Trust in health care: an information perspective
Lars Höglund, Elena Macevičiūtė and T.D. Wilson
Swedish School for Library and Information
Science, University College, 501 90 Borås
E-mail: lars.hoglund; elena.maceviciute; firstname.lastname@example.org
This article reports on an exploratory analysis of existing data obtained through
a national survey carried out in Sweden. The survey questionnaire seeks information
on a wide range of issues, including health care, library use and Internet use. The analysis presented here explores the relationship among these variables and the concept of trust in health care institutions. The results indicate that most of the correlations theoretically suggested were very small and that trust in health institutions in general is high but not strongly related to standard demographic variables found in a general survey of a large population. This exploratory study suggests that more specific indicators of health, experience from health institutions and health-related media exposure are needed to test, in
greater depth, the relationships between information exposure, health and attitudes
towards health institutions.
Large-scale surveys of the general public are not uncommon in the developed world (for
example, there are the British Household Panel Survey, the Polish General Social Survey,
and the various Eurobarometer surveys in the European Union) and surveys of public
health are relatively abundant (for example, the Medical Expenditure Panel Survey in the
USA, the Commonwealth Fund 2001 International Health Policy Survey, and the National
Survey on Health and Medical Care in France). However, it is rather uncommon for either
of these kinds of surveys to include questions relating to access to and use of information
services of any kind.
This article draws upon the national survey of the Swedish population, which has been
implemented annually since 1986 by the SOM Institute of the University of Gothenburg.
This survey includes questions proposed by various research organizations about politics,
society, use of media, public service, the environment, risks, new media technology and
people's leisure activities.
We have chosen to explore, in this article, as far as the data allow, the relationship
between trust in health care in Sweden and a variety of demographic and informational
Public trust in health care
The issue of trust in health care institutions has become of critical importance as governments
seek to justify expenditure on health care and as providers of health care and health
insurance companies (particularly in countries with lower levels of public expenditure on
health) seek to justify their charges . Straten et al.  note that:
Public trust in health care could be defined as being confident that you will be
adequately treated when you are in need of health care.
Straten et al. draw attention to Blendon's contention  that public trust in health care in America had declined in the 20 years prior to the publication of his article, and that this
was the result of institutional changes in health care.
More recently, Orr  notes:
Recent evidence suggests that patients and physicians too often perceive the current
health care system to be unfair or illegitimate because of the ways in which coverage
decisions appear to be made. This perceived unfairness is disrupting trust in the health
care system. Developing a framework that would promote trust in health care
organizations by fostering and demonstrating fairness in coverage decisions would
provide value to all participants in the health care system.
Trust is taken to be important because of its relation to other aspects of behaviour in
relation to health. For example, there is a presumed relationship between trust and
future behaviour, in that persons who do not have trust in the system may seek alternative
therapies or have lower compliance with the therapies recommended by health care
institutions . How trust is composed, however, has at times been taken for granted,
while other researchers have sought to identify the elements of which it is composed
and some researchers have identified links to various aspects of information. Gray ,
for example, suggests that, with a decline in trust, patients will ask for a second opinion,
and seek to identify the 'best physicians' or the 'best hospitals' - all of which require
information for their satisfaction. Mechanic  identifies 'management of confidential
information' and 'openness in providing and receiving information' as two of five dimensions
Straten et al.  carried out both qualitative and quantitative research towards the creation of an instrument to measure the dimensions of public trust in health care. One
of the initial six factors, identified by analysis of interview data, was 'information supply
and communication by care providers'. This dimension also appeared in the follow-up
quantitative analysis, using factor analysis: it provided explanation for only a very small
percentage of the variance (3.7 per cent) but nevertheless was a clearly identifiable factor
which correlated significantly with the other factors, as well as with variables such as
negative or positive experience of healthcare by the persons involved.
Hagström  explores among other topics the information media (namely, press) influence
on healthcare and institutional policy. The study reflects to some extent the complex
interaction between media and decision-makers' behaviour. It also suggests that public
opinion and the general attitude of citizens towards health and medical care problems as
formed by information sources are significant factors in these relationships. In Sweden,
other studies have been conducted relating information provision to the health behaviour
of different population groups [8, 9].
Although the study reported here cannot deal with the relationship between trust and
these specific information variables, we believe there is sufficient evidence of an association
to allow us to examine more general aspects of information behaviour, to discover
whether variables such as library use and Internet access, in general, may be related to
experience of healthcare and trust in healthcare institutions.
