The incremental cost of infections associated with multidrug-resistant organisms in the inpatient hospital setting--A national estimate.
Antibiotic-resistant bacteria pose a major public health threat. The US Centers for Disease Control and Prevention (CDC) estimates that over 2 million individuals become infected with bacteria resistant to antibiotics and that multidrug-resistant organisms (MDROs) account for 23 000 deaths per year. (1) MDROs are microorganisms that are resistant to one or more antimicrobial agents and are most often resistant to all but one or two antimicrobial agents. (2) Most recently, a Thorpe and colleagues find that the share of bacterial infections that are associated with MDROs doubled between 2002 and 2014 in the United States. (3) The last nationally representative hospital study found that hospitalizations with diagnoses for drug resistance increased 359 percent over the period 1997-2006. (4) The rise in MDROs is linked, in part, to inappropriate antibiotic prescribing both in the inpatient and in the outpatient settings.(1,5) A recent study of national trends in antibiotic use during hospitalization found large increases in vancomycin and broad-spectrum antibiotic agents over the period 2006-2012. (6) To combat MDROs, the US government created national strategies, including efforts to decrease the transmission of MDROs in the hospital and the community, to increase the longevity of the current antibiotic supply, and to accelerate research and development of new antimicrobial agents. (7-9)
In this study, we use the largest publicly available all-payer hospital database in the United States to perform the first nationally representative estimate of the hospital-based health care costs associated with MDROs. This is an observational study using hospital discharge data to retrospectively identify patients hospitalized with bacterial infection and estimate the additional cost attributable to infections associated with MDROs among such patients. The counterfactual is represented by patients hospitalized with bacterial infections without diagnosed MDROs. We derive our estimates by comparing hospital stays for bacterial infection patients with and without diagnosed MDRO infections and then using multivarible regression to assess the effect of MDRO infections on the cost of hospitalization after controlling for other patient, clinical, and hospital characteristics. We use the same approach to assess the effect on hospital length of stay. Lastly, we estimate the national incidence of infections associated with MDROs during inpatient hospitalization and combine this with our regression results to estimate the total national costs annually.
2 | BACKGROUND
While the effects of MDROs on morbidity and mortality have been frequently evaluated, little is known about their incremental cost. In its Antibiotic Threat Report, the CDC estimates annual US health care costs from infections due to MDROs at $20 billion, (1) based in part on one of the few available estimates in the literature. In that study, Roberts and colleagues estimated the costs attributable to MDROs in a sample of 1391 hospitalized adults at a single urban public teaching hospital in the year 2000 to be $18 588-$29 069 per patient stay. (10) Because this estimate compared patients with infection with MDROs to patients without any infection and may not be generalizable, it may not accurately represent the true economic burden of MDROs.
A recent study by Thorpe and colleagues estimated the incremental costs attributable to MDROs using data from the Medical Expenditure Panel Survey (MEPS). (11) The authors found that the incremental cost of treating patients with MDRO infections was $1383 in total annual inpatient, outpatient/professional, and pharmaceutical costs per patient over and above the cost of treating patients with bacterial infections who did not show evidence of having MDROs. However, MEPS does not include institutionalized patients, such as patients living in long-term care and nursing homes, and prior research has shown the incidence of MDROs in the institutionalized population to be much higher than that among the general population. (12)
3 | METHODS
3.1 | Data and sample
We conducted a retrospective observational study using cross-sectional data from the National Inpatient Sample (NIS) provided by the Healthcare Cost and Utilization Project and the Agency for Healthcare Research and Quality. (13) The NIS is an annual stratified nationally representative sample of 20 percent of all discharges from US non-Federal, short-term, general, and other specialty hospitals (excluding rehabilitation and long-term care) and includes weights to generate national estimates.(13) The NIS is publicly available and contains de-identified data that are exempt from institutional review board approval. In 2014, the NIS comprised 7.1 million inpatient hospital discharge records (eg, patient demographics, diagnosis and procedure codes, hospital stay characteristics) weighted to represent 35.4 million hospital discharges. We also linked the American Hospital Association hospital information (eg, region, bed size) supplied by HCUP to the discharge records. We used data for the year 2014 and computed national estimates for all results using the inpatient stay discharge weights.
We identified patients with a discharge diagnosis of one or more of the bacterial infections listed in Table 1 during their inpatient stay using ICD-9-CM codes. Similar approaches have been previously validated for identification of patients with bacterial infection during inpatient hospitalization. (14-17)
3.2 | Measuring inpatient hospitalization costs and length of stay
To measure hospitalization costs, we converted the reported charges to costs for each inpatient stay included in our study using the HCUP cost-to-charge ratio (CCR) files. (18) These costs are equivalent to the amount covered by all payers for the hospital facility charges, but not including professional or physician fees. We measured inpatient length of stay as the number of days in the hospital from admission to discharge.
3.3 | Identifying multidrug-resistant organisms
We assigned patients to infections with one of four MDROs using ICD-9-CM discharge diagnosis codes (see Table 1). (a) We identified patients with diagnoses for methicillin-resistant Staphylococcus aureus (MRSA), (b) We identified patients with diagnoses for Clostridium difficile. Previous studies find an acceptable level of sensitivity (75-76 percent) for the use of ICD-9-CM codes to identify MRSA and C. difficile in the hospital setting, (19,20) (c) We identified patients infected with other MDROs using the "V09.x" diagnosis codes for drug-resistant infections. Although these codes have been used in prior studies, (4) they have also shown poor accuracy. (19) Our analysis of Burnham and colleagues' recent study shows that "V09.x" ICD-9-CM codes had a (weighted) sensitivity of 11.9 percent vs the gold standard of laboratory cultures in accurately identifying patients with MDROs during hospitalization. (19) Similarly, a nationally representative study by Thorpe and colleagues of patients with MDROs identified 12 percent of such MDRO patients via ICD-9-CM codes and the other 88 percent of MDRO patients via evidence of antibiotic drug treatment failure. (3) These numbers are close to what we would expect if the numbers of undiagnosed cases of MDROs from Burnham and colleagues' study were generalizable to Thorpe and colleagues' nationally representative sample. Burnham and colleagues' study suggests that for every 1 accurately coded case there are 8.4 additional cases that have not been identified. We used this result in our analysis to extrapolate an upper bound estimate of such hospitalizations, (d) If patients had diagnoses for more than one of these first three resistant pathogens, we assigned them to a separate fourth group of multiple MDROs. Our fifth group was the comparator, patients with infections without any identified MDROs.
