Literature search
Searching literature was conducted from January to March 2021 and updated in December 2022 (Supplementary Appendix 1, shows a detailed search strategy for each database with appropriate search terms).
Description of the studies
Included studies
In addition to an up-to-date electronic search of thirteen databases, we performed manual searches on the reference lists of the included studies and verification of the search for finished studies of published protocols. This led to the discovery of a total of 7086 articles, which were then electronically examined for duplication using the Endnote tool and found to contain 1604 duplicates. The remaining 5482 articles were then extracted to an Excel file for additional title abstract screening, leaving only 40 articles for full-text screening.
Finally, 26 articles with 581 participants were qualitatively retrieved (Supplementary Appendix 2). 23 articles discussed various types of asthma, two discussed COPD and one discussed the two illnesses. In terms of population, 22 publications on adults and 4 research on children were done (82 children included), Table 1 shows the summary of narrative synthesis. While 16 articles completed the meta-analysis process, Fig. 1 shows PRISMA flowchart.
Quality assessment
Figure 2 shows the summary and individual quality of the included studies in this meta-analysis using the ROB2 tool, we found that most of the included studies had a low risk of selection bias (90% had random sequence), performance bias (70% had double blinding), attrition bias (70% had complete outcome data, and reporting bias (90% good reporting of outcomes), while nearly 75% of studies have an unclear risk of selection bias (allocation concealments not reported), and detection bias (blinding of the outcome, not Cleary mentioned), but due to the cross over design of 50% of studies they have a higher risk in other risks of bias.
Quality assessment of the included studies in the meta-analysis.
Primary outcome (lung function improvement indicator)
Forced expiratory volume at 1 s% (FEV1%)
Overall 8 studies were included in this meta-analysis with a total of 220 participants, all reported studies are for asthmatic patients. The pooled effect size SMD (Standard Mean Difference) of FEV1% was statistically significant at 2.7 (95% CI 1.00; 4.39, Tau2 = 3.2415, I2 = 85%, GRADE high), which means that using inhaled heparin make a large effect on FEV1% in asthmatic patients. Then, the Baujat plot was performed to check for outliers and influential causing heterogeneity, we found that Ahmed 1993, Tutuoglu 2001, and Tranfa 2001 are outliers regarding their contribution to heterogeneity (9 to > 15%), but we couldn’t consider them as influential due to their small sample size, (Supplementary Appendix 3, Fig. S1A–C). Figure 3 shows the SMD after performing a leave-one sensitivity analysis of Ahmed 1993, the result pooled effect slightly affected but still large and statistically significant SMD 2.15 (95% CI 0.81;3.50, Tau2 = 1.358) and heterogeneity decreased to I2 = 78%.
Forest plot showing pooled SMD of FEV1% after omitting Ahmed 1993 by leave-one sensitivity analysis.
Subgroup analysis
Supplementary Appendix 2, Fig. S1E subgroup by heparin type shows that UFH is highly effective for improving FEV1% than LMWH (UFH SMD 3.23, 95% CI 0.88; 5.57, I2 = 86%, LMWH SMD 1.38, 95% CI − 0.58; 3.34, I2 = 0%). Figure S1F subgroup by provocation using allergen statistically significant difference for improving FEV1% than exercise (SMD 2.46, 95% CI 0.22; 4.70, I2 = 83%). With different bronchoconstriction inducers. Figure S1G,H,J,N show subgrouping by heparin time, frequency, dose respectively, and treatment type, where studies not mentioned heparin timing had the statically significant effect (SMD 1.22, 95% CI 0.41; 2.03, I2 = 0%), Using heparin once with a dose of 1000IU/Kg alone had the higher and statistically significant SMD 4.3 (95% CI 1.15; 7.45), SMD 3.5 (95% CI 0.47; 6.53), SMD 2.95 (95% CI 0.99; 4.9), respectively. Regarding study publication year category and quality, there was no statistically significant effect while for study design RCT cross-over had SMD 3.23 (95% CI 0.24; 6.22), Fig. S1K–M.
