Systematic reviews and meta-analysis: a structured review of the methodological literature

Convergent and sequential synthesis designs: implications for conducting and reporting systematic reviews of qualitative and quantitative evidence

BACKGROUND

Systematic reviews of qualitative and quantitative evidence can provide a rich understanding of complex phenomena. This type of review is increasingly popular, has been used to provide a landscape of existing knowledge, and addresses the types of questions not usually covered in reviews relying solely on either quantitative or qualitative evidence. Although several typologies of synthesis designs have been developed, none have been tested on a large sample of reviews. The aim of this review of reviews was to identify and develop a typology of synthesis designs and methods that have been used and to propose strategies for synthesizing qualitative and quantitative evidence.

METHODS

A review of systematic reviews combining qualitative and quantitative evidence was performed. Six databases were searched from inception to December 2014. Reviews were included if they were systematic reviews combining qualitative and quantitative evidence. The included reviews were analyzed according to three concepts of synthesis processes: (a) synthesis methods, (b) sequence of data synthesis, and (c) integration of data and synthesis results.

RESULTS

A total of 459 reviews were included. The analysis of this literature highlighted a lack of transparency in reporting how evidence was synthesized and a lack of consistency in the terminology used. Two main types of synthesis designs were identified: convergent and sequential synthesis designs. Within the convergent synthesis design, three subtypes were found: (a) data-based convergent synthesis design, where qualitative and quantitative evidence is analyzed together using the same synthesis method, (b) results-based convergent synthesis design, where qualitative and quantitative evidence is analyzed separately using different synthesis methods and results of both syntheses are integrated during a final synthesis, and (c) parallel-results convergent synthesis design consisting of independent syntheses of qualitative and quantitative evidence and an interpretation of the results in the discussion.

CONCLUSIONS

Performing systematic reviews of qualitative and quantitative evidence is challenging because of the multiple synthesis options. The findings provide guidance on how to combine qualitative and quantitative evidence. Also, recommendations are made to improve the conducting and reporting of this type of review.

Thirty Years of Evidence on the Efficacy of Drug Treatments for Chronic Heart Failure With Reduced Ejection Fraction: A Network Meta-Analysis

BACKGROUND

Treatments that reduce mortality and morbidity in patients with heart failure with reduced ejection fraction, including angiotensin-converting enzyme inhibitors (ACEI), angiotensin receptor blockers (ARB), β-blockers (BB), mineralocorticoid receptor antagonists (MRA), and angiotensin receptor-neprilysin inhibitors (ARNI), have not been studied in a head-to-head fashion. This network meta-analysis aimed to compare the efficacy of these drugs and their combinations regarding all-cause mortality in patients with heart failure with reduced ejection fraction.

METHODS AND RESULTS

A systematic literature review identified 57 randomized controlled trials published between 1987 and 2015, which were compared in terms of study and patient characteristics, baseline risk, outcome definitions, and the observed treatment effects. Despite differences identified in terms of study duration, New York Heart Association class, ejection fraction, and use of background digoxin, a network meta-analysis was considered feasible and all trials were analyzed simultaneously. The random-effects network meta-analysis suggested that the combination of ACEI+BB+MRA was associated with a 56% reduction in mortality versus placebo (hazard ratio 0.44, 95% credible interval 0.26-0.66); ARNI+BB+MRA was associated with the greatest reduction in all-cause mortality versus placebo (hazard ratio 0.37, 95% credible interval 0.19-0.65). A sensitivity analysis that did not account for background therapy suggested that ARNI monotherapy is more efficacious than ACEI or ARB monotherapy.

CONCLUSIONS

The network meta-analysis showed that treatment with ACEI, ARB, BB, MRA, and ARNI and their combinations were better than the treatment with placebo in reducing all-cause mortality, with the exception of ARB monotherapy and ARB plus ACEI. The combination of ARNI+BB+MRA resulted in the greatest mortality reduction.

Symptom Resolution Rates of Posttraumatic versus Nontraumatic Benign Paroxysmal Positional Vertigo

Objective

To determine the rate of symptom resolution in patients with posttraumatic benign paroxysmal positional vertigo (BPPV) and to determine if it differs from resolution rates in patients with BPPV and without a history of head trauma.

