The Multidimensional Illness Severity Questionnaire: Preliminary evaluation of a brief parent-reported measure of illness severity

We conducted a cross-sectional survey and interview-based study of children participating in a research study between April 2012 and May 2013 at a clinical study unit at a large academic medical centre, with one parent per child subject. Eligible children: (i) had enrolled in an eligible research study within the 90 days prior to our study team approaching them; (ii) were between 7 and 17 years old at enrolment; (iii) spoke English as their primary language; (iv) could travel to the hospital for an interview or agreed to a home interview; and (v) had no significant cognitive impairment (defined as functioning at ≥2 years below grade level as assessed by the investigator for the child’s clinical research study). Eligible studies included those that enrolled subjects ≥7 years old and for which the principal investigator of the study granted permission to our study team to approach participants during their study visits. Parents were eligible if their child agreed to participate, they were English speaking, and they gave informed consent. Parents gave permission for their children’s participation and consent for their own participation; children gave assent for their own participation. The study was approved by the institutional review board and conforms to the provisions of the Declaration of Helsinki. Potential subjects were identified through a weekly query of upcoming appointments. We contacted the investigator from each potentially eligible subject’s clinical research study to obtain permission to contact the parent and child for inclusion in this study and to verify eligibility. A study team member approached each potential child–parent pair during their scheduled research visit to provide information, assess eligibility and obtain consent and assent from interested individuals. We developed the MISQ after review of the literature on illness severity and impact of illness perception on decision-making. It addresses five domains of illness severity: (i) threat to life, (ii) functional limitation, (iii) symptom burden, (iv) treatment options and (v) optimism about recovery. Parents rated threat to life, functional limitation and symptom burden on a 4-point ordinal scale as none, mild, moderate or severe and optimism about recovery as very optimistic, optimistic, somewhat optimistic or not at all optimistic, and treatment options on a 5-point scale as excellent, good, fair, poor or none. Higher scores indicate more significant threat to life, limitation and symptom burden; fewer treatment options; and less optimism about recovery. The total scale score could range from 5 to 21. In order to ensure proper understanding of the questions on the MISQ, the first five parents participated in cognitive interviews to allow modification and optimisation of study instruments. Cognitive interview questions were asked after the parent had completed the full study interview and survey to assess how each question on the MISQ was interpreted and whether it was confusing. The MISQ was modified after the first two cognitive interviews to anchor parents on the prior 6 months of their child’s illness. Only participants enrolled after this modification are included in the present analysis. No additional changes were suggested in the remaining three cognitive interviews; these responses are therefore included in this report. In addition to the MISQ, parents’ perceptions of illness severity were assessed using four distinct measures: (i) the Lansky Play Performance Scale (Lansky) 15 ; (ii) a previously validated Severity of Illness Scale (SOIS) 5 ; (iii) modified subscales of the Parent Experience of Child Illness (PECI) 13 ; and (iv) parents’ report of their child’s general health. The Lansky is a functional performance scale for children based on daily play activities of the child. Primarily developed for use in paediatric oncology, it is applicable through age 16 and can be reported and scored by parent observation. Level of play is rated in 10-point increments ranging from fully active, normal (100) to completely disabled (10). The scale has shown good inter-rater reliability (IRR) between two parent raters of the same child (Pearson’s r = 0.65, P < 0.0001). It has also shown discriminant and concurrent validity as established by its ability to distinguish between children with different functional levels in the inpatient, outpatient and healthy child populations (analysis of variance with P < 0.0001), as well as correlation between Lansky performance score and nurse- ( r = 0.75, P < 0.001) and physician- ( r = 0.92, P < 0.0001) reported global ratings. 15 The SOIS was designed for physician and nurse reports of illness severity in a population of cancer patients. It is comprised of six questions based on five dimensions of illness severity: degree of impairment, future outlook, quantity of medical procedures required, number of hospitalisations and ability to participate in activities. Each item is rated on a scale from 1 to 7, with the sum reflecting a unidimensional measure of illness severity (total possible score = 6–42). The SOIS demonstrated good internal consistency (Cronbach’s alpha for physicians = 0.8 and for nurses = 0.79), test–retest reliability (Pearson’s r for 2 weeks = 0.96 and 3 months = 0.92) and IRR (IRR = 0.89 between nurse and physician ratings). The SOIS also demonstrated validity as cancer patients who relapsed or who required bone marrow transplant had higher SOIS scores. 5 In our study, we administered the SOIS to parents. While it has not been previously validated specifically for parent report, the questions overlap with domains in other instruments that have been used for parent report. The PECI, originally developed and tested on parents of children with brain tumours, was developed to assess parent adjustment to their child’s illness. Two subscales of the PECI – guilt and worry and long-term uncertainty – were included in the current study. Both subscales demonstrated good internal consistency (Cronbach’s alpha for guilt and worry = 0.89 and for long-term uncertainty = 0.8) and construct validity as measured by correlations with other measures of distress. Both subscales correlated moderately with internalised subjective burden as measured on the Caregiver Strain Questionnaire (Pearson’s r for guilt and worry = 0.6, P = 0.008 and for long-term uncertainty = 0.46, P < 0.004), and the guilt and worry subscale correlated moderately with intrusion as measured by the impact of event scale ( r = 0.51, P < 0.013). 13 Parents’ reports of their child’s general health were assessed using a visual analogue scale ranging from 1 (worst) to 10 (best) (degree of health their child has ever had) Changes in health trajectory using this scale at various time points in a child’s illness have been shown to relate to decision-making regarding treatment limitations in the clinical setting. 14 After obtaining informed consent, parents participated in interviews related to their child’s research study. Interviews were audio-recorded and subsequently transcribed verbatim for analysis. After the interviews, parents completed the five illness perception measures via self-administered paper surveys. Parents also reported demographic information for themselves and their children. Each parent–child dyad received a $25 gift card for participation. The primary objective of the overall study was to describe differences in parent and child understanding across a variety of research studies. In this report, our main objective is to describe the performance of the MISQ. We also report correlations among all measures of illness severity. Cronbach’s alpha was calculated to assess internal consistency of the MISQ. 16 Descriptive statistics are reported for each measure of illness severity. We calculated Spearman correlations and 95% confidence intervals (CIs) to describe the relationships between measures of illness perception. 16

