Clostridium difficile infection (CDI) is the leading cause of infectious nosocomial diarrhoea but the economic costs of CDI on healthcare systems in the US remain uncertain.
Methods
We conducted a systematic search for published studies investigating the direct medical cost associated with CDI hospital management in the past 10 years (2005–2015) and included 42 studies to the final data analysis to estimate the financial impact of CDI in the US. We also conducted a meta-analysis of all costs using Monte Carlo simulation.
Results
The average cost for CDI case management and average CDI-attributable costs per case were $42,316 (90 % CI: $39,886, $44,765) and $21,448 (90 % CI: $21,152, $21,744) in 2015 US dollars. Hospital-onset CDI-attributable cost per case was $34,157 (90 % CI: $33,134, $35,180), which was 1.5 times the cost of community-onset CDI ($20,095 [90 % CI: $4991, $35,204]). The average and incremental length of stay (LOS) for CDI inpatient treatment were 11.1 (90 % CI: 8.7–13.6) and 9.7 (90 % CI: 9.6–9.8) days respectively. Total annual CDI-attributable cost in the US is estimated US$6.3 (Range: $1.9–$7.0) billion. Total annual CDI hospital management required nearly 2.4 million days of inpatient stay.
Conclusions
This review indicates that CDI places a significant financial burden on the US healthcare system. This review adds strong evidence to aid policy-making on adequate resource allocation to CDI prevention and treatment in the US. Future studies should focus on recurrent CDI, CDI in long-term care facilities and persons with comorbidities and indirect cost from a societal perspective. Health-economic studies for CDI preventive intervention are needed.
The online version of this article (doi:10.1186/s12879-016-1786-6) contains supplementary material, which is available to authorized users.
Abkürzungen
CDI
clostridium difficile infection
CIs
confidence intervals
CO CDI
community-onset CDI
HCF
healthcare facility
HIV
human immunodeficiency virus
HO-CDI
hospital-onset cdi
ICD-9-CM
the international classification of diseases, ninth revision, clinical modification
ICUs
intensive care units
IQR
interquantile range
LTCF
long-term care facility
NIS
national independent sample
SD
standard deviation
US
United States
Background
Clostridium difficile is the leading cause of infectious nosocomial diarrhoea in the United States (US) [1] and the incidence and severity of C. difficile infection (CDI) are increasing [2]. CDI is associated with significant morbidity and mortality; it represents a large clinical burden due to the resultant diarrhoea and potentially life-threatening complications, including pseudomembranous colitis, toxic megacolon, perforations of the colon and sepsis [3‐5]. Up to 25 % of patients suffer from a recurrence of CDI within 30 days of the initial infection. Patients at increased risk of CDI are those who are immuno-compromised, such as those with human immunodeficiency virus (HIV) or who are receiving chemotherapy [6‐8], patients receiving broad-spectrum antibiotic therapy [9, 10] or gastric acid suppression therapy [9, 11], patients aged over 65 years [10], patients with serious underlying disease [12], patients in intensive care units (ICUs) [10], or patients who have recently undergone non-surgical gastrointestinal procedures or those being tube-fed [10].
CDI represents a significant economic burden on US healthcare systems. Infected patients have an increased length of hospital stay compared to uninfected patients, besides there are significant costs associated with treating recurrent infections. A few systematic reviews of cost-of-illness studies on CDI cost are available [13‐21]. These reviews mainly listed the range of reported cost of their respective observation period or were limited by the small number of included studies or inadequate control for confounding factors. No meta-analysis of large number of cost data in the US has been conducted to date. The cost for patients discharged to long-term care facility (LTCF) and recurrent CDI management are understudied. The cost of case management and total financial burden of CDI treatment in the US is therefore underestimated and remains controversial.
Anzeige
The aim of the current study is to conduct a systematic review and meta-analysis of currently available data to identify and quantify the financial burden attributable to CDI, and to further estimate the total economic burden of CDI hospital management in the US.
Methods
Search strategy
English-language databases with online search tools were searched for to offer maximum coverage of the relevant literature: Medline (via the Ovid interface 1946 to July 2015); EMBASE (via the Ovid interface 1980 to July 2015); The Centre for Review and Dissemination Library (incorporating the DARE, NHS EED, and NHS HTA databases); The Cochrane Library (via the Wiley Online Library) and Health Technology Assessment Database (1989 to July 2015).
