Facial nerve palsy (FNP) is characterized by facial weakness caused by various diseases. Among them, Bell’s palsy (BP) and Ramsay-Hunt syndrome (RHS) are the most common causes of FNP [1,2]. The prognosis is generally good in most patients; however, in patients with severe nerve damage, long-term treatment is required, and permanent sequelae can remain in 15–31% of the patients [1,3,4]. Unpredictable long-term prognosis and the possibility of sequelae may worsen patients’ quality of life because of social or psychological problems [5-7].

Although oral corticosteroids are usually recommended as conventional treatments based on their medical evidence, as regards the recovery stage of FNP, no conventional treatments are recommended [8,9]. Therefore, various complementary and alternative treatments have been proposed to support the limitations of conventional treatments. In Korea, where modern medicine and traditional Korean medicine (TKM) coexist, most patients receive both treatments simultaneously. TKM treatment of FNP usually consists of acupuncture, herbs, moxibustion, electroacupuncture, and needle-embedding [10].

For the prognostic prediction of FNP, electrophysiological techniques, such as electromyography, nerve conduction studies, and blink reflex (BR) tests, are commonly used [4,11,12]. The BR is a polysynaptic reflex that consists of an afferent pathway through the trigeminal nerve and an efferent pathway through the facial nerve [13]. Once the supraorbital nerve is stimulated by an electric pulse, an ipsilateral early response (R1) and bilateral late response (R2) including ipsilateral late response (IR2) and contralateral late response (CR2) are observed. These responses reflect the status of the trigeminal nerve, facial nerve, and brainstem [14]. Therefore, latencies to both R1 and R2 tend to be delayed or disappear in FNP [15].

Several studies have shown that BR is a good prognostic indicator in patients with FNP [16,17]. However, no studies with large sample sizes have analyzed long-term prognosis in relation to specific treatment programs. In our previous study, patients with normal or delayed BR showed significantly better recovery rates than those without a response. However, as the sample size was too small to represent the prognosis of patients with FNP and given the limitations of statistical methods, the possible association between the responses was not analyzed [18].

In the present study, we investigated the long-term prognosis of patients with FNP who received integrative TKM treatment using the results of their BR test within a 1-year follow-up period based on a larger clinical dataset. Using multivariable analysis, we also analyzed how each reflex affects the recovery rate.

1. Study design and ethical consideration

This was a retrospective observational study of the medical records of patients using survival analysis methods. Ethical approval was obtained from the Institutional Review Board of Kyung Hee University Korean Medicine Hospital at Gangdong (IRB no. KHNMCOH 2016-03-002). A written informed consent was obtained from the patients for the publication of this study.

2. Participants

Medical records of all patients who were diagnosed with BP or RHS between June 1, 2006, and June 30, 2014, and received integrative TKM treatment were retrospectively retrieved from the electrical medical record system.

The exclusion criteria were as follows: (1) visit to the hospital 6 days from onset, (2) age ≤ 18 years, (3) recurrent facial palsy, (4) bilateral facial palsy, (5) pregnancy, and (6) without records of BR tests.

3. Procedure of treatment and assessment

The integrative treatment program consisted of modern medicinal and TKM treatments. Modern medicinal treatments included oral steroids, antiviral agents, and physical therapy. TKM treatment included manual acupuncture, pharmacopuncture, electroacupuncture, moxibustion, cupping, and herbal medicine. The BR test was conducted 14 (± 2) days after onset, and the House-Brackmann (HB) grade was assessed at every visit until complete recovery was attained. Trained TKM doctors performed the assessment and reached a consensus through discussion (Fig. 1).

Figure 1. Integrative traditional Korean medicine treatment program.

4. Variables

All data, including general characteristics, assessments, and BR results within 1 year from onset, were collected by two independent researchers.

HB grading was used to assess the severity of FNP. Generally, HB grade I indicates complete recovery to normal, and HB grade II signifies good recovery to mild dysfunction. Thus, the times taken to reach HB grades II and I from FNP onset were counted and analyzed.

In the BR test, the results of R1, IR2, and CR2 were used as covariates. Based on the existence of each latency, participants were divided into response (with latency) and nonresponse (without latency) types.

