Patient perceptions of diabetes and diabetestherapy: assessing quality of life

Clare Bradley*

Jane Speight

Department of Psychology, RoyalHolloway, University of London,Egham, UK

*Correspondence to: Prof. C.Bradley, Department of Psychology,Royal Holloway, University ofLondon, Egham, Surrey TW20 0EX,UK.E-mail: c.bradley@rhul.ac.uk

Received: 2 October 2001

Summary

Efforts to prevent complications of diabetes often overlook the impact of the

condition and its treatment on current quality of life (QoL). The Diabetes

Treatment Satisfaction Questionnaire (DTSQ) has proved valuable for

understanding and measuring patients’ treatment satisfaction in assessments

of new treatments and strategies. For example, the DTSQ has demonstrated

improved patient satisfaction with fast-acting insulin lispro versus standard

soluble insulin and with long-acting insulin glargine versus NPH insulin.

However, improvements in treatment satisfaction are often inferred to be

improvements in overall QoL without recognizing the limited scope of the

satisfaction measure. It is necessary to evaluate not only satisfaction with

treatment per se but also the impact of diabetes and its treatment on a broad

range of life domains in order to assess the impact on QoL. The Audit of

Diabetes-Dependent Quality of Life (ADDQoL) measure is a diabetes-specific

instrument that assesses the impact of diabetes on 18 life domains. Use of the

ADDQoL with people with type 1 or type 2 diabetes has shown, on average,

almost universally negative impact of diabetes on all domains. Significant

differences have also been shown in the magnitude of effect between insulin-

treated and non-insulin-treated patients and patients with and without

complications. The negative impact of diabetes on QoL has been observed

despite high levels of treatment satisfaction (as measured by the DTSQ). The

greatest negative impact was observed for the domain ‘Freedom to eat as I

wish’, indicating the strong influence of dietary restrictions on QoL. Studies to

assess the outcomes of treatment approaches designed to improve dietary

flexibility are under way. Copyright # 2002 John Wiley & Sons, Ltd.

Keywords Audit of Diabetes-Dependent Quality of Life; Diabetes Treatment

Satisfaction Questionnaire; insulin glargine; insulin lispro

Introduction

Health professionals, striving to prevent complications of diabetes and

thereby protect quality of life (QoL) in the long term, often overlook the

importance of protecting QoL in the short term. Increased attention is being

given to measures of patient perceptions of the effects of the condition and its

treatment in an effort to improve the overall management of diabetes. Two

instruments that have proved to be effective in distinguishing factors

important for patient satisfaction with treatment and diabetes-specific QoL

are the Diabetes Treatment Satisfaction Questionnaire (DTSQ) [1] and the

Audit of Diabetes-Dependent Quality of Life (ADDQoL) [2].

DIABETES/METABOLISM RESEARCH AND REVIEWSDiabetes Metab Res Rev 2002; 18: S64–S69.Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002 / dmrr.279

Copyright # 2002 John Wiley & Sons, Ltd.

Diabetes Treatment SatisfactionQuestionnaire (DTSQ)

The DTSQ consists of a six-item scale assessing treatmentsatisfaction and two items assessing perceived frequencyof hyperglycemia and hypoglycemia. Together with the22-item Well-Being Questionnaire [3], this instrumenthas been identified by the World Health Organization(WHO) and the International Diabetes Foundation (IDF)as useful in assessing outcomes of diabetes care [4].

The DTSQ has been extensively used to measurepsychological outcomes in diabetes management andhas proved sensitive to changes in patient response to avariety of interventions, including switching from oralagents to insulin injections and switching between insulinpreparations. For example, despite the fact that switchingto injectable insulin is viewed as a major concern amongpatients taking oral agents, use of the DTSQ has shownthat patient satisfaction with treatment usually improvesupon switching to insulin [5–7]. This is not to say thatpatients prefer injections to tablets, but that there aremany other perceived benefits of insulin treatment thatcan lead to improvements in satisfaction with treatment(e.g. increased flexibility).

