• Open Access

A survey of suppression of public health information by Australian governments

Authors


Correspondence to:
Professor C. D'Arcy J. Holman, School of Population Health, The University of Western Australia, Crawley, Western Australia 6009. Fax: (08) 6488 1188; e-mail: dholman@meddent.uwa.edu.au

Abstract

Background: It is cause for concern when a democratically elected government suppresses embarrassing information by hindering public health research or the publication of research findings. We conducted a survey of Australian public health academics to estimate the level of acts of suppression of research by Australian governments, to characterise these events, and to gather views on what interventions might be effective in curbing them.

Methods: A total of 302 academics in 17 institutions completed a postal questionnaire in August 2006 (46% of 652 invited). The instrument sought details of suppression events they had witnessed since 2001.

Results: There were 142 suppression events, including 85 personally experienced by 21.2% (n=64) of respondents. The rates were higher in 2005/06 than in earlier years. No State or Territory was immune from suppression. Although governments most commonly hindered research by sanitising, delaying or prohibiting publications (66% of events), no part of the research process was unaffected. Researchers commonly believed their work was targeted because it drew attention to failings in health services (48%), the health status of a vulnerable group (26%), or pointed to a harm in the environment (11%). The government agency seeking to suppress the health information mostly succeeded (87%) and, consequently, the public was left uninformed or given a false impression. Respondents identified a full range of participative, cognitive, structural and legislative control strategies.

Conclusion: The suppression of public health information is widely practised by Australian governments.

Implications: Systemic interventions are necessary to preserve the integrity of public health research conducted with government involvement.

At the time of the Chernobyl disaster in 1986, the Soviet Government at first kept secret information about the accident, including the impact on health and the number of people affected.1 It then relaxed national radiation protection standards in an effort to whitewash the magnitude of casualty projections and to discharge more people from surveillance programs. This was an example of information suppression, which may be defined as “an active process to prevent data from being created, made available, or given suitable recognition”.2

Many industries suppress information to protect their commercial interests. But when information of legitimate public interest is suppressed at the hands of a democratically elected government, it should be cause for concern. In a democracy, the citizenry has a constitutional right to participate in the appointment of those who govern and to hold the government accountable for its performance. Thus any act of suppression of information that reflects positively or negatively on the performance of government policies and programs is detrimental to the attainment of responsible government. There are now warning signs that Australia may be slipping from its former enviable position of relative freedom from political censorship and official corruption. It has fallen from among the top 10 in international press freedom indices to being outside the top 40.3

We were aware of instances of individual public health researchers whose work had been suppressed, yet beyond these anecdotes very little knowledge existed about the topic in Australia. We therefore set about conducting a national survey of public health academics with the aims of:

  • 1Estimating the overall level, trends and distribution of acts of suppression of public health information by Australian governments.
  • 2Characterising these events with respect to their nature and outcomes.
  • 3Gathering the views of respondents about what interventions would be effective in alleviating suppression by governments of the work of public health researchers.

Methods

Study population

The study population consisted of academics listed online in staff directories of the member institutions of the Australian Network of Academic Public Health Institutions (ANAPHI), which covers most of the significant university-based public health research groups in Australia.4 Of the 19 ANAPHI institutions, one in New South Wales and one in Queensland did not have staff directories online. Eligible participants were academic staff classified at Australian levels B-E (i.e. research fellow or lecturer and above) as it was thought that most academics at these levels would have been involved in research and the publication of their findings. Those with adjunct titles were excluded, whereas those with position descriptors such as ‘director’ or ‘head of school’ were included. In one instance of a large, multi-divisional ANAPHI institution, staff members in schools of nursing and psychology were excluded, whereas those in the schools of exercise and nutrition were included. A total of 679 eligible staff was identified.

