Preliminary results from frequency tables and national quality reports Agenda point 2.2 Guillaume Blanc Lucian Agafitei ESTAT F-4 11-12 May 2016 EHIS Task Force
Outline Unit non-response Mode of data collection Interview duration Sample size Use of proxy respondents Item non-response Problematic variables Preliminary results from 20 national QRs Preliminary results from frequency tables of 23 Member States 11-12 May 2016 EHIS Task Force
Unit non-response Around 50-60% in AT, BE, FI, FR (and LT?) 10% or less in CY, RO, SK Subgroups with higher non-response: Elderly (AT) Men (FI) Young population (FI, CZ) Illiterate (FR) Seriously sick (FR) Non-French speakers (FR) 11-12 May 2016 EHIS Task Force
Mode of data collection Face-to-face interviews Telephone interviews Self-administered questionnaire Use of administrative data BE ✓ - BG CZ EL ES FR(1) HR CY LV(2) LT LU NL(3) AT PL PT(3) RO SE(4) SK FI 11-12 May 2016 EHIS Task Force
Interview duration Face to face interview: 20' 45' Telephone interview: 20' 24' Self-administered questionnaire: 20' 41' Reporting issues: Not always indicated in the quality report Not always presented separately by data collection mode Not always clear if only EHIS part included 11-12 May 2016 EHIS Task Force
Minimum effective sample size MS Achieved sample size (1) Effective sample size (2) Minimum effective sample size (3) Ratio (2) / (3) (4) BE 9,951 4,692 6,500 0.72 BG 6,410 5,008 5,920 0.85 CZ 6,737 6,478 6,510 1.00 EL 8,223 4,691 6,667 0.70 ES 22,842 : 11,620 FR 15,756 11,847 13,110 0.90 HR 5,446 CY 4,958 4,948 4,095 1.21 LV 7,077 9,870 4,555 2.17 LT 5,205 6,426 4,850 1.32 LU 4,004 393? 4,000 0.10? HU NL 7,653 7,289 7,515 0.97 AT 15,771 10,729 6,050 1.77 PL 24,156 20,824 10,690 1.95 PT 18,204 6,515 RO 16,605 8,420 SE 6,200 SK 5,490 5,719 5,370 1.06 FI 6,226 5,330 Out of 12 MS (LU excluded), 5 seem not to comply with EHIS wave 2 regulation 11-12 May 2016 EHIS Task Force
Use of proxies for specific variables Recommendation for not using a proxy for some variables not followed by 6 countries (in 3 countries allowed for all variables) No proxies allowed in 5 countries Proportion of proxy: Less than 10 %: 11 MS More than 10 %: 3 MS (maximum: 36%) 11-12 May 2016 EHIS Task Force
Impact of using proxies Example: Variable PC3 "Need to receive help or more help with self-care activities Premise: 29% of the answers PROXY = 2 or 3 Result : % of Yes (weighted results) PROXY = 1 or 2 or 3 PROXY = 1 PROXY = 2 or 3 54 % 48 % 66 % 11-12 May 2016 EHIS Task Force
Item non-response Huge differences between countries and variables Examples: PL2 (difficulty in seeing): 38% (EE) PL4 (difficulty in hearing): 97% (EE) PN2 (pain interference with work): 45-50% (CZ, EE) PA1 (last time of vaccination against flu): 70-75% (BE, FI), 96% (EE) UN1A (unmet need for health care in the past 12 months due to long waiting list): 49% (FR) HHINCOME: more than 10% in 8 countries (max: 36% in HR) Would be good to give an idea on overall item-non-response 11-12 May 2016 EHIS Task Force