RESUMEN
Marginal analysis evaluates changes in a regression function associated with a unit change in a relevant variable. The primary statistic of marginal analysis is the marginal effect (ME). The ME facilitates the examination of outcomes for defined patient profiles or individuals while measuring the change in original units (e.g., costs, probabilities). The ME has a long history in economics; however, it is not widely used in health services research despite its flexibility and ability to provide unique insights. This article, the second in a two-part series, discusses practical issues that arise in the estimation and interpretation of the ME for a variety of regression models often used in health services research. Part one provided an overview of prior studies discussing ME followed by derivation of ME formulas for various regression models relevant for health services research studies examining costs and utilization. The current article illustrates the calculation and interpretation of ME in practice and discusses practical issues that arise during the implementation, including: understanding differences between software packages in terms of functionality available for calculating the ME and its confidence interval, interpretation of average marginal effect versus marginal effect at the mean, and the difference between ME and relative effects (e.g., odds ratio). Programming code to calculate ME using SAS, STATA, LIMDEP, and MATLAB are also provided. The illustration, discussion, and application of ME in this two-part series support the conduct of future studies applying the concept of marginal analysis.
Asunto(s)
Investigación sobre Servicios de Salud/economía , Modelos Económicos , Modelos Estadísticos , Humanos , Análisis de Regresión , Programas InformáticosRESUMEN
BACKGROUND AND OBJECTIVE: Advanced prostate cancer patients with bone metastasis are predisposed to skeletal complications termed skeletal-related events (SREs). There is limited information available on Medicare costs associated with treating SREs. The objective of this study was to ascertain SRE-related costs among older men with metastatic prostate cancer in the US. METHODS: We analysed patients aged 66 years or older who were diagnosed with incident stage IV (M1) prostate cancer between 2000 and 2007 from the linked Surveillance, Epidemiology and End Results (SEER)-Medicare dataset. A propensity score for the incidence of an SRE was estimated using a logistic regression model including demographic and clinical baseline variables. Patients with SREs (cases) were matched to patients without SREs (controls) based on the propensity score, length of follow-up (i.e. date of prostate cancer diagnosis to last date of observation) and death. Health resource utilization cost differences between cases and controls over time were compared using generalized linear models. Healthcare costs were examined by type of SRE (pathological fracture only, pathological fracture with concurrent surgery, spinal cord compression only, spinal cord compression with concurrent surgery, and bone surgery only) and by source of care (inpatient, physician/non-institutional provider, skilled nursing facility, outpatient and hospice). All costs were adjusted to 2009 US dollars, using the medical care component of the Consumer Price Index. RESULTS: Application of the inclusion criteria resulted in 1,131 metastatic prostate cancer patients with SREs and 6,067 patients without SREs during follow-up. The average age of the sample was 79 years, and 14 % were African American. A total of 928 patients with SREs were matched to 928 patients without SREs. The average health care utilization cost of patients with SREs was US$29,696 (95 % confidence interval [CI] US$24,730-US$34,662) higher than that of the controls. The most expensive SRE group was spinal cord compression with concurrent surgery (US$82,868: 95 % CI US$67,472-US$98,264) followed by bone surgery only (US$37,496: 95 % CI US$29,684-US$45,308), pathological fracture with concurrent surgery (US$34,169: 95 % CI US$25,837-US$ 42,501), spinal cord compression only (US$25,793: 95 % CI US$20,933-US$30,653) and pathological fracture only (US$14,649: 95 % CI US$6,537-US$22,761). The largest cost difference by source of care was observed for hospitalizations (p < 0.01). CONCLUSION: Metastatic prostate cancer patients with SREs incur higher costs compared to similar patients without SREs. SRE costs among older stage IV (M1) prostate cancer patients vary by SRE type, with spinal cord compression and concurrent surgery costing at least twice as much as other SREs.
Asunto(s)
Neoplasias Óseas/economía , Fracturas Óseas/economía , Costos de la Atención en Salud , Neoplasias de la Próstata/economía , Compresión de la Médula Espinal/economía , Anciano , Neoplasias Óseas/complicaciones , Neoplasias Óseas/secundario , Costos y Análisis de Costo , Fracturas Óseas/etiología , Fracturas Óseas/terapia , Humanos , Masculino , Medicare , Estadificación de Neoplasias , Puntaje de Propensión , Neoplasias de la Próstata/patología , Compresión de la Médula Espinal/etiología , Compresión de la Médula Espinal/terapia , Estados UnidosRESUMEN
AIM: The prognostic effects of chemotherapy and various lymph node measures [positive nodes, total node count and the positive lymph node ratio (PLNR)] have been established. It is unknown whether the cancer-specific survival benefit of chemotherapy differs across these nodal prognostic categories. METHOD: This retrospective analysis of linked Surveillance, Epidemiology and End Results (SEER) data and Medicare data (SEER-Medicare)included patients ≥ 65 years of age with a diagnosis of stage III colon cancer between 1997 and 2002. We grouped patients according to the number of positive nodes (N1 and N2), total node count (≥ 12 and < 12 total nodes) and PLNR (below the 75th percentile and at least at the 75th percentile of the PLNR). The end point was colon cancer-specific mortality. RESULTS: Fifty-one per cent (3701) of the 7263 patients received adjuvant therapy during the time period 1997-2002. The mean (standard deviation) number of total nodes examined was 13 (9) and the number of positive nodes identified was 3 (3). Patients with N2 disease, < 12 total nodes examined and a high PLNR had a worse survival at 2, 3 and 5 years following colectomy. Utilization of chemotherapy demonstrated a colon cancer-specific survival benefit (hazard ratio at median follow up = 0.7; P < 0.001) that was consistent and statistically significant across the three nodal prognostic categories examined. CONCLUSION: The benefit of chemotherapy did not vary based on N stage, total node count or PLNR. The results favour a broad-based approach towards increasing the chemotherapy treatment rates in stage III patients of ≥ 65 years of age, rather than an approach that targets clinical subgroups.