Survey instrument

We designed a single-sheet questionnaire following face validation and feedback from piloting an initial draft among academics at our own institution. The instrument prominently contained the following definition:

A suppression event is any circumstance where potentially embarrassing or controversial health information of legitimate public interest was intended to be researched or published, but:

– was withheld or declined funding;

– faced obstruction, including abnormal delays, in being pursued or published; or

– was the object of a request or directive to be modified or sanitised

by an Australian government agency in a manner protecting the interests of government or an identifiable group.

This definition was objective rather subjective with respect to the motivations of governments. It did not require the respondent to determine if the government agency had intended to protect itself or another group, but merely that the manner of the event was such that it had that effect.

The instrument sought information from respondents concerning suppression events affecting their own research and experienced in the last 5.5 years (2001 to mid-2006). It separately enumerated suppression events that respondents had observed affecting other investigators. For each event, it collected data on the year of occurrence; nature of the information suppressed; suspected reasons; whose interests were protected; outcome of suppression; and the State or Territory in which the event occurred. It asked for the respondent's gender, age group, field of expertise and years of experience. Finally, it sought their views on whether suppression in Australia was increasing, decreasing or constant, and what, if anything, should be done to reduce its occurrence.

Participant recruitment and response

The study was approved by The University of Western Australia Human Research Ethics Committee. A letter was sent to eligible participants explaining the study and enclosing a postal version of the instrument and a reply-paid envelope. It was emphasised that participation was voluntary and, given the sensitive nature of the study topic, care was taken to ensure that returns were anonymous. Our commitment to preserving the anonymity of respondents reduced the ability to break down response statistics and results accurately by jurisdiction.

As completed questionnaires came in, they were numbered and dated. We used the dates returned to determine if early respondents differed from late respondents in their likelihood of reporting a suppression event. A reminder letter was sent at three weeks and the total duration of the fieldwork was six weeks.

Of the initial sample of 679 academics, 27 were excluded due to an invalid address or occupational designation and two others did not complete the questionnaire to a useable extent. The final response fraction among those eligible to participate was 46.3% (302/652).

Statistical analysis

The data were coded and comments from respondents were compiled in a text file. An effort was made to identify same events reported by multiple respondents by comparing event characteristics. We calculated the proportion of respondents who experienced at least one event during the last 5.5 years and the incidence rate of total events experienced per person-year during the same period. One event already reported by another respondent was excluded from the incidence rate calculations, but all respondents experiencing them were included in the calculation of the proportion affected. Events were subdivided according to their characteristics and the results were also stratified by characteristics of the academics, including whether they had been early or late responders. Views on suppression trends and possible interventions were subjected to simple descriptive analyses.

An approximate indication of the comparative level of suppression experienced in each Australian State and Territory was calculated as the ratio of events reported in 2001 to mid-2006 divided by the number of public health academics in surveyed ANAPHI institutions in each State and Territory in mid-2006. The validity of the calculation assumed similar growth in the academic communities since 2001 and an even response fraction across the nation. Neither assumption could be validated.

Results

Characteristics of respondents

There were slightly more females (53%) than males. Just over one-half (53%) of respondents were aged 30-49 years, while only 9% were aged less than 30 years. The six most common fields of academic expertise were described as public health (25%), epidemiology (18%), biostatistics (5%), medicine, health promotion and nutrition (each 4%). More than one-third (36%) had up to 10 years of professional experience, a similar proportion (37%) had 11-20 years and the remaining 26% had more than 20 years of experience in their field.

Incidence and profile of suppression events

There were 142 suppression events, of which 85 had been personally experienced by 64 of the 302 respondents during an estimated 1,672 person-years. Thus 21.2% of respondents had personally experienced a suppression event since January 2001 (95% confidence interval 16.6-25.8%) and the average incidence rate was 5.1 (3.9-6.1) events per 100 person-years. A further 24 respondents had observed events affecting other researchers. No respondent reported more than three events. Figure 1 shows the trends in the incidence rates of experienced and observed events from 2001 to mid-2006. The rate of personally experienced events in 2006 was fourfold higher than that based on recall of events in 2001 (RR 4.0 [1.9-8.3]).

