Skip to main content
Key Specialties
Cardiology Diabetology Endocrinology Gastroenterology Geriatrics and Gerontology Gynecology and Obstetrics Hematology Infectious Disease Internal Medicine Nephrology Neurology Oncology Pediatrics Respiratory Medicine Rheumatology Surgery
Congress Coverage
2024 ACC Congress 2023 ESMO Congress 2023 EASD Congress 2023 ERS Congress 2023 ESC Congress 2023 EHA Congress 2023 ASCO Annual Meeting
Clinical Collections
Advances in Alzheimer’s

At a glance: The ONWARDS insulin icodec trials

A round-up of the ONWARDS phase 3 clinical trials evaluating the novel weekly insulin icodec in people with diabetes.
ADVANCED SEARCH
Log in
Top
Journal of Public Health
Published in: Journal of Public Health 4/2017

01-08-2017 | Original Article

Methodology for identification of healthstyles for developing effective behavior change interventions

Authors: Eszter Balku, Gergely Tóth, Erzsébet Nárai, Emese Zsiros, Péter Varsányi, József Vitrai

Published in: Journal of Public Health | Issue 4/2017

Login to get access

Abstract

Aim

The health of the Hungarian population is not as good as it could be according to the socioeconomic development of the country. Since unhealthy behavior is widespread in the population, behavior change seems to be an appropriate tool to improve health in Hungary. To develop effective interventions, it is first necessary to identify homogeneous groups of people from the point of view of behavior change, i.e., healthstyles. Our aim was to develop a suitable survey methodology and segmentation procedure for identifying healthstyles in Hungary. In this article we present the results of our methodological developments.

Subjects and methods

Several blocks of questions were developed based on the COM-B model, which synthesizes the most recognized behavior change models. The questions cover knowledge about and attitude to health, somatic and mental health status, subjective well-being, psychological characteristics, health behaviors, social support, media consumption, health-related information seeking and socio-demographic characteristics. Nationally representative two-stage samples of schoolchildren and adults were drawn. Iterative weighting was applied, which was supplemented with a design weight correcting the design effects. For segmentation, latent cluster analysis was used.

Results

Questionnaires based on the COM-B model were developed and administered to schoolchildren and adults. Several numerical and graphical methods were developed to explore the statistical characteristics of the segmentation models obtained by latent cluster analysis.

Conclusions

The developed survey methodology and segmentation methods were successfully applied to produce statistically stable clusters in both schoolchildren and adults. The statistical methods chosen for segmentation and validation of clusters seem suitable to identify healthstyles.
Footnotes
1
The database contains data about 4438 schools in the 2014–2015 school year. Based on the database the number of fifth classes was 4556, the number of seventh classes was 4680, and the number of ninth classes was 4755.
 
2
Social Renewal Operational Program 6.1.3./B/12/1 “Developing public health communication”
 
3
For a more comprehensive description of the statistical viewpoint of the model-based framework, see Sterba (2009).
 
4
Due to the high number of variables involved (more than 100), application of imputation methods was excluded for theoretical reasons.
 
5
The software version is R 3.2.1., and the computations were performed on a 64-bit Windows 2012 R2 server.
 
6
The ‘poLCA’ package (version 1.4.1) was built under R version 3.2.2.
 
7
In the sense of interpretation, this nature of the LCA method makes the results similar to those that come from a factor model.
 
8
Technically, the model selection in LCA aims to specify the number of clusters.
 
9
Identification of many separate clusters would be counterproductive, and technically it would be too difficult to overview and interpret all of them.
 
10
To describe the axes, a table containing the most important variables defining the axes belongs to each plot, which is not described in this article.
 
11
The adjusted standardized residual is the Pearson residual divided by the standard deviation of all residuals.
 
12
The positive values indicate that the given attribute is typical of a cluster, and in contrast the negative values characterize the cluster by the lack of a given attribute. If the absolute value exceeds two, then we can reasonably say that the given attributes are statistically significant. The higher the absolute value is, the more significant the attributes are. It is important to note that significant values do not mean exclusivity; they denote just the relative differences from the expected values.
 
13
As it can validate the model externally.
 
14
Of course, the identification of the different health-style clusters is not always so obvious as in the example above. (A detailed discussion of these models is not part of this article. We would like to present them in a different publication in the future.)
 
15
Symmetric and asymmetric Goodman-Kruskall λ, Theil’s uncertainty coefficient and Cramer’s V (Liebetrau 1983).
 
16
We have also displayed some of these values on the plots above.
 
17
The disadvantage of the applied simulation method is its time consumption as it was a function of the amount of the input elements. In general, the requested amount of time was more than 1 h per iteration, and 25 iterations were made for each parameter set.
 
18
For the purpose of comparability of the different models, the deviation was normalized to the expected value of the deviation given by the complete random reclassification. For example, the fifth-grade children’s model consists of three clusters with approximately the same size; therefore, the rate of misclassification in a completely random reclassification is approximately 66%—but depends on the sizes of each cluster. (Generally speaking, the expected value of the deviation is a function of the number of clusters and the distribution of the cluster sizes. Without the normalization, the comparison of the different stability models may lead to an erroneous conclusion.)
 
19
As the two examined attributes (rate of omitted variables or observations) interact with each other, the dots are partial values, as is the displayed non-linear model fitted on these partial values.
 
20
For the purpose of evaluating the results, investigating the transition function of the deviations is necessary as it can be interpreted as an indicator of the sensibility.
 
21
Approximately 0.75 is the expected value of the relative deviation from the simple random model omitting 75% of the cases in contrast to omitting 75% of variables—the corresponding values do not exceed 0.4 in either case.
 
