Purpose of the study To reduce the number of unnecessary laboratory tests ordered through a measurement of effects of education and cost awareness on laboratory ordering behaviour by internal medicine residents for common tests, including complete blood cell count (CBC) and renal profile (RP), and to evaluate effects of cost awareness on hospitalisation, 30-day readmission rate and mortality rate.
Study design 567 patients admitted during February, March and April 2014 were reviewed as the control group. Total CBC, CBC with differential and RP tests were counted, along with readmission and mortality rates. Interventions were education and visual cost reminders. The same tests were reassessed for 629 patients treated during 12 months after intervention in 2015.
Results Data showed a significant increase in CBCs ordered after the intervention (mean number per hospitalisation changed from 1.7 to 2.3 (p<0.001)), a decrease in CBCs with differential (mean number changed from 1.7 to 1.2 (p<0.001)) and no change in RPs ordered (mean number, 3.7 both before and after intervention (p=0.23)). No change was found in mortality rate, but the decrease in the readmission rate was significant (p=0.008).
Conclusions Education in the form of cost reminders did not significantly reduce the overall ordering of the most common daily laboratory testing in our academic teaching service. We believe further research is needed to fully evaluate the effectiveness of other education forms on the redundant ordering of tests in the hospital setting.
- cost awareness
- daily laboratory testing
- resident education
Statistics from Altmetric.com
Waste in healthcare has been defined as ‘spending that can be eliminated without reducing the quality of care’.1 The Institute of Medicine estimates that US$690 billion is wasted in the US healthcare system annually (http://www.nationalacademies.org), representing 17.9% of the gross domestic product related to healthcare expenditures.2 Blood tests are a valuable tool for diagnosis and treatment of multiple conditions, but overutilisation of laboratory testing is an important contributor to healthcare waste.3 This problem is well recognised by healthcare providers; however, no consensus exists on what comprises appropriate laboratory ordering. 4 In the Choosing Wisely campaign,5 the Society of Hospital Medicine recommends limiting repetitive measurements of complete blood cell counts (CBCs) and chemistries in the face of clinical and laboratory stability. Despite this recommendation, automatic daily laboratory testing continues to be a high-frequency order in many institutions.6 ,7
Various factors may contribute to overutilisation of laboratory testing, including patient demographic characteristics, length of stay, case complexity, routine clinical practice, cost unawareness and physician inexperience.7 Intervention strategies such as electronic health record alerts, teaching sessions, chart audits, feedback and economic incentives have demonstrated promising results8–10 but require intensive resources. Alerting healthcare providers to laboratory costs has been shown to reduce laboratory ordering; yet, this intervention has not been studied for medical residents,11–13 who often are the ordering providers, particularly in teaching institutions. Strategies aimed at trainees may improve current practice teach responsible habits that last beyond the training years.
To improve the ordering practices of internal medicine residents, we undertook a quality improvement project to reduce the ordering of daily laboratory workups. We aimed (1) to measure the effect of education and cost awareness on laboratory ordering behaviour in internal medicine residents and (2) to quantify the most common laboratory tests performed daily—CBC, CBC with differential and renal profile (RP) which includes assessment of electrolytes, serum urea nitrogen, creatinine, glucose, calcium and phosphorus. We evaluated the effect of cost awareness on patients’ overall cost, 30-day readmission rate and mortality rate.
This quality improvement project was conducted at a 302-bed tertiary teaching hospital. We selected three internal medicine teaching teams for the study. Each team had one third year resident, one second year resident and two interns supervised by a staff-attending physician. The interns placed most of the patient orders, with the two senior residents occasionally placing orders. Each team had a workroom with computer stations where most orders were placed and the required documentation completed.
