Number of arms

Overall, 17 (20%) were split‐mouth studies, 38 (44%) trials were two‐arm studies, 21 (24%) were three‐arm studies, 10 (11%) were four‐arm studies, and one trial described a five‐arm study (Demir 2007).

Treatments and medicaments

The 87 trials described 125 different combinations of pulp treatment (pulpotomy, pulpectomy or direct pulp capping) and medicament.

Duration of follow‐up

The duration of follow‐up was fixed in 78 (90%) trials. Data were assessed at six months in 70 (80%) trials, at 12 months in 59 (68%) trials and at 24 months in 24 (28%) trials.

Rubber dam

In 67 trials (77%), treatments were completed with rubber dam isolation. In four trials, either rubber dam or cotton rolls were used (Ozalp 2005; Waterhouse 2000; Zealand 2010; Zurn 2008); in four trials, cotton rolls were used (Alaçam 1989; Fallahinejad Ghajari 2013; Markovic 2005; Sonmez 2008); and in 12 trials there was insufficient information to determine if a rubber dam or cotton rolls were used (Arikan 2016; Cantekin 2014; Chen 2015; Demir 2007; Goyal 2014; Goyal 2016; Liu 2011; Mortazavi 2004; Nadkarni 2000; Naik 2005; Niranjani 2015; Pinky 2011).

Pulp access

Caries were removed prior to pulpal access in 68 (78%) trials. Pulp was accessed with a high‐speed bur in 37 (43%) trials, a slow‐speed bur in six trials (Aminabadi 2010; Aminabadi 2016; Markovic 2005; Moretti 2008; Ramar 2010; Shabzendedar 2013), a high‐speed followed by a slow‐speed bur in six trials (Casas 2004; Celik 2013; Cuadros‐Fernández 2016; Doyle 2010; Fernández 2013; Nguyen 2017), a combination of slow‐speed bur and excavator in one trial (Nadkarni 2000), a combination of high‐speed bur and round carbide bur in three trials (Fallahinejad Ghajari 2013; Kang 2015; Lourenço 2015a), a high‐speed followed by a combination of slow‐speed bur and round carbide bur in one trial (Ulusoy 2014a), a number 557 round bur in one trial (Kalra 2017), or a handpiece with a round bur (with no precision) followed by a high speed and round carbide bur in one trial (Oliveira 2013a).

Removal of coronal pulp involved an excavator in 35 (40%) trials, a combination of slow‐speed bur and excavator in 11 (13%) trials, a slow‐speed bur in 11 (13%) trials, a high‐speed bur in two trials (Ibricevic 2000; Markovic 2005), a high speed bur followed by excavator in one trial (Celik 2013), round burs numbers ½ and ¼ or excavator in one trial (Grewal 2016), or a number 6 carbide round bur in one trial (Shabzendedar 2013).

In the case of pulpectomy, complete extirpation of the pulp involved barbed broaches, K files or H files.

Haemostasis

Before application of the pulpotomy or direct pulp capping medicament, haemostasis of the pulp stumps was achieved with either dry or moistened (water or saline) cotton wool pellets in 53 trials (61%).

In two trials, haemostasis of the pulp stumps was achieved with techniques that differed according to the group: in Coser 2008, haemostasis was obtained with dry cotton wool pellets in the pulpotomy group and with moistened cotton pellets with saline in the pulpectomy group; whereas, in Doyle 2010, haemostasis was obtained with saline/water flush in the three ferric sulphate arms and with dry cotton pellets in the MTA arm.

In five trials, haemostasis was obtained using other techniques: a sterile cotton pellet soaked in 1.25% sodium hypochlorite solution and placed over the exposure site for 62 seconds without pressure (Demir 2007), a damp sterile cotton pellet (Farsi 2005), a dry sterile cotton pellet and electrofulguration (Fishman 1996), a cotton pellet moistened with 10% sodium hypochlorite maintained for one minute in one group (no haemostasis in the other group, Nakornchai 2010), a cotton pellet moistened with 3% hydrogen peroxide (Shumayrikh 1999), or diode laser (810 nm with the pulsed contact mode of application for two seconds delivered by optical fibre tip and 1.5 W power) (Niranjani 2015). The other trials involved no haemostasis technique or no details about haemostasis.

Irrigation

Irrigation was performed in 48 (55%) trials (in one group only in three trials (Nguyen 2017; Prabhakar 2008; Saltzman 2005). Irrigants used were:

• saline in 33 (38%) trials;

• 0.9% saline solution (Shumayrikh 1999);

• 0.5% saline (Coser 2008);

• 5.25% sodium hypochlorite and distilled water (Al‐Ostwani 2016);

• 2.5% sodium hypochlorite in three trials (Chen 2015; Nakornchai 2010; Trairatvorakul 2008);

• 2.5% sodium hypochlorite and saline in four trials (Arikan 2016; Chandra 2014; Nadkarni 2000; Rewal 2014);

• 1% sodium hypochlorite and saline or water in three trials (Aminabadi 2016; Bezgin 2016; Subramaniam 2011);

• 5% sodium hypochlorite followed by a 0.5% metronidazole solution (Ozalp 2005);

• a mixture of 2.25% sodium hypochlorite (1.5 mL) and 0.12% chlorhexidine gluconate (1.5 mL) (Ramar 2010);

• alternating irrigations of sterile saline and a chlorhexidine solution (Garrocho‐Rangel 2009);

• saline and finally with 2% chlorhexidine (Pramila 2016);

• 3% hydrogen peroxide and sterile saline (Alaçam 1989); and

• water in three trials (Nguyen 2017; Vargas 2006; Yildirim 2016).

Number of visits

Only one intervention session for both groups was necessary for 62 (71%) trials. In six trials (7%), the number of visits was one in one treatment group and two (Akcay 2014; Ansari 2010; Bezgin 2016; Kang 2015; Noorollahian 2008; Sonmez 2008) or one or two (Ibricevic 2000; Nakornchai 2010) in the other groups.

The number of visits was:

• one or two in both groups in two trials (Cantekin 2014; Ozalp 2005);

• two in 12 (14%) trials (Chen 2015; Goyal 2014; Goyal 2016; Grewal 2016; Jayam 2014; Kusum 2015; Mortazavi 2004; Nadkarni 2000; Naik 2005; Shumayrikh 1999; Subramaniam 2009; Subramaniam 2011);

• three in two trials (Arikan 2016; Pinky 2011); and

• four in one trial (Coser 2008).

Description of medicaments used

Intermediate restoration

Formocresol techniques were followed by placement of:

• ZOE in 14 trials (16%);

• ZOE and IRM in eight trials (Agamy 2004; Eidelman 2001; Farsi 2005; Fei 1991; Fernandes 2015; Fuks 1997; Holan 2005; Moretti 2008);

• IRM in six trials (Ansari 2010; Dean 2002; El Meligy 2016; Huth 2005; Shabzendedar 2013; Zealand 2010);

• two successive IRM temporary restorations (formocresol dressing changed after seven days, ZOE placement at third visit) (Coser 2008);

• calcium hydroxide liner and glass‐ionomer cement (Markovic 2005);

• Cavit (Sabbarini 2008);

• ZOE and zinc phosphate cement (Sonmez 2008); and

• ZOE and glass ionomer cement in two trials (Durmus 2014; Yildirim 2016).

The type of intermediate restoration after formocresol technique was not specified in two trials (Alaçam 2009; Goyal 2014).

MTA techniques were followed by placement of:

• IRM in 15 trials (17%);

• ZOE in nine trials (Erdem 2011; Goyal 2016; Jayam 2014; Naik 2005; Noorollahian 2008; Olatosi 2015; Sonmez 2008; Subramaniam 2009; Tuna 2008);

• glass‐ionomer cement in six trials (Celik 2013; Kalra 2017; Liu 2011; Rajasekharan 2017; Uloopi 2016; Yildirim 2016);

• metal‐reinforced glass ionomer cement (Arikan 2016);

• reinforced glass ionomer cement (Bezgin 2016); and

• ZOE and glass ionomer cement (Kusum 2015).

There was no intermediate restoration in three trials after MTA technique (Aeinehchi 2007; Fallahinejad Ghajari 2013; Saltzman 2005).

Calcium hydroxide techniques were followed by placement of:

• ZOE in five trials (Aminabadi 2010; Nadkarni 2000; Niranjani 2015; Tuna 2008; Waterhouse 2000);

• IRM in five trials (Coser 2008; Huth 2005; Moretti 2008; Oliveira 2013a);

• ZOE and IRM (Fernandes 2015);

• glass‐ionomer cement in four trials (Celik 2013; Grewal 2016; Liu 2011; Markovic 2005); and

• dentine adhesive (Garrocho‐Rangel 2009).

The type of intermediate restoration after calcium hydroxide technique was not specified (Alaçam 2009), and there was no intermediate restoration in three trials (Demir 2007; Sonmez 2008; Zurn 2008).

Ferric sulphate techniques were followed by placement of:

• IRM in three trials (Doyle 2010; Huth 2005; Vargas 2006);

• IRM and glass ionomer cement (Cantekin 2014);

• ZOE in six trials (Casas 2004; Erdem 2011; Goyal 2016; Gupta 2015; Ozmen 2017; Yadav 2014);

• ZOE and IRM in two trials (Fei 1991; Fuks 1997);

• ZOE and glass‐ionomer cement (Durmus 2014); and

• MTA followed by a layer of light‐cured glass ionomer (Nguyen 2017).

Eugenol‐free ferric sulphate was followed by placement of:

• Cimpact S (Doyle 2010);

• ZOE then IRM for five days over the ZOE paste for very uncooperative children (Ibricevic 2000);

• calcium hydroxide and glass‐ionomer cement (Markovic 2005); and

• ZOE and zinc phosphate cement (Sonmez 2008).

Ferric sulphate/MTA technique was followed by IRM (Doyle 2010).

Calcium hydroxide cement and calcium sulphate hemihydrate techniques were followed by glass‐ionomer cement restoration (Ulusoy 2014a).

Portland cement technique was followed by:

• IRM in three trials (Lourenço 2015a; Oliveira 2013a; Sakai 2009); and

• ZOE and glass ionomer cement (Yildirim 2016).

NaOCl techniques were followed by IRM in three trials (Fernández 2013; Shabzendedar 2013; Vargas 2006).

The techniques for 2% glutaraldehyde + eugenol + IRM and 2% glutaraldehyde + calcium hydroxide were followed by placement of compomer in Shumayrikh 1999. There was no intermediate restoration after 2% unbuffered glutaraldehyde + ZOE and 2% unbuffered glutaraldehyde + calcium hydroxide techniques in Alaçam 1989. Goyal 2016 reported use of 2% buffered glutaraldehyde followed by ZOE.

EMD technique was followed by ZOE and glass ionomer cement in Yildirim 2016. There was no intermediate restoration in Garrocho‐Rangel 2009 and Sabbarini 2008.

CEM technique was followed by IRM in Malekafzali 2011. There was no intermediate restoration after CEM in Fallahinejad Ghajari 2013 and Khorakian 2014.

Diode laser technique was followed by:

• ZOE in three trials (Gupta 2015; Niranjani 2015; Yadav 2014); and

• ZOE and glass ionomer cement (Durmus 2014).

Er:YAG laser technique was followed by IRM (Huth 2005).

Electrosurgery techniques were followed by placement of:

• ZOE in five trials (Bahrololoomi 2008; Fishman 1996; Gupta 2015; Khorakian 2014; Yadav 2014);

• IRM (Dean 2002); and

• calcium hydroxide (Fishman 1996).

