13. Personalised Approach to Obesity Management in Children
This chapter covers the case study of the Personalised Approach to Obesity Management in Children (PAOMC), a comprehensive, clinical, family-based and personalised childhood obesity intervention targeting children aged seven to 17 years. The case study includes an assessment of PAOMC against the five best practice criteria, policy options to enhance performance and an assessment of its transferability to other OECD and EU27 countries.
Description: the Personalised Approach to Obesity Management In Children (PAOMC) intervention in Estonia is a comprehensive, clinical, family-based and personalised childhood obesity intervention targeting children aged 7 to 17 years. The intervention is delivered in a hospital outpatient clinic setting.
Table 13.1. OECD best practice assessment of PAOMC | |
|---|---|
Criteria | Assessment |
Effectiveness | The intervention successfully impacted participants’ anthropometric measurements and physical activity levels |
Efficiency | Evidence on PAOMC’s efficiency is not available, but, previous analysis by OECD indicates obesity management interventions for children may be cost-effective |
Equity | The intervention addresses a health issue that disproportionally affects poorer children, however, these children are less likely to access PAOMC due to access inequalities |
Evidence-base | PAOMC was evaluated using data collection methods that are reliable and validated, further, the study was designed to reduce selection bias. However, similar to many public health interventions, neither researchers not participants were blinded. |
Extent of coverage | There is not publically available information on participation or dropout rates, however, a high level analysis estimated that nearly 20 000 children could be referred to PAOMC if scaled-up across the country |
Enhancement options: to enhance effectiveness, programme administrators could consider incorporating additional obesity counselling sessions, education and behavioural therapy. To enhance the evidence-base, a control group and a longer-term follow-up could also be included, and more objective evaluation methods could be employed. To enhance equity, the programme could seek to recruit participants from more vulnerable groups, and adapt the intervention to their specific needs. Finally, to enhance extent of coverage, a larger sample size of participants could be recruited in upscaling or adapting this intervention.
Transferability: PAOMC addresses childhood obesity, which is of key political interest in most countries. Further, most countries with available data have either implemented or foresee implementing programs to support physical activity counselling by health professionals indicating greater levels of workforce acceptability of PAOMC. Nevertheless, the success of PAOMC in the target setting will depend on a range of contextual factors, in particular, the willingness and ability of GPs and paediatricians to provide children and their parents with obesity related advice.
Conclusion: although data was not available to assess the intervention fully in terms of costs, evidence-base, equity and extent of coverage, the PAOMC study in Estonia can be considered a best practice in terms of outcomes. To further enhance implementation, programme administrators could take into consideration policy options laid out in this case study, such as including additional behavioural therapy and obesity counselling sessions.
Cardiovascular diseases (CVDs) are a leading cause of death, comprising approximately half of all non-communicable disease (NCD) deaths (Benziger, Roth and Moran, 2016[1]). One of the primary determinants of CVD is obesity, as well as its associated comorbidities (diabetes, hypertension). In 2018, almost 60% of people in OECD countries were overweight, and 25% were obese (OECD, 2019[2]). The adoption of unhealthy behaviours leading to the development of CVD risk factors takes place in early childhood (Peñalvo et al., 2013[3]). However, obesity is largely preventable, highlighting the importance of effective health promotion early on in the life course.
The PAOMC intervention in Estonia is a family-based paediatric obesity intervention targeting children aged 7-17 years with pre-obesity (i.e. overweight) or obesity. The study included 58 children in Tallinn and the surrounding Harju County, and largely took place in the Tallinn Children’s Hospital outpatient clinic. During the intervention, families received education materials, and a personalised assessment of both the parents’ and children’s lifestyle behaviours, as well as of their willingness to change. Moreover, the children received personal counselling, physical activity (PA) assessments and sessions (such as hikes, outdoor exercise activities and gym training), attended bi-weekly education sessions (PA and diet), and set lifestyle goals for a 12 month period.
Children were referred to PAOMC through their family doctor (general practitioner (GP)), general paediatricians, and endocrinologists or by other family initiatives.
This section analyses PAOMC against the five criteria within OECD’s Best Practice Identification Framework – Effectiveness, Efficiency, Equity, Evidence-base and Extent of coverage (see Box 1.1 for a high-level assessment of PAOMC). Further details on the OECD Framework can be found in Annex A.
