SEARCH

SEARCH BY CITATION

Keywords:

  • metastatic bone disease;
  • prevalence;
  • economic burden;
  • metastases

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND

Previous research has documented the prevalence of primary bone cancer; however, there are few data available regarding the impact of metastatic bone disease (MBD) on national expenditure. In this study, the authors quantified the prevalence and direct medical care costs of patients with MBD and the resulting cost impact on U.S. oncology expenditure.

METHODS

Anonymous, patient-level data on health care utilization and cost were obtained from the Thomson Medstat MarketScan research databases. In total, 396,200 patients who were diagnosed with cancer between 2000 and 2004 were selected for the study. Patients with MBD were matched subsequently to non-MBD controls. A 2-part linear regression model was used to compare cases with controls to quantify the incremental cost associated with the disease.

RESULTS

Cancer prevalence in the U.S. during the study period was estimated at 4,861,987 cases annually, and 5.3% (n = 256,137) of those patients had MBD. Rates of MBD were highest in patients with multiple myeloma (28.8%) and lung cancer (15.6%). The mean direct medical cost for all cancers combined was $75,329 for patients with MBD and $31,382 for controls. Regression-adjusted, incremental costs were $44,442 (P < .001) across all cancer types. The incremental cost was highest for patients with multiple myeloma ($63,455) and lowest for patients with lung cancer ($24,946).

CONCLUSIONS

The national cost burden for patients with MBD was estimated at $12.6 billion, which is 17% of the $74 billion in total direct medical cost estimated by the National Institutes of Health, suggesting that MBD is a significant driver of overall oncology cost. Cancer 2007. © 2007 American Cancer Society.

Greater than 10 million Americans currently are living with cancer, either active or in remission, with an additional 1.4 million individuals expected to be diagnosed in 2006.1 Moreover, the annual number of incident cancer cases in the U.S. is expected to double over the next 50 years to 2.6 million, resulting from an aging and expanding population.2 The National Institutes of Health (NIH) estimate overall costs for cancer in 2005 at $210 billion: $74 billion in direct medical expenses, $17.5 billion in indirect morbidity, and $118 billion in indirect mortality.1

With improved medical treatment of many cancers, patients are living longer, which places them at increased risk to develop metastatic disease.1, 3, 4 The skeleton is the third most common target of metastatic cancer and can be the one of the earliest sites afflicted. For example, in breast cancer, bone is both the most common and the earliest site attacked. Ultimately, bone is invaded in 60% to 84% of all cases of metastatic disease, and approximately 70% of those patients experience bone pain.5 Patients with metastatic cancer and bone involvement also are at increased risk for fractures, spinal cord compression, and hypercalcemia, and some scientists have suggested that bone metastases and its associated complications also may contribute to mortality.6, 7

The current standard of treatment for the prevention of skeletal-related events in patients with metastatic bone disease (MBD) is intravenous bisphosphonates. The approved products in this class of compounds include zoledronic acid, pamidronate, and ibandronate (approved in Europe). It is believed that bisphosphonates suppress cancer cell colonization by inhibiting osteoclastic bone resorption, which, otherwise, would release the bone-stored growth factors that feed the cancer cells that colonize bone. Bisphosphonates administered intravenously have demonstrated clinical effectiveness in reducing skeletal complications, although reports on their cost effectiveness have varied. Oral bisphosphonates offer greater convenience and reduced costs, but the results from clinical trials have been mixed.

Despite the emerging importance of MBD in clinical oncology, little information exists regarding its prevalence among cancer patients. A plethora of statistics is available through sources such as the North American Association of Central Cancer Registries, the Surveillance Epidemiology, and End Results (SEER) Program, and American Cancer Society on the prevalence of primary bone cancer; however, to our knowledge, no statistics are available regarding MBD. Similarly, the existing literature regarding cost has not focused on the cost associated specifically with this disease. The objective of the current study was to estimate both the prevalence of diagnosed MBD and its economic burden in the U.S.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Data Source

Data used for the current analyses were derived from the MarketScan Commercial Claims and Encounters (Commercial) and Medicare Supplemental and Coordination of Benefits (Medicare) databases for the period from 2000 to 2004. These databases are derived primarily from employer health insurance plans and have been used widely for diverse health economics research studies.7–10 The data sources contain the inpatient, outpatient, and outpatient prescription drug claims of approximately 14 million individuals of all ages who are covered under a variety of fee-for-service, point-of-service, and capitated benefit plans. Approximately 1 million of the covered lives are from the Medicare database, which consists of individuals with Medicare coverage who have supplemental, employer-funded coverage. For the Medicare population, both the Medicare-paid and employer-paid amounts for services are reported.