LLLT technique was followed by glass ionomer cement in Uloopi 2016 and by ZOE + IRM in Fernandes 2015.

Biodentine technique was followed by ZOE and glass ionomer cement in Kusum 2015, by IRM in Cuadros‐Fernández 2016, and by glass ionomer cement in Rajasekharan 2017. There was no intermediate restoration following Biodentine in three trials (El Meligy 2016; Grewal 2016; Niranjani 2015).

Propolis technique was followed by ZOE and glass ionomer cement in Kusum 2015.

Aloe vera technique was followed by a layer of collagen sponge and glass ionomer cement in Kalra 2017.

Ankaferd Blood Stopper technique was followed by IRM and glass ionomer cement in Cantekin 2014, and by ZOE in Ozmen 2017.

Tempophore was followed by glass ionomer cement in Rajasekharan 2017.

Vitapex technique was followed by IRM placement (Nakornchai 2010) and glass‐ionomer cement (Pramila 2016). There was no intermediate restoration in three trials (Mortazavi 2004; Ozalp 2005; Trairatvorakul 2008).

Metapex technique was followed by ZOE placement in Ramar 2010, ZOE and glass ionomer (Miracle mix) (Subramaniam 2011), and glass ionomer (Al‐Ostwani 2016).

ZOE + Metapex and RC Fill techniques were followed by placement of ZOE (Ramar 2010).

Endoflas technique was followed by placement of ZOE and glass ionomer (Miracle mix) (Subramaniam 2011), ZOE (Rewal 2014), and glass ionomer (Al‐Ostwani 2016).

ZOE technique was followed by glass ionomer in four trials (Al‐Ostwani 2016; Nguyen 2017; Pramila 2016; Subramaniam 2011), thick ZOE paste (Rewal 2014), and no intermediate restoration in six trials (Chandra 2014; Erdem 2011; Mortazavi 2004; Nadkarni 2000; Ozalp 2005; Trairatvorakul 2008).

IRM technique was followed by placement of metal‐reinforced glass ionomer cement (Arikan 2016).

The type of intermediate restoration after calcium hydroxide/iodoform techniques was not specified in Alaçam 2009.

3Mix and ciprofloxacin + ornidazole + minocycline techniques were followed by ZOE or IRM in three trials (Aminabadi 2016; Nakornchai 2010; Pinky 2011).

3Mixtatin and simvastatin were followed by IRM in Aminabadi 2016.

The gutta‐percha/AH‐Plus technique was followed by reinforced glass ionomer cement in Bezgin 2016.

There was no intermediate restoration following ozonated sesame oil‐ZO technique in Chandra 2014.

ZOP technique was followed by glass ionomer in Al‐Ostwani 2016.

There was no intermediate restoration after ZOE technique (Casas 2004), acetone‐based total‐etch adhesive (Demir 2007), Sealapex technique (Ozalp 2005), calcium hydroxide technique (Ozalp 2005), and antibacterial mix technique (Prabhakar 2008).

Final restoration

Final restorations after placement of formocresol were:

• stainless‐steel crown in 18 (21%) trials;

• amalgam in four trials (Bahrololoomi 2008; Erdem 2011; Markovic 2005; Sonmez 2008);

• glass‐ionomer cement and stainless‐steel crown in three trials (Sabbarini 2008; Saltzman 2005; Subramaniam 2009);

• amalgam or stainless‐steel crown in three trials (Ansari 2010; Ibricevic 2000; Ozmen 2017);

• amalgam or glass‐ionomer cement (Aeinehchi 2007);

• glass‐ionomer cement or stainless‐steel crown if the restoration was not satisfactory (Coser 2008);

• composite or stainless‐steel crown (Holan 2005);

• glass‐ionomer cement or composite or stainless‐steel crown (Huth 2005);

• glass‐ionomer cement in two trials (Fernandes 2015; Moretti 2008);

• stainless steel crown and/or glass ionomer restoration and silver amalgam (Jayam 2014);

• glass‐ionomer cement or composite or amalgam, and stainless‐steel crown if indicated (Waterhouse 2000).

Final restoration after placement of formocresol was not mentioned in Alaçam 1989.

Final restorations after placement of MTA were:

• stainless‐steel crown in 21 trials (24%);

• amalgam in five trials (Aeinehchi 2007; Celik 2013; Fallahinejad Ghajari 2013; Sonmez 2008; Tuna 2008), followed by a light‐cured fissure sealant material in one trial (Celik 2013);

• glass‐ionomer cement in two trials (Moretti 2008; Sakai 2009);

• resin modified glass ionomer cement (Oliveira 2013a);

• composite resin in two trials (Bezgin 2016; Liu 2011);

• composite resin or amalgam or stainless (Holan 2005);

• amalgam or stainless‐steel crown (Malekafzali 2011);

• glass ionomer and amalgam (Aminabadi 2016);

• stainless steel crown and/or glass ionomer restoration and silver amalgam (Jayam 2014);

• glass‐ionomer cement and stainless‐steel crown (Subramaniam 2009).

Final restorations after placement of calcium hydroxide were:

• stainless‐steel crown in four trials (Alaçam 2009; Aminabadi 2010; Garrocho‐Rangel 2009; Nadkarni 2000);

• amalgam in four trials (Celik 2013; Demir 2007; Markovic 2005; Tuna 2008) followed by a light‐cured fissure sealant material (Celik 2013);

• glass‐ionomer cement or stainless‐steel crown if the restoration was not satisfactory (Coser 2008);

• glass‐ionomer cement or composite or stainless‐steel crown (Huth 2005);

• glass‐ionomer cement in two trials (Fernandes 2015; Moretti 2008);

• resin modified glass ionomer cement (Oliveira 2013a);

• glass‐ionomer cement and amalgam (Sonmez 2008);

• glass‐ionomer cement or composite or amalgam, and stainless‐steel crown if indicated (Waterhouse 2000);

• glass‐ionomer cement and stainless‐steel crown (Zurn 2008)

• composite resin (Liu 2011); and

• nanohybrid composite resin (Grewal 2016).

Final restorations after placement of ferric sulphate were:

• stainless‐steel crown in eight trials (Cantekin 2014; Doyle 2010; Durmus 2014; Fei 1991; Fuks 1997; Goyal 2016; Gupta 2015; Vargas 2006);

• amalgam in three trials (Erdem 2011; Markovic 2005; Sonmez 2008);

• amalgam or stainless‐steel crown in three trials (Casas 2004; Ibricevic 2000; Ozmen 2017);

• glass‐ionomer cement (Yadav 2014);

• acid etch resin (Nguyen 2017); and

• glass‐ionomer cement or composite or stainless‐steel crown (Huth 2005).

Final restorations after placement of ZOE were amalgam in three trials (Erdem 2011; Mortazavi 2004; Ozalp 2005), stainless‐steel crown in four trials (Nadkarni 2000; Rewal 2014; Subramaniam 2011; Trairatvorakul 2008) and acid‐etch resin (Nguyen 2017).

Final restoration after placement of 2% glutaraldehyde + eugenol + IRM and 2% glutaraldehyde + calcium hydroxide techniques was stainless‐steel crown in one trial (Shumayrikh 1999). Final restoration after placement of 2% unbuffered glutaraldehyde + ZOE and 2% unbuffered glutaraldehyde + calcium hydroxide was not mentioned in one trial (Alaçam 1989). Final restoration after placement of 2% buffered glutaraldehyde was stainless steel crown in one trial (Goyal 2016).

Final restoration after diode laser was stainless steel crown in three trials (Durmus 2014; Gupta 2015; Niranjani 2015) and glass‐ionomer cement in one trial (Yadav 2014).

Final restorations after electrosurgery were stainless‐steel crowns in four trials (Dean 2002; Fishman 1996; Gupta 2015; Khorakian 2014), amalgam in one trial (Bahrololoomi 2008) and glass‐ionomer cement in one trial (Yadav 2014).

Final restoration after placement of acetone‐based total‐etch adhesive, acetone‐based total‐etch adhesive, total‐etching with 36% phosphoric acid and self etch adhesive system was composite in one trial (Demir 2007).

Final restorations after placement of EMD were stainless‐steel crowns in two trials (Garrocho‐Rangel 2009; Yildirim 2016), and glass‐ionomer cement and stainless‐steel crowns in one other trial (Sabbarini 2008).

Final restorations after Er:YAG laser were glass‐ionomer cement and composite or stainless‐steel crowns in one trial (Huth 2005).

Final restoration after LLLT was stainless‐steel crown in one trial (Uloopi 2016) and glass ionomer cement in one trial (Fernandes 2015).

Final restorations after placement of CEM were amalgam or stainless‐steel crowns in one trial (Malekafzali 2011), amalgam in one trial (Fallahinejad Ghajari 2013), and stainless steel crown in one trial (Khorakian 2014).

Final restoration after placement of Portland cement was:

• glass‐ionomer cement in two trials (Lourenço 2015a; Sakai 2009);

• resin‐modified glass‐ionomer cement in one trial (Oliveira 2013a); and

• stainless steel crown in one trial (Yildirim 2016)

Final restoration after placement of calcium hydroxide cement and calcium sulphate hemihydrate was amalgam followed by a light‐cured fissure sealant material in one trial (Ulusoy 2014a).

Final restoration after placement of Biodentine was stainless steel crown in five trials (Cuadros‐Fernández 2016; El Meligy 2016; Kusum 2015; Niranjani 2015; Rajasekharan 2017) and nanohybrid composite resin in one trial (Grewal 2016).

Final restoration was stainless steel crown after placement of propolis (Kusum 2015), Tempophore (Rajasekharan 2017) and aloe vera (Kalra 2017).

Final restoration after Ankaferd Blood Stopper was stainless steel crown in one trial (Cantekin 2014), and amalgam or stainless steel crown in one trial (Ozmen 2017).

Final restorations after placement of Vitapex or Metapex were:

• amalgams in two trials (Mortazavi 2004; Ozalp 2005);

• glass‐ionomer cement and stainless‐steel crowns in one trial (Nakornchai 2010); and

• stainless‐steel crowns in five trials (Al‐Ostwani 2016; Pramila 2016; Ramar 2010; Subramaniam 2011; Trairatvorakul 2008).

Final restorations after placement of 3Mix were glass‐ionomer cement and stainless‐steel crowns in two trials (Nakornchai 2010; Pinky 2011), and glass ionomer and amalgam in one trial (Aminabadi 2016).

Final restoration after placement of 3Mixtatin and simvastatin were glass ionomer and amalgam in one trial (Aminabadi 2016).

Final restoration after placement of Sealapex or calcium hydroxide was amalgam in one trial (Ozalp 2005).

Final restorations after placement of ciprofloxacin + metronidazole + minocycline and ciprofloxacin + ornidazole + minocycline were glass‐ionomer cement and stainless‐steel crowns in one trial (Pinky 2011).

Final restorations after antibacterial mix technique were made of glass‐ionomer cement and composite resin in one trial (Prabhakar 2008).

Final restoration after gutta‐percha/AH‐Plus technique was resin composite in one trial (Bezgin 2016).

Final restoration after placement of calcium hydroxide/iodoform (Alaçam 2009), ZOE (Al‐Ostwani 2016; Casas 2004; Chandra 2014; Pramila 2016), IRM (Arikan 2016), ozonated sesame oil‐ZO (Chandra 2014), ferric sulphate: MTA (Doyle 2010), RC Fill (Pramila 2016; Ramar 2010), ZOE + Metapex (Ramar 2010), Endoflas or Endoflas‐CF (Al‐Ostwani 2016; Rewal 2014; Subramaniam 2011), ZOP (Al‐Ostwani 2016) and 3% or 5% NaOCl (Fernández 2013; Shabzendedar 2013; Vargas 2006) was stainless‐steel crown in 10 trials.