3.4 | Control variables
In order to adjust for factors that may influence bacterial infection patients' costs or length of stay apart from having an MDRO, we controlled for potentially confounding characteristics of the patient, inpatient course of stay, and hospital. First, we controlled for patient demographics, socioeconomic status, and insurance status. Patient demographics were measured by age, sex, and race/ethnicity. We proxied for socioeconomic status using the median household income quartile of each patient's zip code. We controlled for insurance status using primary payer type on the discharge record. Second, we controlled for clinical characteristics of the inpatient stay, including type of bacterial infection, the Elixhauser Comorbidity index, (21,22) the all patient diagnosis-related group (AP-DRG) severity index, inpatient admission source, transfer of patient to another acute care facility, performance of a major surgical/operating procedure, and whether the stay involved maternal or neonatal conditions. Lastly, we controlled for characteristics of the hospital, including US Census Region where the hospital is located, bed size, ownership, urban vs rural setting, and teaching status.
3.5 | Analytic approach
First, we compared hospital stays for patients across the four identified MDROs as well as for patients with no MDROs. We calculated descriptive statistics on hospitalization costs as well on the characteristics of these patients, their inpatient course of stay, the hospitals they were discharged from, and the type of bacterial infection with which they were diagnosed. We used Wald and chi-square tests to assess whether there were statistically significant differences (P < 0.05, two-tailed) between the five mutually exclusive groups.
Next, we estimated two linear regression models to assess the incremental effect of each of the four MDROs on hospitalization costs using a generalized linear model with a gamma distribution and log link function. (23) The first model assessed the unadjusted effect. The second model assessed the adjusted effect on hospitalization costs after controlling for characteristics of the patient, inpatient course of stay, hospital, and bacterial infection type. These results are reported as the additional cost per inpatient stay due to having an infection associated with an MDRO at the population mean among all patients hospitalized with bacterial infection, holding other factors constant. We adjusted our estimates and standard errors to account for the NIS sampling methodology using the Stata survey commands.
In a secondary analysis, we assessed the effect of the four MDROs on hospital length of stay (LOS) by estimating negative binomial regression models to get the unadjusted and adjusted estimates (as in the previous analysis). Results are reported as the additional number of days spent in the hospital per inpatient stay due to having an infection associated with an MDRO at the population mean among all patients hospitalized with bacterial infection.
Finally, we estimated the total annual national inpatient hospitalization costs for each of the four MDROs. Our first step involved estimating the national annual incidence of these MDROs in 2014 by applying the NIS sampling weights to all inpatient stays with one or more diagnoses for an MDRO. Because prior research shows that patients with MDROs are often undercoded with the appropriate "V09.x" codes, we extrapolated an upper bound incidence estimate using the multiplier of 8.418 derived from our analysis of Burnham and colleagues recent study. (19) Next, we used the result from our adjusted cost regression models to generate national cost-per-stay estimates for each of the four MDROs with 95% confidence intervals. We then generated national cost estimates by taking the product of our estimates for incidence and cost per stay, and inflation-adjusting this amount to 2017 US dollars (USD).
In a sensitivity analysis to evaluate the effect of missing variables on our estimates, we calculated descriptive statistics on the inpatient stays with bacterial infection excluded from our study. Specifically, we calculated summary statistics for all variables with observations containing nonmissing values in the excluded sample and compared them on the same variables with the study population.
We performed data analysis using SAS version 9.4 and STATA version 14.
4 | RESULTS
4.1 | Characteristics of the study population
Of the 35.4 million hospital discharges represented in the NIS in 2014, we identified 7 093 157 inpatient stays with diagnosed bacterial infections eligible for inclusion in the study (Appendix S1). We excluded 707 899 inpatient stays (10.0 percent of eligible stays) that were missing data on costs, discharge diagnosis, or other variables included in our analysis. Our final study population consisted of 6 385 258 inpatient stays.
There were 312 385 inpatient stays in our study population with MRSA, 312 310 with C. difficile, 52 590 with another MDRO ("V09.x"), 16 335 with multiple MDROs, and 5 691 637 that show no evidence of any MDROs (Table 2). Inpatient stays with MRSA, C. difficile, another MDRO, and multiple MDROs cost $5253, $7050, $4943, and $21 422 more on average, respectively, than stays with no evidence of any MDROs (P < 0.001). There are statistically significant (P < 0.001) differences between the five groups on all characteristics measured. For example, MRSA is much more common in cellulitis infections that the other MDROs. In addition, other MDROs (V09.x code diagnoses) are much more likely to be found in females, possibly because these other MDROs are more common in urinary tract infections, which are also more common in females. Lastly, it is noteworthy that multiple MDROs are more likely in patients transferred from another acute care facility or in patients with bacteremia or sepsis.
4.2 | Effect of multidrug-resistant organisms on inpatient hospitalization costs
Our adjusted models find that MRSA, C. difficile, another MDRO, and the presence of more than one MDRO are associated with $1718 (95% CI, $1609-$1826), $4617 (95% CI, $4407-$4827), $2302 (95% CI, $2044-$2560), and $3570 (95% CI, $3019-$4122) in additional costs per hospital stay, respectively (Table 3). The mean cost per hospital stay for stays with any diagnosis of bacterial infection is $19 037. Thus, we find that the relative effect of having an MDRO on a hospital stay with a bacterial infection is a 9 percent cost increase for MRSA, a 24 percent cost increase for C. difficile, a 12 percent cost increase for another MDRO, and a 19 percent cost increase for more than one MDRO.
4.3 | Effect of multidrug-resistant organisms on inpatient hospital length of stay
Infection associated with MRSA, C. difficile, another MDRO, and the presence of more than one MDRO is associated with 1.04 (95% CI, 0.99-1.09), 2.62 (95% CI, 2.52-2.73), 1.31 (95% CI, 1.20-1.42), and 1.53 (95% CI, 1.29-1.76) additional days in hospital per stay, respectively, in our adjusted models (Table 4).