5. Statistical methods

Recovery rates of HB grades II and I were calculated using Kaplan-Meier curves, and differences between types were evaluated using log-rank tests. Multivariable proportional hazard Cox regression analysis was performed for variables identified by univariate analysis (p < 0.2). All analyses were performed using PASW Statistics 18 (SPSS Inc.). The cutoff value for significance was set at p = 0.05.

1. Patients

A total of 926 patients were screened, of which 796 fulfilled the criteria. After checking the results of the BR test, patients were classified into response or nonresponse types of R1, IR2, and CR2 (Fig. 2). Table 1 presents the general characteristics of the patients.

Figure 2. Flow diagram of patient selection and response classification. R1, ipsilateral early response; IR2, ipsilateral late response; CR2, contralateral late response.

Table 1 . General characteristics of patients.

Characteristic	Average of number of patients (n = 796)
Age (y)	48.31 ± 15.14
Sex
Male	351 (44.1)
Female	445 (55.9)
Side of palsy
Left	391 (49.1)
Right	405 (50.9)
Diagnosis
Bell’s palsy	726 (91.2)
Ramsay-Hunt syndrome	70 (8.8)
House-Brackmann grade upon onset
Average grade	4.03 ± 0.78
Grade II	24 (3.0)
Grade III	154 (19.3)
Grade IV	394 (49.5)
Grade V	220 (27.6)
Grade VI	4 (0.5)

Values are presented as number (%) or mean ± standard deviation..

2. Analysis of the 1-year recovery rate in all patients

Among the 796 patients with FNP analyzed retrospectively, 538 (67.6%) attained recovery to HB grade II, whereas 315 (39.6%) achieved recovery to HB grade I (Table 2). The 1-year recovery rate to HB grade II and I were 91.9% (95% confidence interval [CI], 90.2–93.6) and 73.6% (95% CI, 70.8–76.4), respectively, in the Kaplan-Meier curves of all patients (Fig. 3).

Figure 3. Kaplan-Meier curves of the 1-year recovery rate to HB grades II (A) and I (B) in all patients. The values at the bottom of the horizontal axis of each graph represent the cumulative recovery rates. HB, House-Brackmann.

Table 2 . Summary of recovery and follow-up.

Summary	Recovery to HB grade II	Recovery to HB grade I
Recovery events	538 (67.6)	315 (39.6)
Censored patients	250 (31.4)	455 (57.2)
Follow-up (d)	25 [17–41]	39 [27–69]
Time of recovery (d)	28 [19–88]	77 [42–OR]
Range of time to recovery (d)	Up to 290	Up to 344

Values are presented as number (%) or median [interquartile range]..

HB, House-Brackmann; OR, out of range..

3. Analysis of the 1-year recovery rate by each blink reflex response

The 1-year recovery rate to HB grade II was 98.5% (95% CI, 97.2–99.8) in R1 response and 82.6% (95% CI, 78.9–86.3) in R1 nonresponse (log-rank p < 0.001; Fig. 4A). R2 response was analyzed by separating IR2 and CR2. The 1-year recovery rate to HB grade II was 98.3% (95% CI, 96.9–99.7) in IR2 response and 85.0% (95% CI, 81.8–88.2) in IR2 nonresponse (log-rank p < 0.001; Fig. 4B). The 1-year recovery rate to HB grade II was 96.5% (95% CI, 94.8–98.2) in CR2 response and 83.6% (95% CI, 79.7–87.5) in CR2 nonresponse (log-rank p < 0.001; Fig. 4C).

Figure 4. Kaplan-Meier curves of the 1-year recovery rate to HB grade II by R1 (A), IR2 (B), and CR2 (C) and to grade I by R1 (D), IR2 (E), and CR2 (F). The values at the bottom of the horizontal axis of each graph represent the cumulative recovery rates. HB, House-Brackmann; R1, ipsilateral early response; IR2, ipsilateral late response; CR2, contralateral late response. ***p-value < 0.001 by the log-rank test.

The 1-year recovery rate to HB grade I was 85.0% (95% CI, 81.8–88.2) in R1 response and 57.6% (95% CI, 52.7–62.5) in R1 nonresponse (log-rank p < 0.001; Fig. 4D). The 1-year recovery rate to HB grade I was 87.9% (95% CI, 84.5–91.3) in IR2 response and 59.5% (95% CI, 55.1–63.9) in IR2 nonresponse (log-rank p < 0.001; Fig. 4E). The 1-year recovery rate to HB grade I was 79.1% (95% CI, 75.8–82.4) in CR2 response and 63.0% (95% CI, 58.0–68.0) in CR2 nonresponse (log-rank p < 0.001; Fig. 4F).