Data Sources

Systematic review of the literature was performed using Medline, EMBASE, and Cochrane databases. English and French articles meeting inclusion criteria and published between 1946 and October 2014 were included.

Review Methods

Data were independently extracted from the articles by 2 reviewers using data collection forms developed a priori. Inclusion and exclusion criteria were decided a priori. Studies were included if they reported on at least 1 case of posttraumatic BPPV (t-BPPV), reported on outcomes of all patients with t-BPPV, had a clearly defined inception point, and provided a clear diagnosis of BPPV (defined a priori by reviewers).

Results

A total of 3017 titles, 362 abstracts, and 67 articles were reviewed, from which 16 articles met inclusion criteria and underwent data extraction. There were a total of 207 patients with posttraumatic BPPV identified. Among the 207 patients, 151 (73%) had resolution of symptoms. The T-BPPV patients may have more multi-canal involvement and may require more repositioning maneuvers for resolution compared to patients with nontraumatic BPPV.

Conclusions

Available evidence does not support the notion that symptom resolution rates in patients with posttraumatic BPPV are worse than those with nontraumatic BPPV. However, well-designed studies with adequate cohorts are lacking. Additional well-executed studies are needed to confirm this lack of difference in resolution rates.

Domestic asbestos exposure: a review of epidemiologic and exposure data

Inhalation of asbestos resulting from living with and handling the clothing of workers directly exposed to asbestos has been established as a possible contributor to disease. This review evaluates epidemiologic studies of asbestos-related disease or conditions (mesothelioma, lung cancer, and pleural and interstitial abnormalities) among domestically exposed individuals and exposure studies that provide either direct exposure measurements or surrogate measures of asbestos exposure. A meta-analysis of studies providing relative risk estimates (n = 12) of mesothelioma was performed, resulting in a summary relative risk estimate (SRRE) of 5.02 (95% confidence interval [CI]: 2.48-10.13). This SRRE pertains to persons domestically exposed via workers involved in occupations with a traditionally high risk of disease from exposure to asbestos (i.e., asbestos product manufacturing workers, insulators, shipyard workers, and asbestos miners). The epidemiologic studies also show an elevated risk of interstitial, but more likely pleural, abnormalities (n = 6), though only half accounted for confounding exposures. The studies are limited with regard to lung cancer (n = 2). Several exposure-related studies describe results from airborne samples collected within the home (n = 3), during laundering of contaminated clothing (n = 1) or in controlled exposure simulations (n = 5) of domestic exposures, the latter of which were generally associated with low-level chrysotile-exposed workers. Lung burden studies (n = 6) were also evaluated as a surrogate of exposure. In general, available results for domestic exposures are lower than the workers' exposures. Recent simulations of low-level chrysotile-exposed workers indicate asbestos levels commensurate with background concentrations in those exposed domestically.

Methodology and reporting of systematic reviews and meta-analyses of observational studies in psychiatric epidemiology: systematic review

BACKGROUND

Relatively little is known of the use of systematic review and synthesis methods of non-randomised psychiatric epidemiological studies, which play a vital role in aetiological research, planning and policy-making.

AIMS

To evaluate reviews of psychiatric epidemiological studies of functional mental disorders that employed synthesis methods such as systematic review or meta-analysis, or other forms of quantitative review.

METHOD

We searched the literature to identify appropriate reviews published during the period 1996 to April 2009. Selected reviews were evaluated using published review guidelines.

RESULTS

We found 106 reviews in total, of which 38 (36%) did not mention method of data abstraction from primary studies at all. Many failed to mention study quality, publication bias, bias and confounding. In 73 studies that performed a meta-analysis, 58 (79%) tested for heterogeneity and of these, 47 found significant heterogeneity. Studies that detected heterogeneity made some allowance for this. A major obstacle facing reviewers is the wide variation between primary studies in the use of instruments to measure outcomes and in sampling methods used.

CONCLUSIONS

Many deficiencies found in systematic reviews are potentially remediable, although synthesis of primary study findings in a field characterised by so many sources of heterogeneity will remain challenging.

