Self-perceived health and burden of diabetes in teenagers with type 1 diabetes: psychometric properties of the Swedish measure ‘Check your health’

Authors


Gunnel Viklund, Section of Nursing research, Department of Medical Sciences, Döbelnsgatan 2C, S-752 37 Uppsala, Sweden.
Tel: +46 707146294 | Fax: +46 851777739 |
Email: gunnel.viklund@karolinska.se

Abstract

The aim of this study was to test the psychometric properties of the instrument ‘Check your health’ in teenagers with type 1 diabetes. The instrument measures ‘self-reported health’ and ‘burden of diabetes’. A convenience sample of 199 teenagers, 12–17 years of age, completed the questionnaires ‘Check your health’ and DisabKids when visiting the diabetes clinic. Forty-seven patients completed the questionnaires at home a second time. In the reliability test, the correlation between test and retest was found to be satisfactory, (0.94–0.62, except for social burden, 0.41). Convergent validity was moderate (0.62–0.38), while the instrument showed good discriminant validity. Self-reported health and burden of diabetes were different in boys and girls, in patients with good or poor metabolic control or who reported high and low disease severity. The domain burden of diabetes turned out to be very sensitive.

Conclusion:  The instrument ‘Check your health’ showed clinical utility in teenagers with diabetes. Reliability and validity tests of the measure showed promising results in Swedish teenagers, and it can probably be used in clinical settings. To further strengthen the convergent validity, it should be compared with other QoL instruments, and to obtain normative values, it has to be used in a larger context.

Introduction

The purpose of diabetes care for young people with diabetes is, referring to the International Society of Pediatric and Adolescent Diabetes’ (ISPAD’s) Consensus Guidelines 2007, to optimize metabolic control (1), prevent acute and late complications (1–3) and optimize health and quality of life (QoL) (2). Metabolic control is measured using HbA1c, while health and QoL are measured using different instruments. We found no instrument that measures burden of diabetes.

Young people with diabetes have reported progressively more impaired perceived health and QoL than healthy children, and gender differences are reported (4–6).

The fact that perceived health and QoL are impaired highlights the importance of routine assessment in children and adolescents with diabetes (7). In a review from 2007, the authors argued that it is just as important to screen for mental health as for other complications from diabetes (8). The ISPAD guidelines recommend that screening should be conducted on a routine basis (2). Burden of diabetes may be an indirect measure of QoL, but we found no literature that can support this.

In a review, De Wit et al. (9) discussed the advantages and disadvantages of generic versus disease-specific instruments. Generic instruments allow comparisons with a healthy population, while disease-specific instruments have more relevance for patients and are more sensitive in detecting clinical changes. The authors evaluated both generic and disease-specific instruments. The generic instruments varied in number of questions between 12 and 87 and the disease-specific instruments between 10 and 52. The time needed to complete the questionnaires varied between 5 and 20 min. The DisabKids Chronic Generic Module-37 (DCGM-37) is the only instrument tested for reliability and validity in different countries, thus allowing a comparison of results across countries and is also the only generic instrument for children with chronic diseases.

If we are to assess QoL and perceived health on a routine basis in a clinical setting, we need disease-specific instruments that are valid and have good reliability, but that also are short and easy to complete. The instrument ‘Check your health’, constructed by the second author (KW), is a diabetes-specific health measure that has been used in clinical settings and found to be easy to use in daily practice. It has also been tested for reliability and validity in adults with diabetes (10).

The questionnaire is constructed as four vertical thermometer scales measuring perceived physical and mental health, social relations and QoL. The construction, (described in the Patients and methods section), also allows the measurement of ‘burden of diabetes’. The instrument can be completed in less than 5 min and is easy to understand.

The aim of this study was to test the reliability and validity of the instrument ‘Check your health’ in young persons (12–17 years) with type 1 diabetes.

Patients and methods

The instrument was first used in a small sample of young people with diabetes, between 12 and 17 years old, as well as with members of two different diabetes teams. Subsequently, some words that were found difficult to understand were changed to improve the instrument’s face validity.

Subjects

A convenience sample of 204 patients with type 1 diabetes from three children’s diabetes centres in central Sweden was recruited. Five were excluded because of diabetes duration less than 6 months. Fifty of the patients were asked to complete the questionnaire a second time and 48 of those replied to the questionnaire. Mean age was 14.7 (range 12–17) years, and mean duration of diabetes was 6.3 (range 0.6–15.7) years. One hundred and fifty-eight patients were receiving multiple daily injections, and 41 used continuous subcutaneous insulin infusion (CSII). Mean HbA1c was 7.5% (SD = 1.3). Participation was voluntary, and written informed consent was obtained from both patients and their parents.

