.g. length of stay in hospital, phlebotomy episodes and rates of hospital re-admission); and resource utilisation.

Design and method, benefits and findings

The study was conducted at a group of six hospitals (denoted as hospitals A, B, C, D, E and F) serviced by a single pathology service that provides comprehensive biomedical laboratory services. A Cerner Powerchart Electronic Medical Record (EMR) system, which enables electronic creation of pathology test orders, was implemented across the six hospitals. In 2013, approximately 80% of pathology test orders across the six study hospitals used the EMR system. These electronic orders were analysed to enable an assessment of pathology test utilisation.

Benefits

This project has the potential to benefit a range of different stakeholders in the healthcare system.

Patients/consumers

The benchmark measures can assist in improving the standardisation of clinicians’ test requesting practices. The project included patient-centric indicators which measured the quality of care (the rates and timing characteristics of repeat testing, the length of stay in the hospital or ED and the rates at which patients presented to a study ED within 28 days of the previous ED presentation).

Clinicians

Pathology services contribute to all branches of medicine. They assist the clinical decision making process and make a critical contribution to the well-being of patients. By profiling pathology utilisation via DRG codes and other patient characteristics, it was possible to control for patient casemix and other confounding variables and compare test selection and utilisation practices in different contexts and according to their impact on outcomes such as LOS.

Hospital pathology laboratories

Measuring, benchmarking and comparing the use of pathology laboratory services is a critical process in the monitoring and quality improvement that all organisations should pursue. Performance benchmarks can enhance quality of practice across different sites.

Hospital management

The analyses of pathology utilisation according to DRG codes enable improved description and comparison of clinician test selection practices and the evaluation of cost effectiveness. Hospitals can compare ordering patterns to those at other sites and use performance data to inform decisions about quality improvement and decision support within their own organisation.

Government Departments of Health and LHDs

Departments of Health and Local Health Districts can use the benchmark results from this project in macro-level decision making. For example, some aspects of clinical decision support (such as the creation and availability of Caresets, or duplicate test order alert parameters in the EMR) may be most effective if applied broadly across entire jurisdictions in the health system. The repeat test timing characteristics and Careset utilisation analyses produced in this project can inform which clinical decision support mechanisms are most likely to yield improvements in pathology requesting, which may favourably impact on patient outcomes.

Findings

Test volume utilisation

When adjusted for hospital, year, casemix (DRG), patient age and sex, the test utilisation generally increased each year between 2008 and 2011. The adjusted rate was higher in 2012 compared to 2011, for Hospitals A, D and F, but not Hospital E. There was a significant reduction in the adjusted mean rate of tests per patient day, from 2012 to 2013, at Hospitals D and E (approximately 0.4 fewer tests per patient day). There was no significant difference in the rate, from 2012 to 2013, at Hospitals A and F.

Variation between clinicians

When focusing on patients who were allocated to the ‘Chest Pain’ DRG, and comparing the variation between clinicians, Hospital D had the lowest median number of pathology tests ordered per patient day, but had the greatest variation between clinicians. Hospital F had the smallest variation between clinicians.

Repeat testing

Overall repeat Electrolytes, Urea, Creatinine (EUC) test rates within 24 hours of the previous test were similar at all study hospitals. This pattern was also found when focusing on patients admitted with the ‘Tracheostomy W/ Vent >95 hours W/ or W/O Cat CC’ DRG, but for the ‘Chest Pain’ DRG, repeat EUC test rates within 24 hours varied considerably between the four study hospitals. The repeat EUC test rate within 24 hours exceeded 20% at Hospital A, but was approximately 5% at Hospital D. A similar pattern was found for repeat Full Blood Count (FBC) tests.

Turn-around times

The Turn-Around Times (TATs) were compared for the Top-10 pathology tests ordered for inpatients registered with the ‘Chest Pain’ DRG and matching populations within the ED (some of whom were admitted as inpatients, and others whose treatment was completed within the ED). Overall, pathology tests ordered for ED patients whose treatment was completed in the ED were processed the quickest, with a median TAT of 49 minutes; the TAT was 52 minutes for ED patients who were eventually admitted and 60 minutes for inpatients. Similarly, the overall variability, as indicated by the Inter-Quartile Range (IQR), was smallest for ED patients whose treatment was completed in the ED (IQR=34 minutes); second smallest for ED patients who were eventually admitted (IQR=40 minutes); and greatest for inpatients (IQR=53 minutes). The same pattern, for both median TATs and variability, was evident for almost all Top‑10 tests considered in the analysis.

Demand management in the ED

Patients presenting with digestive system illnesses accounted for the highest proportion of patients (25%) who had a C-Reactive Protein (CRP) test ordered in the first test order episode. Patients presenting with neurological illnesses accounted for the highest proportion of patients (23%) who had a Creatine Kinase (CK) test ordered in the first test order episode.

EUC and FBC tests were the most frequently ordered tests in the first test order episode for patients located in the acute/resuscitation area for all ED presentations with circulatory, digestive, respiratory, neurological illnesses and system infection/parasites (the Top-5 MDB categories with most ED presentations).

Careset utilisation

Out of 289,417 tests, 34,008 were ordered as part of a Careset (also known as ‘Order Sets’), accounting for 11.8% of tests. ‘Blood Group and Antibody Screen’, containing Blood Group and Antibody Screen, BBT History and Anti-D Antibody, was the most frequently ordered Careset, ordered 4,441 times and accounting for 51.2% of all Caresets ordered. ‘Tracheostomy W/ Vent >95 hours W/ or W/O Cat CC’ was the inpatient DRG with the most number of Caresets ordered at 1,427 (14.2%). However, there were several DRGs where a greater proportion of tests were ordered using Caresets, including ‘Neonate, AdmWt >2499g W/O Significant OR Procs W/O Problem’ with 40.4% of tests ordered as a Careset and ‘Red Blood Cell Disorders W/O Cat or Sev CC’ with 39.4% of tests ordered as a Careset.

ED LOS

Multilevel modelling analyses, which do not constitute evidence for causation, indicated that ED patients who were eventually admitted as inpatients were estimated to have an additional 158.1 minutes length of stay in the ED if any pathology tests were ordered during their ED presentation. For ED patients whose treatment was completed within the ED, they were estimated to have an additional 98.5 minutes length of stay in the ED if any pathology tests were ordered. The utilisation of any imaging procedures during the ED presentation were estimated to increase ED LOS by 37.7 minutes for patients whose treatment was completed within the ED, but reduce the ED LOS by 44.6 minutes for patients who were eventually admitted.

The impact of pathology testing on ED LOS differed according to the laboratory department involved. The impact was greater for pathology tests conducted in the Clinical Chemistry department (an estimated increase of 112.0 minutes in ED LOS) than for tests conducted in Haematology (an estimated increase of 46.1 minutes in ED LOS) and Microbiology departments (an estimated increase of 63.0 minutes in ED LOS).

Resources

Variation in hospital pathology investigations