Natural history of untreated otitis media

OBJECTIVE/HYPOTHESIS

Data from cohort studies and untreated groups in randomized controlled trials can be identified through systematic literature review and synthesized with meta-analysis to estimate natural history of acute otitis media (AOM) and otitis media with effusion (OME).

STUDY DESIGN

Systematic literature review and meta-analysis.

METHOD

Source articles were identified by MEDLINE search through August 2002 plus manual crosschecks of bibliographies and published meta-analyses. Data were abstracted independently by two investigators and combined with random effects meta-analysis to estimate spontaneous resolution, 95% confidence intervals (CI), and heterogeneity. Sensitivity analysis was performed.

RESULTS

Sixty-three articles met inclusion criteria. AOM symptoms improved within 24 hours without antibiotics in 61% of children (95% CI, 50-72%), rising to 80% by 2 to 3 days (95% CI, 69-90%). Suppurative complications were comparable if antibiotics were withheld (0.12%) or provided (0.24%). Children entered recurrent AOM trials with a mean rate of 5.5 or more annual episodes but averaged only 2.8 annual episodes while on placebo (95% CI, 2.2-3.4). No AOM episodes occurred in 41%, and only 17% remained otitis prone (3 or more episodes). OME after untreated AOM had 59% resolution by 1 month (95% CI, 50-68%) and 74% resolution by 3 months (95% CI, 68-80%). OME of unknown duration had 28% spontaneous resolution by 3 months (95%, CI 14-41%), rising to 42% by 6 months (95% CI, 35-49%). In contrast, chronic OME had only 26% resolution by 6 months and 33% resolution by 1 year.

CONCLUSIONS

The natural history of otitis media is very favorable. Combined estimates of spontaneous resolution provide a benchmark against which to judge new or established interventions. The need for surgery in children with recurrent AOM or chronic OME should be balanced against the likelihood of timely spontaneous resolution and the potential risk of learning, language, or other adverse sequelae from persistent middle ear effusion. Further research is needed to identify prognostic factors that can target children unlikely to improve spontaneously for earlier intervention.

Partially hydrolyzed 100% whey protein infant formula and reduced risk of atopic dermatitis: a meta-analysis

OBJECTIVE

A reduced risk of atopic dermatitis (AD) among healthy infants who received 100% whey protein partially hydrolyzed formula (PHF-W) compared with intact protein cow's milk formula (CMF), has been reported in several studies. To validate these observations and estimate the magnitude of this potential association with greater statistical precision, we conducted a meta-analysis of clinical trial and intervention studies.

MATERIALS AND METHODS

A total of 18 articles representing 12 independent study populations met our inclusion criteria.

RESULTS

A statistically significant 44% (summary relative risk estimate [SRRE] = 0.56, 95% confidence interval 0.40-0.77) reduced risk of atopic manifestations, which included AD, was found among infants receiving PHF-W compared with infants receiving CMF. In a subanalysis of 4 studies that reported results specifically for AD and that were considered of superior methodological quality, the incidence of AD was reduced by 55% (SRRE = 0.45, 95% confidence interval 0.30-0.70).

CONCLUSIONS

Regardless of study design, infant population, follow-up time, or study location, individual study findings were consistent because a reduced incidence of AD was reported in all of the reviewed studies. Exclusive breast-feeding should be encouraged as the standard for infant nutrition in the first months of life. For infants who are not exclusively breast-fed, feeding with PHF-W instead of CMF reduces the risk of AD in infants, particularly in infants with a family history of allergy.

Violence and restraint reduction efforts on inpatient psychiatric units

Efforts to reduce aggression and violence and the use of restraint and seclusion have traditionally been through some form of educational program. This paper presents an integrative review of research and quality improvement projects that aimed to reduce aggression/violence or restraint/seclusion through the use of an educational program. Forty-six papers are included in this review. This paper presents summaries and comparisons of the research designs, the content and length of programs, and the outcomes of these programs. From these summaries, trends in relation to design, content, and outcomes are identified, and recommendations for clinicians and researchers are given.

Occupational exposure to rock wool and glass wool and risk of cancers of the lung and the head and neck: a systematic review and meta-analysis

OBJECTIVE

To conduct a review and meta-analysis of risks of cancers of the lung and head and neck (HN) from exposure to rock wool (RW) and glass wool (GW).