Figure 1.

Trends in incidence rates of experienced and observed suppression events in survey respondents from 2001 to mid-2006.

Table 1 shows the profile of experienced and observed suppression events. Many respondents had exercised caution and declined to disclose information in response to some or all of the questions. From what could be gleaned from the responses it appeared that the majority of suppressed information concerned the performance of health services, the health status of a vulnerable population group or a harmful exposure in the environment. Methods of suppression covered the full gamut of the research process from funding to publication. The most common methods were sanitisation of content, delays in release or complete prohibition of research reports. Problems with data access, regulatory approvals and efforts to control the performance of the research, while less frequent, were still well represented. In a small number of instances researchers reported being intimated by threats or accusations, and in one case the suppressor had withdrawn financial support for the respondent's academic position.

Table 1.  Distribution of total experienced and observed suppression events according to event characteristics.
 Number of events%
  1. Notes:

  2. (a) Multiple response item with percentage sum exceeding 100%.

Type of information suppressed  
 Delivery of health services3948
 Population health status2126
 Environmental exposure911
 Health administration911
 Community views22
 Research methods22
 Not disclosed = 60  
Method of suppression  
 Block funding1210
 Refuse regulatory (ethics) approval22
 Deny data access1311
 Impose methods of research22
 Control what is to be reported43
 Sanitise report2823
 Delay publication1613
 Block publication3630
 Threats or accusations65
 Withdraw support of position11
 Not disclosed = 22  
Reason for suppression suspected by researcher  
 Avoidance of unfavourable results5593
 Cost implications of results47
 Not disclosed = 83  
Interests protected by suppressiona  
 Government11581
 Health industry and provider groups4129
 Community groups118
Outcome of suppression  
 Public was uninformed3235
 False impression was created2224
 Research was inadequate1718
 Results were not acted on78
 Researcher resigned22
 Still under negotiation78
 Suppression failed55
 Not disclosed = 50  

Although more than one-half of respondents declined to comment on the underlying reason for suppression, the available responses made clear that avoidance of unfavourable results, whether real or anticipated, was a common occurrence (see Table 1). The following quotations are indicative of the responses: “data not to be released until after elections”; “data challenged routine practice”; “perception by bureaucrat that recommendations would be controversial”; “reason is to avoid ministerial embarrassment, given they have contracted the work”; “would expose poor practice in a rural area”. Cost implications of the research findings were also cited in a few instances. The instrument included a closed multiple-response item on whose interests were protected by the particular government's act of suppression. In four out of five instances it was the interests of the government itself that were protected. The next most common interests protected were those of health industry and provider groups, while the interests of community groups accounted for a small proportion.

According to those responding, the act of suppression was successful in damaging the research communication process on six out of seven occasions (see Table 1). The following were illustrative responses: “information on long and short term risks not available to the public”; “paper toned down – loss of real information that reflect the true state of affairs – feel like I'm hiding the truth”; “no data collected regarding patient safety in … for 10 years”. The remainder of incidents were either still under negotiation or the attempt at suppression was unsuccessful because the results were leaked to the public or published through alternative channels.

Characteristics of researchers subjected to suppression

Table 2 shows the proportion of respondents who experienced suppression according to different researcher characteristics. Those affected were more likely to be males aged 30+ years with 11-20 years of professional experience. Public health academics with less than 10 years or more than 20 years of professional experience were less likely to be involved. There was a tendency for those who responded late (after the reminder) to report a lower incidence of suppression events than the early responders, albeit this difference may have been a chance finding.

Table 2.  Proportion of respondents experiencing suppression events according to respondent characteristics.
 Number of academicsaNumber with eventCumulate incidence
   %ratio(95% CI)
  1. Notes:

  2. (a) Data were missing on some academic respondents.

