Meta-analysis: essential for NPIs practice

Meta-analysis: essential for NPIs practice

Meta-analysis, an indispensable tool for understanding the benefits and risks of non-pharmacological interventions.

More than 100 million Europeans are using alternative medicines according to the 2012 CAMBrella survey. The popularity of these medicines is growing as evidenced by bestsellers, magazine covers, television report hearings and Internet consultations. They are accessible everywhere, in hospitals, in city medicine, in pharmacies, on online sales sites, in specialized shops, and even in supermarkets. A diet here. A dietary supplement there. Drinks, plants, herbal teas, diets, yoga programs, Tai Chi, Qi Gong, stress management techniques, hypnosis methods, psychotherapies, massages, manual therapies, games video, health connected objects, ergonomic pillows… The list seems endless. All these solutions seem good for health and well-being, safe for all manufacturers and professionals, and yet…

A need for clarification

Faced with this expansion for multiple reasons (Ninot, 2013), health authorities such as the High Authority of Health in 2011 and the Academy of Medicine in 2013 call for a better evaluation and monitoring of these solutions. The users through the 2012 CAMBrella survey and the funders (health insurance, complementary health, welfare…) are also waiting for it. What are their indications? What are their real health benefits? Are they more effective if they are combined? What are the risks to health? Are there favorable contexts for use? Who is a “good” answering machine? Are they gateways to abuse of any kind, cults or systematic denial of care? Are they complements or alternatives to conventional treatments? Are they essentially prevention solutions? Should they be prescribed by a doctor? The need for reliable information on their safety and efficiency is urgent. To shed light on these practices and better train professionals and patients, what better than clinical research?

Get out of the nebula of alternative medicine

The evaluation of these solutions located between consumer products and biomedical treatments that have been granted authorization by the ANSM in France, EMA in Europe or FDA in the USA can not be based exclusively on studies of mechanisms (the antioxidant effect for example) and processes explaining the relationship between a patient and a health professional (the placebo effect for example). It can not be limited either to a discipline or a school of thought, still less to the nebula called soft medicine. For example, asserting that dietary supplement or art therapy is good for one’s health does not make sense. It is necessary to dissect the ingredients and the techniques, to know the doses and the durations, to know for which health problem these solutions have a real interest and for what purpose, the prevention, the care or the curative treatment. Thus, we should not talk about hypnosis in general but a method of hypnosis precise to stop smoking. We should talk more about psychotherapy in general, but the method of John Kabat-Zinn, the Mindfulness Based Stress Reduction (MBSR), to reduce stress. As described, these methods can be evaluated and compared. They are called non-pharmacological interventions (NPIs or #NPIs). Each has a name, specific objectives on health indicators, a target population, an explanatory theory, a content (duration, frequency, intensity, techniques/components…), qualified professionals and associated scientific publications (Ninot and Carbonnel, 2016).

Pass the NPIs to the Clinical Research Screen

According to evidence-based medicine developed by Sackett and colleagues in 1996, the evaluation methods for demonstrating the efficacy and safety of a health care solution are first and foremost randomized controlled clinical trials, while other interventional studies systematic reviews and meta-analyses. A clinical trial is a pragmatic experimental study that compares the health benefits and risks of a solution in a group of people to one or more other groups called control or placebo. It asserts that an NPI is effective or not to solve a health problem in a given context, and thus to leave out magical thoughts, amalgams, fashion effects and marketing discourses. It makes it possible to specify indications and good practices, as was the case for the medicine some fifty years ago (Bouvenot, 2006). Their number has been increasing in NPIs since the beginning of the century, and their methodological quality has also been initiated by joint action by researchers such as Glasziou et al. (2008) who invite a better description of NPIs in studies, research collectives such as PRISMA or CONSORT and colleges such as CUMIC (Nizard and Kopferschmitt, 2017). Every year, more than 70,000 publications on studies evaluating the effectiveness of NPIs are produced around the world. There are more than 2 million.

More reliable conclusions with systematic reviews

A systematic review compiles the results of all clinical studies that meet predefined eligibility criteria to answer a question as specific as the effectiveness of an NPI for treating a health problem. It uses a rigorous methodology to minimize collection and replay bias in order to provide reliable results from which conclusions can be drawn and decisions made (Dickersin, 1990). A meta-analysis uses complementary statistical techniques to integrate and summarize the results of studies included in a systematic review (Egger et al., 2001, Cucherat and Leizorovicz, 2017, Sterne et al., 2001). The Cochrane Collaboration provides a rigorous and standardized approach to conducting a meta-analysis. By combining data from all relevant studies, meta-analyzes provide more reliable estimates of the effects of care and prevention strategies than those from a single study (Liberati et al., 2009). Health authorities, national agencies and learned societies rely heavily on these meta-syntheses to make their recommendations for use and to justify additional studies if necessary.

