
==== Front
Cad Saude Publica
Cad Saude Publica
csp
Cadernos de Saúde Pública
0102-311X
1678-4464
Escola Nacional de Saúde Pública Sergio Arouca, Fundação Oswaldo Cruz

10.1590/0102-311XEN141323
01100
Review
Sense of coherence and substance use in adults: a systematic review and meta-analysis
Senso de coerência e uso de substâncias em adultos: revisão sistemática e metanáliseSentido de coherencia y consumo de sustancias por adultos: revisión sistemática y metanálisishttps://orcid.org/0000-0001-9919-8608
Danigno Júlia Freire contributed to the study design, data collection analysis, and writing approved the final version 1
https://orcid.org/0000-0003-4995-4748
Dias Mariane da Silva contributed to the data collection and review approved the final version 1 2
https://orcid.org/0000-0001-9843-412X
Horta Bernardo Lessa contributed to the writing approved the final version 1
1 Universidade Federal de Pelotas, Pelotas, Brasil.
2 Universidade Federal do Rio Grande, Rio Grande, Brasil.
Correspondence J. F. Danigno Universidade Federal de Pelotas. Rua Marechal Deodoro 1160, Pelotas, RS 96020-220, Brasil. juliadanigno@yahoo.com.br
Additional information: ORCID: Júlia Freire Danigno (0000-0001-9919-8608); Mariane da Silva Dias (0000-0003-4995-4748); Bernardo Lessa Horta (0000-0001-9843-412X).

20 9 2024
2024
40 9 e0014132328 7 2023
13 6 2024
28 6 2024
https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License
This study systematically reviews the evidence on the association between sense of coherence (SOC) and substance use during adulthood. Two researchers conducted independent literature searches on the PubMed, LILACS, PsycINFO and Web of Science databases. Original articles assessing SOC and substance use in adults (age > 19 years) were included. Two reviewers independently assessed studies in two phases - initially by reading the title/abstract, then the full text. Discrepancies were resolved by a third reviewer. Estimates were pooled using random-effects models. Bibliographic search identified 21 studies on the association between SOC and substance use in adults. Studies (n = 11) that assessed the association with tobacco smoking found a 0.92 (95%CI: 0.82; 1.01, very low degree of certainty) odds of smoking among those with a high SOC; the association was not modified by age. Individuals with a strong SOC had lower odds of using alcohol (pooled effect: OR = 0.70, 95%CI: 0.50; 0.90, very low degree of certainty); adjustment for confounding variables decreased the magnitude of the association (pooled OR = 0.89, 95%CI: 0.80; 0.98). This systematic review and meta-analysis suggests that a strong SOC protects against substance use among adults regardless of age, with practical implications for preventive interventions and tailored strategies aimed at high-risk individuals. Longitudinal studies are needed to understand the impact of SOC on substance use. Examining interactions with socioeconomic factors and including diverse populations would enhance generalizability.

O objetivo deste estudo foi fazer uma revisão sistemática das evidências sobre a associação entre senso de coerência (SOC) e uso de substâncias na vida adulta. Dois pesquisadores buscaram, independentemente, as bases de dados PubMed, LILACS, PsycINFO e Web of Science. Foram incluídos artigos originais que avaliaram o SOC e o uso de substâncias em adultos (idade > 19 anos). Dois revisores avaliaram, independentemente, os estudos em duas fases - inicialmente com base no título/resumo e, em seguida, no texto completo. Discrepâncias foram resolvidas por um terceiro revisor. Modelos de efeitos aleatórios foram usados para agrupar as estimativas. Identificamos 21 estudos que investigaram a associação entre SOC e uso de substâncias em adultos. Entre os 11 estudos que avaliaram a associação com o tabagismo, a chance de fumar foi de 0,92 (IC95%: 0,82; 1,01, grau de confiança muito baixo) entre aqueles com alto SOC, e a associação não foi modificada pela idade. Indivíduos com forte SOC tiveram menores chances de usar álcool (efeito combinado: OR = 0,70, IC95%: 0,50; 0,90, grau de confiança muito baixo), e o ajuste para confusão diminuiu a magnitude da associação (OR combinado = 0,89, IC95%: 0.80; 0.98). Esta revisão sistemática e metanálise sugere que um forte SOC protege contra o uso de substâncias entre adultos, independentemente da idade, com implicações práticas para intervenções preventivas e estratégias personalizadas, especialmente para indivíduos de alto risco. Estudos longitudinais são necessários para entender o impacto do SOC no uso de substâncias. Examinar as interações com fatores socioeconômicos e incluir diversas populações aumentaria a generalização.

Este estudio realiza una revisión sistemática sobre la asociación entre sentido de coherencia (SOC) y uso de sustancias en la edad adulta. Dos investigadores realizaron búsquedas de forma independiente en las bases de datos PubMed, LILACS, PsycINFO y Web of Science. Se incluyeron artículos originales que evaluaron el SOC y el uso de sustancias por adultos (edad > 19 años). Dos evaluadores examinaron de forma independiente los estudios en dos etapas: primero, el título/resumen; y, después, el texto completo. Las discrepancias fueron resueltas por un tercer evaluador. Se utilizaron modelos de efectos aleatorios para agrupar las estimaciones. Se identificaron 21 estudios sobre la asociación entre SOC y uso de sustancias por adultos. De 11 estudios que evaluaron la asociación con el tabaquismo, la propensión al tabaquismo fue de 0,92 (IC95%: 0,82; 1,01, grado de confianza muy bajo) entre las personas con alto SOC; y la asociación no sufrió cambios según la edad. Las personas con alto SOC tenían menor propensión a consumir alcohol (efecto combinado: OR = 0,70, IC95%: 0,50; 0,90, grado de confianza muy bajo), y el ajuste por confusión disminuyó la intensidad de la asociación (OR combinado = 0,89, IC95%: 0,80; 0,98). Esta revisión sistemática y metanálisis muestra que un alto SOC protege contra el uso de sustancias por adultos, independientemente de la edad, con implicaciones prácticas para las intervenciones de prevención y las estrategias personalizadas, especialmente para las personas con alto riesgo. Se necesitan estudios longitudinales para comprender el impacto del SOC en el uso de sustancias. Analizar las interacciones con los factores socioeconómicos e incluir otras poblaciones podría aumentar la generalización.