Early or late responder    
 Early (1st three weeks)21851231.00
 Late (2nd three weeks)7113150.66 (0.38-1.15)
Gender    
 Female15827171.00
 Male14037261.53 (0.98-2.36)
Age group (years)    
 <30263121.00
 30-4915934211.85 (0.61-5.60)
 50+11426231.98 (0.65-6.04)
Years of experience    
 ≤1010916151.00
 11-2011132291.96 (1.15-3.37)
 >207815191.31 (0.69-2.49)

Suppression events by location

In Figure 2 we show the ratios of experienced suppression events to the number of survey invitees in each Australian State and Territory. These results should be interpreted cautiously. Tasmania did not host an ANAPHI institution at the time of the survey and we remind the reader that one NSW and one Queensland institution were excluded. Respondents assigned events to locations on the basis of where they were working at the time the events occurred. We also emphasise that ‘location’ does not equate with a particular State or Territory government, as the government agency involved may have been the Federal Government or the government of another State jurisdiction.

Figure 2.

Comparison of ratios of experienced suppression events to survey invitees according to location.

Bearing in mind these limitations, the ratio for public health academics located in the Northern Territory at the time of suppression appeared to be inordinately high, amounting to more than fivefold higher than the ratio in Victoria (RR=5.2 [2.1-13.0]). We noted, however, that only six events were experienced in the Northern Territory and that there were larger absolute numbers of events in every Australian State on which data were available.

Researchers' views on suppression

The majority of respondents (68%) were unsure whether suppression events were subject to a trend, whereas 24% thought that suppression was increasing (of which 64% had witnessed an event compared with 29% of the entire sample), 7% thought the rate was constant and 1% believed it was decreasing. When asked what, if anything, should be done to reduce suppression events, the most frequent suggestions were structural or statutory along the lines of promulgation of public policy or regulations to prevent or limit acts of suppression. Lesser numbers favoured communitarian or cognitive interventions involving advocacy or training programs within government. Table 3 lists the frequencies with which different forms of intervention were recommended by respondents.

Table 3.  Frequencies of different interventions to reduce suppression of health information recommended by respondents.
 No. of respondents
Structural and statutory interventions 
 Reforms to government structures or processes29
 Legislation to prohibit suppression28
 Neutral mediating body26
 Increased independence of academic institutions16
 Research register7
 Monitoring by ethics committees4
Communitarian and cognitive interventions 
 Public debate28
 Training programs within government8
 Specific advocacy for researchers7

The following were examples of text responses concerning structural/statutory interventions: “organisation should be made to explicitly state their rationale for changes in writing to improve transparency”; “ensure contracts with government do not include clauses about needing permission prior to releasing information”; “any project funded within or outside government should be registered – register must be public”; “independent mediators involved”; “have a body that can be approached with complaints that has authority to investigate”; “introduce an ombudsman”; “charter on research independence to be signed off by premier/PM/heads of government agencies”; “change performance management of senior bureaucrats”; “automatic dismissal of any CEO responsible for suppression”; “make it illegal under FOI law for government and other agencies who fund public health research to suppress research findings”; “legislate against it at all levels”.

Communitarian/cognitive interventions were illustrated by the following responses: “raise the issue and the downside of this to the public – make them aware”; “constant media presence on the issue”; “build norms to make it unacceptable”; “get deans and vice-chancellors to take a united stand”; “make governments aware of the problem and how it is harmful to the public as well as to the functioning of government as a whole”; “organisational training in transparency/anti-corruption within government organisations”; “reduce climate of fear within government departments”; “establish a supportive organisational culture of telling it the way it is”.

Discussion

A former Commissioner of Health in Western Australia once described public health as “the conscience of the health system”. These survey results show that the conscience is not always permitted to be heard. Just under one-third of the academics had witnessed the suppression of health information by Australian governments in the last 5.5 years and more than one-fifth had experienced such events personally. The events occurred at higher rates in 2005/06 than in earlier years. No State or Territory in which the survey took place was immune. Although governments most commonly hindered research by sanitising, delaying or prohibiting the publication of findings, there was no part of the research process beyond their reach. Most of the affected researchers believed that their work had been targeted because it drew attention to failings in health service delivery, uncovered bad news about the health of a vulnerable group, or pointed to a harmful exposure in the environment. In most instances, the government agency seeking to suppress the health information succeeded and, consequently, the public was left uninformed or given a false impression.