Clinical failure

At six months, data were extractable from 13 RCTs totaling 1048 teeth. In 10 of the 13 trials, there was no clinical failure in any of the participants regardless of the intervention. From the three remaining trials (N = 394 participants), the pooled results showed no statistically significant difference in clinical failure with MTA compared with formocresol. The pooled risk ratio (RR) was 0.37 (95% confidence interval (CI) 0.07 to 1.89). At 12 months, data were extractable from 12 RCTs totaling 740 teeth. In seven of the 12 trials, there was no clinical failure in any of the participants regardless of the intervention. From the five remaining trials, the results showed a statistically significant difference (RR 0.31, 95% CI 0.10 to 0.93) with no evidence of statistical heterogeneity among included trials (I² = 0%). At 24 months, data were extractable from nine RCTs totaling 548 teeth. In five of the nine trials, there was no clinical failure in any of the participants regardless of the intervention. From the four remaining trials, the results showed no statistically significant difference (RR 0.47, 95% CI 0.18 to 1.19; Analysis 1.1).

The results showed no statistically significant difference at any time point in clinical failure when full strength formocresol and 1:5 diluted formocresol results were not pooled (Analysis 2.1; Analysis 3.1).

Radiological failure

At six months, data were extractable from 12 RCTs totaling 922 teeth. In eight of the 12 trials, there was no radiological failure in any participants regardless of the intervention. From the four remaining trials, the pooled results showed a statistically significant difference in radiological failure with MTA compared with formocresol. The pooled RR was 0.38 (95% CI 0.17 to 0.86) with no evidence of statistical heterogeneity among included trials (I² = 0%). At 12 months, results were similar with seven trials providing data for a pooled RR of 0.41 (95% CI 0.19 to 0.89) with no evidence of statistical heterogeneity among included trials (I² = 0%). At 24 months, data were extractable from nine RCTs totaling 548 teeth, with eight trials providing data. The results showed a statistically significant difference (RR 0.42, 95% CI 0.22 to 0.80) with no evidence of statistical heterogeneity among included trials (I² = 18%; Analysis 1.2).

The results showed a statistically significant difference at 12 and 24 months in radiological failure for MTA compared with full strength formocresol (Analysis 2.2) (no statistically significant difference at 6 months); the results showed no statistically significant difference at any point in radiological failure for MTA compared with 1:5 diluted formocresol (Analysis 3.2).

Overall failure

At six months, data were extractable from six RCTs totaling 328 teeth. In four of the six trials, there was no overall failure in any of the participants regardless of the intervention. From the two remaining trials, the results showed no statistically significant difference in overall failure with MTA compared with formocresol (RR 0.23, 95% CI 0.04 to 1.32). Results were similar at 12 months with four trials providing data for an overall pooled RR of 0.48 (96% CI 0.17 to 1.36). At 24 months, data were extractable from seven RCTs totaling 368 teeth, with all seven trials providing data for a pooled RR of 0.50 (95% CI 0.25 to 1.01; Analysis 1.3).

Five of six included trials compared MTA with 1:5 diluted formocresol. One compared MTA with full strength formocresol with no statistically significant difference (Jayam 2014).

Two additional trials, which randomised 32 (Eidelman 2001) and 64 teeth (Holan 2005), did not assess overall failure at a fixed time point but at a mean (range) follow‐up of 13 (6 to 31) and 36 (4 to 74) months, respectively. The RRs were 0.30 (95% CI 0.01 to 6.77) for Eidelman 2001 and 0.19 (95% CI 0.02 to 1.52) for Holan 2005.

Pain

At six months, data were extractable from six RCTs totaling 390 teeth. In five trials, there was no pain in any of the participants regardless of the intervention. From the remaining trial, the results showed no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.91). Results were similar at 12 months, with two trials providing data for a pooled RR of 0.25 (95% CI 0.03 to 2.18). At 24 months, data were extractable from four RCTs totaling 290 teeth. In one of the four trials, there was no pain in any of the participants regardless of the intervention. For the three remaining trials, the pooled results showed no statistically significant difference in pain with MTA compared with formocresol (RR 0.71, 95% CI 0.14 to 3.56; Analysis 1.4).

The results showed no statistically significant difference at any point in pain for MTA compared with full strength formocresol or 1:5 diluted formocresol (Analysis 2.3; Analysis 3.3).

One trial, which randomised 32 teeth, did not assess pain at a fixed time point but at a mean (range) follow‐up of 13 (6 to 31) (Eidelman 2001). There was no pain in any of the participants regardless of the intervention.

Soft tissue pathology

At six months, data were extractable from seven RCTs totaling 410 teeth. In six trials, there was no soft tissue pathology in any participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.91). At 12 months, in three trials, there was no soft tissue pathology in any participants regardless of the intervention. Of the remaining four trials, results showed no statistically significant difference (RR 0.22, 95% CI 0.05 to 1.01). At 24 months, data were extractable from five RCTs totaling 310 teeth, with two trials providing data for a pooled RR of 0.33 (95% CI 0.04 to 3.10; Analysis 1.5).

The results showed no statistically significant difference at any time point in soft tissue pathology for MTA compared either with full strength formocresol or 1:5 diluted formocresol (Analysis 2.4; Analysis 3.4).

In addition, two trials, which randomised 32 (Eidelman 2001) and 64 (Holan 2005) teeth, did not assess soft tissue pathology at a fixed time point but at a mean (range) follow‐up of 13 (6 to 31) for Eidelman 2001 and 38 (4 to 74) months for Holan 2005. There was no soft tissue pathology in any of the participants regardless of the intervention in Eidelman 2001, and for Holan 2005, the RR was 0.94 (95% CI 0.06 to 14.4).

Pathological mobility

At six months, data were extractable from five RCTs totaling 250 teeth. In both trials, there was no pathological mobility in any of the participants regardless of the intervention. At 12 months, data were extractable from four RCTs totaling 200 teeth. For three trials, there was no pathological mobility in any of the participants regardless of the intervention. Results from the remaining trial showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.97). At 24 months, data were extractable from three RCTs totaling 150 teeth. In both trials, there was no pathological mobility in any of the participants regardless of the intervention (Analysis 1.6).

The results showed no statistically significant difference at any time point in pathological mobility between MTA and either full strength formocresol or 1:5 diluted formocresol (Analysis 2.5; Analysis 3.5).

Pathological radiolucency

At six months, data were extractable from 13 RCTs totaling 1010 teeth. In nine of the 13 trials, there was no pathological radiolucency in any of the participants regardless of the intervention. From the four remaining trials, the pooled results showed no evidence of a statistically significant difference in pathological radiolucency with MTA compared with formocresol (pooled RR 0.54, 95% CI 0.27 to 1.08). At 12 months, data were extractable from 11 RCTs totaling 652 teeth. In five of the 11 trials, there was no pathological radiolucency in any participants regardless of the intervention. From the six remaining trials, the pooled results showed evidence of a statistically significant difference in pathological radiolucency between MTA and formocresol (pooled RR 0.43, 95% CI 0.19 to 0.98) with no evidence of statistical heterogeneity among included trials (I² = 0%). At 24 months, data were extractable from eight RCTs totaling 460 teeth. In two of the eight trials, there was no pathological radiolucency in any participants regardless of the intervention. From the six remaining trials, the pooled results showed no evidence of a statistically significant difference in pathological radiolucency with MTA compared with formocresol (pooled RR 0.55, 95% CI 0.25 to 1.22; Analysis 1.7).

The results showed no statistically significant difference at any time point in pathological radiolucency between MTA and either full strength formocresol or 1:5 diluted formocresol (Analysis 2.5; Analysis 3.6).

In addition, two trials, which randomised 32 (Eidelman 2001) and 64 (Holan 2005) teeth, did not assess pathological radiolucency at a fixed time point but at a mean (range) follow‐up of 13 (6 to 31) for Eidelman 2001 and 36 (4 to 74) months for Holan 2005. For one trial, there was no pathological radiolucency in any of the participants regardless of the intervention, and for the other trial, the RR was 0.94 (95% CI 0.06 to 14.4).

Pathological root resorption

At six months, data were extractable from 11 RCTs totaling 866 teeth. In seven of the trials, there was no pathological root resorption in any participants regardless of the intervention. From the four remaining trials, the pooled results showed no evidence of a statistically significant difference in pathological root resorption with MTA compared with formocresol (pooled RR 0.47, 95% CI 0.18 to 1.21). At 12 months, data were extractable from nine RCTs totaling 508 teeth. In five of the trials, there was no pathological root resorption in any participants regardless of the intervention. From the four remaining trials, the pooled results showed no evidence of a statistically significant difference in pathological root resorption between MTA and formocresol (pooled RR 0.26, 95% CI 0.07 to 1.03). At 24 months, data were extractable from six RCTs totaling 338 teeth. In one of the five trials, there was no pathological root resorption in any of the participants regardless of the intervention. From the five remaining trials, the pooled results showed evidence of a statistically significant difference in pathological root resorption between MTA and formocresol (pooled RR 0.25, 95% CI 0.08 to 0.81), with no evidence of statistical heterogeneity among included trials (I² = 0%; Analysis 1.8).

The results showed a statistically significant difference at 24 months in pathological root resorption between MTA and full strength formocresol; the results showed no statistically significant difference at 6 and 12 months between MTA and full strength formocresol, or at any point between MTA and 1:5 diluted formocresol (Analysis 2.6; Analysis 3.7).

A further two trials, which randomised 32 (Eidelman 2001) and 64 (Holan 2005) teeth, did not assess pathological root resorption at a fixed time point but at a mean (range) follow‐up of 13 (6 to 31) months for Eidelman 2001 and 36 (4 to 74) months for Holan 2005. The RR was 0.06 (95% CI 0.00 to 0.92) for Eidelman 2001 and 0.30 (95% CI 0.01 to 6.77) for Holan 2005.

Pulp canal obliteration

At six months, data were extractable from nine RCTs totaling 712 teeth. In six of the trials, there was no pulp canal obliteration in any participants regardless of the intervention. From the three remaining trials, the pooled results showed no statistically significant difference in pulp canal obliteration between MTA and formocresol (pooled RR 1.52, 95% CI 1.00 to 2.30). Results were similar at 12 months, with five of seven trials providing data (RR 1.70, 95% CI 0.81 to 3.57). At 24 months, data were extractable from six RCTs totaling 338 teeth; the pooled results showed a larger risk of pulp canal obliteration with MTA compared with formocresol (RR 2.05, 95% CI 1.07 to 3.94), with no evidence of statistical heterogeneity among included trials (I² = 22%; Analysis 1.9).

The results showed no statistically significant difference at any time point in pulp canal obliteration between MTA and either full strength or 1:5 diluted formocresol (Analysis 2.7; Analysis 3.8).

One additional trial, which randomised 64 teeth, did not assess pulp canal obliteration at a fixed time point but at a mean (range) follow‐up of 36 (4 to 74) months (Holan 2005). The RR was 1.19 (95% CI 0.75 to 1.90).

Dentin bridge formation

At six months, data were extractable from three RCTs totaling 322 teeth. The pooled results showed a greater chance of dentin bridge formation with MTA than with formocresol (RR 18.16, 95% CI 3.63 to 90.91), with evidence of a moderate statistical heterogeneity among included trials (I² = 34%). At 12 months, data were extractable from two RCTs totaling 70 teeth; the pooled results showed no statistically significant difference in dentin bridge formation with MTA compared with formocresol. The pooled RR was 6.00 (95% CI 0.76 to 47.22). Results were similar at 24 months (Analysis 1.10).

