The effect of adjusting the pH of local anaesthetics in dentistry: A systematic review and meta-analysis
PV Aulestia-Viera, MM Braga, MA Borsatti
1 Department of Stomatology, School of Dentistry,
2 Department of Pediatric Dentistry, School of Dentistry, University of São Paulo, São Paulo, Brazil.
Abstract
The acidic nature of commercial local anaesthetics (LAs) can cause pain during infiltration and delay the onset of anaesthesia. It is suggested that adjusting the pH of anaesthetic agents could minimize these effects. This systematic review aimed to evaluate the efficacy of buffered LAs in reducing infiltration pain and onset time in during dental procedures. MEDLINE, Embase, Scopus, and Scielo databases were searched up to April 2017. Randomized controlled trials comparing buffered and unbuffered LAs for intraoral injections were included. Risk of bias was assessed using the Cochrane Collaboration tool. Data upon injection pain and onset time were pooled in a random-effects model. Subgroup analyses compared normal and inflamed tissues, and terminal infiltrations and inferior alveolar nerve (IAN) blocks. Meta-regressions were performed to explain heterogeneity. Fourteen articles were included in this review. Lidocaine with epinephrine was the most used anaesthetic combination. Non-lidocaine studies (n= 2) were not pooled in the meta-analysis. Buffered lidocaine did not result in less pain during intraoral injections: mean difference –6.4 (95% CI –12.81 to 0.01) units in a 0–100 scale. Alkalinized lidocaine did not reduce the onset time in normal tissues when terminal infiltration techniques were used, but resulted in a more rapid onset for IAN blocks (–1.26 minutes) and in inflamed tissues (–1.37 minutes); however, this change may not be clinically relevant, considering the time required to prepare the buffered agent. Studies performed using other anaesthetic salts did not show robust and clinically significant results in favour of alkalinization.
INTRODUCTION
Local anaesthetics (LAs) are essential for pain management in dentistry. They reversibly block nerve conduction, enabling a wide variety of procedures to be performed (Kumar et al. 2015). However, LA injections usually induce fear and anxiety in patients of all ages, are often the reason for postponement of dental visits and even rejection of treatment (Davoudi et al. 2016). Consequently, effective measures to mitigate pain and anxiety and to reduce the need for sedation should be employed.
Patients regularly report a burning and stinging feeling during the injection of anaesthetics. One explanation for this uncomfortable sensation is the acidic nature of commercially available local anaesthetics (pH 4.5, approximately), which are adjusted to this pH to prolong shelf life (Hogan et al. 2011, Malamed et al. 2013, Harreld et al. 2015, Schellenberg et al. 2015, Shurtz et al. 2015). Molecules in the cartridge mostly exist in a water-soluble state and are acidic (RNH+). Conversely, for the anaesthetic to penetrate the nerve sheath, it must be in its un-ionized free base form; then, the H+ ion needs to dissociate from the ionized molecule (Kumar et al. 2015). Since the physiological pH is about 7.4, an increase in H+ in the tissues could cause pain by activating nociceptors such as the acid-sensing ion channels (ASICs) (Cepeda et al. 2010, Saatchi et al. 2016). Buffering the local anaesthetic solution could produce less pain because fewer acid-sensing nociceptors would be activated. In addition, it is believed that by using alkalinized agents, the body takes less time to change the solution from the ionized to the un-ionized form, increasing nerve penetration and producing rapid onset of the anaesthetic effect (Malamed et al. 2013, Shyamala et al. 2016).
The efficacy and onset time of anaesthetics could be affected in inflamed tissues, such as in the cases of pulpitis and abscesses (Malamed et al. 2013). The Henderson-Hasselbalch equation demonstrates that if a local anaesthetic solution is buffered to a pH closer to its pKa, more of the free base form will be available (Balasco et al. 2013). Infected tissues have a pH as low as 5.0, which favours the ionized configuration of the local anaesthetic and reduces anaesthetic penetration into the nerve (Cepeda et al. 2010).
Although many studies have evaluated the alkalinization of LAs with buffering agents, its overall efficacy in dental anaesthesia has not been reviewed comprehensively. This gap in knowledge has discouraged manufacturers and clinicians from investing resources and time in performing the pH adjustment. Former systematic reviews have evaluated the intensity of pain during extraoral injections but none of them have analysed the anaesthetic onset time (Cepeda et al. 2010, Davies 2003, Hanna et al. 2009).
