1 Introduction
Threatened abortion refers to minor vaginal hemorrhage before paroxysmal lower abdominal pain or backache with no cervical dilation, an intact fetal membrane, no discharge of pregnancy tissue, and fetal survival. Threatened abortion is the most common pattern of spontaneous abortion. In clinical practice, the rate of threatened abortion is up to 20%. With lifestyle and dietary habit changes, an increasingly severe environmental pollution, increased occupational and work stress among women, more elderly parturient women and the associated decline in oocyte quality, the occurrence rate of threatened abortion has increased annually,[1] [2,3] [4,5] [6] [7] [8]
2 Methods and analysis
2.1 Search strategy
This study used a method of retrieving published studies through computer retrieval from foreign medical databases (Cochrane Library, PubMed and EMBASE) and domestic databases (SinoMed, CNKI, VIP, Wanfang Data, etc). The search strategy was based on the combination of subject terms and free terms. The Chinese search terms were “Xianzhaoxingliuchan”, “Xianzhaoliuchan”, “Liusuanmei”, “Jianbensanfen”, etc. The English search terms included “Abortion, Threatened”, “Threatened Abortion”, “Phloroglucinol”, “Magnesium sulfate”, etc. All searches were performed in Chinese and English starting from the establishment of each database to April 15, 2018. A manual search of the relevant conference literature and collected papers published after 2010 was also conducted. A supplementary search was also performed, as necessary, to ensure a comprehensive literature collection. Table 1, https://links.lww.com/MD/D35
2.2 Inclusion and exclusion criteria
2.3 Type of study
Randomized controlled trials (RCTs) regarding PHL and MS in the treatment of threatened abortion in Chinese or English were included, regardless of study blinding. This study is based on the preferred reporting items for systematic reviews and meta-analysis (PRISMA statement).
Subjects investigated:
(1) Patients were enrolled regardless of age, sex, and source. Diagnostic criteria were based on the 6th Edition of Gynecology and Obstetrics.[9]
(2) All included pregnant woman were healthy, excluding those with internal or surgical disease during pregnancy or drug combinations.
(3) The observation and control groups (CG) were comparable in gestational weeks, pregnant frequency and abortion frequency, with no statistically significant differences.
Exclusion criteria:
(1) The therapeutic measures did not meet the pre-defined inclusion criteria, and non-target drugs were used;
(2) non-RCT, such as reviews, retrospective studies, animal experiments, phase I clinical trials, etc;
(3) patients with inconsistent baseline data; and
(4) important data or reports were incomplete, and the correspondence author did not respond to inquiry.
Intervention: In the experimental group, an intravenous drip of 40 mg or 80 mg PHL was administered, the dripping speed and dosage were adjusted according to the uterine contractions, and the dosage should be no more than 400 mg within 24 hours. In the CG, after finishing an intravenous drip of 4–5 g MS within 30–60 minutes, a sustained intravenous drip of MS was maintained at a rate of 1–2 g, according to the uterine contractions, and the dosage should not exceed 30 g within 24 hours.
2.4 Outcome indicators
The following related indicators were statistically analyzed after treatment:
(1) the total effective rate (TER) (i.e., effectual, uterine contractions were obviously relieved or disappeared within 12 hours of medication; improved, uterine contractions were relieved to a certain degree within 24 hours of medication; and ineffective, symptoms showed no change or were even aggravated within 48 hours of medication); the total effective rate = (number of effectual cases + number of improved cases)/number of total cases × 100%;
(2) the rate of adverse reactions;
(3) the time to relief of uterine contractions;
(4) the complete relief of uterine contraction symptoms;
(5) the drug onset time;
(6) the duration of the drug treatment; and
(7) the length of hospital stay.
2.5 Data extraction
Two investigators extracted data based on the jointly designed literature data extraction table. The following data were extracted through a double abstraction process:
(1) basic information about the studies, including the title, author, journal and year, and selection level;
(2) study elements, including basic information about subjects, interventions, outcome measures, and design type; and
(3) experimental results, including the total number of participants completing the trial, type of data, and effective dose.