Diagnostic information is recorded by physicians and hospitals to support their claims for reimbursement for particular services, and no diagnoses are recorded on pharmacy claims for prescription drugs. In addition, diagnoses are recorded to the extent that they are required for reimbursement; otherwise, the clinical information available to researchers is limited. The Commercial and Medicare databases generally are representative of the U.S. population in terms of sex (49% men). The mean ages of the Commercial and Medicare populations were 34 years and 74 years, respectively. A greater percentage of MarketScan patients (44%) reside in the southern U.S. compared with the general population.

Overview of Analytical Approach

Descriptive and multivariate analyses were performed with SAS software (version 9.1; SAS Institute Inc., Cary, NC). Two study samples were extracted for the analyses. The first sample consisted of patients who were diagnosed with cancer between 2000 and 2004 and was used to estimate the point prevalence of cancer and the rate of MBD in the population. The second sample consisted of a cohort of cancer patients who were newly diagnosed with MBD and a matched sample of non-MBD controls. The difference in costs between the group of patients with MBD and the control group was used to estimate the economic burden of MBD.

Prevalence estimates

National prevalence estimates were produced for patients with MBD in each of the study years under evaluation. Cancer patient counts were obtained first from the MarketScan databases separately for the commercially insured and Medicare eligible enrollees. Then, these data were projected individually to the national commercially insured population and the national Medicare population using weights that were derived from the National Medical Expenditure Survey. Military, Medicaid, and uninsured populations were assumed to have the same cancer and MBD prevalence as the commercially insured population. Once the national prevalence of cancer by tumor type was established, study rates of MBD by tumor type were applied to estimate the number of patients with MBD nationally.

Incremental cost of MBD

Observation of patients in this study began with the first documented diagnosis of cancer. Patients were required to have at least 12 months of continuous enrollment before the first observed cancer date. The index event was triggered once a patient had a cancer diagnosis for nondiagnostic services on 2 different days within a 6-month time frame. MBD was defined by the presence of the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code for secondary malignant neoplasm of bone and bone marrow (198.5) or for a primary cancer type other than bone cancer accompanied by a secondary diagnosis of bone cancer (170.xx). Patients whose initial presentation of cancer was MBD were excluded from the study. Patients whose first cancer diagnosis was in 2004 also were excluded from the study. The follow-up period was divided into 2 stages. The first stage of follow-up began on the study index date and ended the day before the diagnosis of MBD for cases or the imputed date for controls. The second stage began on the date of diagnosis with MBD for cases or the imputed date for controls and continued until in-hospital death, the end of enrollment, or the end of the database (the maximum length of follow-up was 5 years).

For each selected tumor group, study patients were matched 1:1 to controls according to age, sex, geographic region, payer type, index year, and length of follow-up. Claims of the control groups that spanned the longest possible lengths of available prestudy and follow-up periods (from January 1999 to December 2004) were reviewed to verify the absence of MBD.

Key Outcome Variables and Covariates

A patient-level analytical file was constructed to summarize demographic, clinical, and economic data for study-eligible and matched patients. Study period healthcare utilization and costs were summarized overall and by type of service. Cost estimates were based on paid claims, including insurer and health plan payments, copayments, and deductibles. Summary statistics were standardized as the number of services or dollars per month across all patients and controls, regardless of whether a patient or control had the service. Because the data spanned the years from 2000 to 2004, all costs were adjusted to 2004 price levels using the medical care component of the Consumer Price Index. Patients were classified by initial type of observed cancer (ie, breast, prostate, lung, multiple myeloma, or all other neoplasms [except those classified as benign or carcinoma in situ]), as reported according to the ICD-9-CM diagnosis codes recorded in their healthcare claims data. A Charlson comorbidity index (CCI) was developed to adjust for expected resource consumption associated with major health conditions, such as diabetes, heart or liver disease, or renal failure.11 A chronic disease score also was developed to adjust for resource consumption associated with acute conditions.12

