- Jul 2018
-
europepmc.org europepmc.org
-
On date unavailable, commented:
None
This comment, imported by Hypothesis from PubMed Commons, is licensed under CC BY. -
On 2017 Aug 01, Daniel Quintana commented:
We thank Dr. Grossman for his comments on our manuscript. We are cognisant of the study’s limitations, which we highlighted in our original manuscript. Despite these limitations, we believe our conclusions are still valid but also understand that Dr. Grossman may not agree with our interpretation.
We will now address Dr. Grossman’s two comments in turn, which we have reprinted for clarity:
- Comment 1 from Dr. Grossman: The authors report respiration frequency as the peak frequency in a band range between 0.15-0.40 Hz. Resting respiration rate (i.e. frequency), however, is rarely a constant phenomenon for most people within a resting period of several minutes: some breaths are longer, some are shorter, and the peak frequency does not necessarily reflect average breathing rate; in fact, there are very likely to be different peaks, and only the highest peak would have been used to estimate (or misestimate) average respiratory frequency. Spectral frequency analysis, therefore, is a highly imprecise method to calculate mean breathing frequency (perhaps, the difference in relations found between mentally ill vs. healthy people were merely due to increased variability of respiratory frequency among the ill individuals; see Fig. 1F). In any case, this may be be sufficient to disqualify the main conclusions of the study.
Response: We recognize that spectral frequency may not be an optimal method to calculate mean respiration frequency given the intraindividual variation of respiration rate. However, Levene's Test of Equality of Variances shows that the variances in the clinical and healthy groups are not significantly different [F(1,202) = 1.6, p = 0.21], which suggests that we cannot discard our null hypothesis that the group variances are equal. Thus, variability in mean respiration rates between the groups are unlikely to have confounded our results.
- Comment 2 from Dr. Grossman: However, there is may be even a more serious problem that invalidates the conclusions of this investigation. As already mentioned, the authors examined respiration frequencies only between 0.15-0.40 Hz; this corresponds to a range between 9 and 24 breaths/minute. Already in 1992, we demonstrated among a group of healthy participants that a sizable proportion of participants manifest substantial proportions of resting breathing cycles below 9 cycles per minute: among 16 healthy individuals carefully assessed for respiration rate during a 10-minute resting period, we found that half of the participants showed 1/5 of their total breathing cycles to be slower than 9 cycles/minute (cpm); over 60% of participants showed >10% of their cycles to be slower than 9 cpm (also very likely thqat a substantial proportion of breaths occurred beyond 24 cpm). Thus, accurate estimation of mean resting respiration frequency is also seriously compromised by the insufficient range of frequencies included in the analysis. See Grossman (1992, Fig. 5): Grossman, P. Biological Psychology 34 (1992) 131 -161
Response: To confirm that mean respiration the frequency was not missed or misattributed to non-respiratory frequencies, we re-analysed the data including participants that we originally excluded as they fell outside the 0.15-0.4 Hz range. In total, 2 participants (both from the patient group) had a mean respiratory frequency < 0.15Hz and 4 participants (3 from the patient group and 1 from the healthy control group) had a mean respiratory frequency > 0.4Hz. We also re-analysed absolute high frequency HRV and adjusted the frequency bands accordingly (0.1Hz – 0.4Hz for the 2 participants with a lower than average respiratory rate and, 0.15Hz-0.5Hz for the 4 participants with a higher than average respiratory rate).
We found that including these participants did not change the overall conclusions of the study. While we reported an estimated correlation (p) of −0.29 between HF-HRV and respiration in the patient group [95% CI (−0.53, −0.03)], our updated analysis demonstrated a slightly stronger estimated association -0.47 [95% CI (−0.66, −0.26)]. For the healthy controls, we originally reported an estimated correlation (p) of −0.04 between HF-HRV and respiration in the patient group [95% CI (−0.21, 0.12)]. Our updated analyses demonstrated a close to equivalent estimated association of -0.04 [95% CI (−0.20, −0.12)]. We also originally reported that computing the posterior difference of p between these two tests revealed a 94.1% probability that p was more negative in the clinical group compared to the control group. Running this analysis modestly increased this probability to 99.7%.
Daniel S. Quintana (on behalf of study co-authors)
This comment, imported by Hypothesis from PubMed Commons, is licensed under CC BY. -
On 2017 Apr 25, Paul Grossman commented:
Quintana et al. (2016) suggest that individual difference in respiration rate is only correlated with high-frequency heart-rate variability (HF-HRV), i.e. respiratory sinus arrhythmia (RSA) among seriously mentally ill people, but not among healthy individuals. The data presented has several methodological problems that seem very likely to severely compromise the authors' conclusions:
The authors report respiration frequency as the peak frequency in a band range between 0.15-0.40 Hz. Resting respiration rate (i.e. frequency), however, is rarely a constant phenomenon for most people within a resting period of several minutes: some breaths are longer, some are shorter, and the peak frequency does not necessarily reflect average breathing rate; in fact, there are very likely to be different peaks, and only the highest peak would have been used to estimate (or misestimate) average respiratory frequency. Spectral frequency analysis, therefore, is a highly imprecise method to calculate mean breathing frequency (perhaps, the difference in relations found between mentally ill vs. healthy people were merely due to increased variability of respiratory frequency among the ill individuals; see Fig. 1F). In any case, this may be be sufficient to disqualify the main conclusions of the study.
