GBACC: mixed methods evaluation in 3 phases

1. Getting Better at Chronic Care (GBACC) in North Queensland: a cluster RCT of community health worker care co-ordination in remote FNQ settingsRobyn McDermott, Barbara Schmidt, CillaPreece, Vickie Owens, Sean Taylor, Adrian EstermanPHC RIS Conference, Canberra July 2014

2. Aim: To evaluate the impact of intensive case management for adults with poorly controlled diabetes and other complex conditions, by local Health Workers Design: Cluster RCT over 18 months and mixed methods evaluation over the subsequent 18 months, 2011-2015 Setting: 12 remote FNQ communities in 3 Health Service Districts and 213 participants with poorly controlled diabetes and significant comorbidities Inclusion criteria: Established diabetes in adults, HbA1c >= 8.5%, significant comorbidities Results of Phase 1 (RCT): Significant improvement in some care processes and HbA1c and LDL-C at 18 months. No change in smoking. Significant implementation failure in intervention sites.

3. GBACC: mixed methods evaluation in 3 phases

4. 12 Participating Communities*Intervention sites in phase 1 (randomly allocated) Torres and NPA HHS • Badu* • Bamaga • Injinoo* • New Mapoon • Seisia • Umagico* Cape York HHS • Kowanyama* • Mapoon* • Napranum Cairns and Hinterland HHS • Mareeba (Mulungu) • Mossman Gorge (ACYHC)* • Yarrabah (GYHS)

5. PHASE 1: CONSCONSORT DIAGRAM: GBACCRCT) Enrolment: 12 sites recruited and 327 patients assessed as eligible Baseline data collected, n=213 Excluded: 114 patients declined to participate Group randomisation: 12 sites Intervention: 6 sites (n=100 patients) Received intervention, n=100 Allocated to waitlist group: 6 sites (n=113 patients) Allocation Follow up Lost to follow-up (n=16) • Moved away (12) • Died (4) Lost to follow up (n=6) • Moved away (3) • Died (2) • Withdrew from study (1) Analysis Analysedfor primary outcome, n= 84 (84%) Analysed for primary outcome, n=108 (96%)

6. KEYFINDINGS AT BASELINE Demography of participants • Mean age of participants is 47.6 years • 2/3 of participants are women • After randomization there was no clinical difference between the intervention and control site participants

7. Baseline socio-demographic characteristics of study participants (SD or %)

8. BASELINE CHARACTERISTICSSelf reported - Quality of Life Self reported quality of life (AQoL) at baselinef Life

9. Comparison of Indigenous and non-Indigenous populations with poorly controlled diabetes

10. Comparison of Indigenous and non-Indigenous populations with poorly controlled diabetes:Lipid profiles

11. Clinical care processes at baseline and follow up (%)

12. HbA1c measures at baseline and follow-up by group, absolute values

13. Clinical measures: 212 Indigenous adults with poorly controlled T2DM, baseline and follow-up at 18 months

14. Did it work? (care processes) Care processes (n and % with GPMP, TCA done) at T3 for intervention vs control sites in GBACC among all the participants (N=213) Intervention sites were slightly (26%) more likely to have implemented a GPMP at T3, however this did not reach statistical significance (Odds Ratio = 1.26, 95% confidence interval 0.72-2.22):

15. Implementation fidelity evaluation (measuring type 3 error) during Phase 1 • Major restructure in QH from 2012 • Recruitment freeze • Lack of doctor engagement in care planning • Lack of inclusion of HWs as team users of Patient Information Systems • Since project commencement in the Torres/NPA in 2011 there have been 7 different CEO/COOs, finally resolved with an amalgamation with CYHHS and a new CEO, who is yet to commence (due August 2014) • No pharmacists to perform HMRs • Slow transition to Community Control in Cape York sites, with GBACC HWs unable to access Patient Information Systems • GBACC HWs frequently used for general clinical roles in the absence of regular staff

16. Conclusions and lessons learned • Even though the primary outcome was achieved (1% reduction in HbA1c), significant implementation failure in the 6 intervention sites was evident • The addition of a HW to the team despite good training and skills and cultural competence, in the absence of overarching system change resulted in less than optimum clinical outcomes and low cost-effectiveness • Pragmatic trials of complex interventions are always subject to unpredictable major impacts of the wider policy environment eg change of government and sweeping organisational change, which can overwhelm the trial effect through unintended implementation disruption • Highlights the need for detailed implementation fidelity monitoring as part of the evaluation plan

17. Acknowledgements The GBACC project is funded by NHMRC Partnership project grant number 570149 with the Queensland Government Barbara Schmidt is supported by the APHCRI CRE in Chronic Disease Prevention and Care in Rural and Remote Populations The research reported in this presentation is partly a project of the Australian Primary Health Care Research Institute, which is supported by a grant from the Australian Government Department of Health, under the Primary Health Care Research, Evaluation and Development Strategy. The information and opinions contained in it do not necessarily reflect the views or policy of the Australian Primary Health Care Research Institute or the Australian Government Department of Health.