John F. Johanson, M.D., and Alfred Rimm, Ph.D.
Department of Medicine, University of Illinois College of Medicine, Rockford, Illinois, and Department of Epidemiology and Biostatistics, Medical College of Wisconsin, Milwaukee, Wisconsin
DISCUSSION Characteristics of trials
The results of the meta-analysis can be summarized as follows. At both 3 and 12 months posttreatment, similar numbers of patients treated with IRC or rubber band ligation were asymptomatic. When stratified by severity, similar efficacy was again demonstrated. However, significantly fewer patients undergoing rubber band ligation required additional treatment for symptom recurrence. Although rubber band ligation demonstrated greater long-term efficacy, it was associated with a significantly higher incidence of posttreatment pain. Comparison of sclerotherapy with rubber band ligation likewise demonstrated an increased necessity for retreatment in the sclerotherapy group. Moreover, the overall symptomatic response, as well as the response of second-degree hemorrhoids to sclerotherapy. was not as effective as with rubber band ligation, although the former failed to achieve statistical significance. The frequency of bleeding was comparable among the three treatment techniques.
The use of meta-analysis techniques is becoming an increasingly popular tool to evaluate treatment efficacy, particularly when no individual study has demonstrated a definitive advantage. Despite its growing popularity, questions have been raised regarding its appropriateness (13). The main problem with a meta-analysis lies in the combination of multiple trials which may differ in terms of study design, patient population, methods of patient selection or randomization, and treatment technique. These problems may be minimized by determining specific inclusion criteria, and comparing results within rather than between trials (14). The advantage of the statistical technique utilized in this analysis is that it allows for variation of treatment effects across different studies by providing an objective method of weighting each trial according to the relative value of the information contained therein (11. 13).
The qualitative aspects of a meta-analysis need to be considered as much as the statistical methodology. One must be able, after close inspection of the individual trials, to justify why the combination of trials would
provide any clinically meaningful information (15). In this case, currently available studies failed to demonstrate any statistical difference among response rates to IRC, injection therapy, or rubber band ligation. Moreover, with regard to patient demographics, entrance criteria, treatment techniques, and outcome measurements, the trials were so similar that pooling their results was not a problem.
Hemorrhoids infrequently require operative intervention. With the growing popularity of newer, non-surgical treatment methods, the number of hemorrhoidectomies has declined dramatically (16). Despite numerous controlled clinical trials, however, the optimal nonoperative treatment of hemorrhoids remains unclear. This meta-analysis is the first study that has been able to demonstrate any statistically significant difference among therapeutic techniques: rubber band ligation was more efficacious, long term. Both IRC and rubber band ligation demonstrated comparable response rates at 3 and 12 months after treatment. This is not surprising, because patients were treated initially until their hemorrhoids were adequately obliterated. Significantly more individuals treated with IRC. however, required repeat therapy for symptom recurrence, suggesting that rubber band ligation is more effective long term. When sclerotherapy was compared with rubber band ligation, it was obviously less efficacious. Not only did significantly more patients require repeat therapy, but fewer individuals were asymptomatic after initial treatment. Based on these results, rubber band ligation appeared to be more efficacious than either IRC or sclerotherapy.
The better long-term efficacy in patients treated with rubber band ligation may be related to the depth of tissue destruction. The rubber band, typically placed at the upper end of the hemorrhoid, leads to necrosis of the hemorrhoidal tissue. Sloughing occurs after 7—10 days, causing moderate tissue destruction with scarring and subsequent fixation of the submucosa (1, 17). In contrast, IRC creates a small burn, resulting in minimal tissue injury (i.e., to a depth of only approximately 3 mm) (1, 2). The decreased depth of injury associated with IRC presumably results in less scarring and tissue fixation, which may explain why more individuals treated with IRC required additional therapy.
Although rubber band ligation was more efficacious, it was associated with a significantly greater frequency of posttreatment pain than was IRC. This increased occurrence of pain is not unexpected: two of the three individual trials likewise demonstrated significant differences prior to being pooled (Table 2). Pain is somewhat subjective and may not have been consistent among different individuals. However, in order to be considered a complication of treatment, all of the individual trials required that the pain be severe enough to require follow-up contact with the treating physician. The difference in posttreatment pain between IRC and rubber band ligation may also be the result of differences in depth of tissue injury. Greater tissue depth may result in increased posttreatment pain.
In addition to pain, rubber band ligation has been associated with other potentially life-threatening complications, such as tetanus or liver abscess (18). More recently, there have been several reports of pelvic eellulitis occurring after hemorrhoid banding. Although the number of cases has been relatively small, the outcome has been almost uniformly fatal, questioning the safety of rubber band ligation (18—23). Presumably, a low-grade bacterial infection occurs in the submucosa after placement of the rubber band. Upon sloughing of the hemorrhojdal tissue with associated transmural necrosis, the bacteria invade the deeper tissues, leading to cellulitis and subsequent bacteremia (22, 23). Most of these episodes occur 2—7 days after banding and are associated with the clinical triad of pain, fever, and urinary retention (21—23). Frequently, the clinical course is one of progressive deterioration which is not altered by treatment with antibiotics or even surgical exploration of the banding site (23). Despite the increasing awareness of this complication, no predisposing factors have been firmly established (22, 23). Recommendations proposed to prevent this complication include screening for evidence of immune deficiency, using enemas prior to treatment to cleanse the rectum, preprocedure antibiotics, and the use of appropriate sterile techniques. However, the incorporation of these recommendations greatly increases both the effort and expense associated with rubber band ligation. In contrast, no such life-threatening complications have been reported with the use of IRC. Moreover, the technique is easily performed without the need for antibiotics or sterile techniques.