4.4 | National incidence and cost estimates of multidrug-resistant organisms
An estimated 7 093 157 inpatient stays each year include a diagnosis for bacterial infection--20.1 percent of all stays (Table 5). The national incidence of inpatient stays with bacterial infection that also show evidence of MDROs is 57 865 in our lower bound estimate for infection with another MDRO ("V code" diagnoses) and 487 108 in our upper bound estimate. (The upper bound estimate is our estimate that includes the undercoded MDRO "V code" diagnoses.) We find a national incidence rate of 343 220 for MRSA, 347 230 for C. difficile, and 17 930 for multiple MDROs. This suggests that at least 10.8 percent of stays with bacterial infection--and perhaps as many as 16.9 percent if we account for undercoded infections--show evidence of one or more MDROs. Using the results of our adjusted models for additional costs-per-stay, we estimate the national cost of stays with other MDROs ("V code" diagnoses) to be $133 205 230 (95% CI, $118 276 078-$148 134 423) in our lower bound estimate and $1 121 321 626 (95% CI, $995 648 752-$l 246 996 480) in our upper bound estimate. We find national costs of $589 651 960 (95% CI, $552 241 221-$626 719 994) for MRSA, $1 603 160 910 (95% CI, $1 530 243 183-$1 676 079 838) for C. difficile, and $64 010 100 (95% CI, $54 130 700-$73 907 501) for multiple MDROs. As a result, we estimate the national cost of infections associated with MDROs during inpatient hospitalization to be at least $2.39 billion (95% CI, $2.25-$2.52 billion) and as high as $3.38 billion (95% CI, $3.13-$3.62 billion) if we account for infections undercoded for MDROs.
4.5 | Sensitivity analysis
The results of our sensitivity analysis with descriptive statistics on the inpatient stays with bacterial infection that were excluded from our study due to missing data are shown along with comparisons to our study population in Appendix S2. The excluded inpatient stays we had comparison data on were higher in average costs per stay and more likely to be from hospitals in the West or Midwest as well as from urban teaching hospitals and private nonprofit hospitals. In addition, the excluded inpatient stays were less likely to be emergency department admissions and more likely to be planned admissions or transfers in from another acute care hospital. Furthermore, the patients were more likely to be younger, female, African American or Hispanic, insured by Medicaid or private insurance, and in the upper income quartile. However, we do not see much difference in prevalence of MDROs or in risk scores on the Elixhauser Comorbidity Index or AP-DRG Severity Index between the excluded sample and the study sample. Taken together, these results suggest that the higher costs we see in the excluded sample are not due to clinical factors related to MDROs or disease severity but rather due to regional and hospital-specific cost factors. As a result, the most likely effect of excluding these hospital stays from our study is simply to lessen the base cost of bacterial infection stays without substantially impacting the incremental cost estimate of MDROs.
5 | DISCUSSION
In a large nationally representative population of patients hospitalized in US hospitals with bacterial infection, MDROs are associated with $1718-$4617 in additional costs per inpatient stay and an additional 1.04-2.62 days spent in the hospital compared to patients with a bacterial infection but no MDROs. Approximately 20.1 percent of US hospital stays involve a bacterial infection and a further 10.8-16.9 percent of these stays show evidence (colonization or infection) of an MDRO during that hospitalization. We estimate the national inpatient hospitalization cost of infections associated with MDROs to be approximately $2.4 billion to $3.4 billion annually. As a consequence, infections associated with MDROs during hospitalization result in a substantial cost burden to the entire US health care system beyond the cost of infections not associated with MDROs.
We find the highest incremental and total costs for C. difficile and the lowest incremental costs for MRSA, consistent with estimates from previous reports. (24) The higher costs appear to be driven largely by a higher average LOS, but may also be due to additional testing and increased risk for ICU admission with C. difficile. Prior studies have also shown C. difficile infections to be associated with prolonged length of hospital stay. (25,26) The lowest total costs for multiple MDROs may have been driven by the low incidence rate we find for these infections, possibly due to the undercoding of other MDROs identified by the "Vcode" diagnoses, and the inability to distinguish infection from colonization from the data. (19)
Our incremental cost estimates of $1718 to $4617 for the four MDROs per inpatient stay (compared to patients with a bacterial infection but no MDROs) are more conservative than the case study by Roberts and colleagues that estimated the cost of MDROs per stay to be $18 588 to $29 069. (10) There are three major reasons for this observed difference. First, we take the payer perspective and estimate total costs as the sum of expenditures for health care treatment. Roberts et al took the hospital and societal perspective, estimating hospital costs as patient resource use multiplied by unit costs and societal costs as lost productivity due to excess mortality and time in hospital. Second, the authors' higher estimates represent the difference in costs between the average hospitalized patient and a patient with an infection associated with an MDRO. However, our estimates reflect the incremental cost difference between the average patient hospitalized with bacterial infection and those with a bacterial infection associated with an MDRO, treating the cost of bacterial infection as a "sunk cost." Roberts et al did not limit their study population to patients with bacterial infection, and in their regression models controlled for health care-associated infection, not any bacterial infection, so potentially patients who did not develop an infection were not at risk of developing an infection associated with an MDRO. Additionally, the comparison group in the Roberts et al study were likely healthier than the comparator patients in our study because our comparison patients all had bacterial infections. Evidence for this can be seen in the magnitude of the mean differences in cost and length of stay between patients with and without MDROs in the authors' study vs the much smaller differences in our own (see Table 2). Third, Roberts and colleagues evaluated a sample of 188 adult patients with MDROs and 1203 such patients without MDROs in a single urban public hospital in the year 2000. Our study is larger, assesses different patient populations, and provides a more contemporary perspective since there have been changes in treatment of MDROs. For instance, the average hospital length of stay for infections associated with MDROs declined by approximately 50 percent over the years 2000-2006, and patients hospitalized with infections associated with MDROs gradually became younger and healthier during this time period. (4) As a result, it is likely that the cost of the average patient hospitalized with an associated MDRO declined substantially in the intervening years between Roberts and colleagues' study and our own. In Roberts et al's study, the average length of stay among patients hospitalized with MDROs was 24.2 days, compared to 9.5-15.8 days in our study (see Table 2). In addition, our finding of a mean length of stay of 9.5 days for patients hospitalized with MRSA, for example, is very close to prior findings of 9 and 10 days in hospital. (27,28)
Our incremental cost estimates are, however, higher than the most recent national study by Thorpe and colleagues showing annual incremental MDRO costs of $1383 per patient in total costs and $1275 in hospital costs. (11) Thus, we find that incremental inpatient hospitalization costs for patients with infections associated with MDROs are 1.35-3.62 times greater per patient than that estimated by the most recent national study. This, however, is not surprising considering that study did not include institutionalized and nursing home patients, and these patients have higher rates of the most costly MDRO infection types, such as C. difficile. Taken together, these results suggest the $2.2 billion annual national cost estimate by Thorpe and colleagues is conservative.