We supplemented our data by searching relevant published reports from: National epidemiological agencies, Google search for grey literature and hand searched the reference lists of the included studies. The general search headings identified were: Clostridium difficile, economic, costs, cost analysis, health care costs, length of stay, hospitalization. Examples of the strategy for Medline and EMBASE are listed in Additional file 1.
Study selection
All studies that reported novel direct medical cost and/or indirect costs related to CDI management were included. Review articles, comments, editorials, letters, studies of outbreaks, case reports, posters and articles reporting results from economic modelling of a single treatment measure (i.e. cost effectiveness of faecal transplantation) were excluded in the final analysis. All relevant publications from January 2005 to July 2015 were included in the search. We included the following healthcare settings: hospitals, long-term care facilities and community. Geographical scope covered the US. We did not apply any language restriction. Our predefined inclusion and exclusion criteria are shown in Additional file 1.
Anzeige
Data extraction
Two reviewers (SP, SZ) independently selected the included articles and extracted data. After combining their results, any discrepancies were solved by discussion with HN and MK.
The primary outcomes were CDI-related costs (total costs of those with CDI and other comorbidities) and CDI-attributable costs (total costs of CDI management only, after controlling for the confounders). For studies with control groups (e.g. matched patients without CDI), the CDI-attributable cost extracted was either the cost provided by the articles or calculated by reviewers using the CDI-related cost minus the treatment cost of control groups. The secondary outcome was resource utilization associated with CDI, i.e. CDI-related length of stay (LOS) in hospital and CDI-attributable LOS. The study characteristics of each article were extracted. These included basic publication information, study design, statistical methods, economic data reporting characteristics and population information.
When multiple cost data were presented in a study, we included only one cost estimate for each population subgroup as per the priority below:
a.
Matched data > Unmatched data.
b.
Adjusted model results > Unadjusted model results.
c.
Regression model results > Calculated difference.
d.
Total cost/charges > Subgroup cost/charge (i.e. survivors, died).
e.
Median (Interquantile Range: IQR) > Mean (Standard Deviation, SD).
All costs/charges data were inflated to 2015 US$ equivalent prices adjusted for the Consumer Price Index. If the price year was not reported, it was assumed to be the last year of the data collection period. In cases where charges were reported without cost-to-charge given, costs were estimated using a cost-to-charge ratio of 0.60, which is commonly used value in US health economic studies [22].
Meta-analysis and estimation of national impact
We carried out meta-analysis for cost studies following a Monte Carlo simulation approach, as reported by Jha et al [23] and Zimlichman et al [17], bearing in mind the heterogeneity of the included studies. For each subgroup of CDI, we synthesized the data and reported a point estimate and 90 % confidence intervals (CIs) for the CDI-related cost, CDI-attributable cost and their respective LOS. For each included study, we simulated distribution with pooled results weighted by sample size. We fitted a triangular distribution for each of the included studies based on their reported measures of central tendency and dispersion, i.e. mean and 95 % CI, median and IQR, or median and range. Then we simulated 100,000 sample draws from the modeled distribution of each study. At each iteration, we calculated the weighted average of all included studies. Finally, we reported the mean and 90 % CI from the resulting distribution of the 100,000 weighted average of CDI. This approach facilitated the combination of cost data and eliminated the limitation of combining non-normally distributed data. Monte Carlo simulations were conducted using the Monte Carlo simulation software @RISK, version 7.0 (Palisade Corp).
We estimated the national financial impact of CDI on the US healthcare system, by determining the potential boundaries. The higher boundary was the total number of CDI cases in the US in 2011 extracted from Lessa et al [24], while the lower boundary was the result from a meta-analysis to estimate the total burden of CDI cases in the US [25] (For detailed results see Additional file 1). The total annual cost of CDI management was calculated multiplying the average cost of management per case of CDI, with the total number of CDI cases per year in the US (Fig. 1). We assumed that all CDI cases received treatment in hospital. A point estimate of the final cost (with range) was reported based on a Monte Carlo simulation of 100,000 sample draws.
Fig. 1
Formula for total annual cost calculation
×
Sensitivity analysis
We extracted the total number of CDI patients and CDI-attributable costs from previous studies [25] and reviews [17, 26] to carry out a sensitivity analysis of our total cost estimates.
Quality assessment
The quality of the studies included was assessed mainly based on the complexity of the statistical method (Fig. 2). All studies were included in the final analyses.