Multivariate Analyses

The incremental medical expenditure after MBD diagnosis was estimated using a 2-part regression model that compared MBD patients with a matched control group and adjusted for age, sex, geographic region, relationship to subscriber, urban or rural setting, CCI, expenditure in the period after cancer diagnosis but before MBD, and length of follow-up. The 2-part model consisted of a logistic regression model that predicted the probability of having positive dollars in medical expenditure and a general linear model with an exponential link function that predicted the amount of expenditure among those with expenditure. The predicted probability of having any expenditure (obtained from the first model) was multiplied by the predicted magnitude of the expenditure (which was obtained from the second model) to obtain the total predicted medical expenditure. An estimate of the adjusted incremental cost of MBD was obtained by using the 2-part model to generate a predicted cost difference (MBD vs control) for each member of the MBD cohort and calculating the mean. Cost outliers (n = 35 patients; 0.4%) were excluded from the estimation. Separate models were run for each type of cancer and payer. To assess the statistical significance of the incremental burden, Bonferroni inequality was used to calculate a critical value of 0.0028 that adjusted for multiple testing. Total economic costs of MBD in the U.S. were calculated by multiplying, for each tumor type and payer combination, the regression-adjusted incremental cost per patient by the prevalence of MBD in the U.S. Then, these data were summed to produce the aggregate national cost burden.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Prevalence

Of the 396,200 oncology patients in the MarketScan databases from January 1, 2000, through December 31, 2004, 18,042 patients (4.6%) were identified as having MBD. After applying the projection methodology, the average annual number of U.S. cancer patients estimated during the study period was 4,861,987, with 5.3% (n = 256,137) of these patients projected to have MBD.

Variability in the prevalence of MBD among different types of cancer was considerable. Annual prevalence rates ranged from 3.0% of patients with other types of cancers to 28.8% for those with multiple myeloma. Between these extremes were patients with lung cancer (15.6%), breast cancer (7.4%), and prostate cancer (6.5%). It is important to note that rates of MBD presented in this study are cross-sectional, reflecting the prevalence in the population at a single point in time. Accordingly, they are not intended to reflect the rate of MBD expected over time in any given patient. Table 1 presents estimates of the U.S. prevalence of cancer and MBD in 2004.

Table 1. Estimated Number of Cases with Cancer and Metastatic Bone Disease in the U.S. in 2004 by Payer
Cancer typePrevalence of cancer and MBD, all payersPrevalence MBD by payer, no.
Cancer cases, no.MBD cases, no.MBD rate, %CommercialMedicareOther*
  • MBD indicates metastatic bone disease.

  • *

    Includes the Civilian Health and Medical Program of the Uniformed Services, Medicaid, and uninsured patients.

Breast939,19269,8587.430,79125,30713,761
Prostate735,81247,6076.510,01032,5535044
Lung237,46937,00215.610,14921,7405113
Multiple Myeloma51,38014,78128.8459178772313
Other2,898,13486,8883.029,92840,07316,887
Totals4,861,987256,1375.385,469127,55043,118

Incremental Expenditure

The overall sample for direct cost estimation consisted of 4190 cancer patients without MBD (controls) and 4190 cancer patients with MBD (cases). Table 2 depicts the baseline demographic and clinical characteristics of both populations. The mean age in each cohort was 65.6 years, 53% of patients were men, and between 51% and 52% of patients had Medicare as their primary payer. Twenty-five percent of patients had primary cancer of the lung, 18% had primary cancer of the breast, 12% had primary cancer of the prostate, and 4% had multiple myeloma. The mean (median) length of follow-up was 21 months (18 months), which included 9 months (5 months) from the diagnosis of cancer to the onset of MBD and 12 months (9 months) from diagnosis of MBD to disenrollment or study end. Twenty-four percent of the population was followed for <3 months after diagnosis with MBD, whereas 14% of the population was followed for >2 years. The distribution of cases and controls regionally approximated the United States' population distribution. Patients with MBD were more likely than controls to live in rural areas (P = .049) and to have a history of hypertension (P = .033), diabetes (P = .018), or cerebrovascular disease (P = .001). Accordingly, these patients had a higher CCI (1.8 vs 1.6; P < .0001), although they had lower scores using a broader, medication-based measure of comorbidity, the chronic disease score (4.0 vs 4.4; P < .0001).