Despite the fact that our estimates here are more conservative than the prior study by Roberts and colleagues, MDROs nonetheless add a substantial amount to the cost of hospitalizing patients with bacterial infection and to national health care costs. These are important considerations for policy makers budgeting and planning for efforts to prevent MDROs and develop new antimicrobial agents. (7-9) However, the effort to combat MDROs on a national level is farranging and brings together multiple sectors of the health system and society. These efforts include surveillance of MDRO patterns in health care settings and the community, (1,5,7,29) antimicrobial stewardship by limiting the use of antibiotics in human and animal populations, (1,5,7,9,29) prevention of infections and the spread of MDROs in health care and community settings, (1,2,5,29) precautions to limit exposure when MDROs are detected, (2,29) education of patients and physicians on the risks of inappropriate antibiotic use and on recognition of the signs of infection, (2,9) and lastly, development of new MDRO diagnostic tests as well antibiotics and vaccines to fight MDROs. (1,5,7) There is no doubt that the cost to society and the health care system to combat MDROs is great. Unfortunately, current data on the economic cost to combat MDROs are limited and weak. (30) A systematic review of the literature finds that the cost of measures to combat and eradicate MDROs ranges from $331 to $66 772 US Dollars ([euro]285 to [euro]57 532 Euros) per MDRO-positive patient. (30) As a result, more research is needed on the national cost of efforts to combat and prevent MDROs in the United States and whether such efforts are indeed cost-effective.
While this study highlights the substantial cost attributable to hospitalizing patients with MDROs, there are additional costs not included in our estimates. More research is needed to estimate the incremental costs of postacute, outpatient, professional, and pharmaceutical treatment for patients with MDROs. It is also worth noting that we did not include professional or physician fees in our calculation of the hospitalization costs of MDROs. As a result, our hospitalization cost estimate is conservative. Further research is also needed to identify factors associated with development of MDROs in patients as well as to identify optimal physician care processes that effect better outcomes for patients hospitalized with MDROs. Prior research has shown that involvement of infectious disease specialists and use of evidence-based care processes are associated with lower mortality, fewer readmissions, reduced length of stay, and lower costs for patients hospitalized with infection. (31,32)
Our results are subject to several limitations. First, we rely on ICD-9-CM diagnosis codes to identify patients hospitalized with MDROs instead of laboratory results on bacterial cultures. There are known limitations with this approach. (19,20) However, the accuracy of ICD-9-CM codes for identifying MRSA and C. difficile in the hospital setting is acceptable, (19,20) and we make use of recent research on the sensitivity of "V09.x" ICD-9-CM codes to extrapolate an upper bound estimate for infections with MDROs. Nonetheless, we believe our cost estimates are still likely to be conservative due to missing MDRO diagnoses in hospital discharge records. Second, and relatedly, our counterfactual is represented by patients hospitalized with bacterial infections without diagnosed MDROs. However, due to undercoding of MDROs in the data the counterfactual would include undiagnosed MDROs. The effect of this on our estimates is to lessen the apparent cost differences between patients with bacterial infection but no MDROs and those with MDROs; in other words, we believe our cost estimates for MDROs are conservative. Third, we measure costs for each hospitalization by converting charges to costs because exact cost data are not available; however, we use the HCUP CCR files that are based on actual cost data submitted by the hospitals to CMS and have previously shown good internal validity. (18) Fourth, we are unable to distinguish hospital-acquired infections from community-acquired infections because the presenton-admission flag is missing from the NIS hospital discharge data. However, for our purposes it was not important to identify where the MDRO infection was acquired because we were interested in the total infection costs. Lastly, we excluded 10.0 percent of our eligible population due to missing data. To the extent excluded observations are systematically different from the rest of the population our results may not generalize to this segment of the population. However, the results of our sensitivity analysis suggest that the most likely effect of excluding these hospital stays from our study is simply to lessen the base cost of bacterial infection stays without substantially impacting the incremental cost estimate of MDROs.
6 | CONCLUSION
In a large nationally representative population, we find that 22-34 out of every 1000 inpatient stays at US hospitals are for infections associated with MDROs, which carry significant incremental hospital costs per stay and length of stay.
Joint Acknowledgment/Disclosure Statement: We thank Peter Joski of Emory University for providing advice on data analysis and Scott Fridkin, MD, for his thoughtful review of the draft manuscript. This study was deemed exempt from review by the Saint Louis University Institutional Review Board (letter on file).
CONFLICT OF INTEREST
K.J.J. holds an academic appointment at SLUCOR. K.E.T. serves as Chairman of the Partnership to Fight Chronic Disease. D.J.M. No conflict. J.T.J. No conflict.
Kenton J. Johnston https://orcid.org/0000-0002-7760-203X
(1.) Centers for Disease Control and Prevention. Antibiotic resistance threats in the United States, 2013. 2013. http://www.cdc.gov/drugresistance/threat%0A-report-2013/. Accessed February 2, 2017.
(2.) Health Research and Educational Trust. Preventing MDRO Infections. 2017 Update. 2017:1-28.http://www.hret-hiin.org/Resources/mdro/17/multidrug-resistant-organism-mdro-change-package.pdf. Accessed October 1, 2018.
(3.) Thorpe KE, Joski P, Johnston KJ. Antibiotic-resistant infection treatment costs have doubled since 2002, now exceeding $2 billion annually. Health Aff. 2018;37(4):662-669.
(4.) Mainous AG, Diaz VA, Matheson EM, Gregorie SH, Hueston WJ. Trends in hospitalizations with antibiotic-resistant infections: U.S., 1997-2006. Public Health Rep. 2006;126(3):354-360.