Fig. 2
Quality Assessment Method
×
Anzeige
Results
Search results
The search strategy identified 2671 references from databases. Seven additional references were identified through other sources. After screening the titles, abstracts and relevant full texts (Fig. 3), a total of 42 studies were included in this review.
Fig. 3
PRISMA diagram of economic burden search of C. difficile
×
Study characteristics
The characteristics of the 42 included studies [27‐68] are summarized in Table 1. Cost data collection periods ranged from 1997 to 2012. Most studies (n = 27) used national level databases, with 17 used National Independent Sample (NIS) database and the remaining 10 studies extracted data from various national databases. Fifteen studies were conducted at state level, of which 6 studies only collected data in single hospital. All studies reported cost in hospital level of care, no articles identified in LTCF and community. Nearly all identified references were retrospective hospital database studies (n = 40) and only 1 study was a prospective observational study [29] and another study was a decision tree model [48].
Table 1
Overview of selected references that assessed economic burden attributable to CDI by type of CDI considered in the US
Abbreviations: NR not reported, IBD inflammatory bowel disease, LOS length of stay, ICU intensive care unit, retrosp. retrospective, Comp. complicating, Req. requiring, both requiring and complicating, PD primary diagnosis, SD secondary diagnosis
a Median (Range)
Most studies (n = 15) investigated economic outcomes in all age inpatients. Three studies reported cost data in children less than 20 years old. The mean/median age of the CDI patient groups ranged from 47.4 to 73.0 years. Other studies investigated complicated CDI in high-risk patient groups, such as those with major surgery (n = 16), inflammatory bowel diseases (n = 2), liver or renal disease (n = 4), elderly (n = 2) and ICU patients (n = 1). There was 1 study each in non-surgical inpatients, sepsis inpatients and patients with prolonged acute mechanical ventilation. There was 1 study focusing only on recurrent CDI in the general population.
The sample sizes of included studies ranged from 85 to 7,227,788, with a median sample size of 83,939. A total of 28.8 million inpatient hospital-days were analysed, of which 1.31 million inpatient hospital-days were CDI patients. The median sample size of CDI population was 2938.
Anzeige
The methods to identify CDI varied according to the type of CDI that was assessed in the study. CDI cases were identified either with laboratory test, i.e. positive C. diffcile toxin assay, or hospital discharge diagnosis of C. difficile (primary and/secondary) from administrative datasets using the International Classifications of diseases, Ninth, Clinical Modification, ICD-9-CM 008.45. Clinical diagnosis was also used in two studies.
CDI was classified in three types: Community-onset CDI (CO-CDI) requiring hospitalization, Hospital-onset CDI (HO-CDI) complicating other diseases, or both CDI (Table 2). Most of included studies considered HO-CDI (n = 23) or both CDI types (n = 17). Only four studies investigated CO-CDI only. However, subgroup data of CO-CDI is also available in studies that reported both CDI types.
Table 2
Classification of CDI Cases by Setting of Acquisition
Case definition
Criteria for classification
CO-CDI
- Discharge code ICD-9-CM 008.45 as Primary diagnosis
HO-CDI
- Discharge code ICD-9-CM 008.45 as secondary diagnosis, without a primary diagnosis of a CDI-related symptom (e.g. diarrhea)
- Study population ≥ 48 h of hospitalization
- Symptom onset and/or positive laboratory assay at least ≥ 48 h hospitalization
Both CDI
- No distinction of settings of acquisition
- Discharge code ICD-9-CM 008.45 in any position
Abbreviations: CO-CDI community-onset CDI, HO-CDI hospital-onset CDI, ICD-9-CM The International Classification of Diseases, Ninth Revision, Clinical Modification
CDI costs and LOS
The mean CDI-attributable costs per case of CO-CDI were $20,085 (Range: $7513–$29,662), lower than HO-CDI $34,149 (Range:$1522–$122,318). HO-CDI showed a wider range within which the additional cost for CDI in the general population ranged from $6893 to $90,202 and in high risk groups ranged from $7332 in congestive heart failure patients to $122,318 in renal impairment patients. The mean CDI-attributable LOS was 5.7 days (Range: 2.1–33.4) for CO-CDI, 7.8 (Range:2.3–21.6) days for HO-CDI, and 13.6 (Range: 2.2–16) days for both groups. Cost data and LOS for individual studies are presented in Tables 3 and 4.
