Replications of Published Findings: PCS1
In order to confirm that the data sets that we provide as part of the Common Cold Project are representative of the published findings, we have conducted replication analyses of four of the publications that derived from PCS1 data:
- Cohen, S., Doyle, W. J., Skoner, D. P., Rabin, B. S., & Gwaltney, J. M., Jr. (1997). Social ties and susceptibility to the common cold. Journal of the American Medical Association, 277, 1940-1944. (variable list 1)
- Feldman, P., Cohen, S., Doyle, W. J., Skoner, D. P., & Gwaltney, J. M. (1999). The impact of personality on the reporting of unfounded symptoms and illness. J Pers Soc Psychol, 77, 370-378. (variable list 2)
- Miller, G. E., Cohen, S., Rabin, B. S., Skoner, D. P., & Doyle, W. J. Personality and tonic cardiovascular, neuroendocrine, and immune function. (1999). Brain Behav Immun, 13, 109-123. (variable list 3)
- Cohen, S., Frank, E., Doyle, W.J., Skoner, D. P., Rabin, B. S., & Gwaltney, J. M., Jr. (1998). Types of stressors that increase susceptibility to the common cold in healthy adults. Health Psychology. 17, 214-223. (variable list 4)
For additional information on the data set variables that were used in each analysis, click on the links indicated above.
Inconsistencies Between Data Codes Used in PCS1 Publications and in the Data Set
Objective Colds
The rate of objective colds reported in PCS1 publications is 39.5% (109/276). Using the data contained in the PCS1 Data Set, the frequency of objective colds is slightly higher (40.4% or 111/275). The discrepancy results from modifications made when we created the aggregate data set to (a) the methods we use to compute total adjusted mucus weights and average adjusted nasal clearance times, respectively; and (b) our approach to assigning objective cold status to subjects who are missing pre-challenge cold sign data.
Aggregated mucus weights and nasal clearance times. Prior to computing post-challenge summary scores, daily values are adjusted for baseline by subtracting the pre-challenge value of the relevant variable. Because negative adjusted scores cannot be interpreted, values less than zero are recoded to zero. The previous method for computing post-challenge summary scores differed from the current method in terms of the computational step at which the recoding occurred (see Table 1). Specifically, the former method recoded negative values after scores were aggregated across the five post challenge days (Table 1, top panel), whereas the current method recodes negative values before data are aggregated (Table 1, bottom panel). The latter method improves upon the former because the creation of interpretable daily values allows for analyses to be performed both within- and between-persons.
Table 1. Methods used to compute adjusted summary scores for objective cold sign variables.
| Previous Method |
Summary Score = (day1–day0)+(day2–day0)+(day3–day0)+(day4–day0)+(day5–day0) If score < 0, score = 0 |
| Current Method |
For all post-challenge days: dayX_adj = (dayX – day0) If dayX_adj < 0, dayX_adj = 0 Summary Score = (day1_adj) + (day1_adj) + (day1_adj) + (day1_adj) + (day1_adj) |
Missing pre-challenge data. As described elsewhere, participants are determined to have developed an objective cold if they are both infected and meet criterion level on either of the two objective illness markers: total adjusted mucus weight and average adjusted nasal clearance time. One infected participant in PCS1 (ID=106050) was missing pre-challenge mucus production data. Accordingly the total adjusted mucus weight score could not be computed. To maximize the sample size, cold status initially was assigned to this participant based on the available data, with unadjusted mucus weights being substituted for the adjusted values. We now take a more conservative approach, whereby cold status is assigned to infected subjects with missing baseline only if sufficient information is available (see Table 2). Because the data configuration of participant 106050 matched that described in the last row of Table 2, no objective cold status was assigned to this participant (i.e., value for post.objcold is missing).
Table 2. Current method used to assign objective cold status to infected participants with missing data on either of the two objective illness markers.
| Objective Illness Markers |
Objective Cold? |
|
| Variable 1: Adjusted summary score (i.e., pre-challenge data available) | Variable 2: Unadjusted summary score (i.e., missing pre-challenge data) | |
| Meets criterion | Does not meet criterion | Yes |
| Meets criterion | Meets criterion | Yes |
| Does not meet criterion | Does not meet criterion | No |
| Does not meet criterion | Meets criterion | Cannot be determined |
Subjective Colds
The rate of subjective (Jackson criterion) colds reported in PCS1 publications is 40% (110/276). Using the data contained in the PCS1 Data Set, the frequency of subjective colds is higher (43% or 118/276). The discrepancy results from modifications being made to how total adjusted Jackson scores are computed. The issue is the same as that outlined in Table 1: previous and current methods differed on the computational step at which negative adjusted scores are recoded to zero.