The consistency in study selection between the two evaluators was determined by the Kappa value. Any disagreement was determined by a third party or through mutual discussion. If no consensus was reached due to inadequate information, the literature involved was listed as “to be evaluated”, and a final decision was not made until enough information was obtained after contacting the author(s).
2.6 Quality evaluation
The present study was independently completed by 2 investigators using the bias risk assessment tool before crosschecking the quality of the included trials. The bias risks of the included papers were evaluated in terms of selection bias, performance bias, measurement bias, loss to follow-up bias, and report bias. Each evaluation was considered to suggest a high bias risk, low bias risk or uncertain risk. Disagreement between the 2 investigators in the literature evaluation was resolved through mutual discussion or by a third party.
2.7 Data analysis
The meta-analysis was performed using Stata 12.0 and Rev Man 5.3 software. In the present study, the enumeration data of outcome evaluation were statistically analyzed using the relative risk (RR) and 95% CI as effect sizes. Continuous data were analyzed using standardized mean difference (SMD) and 95% CI; if the same variability had different weights and measures, the data were analyzed by difference in the mean deviation (MD) and 95% CI. Variations in the included documents were called heterogeneity. The heterogeneity of these documents was assessed using the I 2 test, and P > .1 or I 2 < 25% was considered to suggest no statistically significant heterogeneity in the documents. Meta-analysis was performed using a fixed model. P ≥ .1 or 25% ≤ I 2 ≤ 50% was considered to indicate moderate heterogeneity in the documents; P < .1 or I 2 > 50% was considered to suggest high heterogeneity. If data consolidation was clinically feasible, moderate heterogeneity, and high heterogeneity were analyzed using a random effects model, and sensitivity analysis[10]
3 Results
3.1 Search results
In total, 99 relevant documents were found using the search strategy and data collection: the Cochrane Library (n = 0), PubMed (n = 0), Embase (n = 0), SinoMed (n = 59), CNKI (n = 29), WanFang Data (n = 21) and VIP (n = 24). Forty-nine papers were obtained after duplicate checking by using reference management software Note express 2. Papers were ruled out after review of their titles and abstracts due to the following reasons: not associated with the present research (n = 4), clinical and nursing document (n = 1), review (n = 1) and conference data with incomplete data (n = 3). Then, the following papers were excluded: those not meeting the inclusion criteria (n = 18) and repeated publications (n = 1). Finally, 21 RCTs were included (see Fig. 1 ).
Figure 1: The document selection flowchart.
3.2 Study characteristics
A total of 21 RCTs[11–31] [12] [13,14] [15–18] [19,20] [21–24] [25–31] Table 1 shows the basic characteristics of these included studies.
Table 1: The basic characteristics of the 21 included studies.
3.3 Summary of the quality and bias risk of the trials included
The quality of the studies included was low, and they were described as RCTs. Four of the 21 included RCTs reported the specific randomization methods, and 2 of them[21,30] [21,30] Figures 2 and 3 .
Figure 2: Risk of bias.
Figure 3: Risk of bias summary.
3.4 Outcome measures
Total effective rate: A total of 21 studies reported the total effective rate.
Rate of adverse reaction: A total of 21 documents reported the rates of adverse reactions, including flush and fever in 16 documents, dizziness and headache in 16 documents, gastrointestinal discomfort in 14 documents, nausea and vomiting in 9 documents, thirst in 9 documents, perspiration in 8 documents, insomnia in 4 documents, lack of strength in 4 documents, and chest depression in 2 documents.
Time to relieve uterine contraction: A total of 9 documents reported the index. Drug onset time: a total of 7 documents in total reported the index.
The complete relief of uterine contraction symptoms: A total of 4 documents reported this index.
Drug continuous treatment time: A total of 3 documents reported the index.
Length of hospital stay: A total of 2 documents reported the index. The abovementioned outcome measures were analyzed, as shown below.