All included trials compared MTA with 1:5 diluted formocresol.

In addition, the trial by Holan 2005, which randomised 64 teeth, did not assess dentin bridge formation at a fixed time point but at a mean (range) follow‐up of 36 (4 to 74) months. The RR was 2.82 (95% CI 0.12 to 66.82).

Physiological root resorption

At six months, data were extractable from two RCTs totaling 170 teeth. In both trials, there was no physiological root resorption in any of the participants regardless of the intervention. Results were similar at 12 months. At 24 months, one of the two trials showed no cases of physiological root resorption, regardless of the intervention. In the other trial, the results showed no statistically significant difference in physiological root resorption with MTA compared with formocresol (RR 0.33, 95% CI 0.01 to 7.83; Analysis 1.11).

The trial providing data assessed 1:5 diluted formocresol.

Clinical failure

At six months, data were extractable from four RCTs totaling 150 teeth. In three trials, there was no clinical failure in any participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.85). At 12 months, with four trials providing data, the pooled results showed a statistically significant difference in clinical failure between MTA and calcium hydroxide (RR 0.16 (95% CI 0.04 to 0.70), with no evidence of statistical heterogeneity among the included trials (I² = 0%). At 24 months, data were extractable from five RCTs totaling 284 teeth. All trials provided data. The pooled results showed a statistically significant difference in favour of MTA (RR 0.25, 95% CI 0.12 to 0.52), with evidence of a moderate statistical heterogeneity among included trials (I² = 31%; Analysis 5.1).

Radiological failure

At six months, data were extractable from four RCTs totaling 150 teeth. In one trial, there was no radiological failure in any participants regardless of the intervention. In the other trials, the results showed a statistically significant difference in radiological failure in favour of MTA compared to calcium hydroxide (RR 0.08, 95% CI 0.02 to 0.41), with no evidence of statistical heterogeneity among included trials (I² = 0%). Results were similar at 12 months (RR 0.12, 95% CI 0.04 to 0.36). At 24 months, data were extractable from five RCTs totaling 284 teeth. All trials provided data. The pooled results showed a statistically significant difference in favour of MTA (RR 0.14, 95% CI 0.08 to 0.26), with evidence of substantial statistical heterogeneity among included trials (I² = 68%; Analysis 5.2).

Overall failure

At six months, data were extractable from two RCTs totaling 68 teeth. In one trial, there was no overall failure in any of the participants regardless of the intervention. In the other trial, the results showed no statistically significant differences (RR 0.20, 95% CI 0.01 to 3.92). Results were similar at 12 months with all trials providing data (RR 0.34, 95% CI 0.10 to 1.19). At 24 months, the pooled results showed a statistically significant difference in favour of MTA (RR 0.42, 95% CI 0.18 to 0.95), with evidence of moderate statistical heterogeneity among included trials (I² = 36%; Analysis 5.3).

Pain

At six and 12 months, one trial, which randomised 62 teeth, assessed spontaneous pain (Akcay 2014). There was no pain in any of the participants regardless of the intervention. At 24 months, data were extractable from two RCTs totaling 196 teeth. The results showed no statistically significant difference between MTA and calcium hydroxide (RR 0.41, 95% CI 0.09 to 1.73; Analysis 5.4).

Defective restoration (clinically)

One trial, which randomised 139 teeth, assessed amalgam restorations with ditched margins (Celik 2013). At 24 months, there was no statistically significant difference between MTA and calcium hydroxide (RR 0.54, 95% CI 0.08 to 3.71).

Soft tissue pathology

At six months, data were extractable from three RCTs totaling 122 teeth. In one trial, there was no soft tissue pathology in any of the participants regardless of the intervention. In the other trials, results showed no statistically significant difference (RR 0.20, 95% CI 0.02 to 1.62). At 12 months, with all trials providing data, the results were statistically significant and in favour of MTA (RR 0.12, 95% CI 0.02 to 0.62), with no evidence of statistical heterogeneity among included trials (I² = 0%). At 24 months, data were extractable from four RCTs totaling 256 teeth. All trials provided data. The results were statistically significant (RR 0.17, 95% CI 0.06 to 0.47) with no evidence of statistical heterogeneity among included trials (I² = 0%; Analysis 5.5).

Pathological mobility

At six and 12 months, data were extractable from three RCTs totaling 122 teeth. In one trial, there was no pathological mobility in any of the participants regardless of the intervention. In the other trials, results showed no statistically significant difference (RR 0.20, 95% CI 0.02 to 1.62). Results were statistically significant at 12 and 24 months, clearly favouring MTA (RR 0.09, 95% CI 0.01 to 0.66), with no evidence of statistical heterogeneity among included trials (I² = 0%). At 24 months, data were extractable from four RCTs totaling 256 teeth with all trials providing data (Analysis 5.6).

Pathological radiolucency

At six months, data were extractable from four RCTs totaling 162 teeth. For one trial, there was no pathological radiolucency in any of the participants regardless of the intervention. For the remaining trials, results were statistically significant and in favour of MTA (RR 0.10, 95% CI 0.02 to 0.50), with no evidence of statistical heterogeneity among included trials (I² = 0%). Results were similar at 12 (RR 0.14, 95% CI 0.04 to 0.47) and 24 months (RR 0.08, 95% CI 0.03 to 0.22). At 24 months, data were extractable from five RCTs totaling 296 teeth with all trials providing data (Analysis 5.7).

Pathological root resorption

At six months, data were extractable from five RCTs totaling 190 teeth. In one of the trials, there was no pathological root resorption in any of the participants regardless of the intervention. From the other trials, the pooled results showed statistically significant difference in pathological root resorption between MTA and calcium hydroxide (RR 0.10, 95% CI 0.02 to 0.39) with no evidence of statistical heterogeneity among included trials (I² = 0%). Results were similar at 12 (RR 0.07, 95% CI 0.02 to 0.29) and 24 months (RR 0.08, 95% CI 0.03 to 0.18). At 24 months, data were extractable from five RCTs totaling 324 teeth with all trials providing data (Analysis 5.8).

Pulp canal obliteration

At six months, data were extractable from three RCTs totaling 120 teeth. In one of the trials, there was no pulp canal obliteration in any of the participants regardless of the intervention. From the other trials, the results showed a statistically significant difference between MTA and calcium hydroxide (RR 7.77, 95% CI 1.56 to 38.69), showing a higher risk of pulp canal obliteration with MTA than with calcium hydroxide, with no evidence of statistical heterogeneity among included trials (I² = 0%). At 12 months, results were not statistically significantly different with all trials providing data (RR 2.01, 95% CI 0.97 to 4.17). At 24 months, data were extractable from four RCTs totaling 254 teeth, with all trials providing data. The results showed a statistically significantly higher risk of pulp canal obliteration with MTA than with calcium hydroxide (RR 2.05, 95% CI 1.01 to 4.19), with evidence of substantial statistical heterogeneity among included trials (I² = 69%; Analysis 5.9).

Dentin bridge formation

Data were extractable from two RCTs totaling 60 teeth. At six months, the results showed a statistically significant difference and a greater chance of dentin bridge formation when MTA is applied than when calcium hydroxide is applied (RR 0.20, 95% CI 0.05 to 0.84), with no evidence of statistical heterogeneity among included trials (I² = 0%). Results were not statistically significant at 12 and 24 months (RR 0.80, 95% CI 0.37 to 1.74; Analysis 5.10).

Clinical failure

At six months, data were extractable from four RCTs totaling 190 teeth. In three trials, there was no clinical failure in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference in clinical failure between MTA and ferric sulphate (RR 0.08, 95% CI 0.00 to 1.31). At 12 months, data were extractable from three RCTs totaling 130 teeth. In two trials, there was no clinical failure in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference in clinical failure for MTA compared with ferric sulphate (RR 0.20, 95% CI 0.01 to 3.97). Results were similar at 24 months, with all trials providing data (RR 0.52, 95% CI 0.20 to 1.39; Analysis 4.1).

Radiological failure

At six months, data were extractable from four RCTs totaling 190 teeth. In two trials, there was no clinical failure in any of the participants regardless of the intervention. For the two remaining trials, the pooled results showed a statistically significant difference (RR 0.06, 95% CI 0.01 to 0.40) with evidence of moderate statistical heterogeneity among included trials (I² = 30%). At 12 months, data were extractable from three RCTs totaling 130 teeth. In one trial, there was no clinical failure in any of the participants regardless of the intervention. For the remaining two trials, the pooled results showed no statistically significant difference (RR 0.71, 95% CI 0.15 to 3.44). Results were similar at 24 months, with all trials providing data (RR 0.58, 95% CI 0.25 to 1.36; Analysis 4.2).

Overall failure

At six months, data were extractable from four RCTs totaling 190 teeth. In two trials, there was no overall failure in any participants regardless of the intervention. For the two remaining trials, the pooled results showed a statistically significant difference favouring MTA over ferric sulphate (RR 0.06, 95% CI 0.01 to 0.40) with evidence of a moderate statistical heterogeneity among included trials (I² = 30%). At 12 months, data were extractable from three RCTs totaling 130 teeth. In one trial, there was no overall failure in any of the participants regardless of the intervention. For the remaining two trials, the pooled results showed no statistically significant difference (RR 0.71, 95% CI 0.15 to 3.44). Results were similar at 24 months, with all trials providing data (RR 0.78, 95% CI 0.32 to 1.89; Analysis 4.3).

Pain

At six months, data were extractable from three RCTs totaling 160 teeth. In two trials, there was no pain in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 4.00). At 12 months, data were extractable from two RCTs totaling 100 teeth. In the two trials, there was no pain in any of the participants regardless of the intervention (Analysis 4.4).

At 24 months, one trial, which randomised 50 teeth, assessed pain (Fernández 2013). There was no pain in any of the participants regardless of the intervention.

One additional trial, which randomised 111 teeth, did not assess pain at a fixed time point but at a mean (range) follow‐up of 22 (6 to 38) months (Doyle 2010). The RR was 0.36 (95% CI 0.02 to 8.75).

Soft tissue pathology

At six months, data were extractable from two RCTs totaling 110 teeth. In one trial, there was no soft tissue pathology in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 4.00; Analysis 4.5).

At 24 months, one trial, which randomised 50 teeth, assessed soft tissue pathology (Fernández 2013). There was no soft tissue pathology in any of the participants regardless of the intervention.

One additional trial assessed soft tissue pathology at a mean (range) follow‐up of 22 (6 to 38) months (Doyle 2010). There was no statistically significant difference between groups (RR 0.36, 95% CI 0.02 to 8.75).

Adjacent tissues inflammation

One trial, which randomised 50 teeth, assessed adjacent tissues inflammation (Fernández 2013). At six months, there was no adjacent tissue inflammation in any of the participants regardless of the intervention. At 12 and 24 months, the results showed no statistically significance between groups (RR 0.33, 95% CI 0.01 to 7.81; RR 0.20, 95% CI 0.01 to 3.97, respectively).

One additional trial, which randomised 111 teeth, assessed adjacent tissues inflammation at a mean (range) follow‐up of 22 (6 to 38) months (Doyle 2010). There was no statistically significant difference between groups (RR 0.36, 95% CI 0.02 to 8.75).

Pathological mobility

At six months, data were extractable from two RCTs totaling 110 teeth. In one trial, there was no pathological mobility in any of the participants regardless of the intervention. In the other trial, the results showed no statistically significant difference (RR 0.08, 95% CI 0.00 to 1.31; Analysis 4.6).