METHODS

We performed a systematic review and meta-analysis of risk estimates of lung and HN cancer in epidemiologic studies of workers exposed to man-made vitreous fibers (MMVF), specifically RW and GW.

RESULTS

Sixteen estimates of lung cancer risk yielded a summary relative risk (RR) of 1.21 (95% CI = 1.11 to 1.32, based on 1662 exposed cases). Corresponding RRs were 1.26 (95% CI = 1.10 to 1.44) in studies of production workers (with similar risk for RW and GW workers), 1.06 (95% CI = 0.77 to 1.48) in studies of end users, and 1.18 (95% CI = 0.98 to 1.42) in community-based studies. The summary RR for HN cancer was 1.36 (95% CI = 1.13 to 1.63, 414 exposed cases). With a few exceptions, all studies that assessed the risk of lung or HN cancer according to various indices of MMVF exposure failed to detect a dose-risk relation. There was limited evidence of a confounding effect of tobacco smoking. No clear excess of pleural mesothelioma has been reported in MMVF-exposed workers.

CONCLUSIONS

Despite a small elevation in RR for lung cancer among MMVF production workers, the lack of excess risk among end users, the absence of any dose-risk relation, the likelihood of detection bias, and the potential for residual confounding by smoking and asbestos exposure argue against a carcinogenic effect of MMVF, RW, or GW at this time. Similar conclusions apply to HN cancer risk among workers exposed to MMVF.

Primary middle ear Epstein-Barr virus-related lymphoepithelial carcinoma: case reports and systematic review

OBJECTIVES/HYPOTHESIS

To report two cases of primary lymphoepithelial carcinoma in the middle ear and to determine the optimal treatment for such lesions.

STUDY DESIGN/METHODS

Case reports and a systematic review of the literature.

RESULTS

Primary lymphoepithelial carcinoma in the middle ear is an exceptionally rare condition with only two other cases reported in the literature. There appears to be an association with Epstein-Barr virus infection and in those patients originating from the Guangdong province of China, much as is the case for similar lesions found in the nasopharynx. Piecemeal rather than en bloc excision, in combination with adjuvant radiotherapy, appears to adequately control the disease.

CONCLUSIONS

Primary lymphoepithelial carcinoma of the middle ear is a rare lesion, which when treated has a good prognosis.

A systematic review of pediatric sensorineural hearing loss in congenital syphilis

INTRODUCTION

Congenital syphilis is a known cause of progressive sensorineural hearing loss. The prevalence of syphilitic sensorineural hearing loss (SNHL) in childhood is not clearly defined.

OBJECTIVE

To determine the frequency and characteristics of pediatric SNHL following intrauterine infection with or exposure to Treponema pallidum in order to develop evidence-based guidelines for audiologic monitoring.

DATA SOURCES

Medline (1950-March 2008), EMBASE (1980-March 2008), CINAHL (1982-March 2008), BIOSIS Previews (1969-March 2008), and Cochrane databases. Manual search of references of identified articles and book chapters.

STUDY SELECTION

Articles with an inception cohort of children infected with T. pallidum during pregnancy, positive serological identification of syphilis infection in the antenatal period or pathognomonic clinical signs of congenital syphilis infection, and longitudinal serial audiologic evaluations to identify the prevalence and progression of SNHL.

DATA EXTRACTION

Patient information, maternal and infant serologic status, and audiometric data extracted in an independent fashion. Discrepancies resolved through mutual consensus.

DATA SYNTHESIS

Descriptive statistics.

RESULTS

One prospective cohort study met the inclusion criteria. No cases of SNHL in infants with early congenital syphilis treated with antibiotics in the neonatal period were identified.

CONCLUSIONS

There have been no reports of children with confirmed congenital SNHL secondary to in utero syphilis infection. Newborns with positive syphilis serology should have hearing screening performed at birth and receive treatment with an appropriate course of penicillin therapy. Longitudinal hearing screening is recommended for all pediatric patients with congenital syphilis, as further studies documenting longitudinal audiometric data for patients previously treated either fully or partly for congenital syphilis are required.

