Introduction

Perianal fistulas (PF) remain a surgical treatment challenge in colorectal practice due to high recurrence rates and the risk of postoperative incontinence. Treatment of PF depends on the relation to the anal sphincter complex. Traditionally, the fistulas are classified according to Parks’ classification (transsphincteric, intersphincteric, extrasphincteric, and suprasphincteric) [1]. However, treatment of a PF depends mostly on the amount of the sphincter complex that is involved. Therefore, the classification in high and low fistulas has become more common. High perianal fistulas (HPF) are defined as involving the middle and/or upper third parts of the anal sphincter complex. Low perianal fistulas (LPF) are defined as involving the lower one third of the anal sphincter complex or not involving the sphincter muscles at all. The major disadvantage of surgical treatments for PF is the chance of anal sphincter dysfunction, like soiling, incontinence for gas, and liquid and/or solid stool.

At first, fistulotomy was the main surgical procedure for all PF. However, in HPF, this procedure results in a high risk of anal sphincter dysfunction [2]. Consequently, this led to the development of surgical techniques for HPF aimed at sparing anal sphincter function and improving recurrence rates [3–5]. Nowadays, new techniques are still being developed for HPF, since the ideal treatment has not been found yet [6–9].

For LPF, fistulotomy is still considered to be the golden standard of surgical treatment. Success rates have been reported as high as 80–100 % [10–12]. Unfortunately, incontinence rates after fistulotomy might be higher than we assume. This is indicated by several studies that report on incontinence, ranging from soiling to major incontinence, up to 41 % [13, 11, 14]. In contrast to treatments for HPF, development of new treatments for LPF has not been advancing as fast, although some newer techniques have been described [15, 16]. With the development of new techniques lagging behind for LPF and the risks of incontinence being real, we considered the evaluation of our results after fistulotomy appropriate. We therefore assessed healing rate, continence status, and quality of life after fistulotomy for LPF in our region.

Methods

We searched the databases and patient files of our university medical center and its six affiliated hospitals for patients with LPF between 2008 and 2013. A LPF was defined as a fistula traversing the lower one third of the anal sphincter complex or a superficial fistula not involving the striated sphincter complex.

The primary outcome of this study was closure rate after fistulotomy. The secondary outcomes were continence levels, quality of life, time to healing, and time to recurrence. Demographic data, disease etiology, type of fistula, fistula anatomy, and previous fistula surgery were all extracted from patient files. Fistula type was defined as a primary or a recurrent fistula. Both types of fistula were included. Prior fistula surgery was defined as any surgery regarding the fistula, including seton placement. Abscess drainage was excluded.

Closure of the fistula was defined as a visibly closed wound without external fistula openings and without discharge during manual compression. Recurring of these symptoms was defined as a recurrent fistula. If the fistula did not close within 3 months, it was defined as a persisting fistula. The time from operation to closure of the fistula was defined as time to healing. Time to recurrence was defined as the time from healing to the time that the patient file described recurring symptoms.

Information regarding fistula anatomy (location of internal and external openings, location of fistula tract, involvement of the sphincter complex, etc.) was obtained from additional imaging like magnetic resonance imaging (MRI) or from descriptions of the fistula tract in operative reports. MRI was generally only performed in case a HPF was suspected preoperatively.

To identify factors associated with a recurrence, patients with a recurrent LPF were compared to the primary healed patients using a Cox regression analysis. A p value < 0.05 was considered statistically significant. Additionally, a Kaplan-Meier analysis was performed to compare cumulative proportions of patients with recurrence.

Long-term postoperative outcomes regarding incontinence rate and quality of life were obtained using two questionnaires. To assess the long-term postoperative continence status, we used the Vaizey questionnaire [17]. Quality of life was assessed using the Short Form Health Survey 36 (SF-36) [18].

Telephone interviews were used to evaluate the status of the fistula for patients not being in outpatient clinical follow-up anymore. Preoperative and postoperative interviews and examination were performed by the operating surgeons or by the resident surgeon involved with the operation. Either the operating surgeon or the resident surgeon involved with this study performed the telephone interviews. This study was performed according to national and local medical and ethical laws and guidelines. The local medical ethical committees approved this study.

Operative procedure

Patients were positioned in the lithotomy position under general or spinal anesthesia. First, a probe was inserted in the fistula tract, identifying the external and trying to identify the internal fistula openings. An anal retractor was only used if visibility was not good enough to examine the fistula and perform the procedure. The involved medical centers used different types of retractors. If a LPF was confirmed based on the amount of sphincter complex involved after probing the tract, a lay-open procedure of the tract was performed guided by the probe. The fistula track was curetted after the fistulotomy. If a HPF was suspected, the fistulotomy was not performed.

Results

Between January 2008 and June 2013, 537 patients were treated by fistulotomy for a LPF. Mean age at time of operation was 45.5 years (range 5–97); 379 (70.6 %) patients were male. Median total follow-up was 38.9 months (6.0–74.8).

