- The project carried out a systematic review of the cost of CLABSI. After reviewing almost 850 abstracts and over 150 articles in full, 6 articles met inclusion criteria.
- After weighting and adjusting to 2012 dollars, the average CLABSI cost reported in the literature was $70,696 with a range (± two standard deviations) of $40,412–$100,980.
- Assuming baseline rates would have remained stable without the study intervention, an estimated total of 2,187–2,419 CLABSIs were prevented over the course of the project.
- An estimated 290–605 deaths were prevented during the course of the project assuming a 12–25 percent mortality rate.
- An estimated $97,756,628–$244,270,620 in excess costs were averted during the course of the project.
Prior versions of this analysis reported the excess cost per CLABSI at $16,550, an estimate used by the CDC.6 However, to better assess the estimated excess costs averted as a result of the improvement project, a systematic review of the literature was conducted. Although prior systematic reviews have been conducted, this review differed in that it focused solely on the U.S. experience. As such, studies conducted outside of the United States were excluded. In addition, we assumed that the cost of treating adult versus pediatric CLABSIs differs and as such did not include studies solely examining NICU, PICU, or pediatric units.
PubMed, EconLit, Biological Abstract and Science Direct were searched. First, articles were reviewed at the abstract level. Abstracts believed to have appropriate CLABSI cost calculations were retrieved to be reviewed in full. Retrieved articles were then reviewed against inclusion criteria and included articles were abstracted (for a full list of a priori inclusion criteria, go to Appendix A). Costs identified in included studies were adjusted to reflect May 2012 dollars using the consumer price index (CPI) inpatient hospital services index.7 When authors did not indicate what dollar year their estimates were based upon, the year of publication was assumed. A clear consensus does not exist for cost adjustment.8 As such, both the all urban CPI (CPI-U) and CPI inpatient adjustments were calculated and reflected in the summary table. After adjusting to 2012 dollars, a weighted mean was calculated. The number of patients with a CLABSI (i.e., cases) was used to weight the final mean. Although more robust methodologies exist, we were limited by the detail of statistical data reported by authors. This limitation resulted in the use of two standard deviations (plus/minus) to estimate the range of costs based upon the estimated mean.
A total of 841 unique abstracts were identified. Forty-nine articles were retrieved after being deemed relevant at the abstract level, and 109 additional articles were retrieved after conducting a reference review. A total of six articles satisfied inclusion criteria and can be found in Table 11. A flowchart of the article review process can be found in Appendix B, and all excluded articles and reason for exclusion can be found in Appendix C.
|Al-Rawajfah et al.||2012||Length of stay and charges associated with health care acquired bloodstream infections|
|Butler et al.||2010||Attributable costs of enterococcal bloodstream infections in a nonsurgical hospital cohort|
|Cohen et al.||2010||Cost saving from reduced catheter-related bloodstream infection after simulation-based education for residents in a medical intensive care unit|
|Kilgore et al.||2008||Cost of bloodstream infections|
|Shannon et al.||2006||Economics of central line-associated bloodstream infections|
|Warren et al.||2006||Attributable costs of catheter-associated bloodstream infections among intensive care patients in a nonteaching hospital|
Although stringent inclusion/exclusion criteria were applied, the six included articles differed significantly in both estimated costs as well as the methods utilized to derive the costs. Variation can be attributed to factors such as sample size (range: 12-100,851 cases), setting (single hospital site estimate in four studies vs. multi-site estimate), costs considered (e.g., costs billed, actual costs, etc.), and estimation model utilized (e.g., matching procedure). Although a weighted mean was calculated, only one study utilized a large, national administrative database, thereby resulting in greater cases and a larger weight in the averaging process.9
Average costs per CLABSI in each study as well as relevant CPI adjustment weights and adjusted costs can be found in Table 12. The average cost per CLABSI after using the weight and May 2012 CPI inpatient hospital service index adjustment was estimated at $70,696 with a range (plus/minus two standard deviations) of $40,412 to $100,980. This range is similar to that found on the Johns Hopkins CLABSI Opportunity Estimator Web site (range of $40,000 to $117,000 per infection).10
Table 12. Cost per CLABSI of studies meeting inclusion criteria*
|Author||Cases†||Study Mean||Cost in Year||CPI-U Weight||CPI-U
|CPI IP Weight||CPI I
|Al-Rawajfah et al. (2012)||100,851||$68,067.00||2010||1.05||$71,326.16||1.12||$76,093.08|
|Butler et al. (2010)||276||$2,858.87||2007||1.10||$3,150.94||1.39||$3,976.12|
|Cohen et al. (2010)||12||$82,005.00||2008||1.06||$87,061.12||1.30||$106,443.30|
|Kilgore et al. (2008)||12,578||$19,643.00||2006||1.13||$22,271.25||1.48||$29,041.85|
|Shannon et al. (2006)||54||$40,179.00||2006||1.13||$45,554.98||1.48||$59,404.01|
|Warren et al. (2006)||41||$11,971.00||2000||1.33||$15,887.51||2.18||$26,050.71|
CPI-U = all urban consumer price index; CPI IP = consumer price index inpatient hospital services index
To estimate the deaths prevented, a range in mortality was assumed (12-25 percent).11 For each 100 CLABSIs prevented, 12-25 deaths are prevented. To estimate the excess costs averted, the mean cost per CLABSI (using CPI inpatient hospital service adjustment) was utilized: $70,696. Thus, for each CLABSI prevented, $70,696 in treatment costs were averted. To estimate the number of CLABSIs prevented, the number of actual CLABSIs reported was compared with the number of CLABSIs that would have occurred if the pre-intervention ("baseline") rate of CLABSIs per 1,000 line days had persisted:
# CLABSIs prevented = (baseline rate – observed rate) * (#central line days)
These calculations were done on a quarterly basis and totaled over all quarters.
