This study is part of a primary health care research project ("Medical Care in General Practice"; ) funded by the German Ministry of Education and Research. The research ethics committee of the University of G?ttingen approved the study protocol.

We developed a 1-hour interactive training course for participating GPs and practice assistants to cover:

1. patient enrolment and obtaining informed consent,

2. an explanation of the handling of the QL-recorder to participating patients,

3. interpretation and use of results during the consultation.

Training sessions took place within participating practices. We provided brief written manuals for the practice staff as well as interpretation aids to be given to patients.

Answers to open questions were independently analysed and discussed by three researchers (AS, ST, AR) according to the model of inductive category development [²¹]. Using the software Atlas.ti [²²], statements were classified into categories regarding subjective benefits as well as barriers with respect to routine HRQoL assessment. We used a codebook to define resulting categories and anchoring examples. The categories as well as the number of participants who mentioned them are presented in the results section.

Descriptive statistics regarding interview responses, patient characteristics and HRQoL results (frequencies/percentages, 95% confidence intervals CI, means/medians, standard deviations SD and interquartile ranges IQR) were computed using the Statistical Analysis Software package (SAS, Version 9.1).

In response to the invitation, 17 practices (20 GPs) agreed to participate in this study. The practices (8 urban and 9 rural) were spread all over Germany. Three GPs withdrew informed consent later due to personal reasons (severe illness of the practice assistant; change in practice software; lack of time).

According to the practice assistants, virtually all patients who were invited agreed to take part in the study. In total, 523 patients filled in the electronic questionnaires providing 664 assessments (figure 1), with substantial variation between practices (range = 5?205 assessments from 5?158 patients). Out of these, 413 patients completed only one assessment, and 110 patients completed two or more assessments. The total number of follow up assessments was 141, with a maximum of 5 assessments per patient.

A quarter of the patients suffered from a chronic airway disease and consequently answered the SGRQ questionnaire (125/523 patients; 24%); the remainder answered the QLQ-C30. Table 1 shows the characteristics of the 280 patients, 27 practice assistants and 17 GPs who were additionally interviewed by phone. There were no significant differences between the patients who took part in the electronic HRQoL assessment and those who additionally participated in the telephone interviews concerning age (mean: 61 ? 14 SD vs. 62 ? 13 years), diagnosis (chronic airway disease 24% vs. 26%) and distribution of gender (37% vs. 38% male).

The QL-recorders were successfully integrated, typically within half a day, into the 10 different software systems used by the various practices. Rare technical problems could be traced back to instabilities of the wireless networks, but not the QL-recorder itself. No data got lost and results of all HRQoL assessments could be easily exported within a few minutes at the end of the study.

At their initial electronic assessment, patients who completed the QLQ-C30 showed marked impairment compared to the general population in their function scores, global health and symptom scores for fatigue, pain, dyspnea and sleep disorders (figure 2). Similarly, HRQoL of patients with a chronic airway disease was markedly impaired (SGRQ symptoms: median 55.9 [interquartile range IQR 39.6]; activity 53.5 [IQR 43.4]; impact 31.3 [IQR 28.3]; total 39.8 [IQR 30.7]; scale range 0?100 with higher scores indicating more impairment).

As this was a pilot study, the sample size of practices was limited. Thus, on the practice level, it might be most adequate to interpret the results in a qualitative way. At least three types of responses can be distinguished with respect to the practices: (i) Some subscribers of the mailing list announcing the project may have read the invitation, but decided not to take part. Reasons for non-participation might be limited capacity due to workload or scepticism towards new technologies [²³]. (ii) Three practices withdrew informed consent after initially having indicated interest in participation. Reasons for withdrawal included lack of time, change in practice software and severe illness of the practice assistant. Other potential reasons could have been doubt regarding the benefits of electronic HRQoL assessment compared to the effort. (iii) Participating practices were heterogeneous with respect to the GPs' experience, age and gender as well as practice location. This may partly explain the variation in assessment frequencies, which are discussed below in more detail.

Practice assistants reported that virtually all patients who were invited agreed to participate. Hidden decision criteria of practice assistants regarding the selection of patients cannot be ruled out, but were not assessed in the interview. Most patients had little or no experience with computers, and the distribution of age and gender was typical for the general practice population, so we have no clear evidence for a selective invitation, e.g. of younger or more educated patients. Similarly, patients who participated vs. those who did not participate in the telephone interviews showed comparable characteristics.

By means of wirelessly integrated tablet computers, HRQoL data could be easily collected, transferred and automatically printed, making the results available during the same office visit. Thus, several technical and logistic problems such as the patients' inability to handle a mouse or incorrect allocation of patient numbers (IDs) have been successfully solved. Results could be automatically imported into a variety of electronic patient records as recommended by physicians in another study [¹³]. Patients had no difficulty in completing the HRQoL questionnaires on the tablet computer, which confirms other findings [¹⁵].

The majority of participating patients, practice assistants and GPs were satisfied with the electronic HRQoL measurement. GPs appreciated the additional information indicating marked HRQoL impairments in their patients. The assessments showed that most patients had specific limitations e.g. in their physical or role function. Among the QLQ-C30 symptom scales, those for pain, fatigue and sleep disturbance in particular showed clinically significant differences compared to reference values from the general population [²⁴]. These symptoms are often overlooked in daily routine [¹²]. Asthma patients, too, showed an impaired quality of life in the SGRQ compared to the general population [²⁵], with different patterns in symptoms, activity and the impact of the disease. Results for individual patients showed distinct impairments, rather than uniform patterns, which could help the GP to recognise those patients' individual difficulties.

