Study design
The VBSQ (Additional file 1) was designed in cooperation with a clinical chemist and instructors at the Clinical Chemistry Laboratory, Umea University Hospital and researchers at the Dept of Nursing and Dept of Medical Biosciences, Umea University. The VBSQ was based on standard instructions for VBS [13, 14] in line with international recommendations [11, 12]. The questionnaire was in a Swedish version and tested for content, face validity and reliability in a Swedish health care context. The layout of the questionnaire was designed to be easy to read and number of pages was limited, to ensure that the questionnaire could be completed within a reasonable period of time.
Setting
This follow-up study was performed in the PHC and hospital settings in northern Sweden.
Population
Validity
To achieve content validity an extensive dialogue with professionals on different aspects on VBS and questionnaire design was performed. This included senior researchers, registered nurses, enrolled nurses, biomedical technicians, heads of wards, physicians at University Departments and hospital wards as well as researchers with considerable experience in questionnaire design. Instructors from the local laboratory also double-checked that questions were in accordance with guidelines [13, 14]. Criterion validity is applied as we followed guidelines. This is verified as laboratory staff perform more correct VBS according to guidelines [25–27], and therefore can be used to benchmark VBS performance, which is also confirmed in our studies [16–18, 20, 21]. As there are no other questionnaires that can be used as golden standard, concurrent and predictive criterion validity is not applicable [24]. To establish face validity, a focus group with seven enrolled nurses discussed the VBSQ at two occasions. They had considerable experience of VBS in different clinical settings, both in hospitals and in PHCs. In Sweden VBS are most often performed by enrolled nurses, therefore enrolled nurses match as a focus group. Their role was to identify any item that could possibly be misinterpreted. All gave consent to participate in the study.
Reliability
The test and establishment of VBSQ (Additional file 1) stability was conducted in the PHCs setting after the validity process. The head of the PHCs assisted in the distribution and gave permission to perform the survey. Thirty one enrolled nurses, registered nurses, and biomedical technicians of PHCs in two County Councils in northern Sweden were engaged in the test-retest study. Three participants did not answer the VBSQ in the follow up, thus 28 participants were included in the analyses and gave written informed consent to participate in the study. Of these, 64% performed VBS every day, 25% every week and 11% every month or more seldom. The median age was 52 years (Q1 = 44; Q3 = 58), and the median employment time was 7 years, (Q1 = 2; Q3 = 19).
Instrument
Overall the questionnaire was developed to assess practical tasks in accordance with national guidelines [14]. The individual questions were not related to each other i.e., exploratory factor analysis is not applicable. The VBSQ included background characteristics (6 questions), patient identification and collection of specimen (4 questions; 10 items), sample storage and seeking information (2 questions; 7 items) test request management and test-tube labeling (4 questions; 12 items) frequency of error reporting and suggestions (3 questions; 8 items), (Additional file 1). One question was answered yes/no, while the majority were answered by a four-point scale; Never, Seldom, Often, and Always. All "by other mean" items (n = 4) were excluded from the test-retest due to high internal missing rate but were retained in the questionnaire. A careful instruction containing an illustration on how to complete the questionnaire was located on the front page. It is to note that it was clearly pointed out that the respondents were to state how they usually performed VBS practices and not how they knew it was to be performed.
Procedures and data analysis
Validity
Content validation differs from other forms of validity testing. It is not based on the scores from the questionnaire but from discussions between experts whether the questionnaire instrument appears logical to this group of experts [24]. The development and analysis of questions and items of the VBSQ was followed and transcribed by members in the research group. During the questionnaire development, in focus group and experts discussions, extensive efforts were made to ensure that each item was clearly formulated, easy to understand, and could not be misinterpreted. The focus group of enrolled nurses read, answered the questions, and discussed the questions and the items on the VBSQ. The project member wrote down their opinions and amended the questionnaire with help of other project members including instructors from the laboratory. It was also ensured that the questions and items still were according to the guidelines. Thereafter the focus group members reconsidered the questionnaire and had a renewed reflected discussion with project members.
Reliability
The VBSQ was distributed at two different occasions with 3-4 weeks between the test and retest. Data was analyzed by using three measures, Spearman's rank correlation (rs), Kappa coefficient (K) and percentage agreement (%A) in the persons answer. Non-symmetrical K values were manually counted by the author [28]. The items were accepted if they passed at least one of two set criteria [22, 29];
• Criterion one: K ≥ 0.61 = good or rs ≥ 0.7 or %A ≥ 90%
• Criterion two: K ≥ 0.51 = moderate and rs ≥ 0.6 or K ≥ 0.51 = moderate and %A ≥ 80%
Criterion two complemented criterion one in those cases where a combination of the used measures were judged to be acceptable for acceptance of the items. SPSS 18.0 for Windows (SPSS Inc., Chicago, IL) was used for analysis.
Ethics
This study was approved by the Regional Ethical Review Board in Umeå (Dnr 06-104 M). The VBSQ were kept in a locked space and all questionnaires were decoded. Only the researchers had access to the codes and the corresponding names. The participants were assured confidentiality by the information letter and also that they could withdraw from the survey at any time without declaring any reason. They were also informed that data would be presented at group level only.