Abstract

Background

During recent years there has been an intense world-wide debate on the inadequate patient safety in health care [1]. In Sweden alone, it has been estimated that annually approx. 100 000 patients suffer from injuries, nearly 10 000 receives lasting injuries and 3000 die due to mistakes in health care [2]. Studies in other countries show similar figures [1,3-5]. To note is that most injuries are caused by human errors [1-3]. It is therefore important to develop routines and systems to increase patient safety and promote cost effectiveness [2,3,6].

A number of studies suggest that most of the errors in laboratory medicine are linked to the preanalytical phase i.e. before the sample is analyzed in a laboratory. Analytical errors (within the laboratory) and post-analytical errors (reporting and interpretation of results) are less frequent [4,7-10].

Venous blood sampling (VBS) is complex and demands both knowledge and skill [11]. VBS is a frequent procedure in health care [8] and strictly regulated by similar national and international standards [11-14]. The majority of errors in VBS are due to human mistakes which in several cases cause patient harm [4,7]. Errors that can occur during VBS and handling are numerous, and include inaccurate patient and sample identification, inadequately filled tubes [7,9,11], specimen haemolysis [8,9,11,15], and improper handling of anticoagulants [9,11,12]. Standards that refer to best preanalytical practice [11-14] aim to reduce or alleviate many of these errors. In Sweden VBS is usually decentralized and performed by enrolled nurses, registered nurses and biomedical technicians [14]. The VBS standards are not always followed and interventions are needed to ensure patient safety and reduce risk due to VBS errors [15-21]. Our previous studies have investigated practices of VBS staff in hospital wards and primary health care centers (PHCs) by means of a self reported venous blood sampling questionnaire (VBSQ) [16-21]. The initial VBSQ was developed, further modified and used in several studies [16-21].

Questionnaire surveys have several benefits as they are practical to handle, self-administered, economical and give the respondents anonymity. It is also easy to reach a large study group in a large geographic area [22]. When using questionnaires it is important to confirm validity (i.e. how well an instrument measures what it is supposed to measure) [22-24] and reliability (i.e. stability) of the included items [22,23]. However, no validated VBSQ has previously been published. The aim was to test a recently developed questionnaire on self reported venous blood sampling practices for validity and reliability.

Materials and methods

Results

The process of establishing validity resulted in several modifications. The answer alternatives were reduced from five to four (the alternative sometimes was removed) and several questions was also specifically modified to suit the PHCs setting. Focus group meetings resulted in removal of a question about order of test tube collection. A few questions were rephrased and further modified. Each question and items was controlled to be easy to understand and clearly outlined. The number of included questions was judged to be possible to complete in a reasonable amount of time. The final VBSQ included 19 questions out of which 9 had in total 34 underlying items. Thereafter, content validity of the VBSQ was judged as acceptable by experts in VBS and questionnaire design and face validity was established by the focus group with vast experience. Criterion validity was applied as we followed guidelines. This is verified as laboratory staff performs more correct VBS according to guidelines.

When analyzing the VBSQ for reliability, the K and rs of the test-retest varied between high and low which indicated that some of the items needed to be modified or excluded. In total, 82% of the questions with underlying items fulfilled the reliability acceptance criteria (Table ​(Table1).1). The questions or items of criterion one was fulfilled by 71% and criterion two by 68%. Items that did not fulfill the acceptance criteria were 7c, 8b, 8c, 10a, 16a, 18e and 18f. VBSQ items 8b, 8c, and 16a were not stable but important to retain because of the items being interrelated with other underlying items.

Discussion

The low preanalytical error rates noted by the individual laboratory calls for large databases and appropriate techniques for the detection of errors and their consequent reduction [7]. Comparisons of error rates and the effect of interventions have hitherto been possible (although rarely performed) only between laboratories and not between individual hospital wards or PHCs. Most preanalytical procedures that increase the risks for or lead to adverse effects that jeopardize patient safety are performed locally at PHCs and hospital wards [30]. Detailed analyses of occurring errors and risks, present an opportunity to work preventively [31]. To assess risk of near-misses would allow for general and directed corrective educational interventions and also permit comparison and benchmarking of preanalytical practices between wards and PHCs. Thus, focusing on improving the frequency of near misses would thus lead to better opportunities for quality improvement than mere focus on assessment of underreported incidents or registered rare adverse errors [32]. Methods to assess frequently occurring, error-prone everyday tasks at the hospital ward/PHC level are therefore urgently needed. Questionnaire surveys have as far as we know not previously been used to assess how VBS staff follows recommended preanalytical guideline practices in order to improve patient safety. However, this study will give researchers and practitioners a trustworthy VBSQ to survey and/or evaluate VBS practices.

