**3. Results**

A total of 240 selected patients were randomized for double-blind treatment. Two patients in the Houttuynia group did not take medication, thus 120 and 118 patients in the placebo and Houttuynia groups, respectively, were analyzed. A total of 113 and 103 patients completed all study visits, respectively; this means that seven and 13 patients dropped out from each group for various reasons (Figure 1). The data in Table S3 demonstrate patient clinical characteristics at baseline. These data show no significant differences in demographic characteristics between the placebo and Houttuynia groups, comprising about 75% of women with an average age of about 38 years. Results also show no statistical differences in the data at baseline between the two groups in terms of ocular symptoms as well as in signs of MGD and DED.

**Figure 1.** Subject flow chart.

#### *3.1. <sup>E</sup>*ffi*cacy Evaluation*

#### 3.1.1. Primary Outcomes

Total and individual ocular symptom scores gradually decreased in both the groups in the following three visits (Figure 2A). There was a statistically significant difference between the two groups in the score reduction of 1 week (*p* = 0.0002) and 2 weeks (*p* = 0.0020); results reveal a 6.29 ± 0.55 (*p* < 0.0001) and 9.00 ± 0.61 (*p* < 0.0001) reduction in total symptom scores after four weeks of atomization treatment compared with baseline in the placebo and *Houttuynia* groups, respectively, as well as a statistically significant difference between the two (*p* = 0.0018) (Table 1). Results for each individual symptom score show that values for the *Houttuynia* group dropped more than in the placebo group, while reduced values for eye dryness, foreign body sensation, and photophobia were statistically different between the two groups (Table S4).


statistical difference that compared baseline and post-treatment for the same group. *p* Value for Column Factor shows the statistical difference that compared the placebo group and

Houttuynia group. Abbreviations: N, number; SEM, standard error of the mean.


#### *J. Clin. Med.* **2020**, *9*, 4022

**Figure 2.** Change in eye symptom score. (**A**) shows eye symptom scores at baseline, one week, two weeks, and four weeks. (**B**) shows the linear correlation between decreases in eye symptom score after four weeks of treatment and scores at baseline. Points represent the values in this figure. The regression line equation and the correlation coefficient R squared value are also shown in this figure. (**C**) highlights the reduction in eye symptom score after four weeks of therapy in scores <15 and ≥15 (at baseline) in two subgroups. Data are presented as mean ± standard error of the mean (SEM). \* *p* < 0.05, \*\*\* *p* < 0.001. ns, no significance.

Figure 2B reveals a positive linear correlation between eye symptom score reduction after four weeks of treatment and scores at baseline in both groups. Correlation coefficient (r) values were 0.3331 (*p* = 0.0002) and 0.6293 (*p* < 0.0001) in the placebo and *Houttuynia* groups, respectively. We then divided patients into two groups according to the severity of their eye symptoms at baseline (Figure 2C); results show that the reductions in scores after four weeks of therapy in scores < 15 of placebo and *Houttuynia* per group were 4.85 ± 0.45 and 5.32 ± 0.47, respectively, and that there were no statistical differences, while for score ≥ 15, the reductions of placebo and *Houttuynia* group were 8.53 ± 1.14 and 13.07 ± 0.90, respectively. A significant statistical difference was seen between these two groups (95% CI, range between −7.38 and −1.70; *p* = 0.0003) with an interaction *p* value = 0.0057 (Table 2). These results demonstrate that the group with symptom scores ≥ 15 compared to scores < 15 improved more significantly when treated with *Houttuynia* eye drops.

Results show that both meibum quality and TBUT also improved in both groups (Figure 3). Changes in these two indicators after therapy were also significantly statistically different in the two groups compared with the baseline (*p* < 0.0001) (Table 1). Meibum quality scores decreased after therapy; these reductions were 0.57 ± 0.10 and 0.91 ± 0.10 in the placebo and *Houttuynia* groups after four weeks. There were statistical differences between the Houttuynia and placebo groups in the meibum quality score or the reduction value of the score after four weeks of treatment (*p* = 0.0032/0.0091) (Table 1). Patients were divided into four groups on the basis of meibum quality scores at baseline. Results show that the fall in the *Houttuynia* group was higher amongs<sup>t</sup> each subgroup than in the placebo set (*p* value for interaction = 0.0282) (Table 2). Values for TBUT in the *Houttuynia* group were always higher than those in the placebo group after treatment over different lengths of time. Indeed, there was a statistically significant difference between the two groups in the TBUT of 1 week (*p* = 0.0300) and 2 weeks (*p* = 0.0199). After four weeks of treatment, TBUT was 5.60 ± 0.24 s and 6.30 ± 0.22 s (*p* = 0.0172) in the placebo and *Houttuynia* groups, respectively.