The purpose of this systematic review and meta-analysis of randomized clinical trials was to investigate the efficacy of buffering local anaesthetics in reducing infiltration pain and anaesthesia onset time in dentistry.
REVIEW
Materials and Methods
This systematic review was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (Liberati et al. 2009). A review protocol was registered with the public registry of systematic reviews PROSPERO (CRD42016051216).
Literature search
The literature search in MEDLINE, Embase, Scopus and Scielo databases included articles published from inception to April 2017. The search criteria included key words for local anaesthetics, pH regulation, clinical trials, and dentistry (Table 1). Manual search was also performed and no language restrictions were applied.
Studies were included if they were clinical trials using buffered local anaesthetics injected intraorally. Parallel-group, split-mouth, and crossover trials were included. Exclusion criteria included nonrandomized or non-controlled studies, studies that enrolled patients under 18 years of age or those that did not specify the age of the participants, studies varying the anaesthetic concentration between the groups, studies with temperature variation of the anaesthetic solutions between the groups, and studies that did not assess self-reported pain at injection or onset time of anaesthesia.
Data collection and processing
Relevant data were collected using a structured data extraction form. The authors of the included studies were contacted to obtain additional data, when needed. Two researchers reviewed the articles at each stage according to the criteria mentioned above. Possible divergences were discussed until consensus was achieved. Cohen kappa value was calculated to ensure inter-rater reliability, which ranged between 0.85 and 1.0.
The primary outcome of interest was pain intensity during deposition of the local anaesthetic solution.
Since different length scales were used to assess pain in the included articles (visual analogue scales, ordinal scales, and numeric rating scales), all pain scores were converted to a 0–100 scale so that the articles could be compared (Hogan et al. 2011). The secondary outcome was onset time. All measures of time were normalized in minutes. Mean, standard deviation (SD), and sample size of each group (buffered and unbuffered) were collected. When the means or SDs were not available in an article, these were estimated from other published measures of central tendency and dispersion, and P values or the authors were contacted. Table 2 contains information on how the data of each study were processed individually.
Risk of bias was assessed with the Cochrane Collaboration tool (Higgins & Green 2011). Each study was categorized as low, unclear, or high risk of bias in the categories of randomization, allocation concealment, blinding, completeness of outcome data, selective outcome reporting, and other potential biases.
Statistical analyses
Data were combined using a random-effects model (RevMan 5.3; Cochrane Collaboration, Copenhagen, Denmark). Mean differences and 95% confidence intervals (CIs) were used as measures of effect. Heterogeneity was assessed with the Q-Cochran (occurrence of heterogeneity), Tau2 (estimate of between-study variance), and I2 (level of inconsistency) statistics. Sensitivity analysis was performed to check the possible interference of one study or a group of studies in the pooled findings. Subgroup analyses were planned a priori to evaluate the differences in effects between the types of injections (inferior alveolar nerve block or other terminal infiltrations) and local tissue conditions (normal or inflamed). Inflammation was considered in cases of abscess, pulpitis and, periapical lesions. A funnel plot visual analysis and Egger’s test were applied to assess publication bias using Comprehensive Meta-analyses V3 (Biostat Inc., Englewood, NJ, USA).
To identify the possible sources of heterogeneity, meta-regression was used (Comprehensive Meta- analyses V3) along with subgroup analyses. Statistical significance was considered when p < 0.05.
Results
A total of 1075 articles were obtained from the electronic search, and seven articles were obtained from the manual search (Figure 1). Thirty-one articles were selected for full-text assessment; however, three full-texts were not available even after meticulous search and attempts to contact the authors (Al-Sultan 2004, Crose 1991, Kim et al. 2005). Finally, 14 articles were included in this systematic review (Table 2); but three of these did not provide the data necessary for meta-analysis (Lem 1991, Malamed et al. 2013, Phero et al. 2017).
All the studies were published in English and included individuals with good general health.