At 12 and 24 months, one trial, which randomised 50 teeth, assessed pathological mobility (Fernández 2013). There was no pathological mobility in any of the participants regardless of the intervention.

One additional trial, which randomised 111 teeth, assessed pathological mobility at a mean (range) follow‐up of 22 (6 to 38) months (Doyle 2010). There was no statistically significant difference between groups (RR 0.22, 95% CI 0.01 to 4.45).

Pathological radiolucency

At six months, data were extractable from three RCTs totaling 160 teeth. In two trials, there was no pathological radiolucency in any of the participants regardless of the intervention. In the remaining trial, the results showed a statistically significant difference favouring MTA over ferric sulphate (RR 0.03, 95% CI 0.00 to 0.48). At 12 and 24 months, data were extractable from two RCTs totaling 100 teeth. In one trial, there was no pathological radiolucency in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference at 12 and 24 months (RR 0.33, 95% CI 0.01 to 7.81; RR 0.20, 95% CI 0.01 to 3.97; respectively; Analysis 4.7).

One additional trial, which randomised 111 teeth, did not assess pathological radiolucency at a fixed time point but at a mean (range) follow‐up of 22 (6 to 38) months (Doyle 2010). The RR was 0.36 (95% CI 0.02 to 8.75).

Pathological root resorption

At six months, data were extractable from four RCTs totaling 190 teeth. In two trials, there was no pathological root resorption in any of the participants regardless of the intervention. For the two remaining trials, the pooled results showed a statistically significant difference favouring MTA over ferric sulphate (RR 0.07, 95% CI 0.01 to 0.53), with no evidence of statistical heterogeneity among included trials (I² = 8%). At 12 months, data were extractable from three RCTs totaling 130 teeth. In two trials, there was no pathological root resorption in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.97). Results were similar at 24 months, with all trials providing data (RR 0.56, 95% CI 0.12 to 2.51; Analysis 4.8).

One trial, which randomised 111 teeth, did not assess pathological root resorption at a fixed time point but at a mean (range) follow‐up of 22 (6 to 38) months (Doyle 2010). The RR was 0.22 (95% CI 0.07 to 0.71).

Pulp canal obliteration

At six months, data were extractable from three RCTs totaling 140 teeth. In the three trials, there was no pulp canal obliteration in any of the participants regardless of the intervention. At 12 months, data were extractable from two RCTs totaling 80 teeth. In one trial, there was no pulp canal obliteration in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 3.00, 95% CI 0.13 to 70.30). Results were similar at 24 months, with all trials providing data (RR 1.57, 95% CI 0.47 to 5.27; Analysis 4.9).

One additional trial, which randomised 111 teeth, did not assess pulp canal obliteration at a fixed time point but at a mean (range) follow‐up of 22 (6 to 38) months (Doyle 2010). The RR was 1.68 (95% CI 0.93 to 3.04).

Physiological root resorption

One trial, which randomised 50 teeth, assessed physiological root resorption (Erdem 2011). At six and 12 months, there was no physiological root resorption in any of the participants regardless of the intervention. At 24 months, the RR was 0.14 (95% CI 0.01 to 2.63).

Finally, one registered trial assessed clinical and radiographic success at six, nine and 12 months, but no resulting information was published (NCT02783911).

Clinical failure

At six months, data were extractable from six RCTs totaling 332 teeth. In two of the trials, there was no clinical failure in any of the participants regardless of the intervention. From the four remaining trials, the pooled results showed a larger risk of clinical failure with calcium hydroxide compared with formocresol. The pooled RR was 1.98 (95% CI 1.17 to 3.37). The statistical heterogeneity among included trials was substantial (I² = 57%). Results were similar at 12 months (RR 1.87, 95% CI 1.22 to 2.89). At 24 months, data were extractable from three RCTs totaling 150 teeth. The pooled results showed no statistically significant difference (RR 2.18, 95% CI 0.78 to 6.11; Analysis 8.1).

One additional trial, which randomised 76 teeth, did not assess clinical failure at a fixed time point but at an interval follow‐up of zero to six, seven to 12 and 13 to 24 months (Zurn 2008). Between zero and six months, the RR was 1.00 (95% CI 0.06 to 15.41). Results were similar between seven and 12 months and between 13 and 24 months.

Radiological failure

At six months, data were extractable from four RCTs totaling 154 teeth. In one of the trials, there was no radiological failure in any of the participants regardless of the intervention. From the three remaining trials, the pooled results showed a larger risk of radiological failure with calcium hydroxide compared with formocresol (RR 15.48, 95% CI 3.86 to 62.06) with no evidence of statistical heterogeneity among included trials (I² = 0%). Results at 12 months were RR 1.86 (95% CI 1.42 to 2.44) with six trials providing data (332 teeth randomised) and with substantial statistical heterogeneity (I² = 89%), and at 24 months the RR was 3.63 (95% CI 1.73 to 7.61) with three trials providing data (150 teeth randomised) and with no evidence of statistical heterogeneity (I² = 0%) (Analysis 8.2).

The trial by Zurn 2008 did not assess radiological failure at a fixed time point but at an interval follow‐up of zero to six, seven to 12 and 13 to 24 months. Between zero and six months, the RR was 1.00 (95% CI 0.06 to 15.41). Results were similar between seven and 12 months. Between 13 and 24 months, the RR was 9 (95% CI 1.20 to 67.60).

Overall failure

At 12 months, data were extractable from two RCTs totaling 120 teeth. The pooled results showed no statistically significant difference in overall failure with calcium hydroxide compared with formocresol. The pooled RR was 2.41 (95% CI 0.80 to 7.21). Results were similar at 24 months, although the difference was statistically significant and in favour of formocresol (RR 2.93, 95% CI 1.35 to 6.34), with no evidence of statistical heterogeneity (I² = 0%; Analysis 8.3).

One additional trial did not assess overall failure at a fixed time point but at an interval follow‐up of zero to six, seven to 12 and 13 to 24 months (Zurn 2008). Between zero and six months, the RR was 1.00 (95% CI 0.15 to 6.74). Between seven and 12 months, the RR was 4.50 (95% CI 1.04 to 19.47). Results were similar between 13 and 24 months.

Pain

At six months, data were extractable from four RCTs totaling 276 teeth. In three of the trials, there was no pain in any of the participants regardless of the intervention. The results for the remaining trial showed no statistically significant difference in pain with calcium hydroxide compared with formocresol (RR 3.18, 95% CI 0.35 to 29.08). At 12 months, the results were statistically significant (RR 6.30, 95% CI 1.15 to 34.40), in favour of formocresol with no evidence of statistical heterogeneity (I² = 0%; Analysis 8.4).

Soft tissue pathology

At six months, data were extractable from five RCTs totaling 306 teeth. In three of the four trials, there was no soft tissue pathology in any of the participants regardless of the intervention. From the two remaining trials, the pooled results showed no statistical difference in soft tissue pathology with calcium hydroxide compared with formocresol. The pooled RR was 5.14 (95% CI 0.63 to 42.25). At 12 months, the results were statistically significant (RR 6.77, 95% CI 1.23 to 37.10) in favour of formocresol with three trials providing data and with no evidence of statistical heterogeneity (I² = 0%). However, this was not statistically significant at 24 months, with only two trials providing data (Analysis 8.5).

One additional trial, which randomised 76 teeth, did not assess soft tissue pathology at a fixed time point but at an interval follow‐up of zero to six, seven to 12 and 13 to 24 months (Zurn 2008). Between zero and six months, the RR was 1.00 (95% CI 0.06 to 15.41). Results were similar between seven and 12 months and between 13 and 24 months.

Pathological mobility

At six months, data were extractable from four RCTs totaling 238 teeth. In three of the trials, there was no pathological mobility in any of the participants regardless of the intervention. From the remaining trial, the results showed no statistically significant difference in pathological mobility between calcium hydroxide and formocresol (RR 1.21, 95% CI 0.18 to 8.19). At 12 months, results were similar with two of the four trials providing data (RR 1.14, 95% CI 0.40 to 3.31). At 24 months, results were similar with one of two trials providing data (124 teeth randomised) (RR 9.00, 95% CI 0.53 to 153.79; Analysis 8.6).

Secondary caries at the margin (clinically)

One trial, which randomised 84 teeth, assessed secondary caries at the margin (Waterhouse 2000). At six months, the RR was 0.17 (95% CI 0.01 to 3.23). Results were similar at 12 months.

Defective restoration (clinically)

One trial, which randomised 84 teeth, assessed defective restoration (Waterhouse 2000). At six months, there was no statistically significant difference in the number of defective restorations between treatment groups (RR 0.40, 95% CI 0.02 to 9.59); at 12 months, the RR was 1.21 (95% CI 0.08 to 18.72).

Pathological radiolucency

At six months, data were extractable from three RCTs totaling 98 teeth. For one trial, there was no pathological radiolucency in any of the participants regardless of the intervention. For the remaining trials, the pooled results showed no statistically significant difference in pathological radiolucency for calcium hydroxide compared with formocresol (RR 3.78, 95% CI 0.64 to 22.17). Results were similar at 12 months with four trials providing data (276 teeth randomised) (RR 1.90, 95% CI 0.67 to 5.40), and at 24 months with two trials providing data (124 teeth randomised) (RR 3.24, 95% CI 0.79 to 13.28; Analysis 8.7).

One additional trial, which randomised 76 teeth, did not assess pathological radiolucency at a fixed time point but at an interval follow‐up of zero to six, seven to 12 and 13 to 24 months (Zurn 2008). Between zero and six months, the RR was 1.00 (95% CI 0.15 to 6.74). Between seven and 12 months, the RR was 4.50 (95% CI 1.04 to 19.47). Results were similar between 13 and 24 months.

Pathological root resorption

At six months, data were extractable from four RCTs totaling 154 teeth. In one of the trials, there was no pathological root resorption in any of the participants regardless of the intervention. From the remaining trials, the pooled results showed a larger risk of pathological root resorption with calcium hydroxide compared with formocresol (RR 11.87, 95% CI 2.33 to 60.40), with no evidence of statistical heterogeneity among trials (I² = 0%). Results were similar at 12 months, with five of six trials providing data (332 teeth randomised) (RR 6.25, 95% CI 2.04 to 19.14), and at 24 months, with three trials providing data (150 teeth randomised) (RR 4.59, 95% CI 1.33 to 15.81; Analysis 8.8).

One additional trial, which randomised 76 teeth, did not assess pathological root resorption at a fixed time point but at an interval follow‐up of zero to six, seven to 12 and 13 to 24 months (Zurn 2008). Between zero and six months, the RR was 1.00 (95% CI 0.31 to 3.17). Results were similar between seven and 12 months and between 13 and 24 months.

Pulp canal obliteration

At six months, data were extractable from two trials totaling 56 teeth. There was no statistically significant difference (RR 4.00, 95% CI 0.47 to 33.75). At 12 months, data were extractable from three RCTs totaling 140 teeth. The pooled results showed no statistically significant difference in pulp canal obliteration with calcium hydroxide compared with formocresol (RR 2.68, 95% CI 0.91 to 7.95; Analysis 8.9).

One additional trial, which randomised 76 teeth, did not assess pulp canal obliteration at a fixed time point but at an interval follow‐up of zero to six, seven to 12 and 13 to 24 months. Between zero and six months, the RR was 1.00 (95% CI 0.42 to 2.39) (Zurn 2008). Results were similar between seven and 12 months. Between 13 and 24 months, the RR was 1.88 (95% CI 1.29 to 2.72).