The study was approved by the ethics committee at the Karolinska Institute in Stockholm, registration no. 2005/1352-31.

Measures

Check your health

‘Check your health’ measures perceived physical and emotional health, social relations and general QoL on vertical thermometer scales, ranging from 0 to 100, with 0 indicating low perceived health/QoL. No transformation process to obtain standardized scores is required. On the same scales, a person reports what his/her imagined physical and emotional health, social relations and QoL would be if he/she did not have diabetes. The measured difference between, for example physical health with and without diabetes is defined as the physical burden of diabetes. When the difference resulted in a positive value, meaning that, for example imagined physical health without diabetes was reported to be lower than with diabetes, the burden was interpreted as zero. In this study, the marginal values for no burden, low burden, high burden or very high burden are arbitrary.

DCGM-37 and the diabetes-specific module (DCGM-37-DM

The ‘Check your health’ instrument was compared with the DCGM-37 (11,12) and the DCGM-37-DM (13). The DCGM-37 uses Likert scales ranging from 1 to 5. The responses were transformed, resulting in scores ranging from 0 to 100. The questions represent three different domains, (physical health, mental health and social relations), and the total score indicates general QoL. The diabetes-specific module has ten questions that represent the two domains ‘Diabetes acceptance’ and ‘Treatment’ (14), and one question (nr 11a, b, c) that describes perceived illness severity, in this study arbitrarily divided into ‘low severity’ (mean raw score 3–5) and ‘high severity’ (mean raw score 1–2).

The DCGM-37 has shown good test-retest reliability (0.71–0.83) and internal consistency was acceptable (alpha values 0.70–0.87). It showed relatively moderate validity (0.3–0.6) when compared with seven other questionnaires, but showed discriminant validity regarding gender, age and affluence (12).

The diabetes-specific module was tested in 207 persons with diabetes in Europe (countries unknown). The homogeneity of the domains Diabetes acceptance and Treatment showed alpha values of 0.83 and 0.84 respectively (13). Normative values have not yet been reported for the diabetes module.

HbA1c

HbA1c was measured using high-performance liquid chromatography, either as blood collected on filter paper (values were converted to the Mono-S method using the formula (Mono-S = DCCT (15) standard x 1.0678–1.341) or the DCA2000 (Siemens, gives Mono-S values). The normal reference value is <5.2%.

The group differentiation in HbA1c (<6.5%, 6.5–8%, >8%) was based on ISPAD Guidelines for 2007 (1).

Data collection procedure

The patients completed the questionnaires at the outpatient clinic before or after a regular visit, to ensure that they completed the questions without the help from the parents. Forty-eight of the 204 children completed the same questionnaires 2 weeks later at home and returned the retest by mail. Information on HbA1c, duration and medical treatment was collected from the patients’ medical records.

Statistics

All analyses were performed using Stat View 5.0 (16). Mean values (SD) were used when skewness scored less than 1.5 and kurtosis less than 5; otherwise, median values/range was calculated. Floor and ceiling effects were analysed to show the score distribution. This is reported as percentage of results achieving either 0 (floor) or 100 (ceiling). Reliability and convergent validity were analysed using Pearson’s product-moment correlation coefficient, except on the social scale where Spearman’s Rank Correlation was used, as the score distribution was skewed and kurtosis was high.

ANOVA was used in the analysis of discriminant validity, except for the dimension ‘Social relations’ and ‘Social burden’, where Mann–Whitney (gender, disease severity) or Kruskal–Wallis (HbA1c) tests were used.

Results

Reliability

Test-retest reliability was satisfactory and was tested for both self-reported health (0.78–0.66) and self-reported imagined health without diabetes (0.73–0.59) (Table 1).

Table 1.   Correlation coefficient between test and retest measures of the instrument ‘Check your health’
 Physical healthEmotional healthSocial relationsQoLPhysical health without diabetesEmotional health without diabetesSocial relations without diabetesQoL without diabetes
Correlation0.750.7780.7780.6630.7250.6440.710.590
Z-value6.5027.0557.4965.4106.0895.1846.0154.593
Correlation coefficient0.59–0.8510.636–0.8690.673–0.8840.469–0.7960.553–0.8380.443–0.7830.536–0.8260.370–0.747
p-value<0.0001<0.0001<0.0001<0.0001<0.0001<0.00010.0006<0.0001

Scoring

There were no floor effects (number of people scoring 0) on any of the four health dimensions, while the ceiling effect (people scoring 100) ranged from 5% to 30% (Table 2).