Equity
The intervention addresses a health issue that disproportionally affects poorer children – in Estonia 26% of children in the poorest quintile live with pre-obesity or obesity compared to 18% in the highest quintile
Despite being in greater need of obesity prevention programs, children in the lowest income quintile are less likely to see general practitioner compared to the higher income quintile (64% versus 72%)
It is unclear whether specific efforts were made to address other disadvantaged groups (e.g. ethnic minorities)
Effectiveness
PAOMC reduces BMI levels and increases physical activity at rates greater than similar family-based clinical childhood obesity interventions
The PAOMC intervention was generally successful in improving participants’ anthropometric measurements and PA levels. Prior to the intervention, 93% of the children selected for the intervention lived with obesity (BMI ≥ 95 percentile) and 7% with pre-obesity (BMI 85-95 percentile) (Suurorg et al., 2017[4]). Two years after the intervention, 42% had decreased their BMI category, 58% had experienced weight loss, 70% had shown participation in PA sessions, 94% had seen improvements in their sit-up tests and 65% in their six-minute walking test. Nonetheless, 27% did not experience any weight change and 15% saw an increase in weight (Suurorg et al., 2017[4]).
While a comparison can only be carried out at the high-level, given that studies may have used different methodologies and that, across studies, target populations an activities within the intervention may not be fully comparable, the results from the PAOMC intervention seem to be greater than those of similar family-based clinical childhood obesity interventions. Indeed, the Families for Health trial in the United Kingdom, a family-based childhood obesity treatment intervention in a primary care setting, did not result in significant differences in BMI z-score at 12 months between the intervention and control groups (Robertson et al., 2017[5]). Moreover, the High Five for Kids study (also a primary care-based childhood obesity prevention intervention) in the United States was shown to have non-significant change in participants’ BMI (p = 0.15) (Wright et al., 2014[6]). These results point to the potential of the PAOMC programme and underline the importance of comprehensive, family-based and personalised obesity interventions in childhood.
Drawing upon OECD analysis, the potential impact of expanding PAOMC to the national level can be estimated. Specifically, OECD’s 2019 obesity report shows that PA and nutrition interventions targeting school-children could avoid over 65 cases of CVD and 149 cases of diabetes in Estonia between years 2020-50 (OECD, 2019[2]). An additional 0.11 life years (LYs) and 0.98 disability-adjusted life years (DALYs) could also be gained in Estonia per 100 000 people annually from 2020-50. These figures might provide an indication of the potential of the PAOMC programmed if scaled-up in Estonia to a national level. However, the results focus on school-based obesity programmes, rather than on family-based, clinical paediatric obesity interventions, and therefore a range of precautions must be taken in interpreting results.
Efficiency
Analysis by OECD finds obesity management programs targeting school-aged children are cost-effective
Publically available information on the costs of operating PAOMC show the intervention’s overall budget is approximately EUR 25 000 (USD PPP 36 547) per year (Kramer and Suurgorg, 2017[7]). However, it is unclear how many children this covers, therefore a cost per participant is not available.
An analysis of school-based programs designed to reduce rates of overweight and obesity by the OECD can shed light on the potential effect of interventions such as PAOMC. OECD estimates that heathy lifestyle programs targeting school-aged children lead to health expenditure savings of USD PPP 0.01 (EUR 0.01) per capita in health expenditure annually, during the first 30 years after implementation (OECD, 2019[2]). The main reason behind this result is that health care expenditure for NCDs is small in children and young adults due to low incidence rates. Across the whole population, this translates into annual health expenditure savings of USD PPP 12 113. It is expected that in the long term, when children targeted by the intervention reach their 50s’, the impact of the intervention may become larger. Finally, caution should be taken when interpreting these results given PAOMC is implemented in a primary-care setting while OECD analysis relies on findings from school-based interventions.
Equity
PAOMC addresses a health issue that disproportionately affects poorer children, however, it is less likely to reach these children due to access inequalities
PAOMC does not directly target a priority population group however it addresses a health issue that disproportionately affects children from lower-income households. In Estonia, the proportion of children with pre-obesity or obesity was 26% for those in the poorest quintile compared to 18% in the highest quintile (OECD/European Union, 2018[8]). Nevertheless, obesity interventions delivered in a primary care setting may be less likely to reach children living in less affluent families which risks widening existing health inequalities (OECD, 2019[9]). For example, analysis by OECD estimates that after adjusting for needs, in Estonia, 64% of people in the lowest income quintile accessed a GP in the past year compared to 72% in the highest quintile (OECD, 2019[9]).1
It is unclear from the available information whether specific efforts were made to ensure other disadvantaged groups, such as children from different ethnic backgrounds or with a low socio-economic status, accessed PAOMC.
