Study design, study population and self-sampling
This feasibility study used a convenience sample of study participants, recruited at the campus of the Helmholtz Centre for Infection Research in Brunswick, Germany. We invited 1096 individuals to participate through e-mail, flyers or during Campus meetings. Recruitment took place on selected days in January 2017; 180 individuals provided informed and written consent and were enrolled in the study. Inclusion criteria were: older than 18 years and German or English language skills. Individuals with known blood coagulation disorders (e.g. hemophilia), skin diseases on the middle or ring finger of the non-dominant hand, known immunosuppression as well as users of anticoagulant drugs were excluded. Participants received a package consisting of the HS device8 and sampling accessories including two lancets for fingerprick (Medipurpose SurgiLance, SLB250, 2.3 mm , 18 Gauge), study-specific instructions (S3 Appendix), study questionnaire (S4 Appendix).
The study documents were piloted before the study took place to ensure comprehensibility. For this, ten volunteers were interviewed by the study team with either the German or English study documents and asked to perform a demo sample.
A non-responder questionnaire was provided to those who denied participation (S5 Appendix).
All study participants used the device for self-sampling of capillary blood from a fingerprick. The then closed device was, together with the questionnaire and the instruction with the sampling date, returned anonymously via in-house mailing or return boxes on the campus. In addition, some study participants provided venous blood for sampling method comparison. To detect effects specific to the blood application onto the sampling device by study participants (user-error), the radial filter inserts HemaForm-80 Plate Kit (HF) used in the HS device were additionally inoculated with serum obtained through VBP by laboratory staff.
Self-sampling device
The HS cartridge is designed for the collection and transportation of blood8. The box contains an absorbent filter paper with the HF innovation of a fan-shaped construction with eight sheets, covered by an application surface with a small opening for liquids. A lancet with a prick in the finger is used to provide blood. After placing 2–3 drops of blood on the application surface, the cartridge can be closed and shipped. The desiccant dries the sample8.
Questionnaire on feasibility and acceptability
The questionnaire included demographic data such as sex/gender, age group, level of education and previous experience with either self-sampling or blood sampling in others. The status of education was categorized according to ISCED 201128, with level 3 representing the German “Abitur”. Items assessing user-friendliness included overall self-sampling experience, assessment of the level of difficulty, evaluation of pain perception, the preferred method of blood sampling and recommendation of study participation (S4 Appendix).
Laboratory analysis
As a use case, we determined antibody levels for anti-Clostridium tetani toxin IgG to assess the detectability and sensitivity of antibody levels compared to the gold standard blood sampling method. We further used a commercial assay from NovaTec29. Participants did not receive feedback on their laboratory results as the study was anonymized. Serum was obtained via centrifugation of whole blood in serum-specific S-Monovettes from Sarsted (Germany) according to the manufacturer’s protocol30. Aliquots were stored at -80 °C until further use. The radial filter inserts from the Spot-on-Science HF (F0020k, Spot-on-Science, USA) were inoculated with 80 µl of serum samples obtained as described above8. All extraction of dried spots was independent of the inoculation methods and was based on the manufacturer’s recommendation, with adjustments to improve reproducibility and yield. To analyze the HS samples, filter paper blades (including the middle part) are separated with sterile tweezers and scissors and transferred into a sterile Eppendorf tubes. We separated according to the following rule: two blades/tube if the filter paper was fully saturated (resulting in 4 aliquots), four blades/tube if the filter paper was half to 2/3 filled (resulting in two aliquots), and no separation if the filter paper was less than half saturated with blood (resulting in one aliquot). Subsequent extraction from the dried filter material was initialized by adding 250 μl of sterile-filtered PBS and gently shaking the tubes for 1 h at ambient temperature. The separation of re-hydrated samples in solution from the filter material was achieved by dry-spinning the prior bottom-punctured tubes inside another tube using a tabletop centrifuge at 16.000 × g for 1 min. Eluted filter extracts in the lower tube were transferred to a labelled