The aim of this comprehensive meta-analysis was to provide evidence-based data to test whether oral contraceptive (OC) use can promote the incidence of dry socket (DS) in females following impacted mandibular third molar extraction. PubMed, the Cochrane Library, and Elsevier Science Direct databases were searched. The pooled risk ratio (RR) with 95% confidence interval (CI) was calculated using fixed-effects or random-effects model analysis. Heterogeneity among studies was evaluated with the Cochran test and I 2 statistic. Study quality was assessed with the Newcastle–Ottawa scale. Of 70 articles identified in the search, 12 reporting 16 clinical controlled trials were included in this study. The incidence of DS was significantly greater in the OC groups than in the control groups (RR 1.80, 95% CI 1.33–2.43). Subgroup analyses showed that the unit assessed (tooth or patient), the region in which the study was conducted, and the intervention were not related to the incidence of DS in females taking OC after impacted mandibular third molar extraction. The sensitivity analysis showed no significant change when any one study was excluded. Publication bias was also not detected. This study suggests that OC use may promote the incidence of DS in females following impacted mandibular third molar extraction.
The term ‘dry socket’ was first used in 1896 and refers to a potential postoperative complication after tooth removal. Most studies have described the incidence of dry socket (DS) to be in the range of 1–4% for all routine tooth extractions and 5–30% for impacted mandibular third molar extractions. Typically, DS starts 1–3 days after tooth extraction and lasts for 5–10 days. Clinical features of DS are severe throbbing pain that starts 1–3 days after extraction, marked halitosis, and a foul taste. An extraction socket devoid of clot with exposed bone that may be filled with food debris, oedema of the surrounding gingiva, and regional lymphadenitis may also be seen.
Several theories have been postulated for the aetiology of DS. These include bacterial infection, trauma, biochemical agents, and fibrinolysis of the socket blood clot. In spite of this, the precise aetiology of DS remains unclear. Over the past few decades, multiple factors associated with the occurrence of DS have been increasingly recognized, involving both general and local factors, such as age, gender, oral contraceptive (OC) use, menstrual cycle, smoking, length of the operation, pericoronitis, state of eruption, and the use of antimicrobial drugs and mouth rinses.
Among these risk factors, numerous studies on the effect of OC use on the development of DS in females following impacted mandibular third molar extraction have been reported, with conflicting results. Some studies have concluded that the incidence of DS is higher among females taking OCs. However, others have not detected any association. For dental practitioners, a better understanding of the effect of OC use in females undergoing extraction would be beneficial for prudent decision-making on whether and how teeth should be removed.
The meta-analysis is a powerful tool for summarizing the results of different studies by producing a single estimate of the major effect with enhanced precision. Thus, the present meta-analysis was performed to better quantify the magnitude of the association of OC use with the development of DS in females.
Materials and methods
Search methods and key words
The electronic databases of PubMed, the Cochrane Library, and Elsevier Science Direct were searched for relevant studies. An upper date limit of 30 April 2014 was applied; there was no lower date limit. The key words were dry socket OR alveolar osteitis OR fibrinolytic alveolitis OR alveolitis sicca dolorosa OR localized osteomyelitis OR delayed extraction wound healing AND contraceptive OR contraceptives or OC. There were no language restrictions.
Inclusion and exclusion criteria
Inclusion criteria were the following: (1) prospective controlled clinical trials; (2) OC use examined as a risk factor for the development of DS in patients; (3) the measure of effect with its variance was provided or could be calculated indirectly; and (4) similar criteria were used to diagnose DS.
Exclusion criteria were the following: (1) animal study; (2) case report, review, qualitative research, case series, or editorial; (3) the effect of OC use was not investigated; and (4) full text not available.
Data were extracted independently by two review authors (JLX and LS) using a previously prepared data extraction form. The following information was extracted from each study: first author and year of publication, study design, sample size, populations from which the patients and controls were selected, and the potential confounders controlled for.
Quality assessment of the studies included
The Newcastle–Ottawa scale (NOS) was used to assess the quality of the methodology of all studies included. The NOS comprises eight questions assessing categories of selection, comparability, and outcome for which nine possible points (stars) may be awarded. The quality of the studies was determined independently by two reviewers (JLX and LS). Disagreements between the two review authors were resolved by consensus with a third reviewer (RX).
To evaluate if there was a difference in the incidence of DS between the OC groups and control groups, a pooled risk ratio (RR) with the 95% confidence interval (CI) was calculated. All effect sizes and 95% CIs were calculated on the basis of fixed- or random-effects models using Stata 11.0 software (StataCorp LP, College Station, TX, USA). The significance of any discrepancies in the estimates of the effects between OC groups and control groups was assessed by means of Cochran’s test for heterogeneity and a measure of I 2 . If there was no statistical difference for heterogeneity ( P > 0.05), a fixed-effects model was to be used to analyze the data; in the converse situation, a random-effects model was to be applied. To explore the sources of heterogeneity, a subgroup analysis was conducted according to the unit assessed (tooth or patient), the region in which the study was performed, and the intervention. Publication bias was evaluated using Egger’s test. In the case of publication bias, the funnel plot will be asymmetric, or the P -value <0.05 by Egger’s test.