Besides Meta-regression was performed to detect the most predominant predictors of heterogeneity, predictors were the type of heparin and frequency (heparin used once a day for 5 days, ß =  − 3, 95% CI − 6.9; − 0.08), which can explain some of this heterogeneity (R2 = 34.85%).
Forced expiratory volume at 1 s FEV1 (ml)
A total of 7 studies with 218 participants were included in this meta-analysis, all reported studies are for asthmatic patients except Shute for COPD patients. The pooled effect size (Standard Mean Difference) of FEV1 ml was statistically insignificant SMD 2.12 (95% CI − 1.49; 5.72, Tau2 = 13.48, I2 = 91%, GRADE Moderate), besides a substantial heterogeneity that affects the pooled estimate (Supplementary Appendix 2, Fig. S2A). Surprisingly after performing leave one sensitivity analysis by omitting Tuluoglu, 2001, the SMD become statistically significant with a large contribution to the outcome favoring the use of heparin SMD 0.8 (95% CI 0.12; 1.47, Tau2 = 0.252), in addition to a significant decrease in heterogeneity I2% = 62%, Fig. 4. The Baujat plot also confirms this as shown in (Supplementary Appendix 2, Fig. S2B), Tutuoglu is considered as an outlier and slightly influences the pooled effect.
Forest plot illustrates the pooled SMD of FEV1 (ml) after performing leave-one sensitivity analysis, omit Tutuoglu 2001.
Subgroup analysis
Supplementary Appendix 2, Fig. S2E,F subgroup by heparin and disease type shows no statistically significant difference, while Fig. S2F subgroup by provocation type shows that studies do not use allergen-induced bronchoconstriction had a statistically significant effect (SMD 1.43, 95% CI 1.02; 1.83, I2 = 0%). Figure S2H,J,K,O show subgrouping by heparin time, frequency, dose, and treatment type respectively, where subgroup analysis didn’t show any statistically significant difference pooled estimates except for studies not mentioning heparin timing had the statically significant effect and heparin as add on the drug (SMD 1.43, 95% CI 1.02; 1.83, I2 = 0%). Regarding study design and quality, there was no statistically significant effect while for the study publication year category (2015–2021) SMD 1.43 (95% CI 1.02; 1.83, I2 = 0%). Figure S2L–N.
Then, we conducted a meta-regression including study design and publication year category as predictors, and we found that 100% of the intra-studies heterogeneity contributed to study type either parallel or crossover (parallel, ß = 11.33, 95% CI 10.23; 12.4) and year of publication category (the publication year (2015–2021), ß =  − 10.67, 95% CI − 11.97; − 9.5, (2001–2005), ß =  − 0.68, 95% CI − 1.1; − 0.24, (1996–1999), ß =  − 0.8, CI − 1.2; − 0.4).
Secondary outcomes
Other lung function improvement indicators
Forced vital capacity (FVC)
Supplementary Appendix 3, Fig. S4 shows that only 2 studies reported Forced Vital Capacity including 28 COPD and 38 asthmatic patients, in each study alone SMD has a high statistically significant effect favoring using inhaled heparin, but we couldn’t use pooled estimate due to large heterogeneity I2% = 96%. Shute 2017, Tultuoglu 2001 (SMD 1.22 and 11.42, 95% CI 0.4; 2.04 and 8.64; 14.2) respectively.
Peak expiratory flow rate (PEFR)
Similar to FVC, only 2 studies reported PEFR with 108 asthmatic participants but here one favors heparin use Tutuoglu 2001 (SMD 15.23, 95% CI 11.57; 18.89), while Motamed 2021 favors the placebo (SMD − 1.86, 95% CI − 2.43; − 1.3), I2% = 99%, (Supplementary Appendix 3, Fig. S5).