Evidence-base
Strong data collection methods were used to evaluate PAOMC
A cross-sectional, observational study was used to evaluate PAOMC. Measures were taken for anthropometric markers (BMI, height, weight, waist circumference) and physical fitness levels (6-minute walking test and sit-up test), although no information is available on the mediums used. The childrens’ and parents’ desire for behavioural lifestyle change was assessed by the WHO Visual Analogue Scale (VAS) and self-reported questionnaires. Dietary and PA behaviours were determined through assessments led by physicians, surveys and the Yale food addiction scale, a self-reported questionnaire (Suurorg et al., 2017[4]). The use of these tools by qualified health professionals increase the reliability and validity of the evaluation results. For these reasons, the data collection methods used to evaluate PAOMC are considered “strong” against the Quality Assessment Tool for Quantitative Studies framework (see Table 1.1) (Effective Public Health Practice Project, 1998[10]).
The method used to evaluate PAOMC also performed well in regard to the study design as well as reducing selection bias. However, similar to many public health interventions, neither participants nor researchers were blinded, therefore the study is considered “weak” in this area. Further, it was unclear if researchers controlled for confounders.
Extent of coverage
Nearly 20 000 children could be referred to PAOMC by their GP if it were scaled-up across the whole country
There is no publically available evidence to measure participation or dropout rates for PAOMC. However, data on the probability of visiting a GP in the last 12 months is available, which provides a conservative insight into how many children could access PAOMC if it were scaled-up across the whole of Estonia (given children and adolescents can be referred by their GP to PAOMC).
By multiplying the probability of a GP visit in the last year2 by the number of children aged 7-17 years with pre-obesity or obesity, it is estimated that 18 776 children could be referred to PAOMC via their GP if it were scaled-up across the country (out of 30 285 who are eligible) (Statistics Estonia, 2020[11]; WHO, 2016[12]; Eurostat, 2019[13]).3
The PAOMC intervention fits many of the overarching best practice criteria in terms of clinical, family-based childhood obesity programmes. The trial is a comprehensive programme involving the family and wider support networks, which strongly targets both diet and PA. Moreover, it focuses on behaviour change at the family and individual level.
Enhancing effectiveness
Literature on best practices in this field underlines the importance of obesity counselling, education and behavioural therapy in addition to nutrition and exercise (Mead et al., 2017[14]). However, it is important to note that many of these policies fall outside the responsibilities of programme administrators and instead require input from higher-level policy makers (e.g. at the national level). Nonetheless, in upscaling or adapting this intervention, further emphasis on motivational interviewing (MI), positive reinforcement, monitoring, and cognitive restructuring could be considered to enhance effectiveness (Davis et al., 2007[15]). Indeed, a systematic review of the treatment of paediatric obesity found that multicomponent interventions targeting not only diet and PA, but which also included a strong emphasis on behavioural therapy and education achieved the most significant outcomes in terms of reductions in systolic and diastolic blood pressure, BMI, and triglycerides (Rajjo et al., 2017[16]). Moreover, a meta-analysis of the effectiveness of MI concluded that this practice could lead to up to 51% improvement rates in the treatment of problem behaviours (Burke, Arkowitz and Menchola, 2003[17]). MI can help induce behaviour change through guiding individual reflection, as participants are more likely to accept and act on opinions, which they have voiced themselves. Furthermore, shifting participants’ thinking patterns and managing their expectations can lead to higher adherence to dietary change (Burke, Arkowitz and Menchola, 2003[17]).
Enhancing efficiency
Efficiency is calculated by obtaining information on effectiveness and expressing it in relation to inputs used. Therefore policies to boost effectiveness without significant increases in costs will have a positive impact on efficiency.
