To describe methods that can be used to quantify and compare harm caused by surgical interventions in ophthalmology trials.
A published landmark glaucoma trial was used as an exemplar. A consensus-derived classification system of severity of complications was applied to published data of the Tube Versus Trabeculectomy glaucoma trial. The severity grade of each complication was multiplied to the number of patients who incurred that complication to estimate a total harm score for each intervention. Graphical tools were also used to display the differences in complications between trial arms. A review of literature on best practice for reporting harm data was also conducted.
Analyzing treatment harm is challenging with the relatively small number of events and sample sizes used in randomized controlled trials. However, quantification and graphical representation of harm after surgery is possible. Reframing the research question to one for detecting signals of adverse reactions and use of Bayesian analyses can be useful.
Analysis of harm data in clinical trials needs further attention. A severity classification system and a total harm score can be used to quantify harm after glaucoma surgery. Graphical tools can also help interpret complication data.
T he comparison of surgical procedures in randomized trials should take into account not only efficacy but also the frequency and the severity of complications. While efficacy outcomes are often reported in detail, the evaluation of harm associated with interventions tends to be limited to the frequency of itemized complications. Complications are typically recorded as a binary variable, ie, events that do or do not occur during the study follow-up period. Some complications that occur in trials are unrelated to the intervention, as a causal relationship is not required to be reported as a complication, and trying to ascertain which complications are related may be difficult. Some trials report the frequency of serious complications separately because “serious” is a regulated term (eg, serious adverse event) and has a formal definition. Subsequently, within harms classified as serious, there may be a range of severities.
How best to analyze and present harm data collected in a trial is uncertain. The most commonly used approach is to present contingency tables of frequencies by arm and assess rates either subjectively or by using hypothesis testing and P values. The latter option may not be recommended because of the limited statistical power and problems with multiplicity of testing. Graphical approaches and plots to display harm data have been promoted recently as a way to support investigators as they interpret comparisons between study arms and improve the translation of results. ,
To strengthen harm evaluation in glaucoma surgical trials, Sii and associates proposed a consensus-generated classification system for assessing the severity of common complications after glaucoma surgery in general and specifically after trabeculectomy and glaucoma drainage devices. Sii and associates used Delphi methodology and scored severity between 1 and 10, with grade 1 indicating complications of no harm or clinical significance and grade 10 indicating the worst possible outcome. A similar approach has been reported recently to assess complications of retinal detachment surgery.
The Tube Versus Trabeculectomy (TVT) study is a landmark multicenter randomized trial comparing the efficacy and safety of tube shunt (350-mm Baerveldt glaucoma implant) vs trabeculectomy with mitomycin C (0.4 mg/mL for 4 min) in patients who are at high risk of surgical failure (patients who had undergone previous trabeculectomy and/or cataract surgery with intraocular lens implantation). , During the 5-year follow-up period, tube shunt surgery had a higher success rate and a lower overall complication rate than trabeculectomy. There was a significantly higher incidence of early postoperative complications in the trabeculectomy group compared with the tube shunt group. , However, the majority of early complications of trabeculectomy were nonsevere and thus not associated with long-term or permanent harm. The incidence of serious postoperative complications (ie, those associated with a reoperation or vision loss of ≥2 Snellen lines) within 5 years was similar in both groups. The TVT results suggested that tubes may be a better option in patients who have a high risk of trabeculectomy failure.
The purpose of this study was to apply a consensus-derived severity grade and to quantify the postoperative harm caused by glaucoma surgeries after 5 years of follow-up in the TVT study, for example, and to describe possible options to report harm after surgery.
A total of 212 eyes of 212 patients were enrolled in the TVT study, including 107 patients who underwent placement of a tube shunt and 105 patients who had a trabeculectomy with mitomycin C. Postoperative complications during the 5 years of follow-up reported in 2012 were identified from the trial publication. We did not seek unpublished data from the trial.
Complications reported in the TVT study were assigned a severity using a published severity grade ( Table 1 ). In brief, the severity grade by Sii and associates used a Delphi consensus method with a large group of international glaucoma surgeons to quantify severity of all expected complications after glaucoma surgery in general and after trabeculectomy and glaucoma drainage devices in particular. We estimated the total harm score for each intervention as the sum of harm scores of all complications.
|Reported TVT Complications||Different Degrees of Severity a||Estimated Harm for Each Reported Complication b|
|Shallow or flat anterior chamber||3,5||4|
|Persistent corneal edema||5,8||6.5|
|Causing loss of vision|
|Loss of vision ≥2 Snellen lines||7,8,10||8.3|
When the reported data in the TVT study were incomplete or indeterminate, a series of grading rules was applied (Appendix 1). Unrelated complications and those not typical of glaucoma surgery were not considered, eg, retinal detachment. We did not compare statistically the total harm between the 2 groups, as this would be using “complications” rather than “subjects” as the units of analysis, which would be possible only if individual participant data were available. Moreover, some complications typically occur together, eg, choroidal effusion and flat anterior chamber, or severe endophthalmitis and visual loss, requiring different statistical approaches to compare harm between groups.
Descriptive statistics and graphical representation of scores were used. Specifically, a dot plot was used to present the frequency of each type of postoperative complication by treatment arm, as well as the statistical uncertainty (95% confidence interval) and significance of the difference. We did not adjust for multiplicity and used a significance level of .05. A spine plot was also produced to present the frequency distribution of all postoperative complications with a harm score sorted by decreasing levels of severity. We created the figures using Stata software (version 17.0; StataCorp).
Regarding early postoperative complications, 22 of 107 eyes in the tube group developed 7 complications: choroidal effusion, shallow or flat anterior chamber, wound leak, hyphema, aqueous misdirection, suprachoroidal hemorrhage, and vitreous hemorrhage. Thirty-nine of 105 patients in the trabeculectomy group developed 9 early postoperative complications, with decompression retinopathy and cystoid macular edema in addition to the 7 already mentioned. Trabeculectomy is associated with greater harm than tubes (total harm score of 212.6 vs 157.5), mainly because of wound leak and hyphema events ( Table 2 ).
|Early Complication in the TVT Study||Tube||Trabeculectomy||Severity Grade||Total Harm Score|
|Shallow or flat anterior chamber||11||10||10||10||4||44||40|
|Overall harm, per intervention||157.5||212.6|