Provocation concentration of allergen causing 20% fall of FEV1% (PC20) for asthmatic patients
The overall 9 studies with 238 asthmatic participants were included in this meta-analysis. The pooled effect size of PC20 is not statistically significant with a small contribution, SMD = 0.38 (95% CI − 1.2; 1.95, Tau2 = 3.87, I2% = 93%), besides a substantial heterogeneity that affects the pooled estimate. The Baujat plot also confirms this as shown in, Duong 2008 is considered an outlier and slightly influences the pooled effect (Supplementary Appendix 2, Fig. S3A,B). Upon omitting Duong 2008, leave-one sensitivity analysis, the SMD is still statistically insignificant with a high contribution to the outcome favoring the use of heparin (SMD 0.91, 95% CI − 0.15;1.96, I2% = 85%, Tau2 = 1.39), Fig. 5.
Forest plot illustrates the pooled SMD of PC20 after performing leave-one sensitivity analysis, omit Duong, 2008.
Subgroup analysis
Supplementary Appendix 2, subgroup by heparin type (Fig. S3E), allergen type (Fig. S3F), frequency (Fig. S3G), heparin timing (Fig. S3H), year of publication (Fig. S3K), and study design (Fig. S3L) shows no statistically significant difference. While the subgroup by heparin dose (Fig. S3J) shows that UFH dose 1000 Iu/kg had a highly statistically significant effect on PC20 (SMD 1.27, 95% CI 0.26; 2.29, I2 = 74%), and the subgroup by the quality of the study (Fig. S3M) shows that Abstract only studies had a high statistically significant effect on PC20 (SMD 2.06, 95% CI 0.49; 3.63, I2 = 34%). Meta-regression could explain 93.19% of between-studies heterogeneity by 2 predictors of quality of the study and heparin type (Good quality, ß =  − 2.76, 95% CI − 4.4; − 1.2, Poor quality, ß =  − 1.25, 95% CI − 2.7; − 0.2, UFH, ß = 4.9, 95% CI 2.7; 7.27).
Area under the curve (AUC)
Three studies with 78 asthmatic participants reported AUC, Duong 2008, Tahir 1999, and Diamant 1996. Although the individual SMD was statistically significant and favored using heparin (SMD 0.92, 0, 1.92; 95% CI 0.23–1.61, 0, 0.68–3.16) respectively, the pooled estimate is not statistically significant, either by omitting Tahir 1999 no difference was found, and also seems to favor the heparin side (SMD 1.28; 95% CI − 4.85 to 7.41, I2% = 45%) (Supplementary Appendix 3, Fig. S6).
Effect of inhaled heparin on coagulation factors and bleeding
Activated partial thromboplastin time (APTT)
Pavord61, Ashoor40, Lee64, Ahmed63, and Duong42 reported that inhaled heparin did not affect coagulation and partial thromboplastin time.
Plasma anti-factor-X activity
Tahir54 reported no effect of inhaled heparin on plasma factor X (Antifactor X was 0.5 IU/ml before and 1 h after nebulized heparin), while Kwasniewski52 reported that Nandaparin increased anti-factor activity more than the control group (P < 0.05), and no effect on other coagulation parameters was observed 8–9 days after treatment.
Adverse events of inhaled heparin
Headache
Two studies reported headache, Ceyhan62, said that 61.5% of participants had a headache after inhalation of heparin that self-resolved 1–2 h later, while Duong42, only one participant suffer from mild self-limited headache.
Bronchospasm
Ceyhan62 had one participant withdraw from the trial after heparin inhalation due to suffering from bronchospasm.
Serious adverse events
Shute41 and Duong42 reported that they didn’t find any serious adverse events in their participants.
Others
Duong42 reported that total lymphocyte count and eosinophil were increased in the placebo arm than in the heparin arm.
Adverse events
Polosa57, Tranfa53, Diamant58, Garrigo59, Hong60 and Motamed39 they reported that adverse events not seen at all.
Publication bias
FEV1% publication bias was assessed in 2 different ways; visual inspection shows an aggregation of most studies on the left side from the diagonal line indicating asymmetry in the funnel plot. Then further confirmation using Eggers’ test: Linear regression test of funnel plot asymmetry (Test result: t = 3.25, df = 8, P-value = 0.017). Rank correlation test using Begg and Mazumdar’s test is not applicable because of the number of studies less than ten. For FEV1 ml, visual inspection shows an aggregation of most studies on the left side from the diagonal line (Supplementary Appendix 2, Figs. S1O, S2P).