(5.) Marston HD, Dixon DM, Knisely JM, Palmore TN, Fauci AS. Antimicrobial resistance. JAMA. 2016;316(11):1193.
(6.) Baggs J, Fridkin SK, Pollack LA, Srinivasan A, Jernigan JA. Estimating national trends in inpatient antibiotic use among US hospitals from 2006 to 2012. JAMA Intern Med. 2016;30333:1-10.
(7.) President's Council of Advisors on Science and, Technology. Report to the President on combating antimicrobial resistance. 2014. https://www.cdc.gov/drugresistance/pdf/report-to-the-presi-dent-on-combating-antibiotic-resistance.pdf. Accessed February 2, 2017
(8.) The White House. National action plan for combating antibiotic-resistant bacteria. 2015. https://www.cdc.gov/drugresistance/pdf/national_action_plan_for_combating_antibotic-resistant_bacteria. pdf. Accessed February 2, 2017.
(9.) Mehrotra A, Linder JA. Tipping the balance toward fewer antibiotics. JAMA Intern Med. 2016;176(11):1649-1650.
(10.) Roberts RR, Hota B, Ahmad I, et al. Hospital and societal costs of antimicrobial-resistant infections in a Chicago teaching hospital: implications for antibiotic stewardship. Clin Infect Dis. 2009;49(8):1175-1184.
(11.) Thorpe KE, Joski P, Johnston KJ. National health care costs for treating antibiotic-resistant infections exceed $2 billion annually and have doubled over the past 12 years. Health Aff. 2018;37(4):662-669.
(12.) Aliyu S, Smaldone A, Larson E. Prevalence of multidrug-resistant gram-negative bacteria among nursing home residents: a systematic review and meta-analysis. Am J Infect Control. 2017;45(5):512-518.
(13.) Healthcare Cost and Utilization Project (HCUP). HCUP National Inpatient Sample (NIS). Rockville, MD: Agency for Healthcare Research and Quality; 2012. https://www.hcup-us.ahrq.gov/nisoverview.jsp. Accessed January 26, 2017
(14.) Schneeweiss S, Robicsek A, Scranton R, Zuckerman D, Solomon DH. Veteran's affairs hospital discharge databases coded serious bacterial infections accurately. J Clin Epidemiol. 2007;60(4):397-409.
(15.) Barber C, Lacaille D, Fortin PR. Systematic review of validation studies of the use of administrative data to identify serious infections. Arthritis Care Res (Hoboken). 2013;65(8):1343-1357.
(16.) Warren DK, Nickel KB, Wallace AE, Mines D, Fraser VJ, Olsen MA. Can additional information be obtained from claims data to support surgical site infection diagnosis codes? Infect Control Hosp Epidemiol. 2014;35(S3):S124-S132.
(17.) Wilson SE, Turpin RS, Hu XH, Sullivan E, Mansley EC, Ma L. Does initial choice of antimicrobial therapy affect length of stay for patients with complicated intra-abdominal infections? Am Surg. 2005;71(10):816-820.
(18.) Healthcare Cost and Utilization Project (HCUP). HCUP Cost-to-Charge Ratio Files (CCR). Rockville, MD: Agency for Healthcare Research and Quality; 2013. https://www.hcup-us.ahrq.gov/db/state/costtocharge.jsp. Accessed January 27, 2017.
(19.) Burnham JP, Kwon JH, Babcock HM, Olsen MA, Kollef MH. ICD-9-CM coding for multidrug resistant infection correlates poorly with microbiologically confirmed multidrug resistant infection. Infect Control Hosp Epidemiol. 2017;38:1381-1383.
(20.) Goto M, Ohl ME, Schweizer ML, Perencevich EN. Accuracy of administrative code data for the surveillance of healthcare-associated infections: a systematic review and meta-analysis. Clin Infect Dis. 2014;58(5):688-696.
(21.) Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(l):8-27.
(22.) Moore BJ, White S, Washington R, Coenen N, Elixhauser A. Identifying increased risk of readmission and in-hospital mortality using hospital administrative data. Med Care. 2017;55(7):698-705.
(23.) Buntin MB, Zaslavsky AM. Too much ado about two-part models and transformation? Comparing methods of modeling Medicare expenditures. J Health Econ. 2004;23(3):525-542.
(24.) Zhang S, Palazuelos-Munoz S, Balsells EM, Nair H, Chit A, Kyaw MH. Cost of hospital management of Clostridium difficile infection in United States--a meta-analysis and modelling study. BMC Infect Dis. 2016;16(1):447.
(25.) Mitchell BG, Gardner A. Prolongation of length of stay and Clostridium difficile infection: a review of the methods used to examine length of stay due to healthcare associated infections. Antimicrob Resist Infect Control. 2012;1(1):14.
(26.) Magee G, Strauss ME, Thomas SM, Brown H, Baumer D, Broderick KC. Impact of Clostridium difficile-associated diarrhea on acute care length of stay, hospital costs, and readmission: a multicenter retrospective study of inpatients, 2009-2011. Am J Infect Control. 2015;43(11):1148-1153.
(27.) Cosgrove SE, Qi Y, Kaye KS, Harbarth S, Karchmer AW, Carmeli Y. The impact of methicillin resistance in Staphylococcus aureus bacteremia on patient outcomes: mortality, length of stay, and hospital charges. Infect Control Hosp Epidemiol. 2005;26(02):166-174.
(28.) Elixhauser A, Steiner C. Infections with methicillin-resistant Staphylococcus aureus (MRSA) in U.S. hospitals, 1993-2005. HCUP Stat Br #35. 2007;ll(6):868-872. http://www.hcup-us.ahrq.gov/reports/statbriefs/sb35.jsp.
(29.) DAgata EMC, Horn MA, Ruan S, Webb GF, Wares JR. Efficacy of infection control interventions in reducing the spread of multidrug-resistant organisms in the hospital setting. PLoS ONE. 2012;7(2):e30170.
(30.) Birgand G, Moore LSP, Bourigault C, et al. Measures to eradicate multidrug-resistant organism outbreaks: how much do they cost? Clin Microbiol Infect. 2016;22(2):162.el-162.e9.
(31.) Schmitt S, McQuillen DP, Nahass R, et al. Infectious diseases specialty intervention is associated with decreased mortality and lower healthcare costs. Clin Infect Dis. 2014;58(1):22-28.
