This experimental study shows that self-samples and investigator-samples are very similar in testing the presence of S. aureus in nares and pharynxes. For nasal samples, agreement is 93% with a kappa coefficient of 0.85, indicating excellent agreement, for pharyngeal samples agreement is 83% and the kappa coefficient is 0.60, indicating good agreement. This means that self-samples are appropriate for detection of S. aureus and MRSA.
Furthermore, when looking at the discordant pairs, nasal self-samples tend to yield more S. aureus compared to investigator-samples. A rational explanation might be that swabbing is more thoroughly done by persons themselves compared to swabbing by investigators, who press less hard, resulting in fewer microorganisms picked up by the swab and a lower prevalence of S. aureus. For pharyngeal swabs the same can be stated, but results are less clear. This is probably due to the fact that taking swabs from tonsils or tonsillar arches is more complex, compared to swabs from the nares. In addition, the investigators observed that the instructions for pharyngeal samples were less well adhered to than the instructions for nasal samples.
This strengthens the conclusions from Lautenbach et al., although agreement percentages are slightly different (82% for nares and 91% for throat) . Moreover, using the information given in this reference, kappa coefficients can be calculated, which are different from the coefficients found in this study (kappa coefficients of 0.28, CI −0.05-0.60 for nares and 0.80, CI 0.64-0.97 for throat). Possible explanations for these discrepancies can be the different populations studied (US inpatients versus healthy subjects from the Netherlands) or different methods used (strict or random sampling order, type of instructions). Hanselman et al. have used nasal self-sampling, and found S. aureus percentages in US teachers consistent with literature, indicating appropriateness of self-sampling . More studies demonstrate the usefulness of self-samples, albeit from other anatomic sites and for other microorganisms, as human papilloma virus, group-B streptococci, respiratory viruses, and sexually transmitted diseases [7, 13–15].
The validity of nasal and pharyngeal self-sampling cannot be established entirely correctly, as there is no true gold standard. However, when comparing to combined self-samples and investigator-samples per site, nasal self-samples have a sensitivity of 97% (CI 86-100%), and pharyngeal self-samples have a sensitivity of 88% (CI 75-95%). Lautenbach et al. have calculated sensitivities based upon a gold standard of combined samples from nares, throat, axillae, groin and perineum, and found sensitivities of 91% (CI 80-97%) for nasal self-samples, and 67% (CI 53-79%) for throat self-samples .
This study has three limitations. The nursing personnel members are expected to have more prior knowledge of sampling methods compared to the standard outhospital person. This might lead to an overestimation of the adequacy and validity of self-sampling. Moreover, false sampling was not checked for by, for example, placing the swab on a standard sheep blood agar, which detects whether any viable microorganisms are present on the swab. However, the investigators were standing next to the subject when sampling took place, making false sampling unlikely. In addition, S. aureus prevalence corresponds to literature, confirming adequate sampling . Checking for false sampling with blood agar probably is a useful suggestion when using self-samples in a research situation, however. Lastly, three investigators were involved in taking the investigator-samples, which might lead to slightly different sampling techniques. As these investigators are all well trained on sample taking and sampling techniques were discussed beforehand, we believe the effect of this variation is negligible. Data on who took which samples was unfortunately not recorded, giving no opportunity to verify this statement.