A general instrumental variable framework for regression analysis with outcome missing not at random

674 Background: Although the safety of chemotherapy without primary tumor resection (PTR) has been established, questions remain regarding potential survival benefit with PTR. The purpose of this study was to compare mortality with and without PTR among patients with unresectable metastatic colon cancer using nationwide hospital based cancer registry data. Methods: An observational study was conducted of patients with stage 4 colon cancer identified from the National Cancer Data Base (2003-2005). Patients who underwent metastectomy were excluded. Patient, treatment, and hospital data were analyzed. Multivariate Cox regression stratified by receipt of chemotherapy was performed to compare survival with and without PTR. To account for treatment selection bias, Propensity Score Weighting (PSW) and Instrumental Variable (IV) analyses, using hospital-level PTR rate as the instrument, were performed. In order to account for the potential bias associated with early comorbidity or disease burden associated deaths (survivor treatment bias), 1 year landmark analysis was performed. Results: A total of 14,399 patients met inclusion criteria and 6,735 patients were eligible for landmark analysis. PTR was performed in 38.2% of the total cohort and 73.8% of those at landmark. Using multivariate Cox regression analysis, PTR was associated with a significant reduction in mortality (HR 0.39; 95% CI, 0.38-0.41). This effect persisted with PSW (HR 0.4; 95% CI, 0.38-0.43). However, IV analysis showed a much smaller effect, (RR 0.88; 95% CI, 0.83-0.93). While a smaller benefit was seen on landmark analysis using multivariate Cox regression (HR 0.6; 95% CI, 0.55-0.64) and PSW (HR 0.59; 95% CI, 0.54-0.64), IV analysis showed no improvement in survival with PTR (RR 0.97; 95% CI, 0.87-1.06). Stratification by chemotherapy did not alter the results. Conclusions: Among patients with stage IV colon cancer, PTR offered no survival benefit over systemic chemotherapy alone when the IV method was applied at the 1 year landmark. Subject to selection and survivor treatment bias, standard regression analysis may overestimate the benefit of PTR. Future study should focus on identifying patients most likely to benefit from PTR.

Download Full-text

PRS47 COMPARING RISK ADJUSTMENT MODELS: PROPENSITY SCORE MATCHING, STANDARD REGRESSION ANALYSIS AND INSTRUMENTAL VARIABLE METHOD

Value in Health ◽

10.1016/s1098-3015(10)73000-x ◽

2010 ◽

Vol 13 (3) ◽

pp. A204

Author(s):

O Baser ◽

A Dysinger ◽

E Baser

Keyword(s):

Regression Analysis ◽

Propensity Score ◽

Propensity Score Matching ◽

Risk Adjustment ◽

Instrumental Variable ◽

Variable Method ◽

Instrumental Variable Method ◽

Standard Regression

Download Full-text

PRS47 USE OF PROPENSITY SCORE MATCHING, STANDARD REGRESSION ANALYSIS AND INSTRUMENTAL VARIABLE METHOD IN OUTCOMES RESEARCH STUDIES: A COMPARATIVE ANALYSIS

Value in Health ◽

10.1016/s1098-3015(11)72279-3 ◽

2010 ◽

Vol 13 (7) ◽

pp. A327

Author(s):

O Baser ◽

A Dysinger ◽

E Baser ◽

H Yuce

Keyword(s):

Regression Analysis ◽

Comparative Analysis ◽

Propensity Score ◽

Propensity Score Matching ◽

Outcomes Research ◽

Instrumental Variable ◽

Variable Method ◽

Instrumental Variable Method ◽

Standard Regression ◽

Research Studies

Download Full-text

Financial Literacy as a Driver of Financial Inclusion in Kenya and Tanzania

Journal of Risk and Financial Management ◽

10.3390/jrfm14110561 ◽

2021 ◽

Vol 14 (11) ◽

pp. 561

Author(s):

