Improving patient–provider communications: a web‐access national health‐care survey panel

Background

Facilitating dialogue between service users, policy leads and health professionals is essential in any modern health‐care system. In Germany, there has been little tradition in citizen involvement for health‐care policy making and priority‐setting and data on the general publics’ view on the health‐care system are still very limited. Until recently, such data are mostly available through traditional public‐opinion polling and often lack depth, methodological rigour, complexity and independence of source. To overcome this deficiency, the department of health policy at the University of Witten/Herdecke initiated the project GesundheitsPanel in 2003, available at http://www.gesundheitspanel.de. The GesundheitsPanel is an internet survey panel aimed to systematically investigate citizens’ preferences and expectations towards the German health‐care system in an independent scientific context. The panel provides the environment under which specific survey projects can be processed, fielded and administered to samples of panel members.

A health‐care survey panel

Interested citizens visit the internet site of the Gesundheits Panel and sign up for membership. These preliminary members are then invited to participate in a first ‘master’ survey to collect their data. Besides the assessment of demographic information such as occupational and educational background, this master survey also asks for information related to utilization of health‐care resources and self‐rated health.

With full membership, participants agree to be regularly surveyed and consent to membership and data protection rules. The master data available for each panel member can then be used as selection criteria to draw well‐differentiated samples for specific survey projects. In addition, historical survey responses and indicators that measure each member's survey performance can be used for sampling. Panel members are offered incentives for survey participation through a bonus scheme. For example, for responding to a survey with an approximated length of 15–20 min, participants are credited points equivalent to 1.50–2.00 EUR, also depending on complexity of the involved tasks. With accumulation of a certain minimum number of bonus points these can be exchanged for gift certificates. On a personalized panel site, members are informed about their current bonus balance and can manage their gift certificates. Personal data as well as responses to surveys and usage of incentives are kept completely confidential. Data are anonymized for further processing and reported only in aggregate.

Target population and recruitment

Contrary to other ‘health‐care’ panels, which survey health‐care providers or experts, the GesundheitsPanel concentrates on a general population approach.

However, due to media releases in the specialized press, a considerable number of health‐care professionals have also been attracted. Although not actively promoted, this development may offer opportunities to investigate differences in the views of health‐care professionals and the public in specific research projects. Panel members are recruited online, via internet search engines, listing in internet portals and announcements on a number of high‐traffic internet pages. By explicitly mismatching media formats, that is, campaigning ‘offline’ for ‘online’ participation, it is attempted to attract populations other than skilled frequent internet users. Citizens are invited to participate through media releases in the radio, video text, printed press and via hand‐outs. In addition, panel members are offered incentives for recruiting new members.

Internet panels cannot be truly representative for the entire population. The GesundheitsPanel aims to achieve representativeness in certain demographic characteristics such as age, sex, martial status and education by consequently addressing groups commonly underrepresented. Compared with stand‐alone internet surveys, the problem of self‐selectivity is alleviated with the panel design, as members select themselves for panel membership, but not for participation in particular survey projects (Box 1). ¹

Table Box 1  .

Characteristics of the GesundheitsPanel

• ‘General population’ approach

• Cross‐sectional and longitudinal surveys

• Use of interactive preference elicitation techniques,  e.g. conjoint analysis

• Presentation of complex scenarios

• Comprehensive sampling procedures, e.g. multi‐stage, random and quota samples

• Systematic data controlling and prevention of unintended errors

• High motivation of panel members

Research objectives

With its interactive features, the GesundheitsPanel has great potential, both in survey objective and methodology. Research questions can be processed in depth in particular survey projects. A variety of objectives from health economics and health services research may be covered, ranging from evaluation of specific health‐care services and their delivery to issues of fairness and justice and preferences for health‐care resource distribution.

