Clin Microbiol Infect
The economic impact of community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA) remains unclear. We developed an economic simulation model to quantify the costs associated with CA-MRSA infection from the societal and third-party payer perspectives. A single CA-MRSA case costs third-party payers $2277–$3200 and society $7070–$20 489, depending on patient age. In the United States (US), CA-MRSA imposes an annual burden of $478 million to 2.2 billion on third-party payers and $1.4–13.8 billion on society, depending on the CA-MRSA definitions and incidences. The US jail system and Army may be experiencing annual total costs of $7–11 million ($6–10 million direct medical costs) and $15–36 million ($14–32 million direct costs), respectively. Hospitalization rates and mortality are important cost drivers. CA-MRSA confers a substantial economic burden on third-party payers and society, with CA-MRSA-attributable productivity losses being major contributors to the total societal economic burden. Although decreasing transmission and infection incidence would decrease costs, even if transmission were to continue at present levels, early identification and appropriate treatment of CA-MRSA infections before they progress could save considerable costs.
Studies have suggested that community-associated methicillin-resistant Staphylococcus aureus (CA-MRSA, i.e. MRSA colonization and infection not associated with healthcare settings) is a substantial public health problem [1,2]. CA-MRSA strains are common causes of skin and soft tissue infections (SSTIs) in the United States (US) [3,4], with reported outbreaks in many diverse settings and populations, including prisoners, military recruits, and athletes [5,6]. CA-MRSA strains have become endemic [7,8], predominantly causing SSTIs but also necrotizing pneumonia and invasive syndromes such as necrotizing fasciitis, osteomyelitis, septic thrombophlebitis, bacteraemia, and severe sepsis . While studies have shown increases in CA-MRSA infection incidence among veterans  and CA-MRSA SSTI incidence , its overall incidence in the US has not been clearly delineated. Although the Centers for Disease Control and Prevention (CDC) have initiated CA-MRSA public awareness campaigns , there is a dearth of documented community-level efforts to curb transmission.
An extensive literature search for economic studies on CA-MRSA yielded only two, one quantifying the cost impact of an epidemic on Driscoll Children’s health plan  and another focusing on just pneumonia patients . Until CA-MRSA’s overall economic burden is better quantified, it may be difficult for decision makers to determine where CA-MRSA should fall on public health, medical, and scientific priority lists. Without an estimate of the costs associated with CA-MRSA infections, many questions remain. For example, how much should policy makers invest in prevention, education and control? Should insurance companies focus efforts and reimbursement policies on prevention and control? How much should be invested in developing new diagnostic, prevention and treatment interventions? Therefore, we developed an economic computational model to quantify the costs associated with CA-MRSA infection from the third-party payer and societal perspectives.
Fig. 1 outlines the structure of our economic simulation model developed in TreeAge Pro 2009 (Williamstown, MA, USA) to determine the costs associated with a CA-MRSA infection from third-party payer and societal perspectives. Table 1 displays the model inputs. To obtain these values, we conducted an extensive Medline search using key words (‘Methicillin-Resistant Staphylococcus aureus’, ‘Staphylococcus aureus’, ‘Methicillin Resistance’, ‘Community-Acquired Infections’, and ‘CA-MRSA’) to identify studies and excluded those conducted outside the US or among immunocompromised populations. Our study used the CDC epidemiological definition of CA-MRSA to determine each infection type probability. An infection was considered CA-MRSA if the culture was obtained during an outpatient visit or within 48 h of hospital admission. Also within the past year, the patient must not have been admitted to a hospital, nursing home or any other long-term care facility and not have haemodialysis or surgery. Furthermore, an indwelling catheter or a percutaneous device must not have been in place at the time of culture .