4. Multivariable analysis of prognosis indicators associated with the recovery rate

The median recovery time to HB grade II in R1 response was 21 (interquartile range [IQR], 14–34) days compared with 71 (IQR, 32–181) days in R1 nonresponse. The median recovery time to HB grade II in IR2 response was 20 (IQR, 14–33) days compared with 59 (IQR, 27–167) in IR2 nonresponse. The median recovery time to HB grade II in CR2 response was 23 (IQR, 15–50) days compared with 62 (IQR, 29–172) in IR2 nonresponse. The multivariable Cox regression analysis indicated that R1 (hazard ratio [HR], 2.41; 95% CI, 2.23–2.59; p < 0.001) was a significant independent predictive factor for a higher recovery rate to HB grade II (Table 3).

Table 3 . Multivariable Cox regression analysis of recovery to House-Brackmann grade II.

Factor	Median [IQR] (d)	Hazard ratio	95% CI	p-value
R1		2.41	2.23–2.59	< 0.001*
Response	21 [14–34]
Nonresponse	71 [32–181]
IR2		1.45	1.28–1.62	0.031*
Response	20 [14–33]
Nonresponse	59 [27–167]
CR2		1.05	0.90–1.19	0.747
Response	23 [15–50]
Nonresponse	62 [29–172]

IQR, interquartile range; CI, confidence interval; R1, ipsilateral early response; IR2, ipsilateral late response; CR2, contralateral late response..

*p-value < 0.05..

The median recovery time to HB grade I in R1 response was 53 (IQR, 34–109) days compared with 207 (IQR, 71–out of range [OR]) in R1 nonresponse. The median recovery time to HB grade I in IR2 response was 55 (IQR, 34–92) days compared with 194 (IQR, 57–OR) in IR2 nonresponse. The median recovery time to HB grade I in CR2 response was 52 (IQR, 33–89) days compared with 194 (IQR, 57–OR) in IR2 nonresponse. Multivariable Cox regression analysis showed that R1 (HR, 2.21; 95% CI, 1.96–2.45; p = 0.001) and IR2 (HR, 1.88; 95% CI, 1.65–2.10; p = 0.006) were significant independent predictive factors for a higher recovery rate to HB grade I (Table 4).

Table 4 . Multivariable Cox regression analysis of recovery to House-Brackmann grade I.

Factor	Median [IQR] (d)	Hazard ratio	95% CI	p-value
R1		2.21	1.96–2.45	0.001*
Response	53 [34–109]
Nonresponse	207 [71–OR]
IR2		1.88	1.65–2.10	0.006*
Response	55 [34–92]
Nonresponse	194 [57–OR]
CR2		0.78	0.59–0.98	0.220
Response	52 [33–89]
Nonresponse	194 [57–OR]

IQR, interquartile range; CI, confidence interval; R1, ipsilateral early response; IR2, ipsilateral late response; CR2, contralateral late response; OR, out of range..

*p-value < 0.05..

In this study, the long-term prognosis of patients with FNP who received integrative TKM treatment was presented based on actual patient data. Through these data, we found that the presence or absence of the BR response affected the difference in their recovery rates.

The HB grade, which was used to assess recovery in this study, is one of the most frequently used scales in FNP [19]. HB grade II refers to mild dysfunction of facial movements, which usually indicates good recovery, and HB grade I refers to normal facial functioning in all facial areas, indicating complete recovery [20]. Therefore, HB grades II and I are clinically significant treatment goals [21-23].

Our data showed that the estimated recovery rates to HB grades II and I were 91.9% and 73.6%, respectively. Peitersen [1] reported that the complete recovery rates without the treatment of BP and RHS were 71% and 21%, respectively. However, no data presents the baseline severity for comparison [1]. Several clinical trials and observational studies on the effect of steroid and antiviral agents have reported that the recovery rates to HB grade II in BP and RHS were 81.7–89.0% and 58.7%, respectively. Compared with 4.03 of the baseline average HB grade, the baseline average HB grade in those studies was 3.59–3.75 in patients with BP and 4.13 in those with RHS [22,24-26].