A systematic review of neonatal toxoplasmosis exposure and sensorineural hearing loss

INTRODUCTION

The Joint Committee on Infant Hearing 2007 Position Statement includes in utero toxoplasmosis infection as a risk indicator for delayed-onset or progressive sensorineural hearing loss. It is recommended that children with congenital toxoplasmosis infection undergo audiologic monitoring to identify congenital and delayed-onset sensorineural hearing loss.

OBJECTIVE

To determine the prevalence of sensorineural hearing loss and to develop evidence-based guidelines for audiologic monitoring of children born with congenital toxoplasmosis infection.

DATA SOURCES

Systematic search of Medline, EMBASE and Cochrane databases and manual search of references.

STUDY SELECTION

Longitudinal studies reporting an inception cohort identified at birth, with serologic confirmation of toxoplasmosis infection, and long-term serial audiometric evaluation.

DATA EXTRACTION

Independent extraction of patient and audiometric data.

DATA SYNTHESIS

Descriptive statistics.

CONCLUSION

The five studies meeting our inclusion criteria report a prevalence of toxoplasmosis-associated hearing loss from 0% to 26%. Improved treatment regimens for toxoplasmosis may account for this range. Three treatment groups were identified and a subgroup analysis of the compiled data was performed. In children receiving limited or no treatment, the prevalence of toxoplasmosis-associated SNHL was found to be 28%. In children prescribed 12 months of antiparasitic treatment but in whom treatment was not confirmed to have started prior to 2.5 months of age and in whom compliance was not ensured, the prevalence of SNHL was 12%. In children treated with 12 months of antiparasitical therapy initiated prior to 2.5 months of age with serologically-confirmed compliance, the prevalence of SNHL was 0%. Only two longitudinal studies were identified and neither reported any cases of delayed-onset or progressive toxoplasmosis-associated SNHL. Children who have received a 12-month course of antiparasitical therapy initiated prior to 2.5 months with serologically-confirmed compliance should have repeat audiometric evaluation at 24-30 months of age. Children with congenital toxoplasmosis that had no treatment, partial treatment, delayed onset of treatment, or compliance issues should undergo annual audiologic monitoring until able to reliably self-report hearing loss.

Quantitative assessment of red meat or processed meat consumption and kidney cancer

OBJECTIVE

To conduct a quantitative assessment of red meat or processed meat consumption and kidney cancer.

METHODS

We extracted data from 12 case-control studies, three cohort studies, and the Pooling Project of Diet and Cancer publication for which 13 international cohorts were evaluated. Random effects meta-analysis models were used to calculate summary relative risk estimates (SRRE) based on high vs. low intake values. Sensitivity and influence analyses were conducted, including assessments of heterogeneity.

RESULTS

The SRRE for all studies that reported results for red meat (included variables labeled 'red meat' or single red meat items, such as beef, pork, or liver) was 1.12 (95% CI: 0.98-1.29; p-value for heterogeneity=0.015), and the SRRE using only data from prospective cohorts was 1.02 (95% CI: 0.91-1.15) with minimal heterogeneity (p=0.741). Similarly, in a meta-analysis of the five studies that simultaneously adjusted for smoking, BMI, and total energy intake, the SRRE for red meat was 1.02 (95% CI: 0.91-1.15). No significant association was observed in the meta-analysis of processed meat consumption (SRRE=1.07; 95% CI: 0.94-1.23), although a significant association was observed when only data from cohort studies were analyzed (SRRE=1.19; 95% CI: 1.03-1.37).

CONCLUSIONS

Although many of the summary results were positive, all were weak in magnitude, most were not statistically significant, and associations were attenuated among studies that adjusted for key potential confounding factors. In summary, the findings of this meta-analysis are not supportive of an independent relation between red or processed meat intake and kidney cancer.