The electronic databases PubMed, the Cochrane Library, and Elsevier Science Direct were searched for relevant studies up to 30 April 2014. A total of 70 articles were identified after extensive searching. Through title and abstract analysis, 54 articles were excluded for following reasons: no full text available, intervention unclear, not related to the topic, and duplicate information. The full text of the remaining 16 articles was then screened according to the selection criteria. Finally, 12 articles reporting 16 studies were included in the final meta-analysis ( Fig. 1 ).
Description of the studies
Details of the studies included in the final analysis are given in Table 1 . The articles by Butler and Sweet and Nordenram and Grave each reported two studies, and the article by Bonine reported three studies. All of the studies included were cohort studies. The study populations resided in six different countries; seven studies were conducted in the USA and the others were conducted in Sweden, the UK, Italy, Spain, and Denmark. The year of publication ranged from 1974 to 2010. OCs started to come into widespread use from the 1960s onwards. Seven studies used each tooth extraction site as the assessment unit, while another three studies used each patient as the assessment unit; the remaining two studies used either each tooth extraction site or each patient as the assessment unit, as only one tooth was removed at a time in each patient. In all eligible studies, those taking OCs comprised the test group and those not taking OCs comprised the control group. In two studies, all female patients were prescribed prophylactic antibiotics after tooth extractions, while in five trials, chlorhexidine or saline solution rinse was given to patients following tooth extractions. The duration of follow-up for all of the studies included was within the range of 7 days, based on the diagnostic criteria for DS of 2002.
|Author||Location||No. DS/taking OC (test group)||No. DS/not taking OC (control group)||Unit studied||Intervention||Follow-up, days||Age, years||Newcastle–Ottawa scale total score|
|Butler and Sweet, 1977||USA||5/18||6/87||Tooth||Sterile saline solution rinse||4, 5, 6||15–30||5|
|Sweet and Butler, 1978||USA||5/84||8/234||Tooth||Sterile saline solution rinse||4, 5, 6||NR||7|
|Gersel-Pedersen, 1979||Denmark||5/16||12/94||Tooth||CHX rinse||4, 5, 6||NR||7|
|Nordenram and Grave, 1983||Sweden||11/39||7/39||Tooth/patient||Sterile saline solution rinse||1, 4, 7||Test: 18–37, mean 27||5|
|7/39||3/39||Control: 17–34, mean 24|
|Larsen, 1991||USA||3/22||16/52||Tooth||CHX rinse||7||NR||7|
|Bonine, 1995||USA||0/4||8/44||Tooth||Antibiotics/saline solution rinse||7||Mean 22||6|
|Hermesch et al., 1998||USA||21/59||24/111||Patient||None||3–4||NR||5|
|Garcia et al., 2003||Spain||10/87||7/180||Tooth/patient||Antibiotics||5||17–45||7|
|Test: mean 25|
|Control: mean 26|
|Blondeau and Daniel, 2007||UK||10/111||6/80||Patient||None||2–5||NR||6|
|Sivolella et al., 2010||Italy||1/38||1/80||Patient||Antibiotics||7–8||18–45||6|
|Test: mean 26|
|Control: mean 26|
Quality assessment of the studies included
The results of the NOS analysis for cohort studies are included in Table 1 . Most were of moderate quality as a whole, with scores ranging from 5 to 7 stars. Unfortunately not all studies reported the age characteristics of the female patients recruited, and the test groups in several trials were not matched with the controls on age. Hence, judging by the comparability of groups, the quality of the studies included was relatively low.
Heterogeneity of studies
The results of the test for heterogeneity indicated that there was moderate heterogeneity among all the studies included ( P = 0.027; I 2 = 44.8%).
Overall effects and subgroup analysis
The incidence of DS in females was significantly higher in the test groups than in the control groups in a random-effects model (RR 1.80, 95% CI 1.33–2.43) ( Fig. 2 ). Subgroup analyses showed that the unit assessed, the region in which the study was conducted, and the intervention were not related to the incidence of DS in females taking OC following impacted mandibular third molar extraction ( Table 2 ).
|Stratification group||Number||RR (95% CI)||Heterogeneity test||Publication bias|
|Q||P -value||I 2 (%)||t||P -value|
|CHX/saline rinse||7||1.62 (1.10–2.40)||9.22||0.162||34.9||0.04||0.972|