Assessing the impact of a recall system on hospitalisations for patients with diabetes in the Torres Strait, 1998-2000: A retrospective re-evaluation study of a randomised control trial using a different data source

Background: A randomised control trial (RCT) was conducted in the Torres Strait and Northern Peninsula Area (NPA) Health Service District (HSD) during the period 1998-2000. The main intervention strategy for this trial was the introduction of a simple recall/patient reminder system in Primary Health Care Centres (PHCCs) in an attempt to improve the organisation and management of diabetes care. After a twelve-month period, there was a significant reduction (40 per cent) in the number of patients admitted in the intervention sites between baseline and followup. However, as hospitalisation data was derived from the PHC clinic records, it has been suggested that this methodology may have underestimated both the number of patients and hospital episodes captured in the RCT study population.

Aim: To conduct a retrospective re-evaluation study of the hospitalisation component of the original RCT in an attempt to better assess the impact of the clinic recall system for patients with diabetes in the Torres Strait and NPA, 1998-2000. All patients from the RCT were to be tracked manually through three local public hospitals for hospitalisation episodes in a bid to validate the total number of admissions for these Torres Strait Islanders (TSIs) with diabetes.

Methods and Subjects: Subjects included all those patients suffering with diabetes living in the Tones Strait and NPA whose PHC records had been audited during the original RCT. Instead of using PHC clinic patient files, hospital -based patient information systems were used (as a different data source) in order to validate the number of admissions for each individual from the trial. Analysis of the data used the same methodology as the RCT (that is, communities were clustered into intervention and control sites) in order to interpret results in terms of whether or not the implementation of a recall system in the intervention sites had any impact on the hospitalisation rates of patients with diabetes in the Torres Strait.

Results: In comparison with the RCT, the new hospitalisation study found some degree of underestimation in the number of patients identified as having hospitalisations and secondly, considerable underestimation in terms of the number of hospital episodes found. At baseline, the hospital -checking methodologyidentified one extra person and 102 additional episodes of care. At followup, this new methodology found an extra 30 persons and an additional 119 episodes of care. In comparison to the RCT, the new study showed an overall 6.7 per cent increase in the number of individual patients detected having had an admission in the study period. However, there was a large 47.1 per cent increase in extra hospital episodes detected across both audit timeframes.

Discussion: This study supports the main findings of the original RCT. That is, the new study also found a reduction in the proportion of patients hospitalised for any kind of `diabetes -related condition' in the intervention sites between baseline and followup. This reduction was slightly less at 29 per cent in the intervention sites (compared to 40 per cent in RCT across intervention sites) for 'total diabetes -related conditions,' however this new study finding was not statistically significant. The re-evaluation study also found that at followup, those in the intervention sites were 19 per cent significantly less likely to be hospitalised for a diabetes -related condition than those in the control sites (compared with a significant 40 per cent in the RCT). Therefore, admissions for diabetes -related conditions can be reduced or avoided through better organisation and a systematic approach to the management of PHCCs. Although hospital -based information systems may have been more sensitive in detecting patients with hospitalisation episodes and additional episodes for known patients, the tracking of individuals through the public hospital system (without a unique identifier) was problematic and time consuming.

Conclusion: The main aim of the RCT was to improve the organisation and management of PHC in the Tones Strait, in terms of clinical management for the large number of diabetic clients that are served. The original RCT findings suggested that improvements in diabetes care in the intervention sites such as the introduction of a patient recall system was successful in reducing the numbers of patients subsequently admitted to hospital. The re-evaluation study of hospitalisations using a different methodology has also been successful in recording a reduction in the number of people with diabetes being hospitalised from the intervention sites despite a 47 per cent increase in the number of hospital episodes being detected overall.