Demographic characteristics and health information behaviour
The literature relating to health information behaviour reveals a number of associations
with demographic characteristics. Connell and Crawford  found that the amount of
health information received by urban residents from all sources declined with age, but
that older rural women received a great deal of information from a variety of sources and
that the amount declined only slightly with age. Older men received far less information
than younger men. On the other hand, MacHaffie  suggests that persons with severe
and persistent mental illness living in more rural areas with fewer resources are likely to
experience less exposure to health promotion information from all sources. Research has
also shown that men were more likely to perceive lack of knowledge as a barrier to
seeking preventive healthcare in the rural Appalachian region of the USA, whereas older
and less educated persons were more likely to identify cost as a barrier .
Slevin et al. , in a study of a national cancer information service (BACUP), found that use of the service was mainly (80 per cent) by women (although men are marginally
more at risk), both in the 30-49 years age group (52 per cent of enquirers) and in those
over 60 years of age (17 per cent of enquirers). More than 85 per cent of enquirers were
in non-manual occupations and 97 per cent were white (1.2 per cent came from the
Indian subcontinent, compared with 3.6 per cent in the population as a whole). The
authors commented that 'lower social classes make much less use of community health
and preventive services', although again there is a slightly higher incidence of cancer in
the lower socio-economic groups.
Connell and Crawford  found women reported receiving more health information
than men from all sources and attributed this to women's traditional role as caregivers
and 'lay healthcare providers'.
The use of the Internet by various groups has also become a subject of research in
recent years and various relationships with some demographic variables have been found.
Gordon et al.  in a survey of patients attending a rheumatology clinic found that 43 per cent had access to the Internet (27 per cent at home) and that 27 per cent had
searched for medical information. Age was shown to be a factor: those who searched for
information had a median age of 48, while those who did not had a median age of 62.
The patients in this study were predominantly female, but studies of Internet use in
general (at least in the UK) suggest that it is now as likely to be used by women as by
US surveys also show that health information on the Internet is an age-related activity
. More children (65 percent of those aged 10-13) and teenagers (75 percent of those aged 14-17) use the Internet than other age groups. But health information seekers tend
to be ages 30-64 . The latter survey of Americans also shows that finding health
information on the Internet affects how people manage their health. Almost half (48
percent) of users who have gone online for health information say it improved the way
they take care of themselves.
The research company Datamonitor  surveyed over 4500 adults in France,
Germany, Italy, Spain, the UK and the US, and found that 57 percent of respondents had
consulted Internet sources when looking for health information. This compares with 76
percent who consulted their physician, 73 percent who looked for information via television,
books and magazines, and 53 percent who sought advice from family and friends.
The study indicates that between 32 and 34 percent of adults aged 18 to 54 went online
to look for health information during the past 12 months. However, that number falls to
27 percent for adults in the 55-64 age group and to 14 percent for those aged 65 and
over. Datamonitor's research shows that general health sites and sites run by governments
or institutions are considered more credible than those run by pharmaceutical firms,
because they are seen to be more objective.
Harris Interactive  has been tracking the people who search information on health topics ('cyberchondriacs') in the US, France, Germany and Japan for 4 years. Their number
has doubled during this time (54 million in 1998, 110 million in 2002). The survey
summarizes the types of websites visited by cyberchondriacs (Table 1). Most people say
that they look for information on their own, but rely on it only if their doctor confirms
the quality (60 per cent in the US and 47 per cent in Japan). Another group tends to judge the information without consulting doctors (37 per cent in Japan, 46 per cent in Germany). Only in some cases (particularly in Japan) do people look for information only
if their doctor tells them to do this.
Table 1 Websites visited by 'cyberchondriacs' 
|Web site category||US|| France|| Germany|| Japan|
|%|| % ||%|| %|
|Medical journals|| 45|| 45|| 33|| 27|
|Commercial health pages ||44|| 52|| 40|| 32|
|Academic or research institutions || 43|| 50|| 50|| 33|
|Pharmaceutical companies || 34|| 18|| 27|| 23|
|Medical societies || 34|| 21||45|| 35|
|Patient support or advocacy group for specific diseases || 29|| 30|| 42|| 46|
|News media || 29|| 28|| 40|| 33|
|Government sites || 25 ||29|| 24|| 34|
|Hospitals || 16|| 21|| 14|| 36|
|Individual doctors || 11|| 10|| 15|| 25|
A Swedish survey in East Götland found that higher credibility is attributed to health
information from healthcare staff and family than to the same information on the net
. The Riks-SOM survey includes questions relating to many of the factors discussed here, including the main demographic factors, and there is the potential, therefore, to
explore the data from the perspective of both information use and health factors.