|Parameter||Distribution typea||Mean/Median||Standard deviation (SD) or range|
|Skin and soft tissue infection (SSTI; includes non-purulent cellulitis, impetigo, folliculitis, uncomplicated purulent cellulitis and complicated SSTI)|
|Having an SSTI||β||0.78||0.0004|
|Incision and drainage in ambulatory settings for purulent cellulitis (adult)||0.83|
|Incision and drainage in ambulatory settings for purulent and non-purulent cellulitis (paediatric)||0.144|
|Incision and drainage in hospital with purulent cellulitis (adult)||0.98|
|Incision and drainage in hospital for purulent and non-purulent cellulitis (paediatric)||0.478|
|Healthcare utilization if seeking medical care|
|Outpatient facility visit||0.738|
|Emergency room (ER) visit||0.205|
|Hospitalization after outpatient visit||0.021|
|Hospitalization after ER visit||Δ||0.136||0.105–0.147|
|In-hospital mortality (adult)||0.001|
|In-hospital mortality (paediatric)||0|
|Hospitalization (<1 year old)b||γ||3891.2||137.68|
|Hospitalization (1–17 years old)b||γ||3891.2||120.2|
|Hospitalization (18–44 years old)b||γ||5842.57||88.76|
|Hospitalization (45–64 years old)b||γ||7252.78||113.89|
|Hospitalization (65–84 years old)b||γ||7708.01||140.02|
|Hospitalization (85 years old and over)b||γ||7053.17||138.29|
|Incision and drainage in ERb||γ||389.67||228.62|
|Incision and drainage in outpatient settingsb||γ||396.07||159.88|
|Debridement cost in facilityb||γ||436.14||333.29|
|Outpatient/ER visit (hours)||4|
|Hospitalization (median, days)||Δ||4||3–5|
|Outpatient length of therapy (days)||U||5–10|
|Inpatient length of therapy (days)||U||7–14|
|In-hospital mortality (adult)||β||0.12||0.088–0.15|
|In-hospital mortality (paediatric)||0.076|
|Hospitalization (<1 year old)b||γ||39112.2||6234.91|
|Hospitalization (1–17 years old)b||γ||22673.63||5141.96|
|Hospitalization (18–44 years old)b||γ||24 384.77||3621.71|
|Hospitalization (45–64 years old)b||γ||26 327.53||1896.42|
|Hospitalization (65–84 years old)b||γ||20 938.53||864.29|
|Hospitalization (85 years old and over)b||γ||17 284.14||901.43|
|Hospitalization (adult; days)||γ||18.2||16.6|
|Hospitalization (paediatric; median, days)||23.7||6–138|
|Length of therapy (weeks)||U||1–3|
|Other invasive infections|
|Having an invasive infection||0.0625|
|In-hospital mortality (adult)||β||0.1583||0.000231|
|In-hospital mortality (paediatric)||β||.0246||0.0240|
|Hospitalization (<1 year old)b||γ||6581.56||810.71|
|Hospitalization (1–17 years old)b||γ||9377.01||890.03|
|Hospitalization (18–44 years old)b||γ||13 560.29||921.74|
|Hospitalization (45–64 years old)b||γ||14 390.99||787.05|
|Hospitalization (65–84 years old)b||γ||13 691.28||550.31|
|Hospitalization (85 years old and over)b||γ||10 883.61||349.61|
|Hospitalization (adult; days)||Δ||6.0||4–8.5|
|Hospitalization (paediatric; days)||14.2||7.6|
|Length of therapy (weeks)||U||2–3|
|Oral TMP-SMXd daily dose||γ||3.53||0.82|
|Vancomycin IV daily dose/kg (adult)||γ||0.42||0.32|
|Vancomycin IV daily dose/kg (paediatric)||γ||0.84||0.65|
|Home healthcare visit||124.63|
|Hourly wage (median)||16.92|
|Productivity loss due to mortality (years)|
|<1||1 080 911|
|1||1 077 435|
|5||1 062 460|
|10||1 041 071|
|15||1 016 275|
|Parameter||Baseline||Range of Sensitivity Analysis|
|SSTI patients seeking care||0.30||0.10–0.40|
|Incision and drainage in ambulatory settings for purulent cellulitis (adult)||0.83||0.65–0.85|
|SSTI hospitalization (adult)||0.057||0.05–0.45|
|SSTI hospitalization (paediatric)||0.25||0.05–0.45|
Every individual entering the model had a CA-MRSA infection, having one of three major clinical categories (∼90% of all types): SSTI (non-purulent cellulitis, impetigo, folliculitis, uncomplicated purulent cellulitis or complicated SSTI), pneumonia, or other invasive infection (infection of a normally sterile body site, mostly bacteraemia). Some subjects with SSTIs sought medical care and those who did not relied on self-medication with over-the-counter NeosporinTM (neomycin, polymyxin B sulphate and bacitracin) ointment. SST patients who sought care were treated in an outpatient clinic, emergency room (ER), or hospital. All patients with pneumonia or other invasive infection required hospitalization. When probabilities did not sum to one, a Dirchilet distribution was used to normalize the probabilities.
