1. Top of page
  2. Abstract


A major component of the economic burden associated with the treatment of arthritis relates to surgical joint replacements of the hips/knees. The purpose of this study was to describe the recent trend of hip/knee replacements and the associated economic burden.


The Nationwide Inpatient Survey from 1997 to 2004 was analyzed. The International Classification of Diseases, Ninth Revision, Clinical Modification procedure coding was used to identify joint replacement cases.


During 2004, ∼225,900 (95% confidence interval [95% CI] 201,782–250,018) primary hip replacements and 431,485 (95% CI 397,454–465,516) primary knee replacements were performed. This was a 37% increase in primary hip replacements and a 53% increase in primary knee replacements compared with 2000. The number of primary replacement procedures increased equally for males and females; however, the number of procedures increased excessively among persons age 45–64 years. While Medicare remained the major source of payment (55.4% for primary hip replacements, 59.3% for primary knee replacements), private insurance payment experienced a steeper increase. In 2004, the national bill of hospital charges for hip/knee replacements was $26.0 billion, the hospital cost was $9.1 billion, and the amount of reimbursement was $7.2 billion (28% of hospital charges or 79% of hospital cost). Nearly 600,000 hip replacements and 1.4 million knee replacements will be performed in the year 2015.


The number of joint replacement procedures performed is growing faster than ever. The health care community should consider and prepare for this upcoming demand of surgical loads and its associated economical burden.


  1. Top of page
  2. Abstract

Arthritis is a growing public health concern as the population ages. The elderly population in the US is increasing (1) and the proportion of the population age ≥65 years is projected to increase from 12.4% in 2000 to 19.7% in 2030 (2). Because arthritis affects older adults disproportionately, the increased number of older persons will inevitably increase the number of individuals with arthritis. Other groups with higher arthritis prevalence are those who are overweight (3–5). Among adults in the US, nearly 65% are overweight or obese and 30.4% are obese (6). The high levels of obesity remain a significant concern for arthritis.

A major component of the economic burden associated with the treatment of arthritis relates to surgical joint replacements of the hips or knees. When nonsurgical treatment modalities prove unsuccessful, surgical treatment is then considered for pain reduction (7). Joint replacements effectively provide improved physical function as well as pain relief (8), and these surgical treatments have received widespread acceptance. Since 1990, hip and knee replacements have steadily increased in the US (9, 10). As joint replacement became a more popular option, the economic burden resulting from these surgical procedures grew substantially. During the year 2000 in the US, ∼440,000 total hip/knee replacement procedures were performed and their costs were estimated to be nearly $10 billion (11).

With arthritis being recognized as an increasingly important public health concern, an ongoing surveillance of this condition and its costs was listed as a National Arthritis Action Plan (12). The purpose of the current study was to describe the recent trend of hip/knee replacements in the nation. Unlike what has been seen in other countries (13, 14), the literature on the projected future burden in the US is limited. One published study has projected that the national burden of knee revision surgery would be $2 billion by 2030 based on cost data from a single surgeon practice (15). However, the economic burden of primary hip/knee replacement and its future projection still remain understudied. To address this issue, historical data were analyzed to describe/project the burden of hip/knee replacements.


  1. Top of page
  2. Abstract

Data sources.

The Nationwide Inpatient Sample (NIS), a part of the Healthcare Cost and Utilization Project (HCUP), is a database of hospital inpatient stays. The NIS is the only all-payer inpatient care database that includes billing information in the US (16). It contains 5–8 million discharge records per year. In 2004 alone, the NIS contained a probabilistic sample of 8 million records from nearly 1,000 hospitals. Details concerning sampling procedures have been published elsewhere (17). Briefly, the NIS is the principal source for national data concerning characteristics of patients discharged from nonfederal, short-stay hospitals. This annual survey is conducted under the auspices of the Agency of Health Research and Quality. The survey selects probability samples of 20% of US community hospitals to produce a nationally representative sample.

Outcome measures.

The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) primary procedure coding indicative of primary hip/knee replacement was used (81.51 for total hip replacement, 81.53 for revision hip replacement, 81.54 for total knee replacement, and 81.55 for revision knee replacement). The partial hip replacement code (ICD-9-CM 81.52) was not included in this study. Based on hospitalization records, hospital charges and primary sources of payment were analyzed. Hospital charges refer only to the amount the hospital billed, and do not include professional (physician) fees.

Because the hospital discharge database included hospital charges only, the hospital costs and reimbursements were estimated based on Medicare summary information published by Solucient (18). Solucient's Medicare database is based on the Medicare Provider Analysis and Review file (MedPAR), which is produced by the Center for Medicare and Medicaid Services (CMS). MedPAR contains a complete census of all Medicare inpatient discharges (more than 20 million in federal fiscal year 2004). Every US acute-care hospital that discharges at least 1 Medicare patient is represented in Solucient's database. Solucient applies Medicare cost-to-charge ratios to the data in MedPAR. A hospital's cost-to-charge ratios are derived from the worksheet of the Medicare cost report submitted to CMS annually (18).

Medicare provides payments to hospitals for inpatient services through the Diagnosis Related Group (DRG) system. The DRG is a patient-classification scheme used for price-per-case reimbursement. For both primary and revision joint replacements, Medicare reimbursed hospitals under a single DRG (209 major joint and limb reattachment procedures of the lower extremity) in the year 2004. The hospital loss is calculated by the difference between the hospital cost and hospital reimbursed amount.

Statistical analysis.

Because the NIS is a probabilistic sample survey, the sampling weight and sampling design were considered in calculating total hospitalizations and their standard errors correctly. Sampling weights were used to account for the unequal sampling probabilities and to produce estimates for all hospital discharges in the US. The HCUPnet, a Web-based interactive analytic tool, was used to generate statistics from the NIS (19). Because hospital charges were typically skewed to the right, a typical amount (median values) of hospital charges were reported. The magnitude of burden in the future was projected using the trend in recent years. To reflect exponential growth, a linear regression with a log-transformed value (power law) was used in projecting the number of primary procedures in the US in the year 2015. The power laws, the most frequent scaling laws, can be used if the relationship between Y (replacements) and X (year) can be written as Y = αXβ. Two constants α and β are the proportionality and the exponent of the power law, respectively. Because taking logs of both sides gives a straight line, ln(Y) = ln(α) + βln(X), a linear regression model was fitted to the log-transformed data.


  1. Top of page
  2. Abstract

Joint replacements in the US.

The number of primary joint replacements performed in the US was rapidly on the rise (Figure 1). The growth of procedures was especially excessive from 1999, and its growth pattern started to follow an exponential function. In 2004, approximately 225,900 primary hip replacements (95% confidence interval [95% CI] 201,782–250,018) and 431,485 primary knee replacements (95% CI 397,454–465,516) were performed in the nation. This was a 37% increase in primary hip replacements and a 53% increase in primary knee replacements compared with the year 2000 (164,458 total hip replacements and 281,534 total knee replacements) (Table 1). During the same period, the joint revisions were also steadily increasing, although the numbers of procedures performed were smaller (Figure 1). The number of revision hip replacements was between 16% and 21% of primary hip replacements; the number of revision knee replacements was between 8% and 10% of primary knee replacements during this period. Proportions of revision in a given year were up and down without any distinct pattern of change.

thumbnail image

Figure 1. Estimated number of joint replacements performed in the US from 1997–2004 (the Healthcare Cost and Utilization Project, National Inpatient Survey, 2004). The error bar represents the upper and lower bound of the 95% confidence interval.

Download figure to PowerPoint

Table 1. Economic burden of total hip and knee replacements performed in the US*
YearFrequency (95% CI)Median hospital chargesNational bill of hospital charges in million dollars
  • *

    95% CI = 95% confidence interval; ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification.

Primary total hip replacement (ICD-9-CM code 81.51)   
 1997153,080 (141,775–164,385)$20,141$3,387 (3,122–3,652)
 1998154,337 (142,669–166,005)$19,672$3,385 (3,109–3,661)
 1999156,706 (144,864–168,548)$20,497$3,599 (3,305–3,893)
 2000164,458 (150,948–177,968)$22,323$4,231 (3,841–4,621)
 2001184,646 (164,319–204,973)$24,155$5,080 (4,518–5,643)
 2002201,166 (184,296–218,036)$26,699$6,118 (5,585–6,651)
 2003201,545 (185,936–217,154)$29,825$7,012 (6,394–7,631)
 2004225,900 (201,782–250,018)$32,571$8,316 (7,326–9,306)
Revision hip replacement (ICD-9-CM code 81.53)   
 199724,649 (21,680–27,618)$23,126$793 (700–885)
 199831,885 (28,100–35,670)$23,110$841 (725–958)
 199931,124 (27,522–34,726)$23,833$872 (735–1,009)
 200033,990 (29,609–38,371)$26,169$1,048 (890–1,207)
 200138,918 (33,961–43,875)$28,213$1,305 (1,100–1,509)
 200236,856 (32,687–41,025)$30,773$1,383 (1,215–1,550)
 200334,688 (30,860–38,516)$35,131$1,573 (1,298–1,847)
 200437,115 (32,554–41,676)$37,940$1,684 (1,445–1,923)
Primary total knee replacement (ICD-9-CM code 81.54)   
 1997264,331 (245,611–283,051)$19,309$5,618 (5,202–6,034)
 1998251,309 (233,138–269,480)$18,946$5,325 (4,930–5,721)
 1999262,687 (241,201–284,173)$19,700$5,847 (5,324–6,370)
 2000281,534 (260,476–302,592)$21,067$6,612 (6,062–7,162)
 2001313,618 (289,151–338,085)$22,570$7,994 (7,332–8,656)
 2002350,122 (323,803–376,441)$24,705$9,831 (9,072–10,590)
 2003379,719 (353,194–406,244)$26,680$11,867 (10,959–12,775)
 2004431,485 (397,454–465,516)$29,509$14,567 (13,248–15,887)
Revision knee replacement (ICD-9-CM code 81.55)   
 199723,503 (21,092–25,914)$19,946$531 (473–589)
 199823,671 (21,194–26,148)$19,938$537 (475–599)
 199924,971 (21,782–28,160)$21,341$613 (571–704)
 200024,730 (21,980–27,480)$23,240$667 (571–762)
 200127,802 (24,833–30,771)$25,464$843 (732–953)
 200229,948 (26,855–33,041)$28,588$1,008 (893–1,125)
 200331,274 (27,950–34,598)$31,938$1,249 (1,035–1,464)
 200435,048 (31,349–38,747)$35,366$1,457 (1,278–1,637)

Primary hip/knee replacements by demographics.

There was a change in age distribution among persons receiving primary joint replacements. The total number of surgical procedures was highest in persons between 65 and 84 years of age compared with any other age group (Table 2). In 1997, ∼60.2% of the primary hip replacement surgeries and 69.1% of knee replacements were performed for individuals age 65–84 years. Although elderly persons remained the main recipients of hip/knee replacements, joint replacements increased excessively among persons age 45–64 years. Approximately 26.5% of the primary hip surgeries were performed in persons age 45–64 years in 1997 compared with 36.0% in 2004. Compared with the US census data (20), age-specific hip replacements increased 30% (from 30 to 39 per 10,000 people) among persons age 65–84 years and increased 71% (from 7 to 12 per 10,000 people) among persons age 45–64 years. A similar trend was observed in knee replacement procedures. The number of knee replacements increased by 38% (from 60 to 83 per 10,000 people) among individuals age 65–84 years compared with an 83% increase (from 12 to 22 per 10,000 people) among individuals age 45–64 years old. Although it is not shown in the table, there was no pattern of change in proportion to sex among hip/knee replacement recipients during the study period. Between approximately 57.5% and 59.0% of hip replacement recipients were women without any patterns of change over time. The proportion of women among knee replacement recipients also fluctuated between 57% and 64% without any trend over time.

Table 2. Changes in age of hip/knee replacement recipients between 1997 and 2004 in the US
Frequency%*Per 10,000 peopleFrequency%*Per 10,000 peopleFrequency%*Per 10,000 people
  • *

    Column percentage.

  • Per 10,000 population in each age category for the given year.

  • Estimates are not reliable (relative standard error >0.3).

Primary total hip replacement         
 ≤171630.1 161<0.10
Revision hip replacement         
Primary total knee replacement         
Revision knee replacement         

Economic burden of joint replacement.