The possible relationships between the variables of interest to us, that is, health status,
contact with health institutions, library use, Internet use, and trust in health institutions,
are numerous. Consequently, a theoretical model is needed to guide our analysis.
Our initial model of the relationships among these variables is shown in Figure 1. The
model suggests the following theoretical propositions:
P1 Health status will be related to demographic variables such as age, sex and
occupation. Thus, elderly people are more likely than young persons to have a
lower health status, and manual workers more likely than professional workers.
P2 Contact with healthcare institutions will be related to both health status and
demographic factors. Thus, persons in good health are unlikely to seek contact
with healthcare institutions, and demographic variables such as age, occupation
and rural or urban dwelling are likely to have an impact. Men are also less likely
to visit general practitioners, for example, than are women.
P3 The informational variables are likely to be affected by both health status and
demographic variables and, in turn, are likely to have an impact on contact with
healthcare institutions and on trust in healthcare institutions, in that those who
show most use of information resources are likely to have higher trust in health
Given the general nature of the data available, it is not possible for us to identify information
use (i.e., library use and Internet use) specifically for the purposes of seeking health
information, since this use was not specifically identified in the Riks-SOM survey. However,
it is possible to identify general associations that may be explored through more detailed
questioning in relation to healthcare in future surveys.
Riks-SOM, as noted earlier, incorporates questions proposed (and paid for) by a variety of
research organizations. This article, therefore, draws upon questions that have been
proposed for a variety of reasons. However, questions on library and information use were
proposed by the senior author, and these are related to questions proposed by others.
In totality, therefore, the survey was not designed specifically to test the model, but is a
post facto analysis of available data.
The survey is mailed by the Institute to a randomly selected sample of 6,000 persons
between the ages of 15 and 85. In 2001 the response rate was 60.63 per cent, giving
data on 3,638 persons.
The questions on library and information use used in the analysis were:
- How often in the past 12 months have you visited the library?
- How often do you do the following in the library: . . . use a computer to look for
information on the Internet?
- Do you have Internet access (at home, at work, or elsewhere)?
- Do you, or does anyone in your household, subscribe to any daily newspaper?
The questions on healthcare used in the study were:
- How large is your trust in the way the following institutions do their work?
- How large is your trust in the staff within the following institutions?
- Estimate your present health condition from 1 (my worst possible health
condition) to 10 (my best possible).
- How is your current health condition compared with the situation 12 month ago?
Finally, the demographic characteristics used were: age, sex, educational level attained,
residence in rural or urban area, and family income.
The survey is highly representative of the Swedish population: 49.5 per cent of the respondents
were male, and 50.5 per cent were female. This is very close to the official population
figures in December 2002, which gave values of 49.52 per cent for men and 50.48
per cent for women .
The age distribution of the respondents is close to that of the population at large, but
slightly under-represents the age groups 15-39 and 75-85 and over-represents the group
40-75 (Table 2).
Table 2 Age distributions of Swedish population and Riks-SOM respondents
|Swedish population||SOM respondents|
|Age group||No.||%||Age group||No.||%|
The sex and age distributions of the Riks-SOM respondents, therefore, can give us
confidence that the survey data closely represent the Swedish population in general.
Health status was measured on a scale from 1 (poor) to 11 (excellent). The results (Figure
2) show a generally high level of health.
Health status was redefined as low (1-5), medium (6-8) and high (9-11) and a modest
but highly significant correlation between perceived health and age was found (Table 3).
Table 3 Health status 1-3 . ålder4 cross-tabulation
|Health status || ||Ålder4 age groups||Total|
|1 ||Count ||91 ||260 ||134 ||485|
|% within Ålder4 || 12.6 || 13.8 || 15.1 || 13.9|
|2 || Count || 205 || 576 || 345 || 1126|
|% within Ålder4 || 28.4 || 30.5 || 38.9 || 32.2|
|3 || Count || 426 || 1052 || 408 || 1886|
|% within Ålder4 || 59.0 || 55.7 || 46.0 || 53.9|
|Total || Count || 722 || 1888 || 887 || 3497|
|% within Ålder4 || 100.0 || 100.0 || 100.0 || 100.0|
Trust in healthcare
Trust in health institutions was measured on a five-point scale (very strong trust to very
weak trust) and the results showed a generally high level of trust, with 58.4 per cent
choosing the top two trust levels, compared with only 15.6 per cent choosing the bottom
two levels. The results are shown in Figure 3.