Enhancing equity
To enhance equity, to the extent possible, programme administrators are encouraged to undertake a review to determine whether the intervention should be adapted to meet the needs of priority population groups. In order to better understand how different groups of participants benefit from the intervention, future evaluations should break down key indicators, for example, by family socio-economic status and ethnicity. Finally, additional effort to recruit families and children from groups which have lowers level of access to health care is important, particularly if these groups have higher rates of obesity (as outlined under “Efficiency”, nearly 40% of children in the lowest income quintile won’t access a GP and therefore have the opportunity to be referred to PAOMC).
Enhancing the evidence-base
To enhance the evidence-base, future evaluations could consider using a blinded randomised study design if considered ethical in order to better understand the true effect of PAOMC. A longer follow-up would also improve the validity of evaluation results, for example, by collecting data 12 months after the intervention has finished. Moreover, alternatives to questionnaires could be considered to assess dietary and PA habits, as well as behaviour change willingness. Indeed, in order to assess the long-term impact (e.g. after 10 years) of PAOMC on rates of obesity and overweight, it is necessary to gather data according to the same measures, and when possible, with the same individuals (i.e. panel data). Longitudinal panel data is deemed to be the “gold standard”, given that it reduces bias by accounting for differences amongst individuals. However, as this requires long-term funding and support, responsibility for this option lies with high-level policy makers, rather than with the PAOMC study group.
Enhancing the extent of coverage
Given limited information on the extent of coverage for PAOMC, specific polices to boost uptake have not been included. However, in general, efforts to boost health literacy amongst children and parents are likely to increase motivation to participate in programs such as PAOMC.
This section explores the transferability of PAOMC and is broken into three components: 1) an examination of previous transfers; 2) a transferability assessment using publically available data; and 3) additional considerations for policy makers interested in transferring PAOMC.
Previous transfers
To date, PAOMC has not been transferred outside of Estonia, however various personalised obesity intervention targeting children and adults exist. For example, Sweden’s Prescription on Physical Activity (see Chapter 4) intervention is in the process of being transferred to several EU countries.
Transferability assessment
The following section outlines the methodological framework to assess transferability and results from the assessment.
Methodological framework
Details on the methodological framework to assess transferability can be found in Annex A. Indicators from publically available datasets to assess the transferability of PAOMC are listed in Table 1.2. Please note, the assessment is intentionally high level given the availability of public data covering OECD and non-OECD European countries.
Results
Results from the transferability assessment of PAOMC in Estonia to OECD and non-OECD EU countries are in Table 1.3. Overall, PAOMC is likely to have political support from countries given nearly all countries have a specific strategy targeting childhood obesity. In addition, most countries with available data (77%) have either implemented or foresee implementing programs to support physical activity counselling by health professionals indicating greater levels of workforce acceptability of PAOMC (i.e. given the health profession will be more accustomed to providing this service). Data on remaining indicators shows mixed results, further, for these indicators there are high levels of missing data in non-European countries.
To help consolidate findings from the transferability assessment above, countries have been clustered into one of three groups, based on indicators reported in Table 13.3. Countries in clusters with more positive values have the greatest transfer potential. For further details on the methodological approach used, please refer to Annex A.
Key findings from each of the clusters are below with further details in Figure 1.1 and Table 1.4:
Countries in cluster one have population, sector specific, economic and political arrangements in place that support transferring PAOMC, and therefore are less likely to experience implementation barriers.
The majority of countries fall under cluster two, which have political policies in place that support PAOMC. However, prior to transferring PAOMC, countries may wish to consider increasing funding for primary care to ensure the intervention is affordable in the long term. It is important to note that Estonia, which currently operates PAOMC, is in this cluster, indicating although ideal, high levels of spending on primary care is not a pre-requisite for transferring PAOMC.
Remaining countries are in cluster three, where spending on primary care is high, yet changes to the population, sector specific and political contexts may need to be addressed to ensure a successful transfer. For example, by ensuring health professionals receive training on how to lead a healthy lifestyle in countries such as Greece.
Note: Bar charts show percentage difference between cluster mean and dataset mean, for each indicator.
Source: WHO (n.d.[18]), “Global Health Observatory”, https://www.who.int/data/gho; WHO Regional Office for Europe (2021[19]), “2021 Physical Activity Factsheets for the European Union Member States in the WHO European Region”, https://apps.who.int/iris/bitstream/handle/10665/345335/WHO-EURO-2021-3409-43168-60449-eng.pdf; OECD (2020[20]), How’s Life? 2020: Measuring Well-being, https://dx.doi.org/10.1787/9870c393-en; Eurostat (2014[21]), “Self-reported time elapsed since last visit to a medical professional by sex, age and educational attainment level (from 15 to 19 years)”, https://ec.europa.eu/eurostat.