roll-edge tube for storage at -80 °C until further usage. Antibodies against Clostridium tetani toxin were determined using NovaTec’s Clostridium tetani toxin 5S IgG plus – ELISA (PTETG043, NovaTec, Germany). All experiments were done according to the kit manufacturer’s protocol using the Tecan Nanoquant spectrophotometer, extracted filter material and serum from whole blood. The assay is calibrated against the World Health Organization International Standard for Tetanus Immunoglobulin (National Institute for Biological Standards and Control code: TE-3), using five standard concentrations: 0.0, 0.1, 0.5, 1.0, and 5.0 international units per milliliter (IU/mL). Linearity has been demonstrated across this range. Precision is acceptable, with intra-assay coefficient of variation (CV) ranging from 3.46% to 5.34% and inter-assay CV from 9.62% to 13.99%. The diagnostic sensitivity of the assay is 99.22% (95% CI 95.76–99.98%), and diagnostic specificity is 100% (95% CI 76.84–100%). Interpretation of results was based on defined cut-off values, where concentrations below 0.01 IU/mL indicate no protection, 0.01 to 0.10 IU/mL suggest no reliable immunity, 0.11 to 0.50 IU/mL indicate reliable protection with booster recommendation, 0.51 to 1.0 IU/mL suggest reliable protection with follow-up in two years, and values above 1.0 IU/mL indicate long-term protection with retesting recommended after 5 to 10 years29.
Definitions and statistical analysis
Questionnaires were entered into Epi Info™ by two project team members independently and checked for consistency31. Discrepancies were corrected in the data set in agreement with the study management. Missing or unspecified values have been reported in the tables and labelled as such. To describe group differences in the feasibility and acceptance questions of self-sampling, we used significance tests with an alpha level of 0.05. Chi-squared test was used when frequencies were 5 or greater, and Fisher’s Exact test was used for lower frequencies. The sensitivity of the detection of tetanus was calculated from participants who provided blood samples both obtained through self-sampling and VBP.
The Net Promoter Score was calculated for assessing the degree of recommendation of study participation according to Reichheld et al.32. Participants were asked whether they would recommend acquaintances to participate in a study with self-sampling of blood. The score was calculated from the percentage of those who recommend participation (promoters) minus those who do not recommend participation in such a study (detractors). A higher value indicates a stronger recommendation. A Net Promoter Score of > 0 is good, a score of > 50 is excellent and a score of negative values is interpreted as “not recommended”33.
Further, we calculated sensitivity of HS compared to VBP and HF compared to VBP. Pearson correlation coefficient was calculated for the comparison of HS versus VBP and HF versus VBP based on anti-Clostridium tetani toxin IgG concentration (IU/mL). In addition, Bland–Altman plots, as a graphical representation of the relationships between the different sampling strategies, were used. Introduced by Altman and Bland in 1983, Bland–Altman plots provide a method for assessing the agreement between two quantitative measurement methods that evaluate the same parameter. The plot displays the difference between paired measurements on the y-axis versus their average on the x-axis. The plot includes the mean difference (bias) and the limits of agreement, which are defined as the mean difference plus or minus 1.96 times the standard deviation of the differences. This visualization is useful for detecting systematic bias; ideal results show differences close to zero along the y-axis34,35. As sensitivity analysis, we also compared antibody concentration in HS compared to HF. We used R [version 4.3.2.] for all statistical analyses36. The R script can be retrieved from elsewhere (DOI: https://doi.org/10.5281/zenodo.14592165). We used OpenAI’s ChatGPT v2 to review and refine the R code for improved structure and functionality37. For clarity and precision, the English language parts of this manuscript have been edited using DeepL Pro and DeepL Write38. The authors checked the output for correctness and took responsibility for it.
Ethics and data protection
The study was reviewed and approved by the ethics committee of Hannover Medical School by June 21, 2016 (No: 3251–2016). A data protection concept was developed and approved by the internal data protection officer. The Federal Commissioner for Data Protection and Freedom of Information advised on the study. The study was performed in alignment with the Declaration of Helsinki [except study registration]. In addition, the study follows the recommendations for ensuring good epidemiological practice of the German Society for Epidemiology39.