As can be seen from Table 4, the differences in trust related to contact with healthcare
institutions as well as to different social and demographic variables seem to be small.
In fact all the variables tested showed a very small impact on trust in health institutions
and a somewhat greater impact for trust in staff.
Table 4 Trust in healthcare and selected variables: % with high trust (n = 3382)
|% with high trust in staff||% with high trust in health institutions|
|Contact with health care institutions: no ||68 ||(709) ||58 ||(744)|
|Contact with health care institutions: yes || 73 || (2673) || 60 || (2642)|
|Household income: 1 (low) || 71 || (251) || 59 || (252)|
| 2 || 69 ||(568) || 60 || (573)|
| 3 || 67 ||(670) || 56 || (671)|
| 4 || 73 ||(620) || 62 || (628)|
| 5 || 76 ||(485) || 62 || (487)|
| 6 || 73 ||(270) || 57 || (271)|
| 7 (high) || 82 || (128) || 68 || (128)|
|Education: 1 (low) || 69 || (1551) || 57 || (1553)|
| 2 || 71 || (777) || 59 || (794)|
| 3 (high) || 76 || (977) || 65 || (984)|
Media exposure might affect trust, but in this case the more general indicators showed
small correlations with trust. This was also tested in a series of regression analyses: the
results are shown in Table 5.
Table 5 Summary of exploratory regression analysis
|Dependent variable: trust in healthcare institutions, stepwise regression|
|Variables entered || ΣR || ΣR2 || Adj.R2 || Beta || Sig.|
|1 Trust in staff || 0.700 || 0.490 || 0.489 || 0.701 || 0.000|
|2 Age || 0.701 || 0.492 || 0.491 || 0.043 || 0.000|
|Excluded (all other variables): Internet availability, education, subscription to daily newspaper, library visits, contact with health care, present health condition, household income and sex.|
|Dependent variable: trust in staff within health institutions, stepwise regression|
|Variables entered || ΣR || ΣR2 || Adj.R2 || Beta || Sig.|
|1 Subscription to daily newspaper || 0.146 || 0.021 || 0.021 || 0.144 || 0.000|
|2 Present health condition || 0.202 || 0.041 || 0.040 || 0.130 || 0.000|
|3 Sex || 0.052 || 0.049 || 0.046 || 0.093 || 0.001|
|4 Education || 0.231 || 0.053 || 0.050 || 0.069 || 0.011|
Discussion and conclusions
The results do not support the proposed theoretical model. Nor do they seem to support
some earlier findings from more limited populations in other countries. There may be
several reasons for this. First of all, trust in health service institutions in Sweden is high at
this very general level. Among a number of institutions studied for a number of years,
healthcare is normally at the top, along with the universities, the police, the King, and
radio and TV. During the 1990s the trend has been somewhat downwards, indicating a
general decrease in trust for social institutions. Some theories of trust suggest that trustworthiness,
quality, interest and perceived importance together with values and opinions
should explain the level of trust for a specific institution . It does not appear to be possible to trace the complex relationships between different media, their content and the
consequences for individual opinion with these survey data. However, long-term changes
in opinion can be related to media, but this is normally very difficult to show. Many years
of research and debate about the effects of violence in various media illustrate this point.
A further conclusion relates to the need for caution in accepting the results of smallscale
research as representative of populations in general - a point to which we have
already alluded. There is today a fashion for qualitative research in many fields, not excepting
the healthcare arena, and the largely self-selected groups of persons studied are very
unlikely to represent the generality of the population. Where the results of such research
are used to guide public policy, therefore, the results may be other than expected. Where
public resources are to be used to bring about changes in healthcare, it is important that
the research should represent the population at large as well as can be obtained with
current social survey methods. Collaboration with large-scale surveys may be more appropriate
than totally independent but small-scale studies - no matter how much insight into
individual problems is obtained.
Research that does not confirm a proposed theoretical model frequently goes unpublished.
In this case the study results imply that further research and more specific indicators
should be used before the model can be more firmly rejected. However, in spite of
this, the study has shown that existing databanks can be used for explorative research
and model testing to a much larger degree than is normally undertaken. The general
quality and the methodological possibilities with large surveys are normally good. In this
case the survey was not intended to be used for this kind of health-related research. It is
suggested that more and better health indicators should be added in the forthcoming
annual Riks-SOM surveys, as well as more appropriate indicators relating to information
and media use for health-related purposes. We hope to be in a position to influence this
situation for the next survey.