(32.) Goto M, Schweizer ML, Vaughan-Sarrazin MS, et al. Association of evidence-based care processes with mortality in Staphylococcus aureus bacteremia at Veterans Health Administration Hospitals, 2003-2014. JAMA Intern Med. 2017;11(1):146.
Additional supporting information may be found online in the Supporting Information section at the end of the article.
Kenton J. Johnston PhD (1) | Kenneth E. Thorpe PhD (2) | Jesse T. Jacob MD (3) | David J. Murphy MD, PhD (4,5)
(1) Department of Health Management and Policy, Center for Outcomes Research, College for Public Health and Social Justice, Saint Louis University, St. Louis, Missouri
(2) Department of Health Policy and Management, Rollins School of Public Health, Emory University, Atlanta, Georgia
(3) Division of Infectious Diseases, Department of Medicine, Emory University School of Medicine and Emory Antibiotic Resistance Center, Atlanta, Georgia
(4) Department of Medicine, Emory University School of Medicine, Atlanta, Georgia
(5) Office of Quality and Risk, Emory Healthcare, Atlanta, Georgia
Kenton J. Johnston, PhD, Department of Health Management and Policy, Center for Outcomes Research, College for Public Health and Social Justice, Saint Louis University, St. Louis, MO.
Jesse T. Jacob and David J. Murphy contributed equally to this work.
TABLE 1 Definition of bacterial infections and infections associated with multidrug-resistant organisms Bacterial infections (ICD-9-CM Diagnosis Codes) Meningitis (320.x, 049.x) Encephalitis (323.x, 054.3) Cellulitis (681.x, 682.x) Endocarditis (421.x) Pneumonia (481.x, 482.x) Pyelonephritis (590.x) Septic arthritis (711.0x, excluding 711.08) Osteomyelitis (730.0x, 730.1x, 730.2x) Bacteremia (038.x, 790.7) Sepsis/severe sepsis (995.91-92) Surgical site infection (998.59) Urinary tract infection (599.0) Complicated intra-abdominal Infection (531.1,3,5,6; 532.1,3,5,6; 533.1-2,5,6; 534.1-2,5,6; 540.0-1; 56721-23,29,31,38-39,81,89,9; 569.5,81,83; 572; 575.4; 576.1,4) Intestinal infections due to other organisms/enteritis (008.1-5, 008.8) Bacterial infection in conditions classified elsewhere and of unspecified site (041.x) Infections associated with multidrug-resistant organisms (ICD-9-CM Diagnosis Codes) (1) Methicillin-resistant Staphylococcus aureus (038.12, 041.12, 482.42) (2) Intestinal infection due to Clostridium difficile (008.45) (3) Infection with other MDRO (V09.x) (a) (4) With a diagnosis for two or more of (1), (2), and (3) ICD-9-CM, International Classification of Disease, Ninth Revision, Clinical Modification; MDRO, Multidrug-resistant Organism. (a) Any combination of one or more of the following V09.x ICD-9-CM codes: V090, V091, V092, V093, V094, V0950, V0951, V096, V0970, V0971, V0980, V0981, V0990, V0991. TABLE 2 Descriptive statistics (a) on national inpatient stays with bacterial infections and infections associated with multidrug-resistant organisms. National inpatient sample (2014) MRSA (b) Clostridium difficile (c) Inpatient stays in study (g) N = 312 385 N= 312 310 Dependent variables Cost of stay in 2017 USD, (h) 21 213 (35 549) 23 010 (41 393) Mean (SD) Length of stay, Mean (SD) 9.45 (11.81) 10.25 (13.14) Patient characteristics Age, Mean (SD) 56.6 (21.9) 65.2 (19.3) Sex Female 45.6 58.8 Male 54.4 41.2 Race/ethnicity White 72.4 74.8 Black 14.1 12.7 Hispanic 8.9 7.8 Asian 1.5 1.9 Other 3.2 2.8 Income quartile 1 (Bottom) 35.6 27.8 2 28.5 27.7 3 20.5 23.3 4 (Top) 15.4 21.2 Payer Medicare 51.9 65.1 Medicaid 20.7 12.1 Private insurance 18.6 18.6 No insurance 6.2 2.4 Other 2.7 1.8 Inpatient course of stay characteristics Emergency department 70.5 73.9 admission Elective admission 11.4 9.0 Transfer in from acute care 6.8 6.8 hospital Transferred out to acute 3.0 2.4 care hospital Maternal or neonatal- 1.1 0.3 related stay Elixhauser comorbidity 3.5 (2.3) 4.0 (2.1) index, Mean (SD) AP-DRG severity index. 2.7 (1.0) 3.1 (0.8) Mean (SD) Major surgery/operating 29.4 15.2 procedure Bacterial infection type(i) Meningitis 0.1 0.1 Encephalitis 0.1 0.1 Cellulitis 41.4 5.0 Endocarditis 2.1 0.4 Pneumonia 16.1 2.7 Pyelonephritis 1.0 1.3 Septic arthritis 2.9 0.3 Osteomyelitis 11.3 2.0 Bacteremia 31.8 27.7 Sepsis/severe sepsis 24.5 24.6 Surgical site infection 5.5 1.3 Urinary tract infection 16.5 21.6 Complicated intra- 3.1 6.1 abdominal infection Intestinal infections due to 0.1 100.0 other organisms Unspecified bacterial 74.2 11.0 infections Hospital characteristics Region West 17.8 18.1 Northeast 16.6 21.4 Midwest 19.6 21.8 South 46.0 38.7 Bed size Small 17.7 17.0 Medium 29.7 28.6 Large 52.6 54.5 Type Urban teaching 60.2 65.5 Urban nonteaching 28.5 26.7 Rural 11.3 7.8 Ownership Government 13.2 11.5 Private, nonprofit 71.1 76.2 Private, for-profit 15.7 12.3 Other MDRO Multiple (Vcode) (d) MDROs (e) Inpatient stays in study (g) N = 52 590 N= 16 335 Dependent variables Cost of stay in 2017 USD, (h) 20 903 (35 339) 37 381 (56 791) Mean (SD) Length of stay, Mean (SD) 9.47 (11.59) 15.76 (20.26) Patient characteristics Age, Mean (SD) 64.9 (20.4) 64.7 (18.7) Sex Female 61.6 51.5 Male 38.4 48.5 Race/ethnicity White 67.8 69.7 Black 13.5 15.7 Hispanic 12.0 8.9 Asian 2.4 1.9 Other 4.2 3.8 Income quartile 1 (Bottom) 29.9 32.1 2 27.8 28.1 3 21.4 20.7 4 (Top) 20.9 19.0 Payer Medicare 66.1 67.6 Medicaid 14.6 15.0 Private insurance 15.3 13.4 No insurance 2.3 2.0 Other 1.8 2.0 Inpatient course of stay characteristics Emergency department 72.9 72.6 admission Elective admission 10.0 9.7 Transfer in from acute care 6.3 8.6 hospital Transferred out to acute 2.3 3.4 care hospital Maternal or neonatal- 1.0 0.1 related stay Elixhauser comorbidity 3.8 (2.0) 4.5 (2.1) index, Mean (SD) AP-DRG severity index. 