A Markov Decision Model to Guide Treatment of Recurrent Colonic Diverticulitis

Although colonic diverticulitis is a common disorder, there is no clear treatment strategy for patients with recurrent episodes of diverticulitis. We investigated whether colonic resection or conservative or medical treatments have the greatest effects on quality-adjusted life-years (QALYs).

A Markov model simulating patients with 2 episodes of non-surgically treated diverticulitis was used to simulate all relevant outcomes of each treatment strategy. A 1-year cycle length with 10-year follow-up period was used to allow for chance of recurrent diverticulitis. Primary outcome was QALYs gained from each strategy. Factors considered were morbidity, mortality, chance of colostomy formation, risk of recurrence, and persistence of abdominal pain. The probabilities of clinical events were determined by using the best available published data.

A strategy in which colonic resection was performed after 2 episodes of diverticulitis was associated with the lowest quality-adjusted survival (a gain of 8.66 QALYs) and highest chance of stoma formation (1.1%) but the lowest chance of a mild (3.5%) or severe (1.1%) recurrence. The strategies of colonic resection or conservative or medical treatment after the third episode of diverticulitis were comparable in terms of quality-adjusted survival, providing 8.78, 8.76, and 8.74 QALYs, respectively. Probabilistic sensitivity analysis did not change these results. Persistent abdominal complaints were lowest in the medical treatment strategy.

Elective surgery after 2 episodes of diverticulitis should be questioned in terms of QALYs. After the third episode of diverticulitis, surgical or conservative or medical treatments provide similar QALYs, but rates of abdominal symptoms are lower with the medical treatment strategy. This Markov decision model has limitations when the individual patient and physician face a complex decision weighing early and long-term risks and benefits of elective surgery or conservative management.