Dentin bridge formation

Data were extractable from two RCTs totaling 60 teeth. There was a statistically significant difference in favour of calcium hydroxide at six (RR 13.00, 95% CI 1.81 to 93.60) and 12 months (RR 14.00, 95% CI 1.95 to 100.26; Analysis 8.10).

Physiological root resorption

One trial, which randomised 84 teeth, assessed physiological root resorption at 12 months (Waterhouse 2000). There was no statistically significant difference between groups (RR 0.40, 95% CI 0.12 to 1.39).

Secondary caries (radiographically)

One trial, which randomised 84 teeth, assessed secondary caries at 12 months (Waterhouse 2000). There was no statistically significant difference between groups (RR 3.62, 95% CI 0.15 to 86.28).

Clinical failure

At six, 12 and 24 months, data were extractable from two RCTs totaling 122 teeth. At six months, one trial showed no cases of clinical failure in either treatment group. For the second trial, the results showed no statistically difference in clinical failure with calcium hydroxide compared with ferric sulphate (RR 3.40, 95% CI 0.14 to 81.38). At 12 months, the pooled results showed no statistically significant difference in clinical failure with calcium hydroxide compared with ferric sulphate (RR 3.41, 95% CI 0.37 to 31.61). Results were similar at 24 months (RR 3.44, 95% CI 0.90 to 13.18; Analysis 9.1).

Radiological failure

At 12 and 24 months, data were extractable from two RCTs totaling 122 teeth. At 12 months, the pooled results showed no statistically significant difference in radiological failure with calcium hydroxide compared with ferric sulphate. The pooled RR was 1.28 (95% CI 0.53 to 3.13). The direction of effect was the same at 24 months, although the difference was statistically significant (RR 1.97, 95% CI 1.04 to 3.75) in favour of ferric sulphate, and there was no evidence of statistical heterogeneity among included trials (I² = 0%; Analysis 9.2).

Overall failure

At 12 and 24 months, data were extractable from two RCTs totaling 122 teeth. At 12 months, the pooled results showed no statistically significant difference in overall failure with calcium hydroxide compared with ferric sulphate (RR 1.28, 95% CI 0.53 to 3.13). The statistical heterogeneity among included trials was substantial (I² = 54%). At 24 months, the difference was statistically significant (RR 1.97, 95% CI 1.04 to 3.75) in favour of ferric sulphate, with no evidence of statistical heterogeneity among included trials (I² = 0%; Analysis 9.3).

Pathological root resorption

At 12 and 24 months, data were extractable from two RCTs totaling 122 teeth. At 12 months, one of the two trials found no pathological root resorption in any of the participants regardless of the intervention. In the second trial, the results showed no statistically significant difference in pathological root resorption with calcium hydroxide compared with ferric sulphate (RR 0.57, 95% CI 0.05 to 6.05). Results were similar at 24 months, with the two trials providing data (RR 2.29, 95% CI 0.60 to 8.66; Analysis 9.4).

Pulp canal obliteration

One trial, which randomised 28 teeth, assessed pulp canal obliteration (Sonmez 2008). At six and 12 months, there was no pulp canal obliteration in any of the participants regardless of the intervention. At 24 months, the RR was 0.38 (95% CI 0.05 to 3.26).

Clinical failure

At six months, data were extractable from six RCTs totaling 394 teeth. In five RCTs, there was no clinical failure in any of the participants regardless of the intervention. For the remaining trials, there was no statistically significant difference (RR 1.00, 95% CI 0.15 to 6.87). Results were similar at 12 months, with five trials providing data (RR 1.38, 95% CI 0.45 to 4.27). At 24 months, data were extractable from five RCTs totaling 258 teeth. The pooled results for the three trials showed no statistically significant difference in clinical failure (RR 0.83, 95% CI 0.40 to 1.70; Analysis 10.1).

One trial, which randomised 164 teeth, did not assess clinical failure at a fixed time point but at an interval follow‐up of three to 20 and 46 to 48 months (Ibricevic 2000). The RR was 1.00 (95% CI 0.02 to 49.04) at three to 20 months and RR 1.58 (95% CI 0.27 to 9.18) at 46 to 48 months.

Radiological failure

At six months, data were extractable from six RCTs totaling 294 teeth. In two trials, there was no radiological failure in any of the participants regardless of the intervention. From the remaining trials, the pooled results showed no statistically significant difference in clinical failure with ferric sulphate compared with formocresol (RR 0.79, 95% CI 0.32 to 1.92). Results were similar at 12 months, with six trials providing data (394 teeth randomised) (RR 1.33, 95% CI 0.73, 2.42), and at 24 months, with five trials providing data (258 teeth randomised) (RR 1.26, 95% CI 0.71 to 2.24; Analysis 10.2).

One trial, which randomised 164 teeth, did not assess radiological failure at a fixed time point but at an interval follow‐up of three to 20 and 46 to 48 months (Ibricevic 2000). The RR was 1.00 (95% CI 0.42 to 2.36) at three to 20 months. Results were similar at 46 to 48 months.

Overall failure

At six months, data were extractable from four RCTs totaling 184 teeth. In two trials, there was no overall failure in any of the participants regardless of the intervention. From the two remaining trials, the pooled results showed no statistically significant difference in clinical failure with ferric sulphate compared with formocresol (RR 0.53, 95% CI 0.12 to 2.37). Results were similar at 12 months, with four of five trials providing data (284 teeth randomised) (RR 1.16, 95% CI 0.51, 2.64), and at 24 months, with four trials providing data (228 teeth randomised) (RR 1.49, 95% CI 0.74 to 3.01; Analysis 10.3).

Two additional trials, which randomised 96 teeth (Fuks 1997) and 164 teeth (Ibricevic 2000), did not assess overall failure at a fixed time point but at an interval follow‐up of 24 to 35 months and 46 to 48 months, respectively. The RR was 0.44 (95% CI 0.13 to 1.45) for 24 to 35 months (Fuks 1997) and 1.33 (95% CI 0.44 to 4.03) for 46 to 48 months (Ibricevic 2000).

Pain

Data were extractable from four RCTs totaling 230 teeth. In three trials, no participant declared pain regardless of the intervention. For the remaining trial, there was no statistically significant difference at six, 12 (RR 0.33, 95% CI 0.01 to 7.58) and 24 months (RR 0.20, 95% CI 0.01 to 3.85; Analysis 10.4).

Soft tissue pathology

One trial, which randomised 30 teeth, assessed soft tissue pathology at six, 12 and 24 months (Ozmen 2017). There was no statistically significant evidence (RR 0.33, 95% CI 0.01 to 7.58).

Pathologic mobility

One trial, which randomised 30 teeth, assessed soft tissue pathology at six, 12 and 24 months (Ozmen 2017). There was no statistically significant evidence (RR 0.33, 95% CI 0.01 to 7.58).

Adjacent tissue inflammation

One trial, which randomised 50 teeth, assessed adjacent tissue inflammation (Fernández 2013). There was no adjacent tissue inflammation in any of the participants regardless of the intervention. At 12 and 24 months, there was no statistically significant difference (RR 5.00, 95% CI 0.25, 99.16).

Pathological radiolucency

At six months, two trials, which randomised 80 teeth, assessed pathological radiolucency (Fernández 2013, Ozmen 2017). There was no pathological radiolucency in any of the participants regardless of the intervention. At 12 months, data were extractable from four RCTs totaling 230 teeth. In two trials, there was no pathological radiolucency in any of the participants regardless of the intervention. From the remaining trials, the results showed no statistically significant difference in pathological radiolucency with ferric sulphate compared with formocresol (RR 1.80, 95% CI 0.40 to 8.17). Results were similar at 24 months (RR 2.20, 95% CI 0.51 to 9.50; Analysis 10.5).

Two additional trials, which randomised 96 (Fuks 1997) and 164 (Ibricevic 2000) teeth, did not assess pathological radiolucency at a fixed time point but at an interval follow‐up of six to 11, 12 to 23 and 24 to 35 months for Fuks 1997 and three to 20 and 46 to 48 months for Ibricevic 2000. In the two trials, there was no pathological radiolucency in any of the participants regardless of the intervention at the first delays. At 24 to 35 months, the RR was 0.44 (95% CI 0.08 to 2.49) (Fuks 1997) and at 46 to 48 months RR 2.86 (95% CI 0.30 to 26.90) (Ibricevic 2000).

Pathological root resorption

At six months, data were extractable from five RCTs totaling 314 teeth. In three trials, there was no pathological root resorption in any of the participants regardless of the intervention. For the remaining trials, the pooled results showed no statistically significant evidence (RR 0.67, 95% CI 0.12 to 3.84). At 12 months, data were extractable from SIX RCTs totaling 314 teeth. In two trials, there was no pathological root resorption in any of the participants regardless of the intervention. From the remaining trials, the pooled results showed no statistically significant difference in pathological root resorption with ferric sulphate compared with formocresol (RR 1.64, 95% CI 0.53 to 5.08). Results were similar at 24 months with four trials providing data (258 teeth randomised) (RR 1.21, 95% CI 0.50 to 2.96; Analysis 10.6).

Two additional trials did not assess pathological root resorption at a fixed time point but at an interval follow‐up of six to 11, 12 to 23 and 24 to 35 months for Fuks 1997 and three to 20 and 46 to 48 months for Ibricevic 2000. For Fuks 1997, there was no pathological root resorption in any of the participants regardless of the intervention at the first delays. At 24 to 35 months, the RR was 1.31 (95% CI 0.25 to 6.81). In Ibricevic 2000, the RR was 0.95 (95% CI 0.06 to 14.97) at three to 20 months and RR 0.63 (95% CI 0.11 to 3.70) at 46 to 48 months.

Pulp canal obliteration

At six months, data were extractable from three RCTs totaling 134 teeth. In all three trials, there was no pulp canal obliteration in any of the participants regardless of the intervention. At 12 months, data were extractable from three RCTs totaling 134 teeth. In one of the trials, there was no pulp canal obliteration in any of the participants regardless of the intervention. From the two remaining trials, the pooled results showed no statistically significant difference in pulp canal obliteration with ferric sulphate compared with formocresol (RR 0.94, 95% CI 0.54 to 1.64). At 24 months, data were extractable from two RCTs totaling 78 teeth. The pooled results showed no statistically significant difference in pulp canal obliteration with ferric sulphate compared with formocresol (RR 1.24, 95% CI 0.28 to 5.54; Analysis 10.7).

One additional trial, which randomised 96 teeth, did not assess pulp canal obliteration at a fixed time point but at an interval follow‐up of six to 11, 12 to 23 and 24 to 35 months (Fuks 1997). There was no pulp canal obliteration in any of the participants regardless of the intervention at six to 11 or 12 to 23 months. At 24 to 35 months, the RR was 1.64 (95% CI 0.55 to 4.85).

Physiological root resorption

One trial, which randomised 50 teeth, assessed physiological root resorption (Erdem 2011). At six and 12 months, there was no physiological root resorption in any of the participants regardless of the intervention. At 24 months, the RR was 3.00 (95% CI 0.33 to 26.92).

Succedaneous tooth structural anomaly

One trial, which randomised 164 teeth, assessed succedaneous tooth structural anomaly (Ibricevic 2000). At three to 20 and 46 to 48 months, there was no succedaneous tooth structural anomaly in any of the participants regardless of the intervention.