New indicators to assess transferability
Data from publically available datasets is not ideal to assess the transferability of PAOMC. Therefore, Box 1.2 outlines several new indicators policy makers should consider before transferring PAOMC.
In addition to the indicators from secondary sources of data outlined above, the following primary source indicators to measure transferability are recommended.
Population context
What is the ethnicity and cultural diversity of the target population?
What is the level of acceptability of PAOMC amongst parents?
What is the level of health literacy amongst parents? (e.g. knowledge regarding what constitutes health eating, and the impact of healthy eating and exercise on overall health and well-being)
What proportion of school-aged children access primary care services? (Does this figure differ between children with a normal weight and those who are overweight or obese?)
Sector specific context (primary care)
What other obesity management interventions exist for school-aged children?
Have GPs and paediatricians received appropriate training for treating children with obesity?
Do GPs and paediatricians feel comfortable prescribing obesity treatment for children?
Do GPs and paediatricians feel it is their responsibility for prescribing obesity treatment to children?
What proportion of the population access primary care as the entry point to receiving health care?
Is there a culture of health promotion and disease prevention within the health care system?
Over the course of the past three decades, there has been a significant increase in the prevalence of overweight and obesity worldwide. The adoption of inadequate lifestyle behaviours leading to situations of obesity or overweight takes place in early childhood (Peñalvo et al., 2013[3]).The PAOMC intervention seeks to counter such behaviours through a personalised, family-based, paediatric obesity programme in a primary care setting.
The results from this study show that the intervention was successful in positively impacting anthropometric measurements and PA levels amongst children. Details on the intervention’s efficiency were not publically available, however, previous OECD analysis indicates obesity management interventions targeting school-aged children are cost-effective, but produce a population-level impact only in the long-term. The data used to evaluate the effectiveness of PAOMC was derived from a cross-sectional, observational study, which is rated as weak evidence. PAOMC did not directly target a priority population group, nevertheless, it has the potential to reduce health inequalities given it targets a risk factor which disproportionately affects lower-SES children.
PAOMC includes many characteristics considered essential for a successful family-based, clinical obesity interventions in primary care settings. However, further changes, such as incorporating additional behaviour therapy and obesity counselling sessions, could be considered to achieve the intervention’s core objective: reducing obesity and overweight, and improving overall lifestyle behaviours among children in Estonia.
Finally, PAOMC addresses childhood obesity, which is of key political interest, further, it is likely to be supported by health professionals as they are accustomed to providing this type of treatment. Nevertheless, the success of PAOMC in the target setting will depend on a range of contextual factors, in particular, the willingness and ability of GPs and paediatricians to provide children and their parents with obesity related advice.
Box 1.3 outlines next steps for policy makers and funding agencies regarding the PAOMC intervention.
Next steps for policy makers and funding agencies to enhance PAOMC are listed below:
Support future evaluations of PAOMC for example by providing funding to collect participant data beyond the end of the intervention (e.g. 1 year after). By collecting longitudinal (panel) data, policy makers will gain a better understanding of the long-term impact of PAOMC.
Support multipronged policy efforts to boost levels of population health literacy.
Ensure funding to continue the implementation of the intervention as well as for future scale-up and transfer efforts.
Promote findings from the PAOMC case study to better understand what countries/regions are interested in transferring the intervention.
References
[1] Benziger, C., G. Roth and A. Moran (2016), The Global Burden of Disease Study and the Preventable Burden of NCD, Elsevier B.V., https://doi.org/10.1016/j.gheart.2016.10.024.
[17] Burke, B., H. Arkowitz and M. Menchola (2003), “The efficacy of motivational interviewing: A meta-analysis of controlled clinical trials.”, Journal of Consulting and Clinical Psychology, Vol. 71/5, pp. 843-861, https://doi.org/10.1037/0022-006x.71.5.843.
[15] Davis, M. et al. (2007), “Recommendations for Prevention of Childhood Obesity”, Pediatrics, Vol. 120/Supplement 4, pp. S229-S253, https://doi.org/10.1542/peds.2007-2329e.