Table 2. Baseline Demographic and Clinical Characteristics of Cancer Patients With and Without Metastatic Bone Disease*
CharacteristicCancer cases with MBD (N = 4190)Cancer controls without MBD (N = 4190)P
No.%No.%
  • MBD indicates metastatic bone disease; SD, standard deviation; COPD, chronic obstructive pulmonary disease.

  • *

    After initial diagnosis of cancer.

Cancer type
 Breast74117.774117.71.000
 Prostate52012.452012.41.000
 Lung106025.3106025.31.000
 Multiple myeloma1824.31824.31.000
 Other168740.3168740.31.000
Sex
 Men222053.0222053.01.000
 Women197047.0197047.01.000
Age group, y
 0–17270.6270.61.000
 18–34330.8330.81.000
 35–441162.81162.81.000
 45–545431353912.9.896
 55–64129630.9130031.0.925
 65–74104124.8104124.81.000
 75–8494322.594322.51.000
 ≥851914.61914.61.000
Mean ± SD age, y65.6 ± 12.8 65.6 ± 12.8 .771
Insurance plan type
 Commercial206749.3201248.0.229
 Medicare212350.7217852.0.229
Geographic region
 Northeast64215.364215.31.000
 North central151436.1151436.11.000
 South155537.1155537.11.000
 West47911.447911.41.000
Population density
 Rural97623.390121.5.049
 Urban321276.7328578.4.056
Mean ± SD follow-up, mo
 Commercial21 ± 14 21 ± 14 .432
 Medicare21 ± 15 21 ± 15 .882
Baseline medical history
 Anemia2987.13047.3.800
 Cerebrovascular2345.63077.3.001
 Congestive heart failure1874.52225.3.076
 COPD47811.451512.5.211
 Diabetes46311.153312.7.018
 Hypertension128330.6137432.8.033
 Osteoporosis1012.4902.1.421
 Renal disease2937.02736.5.384
Charlson comorbidity index1.77 ± 2.44 1.57 ± 2.08 .000
Chronic disease score4.01 ± 3.68 4.35 ± 3.79 .000

Patients with MBD had significantly higher expenditures (P < .0001) than their control counterparts across all cancer types, regardless of the type of primary payer (Table 3). Regression-adjusted, mean direct medical costs for all cancers combined was $75,329 for cases and was $31,382 for controls. Mean monthly incremental costs ranged from $24,946 for lung cancer to $63,455 for breast cancer. Cost differentials for outpatient expenditure represented between 63% and 71% of the incremental cost burden, depending on the type of cancer and primary payer. The inpatient cost differential ranged from 23% to 38%. The impact of incremental outpatient pharmacy cost was negligible. Figures 1 and 2 represent the cost differential by site of service, cancer type, and payer.

thumbnail image

Figure 1. Incremental expenditure associated with metastatic bone disease (MBD) in commercially insured patients by cancer and service type.

Download figure to PowerPoint

thumbnail image

Figure 2. Incremental expenditure associated with metastatic bone disease (MBD) in Medicare patients by cancer and service type.

Download figure to PowerPoint

Table 3. Total Expenditure per Patient During the Period After Diagnosis With Metastatic Bone Disease: 2004 U.S. Dollars Adjusted and Unadjusted
Cancer typeCasesControlsUnadjusted difference, $Adjusted difference
No.Mean, $SD, $No.Mean, $SD, $DifferenceSEMean, $P
  1. SD indicates standard deviation; SE, standard error.

