Response to Fung et al

We appreciate Dr. Fung and colleagues’ interest in our study.(1) We agree that it is important to examine the role of fiber in the prevention of diverticular bleeding. We did not include such an analysis in the present study since the pathogenesis of diverticulitis and diverticular bleeding likely differs, suggesting distinct etiologies. For example, in an analysis of data from the US Nationwide Inpatient Sample, the prevalence of hospitalizations for diverticulitis and diverticular bleeding showed opposite trends from 2000 through 2010. There were also notable differences according to race and gender.(2) In our cohort, we also found that, in contrast with diverticulitis, fiber was not associated with diverticular bleeding. Comparing women in the highest quintile of dietary intake with those in the lowest quintile, the multivariable-adjusted hazard ratios for diverticular bleeding were 1.09 (95% CI, 0.81–1.46; P-trend=0.46) for total fiber; 1.03 (95% CI, 0.77–1.38; P-trend=0.57) for cereal fiber; 1.11 (95% CI, 0.83–1.47; P-trend=0.63) for fruit fiber; and 1.27 (95% CI, 0.96–1.68; P-trend=0.11) for vegetable fiber. These results highlight the importance of separate investigations of risk factors for diverticulitis and diverticular bleeding.

As a well-accepted approach in nutritional epidemiology, we categorized dietary intake into quantiles,(3, 4) since quantiles are data-driven and not influenced by cut-points selected by the investigator which may maximize statistical significance. Quantiles are also free from assumptions about the dose-response relationship, allowing the reader to visualize whether the association is most compatible with a linear or nonlinear trend. We used quintiles rather than tertiles since the large number of cases in our study facilitated comparisons between finer and more extreme categories of intake. Indeed, although we observed a significant linear trend for total, cereal, and fruit fiber, the hazard ratios were only significantly different in comparing the highest to the lowest quintile. This suggests a potential threshold effect for intake of fiber for achieving benefit.

The instrument administered in our cohort specifically ascertained diverticulitis based on sequentially administered questions designed to distinguish patients with diverticulosis from those with complications of diverticulosis, including diverticulitis and diverticular bleeding. We previously validated these self-reports through medical record review. Nonetheless, we acknowledge the challenge of identifying those with asymptomatic diverticulosis within a large, prospective cohort, since this diagnosis necessitates an endoscopy or imaging study. In several studies based only on individuals who underwent colonoscopy, fiber intake was not significantly associated with uncomplicated diverticulosis.(5, 6) These findings suggest that fiber may not be important in the initial development of diverticulosis, but may play a role in preventing diverticula from becoming infected or inflamed.

We agree that the potentially varying effects of dietary fiber according to food sources are intriguing. We found that a higher whole fruit intake was associated with a reduced risk of diverticulitis, whereas fruit juice was not. This is consistent with most dietary guidelines(7) and findings from large cohort studies(8–10) which encourage higher consumption of fresh fruit but not juices for the prevention of diabetes and diabetic complications. Both in vitro and in vivo studies have shown that structural differences in dietary fiber may induce distinct effects on bacterial community and function.(11–13) Taken together with growing evidence that diet may influence diverticulitis risk through effects on gut microbiota and inflammation,(14) we share Fung et al’s opinion that further research is needed to better understand the mechanisms underlying the distinct effects of fiber from various food sources in diverticulitis prevention.