We developed and used the questionnaire to assess practical performance of VBS of health care personnel [16-21]. Generally, questionnaire surveys are cost effective and easy to coordinate [22] and in our studies we were able to survey almost all VBS personnel distributed throughout the County Council.

In this study, we found that the VBSQ had content validity (determined what it was intended to assess) as both content validity including face validity and criterion validity were judged to be reasonable.

VBSQ were based on standards for VBS and our studies demonstrate that laboratory staff can be used as benchmark in comparison of preanalytical practices [16-18,20,21]. VBS by laboratory staff results in fewer haemolyzed samples than VBS by non laboratory staff [25]. Results from previous surveys showed better performance by the laboratory staff indicate some degree of Criterion validity. Construct validity was not tested since this kind of validity was judged not to be applicable to the investigated questionnaire.

The VBSQ stability determined by the reliability coefficient of the test-retest varied and indicated that some items had to be removed or reformulated. Overall, the stability in the questionnaire was acceptable as 71% of the items fulfilled the acceptance criterion one and 68% criterion two, and 82% fulfilled either of the criteria. However, answers in questionnaires can be affected by memory interference and by changes in knowledge and behavior over time, which influence the stability (test-retest) of the measure. The recommended interval between test and retest is 2 days to 2 weeks [24]. We used an interval of 3-4 weeks so higher stability could have been reached with shorter interval. It is to note that the stability of the items were high considering that we surveyed how VBS was actually performed and not knowledge of how it should be performed according to guidelines. A weakness of using the K for stability assessment is that it takes no account to the degree of disagreement [29]. Thus, complementing with percentage agreement in the persons answer when K is low and rs high is an aid in the stability acceptance decision. For example, in item 12a (low K, high rs and high percentage) and for 15a with (low K, low rs and high percentage) actually only two respondents changed their answers. Items 7c, 18e, 18f will be removed from future VBSQ surveys because of lack of stability. Item 10a will be reformulated due to unexpected dual interpretations by surveyed VBS staff.

The most stable questions were those where VBS were not situation-dependent and the answer therefore expected to reflect the actual practices (example question 12). In contrast patient-related questions such as question 8 when to remove stasis are dependent of the VBS situation and less stable. Items 18e and 18f covering error reporting practices had surprising low rs,%A and K. The loss of respondents was 25%. A few responders wrote that error reports were not an ordinary work tasks and were unable to respond. The VBS personnel are obliged to follow the Swedish National Board of Health and Welfare constitutions [33] which are the basis for the Handbook for health care [14] and for local directives [13]. Also, every staff member has a duty to write error reports [33], therefore question 18 is important to retain but to reformulate.

This VBSQ (Additional file 1) was developed, managed and validated in the Swedish health care context. In other health care settings, with differences such as language, culture and working conditions, necessitates renewal of validity and stability testing of the VBSQ. In particular, performing stability testing with test-retest has an important bearing on questionnaire development prior to surveys and follow-up of interventions.

According to the WHO, a patient safety incident is an event or circumstance that could have resulted, or did result, in unnecessary harm to a patient [34]. Also, ISO technical specification stresses the need of a recurrent review of laboratory non-conformities, errors and incidents, as well as improving the quality of laboratory services and patient safety. The use of reliable quality indicators that effectively evaluate the quality of the steps of the preanalytical phase can thus drive improvement programs for better laboratory services and patient safety [10].

Conclusions

To conclude, the questionnaire has acceptable validity and reliability and could be used for assessment of "near miss" practices that could jeopardize patient safety. After removal of non-stable questions the questionnaire finally contained 19 questions out of which 9 had in total 34 underlying items. The retained non-stable items 8b, 8c, 10a, 16a should be reformulated and thereafter re-validated. A reliable questionnaire to assess "near miss" practices offers several benefits over assessing rare adverse events only. Iterated questionnaire surveys of hospital wards and PHCs VBS staff practices would high-light specific problems and make it possible to follow the effect of corrective actions. Such information is of outmost importance in order to eliminate "near miss" events by error prevention.

Abbreviations

VBS: Venous blood sampling; VBSQ: Venous blood sampling questionnaire; PHC: Primary health care centre.

Competing interests

The authors declare that they have no competing interests.

Authors' contributions

KB, JS, CB and KG proposed the original idea for the study and developed and modified the questionnaire, initially designed also by Olof Wallin. KB, CB and ML executed the statistical analyses. KB drafted the manuscript and all the authors read the manuscript critically for important intellectual content and approved the final manuscript.

Supplementary Material

References