Psychotropic prescribing in long-term care facilities: impact of medication reviews and educational interventions

The objective of this literature review was to evaluate the evidence pertaining to the impact of medication reviews and/or educational interventions on psychotropic drug use in long-term care facilities. A computerized search was conducted using MEDLINE, Cochrane Central Register of Control Trials, CINAHL, EMBASE, International Pharmaceutical Abstracts and PsycINFO, from January 1980 to April 2007. Controlled studies or randomized controlled studies were included for review. The authors identified 26 studies evaluating the impact of medication reviews and/or educational interventions on psychotropic drug use in long-term care facilities. Eleven studies met the inclusion criteria for this review and the data from six of these studies were included in a meta-analysis. The pooled odds ratio (OR) from five studies on hypnotic prescribing showed a decrease in use postintervention (OR = 0.57, 95% confidence intervals [CI] = 0.41-0.79). The pooled OR from five studies on prevalence of antipsychotic prescribing postintervention was not significant (OR = 0.81, 95% CI = 0.63-1.04). Medication reviews and/or educational interventions are effective at reducing psychotropic drug prescribing. However, research on the benefits of these interventions in reducing psychotropic drug use on total health care costs and resident health outcomes is lacking.

A systematic review of the incidence of sensorineural hearing loss in neonates exposed to Herpes simplex virus (HSV)

OBJECTIVE

To develop evidence-based guidelines for appropriate audiological monitoring of children born following exposure to or infection with Herpes simplex virus (HSV) for development of immediate or delayed-onset of sensorineural hearing loss (SNHL).

DATA SOURCES

A Medline search of the 1966-July 2007 database was supplemented by search of the additional database Embase (1980-July 2007). Manual search was conducted of references of identified papers and book chapters.

STUDY SELECTION

Articles were sought that were longitudinal in design, to include an inception cohort of children infected with (or exposed to) HSV who were entered at a similar point at birth and followed over time with serial audiometry to identify hearing loss if it developed.

DATA EXTRACTION

Patient information and audiometric data extraction from relevant articles was performed independently by all three researchers. Discrepancies were resolved by mutual consensus.

DATA SYNTHESIS

Data was analyzed using descriptive statistics.

RESULTS

Three papers reported five children with SNHL following apparent disseminated HSV-2 infections in which other obvious clinical sequelae of HSV infection and co-morbid conditions were present. Audiometric information is lacking regarding onset and progression. There are no reports of delayed-onset SNHL following perinatal or asymptomatic HSV infection.

CONCLUSIONS

The development of SNHL in children with exposure to HSV occurs rarely. Routine serological screening for HSV infection in otherwise healthy neonates newly diagnosed with SNHL is unjustified. There is insufficient data to define the incidence and natural history of SNHL in children with HSV infections. Carefully designed and conducted studies are needed to address this issue.

A meta-analysis of occupational trichloroethylene exposure and liver cancer

OBJECTIVE

Findings from epidemiologic studies of trichloroethylene (TCE) exposure and liver cancer have been inconsistent. To quantitatively evaluate this association and to examine sources of heterogeneity, we conducted a meta-analysis of occupational studies of TCE exposure and liver/biliary tract cancer.

METHODS

We identified 14 occupational cohort studies of TCE exposed workers and one case-control study that met our inclusion criteria. Nine studies specifically identified TCE as a workplace exposure, and were classified as Group I cohort studies. Subcohorts of workers, identified within eight of these studies as more likely exposed to TCE than the total cohort, were analyzed separately.

RESULTS

The combined liver/biliary cancer summary relative risk estimate (SRRE) for all studies was 1.08 (95% CI 0.91-1.29; heterogeneity (H)-P-value=0.12). For the total study populations in the Group I cohorts, the SRRE was 1.14 (95% CI 0.93-1.39; H-P-value=0.05) and for the subcohorts, the SRRE was 1.30 (95% CI 1.09-1.55). Within this subcohort analysis, the association for the European studies of workers from various industries (SRRE=1.38; based on four studies) was higher than the association for the US studies of aerospace and aircraft workers (SRRE=0.97, based on four studies).

CONCLUSION

Although positive associations were observed for some analyses, results were inconsistent across occupational groups (aerospace/aircraft vs. other industries combined), study location, and incidence versus mortality endpoints. In addition, exposure-response trends were not observed consistently across studies. Interpretation is also limited by the potential impact of uncontrolled confounding by other occupational or lifestyle exposures such as smoking or alcohol consumption. Given these limitations, the currently available epidemiologic data are not sufficient to support a causal relation between occupational TCE exposure and liver/biliary cancer.