Table 3
CDI-attributable costs/charges and CDI-related management costs/charges
Abbreviations: CO-CDI community-onset CDI, HO-CDI Hospital-onset CDI, PAMV prolonged acute mechanical ventilation, Cabx concomitant antibiotic use, UC ulcerative colitis, CD Crohn’s disease, IBD inflammatory bowel disease, ICU intensive care unit, CABG coronary artery bypass grafting, VS valvular surgery, BMT, PD primary diagnosis, SD secondary diagnosis, Calculated numbers were marked in Italic, attributable cost = cost of CDI group- cost of control non-CDI group
aQ1-Q3
bMin-Max
Using a Monte Carlo simulation, we generated point estimates and 90 % CI for both cost and LOS; the meta-analysis results are shown in Table 5. The total cost of inpatient management of CDI-related disease was $42,316 (90 % CI: $39,886–$44,765) per case, of which the total CDI-attributable cost was $21,448 (90 % CI: 21,152–21,744) per case. For the inpatient management, the attributable cost for those HO-CDI was $34,157 (90 % CI: $33,134–$35,180), which was 1.5 times as much as CO-CDI management $20,095 (90 % CI: $4991–$35,204).
Table 5
Meta analysis results of cost and LOS of CDI management
Similar patterns were observed in LOS data. The total CDI-related LOS was 11.1 days (90 % CI: 8.7–13.6) and CDI-attributable LOS was 9.7 (90 % CI: 9.6–9.8). The HO-CDI patients had longer CDI-attributable LOS 9.7 days (90 % CI: 9.7–9.7) than CO-CDI patients 5.7 days (90 % CI: 4.1–7.3).
CDI annual national impact estimate
The total burden of healthcare facility CDI in US was estimated 293,300 (Range: 264,200–453,000) cases per year [25]. The total financial burden of CDI inpatient management was estimated to be US$6.3 (Range: $1.9–$7.0) billion in 2015, which required 2.4 million days of hospital stay. The total CDI related disease management cost was nearly doubled at US$12.4 (Range: $3.7–$14.4) billion in 2015 (Table 6). A sensitivity analysis showed that the total CDI-attributable cost ranged from $1.31 to $13.61, which covers our estimates (Additional file 1).
Table 6
Total cost of CDI management in US
Total number of HCF CDI cases per year (2011) [25]
A summary of the quality assessment for statistical methods in included studies is shown in Additional file 1. There were 13 studies of high quality, 21 studies with medium quality and 8 low quality studies.
Discussion
We systematically reviewed 42 published cost studies of CDI case management in the past 10 years (2005–2015) and found a significant financial burden associated with CDI in the US. The total CDI-attributable cost was US$6.3 billion, which is higher than previously reported (range US$1.1–4.8 billion) [14, 16, 17]. The mean cost for CDI-attributable hospitalized patients per case was US$21,448, nearly half of the mean CDI-related inpatient cost.
This review facilitated a meta-analysis of a large number of cost studies for costs related to CDI management and provided an uncertainty range. Zimlichman et al [17] applied this method to calculate CDI cost based on cost data from two cost-of-illness studies (O’Brian 2007 [51] & Kyne 2002 [69]) and obtained a lower cost [2012US $11,285 ($9118–$13,574)] than ours. Our review combined 100-point estimates and ranges from 42 individual studies, which provided more accurate and comprehensive data of the cost result. Despite the methodological heterogeneity in perspectives, treatment procedure and statistical analysis, each included study met our inclusion criteria, which were defined to identify studies that provided real world estimates of costs, therefore the combination of these data with uncertainty range represented a valuable and reliable summary of CDI-related cost.
Furthermore, we evaluated hospital onset CDI and community onset CDI separately. We found that CDI complicating hospitalization cost more than CDI requiring hospitalization and the former had longer attributable hospital stay. Therefore, other factors, such as comorbidity, may contribute to infections and increase the difficulty of CDI treatment.
We estimated that the total cost attributable to CDI management in the US was nearly US$6.3 (Range: $1.9–$7.0) billion, which is similar to Dubberke and Olsen’s estimates at $4.8 billion [14], but significantly higher than other studies (US$ 1.5 billion in Zimlichman et al [17] and $1.1 billion in Ghantoji et al [16]). The later studies reported lower attributable cost per case based on a limited number of studies before 2005, which arguably is out-of-date. To compare with the latest review on global CDI cost (Nanwa et al [26]), this review identified 8 additional studies with recent data. Nanwa et al [26] found that the mean attributable CDI costs ranged from US$8911 to US$30,049, which is similar to our results.