Use of DTSQ in clinical trials of newinsulin analogues

The DTSQ has been used in several clinical trialsevaluating new diabetes treatments, including the newfast-acting insulin analogue, insulin lispro [8–10]. Thispreparation can be injected at the time of eating andoffers several advantages over standard soluble insulin,which is recommended for use 30 min before meals.Insulin lispro allows greater flexibility of mealtimes withless risk of hypoglycemia, which would be likely to followattempts at mealtime flexibility with use of standardsoluble insulin. Use of the DTSQ in these studies showedimprovement in treatment satisfaction with insulin lisproby specifically demonstrating patient-identified advan-tages regarding convenience and flexibility.

The DTSQ has also identified advantages associatedwith use of the new long-acting analogue insulin glarginein randomized trials comparing the agent with NPHinsulin. Insulin glargine is designed to provide 24-hourcoverage of basal insulin needs with a single daily injec-tion, whereas NPH insulin usually needs to be taken twicedaily to achieve 24-hour coverage [11]. Thus, use ofinsulin glargine might be expected to be associated withincreased satisfaction with blood glucose control and/orgreater convenience. In a trial conducted in 474 patientswith type 1 diabetes in the US, changes in treatmentsatisfaction were assessed at Weeks 8, 20, and 28 [12].Although baseline treatment satisfaction levels werehigh in these patients (mean 28.8 of a maximum scoreof 36), a switch to insulin glargine was associated with apronounced improvement in treatment satisfaction com-pared with continued use of NPH insulin, with the

difference becoming significant at Week 20. Similarly, useof the DTSQ in a European trial with 517 patients withtype 1 diabetes (mean baseline satisfaction 28.0) showedthat insulin glargine was associated with improvedtreatment satisfaction at all time points (Weeks 8, 20,and 28), whereas satisfaction deteriorated slightly inNPH-treated patients. Differences between treatmentswere statistically significant throughout the Europeanstudy [13].

It is noteworthy that with the intention of standardizingthe treatment protocol, the European glargine trialrequired switching to standard soluble insulin in allpatients who had been using insulin lispro as their short-acting meal-related insulin, and to the OptiPen1 in thosewho had been using other makes of pens or conventionalsyringes. Treatment satisfaction at the last visit fellsignificantly in patients who had substituted standardinsulin for insulin lispro, and an increase in satisfactionwas noted in those who had substituted the OptiPen1 forconventional syringes. Nevertheless, more favorablesatisfaction scores were observed with insulin glarginethan with NPH insulin in both subgroups. The value of theDTSQ is thus highlighted by its ability to detect changes intreatment satisfaction associated with different basalinsulin use, despite changes in other treatment factorsthat were also measured.

The ceiling effect

Notwithstanding the sensitivity to change of the DTSQ,ceiling effects are often seen with this instrument, wheremaximum or close-to-maximum scores at baseline pro-vide little opportunity for registering improvement insatisfaction with the treatment or strategy being assessed.Thus, for example, in crossover trials of insulin lispro,some improvement in treatment satisfaction was obser-ved in patients allocated to receive insulin lispro first.However, the magnitude of the effect of lispro on treat-ment satisfaction was most evident when patients repor-ted a marked reduction in satisfaction after they had crossedback over to standard soluble insulin treatment [8,10].