2.9 (0.8) 3.4 (0.7) Mean (SD) Major surgery/operating 17.7 24.9 procedure Bacterial infection type(i) Meningitis 0.2 0.2 Encephalitis 0.2 0.5 Cellulitis 9.7 17.6 Endocarditis 0.7 2.0 Pneumonia 5.5 22.9 Pyelonephritis 8.2 1.7 Septic arthritis 0.5 1.7 Osteomyelitis 4.4 10.4 Bacteremia 33.0 49.9 Sepsis/severe sepsis 27.2 40.7 Surgical site infection 3.4 4.4 Urinary tract infection 60.2 37.8 Complicated intra- 7.5 5.5 abdominal infection Intestinal infections due to 0.3 78.6 other organisms Unspecified bacterial 69.2 61.6 infections Hospital characteristics Region West 21.6 21.5 Northeast 18.4 18.1 Midwest 19.6 18.8 South 40.3 41.6 Bed size Small 18.7 15.9 Medium 28.4 29.0 Large 52.9 55.1 Type Urban teaching 63.8 27.3 Urban nonteaching 27.1 27.3 Rural 9.1 6.9 Ownership Government 12.5 12.4 Private, nonprofit 75.6 74.1 Private, for-profit 11.9 13.5 No MDROs P value (f) Inpatient stays in study (g) N = 5 691 637 Dependent variables Cost of stay in 2017 USD, (h) 15 960 (29 381) <0.001 Mean (SD) Length of stay, Mean (SD) 6.99 (9.29) <0.001 Patient characteristics Age, Mean (SD) 61.8 (21.3) <0.001 Sex Female 56.9 <0.001 Male 43.1 Race/ethnicity White 70.8 <0.001 Black 13.3 Hispanic 10.4 Asian 2.1 Other 3.3 Income quartile 1 (Bottom) 31.7 <0.001 2 28.0 3 22.1 4 (Top) 18.3 Payer Medicare 57.5 <0.001 Medicaid 15.8 Private insurance 19.7 No insurance 4.6 Other 2.3 Inpatient course of stay characteristics Emergency department 74.4 <0.001 admission Elective admission 9.8 <0.001 Transfer in from acute care 6.0 <0.001 hospital Transferred out to acute 2.7 <0.001 care hospital Maternal or neonatal- 2.0 <0.001 related stay Elixhauser comorbidity 3.4 (2.1) <0.001 index, Mean (SD) AP-DRG severity index. 2.7 (0.9) <0.001 Mean (SD) Major surgery/operating 20.0 <0.001 procedure Bacterial infection type(i) Meningitis 0.2 <0.001 Encephalitis 0.4 <0.001 Cellulitis 19.9 <0.001 Endocarditis 0.6 <0.001 Pneumonia 4.7 <0.001 Pyelonephritis 4.6 <0.001 Septic arthritis 0.9 <0.001 Osteomyelitis 4.1 <0.001 Bacteremia 29.7 <0.001 Sepsis/severe sepsis 26.0 <0.001 Surgical site infection 2.9 <0.001 Urinary tract infection 40.7 <0.001 Complicated intra- 12.2 <0.001 abdominal infection Intestinal infections due to 2.0 <0.001 other organisms Unspecified bacterial 23.8 <0.001 infections Hospital characteristics Region West 19.1 <0.001 Northeast 18.9 Midwest 20.0 South 42.0 Bed size Small 18.6 <0.001 Medium 29.3 Large 52.1 Type Urban teaching 61.3 <0.001 Urban nonteaching 28.3 Rural 10.5 Ownership Government 12.5 <0.001 Private, nonprofit 72.3 Private, for-profit 15.2 AR, antibiotic-resistant; MDRO, Multidrug-resistant Organism; MRSA, Methicillin-resistant Staphylococcus aureus; SD, standard deviation; USD, U.S. Dollars. (a) Values are presented as percentages unless otherwise noted. (b) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for MRSA (038.12, 041.12, 482.42). (c) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for Clostridium difficile (008.45). (d) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for another MDRO(V09.x). (e) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for two or more of the following: MRSA (038.12, 041.12, 482.42), C. difficile (008.45), and another MDRO (V09.x). (f) Statistically significant difference in means (Wald Test) or proportions (chi-square test) between the five groups. (g) Weighted estimate from the 2014 National Inpatient Sample (NIS) for all stays with an ICD-9-CM discharge diagnosis code for any of the bacterial infections listed in Table 1. After excluding, stays with missing values for the variables listed in this table. (h) Inflation adjusted to reflect the growth in the price of hospital services from the midpoint (June) of 2014 to the midpoint (June) of 2017 using the U.S. city average from the Consumer Price Index. (i) Percentages do not add to 100 because some patients have multiple infection diagnoses on their discharge records. TABLE 3 Impact of infections associated with multidrug-resistant organisms on inpatient hospitalization costs. (In 2017 US dollars) (a) Unadjusted results (b) Additional cost per inpatient stay in US dollars (95% CI) Inpatient stays N = 6 385 258 MRSA only (d) 4862 (4605-5119) Clostridium difficile only (e) 6504 (6229-6780) Other MDRO only (f) 4360 (3742-4977) Multiple MDRO 15 424(14 431-16 417) lnfections (g) Adjusted results (c) Additional cost per inpatient stay in US dollars (95% CI) Inpatient stays N = 6385258 MRSA only (d) 1718 (1609-1826) Clostridium difficile only (e) 4617 (4407-4827) Other MDRO only (f) 2302 (2044-2560) Multiple MDRO 3570 (3019-4122) lnfections (g) Notes: Weighted estimates from the 2014 National Inpatient Sample (NIS) for all stays with an ICD-9-CM discharge diagnosis code for any of the bacterial infections listed in Table 1. After excluding, stays with missing values for the variables included in our regression analyses (the variables listed in Table 2). CI, Confidence interval; MDRO, multidrug-resistant organism; MRSA, methicillin-resistant Staphylococcus aureus. (a) Inflation adjusted to reflect the growth in the price of hospital services from the midpoint (June) of 2014 to the midpoint (June) of 2017 using the U.S. city average from the Consumer Price Index. (b) GLM regression results using a Gamma distribution with log link. All estimates and standard errors have been adjusted to