Keywords:

Sense of Coherence
Substance Use
Smoking
Illicit Drug
Alcohol Drinking
Palavras-chave:

Senso de Coerência
Uso de Substâncias
Fumar
Droga Ilícita
Consumo de Bebidas Alcoólicas
Palabras-clave:

Sentido de Coherencia
Uso de Sustancias
Fumar
Droga Ilícita
Consumo de Bebidas Alcohólicas
==== Body
pmcIntroduction

Sense of coherence (SOC) is a concept introduced by Aaron Antonovsky’s salutogenesis framework. It aims to understand the contributing factors in the development of health and to explain how individuals can manage their lives despite adverse conditions 1 , 2 . SOC comprises three components: comprehensibility (ability to understand an event), manageability (perceived potential to manipulate or resolve the event), and meaningfulness (significance attributed to this event) 2 .

A strong SOC empowers individuals to mobilize internal and external resources to effectively cope with stressors and manage tension, thereby promoting and maintaining their health 1 . Individuals with a strong SOC would be more efficient in creating coping mechanisms and strategies to maintain health in unfavorable situations 2 . Urakawa & Yokoyama 3 observed that SOC is negatively associated with stress levels and positively correlated with the ability to cope with stress. This reduction in stress would positively influence health-related behaviors, contributing to maintain a positive health status.

Evidence suggests that SOC is associated with various health-related behaviors like tobacco smoking 4 , alcohol intake 5 , and illicit drug use 6 . These behaviors are associated with the development of noncommunicable diseases (NCDs) such as cardiovascular diseases, cancer, diabetes, and chronic respiratory conditions, which account for two-thirds of the overall burden of disease in low- and middle-income countries 7 , 8 . In 2019, smoking resulted in 8.71 million attributable deaths (15.4% of all deaths), alcohol use led to 2.07 million attributable deaths among men and 0.40 million among women, and drug use contributed to 0.45 million attributable deaths 8 . SOC would also have an impact in the development of NCDs through health-related behaviors.

It is crucial to recognize and critically assess the methodological limitations within research. Many studies assessed the association between SOC and health-related behaviors without controlling for potential confounding factors. Failure to address known confounders can overestimate the association. For example, low socioeconomic status is associated with lower SOC scores and less favorable behavioral habits. Consequently, analyses that fail to control for confounding by socioeconomic status are susceptible to residual confounding which would overestimate the magnitude of the association.

Currently, only the study by da-Silva-Domingues et al. 9 has reviewed the relation between SOC and substance use, but as part of a broader analysis that focused on health behaviors such as eating habits, time spent on computers, rest periods, as well as smoking, alcohol consumption, and oral health care. However, it specifically evaluates the association between SOC and substance use (tobacco, alcohol, and illicit drugs). It is crucial to recognize how SOC influences both general health behaviors and specifically risky behaviors like substance use, given their significant impact on public health and the development of chronic diseases. While da-Silva-Domingues et al.’s study 9 assessed individuals aged 12 to 30 years, our research seeks to bridge a gap by concentrating on the adult population as defined by the World Health Organization (WHO). By identifying heterogeneity sources and conducting a meta-analysis, the present study intends to offer insights beyond the existing literature which can guide future research and interventions aimed at reducing harmful substance use among adults.

In short, this study reviewed the literature on the association between substance use (tobacco, alcohol, and illicit drugs) and SOC, while exploring sources of heterogeneity.

Methods

This systematic review was registered on PROSPERO (protocol n. CRD42023402776) and conducted according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and AMSTAR2 (A MeaSurement Tool to Assess systematic Reviews) guidelines (Supplementary Material - Boxes 1 and 2; https://cadernos.ensp.fiocruz.br/static//arquivo/suppl-e00141323_3035.pdf). We formulated the following research question: “What is the association between sense of coherence and substance use in adults?” (P = adults; I = high SOC; C = low SOC; O = substance use outcomes).

Search protocol and selection criteria

Bibliographic search was conducted in March 2023 on the PubMed, LILACS, Web of Science, and PsycINFO databases. Search strategies combined the terms for SOC (“sense of coherence” OR salutogen* OR “general resistance resources”) with the following terms for each of the studied outcomes:

Smoking, cigarette smoking, tobacco, and tobacco use disorder;

Alcohol, alcoholism, and alcohol drinking;

Substance use, substance-related disorders, and substance abuse;

Illicit drugs, cocaine, crack, cannabis, amphetamine, and narcotic.

Supplementary Material - Table S1 (https://cadernos.ensp.fiocruz.br/static//arquivo/suppl-e00141323_3035.pdf) shows the search strategy and the number of studies identified.

Inclusion criteria consisted of original articles that evaluated the association of SOC with at least one measure of substance use among adult participants (age > 19 years). We set no restrictions on language, publication date, or the SOC measurement scale used. Papers involving animals, research protocols, editorials, comments, and those with insufficient data were excluded.