The project quarters considered were eight quarters post-intervention, or 24 months. Not all units reported data for all 24 months (to be distinguished from units that reported zero central line days). Thus, it is useful to calculate savings based not only on months of reported data, but under the assumption that for each unit, non-reported months were similar to reported months. In order to estimate the total number of central line days (CLDs) and CLABSIs over the project quarter, these were interpolated for each unit using the available reported data. Mean interpolation was used. For example, if one unit did not report data for the eighth month, the numbers of CLDs and CLABSIs for that month were estimated as the average of the numbers for the seventh and ninth months. If the month with missing data was at the end of the project quarter (i.e., month 24), the numbers of the last available month were carried forward, and if at the beginning of the project quarter (i.e., month 1), the numbers of the first available month were carried back (go to Appendix D for an example). Since CLABSI rates declined over the course of the intervention, using the study average would underestimate projected savings. Thus, data from quarters five and six were used to estimate monthly costs and lives saved over time after the project. Estimates are based on all adult ICU units that participated in the project and had not formally withdrawn. Units that never submitted data but did not formally withdraw were excluded.
Table 13 presents estimated CLABSIs and deaths prevented and excess costs averted using all available data as well as missing data imputation. An estimated total of 2,187 CLABSIs were prevented over the course of the study with a projected continuation of 114 CLABSIs prevented monthly moving forward. After missing values were imputed, this number increased to an estimated total of 2,419 CLABSIs prevented over the course of the study with a projected continuation of 121 CLABSIs prevented monthly.
Estimates of deaths prevented varied as a function of the underlying assumption: 12 percent mortality rate to 25 percent mortality rate. Using all available data, an estimated range of 262–547 deaths were prevented over the course of the study with 14–28 deaths prevented monthly moving forward. After missing value imputation, this number increased to an estimated range of 290–605 deaths prevented over the course of the study with 14–30 deaths prevented monthly moving forward.
A range of estimates for excess costs averted can be found in Table 14. Using all available data and the CPI inpatient hospital service adjusted weighted cost, an estimated $154,612,152 in excess costs were averted with an estimated $8,057,594 in excess costs averted monthly moving forward. After missing value imputation, this number increased to $171,013,624 in excess costs averted with an estimated $8,533,375 in excess costs averted monthly moving forward. In previous reports, using the estimate of $16,550 per CLABSI the estimated excess costs averted would range from $36,194,850 to $40,034,450 (Table 13).
Since the cost estimate is based on an estimated mean, a more conservative approach would be to consider a range of costs. Using all available data and a range that is two standard deviations from the mean, excess costs averted are estimated at between $88,381,044 and $220,540,320. Every month moving forward this translates into an estimated cost savings range of $4,606,015 to $11,509,172. After imputing missing values the total range estimate becomes $97,756,628 to $244,270,620 with every month moving forward estimated at averting $4,877,989 to $12,188,760 in costs.
It should be noted that the finalized estimate of costs is from the hospital perspective, not the patient. As such, excess costs averted such as wages lost were not considered. In addition, estimates do not include costs of death in terms of the value of statistical life which would increase the overall economic impact of the project (see the National Center for Environmental Economics for more information12). Finally, estimated costs do not reflect reductions in CLABSI rates found in non-ICU and pediatric units participating in the project.
Table 13. Estimation of infections and deaths prevented and excess costs averted among participating adult ICUs using all data ("reported") and data after missing value imputation ("estimated") using CDC estimates
|Deaths Prevented||Excess Costs
† CLABSI Prevented represents the sum of Q1 to Q8 and does not include savings outside the timeframe.
‡ Center of Disease Control and Prevention Morbidity and Mortality Weekly Report (Go to: http://www.cdc.gov/mmwr/pdf/wk/mm60e0301.pdf) [Plugin Software Help].
Table 14. Estimation of excess costs averted among participating adult ICUs using all data ("reported") and data after missing value imputation ("estimated")
|CLABSIs Reported/ Estimated||Assumed
|Excess Costs Averted|
6. Morbidity and Mortality Weekly Report (MMWR) March 2011. Vital Signs: Central Line-Associated Blood Stream Infections—United States, 2001, 2008, and 2009. Accessed on: October 18, 2012. Access at: http://www.cdc.gov/mmwr/pdf/wk/mm6008.pdf [Plugin Software Help].
7. Bureau of Labor Statistics, Consumer Price Indexes. Accessed on: October 18, 2012. Access at: http://www.bls.gov/cpi/
8. Scott RD. The Direct Medical Costs of Healthcare-Associated Infections in U.S. Hospitals and the Benefits of Prevention—Centers for Disease Control and Prevention. 2009 Accessed on: October 18, 2012. Access at: http://www.cdc.gov/hai/pdfs/hai/scott_costpaper.pdf [Plugin Software Help].
9. Al-Rawajfah OM, et al. Length of stay and charges associated with health care-acquired bloodstream infections. Amer J Infec Contr 2012;40:227-232.
10. Johns Hopkins CLABSI Opportunity Estimator. Accessed on October 18, 2012. Access at: http://www.hopkinsmedicine.org/quality_safety_research_group/our_projects/
11. Morbidity and Mortality Weekly Report (MMWR) March 2011. Vital Signs: Central Line-Associated Blood Stream Infections—United States, 2001, 2008, and 2009. Accessed on: October 18, 2012. Accessed at: http://www.cdc.gov/mmwr/pdf/wk/mm60e0301.pdf [Plugin Software Help].
12. National Center for Environmental Economics: http://yosemite.epa.gov/ee/epa/eed.nsf/webpages/homepage.