When assignment sequence in crossover trials has asignificant effect on outcomes, the phases of the trialscannot be combined to make optimal use of sample sizeand gain statistical power to detect differences. Instead,the two halves of the trial, before and after crossover,have to be analyzed separately, and power is reduced.Ceiling effects with the DTSQ, in its original status form,can contribute to sequence effects in crossover trials. TheDTSQ-change version (DTSQc) was designed to overcomeceiling effects [14]. This instrument contains the sameeight items as the DTSQ-status version but asks patients toconsider their satisfaction with their current treatmentcompared with their previous treatment. Thus, when usedtogether in the clinical trial setting, the status version,used at baseline, permits assessment of absolute levels ofsatisfaction, while the change version measures relativechange, reflecting increased or decreased satisfaction orno change in satisfaction. Findings in an Austrian study

Patient Perceptions of Diabetes S65

Copyright # 2002 John Wiley & Sons, Ltd. Diabetes Metab Res Rev 2002; 18: S64–S69.

examining insulin lispro versus standard soluble insulin(in a fully flexible functional insulin treatment strategy)indicated that use of the DTSQ-change version increasessensitivity of the DTSQ to changes in patient satisfaction[10], a finding that has since been confirmed in trials ofinsulin glargine [15].

Audit of Diabetes-Dependent Qualityof Life (ADDQoL)

Although the DTSQ has proved to be a valuableinstrument in assessing patient satisfaction with treat-ment, patient satisfaction nevertheless constitutes onlyone outcome of potential importance in evaluating dis-ease and treatment effects. Although treatment satisfac-tion may be an important influence on QoL (and mayeven be considered to be an aspect of QoL), the DTSQdoes not itself measure QoL. An understanding of theeffect of diabetes on QoL requires broader coverage ofaspects of life likely to be influenced by the condition, itstreatment, and any complications. The ADDQoL has beendesigned to meet this need by permitting users to indicatewhether potentially affected domains of life apply to themand to rate the impact of their diabetes on all applicableaspects of life, together with the perceived importance ofeach domain for their QoL.

QoL domains

The design of the ADDQoL was influenced by the philo-sophy underpinning the development of an interview

approach to QoL measurement – the Schedule for

the Evaluation of Individual Quality of Life (SEIQoL),

devised in recognition of the need to measure indivi-

dual perceptions of QoL. The SEIQoL asks respondents

to generate aspects of life that are important to them,

evaluate how good or bad each aspect is currently

felt to be and indicate the importance of each for QoL

[16]. This approach was adapted to address diabetes-

specific issues and presented in a questionnaire format,

resulting in the creation of the initial 13-domain ADDQoL

[2].Continuing development of the ADDQoL has been

influenced by qualitative research to design a measure

of Renal-Dependent Quality of Life (the RDQoL) [17].

Newly designed domains from the RDQoL have been

added to the ADDQoL to extend its relevance to people

with complications of diabetes, and a 20-item ADDQoL

was evaluated in the British Diabetic Association (BDA)-

funded research programme, DIABQoL+ [18]. An 18-

domain instrument is now confirmed for use.An example of an ADDQoL item representing a specific

domain is shown in Figure 1. A ‘not applicable’ (N/A)

option is provided for domains that may not be applicable

to a given individual. Otherwise, users rate the impact of

diabetes on each particular domain and the importance of

that domain for their QoL these two scores are multiplied

to provide a weighted impact score for each applicable

domain (range x9 to +9). An average weighted impact

score is derived by summing the weighted impact scores

for each domain and dividing by the number of applic-

able domains (range x9 to +9). Thus, non-applicable

domains make no contribution to an individual’s score.

Figure 1. Audit of Diabetes-Dependent Quality of Life (ADDQoL) domain-specific item showing assigned scores for magnitudeof effect on domain (x3 to +3) and importance of domain (0 to 3). This is one of the domains for which selection of ‘notapplicable’ (N/A) is possible

S66 C. Bradley and J. Speight

Copyright # 2002 John Wiley & Sons, Ltd. Diabetes Metab Res Rev 2002; 18: S64–S69.

Use of the ADDQoL in hospital diabetesclinics

As part of the DIABQoL+ research, the ADDQoL in itsextended, 20-item form was completed by 795 out-patients attending annual review at one of two UK hospi-tal diabetes clinics. Hardly any impact was shown for thenew ‘Spiritual/religious life’ item, and factor analysisrevealed that it did not load well with other items; as aresult, this item was removed from the measure beforefollow-up use in the DIABQoL+ Programme. One ofthe original items, ‘Others fussing’, was also removedbecause it did not load well on the factor analysis andhad relatively low loadings in the original 13-itemADDQoL.