Additionally, we searched for grey literature in the Google Scholar, CAPES Portal of Theses and Dissertations, and ProQuest databases.

Data extraction (selection and coding)

Study selection was performed in two phases. First, two independent reviewers evaluated the title and abstract of each identified study. Articles considered as possibly eligible for inclusion in the review were retrieved and read in full. Discrepancies between the reviewers regarding the inclusion or exclusion of a paper were resolved by a third reviewer.

Last name of the first author, year of publication, country where the study was conducted, sample size, age and gender of the studied population, study design, scale used for substance use measurement, categorization of substance use, scale used for SOC assessment, SOC categorization, control for confounding factors, and effect measurement along with its 95% confidence interval (95%CI) were extracted. Two independent reviewers extracted the data using a Microsoft Excel spreadsheet (https://products.office.com/). Any discrepancies between reviewers were resolved through consensus or consultation with a third reviewer. Subsequently, the extracted data were transferred to Stata software (https://www.stata.com).

Methodological quality of the included studies was evaluated using the Risk of Bias in Non-randomized Studies of Exposures scale (ROBINS-e) 10 , recommended by the Cochrane Collaboration for assessing effectiveness and safety in nonrandomized intervention trials. This instrument has seven domains of bias categorized by the timing of occurrence: pre-intervention (confounding and selection bias in participant enrollment), at intervention (bias in classifying interventions), and post-intervention (deviations from intended interventions, missing data bias, outcome measurement bias, and bias in selection of reported results). Assessment classified items as low, moderate, severe, or critical risk of bias, or as having no information, following the Cochrane Handbook for Systematic Reviews of Interventions 11 .

Data synthesis and analysis plan

For inclusion in the meta-analysis, the studies had to report a crude or adjusted measure of the association between SOC and substance use. We considered the following reported measures: mean values of SOC for both users and non-users, linear regression coefficients, odds ratios (OR), and prevalence ratios of substance use in different SOC categories along with their corresponding 95%CI or standard errors.

Pooled measure of association was calculated using a random effects model. We conducted two distinct meta-analyses - one focused on substance use outcomes; the other focused on the mean SOC - using Stata. The analyses were stratified according to the type of substance, follow-up rate, sample size, age group, confounding variables (socioeconomic status), study setting (America, Asia, Europe), SOC categorization, smoking and alcohol.

Heterogeneity between studies was assessed using the Q-test and I-square. Estimates were pooled using a random-effects model. Publication bias was assessed using a funnel plot and Egger’s test. We also stratified the analysis according to sample size to further evaluate the impact of publication bias on the estimates.

Certainty of evidence was evaluated using the GRADEpro software (https://www.gradepro.org/). The GRADE (Grading of Recommendations Assessment, Development, and Evaluation) system categorizes evidence quality into four levels - high, moderate, low, and very low - based on considerations such as study design limitations, indirect evidence, inconsistency of results, imprecision of results, and probability of publication bias.

Results

Characteristics of the included studies

Database search identified a total of 566 records (Figure 1). After removing duplicates (n = 206 articles), 360 titles and abstracts were read resulting in 54 studies selected for full text review. In the end, the review sample included 21 articles. Exclusion criteria consisted of articles with outcomes and exposure that were not of interest (n = 14), user sample (n = 6), and lack of data on the association between SOC and substance use (n = 8). Supplementary Material - Box S3 (https://cadernos.ensp.fiocruz.br/static//arquivo/suppl-e00141323_3035.pdf) shows the references excluded after full text review and the respective reason. An additional manual search was performed on the references of the 21 selected articles, but we identified no additional study.

Figure 1 Flowchart of study selection.

Characterization and qualitative synthesis of the selected studies

Table 1 presents the main characteristics of the included studies. Most were cross-sectional (n = 14) and longitudinal (n = 6), conducted in European countries (n = 17) and published from 1987 to 2022. Sample size varied considerably, ranging from 120 to 40,674 participants. Participant age also exhibited a wide range, from 15 to 88 years. SOC was assessed by different instruments, namely: SOC-L9 (n = 2), SOC-7 (n = 1), SOC-3 (n = 3), SOC-29 (n = 7), and SOC-12 (n = 2).

Table 1 Characteristics of the selected studies: odds ratios (OR) for strong sense of coherence (SOC) and their effect on substance use (smoking, alcohol, illicit drugs) in 14 studies and mean difference in SOC regarding substance use (smoking, alcohol) in seven studies.