The GesundheitsPanel started with a survey addressing the planned reform of the German health‐care system after the proposed framework had been presented by health minister Ulla Schmidt. This survey stresses the opening of hospitals for outpatient specialized services, acceptance of the gatekeeper model and priorities for the statutory health insurance benefits catalogue. Surveys projected for the near future include ‘views on medical errors’, ‘risk perception and participation in screening and prevention programmes’ and ‘preferences for priority setting in health‐care’.

Principal survey designs

The opportunity to contact, sample and invite panel members facilitates not only the immediate realization of surveys reflecting current debates but also to collect attitudes and preferences over longer periods of time so that longitudinal and time‐series data can be generated. The panel is designated to employ three principal designs of surveys:

• •Experimental and concomitant methodological surveys that are used to investigate survey methodology, preference elicitation techniques and instrumental quality.
• •Non‐recurring surveys that allow a rapid and in‐depth study in specific (representative) samples.
• •Panel surveys that aim to observe changes in preferences or attitudes over time.

Survey methodology

All common survey techniques can be administered in survey projects within the panel. However, a specific feature of the GesundheitsPanel and the research background at the department is the development and use of interactive preference elicitation techniques, such as conjoint analysis and vignette studies. ² , ³

In brief, such techniques ask the respondent to consider opportunity costs and consider trade‐offs between different attributes of hypothetical scenarios. Data obtained with these methods may be more realistic than traditional ‘attitude’– surveys as most individual and policy decisions require such evaluations of competing options, for example, setting priorities by allocating funds on one programme at the cost of other services.

Computerized surveying is of special value for tasks requiring trade‐offs or the estimation of a point of indifference, such as the standard gamble, person trade‐off or willingness‐to‐pay technique. The GesundheitsPanel uses highly individualized survey designs that incorporate preference elicitation algorithms invisible to respondents. For example, questions, scenarios, design and response codes can be programmed to be mutually dependent of each respondent's responses to prior questions, or even the master data on file. Survey complexity can be decreased significantly, making it easier for a respondent to use.

Data collection and data quality

Key advantages of the web‐accessible resource are that panel members can respond quickly and effectively to the survey. For example, in the first survey (administered to members who have completed the master survey previously), a sample exhaustion of 75% within 3 days of fielding was achieved. After 8 days and one reminder sent, the response rate was about 97.5%.

The panel includes a variety of features to increase data quality and inform researchers on the participation process and performance of the survey.

First, the technology can be used to reduce unintended errors during the survey. Biases introduced by survey presentation can easily be avoided by automated random rotation of questions and response codes. In addition, validity checks can be built in to control for consistency and inform respondents about potentially contradictory responses.

Secondly, performance data such as estimates of time for question or survey completion, question‐specific drop‐out rates, disruption and resumption, are automatically recorded and provide information on respondents’ behaviour during the survey otherwise not available to researchers. Such additional information can be of great value in the assessment and interpretation of survey methods, instruments and data quality. ⁴

Finally, by integrating graphical, other multimedia content and applications, abstract information such as quantitative estimates of risk, probability or mortality rates can be presented while preserving complexity. This has advantages over traditional self‐administered surveys which are often not appropriate for objectives requiring presentation of complex or abstract content, demanding high cognitive performance of respondents.

It should be recognized that extensive use of multimedia content also increases technological requirements. Consequently, an individual's participation may be restricted due to configuration of his/her computer software. ⁵

Dissemination of results

Upon completion of a project, responses to specific survey projects are linked to the master data available for each panel member and may also be related to responses to previous surveys. Results will be presented open to the public at a glance, while detailed results and analyses will be published in relevant professional journals. It is also intended to integrate survey results in teaching and advanced education of medical students where reasonable benefits are expected. This applies, for example, to preferences for health services delivery and the area of communication between patients and providers.

Conclusion

A resource such as the GesundheitsPanel should provide an excellent response to a requirement of health communities to demonstrate effective engagement with service users. The success of the GesundheitsPanel project is mainly determined by the publics’ willingness continually to participate in the surveys and the success in the acquisition of panelists representing a broad range of disciplines. At the time of writing, interest of the general public, providers and the media is high and the motivation and discipline of panel members impressive.