GPs emphasised that the standardised and reproducible HRQoL results helped them to initiate a focused dialogue with the patient, e.g. regarding sensitive topics. As the questionnaires addressed multiple aspects, patients felt the assessment contributed to the physicians' understanding of their personal condition and circumstances. This is in line with other studies showing that patients perceive HRQoL assessments as a valuable support for their care [²⁶,²⁷] and prefer electronic procedures to paper-pencil assessment [¹⁰,²⁸].

Technically, the HRQoL assessment was functional, well accepted and provided usable HRQoL information. Most participants, however, made less practical use of the new tool than expected. Obviously, there are still barriers to overcome. As indicated by other studies, there seems to be a discrepancy between physicians' appraisal of the importance of HRQoL assessment [²⁹] and the intensity of its application in everyday practice [⁶,¹¹]. In our study, the HRQoL assessment was organised by the practice staff and took place within the normal routine, while most previous studies employed research assistants to manage the data collection [³⁰].

A typical single-handed German GP may see 50 to 100 patients per day. There are no specialised practice managers, and the practice assistant must complete all administrative and medical tasks per patient within 3 ? 15 minutes. In the year of our study, legislative changes increased the practice workload by bringing in new documentation requirements and billing system changes. Germany has the shortest consultation times of several European countries [³¹]. While practice assistants considered the effort to simply explain the study aims, and the purpose and handling of the QL-recorder acceptable, additional activities ? including obtaining formal informed consent ? required more time than some practice assistants could afford during busy practice hours. The effort to carry out a HRQoL assessment may be judged positive with the expected benefit in mind, but still be prohibitive given the time pressures of practice reality.

Consistent with this, participating GPs pointed at two primary hindrances: The lack of time to inform patients and to discuss HRQoL data in a busy general practice, and the paucity of resources to alleviate HRQoL deficits. Though the electronic tool reduces workload compared to a paper-pencil measurement, HRQoL assessment still remains an additional task. Time constraints limit the effectiveness of HRQoL assessment if physicians have no capacity to act and appropriately use the information obtained [¹⁴,³²]. One practice however was able to perform a high number of electronic HRQoL assessments. This practice cared for the population of a larger island, and was run by a GP and practice team with special organisational skills, dedication and proven research interest [³³].

The integration of electronic HRQoL assessment into general practices brings with it the prospect of reciprocal transfer of knowledge from patient rated outcomes research into practice and from practice into research. Combining such HRQoL data with information from pseudonymised electronic patient records, (which can be extracted from practice computers after patients' informed consent), would provide a basis for scientific analyses of associations between HRQoL and patients' characteristics, disease and treatment [¹²]. The availability of HRQoL results immediately during the consultation could contribute to patient centred care, help to focus the patient-physician consultation, support the definition of therapeutic goals as well as the evaluation of their achievement, and provide standardised data, which can be compared intra- and inter- individually.

To enhance feasibility and usability of electronic HRQoL assessment, we recommend the following steps:

? Training: Though most participants appreciated the 1-hour training, it might be useful to accompany practice assistants during the first days of electronic HRQoL assessment. This could help to bridge the gap between theory and practice, as HRQoL issues have rarely been part of the medical curriculum [³⁴].

? Printout and interpretation: Additional verbal summaries might be easier to understand compared to graphics. As most HRQoL scales did not exceed the threshold of 50 on average, this reference value may be adequate for cancer patients [³⁵] but does not provide sufficient orientation for general practice.

? Adaption to local needs: Practitioners may be interested in selected aspects of HRQoL depending upon their patient clientele or upon the portfolio of supportive measures they can actually provide. For routine care, questionnaires and result presentations should be tailored to these needs to increase the relevance perceived by the GP, and the probability that documented impairments have actual medical consequences.

? Informed consent: Obtaining patients' written informed consent put an extra burden on practice assistants. In order to make the procedure more convenient, data collection would need to be regarded as a standard component of medical service [³⁶], so that written informed consent would not be required for each assessment.

? Incentives: GPs and practice-assistants received only a small financial allowance within this study. Lack of remuneration for HRQoL assessment and discussion of results is regarded as a barrier to its implementation. Also, regular discussion groups of physicians addressing HRQoL topics might be helpful [²⁰], but could not be realised in our project since participating practices were located in distant German regions.

The availability of tools and training is only the first step as the clinical application of HRQoL assessment represents a complex intervention [³⁷]. Critical questions for any larger project in this area are whether the resources for adequate patient invitation and consideration of HRQoL results in medical decision-making can be provided. A thorough understanding of the clinical workflow, practice requirements and goals are essential for the successful implementation of innovative health information technologies in medical practice [³⁸].

The authors are indebted to all participating patients, practice assistants and GPs. This work was supported by the German Ministry of Education and Research [grant number 01GK0201].

* as defined according to the qualitative content analysis approach.

** number of patients; mentions of several categories per patients possible

* as defined according to the qualitative content analysis approach.

** number of GPs and practice assistants (PA); mentions of several categories per participant possible

* as defined according to the qualitative content analysis approach.

** number of GPs; mentions of several categories per GP possible