[10] Effective Public Health Practice Project (1998), Quality assessment tool for quantitative studies, https://www.nccmt.ca/knowledge-repositories/search/14.
[13] Eurostat (2019), Self-reported time elapsed since last visit to a medical professional by sex, age and educational attainment level.
[21] Eurostat (2014), Self-reported time elapsed since last visit to a medical professional by sex, age and educational attainment level (from 15 to 19 years).
[7] Kramer, I. and L. Suurgorg (2017), Personalized approach in child obesity management, https://janpa-toolbox.eu/files/Estonia-PERSONALIZED%20APPROACH%20IN%20CHILD%20OBESITY%20MANAGEMENT.pdf.
[14] Mead, E. et al. (2017), “Diet, physical activity and behavioural interventions for the treatment of overweight or obese children from the age of 6 to 11 years”, Cochrane Database of Systematic Reviews, https://doi.org/10.1002/14651858.cd012651.
[20] OECD (2020), How’s Life? 2020: Measuring Well-being, OECD Publishing, Paris, https://doi.org/10.1787/9870c393-en.
[9] OECD (2019), Health for Everyone?: Social Inequalities in Health and Health Systems, OECD Health Policy Studies, OECD Publishing, Paris, https://doi.org/10.1787/3c8385d0-en.
[2] OECD (2019), The Heavy Burden of Obesity: The Economics of Prevention, OECD Health Policy Studies, OECD Publishing, Paris, https://doi.org/10.1787/67450d67-en.
[8] OECD/European Union (2018), Health at a Glance: Europe 2018: State of Health in the EU Cycle, OECD Publishing, Paris/European Union, Brussels, https://doi.org/10.1787/health_glance_eur-2018-en.
[3] Peñalvo, J. et al. (2013), “A cluster randomized trial to evaluate the efficacy of a school-based behavioral intervention for health promotion among children aged 3 to 5”, BMC Public Health, Vol. 13/1, https://doi.org/10.1186/1471-2458-13-656.
[16] Rajjo, T. et al. (2017), “Treatment of Pediatric Obesity: An Umbrella Systematic Review”, The Journal of Clinical Endocrinology & Metabolism, Vol. 102/3, pp. 763-775, https://doi.org/10.1210/jc.2016-2574.
[5] Robertson, W. et al. (2017), “Randomised controlled trial evaluating the effectiveness and cost-effectiveness of ‘Families for Health’, a family-based childhood obesity treatment intervention delivered in a community setting for ages 6 to 11 years”, Health Technology Assessment, Vol. 21/1, pp. 1-180, https://doi.org/10.3310/hta21010.
[11] Statistics Estonia (2020), RV021: Population, 1 January by Sex, Year and Age group (5-9, 10-14 and 15-19).
[4] Suurorg, L. et al. (2017), Overview of the project: personalized approach to obesity management in children (Supplement 1), Tallinn Children’s Hospital, Tallinn, Estonia, http://janpa-toolbox.eu/files/Supplementes_Suurorg_web%20Jan%2017th%202017.pdf.
[12] WHO (2016), Prevalence of overweight among children and adolescents, BMI > +1 standard deviations above the median (crude estimate) (%).
[18] WHO (n.d.), Global Health Observatory, https://www.who.int/data/gho (accessed on 25 August 2021).
[19] WHO Regional Office for Europe (2021), 2021 Physical Activity Factsheets for the European Union Member States in the WHO European Region, https://apps.who.int/iris/bitstream/handle/10665/345335/WHO-EURO-2021-3409-43168-60449-eng.pdf.
[6] Wright, D. et al. (2014), “The cost of a primary care-based childhood obesity prevention intervention”, BMC Health Services Research, Vol. 14/1, https://doi.org/10.1186/1472-6963-14-44.
Notes
← 1. Data is for the adult population, parents and non-parents. Nevertheless, it has been used as it is assumed parents, in most cases, are responsible for their child’s health care appointments.
← 2. Data is only available for those aged 15-19 years.
← 3. Estimated population aged 7-17 in Estonia = 156 106 * proportion of children with pre-obesity or obesity in Estonia = 19.4% * the probability of visiting a GP in the past year = 62%. Population data is only available by age groups which don’t identically align with 7-17, therefore the average was used.