3.5 Total effective rate
Twenty-one studies involving 1940 patients (979 cases in the experimental group and 961 cases in the control group) reported the TER.[11–31] P = .019, I 2 = 43.3%). Using a random-effects model, the meta-analysis revealed that the ORR of PHL for threatened abortion was higher than that of MS. The results of the two groups showed a statistically significant difference [n = 21, RR = 0.11, 95% CI (0.05, 0.26), Z = 5.30, P = .000] (see Fig. 4 ). A sensitivity analysis of the results of the included documents found that the combined effective dose showed no change from the original effective dose, suggesting that the meta-analysis result was relatively stable (see Fig. 5 ).
Figure 4: A forest map of the meta-analysis of the TER of phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
Figure 5: A sensitivity map of the TER of phloroglucinol versus magnesium sulfate in the treatment of threatened abortion.
3.6 Adverse reaction rate
(1) Flush and redness. Sixteen of the included studies reported flush and redness.[12–14,16,17–22,24,25,28–31] P = .535, I 2 = 0.0%). Using a random-effects model, the meta-analysis showed that the occurrence rate of flush and redness of PHL was lower than that of MS, with a statistically significant difference [n = 16, RR = 0.09, 95% CI (0.05, 0.16), Z = 7.51, P = .000]. See Figure 6 .
Figure 6: The results of the meta-analysis of the incidence rate of flush and redness associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(2) Dizziness and headache. Sixteen of the included studies reported dizziness and headache.[12–14,16–26,29,30] P = .973, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of dizziness and headache of PHL was lower than that of MS, with a statistically significant difference [n = 16, RR = 0.15, 95% CI (0.08, 0.28), Z = 6.00, P = .000]. See Figure 7 .
Figure 7: The results of the meta-analysis of the incidence rate of dizziness and headache associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(3) Gastrointestinal discomfort: Fourteen of the included studies reported gastrointestinal discomfort.[12–14,16–18,19,23–25,27,29–31] P = .856, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of gastrointestinal discomfort of PHL was lower than that of MS, with a statistically significant difference [n = 14, RR = 0.19, 95% CI (0.10, 0.36), Z = 5.10, P = .000]. See Figure 8 .
Figure 8: The results of the meta-analysis of the incidence rate of gastrointestinal discomfort associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(4) Palpitation: Twelve of the included studies reported palpitation.[12,14,16,18,22,24,26–31] P = .987, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of palpitation of PHL was lower than that of MS, with a statistically significant difference [n = 12, RR = 0.16, 95% CI (0.05, 0.44), Z = 5.07, P = .000]. See Figure 9 .
Figure 9: The results of the meta-analysis of the incidence rate of palpitation associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(5) Nausea and vomiting: Nine of the included studies reported nausea and vomiting.[17,19–23,26,28,29] P = .435, I 2 = 0.0%). By using a fixed-effects model, the meta-analysis showed that the occurrence rate of nausea and vomiting of PHL was lower than that of MS, showing a statistically significant difference [n = 9, RR = 0.26, 95% CI (0.15, 0.47), Z = 4.56, P = .000]. See Figure 10 .
Figure 10: The result of the meta-analysis about the incidence rate of nausea and vomiting associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(6) Dry mouth symptoms: Nine of the included studies reported dry mouth symptoms.[12–14,18,24,25,27,30,31] P = .265, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of dry mouth symptom of PHL was lower than that of MS, with a statistically significant difference [n = 9, RR = 0.16, 95% CI (0.05, 0.44), Z = 3.49, P = .000]. See Figure 11 .
Figure 11: The results of the meta-analysis of the incidence rate of dry mouth associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(7) Perspiration: Eight of the included studies reported PHL.[12–14,18,24,25,30,31] P = .966, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of perspiration of PHL was lower than that of MS, with a statistically significant difference [n = 8, RR = 0.11, 95% CI (0.04, 0.28), Z = 4.58, P = .000]. See Figure 12 .
Figure 12: The results of the meta-analysis of the incidence rate of perspiration associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(8) Insomnia: Among the studies included, 4 reported insomnia.[16,17,23,26] P = .560, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of insomnia of PHL was comparable to that of MS, showing no statistically significant difference [n = 4, RR = 0.42, 95% CI (0.11, 1.58), Z = 1.29, P = .198]. See Figure 13 .