Table 2.   Mean values, SD, range and score distribution of the instrument ‘Check your health’ in 199 teenagers*
 Mean (SD)
Median
Range
Ceiling (%)Floor (%)SkewnessKurtosis
  1. *Dropouts in different domains: 2–6.

Physical health73.6 (17.2)
78.0
20–100
4.60−0.9830.861
Emotional health77.6 (18.4)
80.0
10–100
100−1.2201.806
Social health88.1 (14.5)
90.0
0–100
300.5−2.2217.605
QoL82.3 (16.6)
89.0
15–100
160−1.2091.372
Physical health without diabetes76.9 (17.6)
80.0
0–100
80.5−1.0811.452
Emotional health without diabetes83.1 (15.3)
87.0
10–100
130−1.6474.042
Social health without diabetes89.0 (13.8)
91.0
0–100
300.5−2.5029.748
QoL health without diabetes88.0 (13.4)
90.0
20–100
270−1.7243.993

There was no floor effect on any of the four dimensions in which the respondents imagined not having diabetes, and the ceiling effects ranged from 8 to 30%. Social relations and social burden were positively skewed and had high kurtosis.

Convergent validity

Correlations between ‘Check your health’ and the DCGM-37 were moderate, with correlation coefficients between 0.27 and 0.66 (Table 3).

Table 3.   Correlations between ‘Check your health’ and the DCGM-37
 DisabKids Physical health
R (p-value)
DisabKids Mental healthDisabKids Social relationsDisabKids Total transformed raw scores
  1. The bold values show the comparison between the same domains in the two instruments.

Physical health0.387 (<0.0001)
n = 184
0.427 (<0.0001)
n = 181
0.273 (0.0002)
n = 185
0.422 (<0.0001)
n = 163
Mental health0.501 (<0.0001)
n = 184
0.621 (<0.0001)
n = 181
0.416 (<0.0001)
n = 186
0.582 (<0.0001)
n = 164
Social relations0.332 (<0.0001)
187
0.459 (<0.0001)
n = 183
0.455 (<0.0001)
n = 187
0.463 (<0.0001)
n = 166
QoL0.479 (<0.0001)
n = 185
0.657 (0.0001)
n = 182
0.405 (<0.0001)
n = 187
0.58 (<0.0001)
N = 165

The domain Diabetes acceptance in the DCGM-37-DM was compared with the burden of diabetes (Check your health) and showed low correlations (0.28 and 0.38).

Discriminant validity

Based on previous findings, we hypothesized that girls would have poorer health and higher burden of diabetes than boys would, and that higher HbA1c and more serious disease would be correlated with lower self-reported health and higher burden of diabetes.

The diabetes burden domain of ‘Check your health’ showed discriminant validity on gender, perceived severity of diabetes and HbA1c (Table 4).

Table 4.   Comparison of HbA1c, gender and disease severity with burden of diabetes in 199 teenagers with diabetes
 Physical burden
Mean (SD)
Emotional burden
Mean (SD)
Social burden
Mean (SD)
QoL burden
Mean (SD)
  1. *Dropouts: HbA1c correlations 7–9: gender correlations 5–6: disease severity correlations 14–15.

HbA1c <6.5%1.419 (12.6)
n = 43
−3.045 (13.3)
n = 44
−1.227 (9.8)
n = 44
−2.909 (12.0)
n = 44
HbA1c 6.5–8%0.389 (19.0)
n = 95
−4.734 (10.7)
n = 94
−0.558 (9.0)
n = 95
−4.263 (11.1)
n = 95
HbA1c >8%−14.83 (21.3)
n = 53
−9.415 (12.3)
n = 53
−1.036 (5.1)
n = 55
−10.327 (14.2)
n = 55
F-value/H corr for ties; p-value13.585; p < 0.00014.04; p = 0.0190.877; p = 0.7435.713; p = 0.004
Boys−0.207 (18.4)
n = 92
−3.878 (11.6)
n = 90
−0.883 (7.3)
n = 94
−3.011 (10.7)
n = 92
Girls−6.475 (20.4)
n = 101
−7.317 (12.4)
n = 104
−0.765 (9.0)
n = 102
−7.88 (13.7)
n = 103
T-value/Z-value; p-value2.193; p = 0.02951.833; p = 0.0684−0.1; p = 0.922.714; p = 0.007
Low severity (Raw score 3–5)−3.006 (19.2)
n = 168
−5.268 (12.2)
n = 168
−0.842 (8.8)
n = 171
−5.3 (13.0)
n = 170
High severity (Raw score 1–2)−14.286 (22.4) n = 14−13.143 (11.6) n = 140.714 (2.7) n = 14−9.143 (9.1)
n = 14
T-values/Z-value; p-value−2.078; p = 0.0391−2.308; p = 0.02210.56; p = 0.9553−1.075; p = 0.2839

Discriminant validity in the domain perceived health showed that the lower the HbA1c, the better perceived physical health: girls report lower QoL than boys; and the more perceived disease severity, the worse perceived physical and mental health).