Our results were affected by several sources of bias, but these do not detract from their significance or implications. Only 47% responded to the survey and the higher level of events reported by early responders was consistent with the possibility that those who participated were motivated because they had witnessed suppression events. Thus the risk may have been inflated compared with the experience of the average public health academic. This potential for over-estimation may have been offset by recall bias. Respondents recalled relatively more suppression events from the last 1.5 years than from the four years previously. Whether this was caused by differential recall, a real trend towards a higher rate in 2005/06, or a combination of the two, we cannot say. The research was hampered by its political sensitivity and the reluctance of many respondents to provide details that might identify the suppressor. Therefore, the event profile may have been incomplete due to missing information, although it seems unlikely that information kept secret by some respondents would have been less damning than what was revealed.

The findings of this research were based solely on the perceptions of the researchers who responded to the survey. It is possible that the officers working in government agencies who were involved as the other parties would not perceive that their actions constituted ‘suppression events’ according to our definition. The potential for different perceptions may have been even greater in instances where respondents observed suppression events affecting other researchers. It is also possible, at least in theory, that a suppression event could have occurred for honourable reasons, but with the unintentional side-effect that embarrassing or controversial health information of legitimate public interest was withheld in a manner protecting the interests of government or an identifiable group.

A strength of the research was that events were characterised by the location where the respondent was working when the event occurred and not necessarily their present working location. However, to preserve anonymity the only way these data could be related to denominators of researcher populations ‘at risk’ in each State or Territory was to rely on the numbers of academics invited to participate. This was hardly an ideal measure of the spatial epidemiology of suppression events, but even given the gross limitations the occurrence of suppression events in the Northern Territory appeared to be high relative to the number of researchers who had worked there. It may be relevant that five of the six suppression events occurring in the Northern Territory were concerned with the health status of Indigenous Australians or refugees.

This is thought to be the first study conducted with the primary aim of characterising the frequency and nature of official suppression of information in the Australian health sector. What evidence of the phenomenon exists internationally has been mostly limited to the exposéof individual cases, including the Black report in the United Kingdom,5 the Chernobyl disaster in the Soviet Union,1 the SARS outbreak in China,6 and harmful mercury blood levels in the United States.7 Our findings confirm that the practice is widespread in Australia and is not a problem restricted to isolated aberrations on the part of atypical government officials. Rather, the problem is endemic on a system-wide basis.

Suppression in Australia is not limited to the health sector. A survey of 70 Australian environmental scientists to gauge their input into the formulation of environmental policy found that 11% had been disadvantaged by speaking out publicly and 36% were aware of others who had been disadvantaged.8 In their book Silencing Dissent, Hamilton and Maddison recount a number of cases involving suppression events affecting universities and the research community, especially in the environmental sciences and humanities.9

The design of Australia's system of government has never observed a strict separation of powers between the parliament and the executive. Historically, permanent Public Service heads who reported to ministers, themselves members of a parliament, acted with independence in the general administration of their departments. As permanency in the Public Service has been replaced by short-term contracts, senior executives have come to regard “the politicians as their employers whose needs must be satisfied rather than the Australian nation who really pays them”.10 The increased pressure for senior executives to manage public perceptions through manipulation and, if necessary, suppression of information unfavourable to the government's re-election prospects may be partly at the root of the system-wide failure of democratic principles that this survey has uncovered.