The Infectious Diseases Society of America (IDSA) guidelines and expert opinion guided the treatment regimen for each syndrome :
- • Non-purulent cellulitis (outpatient): 5–10-day course of trimethoprim-sulphamethoxazole (TMP-SMX)
- • Impetigo (outpatient): 2% topical mupirocin ointment twice a day for ∼2 weeks
- • Folliculitis (outpatient): warm-water packs/presses (no cost)
- • Uncomplicated purulent cellulitis (outpatient): incision and drainage (I&D) with or without TMP-SMX
- • Complicated SSTI (hospitalized): intravenous (IV) vancomycin (dosed by age and weight), switched to a 5–10-day course of TMP-SMX 1 day prior to discharge; almost all required I&D
- • Pneumonia: IV vancomycin during hospitalization, switched to TMP-SMX 1 day prior to discharge for a total therapy duration of 1–3 weeks
- • Other invasive infections: IV vancomycin during hospitalization with 2–3 weeks subsequent home health nursing with IV vancomycin administration; paediatric patients could receive an echocardiogram
Selected regimens were conservative (least expensive) and do not represent all possible regimens recommended by IDSA Guidelines or used in practice.
The third-party payer perspective included only direct medical costs (i.e. outpatient/ER visit, hospitalization, and treatment costs), while the societal perspective included both direct and indirect (i.e. productivity losses due to work absenteeism from healthcare visits and hospitalization for individuals or caregivers if patient ≤18 years, and mortality) costs. Median hourly and annual wages for all occupations served as proxies for productivity losses. Work absenteeism was calculated as 4 h missed for an outpatient visit and 8 h per day for the duration of hospitalization (Table 1). Death resulted in the net present value of lost wages for the remainder of the person’s life expectancy based on his/her age . A 3% discount rate adjusted all costs to 2011 US$.
Each simulation fixed a patient’s age, sending 1000 patients with CA-MRSA infections through the model 1000 times (1 000 000 total trials). Subsequent simulations systematically varied patient age (range, <1–85 years).
Simulations determined the cost of a single CA-MRSA infection and an SSTI infection for different patient ages. Multiplying by the number of cases nationally, these costs-per-case were extrapolated to the annual national burden. Annual estimates of US cases came from three studies. Study 1 estimated 94 360 invasive MRSA cases, categorizing 13.7% as community-associated using CDC criteria . Assuming 6.25% of all CA-MRSA infections are invasive , there would be 206 837 CA-MRSA infections per year. Study 2 reported an annual incidence of community-onset MRSA infections (i.e. occurring among persons not hospitalized in the prior year) of 243 per 100 000 . Extending this nationwide resulted in 720 277 CA-MRSA cases per year. Study 3 reported an incidence of 521 CA-MRSA SSTIs per 100 000 in Chicago (presented at the IDSA 2011 annual meeting) , which would translate into an estimated 667.9 per 100 000 for all CA-MRSA infections, resulting in 1 979 869 cases, of which 1 544 298 would be SSTIs. The number of paediatric cases among the estimates was determined using data reporting 36.4%  and 10.1%  of CA-MRSA cases as <18 years old. These estimates were used to determine a range of costs. Study 1 represents the lower estimate, as they used the CDC definition, and Study 3 represents the upper estimate, as they used a 48-h criterion to define CA-MRSA cases. As SSTIs represent ∼75%  of infections, we determined their burden separately using estimates from Study 3 and Study 4, which reported 164.2 per 100 000 CA-MRSA SSTIs .