The number of meta-analyses on NPIs has been increasing since 2010. There are more than 10,000.

Systematic reviews of NPIs are increasing

Meta-analyzes may focus on features of an NPI such as Carayol et al. (2013) on the physical activity program dose in reducing fatigue during breast cancer treatments. This meta-analysis shows a decrease in fatigue experienced with a physical activity program based on the analysis of 17 studies that included 1380 patients (Carayol et al., 2013). It indicates that the benefit on reducing fatigue is less important if the dose of physical activity is greater than 2 hours per week.

Meta-analyzes may focus on the efficacy of a particular NPIs for treating a health problem such as that of Haller et al. (2017) which evaluates the MBSR in addition to biological breast cancer treatments based on 10 studies that included 1709 patients. Compared to routine care, benefits of the MBSR program are found on anxiety and depression. They persist at 6 months and one year after the treatments. Meta-analyzes report on the efficacy of NPIs in preventing falls in the elderly (Rimland et al., 2016) or the treatment of delirium (Cerveira et al., 2017), apathy of individuals dementia (Theleritis et al., 2018), migraines (Probyn et al., 2017), stereotypies of the autistic child (Zarafshan et al., 2017) and depression (Farah et al., 2016).

They can compare multiple NPIs to treat depression in general practice, such as Holvast’s 11 studies that included 1041 patients testing cognitive behavioral therapy (CBT), a physical activity program, psychotherapy based on problem solving, a behavioral change program, and light therapy (Holvast et al., 2017). The authors conclude that CBT appears to be preferable to other NMIs while encouraging further studies to confirm this result. Other meta-analyzes focus on the treatment of pain (Fleming et al., 2016), attention disorders (Catalá-López et al., 2017) and sleep disorders (Hu et al., 2015).

Meta-analyzes may look at indicators other than health benefits and risks such as economic markers. The English study by Woods et al. (2017) on osteoarthritis of the knee shows that acupuncture is the NPI with the best cost-effectiveness ratio based on 88 randomized controlled trials that included 7507 patients. The study compares different NPIs, acupuncture, orthopedic appliances, heat treatment, insoles, interferential therapy, light therapy, manual therapy, neuromuscular electrical stimulation, pulsed electrical stimulation, electromagnetic fields pulsed, static magnets and transcutaneous electrical nerve stimulation. Other meta-analyzes comparing NPIs include for example low back pain (Andronis et al., 2017) or sleep disorders (Slanger et al., 2016).

The quality of a meta-analysis depends on the completeness of census studies

A systematic review or a meta-analysis avoids a frequent bias of the narrative reviews also called state of the art or review of the literature, the subjectified choice of studies. Indeed, in the absence of defined procedures and criteria for identifying studies, the selection of studies may be debatable or even oriented. An incomplete census can skew the results, either by exaggerating the benefits or minimizing them. For example, one researcher showed that a review of the published literature on cholesterol-lowering treatments cited 5 times more positive studies than negative ones. Thus, “reviews of the literature in the field of treatment evaluation are often similar to simple opinions argued by some well-selected trial results (p.312)” according to Cucherat and Leizorovicz (2007). In the field of NPIs, exhaustive enumeration of clinical trials is particularly difficult. Several reasons for this. Authors of meta-analyzes rely on well-known but not exhaustive biomedical databases such as PubMed in the USA. These bases are struggling to identify all publications on NPIs because of their essentially biological focus, and they have to cope with an exponential acceleration in the number of non-pharmacological tests worldwide, with a diversification of communication vectors. scientific, sometimes “complex” strategies of publication of the researchers, to refusals of the promoters of the studies to publish the results. The collection of relevant publications on NPIs may take 6 months. To solve this problem in the field of NPIs, a meta-search engine was created in 2018 by Montpellier’s CEPS collaborative academic platform, Motrial. The online Motrial system lists relevant interventional studies. It sorts and organizes NPIs clinical trial publications from multiple databases based on relevant criteria. It distinguishes the main publication from the secondary publications of a clinical study. It indicates whether the clinical trial has been declared to an ethics committee, a registration of its protocol to the competent authorities and private and / or public funding. It also specifies the promoter and the country(ies) of the study. In short, it automatically realizes in 6 minutes what a researcher does manually in 6 months.

