Longitudinal versus cross-sectional methodology for estimating the economic burden of breast cancer: A pilot study

• Published in: PharmacoEconomics 2004-01, Vol.22 (5), p.285-291
• Main Authors: MULLINS, C. Daniel, JUNLING WANG, COOKE, Jesse L, BLATT, Lisa, BAQUET, Claudia R
• Format: Article
• Language: English
• Subjects: Analysis. Health state; Antineoplastic Agents - economics; Antineoplastic Agents - therapeutic use; Biological and medical sciences; Breast cancer; Breast Neoplasms - drug therapy; Breast Neoplasms - economics; Breast Neoplasms - epidemiology; Cancer; Chemotherapy; Cost-of-illness; Cross-Sectional Studies; Databases, Factual; Drugs; Epidemiology; Female; General aspects; Gynecology. Andrology. Obstetrics; Health technology assessment; Humans; Longitudinal Studies; Mammary gland diseases; Maryland - epidemiology; Medicaid; Medical economics; Medical sciences; Methods; Oncology, Experimental; Pilot Projects; Prescribing; Public health. Hygiene; Public health. Hygiene-occupational medicine; Tumors
• ISSN: 1170-7690; 1179-2027
• DOI: 10.2165/00019053-200422050-00002

Projecting future breast cancer treatment expenditure is critical for budgeting purposes, medical decision making and the allocation of resources in order to maximise the overall impact on health-related outcomes of care. Currently, both longitudinal and cross-sectional methodologies are used to project the economic burden of cancer. This pilot study examined the differences in estimates that were obtained using these two methods, focusing on Maryland, US Medicaid reimbursement data for chemotherapy and prescription drugs for the years 1999-2000. Two different methodologies for projecting life cycles of cancer expenditure were considered. The first examined expenditure according to chronological time (calendar quarter) for all cancer patients in the database in a given quarter. The second examined only the most recent quarter and constructed a hypothetical expenditure life cycle by taking into consideration the number of quarters since the respective patient had her first claim. We found different average expenditures using the same data and over the same time period. The longitudinal measurement had less extreme peaks and troughs, and yielded average expenditure in the final period that was 60% higher than that produced using the cross-sectional analysis; however, the longitudinal analysis had intermediate periods with significantly lower estimated expenditure than the cross-sectional data. These disparate results signify that each of the methods has merit. The longitudinal method tracks changes over time while the cross-sectional approach reflects more recent data, e.g. current practice patterns. Thus, this study reiterates the importance of considering the methodology when projecting future cancer expenditure.