Due to the increasing number of joint replacement procedures performed as well as the increasing amount of hospital charges, the national bill of primary joint replacements grew steeply (Table 1). The hospital charges for joint replacements increased faster than the rate of inflation. The inflation factor from the Consumer Price Index (CPI) between 1997 and 2004 was 1.18 (21). However, the median hospital charge increased from $20,141 to $32,571 (1.62 times) for hip replacements and from $19,309 to $29,509 (1.53 times) for knee replacements. In 2004, approximately $8.3 billion (95% CI $7.3–$9.3 billion) was charged for primary hip replacements and another $14.6 billion (95% CI $13.2–$15.9 billion) was charged for primary knee replacements. Collectively, approximately $22.9 billion was charged for 657,000 primary hip/knee replacement procedures in 2004 alone.

The hospital charges for revision joint replacements also increased over time. In 2004, compared with primary hip/knee replacements, the median hospital charge was ∼$5,000 higher for revision replacements; the median hospital charge for revision hip replacements was $38,000 and for primary hip replacements was $33,000. Likewise, the median hospital charge for revision knee replacements was also $5,000 higher than that of primary knee replacements ($35,000 for revision compared with $30,000 for primary). Approximately $1.68 billion (95% CI $1.4–$1.9 billion) was charged for revision hip replacements and $1.46 billion (95% CI $1.3–$1.6 billion) was charged for revision knee replacements. Jointly, nearly $3.1 billion was charged for 72,000 revision hip/knee replacement procedures in the year 2004. According to Medicare summary information in 2004, the surgery cost and reimbursement for DRG 209 was approximately $12,477 and $9,839, respectively (18). Based on the Medicare data, the hospital costs and the actual transaction of reimbursements in the year 2004 were estimated to be $9.1 billion (729,000 × $12,477) and $7.2 billion (729,000 × $9,839), respectively.

Primary payers for hip/knee replacements.

The economic burden of Medicare for primary hip replacements increased from $2.1 billion in 1997 to $4.6 billion in 2004 (Table 3). However, compared with other sources of payment, the relative burden decreased; Medicare was a principal source of payment for 62.8% of hip replacements in 1997, but declined to 55.4% in 2004. Instead, the burden on private insurance systems increased from 30.4% to 36.2%. Together with the increases in relative shares of payment, an absolute increase in the number of hip replacements, and hospital charges, the burden of private insurance increased to $3.0 billion compared with $1.1 billion in 1997. A parallel trend was observed in knee replacements. During the same period, the proportion of knee replacements primarily paid for by private insurance systems increased from 25.5% to 32.0%. Private insurance payments for knee replacements increased from $1.46 billion in 1997 to $4.64 billion in 2004. A similar pattern was observed for revised hip and knee replacements.

Table 3. Changes in primary payers for the hip and knee replacements between 1997 and 2004 in the US
N*%Total charges, millionN*%Total charges, millionN*%Total charges, million
  • *

    Number performed each year in the US.

  • Column percent.

  • Uninsured or other sources.

Primary total hip replacements         
Revision hip replacements         
Primary total knee  replacements         
Revision knee replacements         

Projection of hip/knee replacements in year 2015.

To fit data to exponential growth (Figure 1), a power law was applied. The recent trend of hip/knee replacements (1999–2004) fit well to the model (R2 = 0.96 for hip replacements; R2 = 0.99 for knee replacements). Based on this model, the number of primary hip/knee replacements in the prediction interval was calculated to the year 2015 (Figure 2). If current trends continue over the next decade, by the year 2015 the annual numbers of primary hip and knee replacements will be greater than 0.5 million and 1.3 million, respectively. The inflation factor from the CPI is estimated to be 1.297 in the year 2015 compared with the year 2004 (21). Therefore, if the exponential growth continues, the national bill of hospital charges will be at least $21 billion for primary hip replacements and another $49 billion for primary knee replacements.

thumbnail image

Figure 2. Projected number of joint replacements in the US from 2005–2015. Solid line = projection; broken line = 95% prediction interval; triangles = observed data points.

Download figure to PowerPoint

If approximately the same proportion of revisions are performed compared with primary procedures (20% of primary hip replacements and 10% of primary knee replacements), nearly 100,000 hip revisions and 127,000 knee revisions are expected to be performed in 2015. Considering an inflation factor of 1.297, the total hospital charges will be at least $4.9 billion for hip revisions and $5.8 billion for knee revisions. Therefore, if the hospital charge grows with the inflation factor, the national bill of hospital charges will be nearly $80.2 billion for all primary and revised hip/knee replacements in the year 2015.