Of the 537 patients, 369 (68.7 %) had a primary fistula, 163 (30.4 %) had a recurrent fistula, and in 5 (0.9 %) patients, the type of fistula was unknown. Fistula anatomy according to Parks’ classification is shown in Table 1.

Table 1 Distribution of fistula anatomy
Full size table

Fistula etiology was cryptoglandular in 357 patients (66.5 %). Thirty-six patients (6.7 %) had a fistula related to Crohn’s disease. For the remaining patients, the disease etiology could not be identified.

MRIs were made preoperatively in 266 patients (49.5 %). Of those 266 MRIs, 242 (91.0 %) confirmed the presence of a LPF. Fistula tract and sphincter involvement was also described in most of the operative reports.

A recurring fistula was seen in 88 (16.4 %) patients of whom 40 (7.4 %) had persisting fistulas. In four (0.7 %) patients, it was unclear if a recurrence occurred (Table 2). This resulted in a success rate of 83.6 %. The median healing time was 37 days (range 6–99). The median time until recurrence was 90 days (range 7–1085).

Table 2 Number and percentages of recurrence after fistulotomy
Full size table

Recurrences were managed by conservative treatment in 12 (13.6 %) patients, with a seton in 16 (18.2 %) patients, a fistulotomy in 47 (53.4 %) patients, a seton + fistulotomy in 1 (1.2 %) patient, and a mucosal advancement flap in 4 (4.5 %). The method of treatment could not be identified in eight (9.1 %) patients. Results after surgery for recurrent LPF are displayed in Table 3.

Table 3 Healing after secondary treatment
Full size table

After the secondary treatment, another 40 fistula remained closed, reaching a secondary healing rate of 90.3 %. The cumulative healing rate at 5 years is 0.81 (95 % confidence interval (95 % CI) 0.71–0.85), as can be seen in the Kaplan-Meier survival curve (Fig. 1).

Fig. 1
figure 1

Healing of fistulas. Remaining cases: 0 months, 533; 20 months, 338; 40 months, 207; 60 months, 72

Full size image

No significant relation was found between gender, Crohn’s disease, an unidentified internal opening or anterior location of the internal opening, and the development of a recurrence. However, a significant relation (p < 0.05) for the development of a recurrence was found if it concerned a recurring fistula (Table 4).

Table 4 Cox regression analysis for recurrence of fistula
Full size table

Preoperatively, seven (1.3 %) patients had known incontinence problems. Five of these patients had complaints of soiling, one patient was incontinent for gas, and one patient was incontinent for solid stool.

The questionnaires were mailed to all 537 patients. After a month, we sent a second questionnaire and contacted the patients that did not respond after the first mailing. In total, 374 patients responded to our questionnaire, resulting in a response rate of 69.6 %. Thirty-four (6.3 %) patients replied that they were not willing to participate. This resulted in 63.3 % completion of the questionnaires. Two (0.4 %) patients turned out to be deceased.

We compared the results of our SF-36 survey to the validated values for the general population [19]. The results of our SF-36 survey are displayed in Fig. 2. Compared to the national values, no significant difference was found in quality of life (Table 5).

Fig. 2
figure 2

Outcomes of the SF-36 questionnaire. PF physical function, RP role physical, BP bodily pain, GH general health, VT vitality, SF social functioning, RE role emotional, MH mental health

Full size image
Table 5 SF-36 outcome: comparison of quality of life
Full size table

The mean Vaizey score was 4.67 (SD 4.80). Major incontinence, defined as a Vaizey score of >6 [3], was seen in 95 (28.0 %) patients. Incontinence levels and their relation to previous surgery are displayed in Tables 6 and 7. We did not find a difference in continence status between patients with and without a preoperative MRI. Data on the number of vaginal deliveries and nulliparous status were unfortunately not available, and its effect on postoperative continence status could therefore not be evaluated.

Table 6 Incontinence levels
Full size table
Table 7 Vaizey score related to previous surgery
Full size table

Discussion

This study on long-term results after fistulotomy for LPF demonstrates a primary healing rate of 83.6 % and a secondary healing rate of 90.3 %. This is consistent with other reports that have demonstrated a high healing rate (>90 %) [20]. Using a Kaplan-Meier analysis, we found a healing rate after 5 years of 0.81 (95 % CI 0.71–0.85).

The healing time after fistulotomy varied widely between several reports, ranging from a median of 3 up to 11.6 weeks [21, 20]. A maximum healing time of 12 weeks is regarded the standard for LPF [22, 21]. The 5-week healing time in our study is therefore similar to healing times in other studies.

Although we found a recurrence rate within the reported recurrence range, it cannot go unnoticed that our primary rate (16.4 %) was slightly higher than the average total recurrence rate of 10 % [23, 2, 20]. By analyzing center-specific data, we found the highest recurrence rate in our university medical center. The higher recurrence rate could be explained by the third line referral position for LPF. There might also be a difference in recurrence per surgeon. Furthermore, it cannot be discarded that the inclusion of patients with Crohn’s disease might contribute to a higher recurrence rate, although these patients were only treated if the Crohn’s disease was in remission. Similar effects are described in reports that also do not exclude Crohn’s disease [10, 24].

