Objective To determine the risk of developing complicated colonic diverticular disease (CCDD) with prior episodes of acute diverticulitis and determine the mortality of the spectrum of CCDD.
Design Population-based cohort study.
Setting Computerised records from the General Practice Research Database linked to Hospital Episode Statistics data from the UK.
Participants Patients and controls registered in the General Practice Research Database from 1990 to 2007.
Main outcome measures Mortality was calculated and Cox regression modelling used to provide adjusted HRs and 95% CI. Logistic regression was used to model the effect of prior acute diverticulitis on the development of complications.
Results 2950 patients (1872 (63.5%) female) had a diagnosis of CCDD (8739 controls). A total of 1042 (35.3%) patients died compared with 2062 (23.6%) controls. Most excess deaths occurred in the first year after the complication. Patients with a perforation/abscess had a 4.5-fold increase in 1-year mortality (HR 4.55, 95% CI 3.74 to 5.52) compared with the general population, whereas those with a fistula or stricture had a 2.5-fold increase in mortality (fistula HR 2.60, 95% CI 1.47 to 4.62; stricture HR 2.41, 95% CI 1.86 to 3.11). Although most patients (2133 (72.3%)) had suffered no prior episodes of acute diverticulitis, increasing episodes of acute diverticulitis were associated with an increased risk of developing a fistula (two or more prior episodes, OR 1.54 95%, CI 1.08 to 2.19), but there was no clear relationship with stricture or perforation/abscess.
Conclusions Although most patients have experienced no prior episodes of acute diverticulitis, fistula formation is preceded by bouts of inflammation. Excess 1-year mortality across the spectrum of CCDD compared with the general population is substantial.
- Diverticular disease
Statistics from Altmetric.com
Significance of this study
What is already known about this subject?
The role of acute diverticulitis in the development of the complications of diverticular disease is unclear.
Guidance on how to prevent complicated diverticular disease does not have a strong evidence base with respect to prior episodes of acute diverticulitis.
There are no population-based estimates of the mortality associated with diverticular stricture and fistula.
What are the new findings?
Most patients have no prior episodes of acute diverticulitis before the development of complicated colonic diverticular disease.
Similar to perforation, there is high excess 1-year mortality associated with a diagnosis of diverticular stricture and fistula.
How might it impact on clinical practice in the foreseeable future?
Our study gives little support to the strategy of prophylactic resection of the colon after two episodes of acute diverticulitis to prevent the development of complicated diverticular disease.
Complications other than perforation also confer a high excess mortality over the general population, and our estimates will guide both patients and practitioners in understanding the natural history of these diseases.
Diverticular disease affects two-thirds of people over the age of 80 years,1 and recent reports suggest that hospital admissions are increasing,2–4 along with operative interventions2 and the incidence of complications.5 6 There is therefore a significant disease burden in terms of morbidity, mortality and cost.5 7 A substantial amount of the disease burden is from the complications of diverticular disease, notably perforation, abscess, fistula and stricture.
Prior estimates of the mortality associated with the spectrum of complicated diverticular disease have been limited to national audit studies and retrospective case series.8–11 Although we have previously shown that most of the excess mortality after perforated diverticular disease is within the first year after diagnosis,5 there are currently few population-based estimates of the mortality associated with the spectrum of complicated diverticular disease.5 Prior studies in this area are small and collected only limited data on confounding factors and consist of secondary care cohorts. In addition, these studies were unable to assess risk factors for the development of complications.
Inflammation, either overt or subclinical, has been reported in resection specimens of patients with diverticular disease,12 13 along with a combination of muscular,14 neuronal and mucosal15 changes, which may be responsible for the development of symptomatic and complicated diverticular disease.16 The natural history of diverticular disease and the development of complications were first detailed by Parks17 over 40 years ago, and subsequently only a few retrospective studies from secondary or tertiary centres have attempted to ascribe risk factors for developing complications.11 18–20 A particular focus of this work has been the role played by inflammation in the form of prior episodes of acute diverticulitis. It has been suggested that early resection after two prior episodes of acute diverticulitis should be used to reduce the morbidity and mortality associated with developing complicated diverticular disease.11 21 Current guidance suggests a more cautious approach to these patients and an increased threshold for elective resection.18 22 Yet neither approach is strongly evidence based because of the absence of relevant data.
We have therefore used data from the General Practice Research Database (GPRD) linked to Hospital Episode Statistics (HES) from the UK to determine whether episodes of prior acute diverticulitis predict occurrence of complications and to estimate the excess 1-year mortality associated with the spectrum of complications of diverticular disease compared with the general population.