Study/Country (Year)	Study design	Sample/Follow-up rate (%)	Mean age (in years)	SOC	Categorization of SOC	Substance use	Scale of substance use	Categorization of substance use	Effect measures (95%CI) *	Control for confounding	
Neuner et al. 16 Germany (2006)	Prospective	2,056 30.2%	34 ± 12	SOC-L9	Quartiles: strongest vs. weakest	Alcohol	AUDIT score	Hazardous alcohol consumption (yes/no). Yes, in men: AUDIT 8-40 points; in women: AUDIT 5-40 points)	OR = 0.51 (0.36; 0.72)	Age, gender, income, education, additional substance use	
Smoking	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers	OR = 0.59 (0.45; 0.78)	
Drugs	Structured self-designed questionnaire	No/Yes. Yes: use of illicit drugs at least 1 to 3 times within the last 12 months	OR = 0.46 (0.33; 0.66)	
Antonovsky et al. 17 Israel (1987)	Cross-sectional	120 (male) 83.64%	41.1	SOC-7	Dichotomous: strong vs. weak	Alcohol	Structured self-designed questionnaire	Drinkers/Non-drinkers. Drinkers: drink once to several times per day	OR = 0.46 (0.18; 1.18)	No	
Tobamidanik & Zabkiewicz 18 United States (2009)	Cross-sectional	4,630	62.4 ± 19	SOC-3	Terciles: strongest vs. weakest	Alcohol	Alcohol dependence - DSM-IV	Drinkers/Non-drinkers. Drinkers: 5 or more drinks consumed at least once per week	OR = 0.30 (0.24; 0.37)	Age, gender, income, education, additional substance use	
Wainwright et al. 19 United Kingdom (2007)	Cross-sectional	18,287 87.40%	41.0-80.0	SOC-3	Dichotomous: strong vs. weak	Smoking	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers	OR = 0.74 (0.63; 0.86)	Age, gender, income, education	
Morita et al. 5 Japan (2014)	Cross-sectional	167 90%	41.9 ± 9.8	SOC-29	Terciles: strongest vs. weakest	Smoking Alcohol	Lifestyle-related questionnaire	Smokers (current)/ex and non-smokers	OR = 1.19 (0.43; 3.34)	Age, gender, income, social support	
Lifestyle-related questionnaire	Drinkers/Non-drinkers. Drinkers: drinking more than 1 gou ** per day	OR = 0.78 (0.33; 1.85)		
Savolainen et al. 20 Finland (2009)	Cross-sectional	8,028 88%	41.8 ± 10.6	SOC-12	Quintiles: strongest vs. weakest	Smoking	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers. Smokers: regular or occasional smokers	OR = 0.75 (0.61; 0.91)	Age, gender, education	
Silarova et al. 4 Slovakia (2014)	Cross-sectional	179 60.1%	58.32 ± 6.54	SOC-13	Dichotomous: strong vs. weak	Alcohol	European Health and Behaviour Survey	Drinkers/Non-drinkers. Drinkers: occasional and regular	OR = 1.01 (0.97; 1.05)	Age, gender, income	
Smoking	European Health and Behaviour Survey		OR = 1.06 (1.00; 1.13)		
van Loon et al. 21 The Netherlands (2001)	Longitudinal	1,431 women 54%	42.6 ± 10.9	SOC-3	Dichotomous: strong vs. weak	Smoking	The Health and Life Experiences Questionary	Smokers (current) and ex/never	OR = 1.16 (0.98; 1.36)	Age	
1,083 men	OR = 1.05 (0.88; 1.25)	
Saade & Marchand 22 Canada (2013)	Longitudinal	7,338 81%	43.82 ± 10.16	SOC-13	Numerical	Alcohol	Canadian norms	Alcohol misuse. Yes: man drinks more than 14 drinks per week or when a woman drinks more than 9 drinks per week	OR = 0.99 (0.98; 1.00)	Age, gender, income, education, marital status, social support	
Von Ah et al. 23 United States (2005)	Cross-sectional	161 40%	19.6 ± 4.09	SOC-29	Numerical	Smoking	4-item tobacco self-efficacy questionnaire	Smokers (current)/ex and non-smokers. Current smoker: an individual who smoked a whole cigarette within the last 30 days	OR = 1.01 (0.97; 1.04)	No	
Ristkari et al. 6 Finland (2005)	Longitudinal	2,314 78.7%	67 ± 11	SOC-13	Quartiles: strongest vs. weakest	Smoking	Structured self-designed questionnaire	Smokers (current) and ex/never. Smokers: smoked during the last 6 months	OR = 0.99 (0.78; 1.26)	No	
Alcohol	Structured self-designed questionnaire	Drinker/Non-drinkers. Drunkers: drunk during the last 6 months	OR = 0.69 (0.49-0.87)		
Drugs	Structured self-designed questionnaire	Yes/No. Yes: drugs during the last 6 months	OR = 0.18 (0.10; 0.33)		
Larm et al. 24 Sweden (2016)	Cross-sectional	40,674 59.2%	53.8 ± 17.9	SOC-13	Terciles: strongest vs. weakest	Alcohol	AUDIT-C	Hazardous alcohol consumption. Yes, in men: AUDIT 8-40 points; in women: AUDIT 6-40 points)	OR = 0.48 (0.36; 0.64)	Yes	
Thomas et al. 25 Sweden (2020)	Cross-sectional	1,007 62.5%	57 ± 7.2	SOC-29	Numerical	Alcohol	Structured self-designed questionnaire	Hazardous alcohol consumption. Drinking more than 9 standard glasses per week for women and more than 14 glasses per week for men, and/or reporting drinking 4 or more standard glasses for women and 5 or more glasses for men on a typical day when drinking	OR = 0.79 (0.64; 0.96)	Age, sex, education, income	
Smoking	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers	OR = 0.84 (0.71; 1.00)		
Poppius et al. 26 Finland (1999)	Longitudinal	4,405 73%	40-55	SOC-29	Terciles: strongest vs. weakest	Smoking	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers	OR = 0.96 (0.82; 1.12)	Age	
Gajdosova et al. 27 Slovakia (2009)	Cross-sectional	830 94.1%	20.5 ± 1.4	SOC-13	Numerical	Smoking	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers	β: -0.61 (-2.06; 0.84)	No	
Ahlstrand et al. 28 Sweden (2022)	Cross-sectional	851 37.3%	28	SOC-13	Numerical	Smoking	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers. No: no or rarely	β: -3.33 (-5.63; -1.03)	Gender, additional substance use	
Alcohol	Structured self-designed questionnaire	Smokers (current)/ex and non-smokers. Current: > once per month	β -1.76 (-3.37-0.15)		
Luszczynska 29 Poland (2002)	Cross-sectional	83 women 100%	35.6 ± 9.0	SOC-29	Terciles: strongest vs. weakest	Smoking	Structured self-designed questionnaire	Cigarettes daily	β: -4.50 (-10.87; 1.86)	No	
Vilela & Alisson 30 Canada (2010)	Cross-sectional	162	25-88	SOC-13	Numerical	Smoking	Structured self-designed questionnaire	Smokers (current) and ex/non-Smokers	β: -4.40 (-13.73; 4.91)	Age, education, marital status	
Alcohol	Structured self-designed questionnaire	Smokers (current) and ex/non-smokers	β: 2.30 (-10.47; 15.06)		
Igna et al. 31 Sweden (2008)	Cross-sectional	841	59.4 ± 8.1	SOC-13	Numerical	Smoking	Structured self-designed questionnaire	Smokers (current) and ex/non-smokers	β: -3.20 (-4.79; -1.61)		
Verešová & Gatial 32 Romania (2010)	Cross-sectional	158	19-25	SOC-29	Numerical	Alcohol	No information	Smokers (current)/ex and non-smokers	β: -38.0 (-47.8; -28.8)	No	
Kouvonen et al. 33 Finland (2008)	Longitudinal	313 5.7%	36-82	SOC-13	Terciles: strongest vs. weakest	Alcohol	Structured self-designed questionnaire	Hazardous alcohol consumption. Excessive drinking leading intoxication twice or more per month vs. less than twice per month	β: -4.24 (-4.27; -4.21)	Age, education, marital status, additional substance use	
95%CI: 95% confidence interval; AUDIT: Alcohol Use Disorders Identification Test; DSM-IV: Diagnostic and Statistical Manual of Mental Disorders - 4th edition.