1 The questions on trust in institutions are part of research by Holmberg and Weibull, who are also
the editors of the main report from the 2001 Riks-SOM survey .
- Anders G. Health against Wealth: HMOs and the Breakdown of Medical Trust. New York: Houghton Mifflin, 1996.
- Straten G F M, Friele R D, Groenewegen P P. Public trust in Dutch health care. Social Science & Medicine 2002; 55; 227-34.
- Blendon R. The public's view of the future of health care. Journal of the American Medical Association 1988; 259; 3587-93.
- Orr A S. Consensus towards trust: improving fairness in health care coverage decisions. Abstract of a paper presented at the 130th Annual Meeting of APHA, Philadelphia, 2002. In Proceedings 9-13. November 2002. http://apha.confex.com, March 2003.
- Gray B. Trust and trustworthy care in the managed care era. Health Affairs 1997; 16; 34-49.
- Mechanic D. Changing medical organization and the erosion of trust. The Milbank Quarterly 1996; 74 (2); 171-89.
- Hagström B. Svensk sjukvård i och under press: pressens raportering om svensk hälso- och sjukvård. Lund: Studentlitteratur, 2002.
- Jarlbro G, Jönsson A, Windamt S. Aids Ett drama i flera akter: en innehållsanalys av svensk press. Lund: Studentlitteratur, 1992.
- Persson E, Sandstrom B, Jarlbro G. Sources of information, experiences and opinions on sexuality, contraception and STD protection among young Swedish students. Advances in Contraception 1992; 8 (1); 41-49.
- Connell C M, Crawford C O. How people obtain their health information: a survey in two Pennsylvania counties. Public Health Reports 1988; 103; 189-95.
- MacHaffie S. Health promotion information: sources and significance for those with serious and persistent mental illness. Archives of Psychiatric Nursing 2002; 16 (6); 263-74.
- Elnicki D M, Morris D K, Shockcor W T. Patient-perceived barriers to preventive health-care among indigent, rural Appalachian patients. Archives of Internal Medicine 1995; 155 (4); 421-4.
- Slevin M L, Terry Y, Hallett N, Jefferies S, Launder S, Plant R, Wax H, McElwain T. BACUP: the first two years. Evaluation of a national cancer information service. British Medical Journal 1988; 297; 669-72.
- Gordon M M, Capell, H A, Madhok R. The use of the Internet as a resource for health information among patients attending a rheumatology clinic. Rheumatology 2002; 41 (12); 1402-5.
- Eysenbach G, Sa E, Diepgen T. Shopping around the Internet today and tomorrow: towards the millennium of cybermedicine. British Medical Journal 1999; 319; 1294.
- US National Telecommunications and Information Administration. A nation online: how Americans are expanding their use of the Internet. 2002. http://www.ntia.doc.gov, March 2002.
- Fox S, Rainie L, Horrigan J, Lenhart A, Spooner T, Burke M, Lewis O, Carter C. The online health care revolution: how the web helps Americans take better care of themselves. Pew Internet & American Life Project, 2000. http://www.pewinternet.org, March 2003.
- Datamonitor. Growth in numbers seeking health info online. NUA surveys, 2002. http://www.nua.com, March 2002.
- Harris Interactive. Four-nations survey shows widespread but different levels of Internet use for health purposes. Health Care News 2002; 2 (11). http://www.harrisinteractive.com, March 2003.
- Garbenby P, Husberg M. Stort intresse för mer hälsoinformation. Läkartidningen 2001; 23; 2814-2816.
- Statistika centralbyrån. Swedish population on 31 December 2002. Stockholm, 2003. http://www.scb.se/statistik, March 2003.
- Holmberg S, Weibull L. Det våras för politiken. Trettiotvå artiklar om politik, medier och samhälle: SOM-undersökningen 2001. SOM-rapport no. 30, 2002. Göteborg: SOM-institutet, Goteborg universitet.
This paper was presented at the iSHIMR meeting, Borås, Sweden in 2003. It is scheduled for publication in Health Informatics Journal.
How to cite this paper
Hölund, L., Macevičiūtė, E. & Wilson, T.D. Trust in health care: an information perspective. Health Informatics Journal, 10(1), 2004. 37-48 [Available at http://informationr.net/tdw/publ/papers/2003iSHIMR.html]