Figure 13: The results of the meta-analysis of the incidence rate of insomnia associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(9) Lack of strength: Four of the included studies reported a lack of strength in patients.[16,19,22,26] P = .534, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of lacking in strength of PHL was shorter than that for MS, with a statistically significant difference [n = 4, RR = 0.34, 95% CI (0.13, 0.86), Z = 2.28, P = .022]. See Figure 14 .
Figure 14: The results of the meta-analysis of the incidence rate of a lack of strength association with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
(10) Chest tightness: Two of the included studies reported chest tightness.[16,29] P = .851, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis showed that the occurrence rate of chest tightness of PHL was lower than that of MS, with a statistically significant difference [n = 2, RR = 0.16, 95% CI (0.02, 1.31), Z = 1.71, P = .088]. See Figure 15 .
Figure 15: The results of the meta-analysis of the incidence of chest tightness associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
3.7 Time required to relieve uterine contractions
Nine of the included studies reported time required to relieve uterine contractions.[12,13,18,24,25,27,29–31] P = .158, I 2 = 32.5%). Using a random-effects model, the meta-analysis suggested that the time required to relieve uterine contractions for PHL was shorter than that for MS, with a statistically significant difference [n = 9, SMD = −0.22, 95% CI (−0.39, 0.05), Z = 2.56, P = .010]. See the results in Figure 16 . A sensitivity analysis of the analysis results of the included documents found that the combined effective did not obviously change from the original effective dose, indicating the stability of the meta-analysis result (see Fig. 17 ).
Figure 16: The results of the meta-analysis of the time required to relieve uterine contractions associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
Figure 17: The sensitivity analysis results for the time required to relieve uterine contractions associated with phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
3.8 Drug onset time
Seven of the included studies reported drug onset time.[12,13,18,24,25,30,31] P = .483, I 2 = 0.0%). Using a fixed-effects model, the meta-analysis suggested that the drug onset time of PHL was later than that of MS, with a statistically significant difference [n = 7, SMD = −0.25, 95% CI (0.09, 0.41), Z = 3.13, P = .002]. See the results in Figure 18 .
Figure 18: The meta-analysis results for the drug onset time of phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
3.9 The complete relief of uterine contraction symptoms
Four of the included studies reported the index.[16,22,23,26] P = .000, I 2 = 91.9%). Using a random-effects model, the meta-analysis suggested that the time required for the complete relief of uterine contraction symptoms was shorter for PHL than for MS, showing a statistically significant difference [n = 4, SMD = −2.08, 95% CI (−3.05, 1.12), Z = 4.24, P = .000]. See the results in Figure 19 . A sensitivity analysis of the results of the included documents found that the combined effective did not obviously change from the original effective dose, indicating the stability of the meta-analysis (see Fig. 20 )
Figure 19: The results of the meta-analysis of the complete relief of uterine contraction symptoms after phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
Figure 20: The sensitivity analysis of the results of the complete relief of uterine contraction symptoms after phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
3.10 Drug continuous treatment time
Three of the included studies reported the index.[15,20,23] P = .000, I 2 = 96.7%). Using a random-effects model, the meta-analysis suggested that the drug continuous treatment time of PHL was shorter than that of MS, with a statistically significant difference [n = 3, SMD = −1.51, 95% CI (−2.93, 0.09), Z = 2.08, P = .038]. See the results in Figure 21 .
Figure 21: The results of the meta-analysis of the drug continuous treatment time of phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
3.11 Length of hospital stay
Two of the included studies reported the index.[11,15] P = .097, I 2 = 63.7%). Using a random-effects model, the meta-analysis suggested that the length of hospital stay of PHL was comparable to that of MS, with no statistically significant difference [n = 2, SMD = −0.27, 95% CI (−0.80, 0.26), Z = 0.99, P = .323]. See the results in Figure 22 .
Figure 22: The results of the meta-analysis of the length of hospital stay after phloroglucinol vs magnesium sulfate in the treatment of threatened abortion.