* OR: odds ratios for sense of coherence (strong) and their effect on substance use (smoking, alcohol, illicit drugs), β: mean difference in sense of coherence regarding substance use (smoking, alcohol);

** Traditional Japanese unit of alcohol beverage.

Risk of bias assessment

Quality assessment of the risk of bias was predominantly moderate (n = 10), eight studies presented critical risk and four low risk of bias (Box 1). Missing data was the main criteria contributing to moderate or critical risk of bias.

Table 2 summarizes the certainty of the evidence for the outcomes included in the meta-analysis.

Box 1 Risk of bias assessment in nonrandomized studies (ROBINS-e - Risk of Bias in Non-randomized Studies of Exposures scale).

STUDY	BIAS DUE TO CONFOUNDING	BIAS IN SELECTION OF PARTICIPANTS INTO THE STUDY	BIAS IN CLASSIFICATION OF INTERVENTIONS	BIAS DUE TO DEVIATIONS FROM INTENDED INTERVENTIONS	BIAS DUE TO MISSING DATA	BIAS IN MEASUREMENT OF OUTCOMES	BIAS IN SELECTION OF THE REPORTED RESULT	OVERALL BIAS	
Antonovsky et al. 17	Serious	Serious	Low	Moderate	Moderate	Moderate	Low	Serious	
Ahlstrand et al. 28	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate	
Gajdosova et al. 27	Serious	Serious	Low	Low	Low	Low	Low	Serious	
Igna et al. 31	Serious	Low	Low	Low	NI	Low	Low	Serious	
Kouvonen et al. 33	Low	Low	Low	Low	Moderate	Low	Low	Moderate	
Larm et al. 24	Low	Low	Low	Low	Low	Low	Low	Low	
Luszczynska 29	Serious	Serious	Low	Low	Low	Low	Moderate	Serious	
Tobamidanik & Zabkiewicz 18	Low	Low	Low	Low	NI	Low	Low	Low	
Morita et al. 5	Low	Moderate	Low	Low	Low	Low	Low	Moderate	
Neuner et al. 16	Low	Moderate	Low	Low	Moderate	Low	Low	Moderate	
Poppius et al. 26	Moderate	Low	Low	Low	Low	Low	Low	Moderate	
Ristkari et al. 6	Serious	Low	Low	Low	Moderate	Low	Low	Serious	
Saade & Marchand 22	Low	Low	Low	Low	Low	Low	Low	Low	
Savolainen et al. 20	Moderate	Low	Low	Low	Low	Low	Low	Low	
Silarova et al. 4	Moderate	Moderate	Low	Low	Moderate	Low	Low	Moderate	
Thomas et al. 25	Low	Low	Low	Low	Moderate	Low	Low	Moderate	
Verešová & Gatial 32	Serious	Serious	NI	Moderate	NI	Moderate	Moderate	Serious	
Vilela & Alisson 30	Low	Moderate	Low	Low	NI	Low	Low	Moderate	
Von Ah et al. 23	Serious	Moderate	Low	Moderate	Serious	Low	Low	Serious	
van Loon et al. 21	Moderate	Low	Low	Low	Moderate	Low	Low	Moderate	
Wainwright et al. 19	Low	Low	Low	Low	Low	Low	Low	Low	
NI: not informed.

Table 2 Certainty of the evidence of the outcomes included in the meta-analysis.

Participants (studies)	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias	Overall certainty of evidence	Relative effect (95% CI)	
Alcohol								
53,855 (8 nonrandomized studies)	Serious	Serious *	Not serious	Not serious	None	⨁◯◯◯ Very low	OR = 0.7 (0.5; 0.9)	
Smoking								
44,494 (11 nonrandomized studies)	Serious	Serious *	Not serious	Not serious	None	⨁◯◯◯ Very low	OR = 0.92 (0.82; 1.01)	
Illicit drugs								
4,370 (2 nonrandomized studies)	Serious	Not serious	Not serious	Serious	Publication bias strongly suspected; strong association **	⨁◯◯◯ Very low	OR = 0.31 (0.04; 0.59)	
95%CI: 95% confidence interval; OR: odds ratio.