A review of treatment interventions in whiplash-associated disorders

In recent years, there has been much debate on the treatment of whiplash-associated disorders (WAD). It is not clear if the treatments commonly employed are effective, and concerns have been raised on the available scientific evidence of many of these treatments. The aim of this study was to review the literature systematically to analyze the evidence basis of many commonly used treatments for patients suffering from WAD, both in the acute and the chronic state. A computer-assisted search of the databases Medline (from 1962 to May 2003), CINAHL (1960-2003), Embase (1976-2003), and Psychinfo (1960-2003) was conducted as well as a check of the reference lists of relevant studies. All randomized controlled trials (RCTs) were retrieved and systematically analyzed with three common instruments of measurement of methodological quality. A qualitative analysis ("best-evidence synthesis") was performed. The methodological quality of 26 RCTs was analyzed. The median quality scores for all three instruments were poor. Based on the degrees of evidence and the practical obstacles, the following treatments can be recommended: Early physical activity in acute WAD, radiofrequency neurotomy, combination of cognitive behavioral therapy with physical therapy interventions, and coordination exercise therapy in chronic WAD. High-quality RCTs are not common in the field of WAD. More research is needed, particularly on the treatment of chronic WAD.

Meta-Analysis

Systematic reviews use explicit and reproducible criteria to assemble, appraise, and combine articles with a minimum of bias. Meta-analysis is a form of systematic review that uses statistical techniques to derive quantitative estimates of the magnitude of treatment effects and their associated precision. Valid meta-analyses address focused questions, use appropriate criteria to select articles, assess the quality and combinability of articles, provide graphic and numeric summaries, consider potential biases, and can be generalized to a meaningful target population. The rate difference, or absolute risk reduction, is the preferred measure of clinical effect size; the reciprocal tells the number needed to treat for one additional favorable outcome. The benefits of meta-analysis over individual trials include greater precision, increased statistical power, and the ability to identify and explore diversity among studies. Threats to validity include heterogeneity, citation bias, publication bias, language bias, and variations in study quality. Because meta-analysis defines rational treatment expectations at a population level, it is an adjunct to, not a substitute for, clinical judgment in the care of individual patients.

Occlusion: Reflections on science and clinical reality

The last 50 years have seen progress in emphasizing scientific evidence as a basis for dental practice, including occlusal therapy. Although a proper understanding of the contributory role of occlusion to temporomandibular disorders should not be minimized, the importance of occlusion in other areas of dental education and practice should not be overlooked. The primary objective of this article is to discuss the nature of this problem as it relates to the duality of science and clinical reality in the evidence-based paradigm, information transfer, quality of evidence, clinical trials, and clinical aspects of occlusion. Some suggested solutions for the problem and thoughts on past and future perspectives of occlusion are expressed in context of inquiry.

The nature and management of labor pain: executive summary

This report describes the background and process for a rigorous project to improve understanding of labor pain and its management, and summarizes the main results and their implications. Labor pain and methods to relieve it are major concerns of childbearing women, with considerable implications for the course, quality, outcome, and cost of intrapartum care. Although these issues affect many women and families and have major consequences for health care systems, both professional and public discourse reveal considerable uncertainty about many questions, including major areas of disagreement. An evidence-based framework, including commissioned papers prepared according to carefully specified scopes and guidelines for systematic review methods, was used to develop more definitive and authoritative answers to many questions in this field. The papers were presented at an invitational symposium jointly sponsored by the Maternity Center Association and the New York Academy of Medicine, were peer-reviewed, and are published in full in this issue of the journal. The results have implications for policy, practice, research, and the education of both health professionals and childbearing women.

The relationship between social support and adolescent mothers' interactions with their infants: a meta-analysis

OBJECTIVE

The purpose of this meta-analysis was to statistically summarize the results of independent quantitative studies regarding the relationship between social support and adolescent mothers' interactions with their infants.

DATA SOURCES

The following sources were selected from 1980 to 1999: citations from bibliographies of previously located articles, dissertations abstracts, the Social Sciences Citation Indexes, and researchers at a regional conference; online sources used were CINHAL, MEDLINE, ERIC, Psych Lit, and Social Work Abstracts.

STUDY SELECTION

Of a potential sample of 31 studies, 14 were excluded because they did not meet the sampling criteria, and 4 were deleted as outliers. The final sample consisted of 13 studies.