In addition, several other items were modified toimprove either their relevance or their readability.‘Sporting, holidays or leisure opportunities’ was modifiedto read ‘My holidays or leisure activities’ (because olderpeople were less inclined to consider sporting opportu-nities as relevant to them). ‘Worries about my future andthe future of others close to me’ was simplified to ‘Worriesabout the future’, because the original was unnecessarilycomplex and did not load well on a forced one-factoranalysis). ‘Problems with travelling (either local or longdistance)’ was changed to ‘Ease of travelling (either localor long distance)’, because the factor loading was on thelow side. ‘My freedom to drink as I wish (includingsweetened tea and coffee, soft drinks, fruit juices, andalcohol)’ was shortened to ‘My freedom to drink as I wish(e.g. sweetened hot and cold drinks, fruit juice, alcohol)’(Flesch reading ease score improved from 66.5 to 76.7).

The new 18-item ADDQoL [19] was further evaluatedin the second-round data collection of the DIABQoL+Programme. Unforced factor analysis with varimaxrotation on the weighted ADDQoL scores produced twofactors with eigenvalues greater than 1, with all items

loading >0.56 on component 1 except ‘Sex life’, ‘Freedomto eat as I wish’, ‘Enjoyment of food’ and ‘Free-dom to drink as I wish’, which loaded >0.41 on compon-ent 2. When a forced one-factor analysis was performed,all 18 items loaded greater than 0.5 (item 14, ‘Depen-dence on others’, had the lowest loading of 0.518).Cronbach’s standardized item alpha was high (0.92),indicating that the 18-item scale is highly reliable.

Table 1 shows mean and median values, standarddeviations (SDs) and ranges of weighted scores. Withinthese scores, there is considerable individual variability,but mean values indicate that across the group, onaverage, not one aspect of life was enhanced by diabetes.All aspects were impaired to some extent, although someindividuals did experience some specific benefits ofdiabetes for their QoL (as indicated by the range ofscores). The greatest negative impact was observed forthe domain ‘Freedom to eat as I wish’ (mean weightedimpact score x3.6); this finding was seen in both insulin-treated and non-insulin-treated patients. Averageweighted impact was more negative for insulin-treatedpatients than for non-insulin-treated patients (p<0.01),although this was not the case for every domain(Figure 2). Patients with diabetes complications reportedsignificantly greater negative impact of diabetes on QoLthan did those without complications (p<0.001). It is ofinterest to note that the overall impact of diabetes on QoLin the study population was profoundly negative, asindicated by an average weighted impact score of x1.96,but accompanying DTSQ findings showed relativelyhigh satisfaction with treatment (mean score 27.76)(Figure 3). Thus, if treatment satisfaction had been usedas an indicator of QoL in this study, the negative impact ofdiabetes on QoL would not have been recognized, and thehigh levels of treatment satisfaction may have beenmisinterpreted to suggest that patients had good QoL[19].

Table 1. Weighted impact scores for quality of life (QoL) domains in the DIABQoL+ study

Life domain Weighted scores

Mean Median SD Range

Freedom to eat as I wish x3.6 x3 3.1 x9 to +9Enjoyment of food x3.0 x2 3.1 x9 to +6Family lifea x2.6 x2 3.0 x9 to +6Working life and work related opportunitiesa x2.6 x2 3.1 x9 to +6Sex lifea x2.6 x1 3.3 x9 to 0The things I could do physically x2.4 x2 2.9 x9 to +4Worries about the future x2.4 x2 3.7 x9 to +9Holidays or leisure activities x2.4 x1 2.8 x9 to +9Freedom to drink as I wish x2.1 x1 2.9 x9 to +9Confidence in my ability to do things x1.8 0 2.6 x9 to +6Friendships and social life x1.8 0 2.6 x9 to +2Motivation to achieve things x1.6 0 2.6 x9 to +6Ease of traveling (local or long distance) x1.4 0 3.2 x9 to +9Physical appearance x1.2 0 2.5 x9 to +9Finances x1.2 0 2.5 x9 to +4Living conditions x1.1 0 2.4 x9 to +3Unwanted dependence on others x1.1 0 2.7 x9 to +9The way society reacts to me x0.9 0 2.0 x9 to +4