* Inconsistency assessment was based on the dissimilarity of effect estimates;

** Only two studies were included.

OR for substance use

Figure 2 summarizes the results from the 21 studies evaluating the association between SOC and substance use. Individuals with high SOC had a 22% lower odds of using any substance compared with individuals with low SOC (pooled effect: OR = 0.78, 95%CI: 0.68; 0.88, very low degree of certainty). When stratified by substance type, we observed that a strong SOC slightly reduced the odds of smoking (n = 11), but the confidence interval included the reference (pooled effect: OR = 0.92, 95%CI: 0.82; 1.01). Regarding alcohol use (n = 8), individuals with a strong SOC had 30% lower odds of using it (pooled effect: OR = 0.70, 95%CI: 0.50; 0.90, very low degree of certainty). Only two studies analyzed the association with use of illicit drugs, showing a pooled OR of 0.31 (95%CI: 0.04; 0.59).

Figure 2 Odds ratios (OR) for sense of coherence (strong) and their effect on substance use type (smoking, alcohol, illicit drugs): 21 studies.

95%CI: 95% confidence interval; ES: effect size. Note: “D+L” refers to the DerSimonian and Laird method for estimating the aggregate mean effect. “I-V” is an Inverse-Variance approach.

Table 3 presents the subgroup analyses according to characteristics of the analyzed studies. Regarding smoking, studies that evaluated young adults showed a similar association (pooled effect: OR = 0.91, 95%CI: 0.74; 1.09) to those that evaluated older individuals (pooled effect: OR = 0.99, 95%CI: 0.75; 1.23). As for alcohol use, studies that adjusted their estimates for confounding variables reported a weaker association (pooled OR = 0.89, 95%CI: 0.80; 0.98) than those reporting crude estimates (pooled OR = 0.51, 95%CI: 0.22; 0.80). The pooled OR remained similar across studies that evaluated different age groups and continents.

Egger’s test results indicated a trend of small-study effects or publication bias, albeit not statistically significant (p = 0.07). However, we must consider that this analysis is underpowered.

Table 3 Sense of coherence (strong) and odds ratios (OR) of substance use: random-effects meta-analyses by subgroup (n = 19).

Subgroup analysis	Smoking	Alcohol	
Number of estimates	Pooled OR (95%CI)	I-squared (%)	p-value *	Number of estimates	Pooled OR (95%CI)	I-squared (%)	p-value *	
Follow-up rate (%)				0.125				0.190	
< 50	2	0.81 (0.40; 1.22)	95.8		1	0.51 (0.33; 0.69)	0.0		
50-70	4	1.02 (0.90; 1.14)	67.4		3	0.70 (0.17; 1.23)	99.4		
> 70	6	0.84 (0.72; 0.96)	51.3		4	0.78 (0.52; 1.04)	78.8		
Sample size				0.016				0.599	
< 1,000	3	1.02 (0.99; 1.05)	0.0		3	0.82 (0.44; 1.19)	59.7		
1,000-1,999	1	0.84 (0.70; 0.98)	0.0		1	0.79 (0.63; 0.95)	0.0		
≥ 2,000	7	0.88 (0.74; 1.03)	81.7		4	0.62 (0.17; 1.07)	99.3		
Participants age				0.605				0.259	
Young	2	0.81 (0.40; 1.22)	95.8		1	0.51 (0.33; 0.69)	0.0		
Adult	6	0.97 (0.83; 1.11)	76.9		6	0.74 (0.51; 0.97)	98.9		
Older adult	1	0.99 (0.75; 1.23)	0.0		1	0.69 (0.50; 0.88)	0.0		
Study design				0.660				0.736	
Cross-sectional	6	0.90 (0.78; 1.01)	84.3		5	0.67 (0.24; 1.10)	98.8		
Longitudinal	5	0.95 (0.75; 1.15)	83.3		3	0.74 (0.41; 1.01)	94.5		
Adjustment for confounding				< 0.001				< 0.001	
No	5	0.94 (0.82; 1.07)	79.9		3	0.48 (0.17; 0.79)	86.4		
Age	3	0.91 (0.69; 1.13)	82.6		1	0.79 (0.63; 0.95)	0.0		
Age + gender + 1 socioeconomic status variable	1	1.06 (1.00; 1.12)	84.3		1	1.01 (0.97; 1.05)	0.0		
Age + gender + additional substance use	2	0.60 (0.43; 0.76)	0.0		3	0.76 (0.35; 1.17)	92.7		
Continent				0.228				< 0.001	
America	1	1.01 (0.98; 1.04)	0.0		5	0.66 (0.28; 1.04)	98.8		
Europe	9	0.90 (0.78; 1.02)	85.9		1	0.99 (0.98; 1.00)	0.0		
Asia	1	1.19 (0.26; 2.64)	0.0		2	0.56 (0.14; 0.97)	0.0		
Categorization of sense of coherence				0.187				0.032	
Numerical	2	0.94 (0.78; 1.10)			2	0.91 (0.71; 1.10)	83.3		
Dichotomous	4	1.00 (0.81; 1.18)			2	0.79 (0.27; 1.32)	78.3		
Terciles	2	0.96 (0.81; 1.11)			2	0.60 (0.42; 0.77)	34.3		
Quartiles	2	0.78 (0.39; 1.17)			2	0.91 (0.71; 1.10)	45.0		
Quintiles	1	0.75 (0.60; 0.90)							
Categorization of smoking				0.012					
Smokers (current)/ex and non-smokers	6	0.81 (0.62; 0.99)	89.7						
Smokers (current) and ex/never	5	1.01 (0.98; 1.05)	0.0						
Categorization of alcohol								0.002	
Drinkers (current)/ex and non-drinker					4	0.79 (0.51; 1.06)	80.2		
Hazardous alcohol consumption					3	0.53 (0.22; 0.84)	93.9		
Alcohol misuse					1	0.99 (0.98; 1.00)	0.0		
Total	11	0.92 (0.82; 1.01)			8	0.70 (0.50; 0.90)			
95%CI: 95% confidence interval.