There was also a correlation between perceived burden of diabetes and self-reported health (Fig. 1) (If fewer than five persons reported ‘Very high burden’, they were calculated together with those reporting ‘High burden’).

Figure 1.

 ‘Check your health’ scores of HRQL in relation to perceived burden of diabetes.

Discussion

‘Check your health’ was found to be easy and quick for young people to complete and showed relatively good results. The instrument was more sensitive in assessing diabetes burden than self-reported health, which is in accordance with the results from tests on adult patients (10).

Test-retest reliability was moderate to good, and good reliability was defined as a correlation coefficient >0.7 (17). Perceived QoL, emotional health and QoL without diabetes had correlation coefficients just below 0.7. The adolescents completed the questionnaire in different environments, which may have influenced the reliability (18). The most valid data were probably received when the questionnaires were completed at the clinic, but it would have been impractical for the families to visit the clinic for the retest 2 weeks later. However, this may have lead to an underestimation of the results on reliability rather than overestimation.

The high ceiling effect, especially on perceived social health and social health without diabetes, correlates well with the DCGM-37 and with other studies regarding adolescents with different chronic conditions (4). This may be a limitation in trying to detect improvements in social relations, and one should consider whether if this domain should be withdrawn from ‘Check your health’ when used in teenagers. On the other hand, the low floor effect enables one to screen for patients with low perceived health or QoL.

The wide range of scores in the four domains of diabetes burden (−75-0) indicates that the scale discriminates between young persons with different levels of perceived burden of diabetes.

As hypothesized, ‘Check your health’ discriminated well between reported disease severities in both physical and emotional health and physical and emotional burden of diabetes. It also revealed a gender difference in reported physical health and QoL as well as burden on QoL. Previous findings have shown a correlation between gender and emotional health (4–6), but this was not the case in this study.

Our study supports previous findings that the lower the HbA1c, the better the QoL. The burden domain discriminated well between teenagers with different HbA1c values, but the correlation with perceived health was weaker. There are indications that metabolic control may correlate with HRQL, but the results of different studies are conflicting (5,6,19).

It seems as though the burden dimension in the instrument ‘Check your health’ is more sensitive than the perceived health dimension, which was also shown in the adult validation test (10).

Convergent validity was moderate in a comparison of the ‘Check your health’ questionnaire with the same four domains in the DCGM-37 (0.66–0.27). This is difficult to explain, because we have shown both good reliability and discriminant validity in the tested instrument.

Perceived mental health in ‘Check your health’ correlated with mental health and QoL in the DCGM-37, while perceived QoL was correlated with both mental health and QoL. There seems to be an overlap between mental health and QoL in the DCGM-37 questionnaire.

‘Check your health’ was found to be very suitable and easy for teenagers to complete, and because the tests were promising it seems important to continue to compare ‘Check your health’ with other measures. The measures KINDL and PedsQl are generic measures with diabetes-specific modules, which may be used in future comparisons (9).

‘Check your health’ is a new instrument and does have some limitations. For instance, it is only available in Swedish. The instrument also has to be used in a larger group of patients to obtain normative scores for both HRQL and burden of diabetes. To allow comparisons between parent and child, a proxy version is needed. Investigating the concordance between a child and a proxy report is often part of the development of a new instrument for children (9). If ‘Check your health’ is to be used as a proxy report, an assessment of concordance is needed.

Conclusions

The instrument ‘Check your health’ is well accepted by young persons with diabetes, is very easy and fast to complete, and may be a suitable tool for measuring HRQL and burden of diabetes. It showed acceptable reliability and validity. The instrument must be used in a larger population, however, to obtain normative scores. The ‘Social relations and social burden’ dimension is more suitable for screening patients with low health or high burden than for detecting improvements in patients.

Acknowledgements

The authors are grateful to all teenagers in the study as well as to the diabetes nurses for assistance in the distribution of the questionnaires to the teenagers.

Funding

This study was supported by the Swedish Diabetes Federation, the Swedish Child Diabetes Foundation and Svenska Diabetesstiftelsen.

Ancillary