  1. Top of page
  2. Abstract

These national estimates found in the NIS were close to those found in the National Hospital Discharge Survey (NHDS) for the same study years (9), and this fact enhances the validity of the estimates that included other years that did not overlap. Although some previous studies reported increasing arthritis prevalence and surgical procedures, the recent increase in joint replacements is alarmingly steeper, and its pattern of change fits to an exponential growth. Some known factors contributing to the increase in joint replacements are an increased elderly population and overweight population. From 2000 to 2004, the population age ≥65 years increased ∼5% (from 35.1 million to 36.8 million) (20). However, during the same period, primary hip replacements and knee replacements increased by 37% and 53%, respectively. The impact of other factors on hip/knee replacements may be significantly larger than that of the aging of the population. Therefore, a projection based on the change in demographic distribution alone may grossly underestimate the future burden. The growth rate of these procedures was also steeper compared with the number of orthopedic surgeons during the same period. According to the American Academy of Orthopaedic Surgeons, only 1,220 more orthopedic surgeons were in practice in 2004 compared with 2000 (17,486 versus 16,266) (22). This means that the increase in orthopedic surgery manpower was only 6.7% (5.8 versus 6.2 orthopedic surgeons per 100,000 population). Therefore, the role of supplier-induced demand was relatively small in this phenomenon. With the advancement of device and surgical techniques, people may perceive orthopedic surgery to be a safer option to consider than before. Further study is needed to address the role of perceived safety in choosing these procedures.

Another noticeable phenomenon is the relatively large increase of replacements among persons age 45–64 years. This trend may be explained by both increases in the overweight population at earlier ages and the change in their expectation of quality of active life. The total number of hip replacements increased by 71% (from 7 to 12 per 10,000 people), and the total number of knee replacements increased by 83% (from 12 to 22 per 10,000). It is not clear why the rate of increase for total knee replacements was slightly higher than for total hip replacements. Further study is needed to better understand this phenomenon. Currently, most replacements last 10–15 years. In an effort to avoid complicated revision operations, doctors encourage people to put off hip replacement as long as possible. However, people may receive hip/knee replacements before reaching the age of 65 to improve their functionality as well as to reduce pain. This new phenomenon has some important implications for both surgical loads and economic burden. The recipients are not eligible for Medicare, yet, accordingly, the increasing number of replacements performed before the age of 65 will add to the burden of private insurance. The major increase in primary replacements among persons age 45–64 years will inevitably amplify the need for hip/knee revisions in the near future as well. Compared with the number of primary replacements, this study demonstrated that approximately 20% of hip revisions and 10% of knee revisions were performed annually. However, these proportions are likely to increase unless a longer-lasting prosthesis is introduced. Due to the complexity of the procedure, revision arthroplasty needs more operative time, requires longer hospital stays, and has a higher complication rate compared with primary arthroplasty (23). The higher hospital resource utilization of revision will add extra loads of burden to our health system.

During the year 2004, the national bill of hospital charges for hip/knee replacements was $26.0 billion and ∼28% ($7.2 billion) of hospital charges were reimbursed to the hospitals. This reimbursed amount was only 79% of the total hospital cost ($9.1 billion), and this deficit suggests that hip/knee replacements were a substantial burden to the hospitals performing these surgeries. Under this DRG 209 reimbursement system, this could be more problematic for the revision of hip/knee replacements, which often require additional hospital resources. The discrepancy between resource utilization and Medicare reimbursement for revision procedures has resulted in substantial financial losses for hospitals that perform these procedures (24). A study reported that the hospital loss on each hip revision during 1995–1999 was nearly $5,402 (25). The amount of loss in the current study was $2,638 (= $12,477 − $9,839) for all primary or revised procedures together. Because revisions are known to be more costly procedures, the present study also suggests that the revision procedures cause financial loss to the hospitals. As pointed out by the president of the American Association of Hip and Knee Surgeons, this could ultimately jeopardize patient access to care as hospitals attempt to limit ongoing losses (26). If hospitals are reluctant to perform these procedures, our health care community may face a challenge in responding to the ever-increasing surgical demand.

