Table 3. Characteristics of included studies potentially suitable for meta-analysis (non-universal health care systems).

Paper	Country of Study	Data Source (s)	Population Included	Years of Diagnosis	Measure of SEP	No of SEP Groups	Treatment Given Within	Age Range	Confounders Controlled For:					Quality Score
									Age	Sex	Stage	Histology	Other
Bradley et al, 2008 [57]	USA	Michigan Cancer Registry and Michigan Medicare and Medicaid data	Medicare and Medicare/Medicaid patients in Michigan	1997–2000	Census tract median household income (high v low)	2	6 months	66–80+	Yes	Yes	Yes	Yes	Co-morbidity, insurance type, ethnicity, urban/rural	4
Davidoff et al, 2010 [58]	USA	SEER cancer registry linked to Medicare data	Medicare patients from 16 SEER registries	1997–2002	Census tract median household income	4	90 days	66–85+	Yes	Yes	Yes	Yes	Co-morbidity, performance status, ethnicity, marital status, rural/urban, prior Medicaid, tumour grade	5
Earle et al, 2000 [59]	USA	SEER cancer registry linked to Medicare data	Medicare patients from 11 SEER registries	1991–1993	Census tract median household income(increase in OR per quintile)	5	4 months	65–104	Yes	Yes	Yes	Yes	Co-morbidity, year of diagnosis, ethnicity, rural/urban, teaching hospital, SEER area	5
Esnoala et al, 2008 [60]	USA	South Carolina central cancer Registry linked to inpatient and outpatient surgery files	South Carolina	1996–2002	Income, zip code level (poverty/not living in poverty)	2	NR	<50–80+	Yes	Yes	Yes	Yes	Co-morbidity, year of diagnosis, insurance type, ethnicity, rural/urban, education, marital status, tumour location	4
Greenwald et al, 1998 [61]	USA	SEER cancer registry	3 (Detroit, San Francisco, Seattle) out of 9 SEER registries	1978–1982	Census tract median household income (increase in OR per decile)	10	NR	< = 75	Yes	Yes	Yes	Yes	Performance status, ethnicity	6
Hardy et al, 2009 [62]	USA	SEER cancer registry linked to Medicare data	Medicare patients from 17 SEER registries	1991–2002	% individuals below poverty line at census tract level	4	NR	65–85+	Yes	Yes	Yes	Yes	Co-morbidity, year of diagnosis, ethnicity, marital status, SEER area, other treatment	5
Hayman et al, 2007 [63]	USA	SEER cancer registry linked to Medicare data	Medicare patients from 11 SEER registries	1991–1996	Census tract median household income	5	4 months/2 years	65–85+	Yes	Yes	Yes	Yes	Co-morbidity, year of diagnosis, ethnicity, SEER area, hospitalisation, teaching hospital, distance to nearest RT centre, receipt of chemotherapy	5
Lathan et al, 2008 [64]	USA	SEER cancer registry linked to Medicare data	Medicare patients from 11 SEER registries	1991–1999	Census tract median household income (inc in OR per quintile)	5	NR	65+	Yes	Yes	Yes	Yes	Co-morbidity, ethnicity, SEER registry, urban, non-profit hospital, patient volume, % of black patients in hospital	5
Polednak, 2001 [65]	USA	Connecticut Tumor Registry (SEER) and inpatient hospital discharge database (HDD)	Connecticut	1992–1997	Census tract poverty rate	5	NR	<55–80+	Yes	Yes	Yes	No	Co-morbidity, ethnicity, marital status	4
Smith et al, 1995 [66]	USA	Virginia Cancer Registry and Medicare claims database	Medicare patients from Virginia cancer registry	1985–1989	Census tract: median household income by race and age	Contin-uousa	6 months	65–85+	Yes	Yes	Yes	Yes	Co-morbidity, ethnicity, county of residence, distance to oncologist	5

Quality scores range from 1 (lowest quality) to 6 (highest quality).

^aOdds ratio for increase per $10,000 income.

CI, confidence interval; non-UHCS, non-universal health care system; NR, not reported; OR, odds ratio; SEER, National Cancer Institute's Surveillance, Epidemiology and End Results database; SEP, socioeconomic position.