To conclude

If meta-analyzes are very promising for distinguishing between true and false about the efficacy and safety of NPIs in solving a health problem (Gueguen et al., 2014), the results of studies must be published and accessible to researchers and clinicians. Work remains to be done in the way both in the production of the results of these studies and in their accessibility to the greatest number.


Académie Nationale de Médecine (2013). Thérapies complémentaires: Leur place parmi les ressources de soins. Paris: Académie Nationale de Médecine.

Andronis L, Kinghorn P, Qiao S, Whitehurst DG, Durrell S, McLeod H (2017). Cost-effectiveness of non-invasive and non-pharmacological interventions for low back pain: a systematic literature reviewApplied Health Economic Health Policy, 15(2), 173-201.

Boutron I, Moher D, Altman DG, Schulz KF, Ravaud P (2008). Extending the CONSORT statement to randomized trials of nonpharmacologic treatment: explanation and elaborationAnnals of Internal Medicine, 148(4), 295-309.

CAMbrella. Programme de recherche européen. A pan-European research network for Complementary and Alternative Medicine. Final Report of CAMbrella Work 2012.

Carayol M, Bernard P, Boiché J, Riou F, Mercier B, Cousson-Gélie F, Romain AJ, Delpierre C, Ninot G (2013). Psychological effect of exercise in women with breast cancer receiving adjuvant therapy: what is the optimal dose needed? Annals of Oncology, 24(2):291-300.

Catalá-López F, Hutton B, Núñez-Beltrán A, Page MJ, Ridao M, Macías Saint-Gerons D, Catalá M, Tabarés-Seisdedos R, Moher D (2017). The pharmacological and non-pharmacological treatment of attention deficit hyperactivity disorder in children and adolescents: A systematic review with network meta-analyses of randomised trialsPLoS One; 12(7), e0180355.

Cerveira CCT, Pupo CC, Dos Santos SS, Santos JEM (2017). Delirium in the elderly: A systematic review of pharmacological and non-pharmacological treatmentsDement Neuropsychology, 11(3), 270-275.

Chen LF, Liu J, Zhang J, Lu XQ (2016). Non-pharmacological interventions for caregivers of patients with schizophrenia: A meta-analysisPsychiatry Research, 235, 123-127.

Clarke K, Mayo-Wilson E, Kenny J, Pilling S (2015). Can non-pharmacological interventions prevent relapse in adults who have recovered from depression? A systematic review and meta-analysis of randomised controlled trialsClinical Psychology Review, 39, 58-70.

Cucherat M, Leizorovicz A (2007). La méta-analyse des essais thérapeutiques: Concepts et interprétation des résultatsMédecine thérapeutique, 13(4), 311-316.

Dickersin K (1990). The existence of publication bias and risk factors for its occurrenceJAMA, 263, 1385-1389.

Egger M, Smith GD, Altman DG (2001). Systematic reviews in health care. Meta-analysis in context. BMJ Books.

Farah WH, Alsawas M, Mainou M, Alahdab F, Farah MH, Ahmed AT, Mohamed EA, Almasri J, Gionfriddo MR, Castaneda-Guarderas A, Mohammed K, Wang Z, Asi N, Sawchuk CN, Williams MD, Prokop LJ, Murad MH, LeBlanc A (2016). Non-pharmacological treatment of depression: a systematic review and evidence mapEvidence Based Medicine, 21(6), 214-221.

Fleming PS, Strydom H, Katsaros C, MacDonald L, Curatolo M, Fudalej P, Pandis N (2016). Non-pharmacological interventions for alleviating pain during orthodontic treatmentCochrane Database Systematic Review, 12, CD010263.

Gertler P, Tate RL, Cameron ID (2015). Non-pharmacological interventions for depression in adults and children with traumatic brain injuryCochrane Database Systematic Review, 12, CD009871.

Glasziou P, Meats E, Heneghan C, Shepperd S (2008). What is missing from descriptions of treatment in trials and reviews? British Medical Journal, 336(7659), 1472-1474.

Gueguen J, Hill C, Barry C (2014). Complementary medicines. In Wiley StatsRef: Statistics Reference Online. John Wiley & Sons, Ltd.

Haller H, Winkler MM, Klose P, Dobos G, Kümmel S, Cramer H (2017). Mindfulness-based interventions for women with breast cancer: an updated systematic review and meta-analysisActa Oncology, 56(12), 1665-1676.