Moreover, all the studies were randomized and double-blinded, except for one study that did not specify blinding (Shyamala et al. 2016). Eight studies with a cross-over or split-mouth design (Lem 1991, Primosch & Robinson 1996, Whitcomb et al. 2010, Hobeich et al. 2013, Malamed et al. 2013, Comerci et al. 2015, Shurtz et al. 2015, Phero et al. 2017) tested experimental and control conditions in the same person, whereas six studies had a parallel design (Al-Sultan et al. 2006, Kashyap et al. 2011, Balasco et al. 2013, Harreld et al. 2015, Schellenberg et al. 2015, Shyamala et al. 2016). Pain was measured on a variety of scales (Table 2), while the onset of anaesthesia was assessed by electric pulp testing (Whitcomb et al. 2010, Hobeich et al. 2013, Shurtz et al. 2015) or soft tissue probing (Al-Sultan et al. 2006, Kashyap et al. 2011, Shyamala et al. 2016).
Lidocaine was used as the anaesthetic agent in 12 studies; bupivacaine was used in one study (Shyamala et al. 2016), and articaine was used in one study (Shurtz et al. 2015). These two non- lidocaine studies were not pooled in the meta-analysis. All of the studies used epinephrine as the vasoconstrictor. In five studies, the oral mucosa was anaesthetized with a topical anaesthetic (benzocaine) before infiltration (Balasco et al. 2013, Comerci et al. 2015, Harreld et al. 2015, Schellenberg et al. 2015, Shurtz et al. 2015). One study was carried out in patients with periapical lesions (Al-Sultan et al. 2006); two studies were performed in patients with abscesses (Balasco et al. 2013, Harreld et al. 2015), and one study was performed in patients with pulpitis (Schellenberg et al. 2015). Patients in the other studies were considered to have normal local tissue conditions.
The buffer agent used in all the studies was sodium bicarbonate (SB). The buffered anaesthetic solution was always prepared immediately before injection, manually (Lem 1991, Primosch & Robinson 1996, Al-Sultan et al. 2006, Whitcomb et al. 2010, Kashyap et al. 2011, Phero et al. 2017) or by using a precision mixing device (Balasco et al. 2013, Hobeich et al. 2013, Malamed et al. 2013, Comerci et al. 2015, Harreld et al. 2015, Schellenberg et al. 2015, Shurtz et al. 2015, Shyamala et al. 2016). The most frequent buffer concentration was 8.4%, which was used in 12 studies. Most of the studies reported the final pH of the buffer anaesthetic solution which ranged between 6.5 and 7.59.
None of the included studies reported the presence of precipitate at these concentrations.
Characteristics of the included studies are provided in Table 2.
Only four studies had an overall low risk of bias (all categories evaluated as low risk). Three studies had a “high risk” score for reporting bias, because one or more outcomes of interest (pain and onset time) were reported incompletely, and thus they could not be entered in the meta-analysis. Sequence generation and blinding of parties were always mentioned; however, the mechanism was often not specified and was assessed as unclear. The allocation concealment process was not explained in over 50% of the studies (Figure 2).
Quantitative synthesis
Nine studies were pooled in the meta-analysis of self-reported pain for buffered versus unbuffered local anaesthetic injection. Two studies had more than two treatment arms. These treatments were split and referred to as Comerci 2015a; Comerci 2015b (Comerci et al. 2015) and Hobeich 2013a; Hobeich 2013b (Hobeich et al. 2013). Buffering lidocaine to values close to that of the physiological pH did not result in significant pain reduction, and considerable heterogeneity was observed (Figure 3). Removing the study whose mean was farther from the overall mean difference (Kashyap et al. 2011) did not change the results.
The use of different pain scales helped in explaining a major part of the heterogeneity. The study design (cross-over or split-mouth versus parallel trials), the use of topical anaesthetics, buffer concentration, and type of injection failed to explain the heterogeneity of the results for the pain outcome (Table 3).
Four studies were pooled in the meta-analysis of onset time for buffered versus unbuffered local anaesthetics. One study was split and evaluated as: Hobeich 2013a and Hobeich 2013b (Hobeich et al. 2013). Buffering lidocaine to close to the physiological pH did not result in a significant reduction in onset time in normal tissues when terminal infiltration techniques were used (Figure 4). Buffering the LA resulted in reduced onset time when administered in inferior alveolar nerve (IAN) blocks (Figure 4B). The only study conducted in inflamed tissue (Al-Sultan et al. 2006) also reported a significant reduction in onset time (Figure 4A).