General Practice Research Database
The GPRD contains diagnostic and prescription data for over 13 million people of the general population in the UK, with 3.4 million active patients contributing data. Originally developed in 1987, the database represents a prospectively collected source of continuous data on illness in general practice in the UK, where all patient care is coordinated by the general practitioners. The data collected are audited regularly and the participating general practices are subjected to a number of quality checks to ensure that they are ‘up to standard’ for research purposes. The database consists of observations and diagnoses made by general practitioners, as well as information sent to them from hospitals such as pathology and radiology reports and discharge letters. Diseases are coded within the GPRD using Read or Oxford Medical Indexing System (OXMIS) codes. Read codes are a coded thesaurus of clinical terms that have been used by clinicians to record data in IT systems in the primary care setting since 1985. OXMIS codes form a hierarchical coding system that has been used to record data in primary care.
Hospital Episode Statistics
HES is a data source containing details of all admissions to NHS hospitals in England, including demographic data and information about diagnoses and operations. It has collected data since 1989, with more than 12 million new records added each year. Records are coded using a combination of International Statistical Classification of Disease and Related Health Problems (ICD-10) for primary diagnosis at discharge along with Office of Population, Censuses and Surveys Classification of Surgical Operations and Procedures (OPCS 4) detailing procedures performed.
In over 200 general practices in England within the GPRD, patients' GPRD and HES data have been linked. The data contain information on admissions from 1997 to 2007.
We identified a cohort of patients with incident complicated diverticular disease who had a Read or OXMIS code for stricture, perforation/abscess or fistula due to colonic diverticular disease. To ensure all cases were captured, we also included all cases of diverticular disease plus a procedure code for any left-sided surgical procedure or the formation of a stoma during ‘up to standard’ data between 1990 and 2007. We excluded all patients with a code for colorectal cancer and those under the age of 45 years at index date. A case was defined as incident if no prior record of complicated colonic diverticular disease was entered into the record within 90 days of original registration at the practice. This time period has been shown to be valid for acute diagnoses.23 The control population was allowed to have a prior history of diverticulosis, diverticular disease and acute diverticulitis, but not a specific code for a complication.
From this cohort, we then identified patients with each complication of stricture, abscess/perforation and fistula. Those with an abscess/perforation were identified using a previously validated method.5 Those with a fistula were identified using a code for a colovesical or colovaginal fistula plus a code for diverticular disease. Those with a stricture were identified as those with a code for a stricture plus a code for diverticular disease. All cases with available HES-linked data were identified in the HES data using ICD-10 codes for diverticular perforation (K572), those with an OPCS 4 code for a Hartmann's procedure, and those with an ICD-10 code for a fistula or a stricture in their primary diagnosis field for each episode. Any cases identified in secondary care from the HES-linked period that were not identified from the primary care data were added to the respective disease cohort. Only patients with an exclusive diagnosis of one of the complications of diverticular disease were taken forward in the analysis.
Acute diverticulitis was defined as any case with a Read/OXIMIS code for acute diverticulitis. A further group was defined of patients with a code for a consultation of acute diverticulitis linked to a code for the same consultation with a prescription of antibiotics.
The population cohort consisted of three individually matched patients within the same practice as the case matched by age and gender. The cases from the disease cohort were excluded from the population cohort. The study is a matched cohort design, and each of the population controls had to be alive and contributing data to the GPRD at the index date of the case, which is effectively matching on calendar period.
We carried out an historical cohort study design using the disease and population cohorts. All patients with an incident diagnosis between 1990 and 2007 were included in the study population except those who had died at the index date.
We analysed data on all patients with complicated diverticular disease and then by complication type (perforation/abscess, stricture and fistula). The patients were censored at the date of recording of death, transferring out of the practice or the practice no longer being deemed up to standard, whichever came first. Our primary outcome measure was mortality, and we examined the effects of age (in 10-year bands), sex and comorbidity using the Charlson index.24 We then compared mortality in the disease cohort with the general population cohort using Cox regression analysis, controlling for potential confounders (age, gender, comorbidty, smoking, episodes of acute diverticulitis, and body mass index (BMI)) and checking the proportional hazards of our final models using log-log plots. The majority of deaths from perforated diverticular disease occur in the first year after diagnosis,5 and therefore we planned an analysis of the data using 1-year mortality.
Nested case–control study of acute diverticulitis
We carried out a nested case–control study of only cases of complicated diverticular disease. We used logistic regression analysis to examine the effect of prior episodes of acute diverticulitis on developing each type of complication using the other cases as controls, as acute diverticulitis was a rare event in the general population. We included in the model the effects of age, sex, smoking, BMI and comorbidity, as they have previously been identified as possible risk factors for the development of complications.
Analysis was performed using StataMP V.11 (64 Bit) software package. The study had approval from the Independent Scientific Advisory Committee approval board who provide scientific advice to the Medicines and Healthcare products Regulatory Agency (MHRA) on study design and advise if further approval is required from the Multi-centre Research Ethics Committee outside the MHRA's current approval for observational studies.
Analysis of HES-linked data sensitivity
We planned a separate sensitivity analysis of only HES cases to estimate any differences from our overall estimates, should this be apparent. For this analysis, only those patients with HES-linked data were used to allow comparison with the cohort without linked data to ensure there was no systematic differences between those with and without linked data.
We have previously received anonymised case records for a selection of patients with a diagnosis of complicated diverticular disease from our previous cohort of patients with complicated diverticular disease5 derived from the GPRD. A case was considered a valid case if there was a recording of a diagnosis of perforation/abscess, stricture or fistula with a clinic letter or discharge summary that confirmed the diagnosis within 1 year of the diagnosis.
We identified 2950 patients with a diagnosis of complicated diverticular disease (and 8739 controls). The mean prior data time to diagnosis was 8.00 (SD 5.05) years in the general population and 7.99 (SD 5.05) years in the disease cohort. The disease cohort was more likely to be women than men (63.5% vs 36.5%, χ2 p<0.001). The disease cohort was more likely to have serious comorbidity (Charlson index >2 35.1% vs 23.0%, χ2 p<0.001), be ever smokers (41.5% vs 29.7%, χ2 p<0.001) and have a BMI >30 kg/m2 (16.9% vs 12.1%) (table 1).
In total, 1042 (35.3%) of the disease cohort died over the follow-up period compared with 2062 (23.6%) of the general population cohort. Most of the excess mortality occurred in the first year after diagnosis, as shown in figure 1.
Analysis of the mortality in the first year after diagnosis of a perforation or abscess revealed a 4.5-fold increase in mortality compared with the general population (HR 4.55, 95% CI 3.74 to 5.52) (table 2). Those with a fistula had a 2.6-fold increase in mortality (HR 2.60, 95% CI 1.47 to 4.62) and had a 40% increased mortality in subsequent years compared with the general population (HR 1.4, 95% CI 1.04 to 1.91). A 2.4-fold increase in mortality was observed for those patients with a diagnosis of a diverticular stricture (HR 2.41, 95% CI 1.86 to 3.11). The 1-year mortality of all cases was 12.8% (377/2932), with the greatest 1-year mortality associated with abscess and perforation at 15.5% (239/1546).
The disease cohort was far more likely to have a recorded episode of prior acute diverticulitis than the general population. Notably, across the spectrum of complications of diverticular disease, most patients (72.3%) had no prior recording of an episode of acute diverticulitis (table 1). When the diagnosis of acute diverticulitis was restricted to those patients with a Read or OXIMIS code for acute diverticulitis plus an antibiotic prescription for that consultation, 90.2% of patients across the spectrum of complicated diverticular disease had no prior episodes recorded (table 1).
Nested case–control study
Those with a fistula experienced an increased number of prior episodes of acute diverticulitis compared with those with stricture or perforation/abscess (table 3).There was evidence of a dose–response relationship (increasing risk of developing a fistula with increasing number of prior episodes of acute diverticulitis) between increasing prior episodes of acute diverticulitis and the development of a fistula, with those with two or more prior episodes being 1.5 times more likely to develop a fistula than the other complications (OR 1.54, 95% CI 1.08 to 2.19) (table 3).
Analysis of HES data sensitivity
When we restricted our mortality analysis to those identified only in HES, the estimates for mortality were similar. For example, overall mortality in HES was 1.5 times that of the general population, controlling for age, sex and comorbidity (HR 1.46, 95% CI 1.25 to 1.71), compared with 1.5 times in cases identified from GPRD (HR 1.54, 95% CI 1.42 to 1.66).
Case note review
We previously identified a cohort of patients with complicated diverticular disease from which we validated 2.0% (70/3485) of cases.5 We showed that 58 (84.3%) of 70 cases had either a clinic letter or discharge summary from secondary care as evidence of their diagnosis of complicated diverticular disease. Our cohort of cases of perforated diverticular disease, as defined above, captured 20 of the cases validated in this larger cohort, representing 2.2% of our patients with perforation. Of these cases, 18 (90%) had evidence from secondary care of a diagnosis of perforated diverticular disease. Our diagnosis of fistula captured 12 cases, all of which had evidence of a diagnosis from secondary care. We identified 12 cases of stricture, of which 11 (92%) had evidence from secondary care. A total of eight cases with abscess were identified, of which seven (88%) had evidence of a diagnosis from secondary care.