A similar approach estimated the annual burden for two particularly high-risk sub-populations: jail (prevalence, 4.5–79.7% ) and military populations (prevalence, ≤5% ). David et al.  evaluated SSTI incidence and aetiology among detainees at Cook County Jail, reporting 15.84 MRSA SSTIs per 1000 detainee-years (mean age, 33 years) with a census of 10 000 detainees at a given time. An army installation study estimated 35 CA-MRSA SSTI cases per 1000 soldiers (mean age, 22 years) . These incidence estimates multiplied by cost-per-SSTI-case determined the annual economic burden in these sub-populations.
One-way sensitivity analyses systematically varied key parameters one at a time throughout their ranges (Table 1). Monte Carlo probabilistic sensitivity analysis simultaneously varied all parameters throughout their ranges.
Cost to third-party payers
Table 2 shows the median and 95% range of the total direct medical costs of a CA-MRSA case from our simulations. Varying the SSTI care-seeking probability (10–40%) altered this cost in most cases by <$300 (e.g. median $2698 with a 10% care-seeking probability and $2938 with a 40% probability for a 20-year-old). With relatively modest outpatient costs-per-case (median <$100), hospitalization was the primary cost-driver.
|Age||Outpatient healthcare costs||Inpatient healthcare costs||Total third-party payer costs (total direct)b||Mortality costsc||Indirect costs (total productivity losses)d||Total societal costs (total direct and indirect)b|
|<1||27 (24–32)||3123 (2547–3713)||3200 (2602–3862)||6519 (2172–13 038)||6923 (2570–12 345)||10 212 (5463–15 979)|
|1||29 (25–37)||2352 (1943–2761)||2373 (2003–2759)||6509 (2168–13 011)||6902 (2565–12 382)||9361 (4848–15 896)|
|5||33 (26–46)||2410 (2004–2812)||2435 (2060–2850)||6419 (2138–11 769)||6806 (2535–12 145)||9291 (4741–15 814)|
|10||39 (29–59)||2469 (2099–2918)||2517 (2138–2949)||6290 (2096–11 535)||6679 (2479–11 972)||9270 (4887–15 487)|
|15||74 (64–83)||2594 (2168–3062)||2635 (2210–3075)||6141 (2046–11 260)||6535 (2449–11 649)||9183 (4908–14 732)|
|20||73 (65–84)||2787 (2371–3236)||2849 (2458–3328)||17 387 (9660–26 085)||17 418 (10 677–26 067)||20 489 (13 031–28 745)|
|35||74 (64–84)||2770 (2363–3218)||2861 (2409–3304)||15 976 (9762–23 958)||16 013 (9840–23 925)||19 166 (11 938–27 381)|
|50||74 (64–84)||2985 (2539–3416)||3068 (2602–3551)||12 921 (7180–19 387)||12 961 (7284–18 695)||16 048 (10 730–22 066)|
|65||74 (64–84)||2625 (2227–3032)||2711 (2321–3115)||9388 (5216–14 079)||9414 (5298–14 088)||12 200 (8045–16 605)|
|85||74 (63–83)||2183 (1882–2518)||2277 (1946–2588)||4666 (2593–6750)||4704 (2908–6983)||7070 (4865–9665)|
Cost to society
As Table 2 shows, societal costs were four to seven times higher than third-party costs, as the vast majority came from productivity losses (i.e. individuals or caregivers missing work plus lost productivity from infection-related deaths). Again, the medical care-seeking probability did not have a substantial impact (increasing by ≤$500). A child not surviving has higher productivity losses than an adult; however, we observe greater productivity losses for adults due to higher mortality rates.