The study sample included 31 parent–child dyads. Interviews were initiated with 33 families; however, responses from two parent–child dyads were excluded because we determined after initiating the interviews that they did not meet inclusion criteria due to language ( n = 1) and cognitive limitations ( n = 1). The remaining 31 parent–child dyads were included in the interview study. The response rate among eligible and approached families was 31/76 (41%; Fig. 1 ). Of the 31 children in the total study sample, 20 were enrolled in their research study due to illness; 11 were enrolled as healthy controls and are excluded from this analysis. The 20 ill children are the focus of the present report. The mean age of child participants was 12.4 years ( Table 1 ). Just over one-third of child participants ( n = 4, 36%) were female. All children were English speaking, and the majority (68%) self-identified as white, non-Hispanic. Ten (50%) had previously participated in research studies. The majority of parent respondents were female ( n = 18, 90%), married ( n = 12, 60%) and had completed at least 4 years of college (55%). Children ( n = 31) were enrolled from protocols that included spinal muscular atrophy ( n = 2), muscular dystrophy ( n = 3), endometriosis ( n = 6), short stature ( n = 4), asthma ( n = 9), autism spectrum disorder ( n = 6) and sickle cell disease 1 ; those with illness ( n = 20) included spinal muscular atrophy ( n = 2), endometriosis ( n = 4), short stature ( n = 1), asthma ( n = 9), autism spectrum disorder ( n = 3) and sickle cell disease ( n = 1). Items on the MISQ and responses are reported in Table 2 . Detailed responses are shown in Appendix S1 ( Supporting Information ). Scores for functional limitation (mean = 2.25, standard deviation (SD) = 0.79), symptom burden (mean = 2.85, SD = 0.93), treatment options (mean = 2.2, SD = 1.28) and optimism about illness (mean = 2.5, SD = 1.19) were higher than scores for threat to life (mean = 1.7, SD = 1.03). The MISQ had adequate internal consistency (Cronbach’s α = 0.76). 17 Table 3 summarises the responses on all illness severity measures. Higher scores on the MISQ, SOIS and PECI and lower scores on the Overall Health and Lansky indicate greater illness severity. Mean scores were 11.5/21 on the MISQ, 16.1/42 on the SOIS, 30.2/56 on the PECI, 87/100 on the Lansky and 6.5/10 on the Overall Health Scale. Correlations among measures of illness severity are reported in Table 4 . Although CIs are wide due to the limited sample size, the MISQ correlated strongly and in the expected directions with the PECI (Spearman’s rho = 0.75, 95% CI = 0.44–0.9), Lansky (Spearman’s rho = −0.75, 95% CI = −0.9, −0.46) and parent rating of overall health (Spearman’s rho = −0.76, 95% CI = −0.9, −0.46). In contrast, the MISQ did not correlate significantly with the SOIS.