Asunto(s)
Quimioterapia Adyuvante , Neoplasias del Colon/tratamiento farmacológico , Neoplasias del Colon/mortalidad , Anciano , Anciano de 80 o más Años , Neoplasias del Colon/patología , Femenino , Humanos , Metástasis Linfática , Masculino , Medicare , Estadificación de Neoplasias , Pronóstico , Modelos de Riesgos Proporcionales , Estudios Retrospectivos , Programa de VERF , Análisis de Supervivencia , Estados Unidos/epidemiologíaRESUMEN
OBJECTIVE: Our objective was to determine how patient demographics and outpatient referrals to specialized dementia (DEM) or mental health (MH) clinics influence receipt of anti-dementia (AD), antidepressant (ADEP), antipsychotic (APSY) and sedative-hypnotic (SEDH) medications among veterans with dementia. DESIGN: Retrospective, cross-sectional observational study. SETTING: Veterans Affairs Maryland Health Care System (VAMHCS). PARTICIPANTS: Veterans aged ≥ 60 years with Alzheimer's or related dementia diagnosis after 1999 with minimum of one-year follow-up or death were included. MEASUREMENTS: Retrospective analysis of VAMHCS electronic medical records were used to determine predictors of AD, ADEP, APSY, and SEDH prescribing using logistic regression models that examined visits to DEM or MH clinics, patient age, follow-up time, race/ethnicity and marital status. RESULTS: Among 1209 veterans with average follow-up of 3.2 (SD 1.9) years, 36% percent had MH visits, 38% had DEM visits and 19% visited both clinics. DEM visits were associated with AD and ADEP but not APSY medication receipt (OR(AD:DEM) = 1.47, 95% CI = (1.052, 2.051); OR(ADEP:DEM) = 1.66, 95% CI = (1.193, 2.302); OR(APSY:DEM) = 1.35, 95% CI = (0.941, 1.929)). MH visit was associated with ADEP and APSY medication receipt (OR(AD:MH)\ = 1.16, 95% CI = (0.821, 1.631); OR(ADEP:MH) = 2.83, 95% CI = (2.005, 4.005); OR (APSY:MH) = 4.41, 95% CI = (3.109, 6.255)). CONCLUSION: In the VAMHCS dementia population, visits to DEM or MH specialty clinics increase the odds of receiving AD, ADEP, and APSY medications.
Asunto(s)
Instituciones de Atención Ambulatoria/clasificación , Atención Ambulatoria/estadística & datos numéricos , Demencia/tratamiento farmacológico , Pautas de la Práctica en Medicina/estadística & datos numéricos , Veteranos , Anciano , Anciano de 80 o más Años , Enfermedad de Alzheimer/tratamiento farmacológico , Instituciones de Atención Ambulatoria/estadística & datos numéricos , Estudios Transversales , Utilización de Medicamentos , Registros Electrónicos de Salud , Femenino , Humanos , Masculino , Maryland , Servicios de Salud Mental , Persona de Mediana Edad , Psicotrópicos/uso terapéutico , Estudios Retrospectivos , Estados Unidos , United States Department of Veterans AffairsRESUMEN
INTRODUCTION: Predictors of hospital discharges against medical advice (AMA), including race/ethnicity, have been examined previously. However, the predictive effect of an admission for rehabilitative care has not been examined in an inpatient stroke population, nor has the impact of race/ethnicity on this relationship been reported. METHODS: Live hospital discharges with a primary diagnosis of stroke from 2000-2005 were identified in a longitudinal dataset. The outcome of interest was a discharge AMA. A hierarchical logistic model was estimated to examine the effect of race/ethnicity and rehabilitative care on the outcome while controlling for patient and hospital characteristics. RESULTS: A total of 569 of the 79,561 stroke admissions (0.7%) ended in a discharge AMA. There were 1,565 admissions for rehabilitative care and 32% of patients were non-Caucasian. Among Caucasians, adjusted odds of a discharge AMA were higher for patients with an admission for rehabilitative care (AOR = 3.83, p < 0.0001). Among those not admitted for rehabilitative care, non-Caucasian patients were more likely to leave AMA (AOR = 1.4; p = 0.0005). CONCLUSIONS: This study identifies dependence between race/ethnicity, rehabilitative care, and discharges AMA. More research is needed to understand the implications of differential rates of discharges AMA among race/ethnic groups and across patient care settings.