Ashenafi Fanta ◽

Kingstone Mutsonziwa

Keyword(s):

Regression Analysis ◽

Financial Literacy ◽

Financial Services ◽

Instrumental Variable ◽

Financial Inclusion ◽

Demand Side ◽

Financial Products ◽

Instrumental Variable Regression ◽

Per Se ◽

Using Data

Efforts are being exerted in many developing countries to promote financial inclusion by increasing individuals’ access to financial products and services. However, literature suggests that increasing the supply of financial products and services per se may not help in expanding financial inclusion unless concerted efforts are exerted in enhancing financial literacy. This is because financially literate individuals are more likely to appreciate the value of financial services and hence take up financial products. This paper reports the link between financial literacy and inclusion using data from a demand side financial inclusion survey conducted in Kenya and Tanzania in 2016 covering a total of 6029 individuals. Results from our instrumental variable regression analysis confirmed that financial literacy is a strong driver of financial inclusion. This implies that efforts to promote financial inclusion need to be accompanied with financial literacy campaigns in both countries.

Download Full-text

On the Relationship between Length of Stay and Total Trip Expenditures: A Case Study of Instrumental Variable (IV) Regression Analysis

Tourism Economics ◽

10.5367/te.2014.0357 ◽

2015 ◽

Vol 21 (2) ◽

pp. 357-367 ◽

Cited By ~ 11

Author(s):

Christer Thrane

Keyword(s):

Regression Analysis ◽

Length Of Stay ◽

Instrumental Variable ◽

The Relationship

Download Full-text

Dexmedetomidine and Mortality From Sepsis Requiring Mechanical Ventilation: A Japanese Nationwide Retrospective Cohort Study

Journal of Intensive Care Medicine ◽

10.1177/0885066620942154 ◽

2020 ◽

pp. 088506662094215

Author(s):

Shotaro Aso ◽

Hiroki Matsui ◽

Kiyohide Fushimi ◽

Hideo Yasunaga

Keyword(s):

Mechanical Ventilation ◽

Logistic Regression ◽

Regression Analysis ◽

Propensity Score ◽

Logistic Regression Analysis ◽

Instrumental Variable ◽

Diagnosis Procedure Combination ◽

Instrumental Variable Analysis ◽

Mechanically Ventilated ◽

Variable Analysis

Background: Dexmedetomidine has a mild sedative effect and may reduce mortality in mechanically ventilated critically ill patients. However, few studies have examined the effects of dexmedetomidine in patients with sepsis who require mechanical ventilation. The aim of this study was to investigate the association between dexmedetomidine and mortality in patients with sepsis requiring mechanical ventilation, using a nationwide inpatient database in Japan. Methods: Using the Diagnosis Procedure Combination database from July 1, 2010, to March 31, 2016, we identified adult patients with sepsis who required mechanical ventilation for more than 2 days. Patients were divided into those who received dexmedetomidine and those who received midazolam or propofol within 1 day after admission. Logistic regression analysis, propensity score-matched analysis, and instrumental variable analysis were performed to compare all-cause 28-day mortality and duration of mechanical ventilation between the groups. Results: In total, 50 671 were eligible patients, including dexmedetomidine group (n = 13 759) and propofol or midazolam group (n = 36 912). The dexmedetomidine group had significantly lower all-cause 28-day mortality compared with the group receiving midazolam or propofol, as shown by the logistic regression analysis (odds ratio [OR]: 0.78; 95% confidence interval [CI]: 0.73-0.84), the propensity score–matched analysis (OR: 0.85; 95% CI: 0.80-0.91), and the instrumental variable analysis (OR: 0.64; 95% CI: 0.57-0.73). The duration of mechanical ventilation in the dexmedetomidine group was significantly shorter than that in the midazolam or propofol group. Conclusions: Dexmedetomidine was associated with a reduction in all-cause 28-day mortality and duration of mechanical ventilation.

Download Full-text