How parents perceive their child’s illness severity influences how they cope with the illness, engage with treatment and seek out needed supports. An instrument that is easily and quickly completed by parents and that applies across diverse diseases and ages is critical to advancing paediatric health services and outcomes research. Preliminary evaluation of the MISQ suggests that it is an easy-to-administer and internally consistent multidimensional assessment of illness severity that holds promise in meeting this need, potentially for real-time use in assessing parent illness perceptions. Further studies involving larger and more diverse populations are needed to confirm its reliability and validity as a measure of this essential construct.

Multiple studies highlight the importance of the perception of illness severity as perception relates to key patient-related outcomes, including compliance and willingness to engage in treatment and seek essential resources. 2 - 4 Patient- or parent-reported measures of illness severity are essential to the conduct of health services and outcomes research. We designed and conducted a preliminary analysis of a new measure to assess parents’ perceptions of their child’s illness severity, which we conceptualised as a multidimensional construct related to prognosis, symptom and treatment burden and optimism about recovery. In the absence of an existing instrument usable across the range of childhood illness and that encompasses the multiple domains of illness severity, we designed a measure to apply across illness types. We evaluated its internal consistency as a measure of reliability, as well as its concurrent validity, against four other parent-reported measure of illness severity. The strong correlations between the MISQ and other measures of illness severity, including overall health rating, functional performance and parental guilt and worry, along with its adequate internal consistency and ease of administration, suggest that it is a promising new measure to assess parent-reported illness severity across variable disease states and child ages. In addition, it appears to capture a variety of illness constructs, including ‘severity’, in its more traditional sense related to prognosis and symptom burden but extending to functional performance and parent optimism. The MISQ promises to capture these constructs in a brief, simple-to-administer five-question assessment. These observations support its potential use in future health outcomes studies involving paediatric patients. Although this study supports the potential of the MISQ in assessing parent perception of their child’s illness severity, some limitations should be noted. First, as this was a single institution study with a relatively homogenous study population, further study across a range of populations is required to assess generalisability. In addition, although we demonstrated adequate internal consistency and concurrent validity, we did not assess other dimensions of reliability, including test–retest and IRR, or correlations between MISQ scores and physician or nurse ratings of illness. Future studies in larger samples should address these questions. As the MISQ correlates with several measures of illness perception, we believe it is usable in its current form, although larger studies may demonstrate a need for modification. Finally, although the sample size was large enough to demonstrate correlations between multiple measures, a larger sample size would allow more precise estimates of correlations. Several challenges could have contributed to the lower sample size in this study. Recruitment occurred through an outpatient research centre, which limited the times to approach families when they were present for another study visit. In addition, as parents were already participating in another study, the added burden of additional questionnaires and time for individuals who had planned for a study visit may have led to limited participation. Importantly, while this pilot was conducted with a small sample size, conditions ranged across varying levels of severity, including developmental conditions such as autism to chronic conditions such as asthma and life-limiting conditions such as spinal muscular atrophy.

Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Appendix S1. Detailed responses to the Multidimensional Illness Severity Questionnaire.