Although democracy and responsible government are held by the High Court to be enshrined in the Australian Commonwealth Constitution,11 it is questionable if a remedy for the suppression of health information by governments is to be found in constitutional law. The Constitution affords a guarantee of freedom of political communication, implied by the entire system of responsible government created by provisions for the Senate, House of Representative, ministers, altering the Constitution and by other sections.11 However, in the cases of Lange vs. ABC and Levy vs. Victoria,12,13 the High Court read down Australian free speech jurisprudence. In the words of Justice McHugh: “The freedom protected by the Constitution is not, however, a freedom to communicate. It is a freedom from laws that effectively prevent the members of the Australian community from communicating with each other about political and government matters relevant to the system of representative and responsible government provided for by the Constitution.”14

Researchers themselves may contribute to suppression by compromising their independence. It is not uncommon for public health academics to be offered commissioned research contracts or to receive approvals for access to health data that are conditional upon the right of a government agency to withhold the research results from publication. We question the ethics of entering into such arrangements, which have potential to cause severe result-dependent publication bias. The NHMRC/AVCC Australian Code for the Responsible Conduct of Research is internally inconsistent on this point. Section 4.4 stresses that researchers have a responsibility to disseminate a full account of their findings, whereas section 4.12 states that any restrictions on communications that have been agreed with the sponsor must be honoured.15

Given that the sources of power wielded by government officials in suppressing health information are not those specifically legislated by a parliament, but derive indirectly from government control of funding, raw data access and collaborations of researchers with government employees, such as public sector service providers, the solution to the problem must be sought in a systemic intervention, the options for which were enumerated by the survey respondents.

In proposing what should be done, respondents identified a full range of participative, cognitive, and structural and legislative strategies. We agree with the majority of respondents that a comprehensive structural solution should be the ultimate aim, covering stages concerned with both the inputs and outputs of research. However, this is unlikely to occur without first taking the problem out of obscurity and into the public light where its obnoxious character in harming the Australian community can be exposed. Publicity will be effective in giving researchers courage to adhere steadfastly to their roles as expert commentators who are independent of government and to speak out when that role is compromised.16 Most importantly, the public must be apprised of the ills of suppression so that they will then voice their concerns. We summarise the control measures synthesised from the survey results and other sources as the following recommendations:

  • 1Work with the media to conduct an extended campaign to make the presence of the problem known to the public.
  • 2Advocate for the adoption by Australian governments of policies that match or exceed the OECD Guidelines for Managing Conflict of Interest in the Public Service.17
  • 3Advocate for the explicit inclusion of guidelines for the avoidance of result-dependent research funding and publication in working with governments in the AVCC/NHMRC Australian Code for Responsible Conduct of Research.15
  • 4Advocate for scientific journals to require statements from publishing authors that their work has not been subject to result-dependent funding or veto of publication by a government.18,19
  • 5Advocate for the joint adoption of a charter by government health agencies and research and academic institutions that supports the independent role of health researchers in evaluating the health system.
  • 6Promote a role of parliamentary ombudsmen or women in mediating the resolution of complaints by researchers concerning suppression.
  • 7Work with government ministries and agencies in synergistic partnerships that lead to better-quality programs for which ministers and other government officials receive due recognition.
  • 8Work with government ministries, agencies and the media to promote a culture that avoids blame and values constructive criticism.
  • 9Promote a role of institutional ethics committees in scrutinising the ethical behaviour of government agencies and researchers who work together.
  • 10Establish a surveillance system to monitor the occurrence of suppression events and to report on their trends.

The suppression of health information as practised by Australian governments represents very poor government practice in its milder forms and official corruption in more extreme cases. It is certainly unbecoming of a nation that prides itself on being an enlightened, democratic society where the government is held accountable for protecting and improving the lives of the people. The gravest danger faced by the readers of this journal is that we may become complicit in a conspiracy of silence whereby the results of this nationwide survey are greeted with scepticism or tabooed by those among us who are fearful of what speaking out could mean for our research careers. But if ever there was a duty for public health interests in Australia and New Zealand to speak out, we suggest it is on this very topic. For without an assurance of the freedom of public health researchers to communicate new information on environmental harms, the performance of services and the health status of vulnerable groups, the mission of public health will become futile.

Ancillary