3.12 Publication bias
Egger test was performed to test the publication bias of two outcome measures (TER and time required to relieve uterine contractions) in more than 7 included studies, and I 2 ≥ 30 using Stata 12.0 statistical software. The TER suggested certain publication bias (P = .013 < .05, 95% CI, 0.6333977, 4.670988). The results on the time required to relieve uterine contractions revealed no obvious publication bias (P = .555 > .05, 95% CI, −7.458848, 12.75799) (see Figs. 23 and 24 ).
Figure 23: The TER of phloroglucinol vs magnesium sulfate, according to Egger test, in the treatment of threatened abortion.
Figure 24: The time required to relieve uterine contractions associated with phloroglucinol versus magnesium sulfate, according to Egger test, in the treatment of threatened abortion.
4 Discussion
The occurrence rate of threatened abortion is high, and it is primarily early abortion. In particular, threatened abortion is more likely to happen in patients with a medical history of recurrent abortion. Some studies have shown that early vaginal hemorrhage is closely associated with perinatal hemorrhage and low-weight infant. Drug treatment can significantly lower the occurrence rate of antenatal hemorrhage, intrauterine fetal distress or low-weight infant.[32]
The most striking characteristic of PHL is that such medicine can relieve smooth muscle spasm without causing a series of adverse reactions, avoid damage to cardiovascular functions, take effect rapidly and lead to high patient compliance.[33] [34]
(1) Results: Although MS requires a shorter drug onset time than PHL in treating threatened abortion, the TEF, time required to relieve uterine contractions, complete relief of uterine contraction symptoms, duration of drug treatment and length of hospital stay are better for PHL than for MS. More importantly, the adverse reaction rate is much lower for PHL than that for MS. Most independent trials have reported contrasting effects; however, some meta-analyses have suggested high heterogeneity. Although sources of heterogeneity have been explored through sensitivity analysis, it is difficult for many sensitivity analyses to identify exact sources of heterogeneity.
(2) Quality of evidence: Evidence-based medicine highlights the importance of evidence. The workgroup responsible for evaluating quality of evidence will have an important influence on the validity of conclusion. In the present study, quality evaluation was performed using the RevMan 5.3 risk of bias table. There has been no completely reliable conclusion, however, due to the limited patient cases in the study and the poor methodological quality of the studies included.
(3) Potential bias in systematic review[35–37]
Limitations and advantages: Although PHL is expected to have a higher clinical application value than MS in the treatment of threatened abortion, its maternal-fetal safety in late threatened abortion necessitates a tremendous amount of research. To obtain more comprehensive data about the maternal-fetal influence of the drug, it is necessary to perform stratified RCT for threatened abortion of varying degrees at different gestational ages, improve the outcomes of mother and infant after the administration of tocolytic agent and provide long-term follow-up for newborn growth and development. Conclusions of meta-analyses are more convincing than the conclusions of individual clinical trials and serve as a reference for clinical decision-making. However, a meta-analysis still cannot substitute for large-scale, multicenter clinical RCT.
5 Conclusion
PHL can significantly control the symptoms of threatened abortion and shorten the treatment time without increasing the occurrence rate of adverse reactions compared to MS. However, the conclusion of the present study is limited by the small sample size. Thus, the results must be further verified by large-sample, multicenter, high quality clinical RCT.
Author contributions
Conceptualization: Jingzhi Guan.
Data curation: Shaofei Yuan, Jingzhi Guan.
Formal analysis: Fengli Gao, Zhong Xin.
Investigation: Haijun Guo, Suqin Shi.
Methodology: Shaofei Yuan, Fengli Gao, Lei Shi, Xia Yang, Jingzhi Guan.
Resources: Shaofei Yuan, Fengli Gao, Zhong Xin, Lei Shi.
Software: Shaofei Yuan, Fengli Gao, Xia Yang, Jingzhi Guan.
Supervision: Jingzhi Guan.
Writing – original draft: Shaofei Yuan, Fengli Gao, Zhong Xin, Haijun Guo, Suqin Shi, Lei Shi, Xia Yang.
Writing – review & editing: Jingzhi Guan.
Jingzhi Guan orcid: 0000-0002-3086-477X.