aItems contain ‘not applicable’ (N/A) response option.SD, Standard deviation.

Patient Perceptions of Diabetes S67

Copyright # 2002 John Wiley & Sons, Ltd. Diabetes Metab Res Rev 2002; 18: S64–S69.

Concluding comments

In general, these findings indicate that the ADDQoL is

sensitive to effects of diabetes (including both its

treatment and complications) that cannot be captured

by the measurement of treatment satisfaction alone. The

ADDQoL identifies more negative psychological outcomes

than the DTSQ, and thus it is likely to be even more

Figure 2. Mean weighted impact scores for 18 domains in Audit of Diabetes-Dependent Quality of Life (ADDQoL) comparinginsulin-treated and non-insulin-treated patients in the DIABQoL+ study. p Values indicate significance of differences betweengroups (*p<0.05; **p<0.01; ***p=0.001)

Figure 3. Mean Diabetes Treatment Satisfaction Questionnaire (DTSQ) treatment satisfaction scores (left) and Audit of Diabetes-Dependent Quality of Life (ADDQoL) average weighted impact scores (right) among insulin-treated and non-insulin-treatedpatients in the DIABQoL+ study (**p<0.01). Reprinted, with permission, from Speight J, Bradley C. Diabetologia 2000.Copyright # Springer-Verlag

S68 C. Bradley and J. Speight

Copyright # 2002 John Wiley & Sons, Ltd. Diabetes Metab Res Rev 2002; 18: S64–S69.

sensitive to improvements following change to a newtreatment that protects aspects of life that are importantfor QoL. In particular, the DIABQoL+ study has shownthat restrictions on dietary freedom have a major negativeimpact on QoL, suggesting that treatments that increasedietary freedom without loss of metabolic control willimprove QoL for many patients. This theory is beingtested in a trial examining the effects of glimepiride – aonce-daily, food-activated oral sulfonylurea that permitsflexibility of meal timing – and in a trial by the DoseAdjustment For Normal Eating (DAFNE) study groupevaluating the Dusseldorf approach to insulin treatmentdesigned to increase dietary flexibility [20,21].

References

1. Bradley C (ed). Diabetes Treatment Satisfaction Questionnaire.In Handbook of Psychology and Diabetes: A Guide to PsychologicalMeasurement in Diabetes Research and Practice. HarwoodAcademic, 1994; 111–152.

2. Bradley C, Todd C, Gorton T, Symonds E, Martin A, Plowright R.The development of an individualized questionnaire measure ofperceived impact of diabetes on quality of life: the ADDQoL.Qual Life Res 1999; 8: 79–91.

3. Bradley C. The Well-being Questionnaire. In Handbook ofPsychology and Diabetes: A Guide To Psychological Measurementin Diabetes Research and Practice, Bradley C (ed.). HarwoodAcademic, 1994; 89–109.

4. Bradley C, Gamsu DS. Guidelines for encouraging psychologicalwell-being: report of a working group of the World HealthOrganization Regional Office for Europe and InternationalDiabetes Federation European Region St Vincent DeclarationAction Programme for Diabetes. Diabet Med 1994; 11: 510–516.

5. Jennings AM, Lewis KS, Murdoch S, Talbot JF, Bradley C, WardJD. Randomized trial comparing continuous subcutaneousinsulin infusion and conventional insulin therapy in type IIdiabetic patients poorly controlled with sulfonylureas. DiabetesCare 1991; 14: 738–744.

