To the Editor:

• “... what a difference can be made and how bureaucracies can stuff things up”.

• “... systematic testing and treatment of people with high blood pressure and kidney disease dramatically improved blood pressure and resulted in a 50% reduction of deaths”.

• “... excellent results were achieved by good management and they were lost when intensity of management was relaxed”.

The above quotes are from an episode of The health report broadcast late last year on Radio National.1 The episode, which described a deterioration in the health of an Indigenous community after a chronic disease treatment program was handed over to a community health board, caused me to take a closer look at the articles in the Journal by Hoy and colleagues on which the claims were based.2,3 I found several issues of concern.

The small numbers of deaths each year in the study community and the analysis and presentation of the death data mean that the conclusions about trends in mortality over time are tenuous. This is highlighted by the discrepancies between the two articles in the terminology used to classify deaths, in the numbers of deaths reported, and in the trends over time. Discrepancies in terminology or numbers of reported deaths are not explained. The declining trend in the number of “natural” deaths described in the 2000 article is not apparent in the “non-renal” deaths in the 2005 article. The rate of “non-renal” death for the period 1996–97 to 1998–99 reported in the 2005 article appears to be increasing rather than declining, as described in the 2000 article (rates for earlier years are not presented in either article). It is clear that, with these small numbers, the reclassification or misclassification of a single death can affect the trends in “renal death” or end-stage renal disease over time, and that the use of “rolling averages” hides the year-to-year variability that would be expected in these data.

The trend over time in the key intermediate outcome indicator of blood pressure control does not support the conclusion regarding impact of the “handover” on the program. The data presented in the 2005 article show a decline in control commencing in the third year. An earlier analysis of the same data showed the decline in blood pressure control began as early as the second year after entry into the program.4 Neither analysis shows any clear change in the declining trend in blood pressure control around the time of “handover” of the program.

While the discussion of the findings of the 2005 article is circumspect, at the time of interview, Hoy conspicuously did not deny the statement of The health report host that the primary cause of the apparent loss of the early impact of the program was the bureaucracy “stuffing up”. The article makes some important points about the operation of chronic disease programs, but makes no mention of the commonly experienced difficulties of sustaining health programs,5,6 or the research requirements for understanding sustainability.7

These issues raise serious questions about the validity of the conclusions and the simplistic claims arising from the articles.

In reply: I appreciate the feedback on the 2000 and 2005 articles describing the dynamics and outcomes of the “Tiwi treatment program”.1,2

Thorough and timely identification and enumeration of deaths is a problem, especially for people not enrolled in the treatment program. Without a register of such people, systematic checking of their fate was not possible. The additional “non-renal” deaths in the community-at-large presented in our 2005 article, compared with previous articles, seem to have been captured largely by the broad net spread by the Tiwi Health Board when it assumed responsibility for its primary care services, in an attempt to identify all its potential clients. This process identified several hundred more people than expected and captured additional deaths, several dating back years. The precise definition of a community member is also a problem, especially for people living permanently or intermittently elsewhere (eg, in Darwin or other communities).

The broadened definition of “renal deaths” in the 2005 article,2 which accommodates people who died with renal failure but did not begin dialysis, more fully represents the impact of renal disease. Conversely, recording only those who began dialysis allows estimates of the impact on health services and potential savings from better management.3 Both approaches have their place. Rolling averages, which indeed have limits, were used in view of the overall small and erratically spaced number of terminal events in any year.

The figures we reported in our 2005 article did not show a deterioration in blood pressure at Year 2, either in the treatment group as a whole, or in the smaller cohort followed for a full 6 years.2 An earlier analysis, which largely embraced the active years of the program, also showed that blood pressure at Year 3 was not significantly different from that at Year 2 (systolic blood pressure, P = 0.68) (Box). With time, the number of people who had moved through 3 years of treatment increased, and the timing of their 3-year blood pressure measurements moved from a mix of 1998–1999 to 1999–2002, when, as program dynamics suggest, intensity of management was relaxed, and mean values deteriorated, as we reported in 2005.

The blood pressure measurements in the report by Bailie’s group5 were compiled from a review of paper-based medical records, the clinic’s newly implemented Coordinated Care Trial Information System, and the Territory’s Information System (Systematic Health Information Logically Organised), as well our from our treatment program database. Those blood pressures were allocated time definitions in a different way, and the summary data were derived from adjusted predictions from cross-sectional time series modelling, rather than from factual recordings at the stated intervals.5

I did not solicit the interview for The health report, nor determine its directions nor the resulting headlines. However, the under-resourcing of primary care relative to needs in remote Aboriginal settings, and the lack of stability in the organisations in which it is delivered, are very detrimental. I regret that, once the Tiwi Health Board was constituted, it was not mentored and supported through its difficulties. More recently, the fledgling community-controlled Gulf Health Service in the Borroloola region of the Northern Territory met a similar fate. Chronic disease remains underserviced in both these regions, where the people are among the sickest in Australia.