The results of this meta-analysis demonstrated that rubber band ligation is the more efficacious treatment technique long term, i.e., fewer patients required additional therapy for symptom recurrence. This apparent therapeutic advantage must be examined in light of the increased rate and severity of complications associated with rubber band ligation. IRC is nearly as efficacious, but is significantly less painful, and consequently may be more agreeable to the patient. If efficacy was the only consideration, rubber band ligation would seem to be the optimal choice, depending upon the experience and desires of both the patient and treating physician. The most efficacious therapy, however, may not be the optimal one if the risks of potential complications outweigh the benefits of treatment. When the potentially life-threatening complications associated with rubber band ligation are considered, IRC would appear to be the optimal nonoperative therapy for first- and second-degree hemorrhoids.
REFERENCES
1.Smith LE. Hemorrhoids: A review of current techniques and management. Gastroenterol Clin North Am 1987:16:79—91.
2.Dennison AR, Wherry DC, Moms DL. Hemorrhoids: Nonoperative management. Surg Clin North Am 1988:68:1401—9.
3.Weinstein Si. Rypins EB. Houck I. et al. Single session treatment for bleeding hemorrhoids. Surg Gynecol Obstet 1987; 165:
479—82.
4.Templeton JL, Spence RAJ. Kennedy TL. et al. Comparison of infrared coagulation and rubber band ligation for first- and second-degree hemorrhoids: A randomized prospective clinical trial. Br Med J 1983:286:1387—9.
5.Ambrose NS, Hares MM. Alexander-Williams I. et al. Prospective randomized comparison of photocoagulation and rubber band ligation in treatment of hemorrhoids. Br Mcdi 1983:286:
1389—91.
6.Walker AJ, Leicester Ri. Nicholls RI, et aI. A prospective study of infrared coagulation, injection, and rubber band ligation in the treatment of hemorrhoids. mi Colorect Dis 1990:5:113—6.
7.Gartell PC. Sheridan RI. McGinn FP. Outpatient treatment of hemorrhoids: A randomized clinical trial to compare rubber band ligation with phenol injection. Br I Surg 1985:72:478—9.
8.Sim AIW, Murie IA. Mackenzie I. Comparison of rubber band ligation and sclerosant injection for first- and second-degree hemorrhoids: A prospective clinical trial. Acta Chir Scand 1981:
147:717—20.
9.Sim AJW, Murie IA. Mackenzie I. Three-year follow-up study on the treatment of first- and second-degree hemorrhoids by sclerosant injection or rubber band ligation. Surg Gynecol Obstet
1983:157:534—6.
10.Ambrose NS. Morris D. Alexander-Williams I, et al. A randomized trial of photocoagulation or injection sclerotherapy for the treatment of first- and second-degree hemorrhoids. Dis Colon Rectum 1985:28:238—40.
11.DerSimonian R. Laird N. Meta-analysis in clinical trials. Controlled Clin Trials 1986:7:177—88.
12.iohanson IF, Sonnenberg A. The prevalence of hemorrhoids and chronic constipation: An epidemiologic study. Gastroenterology
1990:98:380—6.
13.Berlin IA. Laird N. Sacks HS. et al. A comparison of statistical methods for combining event rates from clinical trials. Statist Med 1989:8:141—51.
14.Oldridge NB. Guyatt GH. Fischer ME. et al. Cardiac rehabilitation after myocardial infarction: Combined experience of randomized clinical trials. JAMA 1988:260:945—50.
15.Goodman SN. Have you ever meta.analysis you didn"t like" Ann Inter Med 1991:114:244—6.
16.Iohanson iF. Sonnenberg A. Temporal changes in the occurrence of hemorrhoids in the United States and England. Dis Colon Rectum 1991:34:585—93.
17.LefTE. Hemorrhoids: Current approaches to an ancient problem. Postgrad Med 1987:82:95—101.
18.Wechter DG. Luna GK. An unusual complication of rubber band ligation of hemorrhoids. Dis Colon Rectum 1987;30:
137—40.
19.Marshman D. Huber P1, Timmerman W, et al. hemorrhoidal ligation: A review of efficacy. Dis Colon Rectum 1989:32:
369—77.
20.Shemesh El. Kodner Ii, Fry RD. et aI. Severe complication of rubber band ligation of internal hemorrhoids. Dis Colon Rectum
1987:30:199—200.
21.Quevedo-Bonilla G, Farkas AM. Abcarian H. et al. Septic complications of hemorrhoidal banding. Arch Surg 1988:123:650—1.
22.Clay LD III, White Ii Jr. Davidson IT. et al. Early recognition and successful management of pelvic cellulitis following hemorrhoidal banding. Dis Colon Rectum 1986:29:579—81.
23.Scarpa Fl. Hillis W. Sabetta JR. Pelvic cellulitis: A life-threatening complication of hemorrhoidal banding. Surgery 1988:
103:383—5.