However, there is may be even a more serious problem that invalidates the conclusions of this investigation. As already mentioned, the authors examined respiration frequencies only between 0.15-0.40 Hz; this corresponds to a range between 9 and 24 breaths/minute. Already in 1992, we demonstrated among a group of healthy participants that a sizable proportion of participants manifest substantial proportions of resting breathing cycles below 9 cycles per minute: among 16 healthy individuals carefully assessed for respiration rate during a 10-minute resting period, we found that half of the participants showed 1/5 of their total breathing cycles to be slower than 9 cycles/minute (cpm); over 60% of participants showed >10% of their cycles to be slower than 9 cpm (also very likely thqat a substantial proportion of breaths occurred beyond 24 cpm). Thus, accurate estimation of mean resting respiration frequency is also seriously compromised by the insufficient range of frequencies included in the analysis. See Grossman (1992, Fig. 5): Grossman, P. Biological Psychology 34 (1992) 131 -161
It is also unfortunate that the authors merely cited a single investigation that unusually showed no relation between individual differences in respiration frequency and RSA magnitude (i.e. Denver et al., 2007), but none of the many studies that have found correlations in the range of r's= 0.3-0.5; e.g. https://www.researchgate.net/publication/279615441_Respiratory_Sinus_Arrhythmia_and_Parasympathetic_Cardiac_Control_Some_Basic_Issues_Concerning_Quantification_Applications_and_Implications
http://journal.frontiersin.org/article/10.3389/fphys.2016.00356/Fülle
https://pdfs.semanticscholar.org/6e44/e75dd2061a43cc69a4354171540e8a98e6a5.pdf
The Denver et al. study, additionally, used the same methods inaccurately to calculate respiration rate.
Paul Grossman Pgrossman0@gmail.com
This comment, imported by Hypothesis from PubMed Commons, is licensed under CC BY. -
On 2017 Apr 24, Paul Grossman commented:
None
This comment, imported by Hypothesis from PubMed Commons, is licensed under CC BY.
-
- Feb 2018
-
europepmc.org europepmc.org
-
On 2017 Apr 24, Paul Grossman commented:
None
This comment, imported by Hypothesis from PubMed Commons, is licensed under CC BY. -
On 2017 Apr 25, Paul Grossman commented:
Quintana et al. (2016) suggest that individual difference in respiration rate is only correlated with high-frequency heart-rate variability (HF-HRV), i.e. respiratory sinus arrhythmia (RSA) among seriously mentally ill people, but not among healthy individuals. The data presented has several methodological problems that seem very likely to severely compromise the authors' conclusions:
The authors report respiration frequency as the peak frequency in a band range between 0.15-0.40 Hz. Resting respiration rate (i.e. frequency), however, is rarely a constant phenomenon for most people within a resting period of several minutes: some breaths are longer, some are shorter, and the peak frequency does not necessarily reflect average breathing rate; in fact, there are very likely to be different peaks, and only the highest peak would have been used to estimate (or misestimate) average respiratory frequency. Spectral frequency analysis, therefore, is a highly imprecise method to calculate mean breathing frequency (perhaps, the difference in relations found between mentally ill vs. healthy people were merely due to increased variability of respiratory frequency among the ill individuals; see Fig. 1F). In any case, this may be be sufficient to disqualify the main conclusions of the study.
However, there is may be even a more serious problem that invalidates the conclusions of this investigation. As already mentioned, the authors examined respiration frequencies only between 0.15-0.40 Hz; this corresponds to a range between 9 and 24 breaths/minute. Already in 1992, we demonstrated among a group of healthy participants that a sizable proportion of participants manifest substantial proportions of resting breathing cycles below 9 cycles per minute: among 16 healthy individuals carefully assessed for respiration rate during a 10-minute resting period, we found that half of the participants showed 1/5 of their total breathing cycles to be slower than 9 cycles/minute (cpm); over 60% of participants showed >10% of their cycles to be slower than 9 cpm (also very likely thqat a substantial proportion of breaths occurred beyond 24 cpm). Thus, accurate estimation of mean resting respiration frequency is also seriously compromised by the insufficient range of frequencies included in the analysis. See Grossman (1992, Fig. 5): Grossman, P. Biological Psychology 34 (1992) 131 -161
It is also unfortunate that the authors merely cited a single investigation that unusually showed no relation between individual differences in respiration frequency and RSA magnitude (i.e. Denver et al., 2007), but none of the many studies that have found correlations in the range of r's= 0.3-0.5; e.g. https://www.researchgate.net/publication/279615441_Respiratory_Sinus_Arrhythmia_and_Parasympathetic_Cardiac_Control_Some_Basic_Issues_Concerning_Quantification_Applications_and_Implications
http://journal.frontiersin.org/article/10.3389/fphys.2016.00356/Fülle
https://pdfs.semanticscholar.org/6e44/e75dd2061a43cc69a4354171540e8a98e6a5.pdf
The Denver et al. study, additionally, used the same methods inaccurately to calculate respiration rate.
Paul Grossman Pgrossman0@gmail.com
This comment, imported by Hypothesis from PubMed Commons, is licensed under CC BY.
-