As almost all households have a telephone [17
] random dialling should have produced a representative sample. However, the growth in telephone marketing and requests to support charities has increased citizen resistance to participation in a telephone survey which will have added some non-respondent bias. The survey was undertaken in Wellington and although the sample appeared to be over-represented by females and those people in the 26 to 45 years age group compared with Wellington data the differences did not indicate serious bias in the sample as they were not statistically significant (at the p = 0.05 level, Chi-square test). However, this does not prove that the sample is unbiased. We have no statistical data on those who refused to take part in the survey or those who had not searched the internet for health information.
The telephoned group had greater numbers than expected in the older age groups than the mall intercept respondents (Chi-square p < 0.05). Early evening (between 6 PM and 9 PM and Sunday afternoons were the best periods for the telephone survey, while the mall intercept interviews were conducted during the day. We consider the two data collection modes combined, captured a more representative sample than one mode alone. No other statistically significant differences between survey modes were found.
Although unavailability of data precluded an investigation of differences in income and educational status between the sample and the general population, international studies suggest that users of the internet have higher incomes and are younger than the general population. US and Australian researchers have found that those using the internet to find health information tended to be younger, were better educated, and had higher incomes than non-users [18
The Wellington study sample is not necessarily representative of New Zealand as a whole. Wellington has higher income levels, lower unemployment, younger average age and higher levels of education than the rest of the country. Households in Wellington have greater access to the internet than other households in New Zealand [10
]. The cost of a GP consultation in Wellington is probably higher than the New Zealand average but this was unable to be quantified as Wellington specific data were not available.
Most of our sample (83%) searched for information at home compared with 33% at work. A British survey [7
] also found that most people (66%) searched the internet at home for health information and 28% from work. As personal use of the internet is discouraged in most work places and given the sensitive nature of some of the information sought it was not unexpected that most searches were conducted in the privacy of the home and that the information was for either themselves or a family member. The American Medical Association [23
] note that personal privacy is the most important concern of users of medical information websites.
The top scoring search topic was general health and nutrition followed by information on a specific illness. Broadly similar results are reported in the international literature [7
]. Very few health professionals had suggested a website to respondents.
Respondents valued the information highly (more than a general practitioner consultation), generally found it useful, and most took some action as a result of the knowledge gained. The three most frequent actions taken after finding the relevant information related to, talking with someone, changing some aspect of lifestyle or consulting a mainstream medical or health professional. A low proportion of searchers (13%) did nothing. A US study found that 59% of users of the internet for health information did not discuss the findings with their doctor, but those that did valued the information more highly than those that did not [19
]. Australian research [22
] found that 19% of searchers used the information as a second opinion, 16% discussed it with their doctor or pharmacist and 11% changed the way in which they managed their healthcare. Research in the UK [7
] found that 93% of searchers found the information useful, 57% took some action to improve their health (mainly lifestyle changes), and 51% found information not provided by their doctor.
Our study is unique in that we asked Wellington respondents to assign a money value to the information found. The main limitations of our research are that respondents could have had difficulty in accurately recalling and estimating search times and values. Other limitations of the study were our inability to obtain precise income information (rather that income tax bands) and uncertainty that the sample was representative of the population that has used the internet to search for health information. We assumed that the value respondents assigned to information found could be used as a proxy for willingness to pay, and that the time costs of a search could be estimated from income levels. Sensitivity analysis revealed that net benefits of the information were sensitive to changes in the value of benefits but not to cost changes.
Although respondents stated that as a result of the information found they had made lifestyle or health related changes it is not known if these actions resulted in improved health outcomes and quality of life [22
]. With the exception of clearly beneficial lifestyle changes such as reduced consumption of alcohol and tobacco, more information would be necessary to evaluate the health impact of the actions taken.
The internet is rapidly evolving, transcends national borders, and is not owned and cannot be controlled by any one country or individual. Dissemination of information via the internet is potentially lower cost than through traditional media. Content can be quickly updated and instantaneously published to the world-wide-web. However, the quality and unbiased nature of the information accessible on the world-wide-web cannot be guaranteed, and consumers may be unable to make an informed choice based on this information alone.
Poor quality information and advice on the world-wide-web, if followed, could be deleterious to health. Harm to individuals and wastage of health care resources may be caused through non compliance with health care professionals' advice, inappropriate/over/under treatment. Good quality information may however improve understanding of illness, increase compliance and reduce waste.
The internet has an important and growing role in the efficient and equitable provision of health-related information, and presents a partial solution to the problem of asymmetric information. Never-the-less governments should not necessarily devote scarce health care resources in attempting to regulate, censor or build their own health information websites that contain every conceivable item of health information. Governments can help consumers by providing a website containing information on treatment options for common illnesses, guidance to help citizens evaluate the quality of web-based information and links to other useful and reputable websites. The Harvard School of Public Health [28
] has published a consumer's guide for evaluating health information and criteria for evaluating health related websites published in the BMJ [29
] and the American Medical Association has developed guidelines for health information websites [23
]. Such guidelines could be modified to suit the needs of New Zealand society and posted on a health information website. As a result of reviewing the literature and undertaking this study we have developed table . This table lists some important internet health information characteristics, consequences of variable information quality, and suggested criteria for an "official" health information website.
Internet health information: characteristics, consequences of information quality and suggested criteria for an "official" health information website