Lack of identifying the internal opening [2], female gender, recurrent fistula surgery [24], and location off the midline [2] are reported to be factors associated with the development of a recurrence. In our series, the only significant factor associated with a recurrence was a recurring fistula. The reasons for our different findings are not clear.

Postoperative quality of life results were not significantly different from the general population. Furthermore, the long-term incontinence levels were low with a mean Vaizey score of 4.66 (SD 4.75). However, we did find major incontinence in 95 (28.0 %) patients [3].

Incontinence rates vary significantly between studies from 2.4 up to 64 % [24, 25], but most series report incontinence rates ranging between 30 and 40 % [23, 10, 11]. Similar to these reports, most incontinence problems consisted of soiling or flatus. Standardized incontinence scales are unfortunately rarely used in older studies. Reports that do use a standardized scale report average Vaizey scores of >6.5, which is higher than our series (4.66) [11]. Major incontinence is usually described as a Vaizey score >6 [3]. The Vaizey questionnaire consists of six questions scored 1 (mild incontinence) to 4 (severe incontinence) with a maximum score of 24. If a patient has only minor complaints, it may not affect daily functioning, although it may result in a score >6 on the questionnaire. Therefore, it is important to distinguish a very high Vaizey score from a Vaizey score of just >6 with only minor complaints.

The reason for the high percentage of major incontinence in our study remains unclear; however, several possibilities come to mind. Even though no significant relation was found between the Vaizey score and the number of previous operations, it is likely that previously operated patients have a more damaged sphincter complex and a higher cumulative risk of postoperative incontinence. Secondly, if during the operation a HPF is suspected, even if the MRI contradicts, a fistulotomy is not performed and a seton is placed. This, however, does not exclude the possibility that some of the fistulotomies were performed in patients that had a HPF, making the risk of postoperative incontinence significantly higher. Besides, about half of the patients did not have a preoperative MRI and the surgeon had to rely on operative findings.

As previously described, we unfortunately miss the data on vaginal delivery status for our female patients. It is known that this can negatively affect sphincter function. Unfortunately, we cannot say if this influenced our postoperative results.

Another factor that has to be taken into account is the pre- and postoperative bowel function, specifically for the patients with CD, because irritable bowel with frequent diarrhea may also affect the continence status. However, the number of patients, both pre- and postoperatively, with these symptoms was limited to only a couple. The high level of postoperative continence issues can therefore not contribute to this.

Although we found a major incontinence in 28.0 % of the patients, quality of life was not significantly different from the general population. Long-term postfistulotomy quality of life is more likely affected by more recent life events instead of a limited surgical procedure years ago [26]. We performed a long-term quality of life survey to assess the current health status of our patients. However, after assessing the results, the usefulness of these quality of life results can be discussed. Either patients are satisfied with their quality of life regardless of their continence status or the quality of life questionnaire is not the right instrument to use.

A limitation of this study was the retrospective design. Due to this design, pre- versus postoperative comparison of secondary outcomes was not possible. Another limitation was the use of telephone interviews to assess long-term follow-up, which may have resulted in bias and maybe missed recurrences.

Fistulotomy is still considered to be the golden standard for the treatment of LPF. However, incontinence rates may be higher than we expected. The amount of sphincter involved by the fistula might be underestimated during operation. Therefore, we believe that it is important to lower the chance of anal sphincter dysfunction as much as possible. Patients should be optimized for the operation to lower the chance of a needed reoperation. Active Crohn’s disease needs to be treated first. Besides, patients should be encouraged to stop smoking, since we know smoking influences wound healing [27], although contradictory results are found for the influence of smoking on the healing of PF [28, 11].

Preoperative imaging, either using MRI or (endoanal) ultrasound, of the fistula tract might be useful to determine the amount of sphincter muscle involved. This could be advantageous during preoperative planning. Because we know that the higher the internal fistula opening is located, and the more sphincter muscle is involved, the higher the risk of postoperative incontinence will be. In cases of a relatively high internal opening and large involvement of sphincter muscle, it could be better to change tactics and choose an operative technique that is developed for HPF. However, there is no data supporting this statement.

Fistulotomy seems to be a reasonably good treatment for LPF with a healing rate of 0.81 (95 % CI 0.71–0.85) after 5 years and a minimal effect on continence status (mean Vaizey score 4.67). However, major incontinence is still reported in 28.0 % of patients. Although this rate might be an overestimation on its influence in daily functioning, these levels of incontinence are disturbing for a small procedure regarded to be the golden standard for LPF. While many new techniques are being developed for HPF, the development of new LPF treatments is lagging behind when this could lower major incontinence levels.