Data were collected on all patients admitted to the three target internal medicine teams. The total numbers of CBCs, CBCs with differential and RP tests per patient per hospitalisation were recorded. In addition, we examined demographic characteristics, hospital readmission rates (defined as readmission <30 days before encounter admission) and mortality rate (defined as deaths during the hospitalisation). We selected February, March and April 2014 as our control period for collecting baseline data before an intervention. The Charlson Comorbidity Index, a 1-year mortality predictor based on multiple medical conditions, was used to compare the comorbidities of the control and intervention groups. Thereby, data on the presence of these conditions were collected of all patients in baseline and postintervention groups.
A 30 min teaching session for internal medicine residents was held in December 2014, in which the quality improvement team presented the current literature encompassing laboratory ordering practices. This literature included coverage of the waste produced by excessive laboratory ordering and the interventions tried thus far, both successful and unsuccessful. The objectives of the project were delineated to the residents, including the anticipated timeline. The session concluded by answering questions and addressing comments or concerns of the residents regarding the information presented.
Starting in February 2015, laminated cards with the considerations and costs of the selected three laboratory tests (figures 1 and 2) were placed on each resident's computer. The cards were left in place for 3 months. At the conclusion of this period, the team performed a chart review of each patient admitted to the internal medicine teaching service in February, March and April 2015 and obtained the same information in data collection as collected for these months in 2014. Patients requiring time in the intensive care unit were excluded from the study.
Statistics for continuous variables were reported as mean (SD) and median (range); categorical variables were reported as frequency and percentage. Demographic information, laboratory test orderings and outcome variables were compared between the patient subgroup of February, March and April 2014 and the subgroup of 2015. Continuous variables were compared between the 2 years with Wilcoxon rank-sum test; categorical variables were compared with χ2 test. All tests were two-sided, and the α level was set at 0.05 for statistical significance.
The study followed the guidelines of Strengthening the Reporting of Observational Studies in Epidemiology checklist.
During February, March and April, the total numbers of patients seen by internal medicine residents were 567 and 629 for 2014 and 2015, respectively (table 1). Between the 2 years, the average hospital length of stay was 4.7 and 4.8 days; the mean average age was 64.4 and 66.3 years, respectively (p=0.11). Sex was similar as well: 50.2% female and 49.8% male (p=0.68). No statistical difference on the Charlson Comorbidity Index was found between the baseline and intervention groups (p=0.32).
When comparing the CBC tests ordered per hospital stay between the baseline and intervention groups, we found a significant increase in mean number for the patients after the intervention (mean, 1.7 and 2.3 (p<0.001)) (figure 3). However, a concomitant significant decrease was found in CBCs with differential (mean, 1.7 and 1.2 (p<0.001)) (table 2). RP was the most commonly ordered among the three tests, with a mean of 3.7 per hospital stay per patient; however, the mean was identical for both years, and no significant difference was found between the two groups (p=0.23). The mean total number of laboratory tests per patient hospitalisation was 7.2. Although differences were seen in the numbers of CBC and CBC with differential orderings, the total amount of laboratory tests between the two groups was not statistically different (p=0.38).
Laboratory testing at discharge
We found a similar pattern in laboratory testing on the discharge day. Ordered CBCs increased significantly between the control and intervention groups (p=0.002), while the CBCs with differential decreased significantly (p<0.001). RP ordering again was similar between the groups (p=0.27).
Readmission, mortality rate and cost
When comparing death and readmission rates between the control and intervention groups, we found no difference in mortality rate. However, a statistically significant decrease was seen in readmission rate (p=0.008) in the intervention group. The overall mean cost for all laboratory tests was US$632.40 per patient hospitalisation, but no statistical difference occurred in the overall cost for all 3 months between the control and intervention groups (p=0.14).
We found that our quality improvement project of education and cost reminders did not significantly reduce the overall ordering of the most common daily laboratory testing for patients in the hospital on an internal medicine teaching service. To bring the highest value to the current practice of medicine, healthcare professionals need to optimise resources with highest cost effectiveness without adverse effects on health outcomes and high-quality care. For several reasons, excessive laboratory testing may represent an economic burden for the hospital and an unnecessary risk for patients. The reasons include unnecessary cost to the healthcare system and unnecessary discomfort and blood loss to the patient.14 Although laboratory testing may deliver information to providers about a patient's internal body processes to help identify and treat a wide variety of medical conditions, repeated testing in the setting of clinical stability is not indicated. In addition, cost awareness should always be encouraged and test costs considered.