Commercial payers
 Breast443108,147119,48944023,06138,41585,086597785,205<.001
 Prostate12087,62397,95411522,54238,80065,081964673,599<.001
 Lung51190,93789,70449755,18486,81735,753556836,195<.001
 Multiple Myeloma83199,889207,09379116,830175,80583,05930,452106,763.003
 Other907102,395123,85887827,12865,52675,267470680,952<.001
 All cancers2064103,728120,070200936,43875,99467,290315469,061<.001
Medicare
 Breast29658,98577,29129926,43654,06732,549547332,506<.001
 Prostate39946,80852,85640518,99530,96427,812306830,872<.001
 Lung54941,41760,53756129,75951,32211,658337614,375<.001
 Multiple Myeloma9978,25394,08310359,35790,20218,89612,97724,909.131
 Other77945,11364,64180724,41354,26920,700300323,049<.001
 All cancers212247,96265,723217526,71752,89521,245182421,711<.001
All payers
 Breast73988,402107,29273924,42645,39863,976429063,455<.001
 Prostate51956,28168,22352019,78132,85436,500332940,276<.001
 Lung106065,28779,902105841,70071,36423,588329624,946<.001
 Multiple Myeloma182132,615166,08818284,304136,82248,31216,03057,720.001
 Other168675,677104,623168525,82660,38549,852295552,350<.001
 All cancers418675,329100,249418431,38265,18643,947185344,442<.001

The national cost burden for patients with MBD in 2004 dollars was estimated at $12.6 billion, or 17% of the $74 billion in total direct medical cost estimated by the NIH. Table 4 presents the national cost burden by cancer type and payer.

Table 4. Estimated 2004 U.S. Expenditures: Patients with Metastatic Bone Disease
Cancer typeExpenditures, $
Employer-sponsored, commercialEmployer-sponsored, medicare eligibleOther payers or uninsured status*Total
  • *

    Other payers include Medicaid and coverage for veterans, military members, and dependents.

Breast2,542,882,597788,006,5531,136,490,3384,467,379,488
Prostate661,158,755914,354,133333,128,4191,908,641,307
Lung339,805,603314,386,795171,212,893825,405,291
Multiple Myeloma506,743,929188,044,250255,325,673950,113,852
Other2,343,939,927876,488,6161,322,536,8354,542,965,378
All cancers6,394,530,8113,081,281,0013,218,694,15812,694,505,970

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Despite the growing impact of MBD in clinical oncology, there is little information regarding the prevalence, utilization, and costs of this disorder among cancer patients. To address these gaps, this study utilized a database that reflected the healthcare experience of approximately 14 million people of all ages who were covered by employer-paid commercial or Medicare supplemental insurance, integrating inpatient, outpatient, and pharmaceutical drug claims. We estimated the prevalence of MBD as well as the incremental, direct medical costs associated with the disease. By comparing patients who had MBD with a similar cohort without the disease, our study accounted for all the excess costs associated with the disease and its complications. To our knowledge, the current study represents one of the first to include detailed, direct, medical, tumor-specific cost information for MBD.

These results were extrapolated to provide an estimate of the overall economic burden of MBD in the U.S. Our findings suggest that 250,000 patients were afflicted with MBD in 2004. Regression-adjusted, incremental, per-patient costs across cancer type and payer were $44,442. This suggests a national economic burden of $12.6 billion, or 17% of the NIH-reported $74 billion in direct medical costs for cancer. Because the number of individuals aged ≥65 years is expected to increase from the 35 million observed in the 2000 Census to 80 million in 2040, it is likely that the cost of MBD will increase accordingly.

Since 1990, several studies examining the cost of cancer in Western Europe and North America have been published.13–35 The design and intent of these studies vary. In general, the studies 1) estimate the economic burden nationally,13–18 2) estimate the cost per patient of specific tumor types,13–15, 19–23 3) examine the cost impact of stage of disease or of metastatic disease,16, 17, 19, 22–30 or 4) evaluate the cost impact of different treatments on specific tumor types.30–35 Methodologically, some studies have used a control group to estimate the incremental cost attributable to cancer, whereas other studies have itemized costs that the investigators considered cancer-related or attributed to cancer the total healthcare cost of patients with the disease. Several studies are specific to 1 payer and, thus, may not be generalizable to the larger population of cancer patients.