Costs of a CA-MRSA SSTI
At baseline hospitalization values (5.7% for ≥18 years, 25% for <18 years), SSTI costs ranged from $202 (<1 year old) to $326 (65 years) for society and $168 (<1 year old) to $292 (65 years) for third-party payers. Fig. 2(a) shows how SSTI costs (societal perspective) trend with age and hospitalization rate. Approximately 85–90% of SSTI costs are direct medical costs. SSTI costs vary by age, this being largely attributable to adults’ higher mortality. Hospitalization rate was a much stronger cost-driver among adults than children (e.g. increasing from 5% to 45% only increased societal costs by $261 for a 1-year-old, but increased costs by $490 for a 65-year-old).
Fig. 2(b) shows the SSTI cost trend with hospitalization rate and care-seeking probability for a 23-year-old. The median cost-per-case when 40% of patients seek care ($370) was almost four times higher than when 10% sought care ($97). The cost more than doubled when hospitalization increased from 5% to 45%.
Table 3 shows the estimated annual total US and sub-population burdens for all CA-MRSA infections and SSTIs. CA-MRSA yielded average annual US costs of ≥$560 million for third-party payers. High productivity losses meant societal costs substantially exceeded third-party costs (≥$2.7 billion).
|Incidence estimate||Third-party payer perspective||Societal perspective|
|All CA-MRSA infections annual US burden|
|36.4% of cases are paediatric||560 (478–644)||2685 (1621–3464)|
|10.1% of cases are paediatric||562 (469–632)||2961 (1494–3989)|
|36.4% of cases are paediatric||1951 (1665–2244)||9350 (5646–12 063)|
|10.1% of cases are paediatric||1958 (1630–2187)||10 311 (5205–13 892)|
|36.4% of cases are paediatric||5363 (4577–6169)||25 701 (15 520–33 159)|
|10.1% of cases are paediatric||5509 (5159–6011)||28 343 (14 308–38 186)|
|SSTI annual US burdena|
|521 per 100 000b||343 (259–451)||393 (312–503)|
|164.2 per 100 000b||108 (82–142)||124 (98–159)|
|15 per 1000b||14||16|
|25 per 1000b||23||26|
|35 per 1000 ||32||36|
|10 per 1000b||6||7|
|12 per 1000b||7||8|
|15.84 per 1000 ||10||11|
The estimated annual societal cost to Cook County Jail was $140 265 (65% of outpatient purulent cellulitis cases required I&D) and $146 525 when 85% required I&D; annual direct medical costs ranged from $124 338 to $130 659 (results not shown). Table 3 provides cost estimates for various CA-MRSA incidence rates in jails around the country (748 728 inmates in 2009–2010 ).
Estimated annual societal costs to an army installation  ranged from $834 848 (65% of outpatient purulent cellulitis cases required I&D) to $874 797 (85% required I&D), with annual direct medical costs ranging from $737 618 to $780 354 (results not shown). Table 3 shows the annual costs for all US Army installations.
Our study suggests that CA-MRSA confers a substantial economic burden, greater than many other acute infectious diseases that have garnered the attention of policy makers, scientists and manufacturers. For example, the cost per CA-MRSA infection ($7070–$20 489) is two to five times that of an influenza case ($3000–$4000 ), three to ten times that of food-borne illness (∼$1851 ) or pertussis ($1952 per case ), and over 17 times that of Lyme disease ($397–$923 ). Moreover, because the incidence of some of these diseases is relatively low (e.g. pertussis and Lyme disease), the annual economic burden of CA-MRSA infections is much higher in comparison, ∼2–13 times higher than pertussis (among adults and adolescents) and 8–17 times higher than Lyme disease. These numbers help justify investment in effective CA-MRSA prevention and control and imply that even minimal investment could generate favourable returns for policy makers, and military and US jail systems. Our results can also help guide funders and manufacturers in establishing priorities.
Our results show that CA-MRSA may have hidden costs that may be missed by certain estimation methods. Because direct medical costs are only a minority (∼25%) of the total cost for all CA-MRSA infections, many hospital and insurance databases may not capture productivity losses, thus greatly underestimating costs. Additionally, a substantial proportion of costs come from the small fraction of deaths. Therefore, focusing on the majority of cases that carry relatively low costs (e.g. over-the-counter medications, clinic visits, course of antibiotics, and perhaps a half-day of lost productivity) overlooks the impact of complicated cases that rapidly accrue costs. Appropriate measures could prevent the bulk of CA-MRSA-associated costs. Reducing transmission and infection incidence would decrease costs. Even if infection incidence was to remain at present levels, early identification and proper treatment of infections before they progress could save considerable costs.
Therefore, research and policy development could best proceed concurrently in two directions: (i) developing and implementing measures to reduce transmission, and (ii) identifying and treating uncomplicated infections early to ensure they do not become invasive. These require a better understanding of CA-MRSA transmission dynamics and involve determining who may be at increased risk of disease progression and more complicated infections. Studies have suggested that crowded environments, lack of hygienic conditions, sharing contaminated objects (e.g. towels) and compromised skin integrity lead to increased transmission . Additionally, individuals with compromised immune systems, prior antibiotic use or other co-morbidities may be at higher risk [6,32]. However, much remains unknown. Additionally, considerable variability remains in antimicrobial management [14,33]; evolving resistance patterns may further alter the therapeutic landscape. Considerable debate remains over the definitions and incidence of CA-MRSA infections. CA-MRSA infection criteria have varied in the literature, from the narrower epidemiological criteria of the CDC definition to broader criteria, such as the 48-hour criterion, which classifies CA-MRSA infections diagnosed among inpatients cultured within 48 h of admission to a hospital and all outpatients, with the broader definition leading to a higher estimated incidence . Since our study demonstrates that annual US burden is rather sensitive to such definitions, future studies should clearly state the definition criteria used. Despite the variability in costs, our study describes the general magnitude of the problem (i.e. CA-MRSA at least costs several billion dollars, perhaps much more, to society and at least half a billion to third-party payers). Getting a consensus definition of CA-MRSA and obtaining better infection incidence data would further hone this cost estimate.
All models, by definition, are simplifications of real life and cannot include every possible CA-MRSA infection outcome . Our model’s parameter values came from studies with varying rigor and study populations (which may not be comparable) and may change as new studies emerge. For simplicity, we divided infections into discrete syndromic categories, while infections may involve multiple categories or other severe manifestations.
Endeavouring to be conservative (e.g. using the least expensive treatment options and the CDC definition) is likely to underestimate CA-MRSA’s economic burden. Also, our productivity loss calculations assumed a 40-h work week, missed work only during outpatient visits or hospitalization, and did not account for decreased productivity while recovering. Our model did not account for all possible procedures (e.g. pleural drainage and video-assisted thoracoscopic surgery) and permanent disability (e.g. amputation) that may result from severe infections. Our model did not include infection control interventions for CA-MRSA carriers or in-hospital transmission, or their associated costs.
The considerable economic burden of CA-MRSA infections may justify further investment in prevention and control. Much of the overall burden stems from productivity losses. A substantial proportion comes from a minority of cases resulting in death. Although decreasing transmission and infection incidence would decrease costs, early identification and appropriate treatment of infections before they progress could save considerable costs. Therefore, research and policy development may involve developing and implementing measures to reduce transmission, identify infections early, and prevent minor infections from progressing. Although decision makers may have been aware of some of these issues, quantifying their general magnitude could help motivate, plan for and guide investment in relevant interventions and overcome the current general dearth of community-level interventions.
This study was supported by the National Institute of General Medical Sciences Models of Infectious Disease Agent Study (MIDAS) grants 5U54GM088491-02 and 5U01GM087729-03 and by the National Institute of Allergy and Infectious Diseases grant 1RC4AI092327-01. The funders had no role in the design and conduct of the study, collection, management, analysis and interpretation of the data, and preparation, review or approval of the manuscript.
The authors are not aware of any significant conflicts of interest.