HAS (2011). Développement de la prescription de thérapeutiques non médicamenteuses validées. Paris: HAS.

Holvast F, Massoudi B, Oude Voshaar RC, Verhaak PFM (2017). Non-pharmacological treatment for depressed older patients in primary care: A systematic review and meta-analysisPLoS One, 12(9), e0184666.

Hu RF, Jiang XY, Chen J, Zeng Z, Chen XY, Li Y, Huining X, Evans DJ (2015). Non-pharmacological interventions for sleep promotion in the intensive care unitCochrane Database Systematic Review, 10, CD008808.

Khan F, Amatya B, Bensmail D, Yelnik A (2017). Non-pharmacological interventions for spasticity in adults: An overview of systematic reviewsAnnals of Physical Rehabilitation Medicine, S1877-0657(17), 30415-3.

Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JPA, Clarke M, Devereaux PJ, Kleijnen J, Moher D (2009). The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: Explanation and elaborationPLoS Medicine, 6(7), e1000100

Ninot G, Carbonnel F (2016). Pour un modèle consensuel de validation clinique et de surveillance des interventions non médicamenteuses (INM)Hegel, 6(3), 273-279.

Ninot G (2013). Démontrer l’efficacité des interventions non médicamenteuses: Question de points de vue. Montpellier: Presses Universitaires de la Méditerranée.

Nizard J, Kopferschmitt J (2017). Collège Universitaire de Médecine Intégrative et ComplémentaireHegel, 7(4), 327-330.

Probyn K, Bowers H, Mistry D, Caldwell F, Underwood M, Patel S, Sandhu HK, Matharu M, Pincus T (2017). Non-pharmacological self-management for people living with migraine or tension-type headache: a systematic review including analysis of intervention componentsBritish Medical Journal, 7(8), e016670.

Ravnsko U (1992). Cholesterol lowering trials in coronary heart disease: frequency of citation and outcomeBritish Medical Journal, 305, 15-19.

Rimland JM, Abraha I, Dell’Aquila G, Cruz-Jentoft A, Soiza R, Gudmusson A, Petrovic M, O’Mahony D, Todd C, Cherubini A (2016). Effectiveness of non-pharmacological interventions to prevent falls in older people: A systematic overviewPLoS One, 11(8), e0161579.

Sackett DL, Rosenberg WM, Gray JA, Haynes RB, Richardson WS (1996). Evidence based medicine: what it is and what it isn’tBritish Medical Journal, 312(7023), 71-72.

Slanger TE, Gross JV, Pinger A, Morfeld P, Bellinger M, Duhme AL, Reichardt Ortega RA, Costa G, Driscoll TR, Foster RG, Fritschi L, Sallinen M, Liira J, Erren TC (2016). Person-directed, non-pharmacological interventions for sleepiness at work and sleep disturbances caused by shift work. Cochrane Database Systematic Review, (8), CD010641.

Sterne JA, Egger M, Smith GD (2001). Systematic reviews in health care: Investigating and dealing with publication and other biases in meta-analysisBritish Medical Journal, 323(7304), 101-105.

Theleritis C, Siarkos K, Politis AA, Katirtzoglou E, Politis A (2018). A systematic review of non-pharmacological treatments for apathy in dementiaInternational Journal of Geriatric Psychiatry, 33(2), e177-e192.

Treanor CJ, McMenamin UC, O’Neill RF, Cardwell CR, Clarke MJ, Cantwell M, Donnelly M (2016). Non-pharmacological interventions for cognitive impairment due to systemic cancer treatmentCochrane Database Systematic Review,  8, CD011325.

Woods B, Manca A, Weatherly H, Saramago P, Sideris E, Giannopoulou C, Rice S, Corbett M, Vickers A, Bowes M, MacPherson H, Sculpher M (2017). Cost-effectiveness of adjunct non-pharmacological interventions for osteoarthritis of the kneePLoS One, 12(3), e0172749.

Zarafshan H, Salmanian M, Aghamohammadi S, Mohammadi MR, Mostafavi SA (2017). Effectiveness of non-pharmacological interventions on stereotyped and repetitive behaviors of pre-school children with autism: A systematic reviewBasic Clinical Neuroscience, 8(2), 95-103.

To reference this Blog en Sante © article

Ninot G (2018). Meta-analysis: essential for NPIs practice. Blog en Santé, A86.

© Copyright 2018 Grégory Ninot. All rights reserved.

Leave a Reply

Your email address will not be published. Required fields are marked *