The study design and the method used to measure the onset of anaesthesia explained a large amount of heterogeneity, as observed in the subgroup analysis (a posteriori) and meta-regression (Table 3). In parallel-group studies that used gingival probing to assess the onset of anaesthesia, the decrease in onset time with buffered lidocaine compared with the unbuffered type was –1.27 minutes (95% CI –1.41 to –1.12; I2 = 0%) and it was -0.01 minutes (95% CI -0.39 to 0.38; I2 = 0%) in cross-over or split mouth trials that applied electric pulp testing for onset evaluation (p< 0.00001 between the subgroups). Buffer concentration failed to explain the heterogeneity of the results for onset time (Table 3).
The visual inspection of funnel plots and Egger’s test did not suggest the presence of publication bias related to pain (p= 0.46) or onset time (p= 0.26) in the total sample analyses.
DISCUSSION
Commercial cartridges of local anaesthetics with adrenaline are usually up to 1000 times more acidic than the physiological pH (Frank & Lalonde 2012). This is a probable cause for pain on injection and delay in the onset of anaesthesia, especially in the presence of local inflammation. In an attempt to avoid these inconveniences, the concomitant use of buffers has been studied.
The present meta-analysis reveals that increasing the pH of lidocaine close to that of the physiological pH does not decrease the pain associated with its injection. The overlap with the line of no effect is subtle (Figure 3); nevertheless, a slight change in pain (6 points on a 100-point scale) may not be noticeable by most patients. Previous studies have determined that the minimal perceptible difference in pain is about 15 points on a 0–100 scale (Cepeda et al. 2003, Jensen et al. 2003, Kendrick & Strout 2005, Bijur et al. 2010).
Although former systematic reviews have already explored the impact of buffered anaesthetics on pain during injection, some methodological differences should be addressed. Cepeda et al. (2010) and Hanna et al. (2009) conducted meta-analyses including randomized clinical trials that used lidocaine for a variety of extraoral procedures. They found that buffered lidocaine decreased the pain on injection by about 1-unit on a 10-point scale, a larger effect in favour of the alkalinized group compared to the present study. This difference may be related to the site of injection, as injection pain is usually more intense in the skin than in the oral mucosa because of the thicker and denser layers of the skin (Karkut et al. 2010). In another systematic review, Davies (2003) included studies that exclusively involved emergency department practices and concluded that although buffered anaesthetics were found to be favourable in most of the studies, certain sites (including intraoral and penile blocks) seemed to be less affected.
Besides the absence of a significant effect on pain reduction, the onset time did not show a considerable decrease with the use of buffered lidocaine for terminal infiltrations in normal tissues, suggesting that in these conditions, the body’s buffering capacity may increase pH rapidly enough to make anaesthetic buffering clinically ineffective (Shurtz et al. 2015). A significant reduction was observed in the group of studies that performed IAN blocks (-1.26 minutes) and in the subgroup that evaluated inflamed tissues (-1.37 minutes). However, this magnitude of reduction is not a relevant clinical time gain as 1.3 minutes or more would be the time required to prepare the buffered LA at chairside. Therefore, affirming that this effect is robust and clinically relevant is not possible.
The non-lidocaine studies had mixed results (Table 2). The use of buffered articaine was not associated with significant changes for pain on injection or onset of pulpal anaesthesia (Shurtz et al. 2015). On the other hand, buffered bupivacaine resulted in less pain on injection and faster onset of soft tissue anaesthesia (Shyamala et al. 2016). Nevertheless, the mean difference in the onset time between buffered and unbuffered bupivacaine was 1.04 minutes. Thus, similar questions on the clinical significance of this reduction arise. In the same study, the pain on injection was assessed with a numerical scale (10-point VAS); however, the results were presented and statistically analysed in a categorical manner (Chi square test), which makes the pain data more subjective.
