Rumination, Reflection and Distress: An 8-Month Prospective Study of Colon-Cancer Patients

RRQ, rumination: At T1, 8 participants had missing data with an average of 1.38 missing items (range 1–2). RRQ, reflection: At T1, 5 participants had missing data with 2 participants having all items missing (and thus were not included in analyses) and the remaining 3 participants all having 1 missing item. At T2, no participants had missing data for the rumination or the reflection subscale of the RRQ. IES, intrusion subscale: At T1, 5 participants had missing data for the IES intrusion subscale with an average of 4 missing items (range 1–7; 3 participants were not included in analyses because of too many missing items). IES, avoidance subscale: At T1, 7 participants had missing data with an average of 4 missing items (range 1–8; 3 participants were not included in the analyses because of too many missing items). At T2, 1 participant had 1 missing item for the IES intrusion subscale and 1 participant had 1 missing data point for the IES avoidance subscale. BDI-II: At T1, 6 participants had missing data with 1 participant missing 11 items (and thus not included in analyses) and the remaining 5 participants missing between 1 and 6 items (average 2.8 items). At T2, 4 participants had missing data for the BDI-II with an average of 3.75 missing items (range 1–10). Thus, more replacement of missing values was needed for the T1 data.

Because the results for reflection were in disagreement with expectations, we examined the relation between reflection and an alternative measure of distress. As a part of the study participants described the event where they received the cancer diagnosis at both T1 and T2. We counted positive and negative affect words in the descriptions the participants made of the event where they received the diagnosis. We coded this as “0” for no mentioning of positive/negative affect words and “1” for the mentioning of positive/negative affect words. The agreement between the two independent co-raters was high (κ = 0.80). We analyzed these data in two regression analyses using reflection at T1 to predict mentioning of positive or negative affect words in T2 descriptions while controlling for mentioning of positive or negative affect words in T1 descriptions. For negative affect words, the overall model was significant (F(2, 48) = 8.84, p < 0.05). However, reflection did not reach significance, although the direction of the relationship indicated that more reflection was associated with higher likelihood of mentioning negative affect words (β = 0.16, p > 0.05). For positive affect words, the overall model was significant (F(2, 48) = 6.38, p < 0.05). Reflection was associated with less likelihood of mentioning positive affect words at T2, while controlling for T1 positive affect words (β = −0.26, p = 0.05). We believe that these additional analyses further support the idea that reflection is associated with reduced well-being over time in the present study.