**Table 2.** *Cont.*

Score reduction value = pre-treatment value—post-treatment value. Abbreviations: N, number; SEM, standard error of the mean; CI, confidence interval.

**Figure 3.** Change in meibum quality and tear break-up time. (**A**) shows the meibum quality score reduction values after one week, two weeks, and four weeks compared with the baseline. (**B**) shows tear break-up time at baseline, one week, two weeks, and four weeks. Data are presented as mean ± SEM. \* *p* < 0.05, \*\* *p* < 0.01, ns, no significance.

#### 3.1.2. Secondary Outcomes

Secondary outcomes in the two groups also improved after four weeks of treatment (Figure 4A–F). Meibum expressibility scores in the *Houttuynia* group subsequent to treatment remained continuously low compared to the placebo group and there were no statistical differences. Patients were then divided into four groups according to their middle meibomian gland expression scores at baseline; this enabled us to demonstrate that decreases in the *Houttuynia* group among each subgroup were higher than those in the placebo group and there were statistical differences for interactions (*p* = 0.0051) (Table 2).

**Figure 4.** Changes in secondary outcome measures. (**A**–**E**) illustrate meibum expressibility scores, eyelid margin change, corneal staining, and Schirmer I test results, respectively. These data are shown at baseline as well as after one week, two weeks, and four weeks. (**C**,**F**) show reduction values for conjunctiva congestion and meibomian gland dropout scores after one week, two weeks, and four weeks compared with the baseline. Data at each point are presented as mean ± SEM. ns, no significance.

Eyelid margin change scores in the *Houttuynia* group were also continuously low compared to the placebo group, although there were no statistical differences. A positive correlation was found between decreases in eyelid margin change scores after four weeks of treatment and these scores at baseline in the two groups. The r-values in this case were 0.7707 (*p* = 0.1272) and 0.9890 (*p* = 0.0014) in the placebo and *Houttuynia* groups, respectively (Table S5). The eyelid margin score reduction of the *Houttuynia* group was more overall than that of the placebo group in each subgroup case, encapsulating a statistically significant interaction difference (*p* = 0.0158) (Table 2).

Results show that conjunctival congestion improved following atomization treatment. Decreased values of the placebo and *Houttuynia* groups were 0.61 ± 0.07 (*p* < 0.0001) and 0.63 ± 0.06 (*p* < 0.0001) in one subgroup (score = 1 at baseline) and 0.86 ± 0.17 (*p* < 0.0001) and 1.05 ± 0.20 (*p* < 0.0001) in another subgroup (score = 2 at baseline) (*p* value for interaction = 0.6094) (Table 2). The corneal staining score also reduced after treatment: values were 1.07 ± 0.37 and 1.13 ± 0.48 in the placebo and *Houttuynia* groups, respectively. There was no statistical difference in the subgroup at the last visit (corneal staining score > 1 at baseline) (Table 2). Volumes for the Schirmer I test were 7.74 ± 0.32 mm/5 min and 8.45 ± 0.50 mm/5 min (*p* value for interaction = 0.6298) in the placebo and *Houttuynia* groups, respectively (Table 1). Meibomian gland dropout score reduction gradually increased; score reduction in the *Houttuynia* group was higher than in the placebo group and there was no statistical difference.

Thus, taking all these data together, they sugges<sup>t</sup> that treatment with *Houttuynia* was more effective than treatment with placebo.

#### *3.2. Safety Assessments*

Visual acuity and IOP values for the two groups remained basically unchanged following medication. Patients in one center (Peking Union Medical College Hospital) were assessed for blood/urine routines and liver/kidney functions but all indicators remained basically unchanged following medication. Although some patients exhibited abnormal indicators after medication, there were no clinically significant differences in index values and indicators returned to normal in the follow-up.

The data presented in Table S6 reveal that seven adverse event cases were seen throughout the test. Four of these were in the placebo group (i.e., animal wool allergy, upper respiratory tract infection, blurred vision, and allergic conjunctivitis), and the incidence rate of these events was 2.50%, while three further cases were in the *Houttuynia* group (i.e., upper respiratory tract infection, conjunctivitis, and allergic conjunctivitis). The incidence rate for the cases was 2.54% and data show no statistically significant difference between the two groups. Two subjects in the placebo group and one in the Houttuynia group were withdrawn from the trial due to these adverse events. Indeed, as all adverse event cases either recovered or were relieved, we consider these irrelevant to the experimental treatment.