Clinical failure

At six months, data were extractable from two RCTs totaling 110 teeth. In the two trials, there was no clinical failure in any of the participants regardless of the intervention. At 12 months, data were extractable from two RCTs totaling 110 teeth. In one trial, there was no clinical failure in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 4.39, 95% CI 0.22 to 87.82; Analysis 11.1).

At 24 months, one trial, which randomised 50 teeth, assessed clinical failure (Fernández 2013). The results showed no statistically significant difference (RR 3.00, 95% CI 0.13 to 70.30).

Radiological failure

At six months, data were extractable from two RCTs totaling 110 teeth. The results showed no statistically significant difference (RR 0.55, 95% CI 0.22 to 1.39). Results were similar at 12 months (RR 0.42, 95% CI 0.17 to 1.02; Analysis 11.2).

At 24 months, one trial, which randomised 50 teeth, assessed radiological failure (Fernández 2013). The results showed no statistically significant difference (RR 1.50, 95% CI 0.27 to 8.22).

Overall failure

One trial, which randomised 50 teeth, assessed overall failure (Fernández 2013). The results showed no statistically significant difference at six months (RR 3.00, 95% CI 0.33 to 26.92), 12 and 24 months (RR 1.50, 95% CI 0.27 to 8.22).

Pain

At six and 12 months, data were extractable from two RCTs totaling 110 teeth. In the two trials, there was no pain in any of the participants regardless of the intervention (Analysis 11.3).

At 24 months, one trial, which randomised 50 teeth, assessed pain (Fernández 2013). There was no pain in any of the participants regardless of the intervention.

Soft tissue pathology

At six and 12 months, data were extractable from two RCTs totaling 110 teeth. In the two trials, there was no soft tissue pathology in any of the participants regardless of the intervention (Analysis 11.4).

At 24 months, one trial, which randomised 50 teeth, assessed soft tissue pathology (Fernández 2013). There was no soft tissue pathology in any of the participants regardless of the intervention.

Adjacent tissue inflammation

At six months, data were extractable from two RCTs totaling 110 teeth. In the two trials, there was no adjacent tissue inflammation in any of the participants regardless of the intervention. At 12 months, data were extractable from two RCTs totaling 110 teeth. The pooled results showed no statistically significant difference (RR 0.31, 95% CI 0.03 to 2.91; Analysis 11.5).

At 24 months, one trial, which randomised 50 teeth, assessed clinical failure (Fernández 2013). The results showed no statistically significant difference (RR 0.50, 95% CI 0.05 to 5.17).

Pathological mobility

At six and 12 months, data were extractable from two RCTs totaling 110 teeth. In the two trials, there was no pathological mobility in any of the participants regardless of the intervention (Analysis 11.6).

At 24 months, one trial, which randomised 50 teeth, assessed pathological mobility (Fernández 2013). There was no pathological mobility in any of the participants regardless of the intervention.

Pathological radiolucency

At six months, data were extractable from two RCTs totaling 110 teeth. In one trial, there was no pathological radiolucency in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.88, 95% CI 0.06 to 13.35). Results were similar at 12 months, with the two trials providing data (RR 0.56, 95% CI 0.07 to 4.17; Analysis 11.7).

At 24 months, one trial, which randomised 50 teeth, assessed clinical failure (Fernández 2013). The results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.97).

Pathological root resorption

At six months, data were extractable from two RCTs totaling 110 teeth. The pooled results showed no statistically significant difference (RR 0.51, 95% CI 0.18 to 1.42). Results were similar at 12 months (RR 0.38, 95% CI 0.15 to 1.01; Analysis 11.8).

At 24 months, one trial, which randomised 50 teeth, assessed clinical failure (Fernández 2013). The results showed no statistically significant difference (RR 1.50, 95% CI 0.27 to 8.22).

Clinical failure

At six months, data were extractable from three RCTs totaling 130 teeth. The pooled results showed no statistically significant difference (RR 0.23, 95% CI 0.04 to 1.30). Results were similar at 12 months, with two trials providing data (100 teeth randomised) (RR 0.20, 95% CI 0.02 to 1.62; Analysis 12.1).

Radiological failure

At six months, data were extractable from three RCTs totaling 130 teeth. The pooled results showed no statistically significant difference (RR 0.89, 95% CI 0.38 to 2.12). Results were similar at 12 months, with two trials providing data (100 teeth randomised) (RR 0.91, 95% CI 0.44 to 1.92; Analysis 12.2).

Pain

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 0.20, 95% CI 0.03 to 1.60; Analysis 12.3).

Pathological radiolucency

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 0.25, 95% CI 0.03 to 2.08; Analysis 12.4).

Pathological root resorption

At six months, data were extractable from twoRCTs totaling 50 teeth. In one trial, there was no pathological root resorption in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 1.50, 95% CI 0.29 to 7.73; Analysis 12.5).

Clinical failure

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 0.56, 95% CI 0.13 to 2.34; Analysis 13.1).

At 12 months, one trial, which randomised 20 teeth, assessed clinical failure (Gupta 2015). The results showed no statistically significant difference (RR 1.00, 95% CI 0.17 to 5.77).

Radiological failure

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 1.25, 95% CI 0.38 to 4.12; Analysis 13.2).

At 12 months, one trial, which randomised 20 teeth, assessed radiological failure (Gupta 2015). The results showed no statistically significant difference (RR 1.00, 95% CI 0.17 to 5.77).

Pain

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 0.56, 95% CI 0.13 to 2.34) (Analysis 13.3).

At 12 months, one trial, which randomised 20 teeth, assessed pain (Gupta 2015). The results showed no statistically significant difference (RR 1.00, 95% CI 0.17 to 5.77).

Pathological mobility

At six months, data were extractable from two RCTs totaling 50 teeth. In the two trials, there was no pathological mobility in any of the participants regardless of the intervention (Analysis 13.4).

At 12 months, one trial, which randomised 20 teeth, assessed pathological mobility (Gupta 2015). There was no pathological mobility in any of the participants regardless of the intervention.

Pathological radiolucency

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 0.25, 95% CI 0.03 to 2.08; Analysis 13.5).

At 12 months, one trial, which randomised 20 teeth, assessed pathological radiolucency (Gupta 2015). The results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.70).

Pathological root resorption

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 2.20, 95% CI 0.54 to 8.88; Analysis 13.5).

At 12 months, one trial, which randomised 20 teeth, assessed pathological root resorption (Gupta 2015). The results showed no statistically significant difference (RR 5.00, 95% CI 0.27 to 92.62).

Pulp canal obliteration

At six months, data were extractable from two RCTs totaling 50 teeth. In the two trials, there was no pulp canal obliteration in any of the participants regardless of the intervention (Analysis 13.6).

At 12 months, one trial, which randomised 20 teeth, assessed pulp canal obliteration (Gupta 2015). There was no pulp canal obliteration in any of the participants regardless of the intervention.

Clinical failure

At six months, data were extractable from two RCTs totaling 50 teeth. In one trial, there was no clinical failure in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.70; Analysis 15.1).

At 12 months, one trial, which randomised 20 teeth, assessed clinical failure (Gupta 2015). The results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.70).

Radiological failure

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 0.64, 95% CI 0.19 to 2.18; Analysis 15.2).

At 12 months, one trial, which randomised 20 teeth, assessed radiological failure (Gupta 2015). The results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.70).

Pain

At six months, data were extractable from two RCTs totaling 50 teeth. In one trial, there was no pain in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.70; Analysis 15.3).

At 12 months, one trial, which randomised 20 teeth, assessed pain (Gupta 2015). The results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.70).

Pathological mobility

At six months, data were extractable from two RCTs totaling 50 teeth. In the two trials, there was no pathological mobility in any of the participants regardless of the intervention (Analysis 15.4).

At 12 months, one trial, which randomised 20 teeth, assessed pathological mobility (Gupta 2015). There was no pathological mobility in any of the participants regardless of the intervention.

Pathological radiolucency

At six months, data were extractable from two RCTs totaling 50 teeth. In the two trials, there was no pathological radiolucency in any of the participants regardless of the intervention (Analysis 15.5).

At 12 months, one trial, which randomised 20 teeth, assessed pathological radiolucency (Gupta 2015). There was no pathological radiolucency in any of the participants regardless of the intervention.

Pathological root resorption

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 0.64, 95% CI 0.19 to 2.18; Analysis 15.6).

At 12 months, one trial, which randomised 20 teeth, assessed pathological root resorption (Gupta 2015). The results showed no statistically significant difference (RR 0.20, 95% CI 0.01 to 3.70).

Pulp canal obliteration

At six months, data were extractable from two RCTs totaling 50 teeth. In the two trials, there was no pulp canal obliteration in any of the participants regardless of the intervention (Analysis 15.7).

At 12 months, one trial, which randomised 20 teeth, assessed pulp canal obliteration (Gupta 2015). There was no pulp canal obliteration in any of the participants regardless of the intervention.

Clinical failure

At six and 12 months, data were extractable from two RCTs totaling 150 teeth. In the two trials, there was no clinical failure in any of the participants regardless of the intervention (Analysis 16.1).

At 24 months, one trial, which randomised 50 teeth, assessed clinical failure (Fernández 2013). The results showed no statistically significant difference (RR 3.00, 95% CI 0.13 to 70.30).

Radiological failure

At six months, data were extractable from two RCTs totaling 150 teeth. The pooled results showed no statistically significant difference (RR 1.29, 95% CI 0.33 to 5.08). Results were similar at 12 months (RR 1.86, 95% CI 0.52 to 6.59; Analysis 16.2).

At 24 months, one trial, which randomised 50 teeth, assessed radiological failure (Fernández 2013). The results showed no statistically significant difference (RR 3.00, 95% CI 0.33 to 26.92).

Overall failure

At six, 12 and 24 months, one trial, which randomised 50 teeth, assessed radiological failure (Fernández 2013). The results showed no statistically significant difference (RR 7.00, 95% CI 0.38 to 128.87).

Pain

At six and 12 months, data were extractable from two RCTs totaling 150 teeth. In the two trials, there was no pain in any of the participants regardless of the intervention (Analysis 16.3).

At 24 months, one trial, which randomised 20 teeth, assessed pain (Fernández 2013). There was no pain in any of the participants regardless of the intervention.

Soft tissue pathology

At six and 12 months, data were extractable from two RCTs totaling 150 teeth. In the two trials, there was no soft tissue pathology in any of the participants regardless of the intervention (Analysis 16.4).

At 24 months, one trial, which randomised 20 teeth, assessed soft tissue pathology (Fernández 2013). There was no soft tissue pathology in any of the participants regardless of the intervention.

Adjacent tissue inflammation

One trial, which randomised 50 teeth, assessed adjacent tissue inflammation (Fernández 2013). At six and 12 months, there was no adjacent tissue inflammation in any of the participants regardless of the intervention. At 24 months, the results showed no statistically significant difference (RR 3.00, 95% CI 0.13 to 70.30).

Pathological mobility

At six and 12 months, data were extractable from two RCTs totaling 50 teeth. In the two trials, there was no pathological mobility in any of the participants regardless of the intervention (Analysis 16.5).

At 24 months, one trial, which randomised 20 teeth, assessed soft tissue pathology (Fernández 2013). There was no pathological mobility in any of the participants regardless of the intervention.

Pathological radiolucency

At six and 12 months, data were extractable from two RCTs totaling 50 teeth. In the two trials, there was no pathological radiolucency in any of the participants regardless of the intervention (Analysis 16.6).

At 24 months, one trial, which randomised 20 teeth, assessed pathological radiolucency (Fernández 2013). There was no pathological radiolucency in any of the participants regardless of the intervention.