A few studies have estimated the costs associated with metastases primarily in patients with breast cancer. Wai et al calculated the costs of treating patients with incurable breast cancer in the Canadian health system from 1995 to 1996. The mean total cost to the health system was $36,474.33 (Canadian) per patient (95% confidence interval, $29,752–43,196 per patient). Berkowitz et al also estimated the lifetime direct costs of treating metastatic breast cancer (MBC). Using data from a variety of sources, those authors estimated that the lifetime costs of treating MBC were $60,000. Rao et al used Medicare data to assess the cost of illness in patients with MBC who were diagnosed between 1997 and 1999: Using an approach similar to the current study, in total, 397 MBC patients were identified and followed for an average of 16.2 months. The mean total cost was $35,164 per MBC patient and $4176 per individual in the control group. Lamerato et al estimated the incremental, annual economic burden of recurrent breast cancer in 2003 dollars at $66,909 ($79,253 vs $12,344, respectively). Studies examining the impact of treatment choices during the last year of life for cancer patients have provided an indirect measure of the cost of disease progression: Berger et al projected mean expenditures between $76,446 and $90,935 during the final year of life.

The use of SEER-reported, 5-year prevalence statistics from 2001 to 2003 to project 2004 rates yields an estimated number of cancer cases in the U.S. of 4,464,163. Prevalence estimates in the current study are re 8.9% higher than this projection. This discrepancy in prevalence estimates is not consistent by tumor types. Prevalence estimates in the current study are lower for prostate and lung cancers, higher for multiple myeloma and other cancers, and almost identical for breast cancer. Using the SEER-estimated prevalence data reduces the projected number of cases of MBD in the current study by 0.5%. The National Cancer Institute estimates that 9.8 million individuals with a history of cancer were alive in 2001. Differences between this estimate and reported prevalence estimates reflect the growing number of Americans who have a history of cancer but remain in remission.

In interpreting our findings, several factors should be considered. Although the current study was based on a large and diverse sample of cancer patients, it was not a random population sample, and the sources of the data are worth reviewing. The MarketScan databases comprise employer-sponsored coverage for active employees, dependents, and retirees. For the current study, we did not obtain specific prevalence or per-patient costs for other payers (Medicaid, military, individual) or for the uninsured. Prevalence and cost data for these populations were presumed to be similar to the commercially insured cohort. To the extent that this assumption is inaccurate, it is possible that the true mean cost is either higher or lower than our estimate.

Several other limitations of this study should be considered when interpreting the results. Specific histologic findings, which are required for American Joint Committee on Cancer tumor staging, generally are not available in claims data. The Medstat disease-staging algorithm can distinguish between metastatic and nonmetastatic disease for most cancer types. However, because the current study included all cancer types, this algorithm was not applied. Accordingly, the greater costs observed in the MBD cohort may reflect in part the differences in cancer stage between the MBD and non-MBD cohorts.

Brown et al. noted that estimating the cost burden in patients with short-term survival rates (those who survive for <18 months) presents a problem analytically, because there may not be enough months of survival to establish full-term initial-care and terminal-care phases. Given the length of follow-up in the current study and our lack of access to survival data, mean per-patient costs may have been understated or overstated. Finally, although the efficacy of treatment with intravenous bisphosphonates has been established as effective in reducing the frequency of skeletal-related events in patients with metastatic bone disease, the cost impact of this class of drugs was beyond the scope of the current study.

In conclusion, to our knowlege, the current study is among the first to include detailed direct, medical, tumor-specific cost information for MBD. Moreover, it addresses many of the shortcomings of previous studies on the overall cost of metastasis given its large, diverse sample of oncology patients, comprehensive data capture based on administrative databases, and use of a control group without MBD to estimate the total incremental costs of the disease. The results suggest that the cost associated with MBD is a significant driver of overall oncology cost with important implications for healthcare delivery systems both in the U.S. and worldwide.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We acknowledge the contributions of Dr. Susan Zelt, Greg Lenhart, and Laurie Costa for their critical editing, reviewing, and analysis.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES