Longitudinal assessment of colonoscopy adverse events in the prospective Cooperative Studies Program no. 380 colorectal cancer screening and surveillance cohort

Abstract

Background and Aims:

Data are limited regarding colonoscopy risk during long-term, programmatic colorectal cancer screening and follow-up. We aimed to describe adverse events during follow-up in a colonoscopy screening program after the baseline examination and examine factors associated with increased risk.

Methods:

Cooperative Studies Program no. 380 includes 3,121 asymptomatic Veterans aged 50–75 who underwent screening colonoscopy between 1994–97. Peri-procedure adverse events requiring significant intervention were defined as major events (other events were minor) and tracked during follow-up for at least 10 years. Multivariable odds ratios (OR) were calculated for factors associated with risk of follow-up adverse events.

Results:

Of 3,727 follow-up examinations in 1,983 participants, adverse events occurred on 105 (2.8%) examinations in 93 individuals, including 22 major and 87 minor events (examinations may have had >1 event). Incidence of major events (per 1,000 examinations) remained relatively stable over time with 6.1 events at examination two, 4.8 at examination three, and 7.2 at examination four. Examinations with major events included one perforation, three gastrointestinal (GI) bleeds requiring intervention, and 17 cardiopulmonary events. History of prior colonoscopic adverse events was associated with increased risk of events (major or minor) during follow-up (OR 2.7; 95% confidence interval, 1.6–4.6).

Conclusions:

Long-term programmatic screening and surveillance was safe, as major events were rare during follow-up. However, serious cardiopulmonary events were the most common major events. These results highlight the need for detailed assessments of comorbid conditions during routine clinical practice, which could help inform individual decisions regarding the utility of ongoing colonoscopy follow-up.

Introduction

Current guidelines recommend the use of screening colonoscopy for average-risk patients beginning at 45–50 years of age.^1–3 Screening colonoscopies are an evidence-based modality for decreasing colorectal cancer (CRC) risk and are generally considered low-risk, with a pooled major adverse event rate of 2.8 per 1000 procedures (95% confidence interval (CI), 1.5–5.2).^4–13 After an initial screening colonoscopy, individuals enter a CRC prevention program, which includes follow-up examinations in patients with and without neoplasia. Yet, data are limited regarding the risk of adverse events over time across multiple examinations in individuals undergoing long-term programmatic colonoscopy follow-up, especially as they age and accumulate medical comorbidities. Given the limited evidence supporting the benefits of follow-up colonoscopy after the initial examination,¹⁴ it is important to accurately characterize the longitudinal risk of these invasive procedures to help inform clinicians and patients about the safety of ongoing screening and surveillance.

Current estimates of colonoscopy risk are based on studies of a single screening or surveillance colonoscopy, which were typically evaluated either retrospectively at single academic institutions^15–20 or through large population-based databases^21–28. Prior single-center studies were often limited in sample size or had variations in adverse event reporting owing to institutional-specific practices. Population-based studies, on the other hand, relied on procedural and diagnostic coding from administrative data, and thus may have incompletely captured target populations and lack important clinical detail. Specifically, these studies considered colonoscopy adverse events as isolated events unrelated to prior factors or procedures, as they often lacked individual-level follow-up data across multiple examinations. This limits the ability to adequately characterize ongoing colonoscopy risk based on current and past factors, which is necessary to inform risk-stratified follow-up recommendations.

To address these knowledge gaps, we used data from Cooperative Studies Program Study #380, a well-annotated prospective CRC screening colonoscopy cohort, to describe the risk of adverse events across multiple follow-up colonoscopies after a baseline examination and to examine the patient and endoscopic factors associated with adverse event risk during follow-up.

Methods

Study Design and Patient Population

Data were collected from the Department of Veterans Affairs (VA) Cooperative Studies Program Study #380, a prospective CRC screening cohort of 3,121 asymptomatic veterans aged 50–75 years old who underwent a screening colonoscopy from 1994 to 1997 across 13 medical centers. The original purpose of Cooperative Studies Program no. 380 was to describe the distribution of screening examination findings and examine the relationship between these findings and varying surveillance intervals with long-term outcomes. The initial follow-up examination frequency was partially guided by a five-year randomization protocol determined by baseline colonoscopy findings. Participants were followed for ≥10 years or until death. The complete methods and primary results of Cooperative Studies Program no. 380 have been previously published.^{4, 5, 29}

At the initial Cooperative Studies Program no. 380 examination, adverse events were documented up to 30 days post-procedure by study investigators and nurses.³⁰ During the scheduled follow-up phase, study nurses called participants 24 to 72 hours after an examination to assess adverse events. Participants were also instructed to contact study personnel regarding any adverse event concerns. When funding for study nurses expired, adverse events were tracked centrally by interacting with local study investigators or reviewing each study site’s patient files (mainly including peri-procedural adverse events).

For the current analysis, Cooperative Studies Program no. 380 patients with at least one follow-up colonoscopy after the baseline examination were included, with one exception. During an adjudication process, it became apparent that one study site frequently recorded clinically insignificant events (e.g., “preprocedural nasal cannula placement”) during colonoscopy procedures as ‘other’ minor events. This site was excluded to avoid finding artificial relationships with follow-up colonoscopy adverse events. No other relevant site-level variations or patterns of colonoscopy complications were identified. Otherwise, all other follow-up colonoscopies were included, regardless of completion status or time from prior examination. Incomplete examinations (failure to reach cecum) or repeated examinations occurring within 6 months of prior could have been driven by a previous adverse event, and thus were counted as separate examinations in our analysis. The study was approved by the Durham VA Medical Center Institutional Review Board (MIRB #1872) and included a waiver of informed consent. All methods were executed in accordance with relevant guidelines and regulations.

Outcome Measures

Adverse events were classified as major or minor events. Events were reported to each institution’s institutional review board and the Cooperative Studies Program no. 380 Data Monitoring Board. Major events included colonic perforation; gastrointestinal (GI) bleeding requiring hospitalization, surgery, or transfusion; cardiopulmonary events (including peri-procedural apnea; hypotension requiring intervention such as IV resuscitation and/or procedure termination; various cardiac arrhythmias requiring intervention; myocardial ischemia/angina; need for cardiopulmonary resuscitation); or ‘other’ as determined by study investigators. Minor events were reported as GI bleeding that did not require hospitalization or transfusion, transient hypotension which did not require specific intervention (systolic blood pressure < 100 mmHg), transient oxygen desaturation which did not require specific intervention (saturation < 88%), abdominal bloating or pain lasting more than two hours after completion of the study, abdominal pain which prevented completion of the colonoscopy, or ‘other’ as determined by study investigators. To model the impact of specific demographic, clinical, or procedural factors on follow-up risk, a composite outcome of major and/or minor event(s) was created, dichotomized as yes or no, since there were too few major outcomes to model as a primary outcome.

Covariates

Prior to the baseline examination, participants completed a series of self-reported questionnaires detailing demographics, medical and surgical history, and family gastrointestinal history. Baseline and follow-up intraprocedural data were collected by the performing endoscopist and verified by the study team. Prior complications were categorized in three clinically relevant ways: 1) Any prior major or minor events (yes/no), 2) Timing of any previous adverse event (immediately prior examination/remote examination/no prior), or 3) Number of prior examinations with any adverse events (0, 1, 2+). Finally, given the presumed subjective differences between reporting events across sites, particularly for minor events, site-level variations were also included as an adjusted covariate.

Statistical Analysis

Adverse event rates and 95% CIs were calculated per 1,000 follow-up examinations, which evaluated only the most severe event per examination. Additional itemization, which included multiple events per examination, was also reported by specific adverse event. To calculate the impact of specific factors on follow-up risk, we used longitudinal mixed models which account for multiple follow-up colonoscopies occurring at different time points.³¹ Models were adjusted for demographic, baseline clinical, and procedural-specific characteristics selected a priori (including age, sex, race, VA site, comorbid conditions, prior adverse events, daily non-steroidal anti-inflammatory use, adequacy of bowel preparation, performance of polypectomy, and randomization group). To reflect the varying intensity of surveillance based on baseline examination findings as prespecified in the original study design, models were also adjusted for randomization group categorized as either a specific surveillance schedule (examinations at 2 and 5 years or 5 years only) or usual care per discretion of provider (all other examination schedules). Finally, a patient-level intercept was included to account for unmeasured patient factors that are correlated across examinations. Statistical significance was defined as p <0.05. All analyses were performed using R 4.05 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Patient and Procedure Characteristics

Of the initial Cooperative Studies Program no. 380 cohort, 2,171 participants had at least one follow-up examination. After excluding one site as described above (n=188), 1,983 participants remained who underwent a total of 3,727 follow-up colonoscopies (Table 1). There were no relevant qualitative differences in baseline adverse events that may have impacted inclusion between those who had follow-up examinations versus those who had a baseline examination but did not adhere to follow-up.²⁹ The cohort was predominately male (n=1,916, 96.6%) and Caucasian (n=1,661, 83.8%), with body mass index (BMI) >25 kg/m² (n=1687, 85.1%) and >1 comorbid condition (n=1844, 93.0%). In total, 3,727 follow-up examinations were performed, with 1,045 participants having received at least two follow-up examinations after baseline, and 423 participants with at least three follow-up examinations after baseline.

Table 1:

Baseline Demographic and Clinical Characteristics of Included Cooperative Studies Program no. 380 Participants (N = 1983)

Variable	Baseline Frequency, n (%)
Demographics
Age, years, n (%)
50–59	725 (36.6)
60–69	952 (48.0)
70–75	273 (13.8)
76–80+	33 (1.7)
Race, n (%)
Caucasian	1661 (83.8)
African American	178 (9.0)
Hispanic	93 (4.7)
Other	48 (2.4)
Missing	3 (0.2)
Sex, n (%)
Male	1916 (96.6)
Female	67 (3.4)
Baseline Clinical factors
Body mass index* (kg/m2), n (%)
<=25	295 (14.9)
>25–29	878 (44.3)
>=30	809 (40.8)
History of diverticulosis, n (%)
No	1925 (97.1)
Yes	58 (2.9)
Diabetes
No	1614 (81.4)
Yes	369 (18.6)
Comorbid conditions, n (%)
0	139 (7.0)
1	362 (18.3)
2	447 (22.5)
3+	1035 (52.2)
Baseline Smoking, n (%)
Never smoker	486 (24.5)
Former smoker	1082 (54.6)
Current daily smoker	415 (20.9)
Prior abdominal surgery, n (%)
No	1870 (94.3)
Yes	113 (5.7)
Daily NSAID use, including aspirin, n (%)
No	940 (47.4)
Yes	1043 (52.6)
Randomization group ‡
Usual care	917 (46.2)
Surveillance colonoscopy	1066 (53.8)

^*Missing data excluded

^‡Follow-up examination frequency was initially guided by a randomization protocol based on screening colonoscopy findings.

The number of individuals receiving a baseline and each follow-up examination, as well as the number of adverse events associated with these examinations, is shown in Figure 1. Overall, events were identified on 105 (2.8%) follow-up examinations among 93 individuals, including 22 (0.59%) major and 87 (2.3%) minor events. Across follow-up examinations, the frequency of major events remained approximately stable between 0.5% and 0.7%, while reports of minor events tended to decrease over time from 2.9% at examination two to 1.0% by the fourth examination. Among patients who did not receive a follow-up examination, 98 (10.2%) experienced any adverse event on the baseline examination, including five (0.58%) with major and 94 (10.9%) with minor events. On the other hand, of those who received any follow-up, a similar proportion (n=193, 9.7%) had adverse events identified on baseline colonoscopy, including 8 (0.40%) with major and 185 (9.3%) with minor events.

Figure 1: Descriptive Cohort Diagram for Included Cooperative Studies Program no. 380 Participants with Baseline Examinations and Follow-up Examinations.

*699 total examinations in 423 participants

^#Total number is not equal to major + minor major adverse events because some individuals had both major and minor adverse events occur on the same examination

Incidence of Major Adverse Events

The incidence of any adverse event during follow-up (per 1,000 examinations) was 28.2 events (95% CI, 23.2–34.1), including 5.9 major events (95% CI, 3.8–9.1) as shown in Table 2. Supplemental Table 1 demonstrates the incidence of major events remained relatively stable over time, with 6.1 major events at examination two (95% CI, 3.3–10.9), 4.8 events at examination three (95% CI, 1.8–11.8), and 7.2 events at examination four and beyond (95% CI, 2.6–17.6). Total numbers only represent single examinations, which may include multiple occurrences of major events. One follow-up examination experienced a colonic perforation (0.27 events, 95% CI, 0.014–1.7), with a polypectomy occurring on the same examination. Three events of lower GI bleeding requiring hospitalization (0.81, 95% CI, 0.21–2.6) occurred, including two of these participants requiring transfusion (0.54, 95% CI, 0.09–2.2) and one requiring surgery (0.27, 95% CI, 0.014–1.7); all examinations had polypectomy. Seventeen follow-up examinations resulted in at least one major cardiopulmonary adverse event (4.6, 95% CI, 2.8–7.5). Of these examinations, ten examinations reported in hypotension requiring intervention (2.7, 95% CI, 1.4–5.1), one examination with peri-procedural apnea (0.27, 95% CI, 0.014–1.7), six examinations resulted in cardiac arrhythmia (1.6, 95% CI, 0.66–3.7), three examinations with bradycardia requiring atropine (0.81, 95% CI 0.21–2.6), and one vasovagal event during follow-up (0.27, 95% CI, 0.014–.74). The remaining examination experiencing an ‘other’ major event was due to hypoglycemia post-colonoscopy preparation requiring hospitalization. No follow-up examinations resulted in death or the need for cardiopulmonary resuscitation.

Table 2.

Number and Rate of Examinations with Adverse Events Among Baseline and Follow-up Colonoscopies

	Baseline colonoscopy (n = 2848)		All follow-up colonoscopies (n = 3727)
	Events* (n)	Event rate per 1,000 (95% CI)	Events* (n)	Event rate per 1,000 (95% CI)
Any adverse event, n (%)	291	102.18 (91.42–114.03)	105	28.17 (23.21–34.14)
Major adverse events, n (%)	13	4.57 (2.54–8.02)	22	5.90 (3.80–9.08)
Perforation	0	0 (0–1.68)	1	0.27 (0.014–1.74)
Major GI Bleeding	2	0.70 (0.12–2.83)	3	0.81 (0.21–2.56)
Major Cardiopulmonary Events	11	3.86 (2.03–7.13)	17	4.56 (2.75–7.46)
Other	4	1.40 (0.45–3.86)	1	0.27 (0.014–1.74)
Minor adverse events, n (%)	279	97.96 (87.42–109.61)	87	23.3 (18.84–28.85)

^*Examinations may have had >1 adverse event, but only the single worst occurrence is included in the total

Of the 13 total individuals with a major event at baseline, none had a major event on any follow-up examination; however, one individual without a major event at baseline experienced the same three major events during follow-up (bradycardia responsive to atropine). Otherwise, in the ten individuals with a major event during follow-up and polypectomy on same examination, three had bleeding requiring hospitalization, including two with transfusion and one requiring surgery. In the others, four had hypotension requiring intervention, two had cardiac arrhythmias, and one had perforation. Further, of these ten individuals with major event and polypectomy, four had a polyp sized ≥10mm removed (including three in the right colon) resulting in two instances of hospitalization for bleeding and two instances of hypotension requiring intervention. Finally, of the eight patients with major events and on daily acetylsalicylic acid/nonsteroidal anti-inflammatory drugs (ASA/NSAIDs), two had bleeding requiring hospitalization while three had hypotension requiring intervention. Additional details for both major and minor event rates are described in Table 2.

Risk Factors for Adverse events

Of the 22 individuals with major adverse events during follow-up, only one (4.5%) had a prior major complication. Although there were too few major events to perform meaningful statistical analysis, those with major events were more likely to have had any prior adverse event (2.5% vs 0.34%), be older (1.7% in those 76–80+ years vs 0.38% in those 50–59 years), have poor prep (1.4% vs 0.59% in those with good prep), and have undergone polypectomy (3.8% vs 0.35%). Trends were largely similar with minor reported adverse events (Supplemental Table 2).

The risk of any adverse events, including major and/or minor, at follow-up was not associated with any demographic, baseline clinical, or intraprocedural factors (Table 3). However, compared to no prior adverse events, a history of any prior reported adverse event was associated with higher odds of follow-up adverse events (OR 2.7; 95% CI, 1.6–4.6). Follow-up adverse event risk was impacted by the timing of the prior adverse event, with the greatest risk in those who had an adverse event on the examination immediately prior (OR 3.1; 95% CI, 1.7–5.7) rather than a remote examination (OR 2.1; 95% CI, 0.91–4.8), when compared with those with no prior adverse events (Table 4). An increasing number of prior examinations with adverse events was also associated with increased follow-up adverse event risk compared with no prior adverse events (2+ examinations: OR 3.9; 95% CI, 1.3–12.0; 1 examination: OR 2.6; 95% CI, 1.5–4.6).

Table 3.

Adjusted Odds of Follow-Up Colonoscopy Adverse Event Risk by Individual Characteristics

Predictive Factors for Follow-up Adverse Events	Number of Follow-up Examinations with Any Adverse Events N=105	Total Number of Follow-up Examinations N=3727	Adjusted Odds Ratio† (95% CI)	p-value
Prior Adverse Events
Prior Major or Minor Adverse Event, n (%)
No	69 (2.10)	3279	Reference
Yes	36 (8.04)	448	2.71 (1.58–4.64)	<0.001
Demographics
Age, years, n (%)
50–59	27 (2.07)	1305	Reference
60–69	49 (2.68)	1828	1.04 (0.58–1.88)	0.88
70–75	25 (4.67)	535	1.31 (0.64–2.66)	0.46
76–80+	4 (6.78)	59	1.50 (0.38–5.90)	0.56
Race, n (%)
Caucasian	93 (2.98)	3119	Reference
African American	6 (1.72)	349	0.58 (0.21–1.58)	0.28
Hispanic	3 (1.78)	169	0.89 (0.22–3.58)	0.87
Other	3 (3.45)	87	1.46 (0.36–5.86)	0.59
Missing	0 (0)	3	—	—
Sex, n (%)
Male	98 (2.72)	3614	Reference
Female	7 (6.19)	113	1.79 (0.66–4.83)	0.25
Clinical factors
History of diverticulosis, n (%)
No	102 (2.81)	3632	Reference
Yes	3 (3.16)	95	1.01 (0.23–4.33)	0.99
Comorbid conditions, n (%)
0	0 (0)	0
1	45 (2.61)	1726	Reference
2	39 (2.70)	1446	0.96 (0.56–1.67)	0.90
3+	21 (3.78)	555	1.10 (0.55–2.18)	0.79
Baseline Smoking, n (%)
Never smoker	32 (3.62)	883	Reference
Former smoker	58 (2.82)	2055	0.80 (0.46–1.41)	0.44
Current daily smoker	15 (1.90)	789	0.66 (0.30–1.44)	0.29
Prior abdominal surgery, n (%)
No	99 (2.86)	3465	Reference
Yes	6 (2.29)	262	0.82 (0.30–2.24)	0.70
Daily NSAID use, including aspirin, n (%)
No	48 (2.66)	1802	Reference
Yes	57 (2.97)	1922	0.88 (0.53–1.45)	0.61
Randomization group ‡
Usual care	36 (2.30)	1565	Reference
Surveillance colonoscopy	69 (3.19)	2162	1.35 (0.81–2.25)	0.24
Intraprocedural factors
Adequacy of bowel prep, n (%)
Good	80 (3.37)	2376	Reference
Fair	14 (2.66)	526	0.94 (0.47–1.86)	0.86
Poor	6 (2.79)	215	1.04 (0.38–2.80)	0.94
Polypectomy, n (%)
No	53 (2.54)	2086	Reference
Yes	52 (3.17)	1641	0.9 (0.57–1.50)	0.74

^†Model adjusted for any prior adverse events categorized as yes/no, site, and colonoscopy randomization group, along with all demographic, clinical, and procedural factors covariates

^‡Follow-up examination frequency was initially guided by a randomization protocol based on screening colonoscopy findings

Table 4.

Adjusted Odds of Follow-Up Colonoscopy Adverse Event by Prior Adverse Events Categorization

Prior Adverse Events Categorization	Follow-up Adverse Event of Any Kind
	Yes (N=105)	No (N=3622)	Adjusted Odds Ratio* (95% CI)	p-value
Timing of adverse event, n (%)
No prior adverse event	69	3210	Reference
Remote examination	11	182	2.09 (0.91–4.80)	0.084
Immediately prior examination	25	230	3.13 (1.71–5.71)	<0.001
Number of prior examinations with adverse events, n (%)
0	69	3210	Reference
1	29	373	2.59 (1.46–4.60)	0.0011
2+	7	39	3.91 (1.27–12.00)	0.017

^*Models adjusted for age, sex, race, site, comorbid conditions, daily non-steroidal anti-inflammatory use (including aspirin), adequacy of bowel preparation, performance of polypectomy, and colonoscopy randomization group. Separate adjusted models were used to estimate the odds ratios for each categorization of prior adverse events

Discussion

In this study of patients undergoing longitudinal screening and surveillance colonoscopy, we identified few major events at follow-up colonoscopy (0.59%), with a stable occurrence over time of approximately 5–7 major events per 1,000 examinations. Serious cardiopulmonary events were the most common major events (4.6 per 1,000 examinations), while major events related directly to the colonoscopy such as perforation (0.27 events per 1,000 examinations) and serious GI bleeding (0.81 events per 1,000 examinations) were much less frequent. Further study is needed to characterize the relationship of cardiopulmonary events to the overall risk of long-term colonoscopy follow-up, with particular attention to the relationship with increasing age and medical comorbidities. Otherwise, most adverse events were minor in severity (2.3%), with the rate of minor events decreasing across examinations. Although several potential explanations exist, including under-reporting over time, this finding highlights the importance for robust, systematic tracking of adverse events during routine practice, with standardized definitions, to identify quality improvement opportunities to mitigate longitudinal adverse event risk.

In this analysis, we describe a wide array of GI and non-GI major adverse events associated with follow-up colonoscopies, which expands on previous reports that were based on aggregated data of all examination indications and regardless of number. A 2008 systematic review of 12 studies, including 57,742 average-risk screening colonoscopies, found a pooled serious adverse event rate of 2.8 per 1000 examinations (95% CI, 1.5–5.2),¹³ with serious events generally defined to include perforations, hemorrhage, diverticulitis, severe abdominal pain, cardiovascular events, and death. Our slightly higher major event rate is likely attributable to our additional inclusion of hypotension requiring intervention (2.7 events per 1000 examinations), as originally defined in the Cooperative Studies Program no. 380 protocol. However, this increased rate of cardiopulmonary events (4.6 events per 1,000 examinations), particularly compared with the lower rates of events as a direct result of colonoscopic instrumentation such as perforation (0.27 events per 1,000 examinations) and serious GI bleeding (0.81 events per 1,000 examinations), highlights the importance of detailed assessments of comorbid conditions during routine practice in an aging population over time. Future work will need to evaluate the contribution of accumulating medical comorbidities on cardiopulmonary risk, which could help inform individual decisions regarding the safety of ongoing colonoscopy follow-up.

This study is one of the first to perform serial analyses for specific adverse event risk at each colonoscopy in individual participants undergoing multiple examinations, which is a unique contribution to the literature. The major event rate (per 1,000 examinations) in our analysis remained stable and rare across multiple examinations (6.1 at examination two, 4.8 at examination three, and 7.2 at examination four and beyond). Furthermore, follow-up adverse event rates across multiple examinations attributed directly to the colonoscopy procedure remained consistent with current quality benchmarks set by the American Society for Gastrointestinal Endoscopy (ASGE)/American College of Gastroenterology (ACG) Task Force. Our longitudinal follow-up cohort experienced an overall perforation incidence of 1/3727 examinations (0.03%) and major bleeding incidence of 3/3727 examinations (0.08%), both well below the ASGE performance targets of <0.1% and <1%, respectively.³² Although current guideline metrics do not distinguish between screening and surveillance examinations or based on an individual’s examination number, our work in conjunction with baseline Cooperative Studies Program no. 380 adverse event data published by Nelson et al. suggest these target metrics are reasonable regardless of indication or examination number.³⁰ Indeed, our study suggests that follow-up with colonoscopy is safe, with few adverse events in a population undergoing long-term, programmatic screening and surveillance. Finally, given that all direct colonoscopy related major events were noted to have had polypectomy on the same examination, ongoing efforts to refine polyp removal techniques may continue to improve programmatic colonoscopy safety.⁴⁴

Understanding other factors associated with follow-up colonoscopy adverse events is essential to personalizing recommendation for ongoing screening and surveillance. Although we could not estimate the precise risk of major events at each follow-up examination based on individual risk factors due to few events, well-documented evidence in the literature supports specific risk factors for adverse events such as polyp removal,^{13, 21, 27, 33–36} medical comorbidities,^{21, 24, 33, 35, 37} and poor bowel preparation.^{35, 38, 39} Additionally, in our analysis evaluating any follow-up adverse event (major and/or minor), we found that prior reported colonoscopy adverse events increased the risk for follow-up adverse events (OR 2.74), with more recently reported adverse events (OR 3.13 if adverse event on the immediately prior examination) and a higher number of past adverse events (OR 3.91 if 2+ prior examinations with any adverse event) having a greater impact. This longitudinal association between adverse events could be due to patient-level factors such as recurrent/difficult polyps, tortuous colons, altered anatomy, or issues with sedation. Importantly, the quality of baseline screening in Cooperative Studies Program no. 380 was consistent with current quality metrics, including adenoma and polyp detection rates, and continued to be monitored during the research phase of the protocol.²⁹ However, overall adverse event rates were largely driven by minor events, which substantially decreased over time (from 2.9% at examination two to 1.0% by examination four). Although colonoscopy may be getting safer over time with improved technology (e.g., use of CO₂ insufflation to minimize barotrauma), innovative techniques (e.g., use of “cold” instead of “hot” snares for polypectomies), or selection of healthier patients during programmatic follow-up, these results were likely impacted by under-reporting of minor events over time which may be seen as routine and manageable. Similar to our study, published estimates of minor events are highly variable, but can be as high as 33%.^{16, 35, 40, 41} Given the concern for haphazard reporting, our results regarding the impact of prior adverse events on current risk should be considered preliminary point estimates that deserve more study, as even minor reported events are important for clinical decision making and may reduce willingness to undergo future examinations.^{16, 35, 42, 43} Recent guidelines now recommend implementing processes to systematically define and track colonoscopy adverse events during screening and surveillance, including relevant patient, procedure, and facility factors.⁴⁴ Rigorously identifying and mitigating factors associated with longitudinal adverse events will help to continually optimize the safety of ongoing colonoscopy follow-up and inform individualized CRC prevention recommendations.

This study has limitations. Adverse events were only systematically tracked within a few days peri-procedurally, so events that relied on patient reporting or occurred up to 30 days later could have been missed.²¹ In terms of risk factors for follow-up adverse events, our cohort lacked sufficient major event outcomes to model this risk separately. Certain information on relevant risk factors was lacking, such as anticoagulant use, and lack of data on the change in specific clinical factors over time may have contributed to the lack of association between adverse event risk and certain clinical characteristics. Unmeasured bias may also have led to the selection of healthier, lower risk patients for follow-up colonoscopy over time, resulting in more conservative odds ratio estimates. While compliance for a first follow-up examination did not seem to be affected by the occurrence of an adverse event at baseline in our population, the initial randomization protocol may have encouraged follow-up adherence to a higher degree than what would occur in the real-world setting. Furthermore, despite the consistency of our results with prior studies, our definitions of major and minor adverse events may differ from other studies which could decrease the external validity of our effect sizes. Finally, while current evidence suggests that females are more likely to report colonoscopic adverse events,^{35, 38, 45} our population was majority male which may limit generalizability. A similar analysis in a larger, more representative screening cohort should be performed to further investigate the findings reported in our analysis.

In summary, this study provides valuable insight for longitudinal CRC screening and surveillance programs. Long-term colonoscopy follow-up was generally safe, with few major events. However, serious cardiopulmonary events were the most common major events. Further study is needed to better characterize the relationship of cumulative risk for cardiopulmonary events with increasing age or medical comorbidities. Additionally, validation of the impact of prior adverse events or other associated factors on follow-up complication risk in larger cohorts is needed. Overall, these results highlight recent efforts to improve programmatic tracking of adverse events during routine screening and surveillance.⁴⁴ These efforts should inform clinical decision support tools which seek to identify individuals who may be less likely to benefit from ongoing colonoscopy follow-up or who may need special precautions when performing the follow-up examination.

Acronyms and abbreviations list

CRC

colorectal cancer

OR

odds ratios

GI

gastrointestinal

CI

confidence interval

VA

Department of Veterans Affairs

CSP #380

Cooperative Studies Program Study #380

BMI

body mass index

ASA

acetylsalicylic acid

NSAIDs

Nonsteroidal Anti-inflammatory Drugs

ASGE

American Society for Gastrointestinal Endoscopy

ACG

American College of Gastroenterology

Data Materials:

The data pertaining to the current study are available upon request at https://www.vacsp.research.va.gov/CSPEC/Studies/INVESTD-R/CSP-380-Risk-Factor-Colonic-Adenomas.asp.