We compared the efficacy to collect specimen for viral detection between fNS and NPA. According to Landis and Koch's  scale there was a substantial agreement between the methods, to NPA's favor, and NPA was significantly more efficient collecting epithelial material approximated by quantification of ACTB. Even if removal of the cellular component of NPA does not interfere with quantification of some respiratory viruses , a higher cell count might indicate a higher yield of specimen. Even though not statistically significant, the Ct value negatively correlated to the number of cells collected, and thus, the finding of NPA being superior to the fNS in collecting cells invite us to speculate that the fNS sometimes may collect inadequate amount of specimen in order to reach the PCR method's detection limit. Furthermore, several reports have suggested that sensitivity increases when the nasopharynx is sampled instead of the nasal cavity [2, 5, 11, 14].
Thus, the difference in sensitivity for the fNS and NPA could have several reasons; biological and technical. One obvious biological reason is the fact that two different anatomical sites are investigated; one technical explanation could be that suction is superior scratching for collecting adequate amount material. The production of mucus that transports the viral nucleic acids in an anterior direction to the fNS's sampling site may be an important parameter. In this patient category, the lack of a competent immune response could reduce the mucus-production. Taking the above into account, the depth of the sampling with the fNS probably plays an important role. However, a deeper penetration could cause even more discomfort than does the NPA.
In contrast to our findings, several groups have reached good sensitivity using the fNS in the outer part of the nose and concluded it comparable to NPA [1, 4, 9]. Sung et al who used an insertion depth of only 1-1.5 cm summarized the fNS an even better option when using PCR.
This study has several limitations. The time interval between fever onset and sampling varied between the patients, and they did not necessarily have respiratory tract symptoms. Furthermore, the patients suffered from different underlying diseases. However, this reflects the clinical reality and as both sampling methods were used at the same time point in the same patient, the comparison is still valid. The foremost important limitation is the low number of positive samples on which the calculation of sensitivity is based. However, the upper limit in the 95% confidence interval for the overall sensitivity was 85% which is still a questionable value for replacing an established method.
In summary, the NPA was superior to the fNS in collecting epithelial cells. Furthermore, we also found lower sensitivity for the fNS than for NPA in detecting respiratory viruses in these immunocompromised adults. For detection of rhinovirus it was slightly better, but still not appropriate, according to our judgment, for clinical use in this patient category. With an insertion depth comparable to what other groups have used we found lower sensitivity for the fNS than previous investigators. Less mucus production and consequently less anterior transport of viral nucleic acid in immunocompromised individuals may explain this finding.