Studies have demonstrated poor awareness of cost by physicians and other healthcare professionals.15 One recent study showed that residents' cost awareness can be increased through comprehensive visual displays.14 However, that study did not specifically address whether increased awareness resulted in residents changing their clinical decision-making habits or their test ordering. Our study confirmed that awareness of laboratory test cost did not change overall ordering behaviour. Yet, we find it interesting that residents were able to significantly decrease the total number of CBCs with differential ordered per patient hospitalisation but appeared to compensate with a significant increase in total CBC orders. This outcome suggests that residents may have recognised that a more expensive CBC with differential may not add incremental value to a patient's medical management. However, the interventions did not improve overall total number of daily laboratory tests ordered.
Several studies have evaluated the effect of education on healthcare costs and test ordering behaviour.7 ,11 ,16 ,17 One study performed in Australia demonstrated a decrease in the number of inappropriate tests ordered by trainees after an educational intervention and a feedback strategy were applied.7 This change in behaviour was transient, and its effect waned by the end of the semester following the intervention. Two subsequent studies had conflicting results. Bates et al 11 showed that implementation of a computerised display of test costs failed to result in a statistically significant change in test orders at a teaching institution. In contrast, a separate study using the same strategy at a non-academic institution found a 9.1% reduction in tests ordered.16 Multiple other studies have shown a high variability in results after education strategies, demonstrating that other factors may also influence ordering behaviour.7 ,17
One explanation for the large number of laboratory tests seen in our study population is the fact that our institution uses an electronic health record, and this helps facilitate the ease of ordering. After an order for a daily laboratory test is placed in the computer, it is continued automatically until it is manually stopped. In this case, tests may not be reassessed daily to determine whether they are truly needed. Because our study showed that education and cost awareness were not sufficient to change this ordering practice in residents, an alternative solution could be to modify the ordering system to have an expiration date.8 ,18 A second explanation is the fact that most ordering in our teaching services is done by interns. Lack of experience has been found to affect ordering.7 Possible contributors are insecurity in practice, supervisor pressure and lack of awareness about adverse effects and costs. Residency programmes are an apt environment for effective feedback by more experienced attending physicians, which have been shown to reduce laboratory testing in other settings.10
Our study has limitations. Although Charlson Comorbidity Index is a validated tool to measure overall comorbidities of a population,19 we could not distinguish whether there was an increase in CBC ordering because the individual patients had more serious comorbidities that required more frequent monitoring and laboratory testing. In addition, we did not independently measure the effectiveness of the educational portion of our intervention with any independent evaluation. It may be that an alternative education could have been more effective. We also did not survey the residents for their opinions on the effectiveness of the intervention or whether they believed they had changed their practice because of the intervention. This information could have added to the understanding behind the failure of change. Finally, a small change occurred in the resident personnel between the baseline measurement and the postintervention measurement, because the intervening time crossed the graduation date of the academic year. Therefore, the outgoing third year residents were no longer followed in the data, and the incoming intern class was a part of the intervention. However, the basic processes of ordering and patient care, as well as team attendings, did not change. This likely had minimal effect.
Our results support the hypothesis that a programme of education and cost awareness has no significant impact on ordering behaviour of medical residents. Future research should focus on alternative interventions if significant effect is to be made.
Neither education nor cost awareness is an effective intervention to reduce automatic ordering of daily laboratory tests by residents. Further studies should be undertaken to find tactics that reduce unnecessary test ordering.
Current research questions
Does eliminating automatic daily orders change ordering behaviour in medical residents?
What combination of interventions are more likely to produce behavioural changes?
Do incentives have an impact on medical residents’ laboratory ordering behaviour?
Contributors All authors were involved in the quality improvement project and contributed to the writing of this manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
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.