Pathological root resorption

At six months, data were extractable from two RCTs totaling 50 teeth. The pooled results showed no statistically significant difference (RR 1.29, 95% CI 0.33 to 5.08; Analysis 16.7). Results were similar at 12 months (RR 1.86, 95% CI 0.52 to 6.59).

At 24 months, one trial, which randomised 20 teeth, assessed pathological root resorption (Fernández 2013). The results showed no statistically significant difference (RR 3.00, 95% CI 0.33 to 26.92).

Clinical failure

At six months, data were extractable from two RCTs totaling 100 teeth. The results showed no statistically significant difference (RR 0.80, 95% CI 0.23 to 2.83; Analysis 17.1).

At 12 and 24 months, one trial, which randomised 70 teeth, assessed clinical failure (Yildirim 2016). There was no statistically significant difference (RR 3.00, 95% CI 0.33 to 27.46).

Radiological failure

At six months, one trial, which randomised 30 teeth, assessed radiological failure (Sabbarini 2008). There was a statistically significant difference in favour of EMD (RR 0.46, 95% CI 0.24 to 0.88).

At 24 months, one trial, which randomised 70 teeth, assessed radiological failure (Yildirim 2016). There was no statistically significant difference (RR 1.40, 95% CI 0.49 to 3.99).

Overall failure

At six, 12 and 24 months, one trial, which randomised 70 teeth, assessed overall failure (Yildirim 2016). There was no statistically significant difference (RR 3.00, 95% CI 0.33 to 27.46).

Pain

At six months, data were extractable from two RCTs totaling 100 teeth. The results showed no statistically significant difference (RR 0.40, 95% CI 0.08 to 1.92; Analysis 17.2).

At 12 and 24 months, one trial, which randomised 70 teeth, assessed pain (Yildirim 2016). There was no statistically significant difference (RR 1.00, 95% CI 0.07 to 15.36).

Soft tissue pathology

At six months, data were extractable from two RCTs totaling 100 teeth. In one trial, there was no soft tissue pathology in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 2.00, 95% CI 0.19 to 21.06; Analysis 17.3).

At 12 and 24 months, one trial, which randomised 70 teeth, assessed clinical failure (Yildirim 2016). There was no statistically significant difference (RR 2.00, 95% CI 0.19 to 21.06).

Pathological mobility

At six months, data were extractable from two RCTs totaling 100 teeth. In one trial, there was no pathological mobility in any of the participants regardless of the intervention. For the remaining trial, the results showed no statistically significant difference (RR 5.00, 95% CI 0.26 to 96.13; Analysis 17.4).

At 12 and 24 months, one trial, which randomised 70 teeth, assessed pathological mobility (Yildirim 2016). There was no statistically significant difference (RR 5.00, 95% CI 0.25 to 100.53).

Pathological radiolucency

At 24 months, one trial, which randomised 70 teeth, assessed pathological radiolucency (Yildirim 2016). There was no statistically significant difference (RR 1.67, 95% CI 0.43 to 6.45).

Pathological root resorption

At 24 months, one trial, which randomised 70 teeth, assessed pathological root resorption (Yildirim 2016). There was no statistically significant difference (RR 1.00, 95% CI 0.15 to 6.71).

Clinical failure

Ozalp 2005, which randomised 40 teeth, assessed calcium hydroxide versus ZOE based on clinical failure at six and 12 months; there was no statistically significant difference at either time point (six months: RR 3.00, 95% CI 0.13 to 69.52; 12 months: RR 9.00, 95% CI 0.52 to 156.92).

Nadkarni 2000, which randomised 70 teeth, assessed calcium hydroxide versus ZOE based on clinical failure at nine months. There was no statistically significant difference (RR 0.25, 95% CI 0.03 to 2.13).

Radiological failure

At six months, data were extractable from two RCTs totaling 110 teeth. The pooled results showed no statistically significant difference in radiological failure between calcium hydroxide and ZOE (RR 2.50, 95% CI 0.50 to 12.50; Analysis 18.1).

Ozalp 2005 provided data on radiological failure at 12 months. There was no statistically significant difference (RR 9.00, 95% CI 0.52 to 156.92).

Overall failure

Nadkarni 2000, which randomised 70 teeth, provided data on overall failure. There was no statistically significant difference at nine months (RR 0.50, 95% CI 0.10 to 2.56). Data at other time points were unavailable.

Pain

At six months, data were extractable from two RCTs totaling 110 teeth. The pooled RR was 1.00 (95% CI 0.14 to 6.90; Analysis 18.2).

Ozalp 2005, which randomised 40 teeth, provided data on pain at 12 months. There was no statistically significant difference (RR 3.00, 95% CI 0.13 to 69.52).

Pathological mobility

At six months, data were extractable from two RCTs totaling 110 teeth. The pooled RR was 1.00 (95% CI 0.14 to 6.90; Analysis 18.3).

Ozalp 2005, which randomised 40 teeth, also provided data on pathological mobility at 12 months. There was no statistically significant difference (RR 3.00, 95% CI 0.13 to 69.52).

Pathological radiolucency

Two trials (110 teeth) provided data on pathological radiolucency at six months (Nadkarni 2000; Ozalp 2005). The pooled results showed no statistically significant difference between calcium hydroxide and ZOE (RR 1.50, 95% CI 0.26 to 8.72; Analysis 18.4).

Ozalp 2005 provided data at 12 months. There was no statistically significant difference (RR 5.00, 95% CI 0.26 to 98.00).

Pathological root resorption

Ozalp 2005 provided data on pathological root resorption. There was no pathological root resorption in any of the participants regardless of the time of assessment.

Filling material anomaly

Ozalp 2005, which randomised 40 teeth, provided data on filling material anomaly. There was no filling material anomaly in any of the participants regardless of the time of assessment.

Clinical failure

Data were extractable from two trials, totaling 62 teeth (Al‐Ostwani 2016; Subramaniam 2011). There were no statistically significant differences at six (RR 0.60, 95% CI 0.08 to 4.29) and 12 months (RR 0.71, 95% CI 0.15 to 3.33; Analysis 19.1).

Radiological failure

At six and 12 months, data were extractable from two trials, totaling 62 teeth. There was no statistically significant difference (RR 1.00, 95% CI 0.31 to 3.27; Analysis 19.2).

Pain

At six and 12 months, one trial, which randomised 30 teeth, assessed pain (Subramaniam 2011). There was no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.58).

Pathologic mobility

At six and 12 months, one trial, which randomised 30 teeth, assessed pathologic mobility (Subramaniam 2011). There was no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.58).

Soft tissue pathology

At six and 12 months, one trial, which randomised 30 teeth, assessed soft tissue pathology (Subramaniam 2011). There was no soft tissue pathology in any of the participants regardless of the intervention.

Adjacent tissue inflammation

At six and 12 months, one trial, which randomised 30 teeth, assessed adjacent tissue inflammation (Subramaniam 2011). There were no adjacent tissue inflammation in any of the participants regardless of the intervention.

Pathologic radiolucency

At six and 12 months, data were extractable from two trials, totaling 62 teeth. There was no pathologic radiolucency in one trial; for the remaining trial, there was no statistically significant difference (RR 1.33, 95 CI 0.35 to 5.03) (Al‐Ostwani 2016).

Pathologic root resorption

At six and 12 months, one trial, which randomised 30 teeth, assessed adjacent tissue inflammation (Subramaniam 2011). There was no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.58).

Clinical failure

At six months, data were extractable from two trials totaling 92 teeth. In one trial, there was no clinical failure in any of the participants regardless of the intervention. For the remaining trial, there was no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.58; Analysis 20.1).

At 12 months, one trial, which randomised 30 teeth, assessed clinical failure (Subramaniam 2011). There was no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.58).

Radiological failure

At six months, data were extractable from two trials totaling 92 teeth. There was no statistically significant difference (RR 2.02, 95% CI 0.79 to 5.15; Analysis 20.2).

At 12 months, one trial, which randomised 30 teeth, assessed radiological failure (Subramaniam 2011). There was no statistically significant difference (RR 0.35, 95% CI 0.02 to 8.39).

Pain

At six months, data were extractable from two trials totaling 92 teeth. In one trial, there was no pain in any of the participants regardless of the intervention. For the remaining trial, there was no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.58; Analysis 20.3).

At 12 months, one trial, which randomised 30 teeth, assessed pain (Subramaniam 2011). There was no statistically significant difference (RR 0.35, 95% CI 0.02 to 8.39).

Soft tissue pathology

At six months, data were extractable from two trials totaling 92 teeth. In one trial, there was no soft tissue pathology in any of the participants regardless of the intervention. For the remaining trial, there was no statistically significant difference (RR 0.33, 95% CI 0.01 to 7.58; Analysis 20.4).

At 12 months, one trial, which randomised 30 teeth, assessed soft tissue pathology (Subramaniam 2011). There was no statistically significant difference (RR 0.35, 95% CI 0.02 to 8.39).

Pathological mobility

At six months, data were extractable from two trials totaling 92 teeth. In the two trials, there was no pathological mobility in any of the participants regardless of the intervention (Analysis 20.5).

At 12 months, one trial, which randomised 30 teeth, assessed pathological mobility (Subramaniam 2011). There was no pathological mobility in any of the participants regardless of the intervention.

Pathological radiolucency

At six months, data were extractable from two trials totaling 92 teeth. There was no statistically significant difference (RR 2.02, 95% CI 0.79 to 5.15; Analysis 20.6).

At 12 months, one trial, which randomised 30 teeth, assessed pathological radiolucency (Subramaniam 2011). There was no statistically significant difference (RR 0.35, 95% CI 0.02 to 8.39).

Pathological root resorption

At six months, data were extractable from two trials totaling 92 teeth. In the two trials, there was no pathological root resorption in any of the participants regardless of the intervention (Analysis 20.7).

At 12 months, one trial, which randomised 30 teeth, assessed pathological root resorption (Subramaniam 2011). There was no pathological root resorption in any of the participants regardless of the intervention.

Succedaneous tooth anomaly (radiographically)

One trial, which randomised 30 teeth, assessed Metapex versus Endoflas based on unerupted succedaneous tooth anomaly (radiographically) (Subramaniam 2011). There was no cases in any of the participants regardless of the delay.

Clinical failure

At six months, data were extractable from four RCTs totaling 287 teeth. In three of the trials, there was no clinical failure in any of the participants regardless of the intervention. From the remaining trial, the results showed no statistically significant difference in clinical failure with Vitapex compared with ZOE (RR 0.33, 95% CI 0.01 to 7.84). At 12 months, in two trials, there was no clinical failure in any of the participants regardless of the intervention. From the two remaining trials, the pooled results showed a statistically significant difference (RR 4.75, 95% CI 1.21 to 18.55) with evidence of statistical heterogeneity among included trials (I² = 78%; Analysis 21.1). Pramila 2016 (86 teeth randomised) assessed clinical failure at 30 months, and reported no failures.

Radiological failure

At six months, data were extractable from four RCTs totaling 287 teeth. In one of the trials, there was no radiological failure in any of the participants regardless of the intervention. From the three remaining trials, the pooled results showed no statistically significant difference in radiological failure with Vitapex compared with ZOE (RR 2.36, 95% CI 0.86 to 6.50). At 12 months, the pooled results showed a statistically significant difference (RR 6.56, 95% CI 2.58 to 16.67), with evidence of statistical heterogeneity among included trials (I² = 82%; Analysis 21.2). At 30 months, one trial, which randomised 86 teeth, assessed radiological failure, and there was a statistically significant difference (RR 19.00, 95% CI 1.14 to 316.52) (Pramila 2016).

Overall failure