6. Taylor R, Foster D, Kyne-Grzebalski D, Vanderpump M. Insulinregimens for the non-insulin dependent: impact on diurnalmetabolic state and quality of life. Diabet Med 1994; 11:551–557.

7. Witthaus E, Stewart J, Bradley C. Improved psychologicaloutcomes after initiation of insulin treatment in patientswith type II diabetes. Diabetologia 2000; 43 (Suppl. 1): A205(Abstract 787).

8. Janes JM, Bradley C, Rees A. Preferences for, and improvements

in aspects of quality of life (QOL) with, insulin lispro in amultiple injection regimen (Abstract). Diabetologia 1997; 40(Suppl. 1): A353.

9. Renner R, Pfutzner A, Trautmann M, Harzer O, Sauter K,Landgraf R, on behalf of the German Humalog-CSII StudyGroup. Use of insulin lispro in continuous subcutaneous insulininfusion treatment. Diabetes Care 1999; 22: 784–788.

10. Howorka K, Pumprla J, Schlusche C, Wagner-Nosiska M,Shabmann A, Bradley C. Dealing with ceiling baseline treatmentsatisfaction level in patients with diabetes under flexible,functional insulin treatment: assessment of improvements intreatment satisfaction with a new insulin analogue. Qual Life Res2000; 9: 915–930.

11. Heinemann L, Linkeschova R, Rave K, Hompesch B, Sedlak M,Heise T. Time–action profile of the long-acting insulin analoginsulin glargine (HOE901) in comparison with those of NPHinsulin and placebo. Diabetes Care 2000; 23: 644–649.

12. Bradley C, Witthaus E, Stewart J. Treatment satisfaction andpsychological well-being in patients with type I diabetes, treatedwith a new long-acting insulin insulin glargine. Diabetes 1999;48 (Suppl. 1): A353 (Abstract 1547).

13. Witthaus E, Stewart J, Bradley C. Treatment satisfaction andpsychological well-being with insulin glargine compared withNPH in patients with type 1 diabetes. Diabet Med 2001; 16:619–625.

14. Bradley C. Diabetes treatment satisfaction questionnaire.Change version for use alongside status version providesappropriate solution where ceiling effects occur. Diabetes Care1999; 22: 530–532.

15. Bradley C, Plowright R, Stewart J, Witthaus E. Diabetestreatment satisfaction questionnaire (change) in English andGerman evaluated in insulin glargine trials. Diabetologia 2000;43 (Suppl. 1): A196 (Abstract 754).

16. McGee HM, O’Boyle CA, Hickey A, O’Malley K, Joyce CR.Assessing the quality of life of the individual: the SEIQoL with ahealthy and a gastroenterology unit population. Psychol Med1991; 21: 749–759.

17. Bradley C. Design of a renal-dependent individualized quality oflife questionnaire. Adv Perit Dial 1997; 13: 116–120.

18. Speight J, Barendse S, Bradley C. The DIABQoL+ Programme.Proc Br Psych Soc 1999; 7: 35.

19. Speight J, Bradley C. ADDQoL indicates negative impact ofdiabetes on quality of life despite high levels of satisfaction withtreatment. Diabetologia 2000; 43 (Suppl. 1): A225 (Abstract864).

20. Muhlhauser I, Bruckner I, Berger M, et al. Evaluation of anintensified insulin treatment and teaching programme asroutine management of type 1 (insulin-dependent) diabetes.The Bucharest-Dusseldorf Study. Diabetologia 1987; 30:681–690.

21. The DAFNE Group. The DAFNE (Dose Adjustment For NormalEating) Trial: the end of the diabetic diet in Type 1 diabetes.Diabet Med 2002; 19 (Suppl. 2): 3.

Patient Perceptions of Diabetes S69

Copyright # 2002 John Wiley & Sons, Ltd. Diabetes Metab Res Rev 2002; 18: S64–S69.