In this study, we only assessed the quality of study emphasizing statistical methods and did not use the modified economic evaluation guideline as other COI systematic reviews. Cost and LOS estimation of healthcare-associated infections has the potential to be misleading if the confounders such as patients’ comorbidities or daily severity of illness were not properly controlled for. Using either the matching design or multivariable regression analysis allows to control known confounders and may, in part, address selection bias [70]. We found that whether advanced statistical methods were used and described was crucial for the assessment of data quality, which has not be fully captured by the existing quality assessment tool. Therefore in this study we assessed quality of included studies using this new method. Moreover, Nanwa et al [26] has evaluated the methodological completeness of most included studies (34 out of 42); we agree with their recommendations regarding possible improvement of future cost-of-illness study. However, we need to bear in mind that cost effects or excess LOS are still likely to be overestimated if the interval to onset of HAI is not properly accounted for in the study design or analysis [70].
Our systematic review has some limitations. First, all included studies reported direct medical costs from hospital perspective, therefore indirect cost to patients and society and costs of additional care after hospital discharge, have not been captured. No studies reported indirect cost (productivity loss due to work day losses) of patients or care-givers, and we failed to identify studies assessing cost of CDI in long-term care facilities, where about 9 % of CDI patients were discharged to for an average of 24 days of after-care. This would result in an additional US$141 million burden on the healthcare system and society due to LTCF transfers [14]. Second, we did not separate primary CDI from recurrent CDI cost in our review because only two studies reported cost specifically to recurrent CDI $12,592 (Range: $9752, $15,919) [2]. Moreover, we found it difficult to exactly match the CDI case definition in cost study (e.g. ICD10 Code primary diagnosis and secondary diagnosis) with the case definition in epidemiology studies (e.g. community onset, hospital onset), therefore we did not estimate CDI patients managed at outpatient and community settings due to lack of both epidemiology and economic data. The total costs of CDI management may be higher than our current estimate. Fourth, unlike other published reviews, we did not include cost studies from countries other than the US nor facilitate any international comparison. This study initially aimed to identify cost-of-illness studies in North America, but we did not find any studies reporting cost data from Canada. This is likely because we restricted our search to English language databases. Therefore the cost of CDI management in Canada remains unknown. However, we did not apply any language restrictions to the current review.
Effective prevention can reduce the burden of diseases. Strategies have been promoted such as appropriate use of antimicrobials, use of contact precautions and protective personal equipment to care for infected patients, effective cleaning and disinfection of equipment and the environment, and early recognition of disease as primary prophylaxis [71]. As CDI is an infectious disease, the population at risk would benefit from an effective vaccine, which is currently under development [72, 73].
More cost of illness studies for recurrent CDI, or in LTCF, and indirect cost from a societal perspective are needed in the future. We would also recommend that published studies report their methods and include point estimates with uncertainty range. Further economic studies for CDI preventive interventions are needed.
Conclusion
This review indicates that CDI places a significant financial burden on the US healthcare system. In addition, our findings suggest that the economic burden of CDI is greater than previously reported in the US. This review provides strong evidence to aid policy-making on adequate resource allocation to CDI prevention and treatment in US.
Acknowledgements
We gratefully acknowledge the comments and suggestions from Guy De Bruyn, Clarisse Demont, Kinga Borsos (Sanofi Pasteur) during manuscript preparation. We thank Sanofi Pasteur for financial support for this work. The findings and conclusions in this report are those of the authors and do not necessarily represent the official views or policies of Sanofi Pasteur.
Funding
Sanofi Pasteur funded this study.
Availability of data and materials
The datasets supporting the conclusions of this article are included within the article and its Additional file 2.
Authors’ contributions
Study design (MK, HN, AC); data collection (SZ, SP, EB); data analysis (SZ, EB); data interpretation (SZ, EB, HN, AC, MK); development of initial draft manuscript (SZ, EB, HN), critical revisions for intellectual content of manuscript (SZ, SP, EB, HN, AC, MK); study supervision (HN, MK). All authors reviewed and approved the final draft of manuscript.
Competing interests
SP, AC, MK are employees of Sanofi Pasteur.
Consent for publication
Not applicable.
Ethics approval and consent to participate
Not applicable.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