High values of heterogeneity could represent a limitation of the meta-analysis, although the use of a random-effects model incorporated this heterogeneity and reported an average intervention effect across the populations being studied (Hogan et al. 2011). Heterogeneity may be influenced by the nature of the main outcome – pain – which is a body response to injury and is influenced by a person’s psychology and memories (Astramskaite & Juodzbalys 2017). The presence of heterogeneity may also reflect the clinical heterogeneity caused by multiple factors, such as the use of topical anaesthetics, study design, pain scale, population characteristics (i.e. age, gender, oral health status), individual pain threshold, needle size, speed of injection, and temperature of the anaesthetic solution (Cepeda et al. 2010).
Heterogeneity concerning pain on injection was mainly explained by the different scales used. It has been pointed out that the Visual Analogue Scale (VAS), the Numerical Rating Scale (NRS), and ordinal scales are valid and reliable. While VAS and NRS are highly correlated, the correlation coefficients between VAS or NRS and ordinal scales vary between studies (Williamson & Hoggart 2005, Hjermstad et al. 2011, Bahreini et al. 2015). The normalization of the different scales used in the included studies enabled their comparison in the meta-analysis; however, the lack of a perfect correlation between the different scales may have influenced the heterogeneity of the results.
The number of response options used in a scale is also important. A small number of categories in a scale requires a larger change in pain before the score changes. In contrast, a scale that presents too many levels could also be fruitless as it may have more levels of discrimination than most patients use (Hjermstad et al. 2011, Williamson & Hoggart 2005). Further clinical trials may standardize the use of an appropriate pain scale (i.e. a 0–10 NRS) as this factor might have a significant influence on the results.
The method employed to measure the onset of anaesthesia explained the heterogeneity of the results for onset time. While gingival probing evaluates soft tissue anaesthesia, which is more superficial, electric pulp testing evaluates pulpal anaesthesia, which implies the diffusion of the anaesthetic through the cortical bone. An exception is the IAN block where the anaesthetic is supposed to reach the nerve before its entry into the mandibular canal, requiring anaesthetic diffusion mainly in the soft tissue. The subgroup analysis showed that the buffered lidocaine was effective in reducing the onset time only in the group that evaluated soft tissue anaesthesia (–1.27 minutes) and in the group that performed IAN blocks (–1.26 minutes), suggesting that the buffered solutions can lead to a faster onset in soft tissue anaesthesia, but not necessarily in more profound anaesthesia, as in the case of the pulp (Hobeich et al. 2013).
The study design also explained the heterogeneity in the onset time outcome. Parallel studies had greater inter-individual variation, as observed in the real clinical situations. This difference is reduced in cross-over and split-mouth studies, where inter-individual variation is eliminated. These three causes of heterogeneity (pain scale, onset assessment method, and study design) are simple adjustments that could be made in the methodology of clinical trials and lead to more comparable studies in future meta-analyses.
In addition to onset time, mixed results about other parameters related to anaesthetic efficacy have been reported in the dental literature. Shyamala et al. (2016) indicated that the anaesthetic duration remains unchanged when the LA is buffered; conversely, Brennan et al. (1987) affirmed that the duration is diminished. Rood (1977), Parirokh (2016), Saatchi et al (2015) and Schellenberg et al. (2015) observed that alkalinization did not increase the anaesthetic success in the presence of acute inflammation, and Shurtz et al. (2015) and Whitcomb et al. (2010) did not find a significant benefit in normal tissues either. However, Saatchi et al. (2016) reported that a buccal infiltration of Epinephrine bitartrate increased the success rate of inferior alveolar nerve block in the presence of pulpitis.
To reduce the risk of bias, future studies should be more meticulous in explaining the process of sequence generation and allocation concealment, and should clearly report central tendency and dispersion measures of their outcomes, even when no statistical significance is found between the groups.
CONCLUSIONS
On the basis of this review, routine alkalinization of local anaesthetics is not recommended in dentistry. Adjusting the pH of lidocaine was not effective in reducing the pain of intraoral injections in normal or inflamed tissues. The onset of anaesthesia was not improved by using buffered lidocaine in normal tissues. A slight reduction of the onset time was observed in inflamed tissues and when the IAN block technique was used; however, this change may not be clinically relevant considering the time required to prepare the buffered agent. Studies performed using other anaesthetic salts did not show robust and clinically significant results in favour of alkalinization.