Summary of findings
This study has shown that most patients with complicated colonic diverticular disease had no prior history of acute diverticulitis, but those with a fistula were most likely to have had at least one prior episode of prior acute diverticulitis. The risk of fistula increased in a dose–response fashion with increasing episodes of prior acute diverticulitis, whereby those with two or more prior episodes had a 50% increase in risk of fistula compared with developing either stricture or perforation/abscess. There is an increase in mortality in patients with any of the complications of diverticular disease compared with the general population in the first year after diagnosis. The excess mortality in those with perforation or abscess was 4.5 times greater than the mortality of the general population, whereas, in the patients with fistula or stricture, this risk was ∼2.5-fold. The increased mortality was confined to the first year except in those with a fistula, who had a 40% increase in mortality in subsequent years.
Strengths and limitations
For the first time we have used linked UK primary and secondary care data, which is a major advantage when studying complicated diverticular disease. Among the advantages are the security of the diagnosis and the timing of the complication, but probably more important is the ability to document the detail of the progression of the disease. This is compared with previous studies, which have only had secondary or tertiary care data. Previous studies have shown that up to 20% of cases of perforated diverticular disease may not have a primary care record of the diagnosis.5 It is still possible therefore that we have missed cases, as not all cases had linked data, from the cohort with a diagnosis of complicated diverticular disease made in secondary care but not recorded in primary care before 1997 or in unlinked practices. This number will represent only a small number of actual cases. Our comparison of the hospital-only defined cases and the overall cohort showed no important differences in the point estimates for mortality, suggesting that there is no systematic difference in data from the HES-linked and non-linked practices.
The coding of complicated diverticular disease is far more detailed in Read/OXIMIS coding, which is a clear strength of using primary care data. Although this is somewhat limited in ICD-10, the secondary care data have the advantage of OPCS4 coding of procedures. The two linked together are therefore superior to either type used alone. Our validation study shows that, in the validated cases, most patients have evidence from secondary care to support the diagnosis made in primary care of abscess/perforation, fistula and stricture. Lifestyle risk factors for the development of complications such as BMI and smoking were not recorded for approximately a third of those in the study. Those with missing data were classified as having missing data and were included in the analysis.
Mortality associated with the spectrum of complicated diverticular disease has previously only been described in national audit studies and retrospective cases series from secondary care.8–10 We have previously described the mortality of perforated diverticular disease using a separate cohort, but were unable to describe the effect of prior episodes of acute diverticulitis5 or the mortality associated with abscess, fistula and stricture. The present study shows that the mortality associated with complicated diverticular disease is large, yet it varies substantially depending on the type of complication, which is an important clinical finding. Our previous study of the mortality associated with perforated diverticular disease reported a sixfold (HR 5.6, 95% CI 4.7 to 6.8) increase in 1-year mortality, which is similar to the mortality reported for those with a perforation and abscess in this study (HR 4.55, 95% CI 3.74 to 5.52). Prior studies are smaller than ours, with the largest having just 538 cases.9 Although a study of 300 cases of complicated diverticular disease containing 23 cases of faecal peritonitis showed a mortality in these of 48%,8 the risk of in-hospital mortality associated with a stricture was estimated using just 31 cases.8 10 These studies were clearly limited by their size and could not report on the role of acute diverticulitis. The present population-based study suggests that the risk of death is greatest with perforation/abscess followed by fistula and stricture, with an excess mortality continuing only after a diagnosis of a fistula. Previous studies with mixed populations have also observed significant mortality among this group of patients.25 26
Initial studies and the subsequent advice based on them suggested that elective resection of the colon should be performed after a second episode of acute diverticulitis, as it was thought to herald the likelihood of a complication,17 21 yet this view appears to have recently changed and elective resection is only to be advised on a patient-by-patient basis.19 20 22 27 28 Before this latter advice was published, there was evidence of increasing resection rates.2 Our results show that only a minority of patients have a prior history of more than two episodes of acute diverticulitis (12.9%), and, although this is far more than in the general population, it is still only a small number of cases and therefore the current cautious approach to surgical intervention would appear sensible.
Funding JW is funded by a UK National Institute of Health Research Clinician Scientist Fellowship. The dataset was acquired through the Medical Research Council scheme for academic use of GPRD data. The linked data and validation study was funded in part by the BUPA Foundation. The funders had no role in the design and conduct of the study, collection, management, analysis, and interpretation of the data, and preparation, review or approval of the manuscript.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.