account for the NIS sampling methodology. (c) With the same specifications as the unadjusted model, but adjusted to control for the covariates shown in Table 2 (patient characteristics, inpatient stay characteristics, bacterial infection type, and hospital characteristics). (d) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for MRSA (038.12, 041.12, 482.42). (e) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for C. difficile (008.45). (f) With an ICD-9-CM discharge diagnosis code for a bacterial infection and foranother MDRO (V09.x). (g) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for two or more of the following: MRSA (038.12, 041.12, 482.42), C. difficile (008.45), and another MDRO (V09.x). TABLE 4 Impact of infections associated with multidrug-resistant organisms on inpatient hospital length of stay Unadjusted results (a) Additional length of stay in days (95% CI) Inpatient stays N = 6 385 258 MRSA only (c) 1.98(1.90-2.05) Clostridium difficile only (d) 2.61 (2.54-2.69) Other MDRO only (e) 1.91 (1.74-2.08) Multiple MDRO 5.63 (5.31-5.96) infections (f) Adjusted results (b) Additional length of stay in days (95% CI) Inpatient stays N = 6 385 258 MRSA only (c) 1.04(0.99-1.09) Clostridium difficile only (d) 2.62 (2.52-2.73) Other MDRO only (e) 1.31 (1.20-1.42) Multiple MDRO 1.53(1.29-1.76) infections (f) Notes: Weighted estimates from the 2012-2013 National Inpatient Sample (NIS) for all stays with an ICD-9-CM discharge diagnosis code for any of the bacterial infections listed in Table 1. After excluding, stays with missing values for the variables included in our regression analyses (the variables listed in Table 2). CI, confidence interval; MDRO, Multidrug-resistant Organism; MRSA, methicillin-resistant Staphylococcus aureus. (a) Negative binomial regression results. All estimates and standard errors have been adjusted to account for the NIS sampling methodology. (b) With the same specifications as the unadjusted model, but adjusted to control for the covariates shown in Table 1 (patient characteristics, inpatient stay characteristics, bacterial infection type, and hospital characteristics). (c) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for MRSA (038.12, 041.12, 482.42). (d) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for C. difficile (008.45). (e) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for another M DRO (V09.x). (f) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for two or more of the following: MRSA (038.12, 041.12, 482.42), C. difficile (008.45), and another MDRO (V09.x). TABLE 5 Annual national inpatient cost estimates of infections associated with multidrug-resistant organisms. (In 2017 US dollars) (a) Total inpatient stays, N (%) (b) 35 358 818 (100.0) Inpatient stays with bacterial infection, N (%) (c) 7 093 157 (20.1) (1) Inpatient (2) Additional stays with MDROs cost per MDRO Stay Number (%) Point estimate (95% CI) Methicillin-resistant 343 220(1.0) 1718 (1609-1826) Staphylococcus aureus (d) Clostridium difficile (e) 347 230(1.0) 4617 (4407-4827) Other drug-resistant 57 865(0.2) 2302 (2044-2560) microorganism (f) Other drug-resistant 487 108 (1.4) 2302 (2044-2560) microorganism upper Multiple antibiotic-resistant 17 930(0.1) 3570 (3019-4122) infections (h) (3) Total cost of MDRO = (1) x (2) Point estimate (95% CI) Methicillin-resistant 589 651 960 Staphylococcus aureus (d) (552 241 221-626 719 994) Clostridium difficile (e) 1 603 160 910 (1 530 243 183-1 676 079 838) Other drug-resistant 133 205 230 microorganism (f) (118 276 078-148 134 423) Other drug-resistant 1 121 321 626 microorganism upper (995 648 752-1 246 996 480) Multiple antibiotic-resistant 64 010 100 infections (h) (54 130 700-73 907 501) MDRO, Multidrug-resistant Organism; NIS, National Inpatient Sample. (a) Inflation adjusted to reflect the growth in the price of hospital services from the midpoint (June) of 2014 to the midpoint (June) of 2017 using the U.S. city average from the Consumer Price Index. (b) Weighted estimates from NIS, using the discharge weights on all inpatient stay discharges in NIS, 2014. (c) With an ICD-9-CM discharge diagnosis code for any of the bacterial infections listed in Table 1. (d) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for MRSA (038.12, 041.12, 482.42). (e) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for C. difficile (008.45). (f) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for another MDRO (V09.x). (g) Upper bound estimate extrapolated from our analysis of Burnham et al (2017). In that study, V09.x ICD-9 codes had a (weighted) sensitivity of 11.88% vs the gold standard of laboratory cultures in accurately identifying patients with MDROs during hospitalization. This implies that for every 1 accurately coded case of a MDRO infection hospitalization using the V09.x ICD-9 codes there are 8.418 additional cases that have not been identified. (h) With an ICD-9-CM discharge diagnosis code for a bacterial infection and for two or more of the following: MRSA (038.12, 041.12, 482.42), C. difficile (008.45), and another MDRO (V09.x).
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||RESEARCH ARTICLE|
|Author:||Johnston, Kenton J.; Thorpe, Kenneth E.; Jacob, Jesse T.; Murphy, David J.|
|Publication:||Health Services Research|
|Date:||Aug 1, 2019|
|Previous Article:||Medicare expenditures attributable to dementia.|
|Next Article:||Association of Medicare's Bundled Payments for Care Improvement initiative with patient-reported outcomes.|