After thorough investigation, CMS officially adopted new ICD-9-CM diagnosis and procedure codes related to revision total replacements and published them in the Federal Register on August 12th, 2005 (27). CMS eliminated DRG 209 and created 2 new DRGs: 544 (major joint replacement or reattachment of lower extremity) and 545 (revision of hip or knee replacement). These codes went into use beginning in October 2005. CMS states in this Federal Register, “We believe that the creation of the new DRGs for revisions of hip and knee replacements should resolve payment issues for hospitals that perform the more difficult revisions of joint replacements” (27). Under the CMS proposed DRG structure, the amount of reimbursement was $11,018 for total joint replacements (DRG 544) and $13,922 for revision joint replacements (DRG 545) in 2005. In 2004, the reimbursement and hospital costs for DRG 209 were $9,839 and $12,477, respectively. Because Medicare is the single largest payer of joint replacements and other payers often use Medicare reimbursement in their practice, the creation of a new DRG for revisions of hip and knee replacements is expected to provide financial relief for hospitals. To date, the impact of this new DRG system is not yet reported.

Previously, a study projected the number of primary hip and knee replacements to be 453,000 and 2.16 million, respectively, in the year 2030 (28). However, taking the recent trend into account, the current study cautiously concludes that these projected numbers could be reached a decade sooner than the year 2030. Because the current study demonstrated overwhelmingly faster growth of hip/knee replacements compared with rates of aging of the population in the past, the future projection based on the aging of the population will underestimate the future burden.

Both strengths and weaknesses of this study relate to the characteristics of the NIS databases. The strength of the NIS data relates to the large sample size. Despite the use of a probabilistic sampling technique, the large sample size yields a reasonably tight confidence interval. The projection was based on the pattern found in recent years by assuming that the social/health environment would change at the same rate. Projection based on known risk factors would be useful to understand the percentage of variances explained by the projection model. However, the factors driving this recent increase are not well described yet. Other measures such as tolerance of pain and people's perception of safety of procedures are often hard to quantify. The prediction was made to better understand the direction of our utilization of arthroplasty. However, it should be interpreted within the limitation of projection. The nature of any projection is the extrapolation of a model; therefore, the reliability becomes smaller for long time interval predictions.

In estimating the reimbursement and hospital costs in the nation, Medicare summary information was used. Medicare is the single largest payer for these procedures. However, economic burden on the US health care system would be reflected more accurately if databases included actual reimbursements from all types of payers. Another limitation of this study is the inability to include the data for the last 2 years and the type of prosthesis. Many other countries use their arthroplasty registry data to follow the trends or the survivorship of the arthroplasties (29–31). However, because an arthroplasty registry does not exist at the national level in the US at present time, either NHDS or NIS data should be used to describe the burden of arthroplasty in the nation. Due to the large scale of data collection, NHDS or NIS data are available to researchers approximately 2 years later than the referenced year. Therefore, dissemination of timely information is greatly compromised. Establishment of a national joint replacement registry may provide timely surveillance reports as well as other critically important information such as the survivorship of various prostheses (32). A less-invasive surgery for hip and knee replacements can be performed as outpatient procedures for selected patients (33).

Less-invasive surgery encompasses both small-incision techniques and minimally invasive techniques. Small- incision hip and knee replacement surgery entails performing the conventional approach through a smaller skin incision than historical incisions. Minimally invasive surgery uses not only a smaller incision, but also new exposure techniques, which are less invasive to soft tissue and bone (34). Because the current study utilized the NIS database, it did not capture the number of surgeries performed on outpatients. This inability to track the utilization of hip/knee replacements in the nation may be another reason that the arthroplasty registry is needed.

In conclusion, the results of the present study demonstrate that the burden resulting from hip/knee joint replacement is not only substantial but also increasing at an alarmingly steep rate. Public health education is critically important to reduce the proportion of people who are overweight/obese as well as to manage arthritis at earlier stages. At the same time, given the steeply increasing trend of joint replacements and the expected number of joint revisions needed, the health care community should be prepared for this upcoming demand of primary/revision surgical loads and its economic burden on government and private insurance systems.


  1. Top of page
  2. Abstract

Dr. Kim had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Kim.

Acquisition of data. Kim.

Analysis and interpretation of data. Kim.

Manuscript preparation. Kim.

Statistical analysis. Kim.


  1. Top of page
  2. Abstract