22
The final model for the seventh-grade children is incomparably less stable than the others. One possible assumption is that the confusion of responses was related to the stage of personality development (Candida 2013), but to declare it with certainty we need more investigations.
 
Literature
Bock HH (1987) On the interface between cluster analysis, principal component analysis, and multidimensional scaling. In: Bozdogan and Gupta (eds) Multivariate statistical modeling and data analysis. Springer Netherlands, p 17–34. doi:10.​1007/​978-94-009-3977-6_​2 CrossRef
Candida CP (2013) Looking forward through the lifespan: developmental psychology. Pearson Higher Education, Melbourne
Cantril H (1965) The pattern of human concerns. Rutgers Un, New Brunswick
Costa FM, Jessor R, Turbin MS, Dong Q, Zhang H, Whang C (2005) The role of social contexts in adolescence: context protection and context risk in the United States and China. Appl Dev Sci 9:67–85. doi:10.​1207/​s1532480xads0902​_​3 CrossRef
Dziak JJ, Coffman DL, Lanza ST, Li R (2012) Sensitivity and specificity of information criteria. PeerJ Inc., Pennsylvania
European Commission (1999) A pan-EU survey on consumer attitudes to physical activity. Body-weight and Health, Luxembourg
Gebauer F (2008) Applicability of the Mérei’s “multidimensional sociometry” today. Budapesti Nevelő 2:67–72 [in Hungarian]
Gordon W, Langmaid R (1988) Qualitative market research: a practitioner’ and buyer’s guide. Gower, London
Jessor R, In Bonino S, Cattelino E, Ciairano S (2005) Adolescents and risk: behaviors, functions, and protective factors. New York
Lazarsfeld P (1950a) The logical and mathematical foundations of latent structure analysis. In: Stouffer SA (ed) Measurement and prediction, the American soldier: studies in social psychology in World war II. Princton University Press, Princeton, pp 362–412
Lazarsfeld P (1950b) The interpretation and computation of some latent structures. In: Stouffer SA (ed) Measurement and prediction, the American soldier: studies in social psychology in world war II. Princton University Press, Princeton, pp 413–472
Liebetrau AM (1983) Measures of association. Quantitative applications in the social sciences series no. 32. Sage Publications, Newbury Park
Macías YF, Glasauer P (2014) Guidelines for assessing nutrition-related Knowladge, attitudes and practices. Food and Agriculture Organization of the United Nations, Rome
Majerszki N (2014) The Z generation of the twenty first century. Verpelét
Morris J, Marzano M, Dandy N, O’Brien L (2012) Theories and models of behaviour and behaviour change. For Sustain Behav Behav Chang Theor :1–27
Nárai E (2009) Healthstyle segmentation, attitudes to health. Eötvös Loránd University
Ngai EWT, Xiu L, Chau DCK (2009) Application of data mining techniques in customer relationship management: a literature review and classification. Expert Syst Appl 36:2592–2602CrossRef
Pikó B (2008) Value orientation and health behavior in adolescents. Iskolakultúra :62–86
Rokeach M (1973) The nature of human values. Free Press, Michingan
Schwarzer R, Jerusalem M (1995) General self-efficacy scale. Meas Heal Psychol A user’s portfolio Causal Control beliefs. :35–37. doi: 10.​1037/​t00393-000
Utility C, Properties P (2003) AUDIT-C questionnaire. Intern Med 163:2–3
Vermunt JK (2003) Applications of latent class analysis in social science research. In: Nielsen TD, Nevin Lianwen Z (eds) Symbolic and quantitative approaches to reasoning with uncertainty. Springer, Berlin Heidelberg, pp 22–36CrossRef
Vermunt J, Magidson J (2002) Latent class cluster analysis. In: Hagenaars J, McCutcheon A (eds) Applied latent class analysis. Cambridge University Press, Cambridge, pp 89–106CrossRef
Wardle AJ, Steptoe A, Wardle J (2003) Socioeconomic differences in attitudes and beliefs about healthy lifestyles. Epidemiol Commun Health 57:440–443CrossRef
Weber EU, Blais A-R, Betz NE (2002) A domain-specific risk-attitude scale: measuring risk perceptions and risk behaviors. J Behav Decis Mak 15:263–290. doi:10.​1002/​bdm.​414 CrossRef
Zanotti CCC, Currie AMD, Roberts MDLC, Barnekow OSRFSV (2010) Social determinants of health and well-being among young people. Health Policy Child Adolesc 272
Metadata
Title
Methodology for identification of healthstyles for developing effective behavior change interventions
Authors
Eszter Balku
Gergely Tóth
Erzsébet Nárai
Emese Zsiros
Péter Varsányi
József Vitrai
Publication date
01-08-2017
Publisher
Springer Berlin Heidelberg
Published in
Journal of Public Health / Issue 4/2017
Print ISSN: 2198-1833
Electronic ISSN: 1613-2238
DOI
https://doi.org/10.1007/s10389-017-0799-y

Other articles of this Issue 4/2017

Journal of Public Health 4/2017 Go to the issue

Original Article

Effect of basic periodontal treatment on glycemic control and inflammation in patients with diabetes mellitus type 1 and type 2: controlled clinical trial

Original Article

Diagnostic accuracy of different methods of early detection of chronic kidney disease

Original Article

Procedures in child deaths in The Netherlands: a comparison with child death review

Original Article

Women’s life span and age at parity

Erratum

Erratum to: Genetics in the medical and psychosocial practice: on the role of value hierarchies in counselling

Original Article

The cost-effectiveness of population Health Checks: have the NHS Health Checks been unfairly maligned?