* p-value - test of group differences.

Mean SOC

Seven studies assessed the mean SOC among substance users and non-users assessed using the SOC-13 scale. The pooled mean difference was -3.50 points on the SOC scale (95%CI: -5.47; -1.53) (total score can range from 13 to 65 points) for substance users compared with non-users (Figure 3).

Figure 3 Mean difference in sense of coherence regarding substance use (smoking, alcohol): nine studies.

95%CI: 95% confidence interval; ES: effect size. Note: “D+L” refers to the DerSimonian and Laird method for estimating the aggregate mean effect. “I-V” is an Inverse-Variance approach.

Other measures - regression coefficients

One study estimated the beta coefficient, indicating that a 0.1-point decrease on the SOC scale corresponds to a 1-point increase on the Alcohol Problem Index (a Stern index scale).

Discussion

This systematic review identified 21 studies on the association between SOC and substance use in adults. Our findings indicate a negative association between SOC and substance use, suggesting that individuals with a strong SOC are less likely to use alcohol and illicit drugs.

Regarding confounding control, half of the studies included in the meta-analyses did not adjust their estimates for confounding by socioeconomic status or other variables. As for alcohol use, studies that reported crude estimates showed a stronger association compared with those that controlled for confounding variables, thereby suggesting that confounding overestimated the association between SOC and substance use. Socioeconomic and demographic factors play a crucial role in shaping meaningful experiences that contribute to developing a strong SOC in adulthood 12 . Socioeconomic status has been positively associated with SOC 13 but negatively associated with substance use 14 , 15 . Consequently, socioeconomic status would overestimate the magnitude of the association between SOC and substance use. As previously described, the magnitude of the association between SOC and alcohol use was weak in those studies that controlled for confounding variables. Thus, further studies evaluating the association of SOC with substance should adjust their estimates to socioeconomic variables.

This study has several strengths, such as the independent literature search conducted by two authors. All studies included collected self-reported data on both substance use and SOC, thus minimizing the occurrence of information bias. Additionally, only studies using validated and standardized instruments to assess SOC were included, thus reducing the possibility of misclassification. One limitation concerns the small number of identified studies measuring issues such as illicit drug use.

Analysis revealed a more consistent association of SOC with alcohol use than with smoking; however, precise measurement of this relation for illicit drugs was hindered by the limited number of studies available. Scarcity of evidence prevents definitive conclusions about the association between SOC and illicit substance use. This research gap suggests that SOC may be more strongly linked to behaviors related to more severe substances like illicit drugs and hazardous alcohol consumption, but confirming this hypothesis would require more comprehensive and specific investigation. Acknowledging this limitation highlights the pressing need for future research focused on understanding the correlation between SOC and use of different types of substances, particularly those considered severe.

Certainty of evidence as assessed by the GRADE system was very low, thus more robust study designs such as randomized clinical trials and longitudinal studies are needed to better understand the impact of a strong SOC on substance use (alcohol, smoking, and illicit drugs). Future research could examine how specific socioeconomic aspects interact with SOC to influence substance use, looking at factors like income, education, and occupation individually. Increasing the certainty of evidence through rigorous study designs is crucial for elaborating effective interventions and public health policies aimed at reducing harmful substance use and promoting overall well-being. Including diverse populations in research, considering ethnicity, culture, and geography, would enhance result generalizability and provide a more comprehensive understanding of how SOC operates across various demographic groups. Moreover, the consistency of our findings with another recent review involving adolescent and young adult populations highlights SOC as a potential protective factor against harmful substance use, reinforcing the clinical and public health relevance of these findings.

Overall, this systematic review and meta-analysis suggest that a strong SOC protects against substance use (alcohol, smoking, and illicit drugs) among adults, regardless of age. As practical implications, these findings suggest that early identification of individuals with low SOC may indicate the need for preventive interventions related to substance use. Recognizing and addressing a diminished SOC early could guide tailored interventions aimed at fortifying resilience and mitigating the risk of substance-related issues, especially in high-risk individuals.