DATA EXTRACTION

Each study was coded for methodologic and substantive variables, including quality indicators.

DATA SYNTHESIS

Most of the studies were cross sectional in design and used a variety of measures. Together, the studies included a sample of 823 mothers.

CONCLUSIONS

A medium effect size and a significant relationship between social support of adolescent mothers and their interactions with their infants were established. No difference was found in the relationship when support was provided by the family or through a network.

Developing an evidence base for interdisciplinary learning: a systematic review

AIM OF THE STUDY

The overall aim of the study was to explore the feasibility of introducing interdisciplinary education within undergraduate health professional programmes. This paper reports on the first stage of the study in which a systematic review was conducted to summarize the evidence for interdisciplinary education of undergraduate health professional students.

METHODS

Systematic reviews integrate valid information providing a basis for rational decision making about health care which should be based on empirical and not anecdotal evidence. The accepted principles for systematic reviews were adapted in order to allow integration of the literature to produce recommendations for educational practice and guidelines for future research.

FINDINGS

The literature on interdisciplinary education was found to be diverse, including relatively small amounts of research data and much larger amounts of evaluation literature. Methodological rating schemes were used to test for confounding influences in the research studies. The number of studies found was 141 but only 30 (21%) were included in the analysis because of lack of methodological rigour in the research and poorly developed outcome measures.

CONCLUSIONS

Student health professionals were found to benefit from interdisciplinary education with outcome effects primarily relating to changes in knowledge, skills, attitudes and beliefs. Effects upon professional practice were not discernible and educational and psychological theories were rarely used to guide the development of the educational interventions.

Pilocarpine for radiation-induced xerostomia in head and neck cancer

This article examines the use of pilocarpine hydrochloride for radiation-induced xerostomia in patients with head and neck cancer. Four randomized controlled trials involving 401 patients met the authors' predefined inclusion criteria and were critically appraised. Three studies used topical pilocarpine as a mouthwash. Outcome measurements were both quantitative and qualitative. The studies reported statistically significant differences in favour of pilocarpine-stimulated treatment groups. Subjective improvements in feelings of oral dryness, speaking and chewing were reported in all studies. However, sample sizes varied and in all cases sampling methods were poorly defined. The pilocarpine doses tested produced side effects, but those reported were easily tolerated. The persistent findings of symptomatic improvement following pilocarpine use merit consideration, but there is insufficient evidence from these studies alone to generalize results to the wider population.

Assessment of published reliability studies for cervical spine range-of-motion measurement tools

OBJECTIVE

To assess the reliability of tools to measure cervical spine range of motion in clinical settings and discuss the necessary components for reliability studies.

DATA SOURCES

Database searches included Bandolier, Bath Information and Data Services including Index of Scientific and Technical Proceedings, British Nursing Index, Cumulative Index to Nursing and Allied Health Literature, English National Health Care Database, MEDLINE, Occupational Therapy Index, Physiotherapy Index, and Rehabilitation index for English language articles from 1966. In addition, citations were searched.

STUDY SELECTION

Studies were selected that assessed the tool for intraobserver or interobserver reliability, evaluated it on movements of flexion/extension, lateral flexion, or rotation, and measured range of motion of the whole cervical spine.

DATA EXTRACTION

All papers were read by one nonclinical researcher with a data extraction sheet. A consultant rheumatologist and a physiotherapist were each asked to read a sample of the papers to give a clinical viewpoint.

DATA SYNTHESIS

Evidence for the reliability of measurement tools was assessed qualitatively based on the quality of the study designs, appropriateness of analysis, and strength of the reliability based on reported intraclass correlation coefficients (the most appropriate analysis technique for reliability studies of this nature). Measurement tools were found to have not been fully tested for reliability, particularly in terms of adequate sample size and appropriate analysis techniques. There were also wide variations in the research design, including the protocol for movement, the characteristics of observers and study population, whether warm-ups were allowed, whether the movement was active or passive, and time intervals between repeated measurements.

CONCLUSION

Although a range-of-motion device has shown promise in reliability and has many advocates, its practicality for clinical use is questionable. Further work must be performed on all measurement tools. Researchers need to produce more rigorous studies and consider the issues discussed here.