Although significant comparison was difficult because of demographic heterogeneity, we found that integrative TKM treatment might increase the recovery rate of FNP.

The BR test is a simple, noninvasive, and inexpensive test. Because a delay in or loss of the BR response is associated with lesions of the facial nerve in patients with FNP, it may be used as a diagnostic method [15]. Several clinical studies have demonstrated the diagnostic value of the BR test in terms of its sensitivity and specificity [12,27,28]. The BR response is a reliable factor, particularly in the acute stage between 7 and 10 days after onset [29-31]. In our previous study, although the normal and delayed types in the BR test showed good prognoses (100% recovery rate) within 3 months of onset, the nonresponse type had a relatively poor prognosis (87.8% recovery rate) until 6 months after onset [18].

Other studies have proved the prognostic value of the BR test in patients with FNP [16,17]. However, these studies enrolled < 100 participants, and the follow-up period was not long enough (52 days [mean time] and 3 months in the two studies, respectively). Our previous study is also limited by its small sample size (n = 163), short follow-up period (up to 6 months), and statistical methods used. Therefore, we followed up a larger number of participants (n = 796) for up to 1 year to verify the diagnostic value of the BR test in patients with FNP.

Previously, we divided patients with FNP into three groups (normal latency; normal/delayed latency; delayed/no signal; absent). We grouped patients with a delayed signal in the BR test, which fulfill the standards of the normal BR test suggested by Oh [32], into the normal group, with a delayed signal more than the normal group into the delayed group, and with no signal into the absent group.

With time, as treatment progressed, the results of the normal and delayed groups showed no significant difference. Both groups recovered very well, whereas the absent group recovered incompletely. Therefore, in this study, patients were divided into two groups (response and nonresponse types) according to the presence of latency.

R1, IR2, and CR2 abnormalities in FNP are commonly due to axonal loss of the efferent pathway. Therefore, these factors showed similar values as prognostic factors. Considering that these factors are non-independent, multivariable analysis was performed, and R1 was found to be a significant factor for both HB grades II (HR, 2.41) and I (HR, 2.21). In addition, IR2 was a significant factor in HB grade I (HR, 1.88).

The strengths of this study lie in its therapeutic homogeneity, as it was restricted to patients receiving identical treatment beginning at the acute stage, its long follow-up period of 1 year, and usability of data, which can be used to estimate the recovery rate with time. Furthermore, by applying simple criteria, such as response vs. nonresponse, we demonstrated the practicality of BR in clinical practice.

However, this study has some limitations. Given the retrospective nature, the date when the BR test was performed may not exactly match the date on which the patient had recovered. Furthermore, follow-up could not be completed in many cases, leading to data limitations despite trying to overcome the same using censored data in survival analysis methods. Therefore, a prospective external validation study is necessary to confirm data accuracy. In addition, it focused only on the indicators (R1, IR2, and CR2) of the BR test without considering other patient factors, such as baseline characteristics and treatment variables. Although most patients received treatment according to the same integrative treatment program (acupuncture, pharmacopuncture, electroacupuncture, moxibustion, cupping, herbal medicine, and modern medicinal treatments included oral steroids, antiviral agents, and physical therapy), there may be variations in the frequency, number, and specific types of treatments based on the severity of the patient’s condition. Therefore, future prospective studies with controlled treatment methods as well as additional studies considering various factors, such as age, sex, demographic characteristics, comorbidities, and other neurophysiological tests, are needed.

In summary, BR in the acute stage was significantly associated with the recovery rate in patients with FNP. The R1 response was a significant prognostic indicator for the recovery to HB grades II and I. Thus, it may be used to predict good or poor prognoses based on the response of the BR in actual clinical situations.

Conceptualization: BG. Data curation: JK. Formal analysis: BG. Funding acquisition: SN, YB. Investigation: BG. Project administration: SN, YB. Supervision: SN, YB. Writing – original draft: BG. Writing – review & editing: JP, JK, DK, SN, YB.

The authors declare no conflict of interest.

This study was supported by the Traditional Korean Medicine R&D program funded by the Ministry of Health and Welfare through the Korea Health Industry Development Institute (KHIDI) (RS-2020-KH087887).

The present study was approved by the Institutional Review Board of Kyung Hee University Hospital at Gangdong (IRB no. KHNMCOH 2016-03-002). A written informed consent was obtained from the patients for the publication of this study.