Supplementary Material
==== Refs
References

1 Antonovsky A Health, stress and coping. London Jossey-Bass 1979
2 Antonovsky A Unraveling mystery of health: how people manage stress and stay well. San Francisco Jossey-Bass 1987
3 Urakawa K Yokoyama K Sense of coherence (SOC) may reduce the effects of occupational stress on mental health status among Japanese factory workers. Ind Health 2009 47 503 508 19834259
4 Silarova B Nagyova I Rosenberger J Studencan M Ondusova D Reijneveld SA Sense of coherence as a predictor of health-related behaviours among patients with coronary heart disease. Eur J Cardiovasc Nurs 2014 13 345 356 23828020
5 Morita Y Ohta M Inoue T Honda T Konno Y Eguchi Y Sense of coherence is significantly associated with both metabolic syndrome and lifestyle in Japanese computer software office workers. Int J Occup Med Environ Health 2014 27 967 979 25556337
6 Ristkari T Sourander A Helenius H Nikolakaros G Salanterä S Multimäki P Sense of coherence among Finnish young men - a cross-sectional study at military call-up. Nord J Psychiatry 2005 59 473 480 16316900
7 GBD 2019 Risk Factors Collaborators Global burden of 87 risk factors in 204 countries and territories, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet 2020 396 1223 1249 33069327
8 Institute for Health Metrics and Evaluation GBD results. https://vizhub.healthdata.org/gbd-results/ 13/Jun/2023
9 da-Silva-Domingues H Del-Pino-Casado R Palomino-Moral PÁ López Martínez C Moreno-Cámara S Frías-Osuna A Relationship between sense of coherence and health-related behaviours in adolescents and young adults: a systematic review. BMC Public Health 2022 22 477 477 35272648
10 Higgins JPT Thomas J Chandler J Cumpston M Li T Page MJ Cochrane handbook for systematic reviews of interventions. 2nd Chichester John Wiley & Sons 2019
11 Higgins JPT Thomas J Chandler J Cumpston M Li T Page MJ Cochrane handbook for systematic reviews of interventions, version 6.4 (updated August 2023). http://www.training.cochrane.org/handbook 18/Apr/2024
12 Bernabé E Watt RG Sheiham A Suominen-Taipale AL Nordblad A Savolainen J The influence of sense of coherence on the relationship between childhood socioeconomic status and adult oral health-related behaviours. Community Dent Oral Epidemiol 2009 37 357 365 19614720
13 Togari T Inoue Y Oshima G Abe S Hosokawa R Takaku Y Socioeconomic status and the sense of coherence among Japanese people living with HIV. Int J Environ Res Public Health 2022 19 7673 7673 35805330
14 Hiscock R Bauld L Amos A Fidler JA Munafò M Socioeconomic status and smoking: a review. Ann N Y Acad Sci 2012 1248 107 123 22092035
15 Probst C Kilian C Sanchez S Lange S Rehm J The role of alcohol use and drinking patterns in socioeconomic inequalities in mortality: a systematic review. Lancet Public Health 2020 5 e324 ee32 32504585
16 Neuner B Miller P Maulhardt A Weiss-Gerlach E Neumann T Lau A Hazardous alohol consumption and sense of coherence in emergency department patients with minor trauma. Drug Alcohol Depend 2006 82 143 150 16216439
17 Antonovsky H Hankin Y Stone D Patterns of drinking in a small development town in Israel. Br J Addict 1987 82 293 303 3471260
18 Tobamidanik L Zabkiewicz D Indicators of sense of coherence and alcohol consumption-related problems: the 2000 U.S. National Alcohol Survey. Subst Use Misuse 2009 44 357 373 19212927
19 Wainwright NW Surtees PG Welch AA Luben RN Khaw KT Bingham SA Healthy lifestyle choices: could sense of coherence aid health promotion? J Epidemiol Community Health 2007 61 871 876 17873222
20 Savolainen J Suominen-Taipale A Uutela A Aromaa A Härkänen T Knuuttila M Sense of coherence associates with oral and general health behaviours. Community Dent Health 2009 26 197 203 20088216
21 van Loon AJM Tijhuis M Surtees PG Ormel J Personality and coping: their relationship with lifestyle risk factors for cancer. Pers Individ Dif 2001 31 541 553
22 Saade SL Marchand A Work organisation conditions, alcohol misuse: the moderating role of personality traits. Work 2013 44 191 200 22927607
23 Von Ah D Ebert S Ngamvitroj A Park N Kang DH Factors related to cigarette smoking initiation and use among college students. Tob Induc Dis 2005 3 27 40 19570288
24 Larm P Åslund C Starrin B Nilsson KW How are social capital and sense of coherence associated with hazardous alcohol use? Findings from a large population-based Swedish sample of adults. Scand J Public Health 2016 44 525 533 27113963
25 Thomas K Nilsson E Festin K Henriksson P Lowén M Löf M Associations of psychosocial factors with multiple health behaviors: a population-based study of middle-aged men and women. Int J Environ Res Public Health 2020 17 1239 1239 32075162
26 Poppius E Tenkanen L Kalimo R Heinsalmi P The sense of coherence, occupation and the risk of coronary heart disease in the Helsinki Heart Study. Soc Sci Med 1999 49 109 120 10414844
27 Gajdosova B Orosova O Geckova AM Tavel P van Dijk JP Personality dimensions, sense of coherence and self-esteem as risk/protective factors for smoking among university students. Stud Psychol (Bratisl) 2009 51 183 192
28 Ahlstrand I Larsson I Larsson M Ekman A Hedén L Laakso K Health-promoting factors among students in higher education within health care and social work: a cross-sectional analysis of baseline data in a multicentre longitudinal study. BMC Public Health 2022 22 1314 1314 35804344
29 Luszczynska A Sense of coherence and smoking in a sample of abused women. Psychol Rep 2002 90 3 2 1123 1128 12150396
30 Vilela LD Allison PJ An investigation of the correlates of sense of coherence in a sample of Brazilians with head and neck cancer. Oral Oncol 2010 46 360 365 20371202
31 Igna CV Julkunen J Ahlström R Sense of coherence relates with smoking. J Health Psychol 2008 13 996 1001 18987071
32 Verešová M Gatial V Sense of coherence and proactive coping of university students: relations with alcohol and tobacco use frequency. Psychol Health 2010 25 360 360
33 Kouvonen AM Väänänen A Woods SA Heponiemi T Koskinen A Toppinen-Tanner S Sense of coherence and diabetes: a prospective occupational cohort study. BMC Public Health 2008 8 46 46 18254945
